linux/mm/madvise.c

1556 lines
40 KiB
C
Raw Normal View History

License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 14:07:57 +00:00
// SPDX-License-Identifier: GPL-2.0
/*
* linux/mm/madvise.c
*
* Copyright (C) 1999 Linus Torvalds
* Copyright (C) 2002 Christoph Hellwig
*/
#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/syscalls.h>
#include <linux/mempolicy.h>
#include <linux/page-isolation.h>
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
#include <linux/page_idle.h>
#include <linux/userfaultfd_k.h>
#include <linux/hugetlb.h>
#include <linux/falloc.h>
#include <linux/fadvise.h>
#include <linux/sched.h>
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
#include <linux/sched/mm.h>
mm: move anon_vma declarations to linux/mm_inline.h The patch to add anonymous vma names causes a build failure in some configurations: include/linux/mm_types.h: In function 'is_same_vma_anon_name': include/linux/mm_types.h:924:37: error: implicit declaration of function 'strcmp' [-Werror=implicit-function-declaration] 924 | return name && vma_name && !strcmp(name, vma_name); | ^~~~~~ include/linux/mm_types.h:22:1: note: 'strcmp' is defined in header '<string.h>'; did you forget to '#include <string.h>'? This should not really be part of linux/mm_types.h in the first place, as that header is meant to only contain structure defintions and need a minimum set of indirect includes itself. While the header clearly includes more than it should at this point, let's not make it worse by including string.h as well, which would pull in the expensive (compile-speed wise) fortify-string logic. Move the new functions into a separate header that only needs to be included in a couple of locations. Link: https://lkml.kernel.org/r/20211207125710.2503446-1-arnd@kernel.org Fixes: "mm: add a field to store names for private anonymous memory" Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Colin Cross <ccross@google.com> Cc: Eric Biederman <ebiederm@xmission.com> Cc: Kees Cook <keescook@chromium.org> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:06:07 +00:00
#include <linux/mm_inline.h>
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
#include <linux/string.h>
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
#include <linux/uio.h>
ksm: the mm interface to ksm This patch presents the mm interface to a dummy version of ksm.c, for better scrutiny of that interface: the real ksm.c follows later. When CONFIG_KSM is not set, madvise(2) reject MADV_MERGEABLE and MADV_UNMERGEABLE with EINVAL, since that seems more helpful than pretending that they can be serviced. But when CONFIG_KSM=y, accept them even if KSM is not currently running, and even on areas which KSM will not touch (e.g. hugetlb or shared file or special driver mappings). Like other madvices, report ENOMEM despite success if any area in the range is unmapped, and use EAGAIN to report out of memory. Define vma flag VM_MERGEABLE to identify an area on which KSM may try merging pages: leave it to ksm_madvise() to decide whether to set it. Define mm flag MMF_VM_MERGEABLE to identify an mm which might contain VM_MERGEABLE areas, to minimize callouts when forking or exiting. Based upon earlier patches by Chris Wright and Izik Eidus. Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Chris Wright <chrisw@redhat.com> Signed-off-by: Izik Eidus <ieidus@redhat.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Rik van Riel <riel@redhat.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Lee Schermerhorn <lee.schermerhorn@hp.com> Cc: Avi Kivity <avi@redhat.com> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-09-22 00:01:57 +00:00
#include <linux/ksm.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/blkdev.h>
#include <linux/backing-dev.h>
#include <linux/pagewalk.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/shmem_fs.h>
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
#include <linux/mmu_notifier.h>
#include <asm/tlb.h>
#include "internal.h"
mm: create new mm/swap.h header file Patch series "MM changes to improve swap-over-NFS support". Assorted improvements for swap-via-filesystem. This is a resend of these patches, rebased on current HEAD. The only substantial changes is that swap_dirty_folio has replaced swap_set_page_dirty. Currently swap-via-fs (SWP_FS_OPS) doesn't work for any filesystem. It has previously worked for NFS but that broke a few releases back. This series changes to use a new ->swap_rw rather than ->readpage and ->direct_IO. It also makes other improvements. There is a companion series already in linux-next which fixes various issues with NFS. Once both series land, a final patch is needed which changes NFS over to use ->swap_rw. This patch (of 10): Many functions declared in include/linux/swap.h are only used within mm/ Create a new "mm/swap.h" and move some of these declarations there. Remove the redundant 'extern' from the function declarations. [akpm@linux-foundation.org: mm/memory-failure.c needs mm/swap.h] Link: https://lkml.kernel.org/r/164859751830.29473.5309689752169286816.stgit@noble.brown Link: https://lkml.kernel.org/r/164859778120.29473.11725907882296224053.stgit@noble.brown Signed-off-by: NeilBrown <neilb@suse.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Tested-by: David Howells <dhowells@redhat.com> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-05-10 01:20:47 +00:00
#include "swap.h"
struct madvise_walk_private {
struct mmu_gather *tlb;
bool pageout;
};
/*
* Any behaviour which results in changes to the vma->vm_flags needs to
* take mmap_lock for writing. Others, which simply traverse vmas, need
* to only take it for reading.
*/
static int madvise_need_mmap_write(int behavior)
{
switch (behavior) {
case MADV_REMOVE:
case MADV_WILLNEED:
case MADV_DONTNEED:
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
case MADV_DONTNEED_LOCKED:
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
case MADV_COLD:
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
case MADV_PAGEOUT:
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
case MADV_FREE:
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
case MADV_POPULATE_READ:
case MADV_POPULATE_WRITE:
mm/madvise: introduce MADV_COLLAPSE sync hugepage collapse This idea was introduced by David Rientjes[1]. Introduce a new madvise mode, MADV_COLLAPSE, that allows users to request a synchronous collapse of memory at their own expense. The benefits of this approach are: * CPU is charged to the process that wants to spend the cycles for the THP * Avoid unpredictable timing of khugepaged collapse Semantics This call is independent of the system-wide THP sysfs settings, but will fail for memory marked VM_NOHUGEPAGE. If the ranges provided span multiple VMAs, the semantics of the collapse over each VMA is independent from the others. This implies a hugepage cannot cross a VMA boundary. If collapse of a given hugepage-aligned/sized region fails, the operation may continue to attempt collapsing the remainder of memory specified. The memory ranges provided must be page-aligned, but are not required to be hugepage-aligned. If the memory ranges are not hugepage-aligned, the start/end of the range will be clamped to the first/last hugepage-aligned address covered by said range. The memory ranges must span at least one hugepage-sized region. All non-resident pages covered by the range will first be swapped/faulted-in, before being internally copied onto a freshly allocated hugepage. Unmapped pages will have their data directly initialized to 0 in the new hugepage. However, for every eligible hugepage aligned/sized region to-be collapsed, at least one page must currently be backed by memory (a PMD covering the address range must already exist). Allocation for the new hugepage may enter direct reclaim and/or compaction, regardless of VMA flags. When the system has multiple NUMA nodes, the hugepage will be allocated from the node providing the most native pages. This operation operates on the current state of the specified process and makes no persistent changes or guarantees on how pages will be mapped, constructed, or faulted in the future Return Value If all hugepage-sized/aligned regions covered by the provided range were either successfully collapsed, or were already PMD-mapped THPs, this operation will be deemed successful. On success, process_madvise(2) returns the number of bytes advised, and madvise(2) returns 0. Else, -1 is returned and errno is set to indicate the error for the most-recently attempted hugepage collapse. Note that many failures might have occurred, since the operation may continue to collapse in the event a single hugepage-sized/aligned region fails. ENOMEM Memory allocation failed or VMA not found EBUSY Memcg charging failed EAGAIN Required resource temporarily unavailable. Try again might succeed. EINVAL Other error: No PMD found, subpage doesn't have Present bit set, "Special" page no backed by struct page, VMA incorrectly sized, address not page-aligned, ... Most notable here is ENOMEM and EBUSY (new to madvise) which are intended to provide the caller with actionable feedback so they may take an appropriate fallback measure. Use Cases An immediate user of this new functionality are malloc() implementations that manage memory in hugepage-sized chunks, but sometimes subrelease memory back to the system in native-sized chunks via MADV_DONTNEED; zapping the pmd. Later, when the memory is hot, the implementation could madvise(MADV_COLLAPSE) to re-back the memory by THPs to regain hugepage coverage and dTLB performance. TCMalloc is such an implementation that could benefit from this[2]. Only privately-mapped anon memory is supported for now, but additional support for file, shmem, and HugeTLB high-granularity mappings[2] is expected. File and tmpfs/shmem support would permit: * Backing executable text by THPs. Current support provided by CONFIG_READ_ONLY_THP_FOR_FS may take a long time on a large system which might impair services from serving at their full rated load after (re)starting. Tricks like mremap(2)'ing text onto anonymous memory to immediately realize iTLB performance prevents page sharing and demand paging, both of which increase steady state memory footprint. With MADV_COLLAPSE, we get the best of both worlds: Peak upfront performance and lower RAM footprints. * Backing guest memory by hugapages after the memory contents have been migrated in native-page-sized chunks to a new host, in a userfaultfd-based live-migration stack. [1] https://lore.kernel.org/linux-mm/d098c392-273a-36a4-1a29-59731cdf5d3d@google.com/ [2] https://github.com/google/tcmalloc/tree/master/tcmalloc [jrdr.linux@gmail.com: avoid possible memory leak in failure path] Link: https://lkml.kernel.org/r/20220713024109.62810-1-jrdr.linux@gmail.com [zokeefe@google.com add missing kfree() to madvise_collapse()] Link: https://lore.kernel.org/linux-mm/20220713024109.62810-1-jrdr.linux@gmail.com/ Link: https://lkml.kernel.org/r/20220713161851.1879439-1-zokeefe@google.com [zokeefe@google.com: delay computation of hpage boundaries until use]] Link: https://lkml.kernel.org/r/20220720140603.1958773-4-zokeefe@google.com Link: https://lkml.kernel.org/r/20220706235936.2197195-10-zokeefe@google.com Signed-off-by: Zach O'Keefe <zokeefe@google.com> Signed-off-by: "Souptick Joarder (HPE)" <jrdr.linux@gmail.com> Suggested-by: David Rientjes <rientjes@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Chris Kennelly <ckennelly@google.com> Cc: Chris Zankel <chris@zankel.net> Cc: David Hildenbrand <david@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: Hugh Dickins <hughd@google.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-07-06 23:59:27 +00:00
case MADV_COLLAPSE:
return 0;
default:
/* be safe, default to 1. list exceptions explicitly */
return 1;
}
}
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
#ifdef CONFIG_ANON_VMA_NAME
2022-03-05 04:28:51 +00:00
struct anon_vma_name *anon_vma_name_alloc(const char *name)
mm: add anonymous vma name refcounting While forking a process with high number (64K) of named anonymous vmas the overhead caused by strdup() is noticeable. Experiments with ARM64 Android device show up to 40% performance regression when forking a process with 64k unpopulated anonymous vmas using the max name lengths vs the same process with the same number of anonymous vmas having no name. Introduce anon_vma_name refcounted structure to avoid the overhead of copying vma names during fork() and when splitting named anonymous vmas. When a vma is duplicated, instead of copying the name we increment the refcount of this structure. Multiple vmas can point to the same anon_vma_name as long as they increment the refcount. The name member of anon_vma_name structure is assigned at structure allocation time and is never changed. If vma name changes then the refcount of the original structure is dropped, a new anon_vma_name structure is allocated to hold the new name and the vma pointer is updated to point to the new structure. With this approach the fork() performance regressions is reduced 3-4x times and with usecases using more reasonable number of VMAs (a few thousand) the regressions is not measurable. Link: https://lkml.kernel.org/r/20211019215511.3771969-3-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:06:03 +00:00
{
struct anon_vma_name *anon_name;
size_t count;
/* Add 1 for NUL terminator at the end of the anon_name->name */
count = strlen(name) + 1;
anon_name = kmalloc(struct_size(anon_name, name, count), GFP_KERNEL);
if (anon_name) {
kref_init(&anon_name->kref);
memcpy(anon_name->name, name, count);
}
return anon_name;
}
2022-03-05 04:28:51 +00:00
void anon_vma_name_free(struct kref *kref)
mm: add anonymous vma name refcounting While forking a process with high number (64K) of named anonymous vmas the overhead caused by strdup() is noticeable. Experiments with ARM64 Android device show up to 40% performance regression when forking a process with 64k unpopulated anonymous vmas using the max name lengths vs the same process with the same number of anonymous vmas having no name. Introduce anon_vma_name refcounted structure to avoid the overhead of copying vma names during fork() and when splitting named anonymous vmas. When a vma is duplicated, instead of copying the name we increment the refcount of this structure. Multiple vmas can point to the same anon_vma_name as long as they increment the refcount. The name member of anon_vma_name structure is assigned at structure allocation time and is never changed. If vma name changes then the refcount of the original structure is dropped, a new anon_vma_name structure is allocated to hold the new name and the vma pointer is updated to point to the new structure. With this approach the fork() performance regressions is reduced 3-4x times and with usecases using more reasonable number of VMAs (a few thousand) the regressions is not measurable. Link: https://lkml.kernel.org/r/20211019215511.3771969-3-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:06:03 +00:00
{
struct anon_vma_name *anon_name =
container_of(kref, struct anon_vma_name, kref);
kfree(anon_name);
}
2022-03-05 04:28:51 +00:00
struct anon_vma_name *anon_vma_name(struct vm_area_struct *vma)
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
{
mmap_assert_locked(vma->vm_mm);
2022-03-05 04:28:51 +00:00
return vma->anon_name;
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
}
/* mmap_lock should be write-locked */
2022-03-05 04:28:51 +00:00
static int replace_anon_vma_name(struct vm_area_struct *vma,
struct anon_vma_name *anon_name)
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
{
2022-03-05 04:28:51 +00:00
struct anon_vma_name *orig_name = anon_vma_name(vma);
mm: add anonymous vma name refcounting While forking a process with high number (64K) of named anonymous vmas the overhead caused by strdup() is noticeable. Experiments with ARM64 Android device show up to 40% performance regression when forking a process with 64k unpopulated anonymous vmas using the max name lengths vs the same process with the same number of anonymous vmas having no name. Introduce anon_vma_name refcounted structure to avoid the overhead of copying vma names during fork() and when splitting named anonymous vmas. When a vma is duplicated, instead of copying the name we increment the refcount of this structure. Multiple vmas can point to the same anon_vma_name as long as they increment the refcount. The name member of anon_vma_name structure is assigned at structure allocation time and is never changed. If vma name changes then the refcount of the original structure is dropped, a new anon_vma_name structure is allocated to hold the new name and the vma pointer is updated to point to the new structure. With this approach the fork() performance regressions is reduced 3-4x times and with usecases using more reasonable number of VMAs (a few thousand) the regressions is not measurable. Link: https://lkml.kernel.org/r/20211019215511.3771969-3-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:06:03 +00:00
2022-03-05 04:28:51 +00:00
if (!anon_name) {
vma->anon_name = NULL;
anon_vma_name_put(orig_name);
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
return 0;
}
2022-03-05 04:28:51 +00:00
if (anon_vma_name_eq(orig_name, anon_name))
return 0;
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
mm: prevent vm_area_struct::anon_name refcount saturation A deep process chain with many vmas could grow really high. With default sysctl_max_map_count (64k) and default pid_max (32k) the max number of vmas in the system is 2147450880 and the refcounter has headroom of 1073774592 before it reaches REFCOUNT_SATURATED (3221225472). Therefore it's unlikely that an anonymous name refcounter will overflow with these defaults. Currently the max for pid_max is PID_MAX_LIMIT (4194304) and for sysctl_max_map_count it's INT_MAX (2147483647). In this configuration anon_vma_name refcount overflow becomes theoretically possible (that still require heavy sharing of that anon_vma_name between processes). kref refcounting interface used in anon_vma_name structure will detect a counter overflow when it reaches REFCOUNT_SATURATED value but will only generate a warning and freeze the ref counter. This would lead to the refcounted object never being freed. A determined attacker could leak memory like that but it would be rather expensive and inefficient way to do so. To ensure anon_vma_name refcount does not overflow, stop anon_vma_name sharing when the refcount reaches REFCOUNT_MAX (2147483647), which still leaves INT_MAX/2 (1073741823) values before the counter reaches REFCOUNT_SATURATED. This should provide enough headroom for raising the refcounts temporarily. Link: https://lkml.kernel.org/r/20220223153613.835563-2-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: Michal Hocko <mhocko@suse.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Alexey Gladkov <legion@kernel.org> Cc: Chris Hyser <chris.hyser@oracle.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Peter Collingbourne <pcc@google.com> Cc: Sasha Levin <sashal@kernel.org> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xiaofeng Cao <caoxiaofeng@yulong.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-05 04:28:55 +00:00
vma->anon_name = anon_vma_name_reuse(anon_name);
2022-03-05 04:28:51 +00:00
anon_vma_name_put(orig_name);
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
return 0;
}
#else /* CONFIG_ANON_VMA_NAME */
2022-03-05 04:28:51 +00:00
static int replace_anon_vma_name(struct vm_area_struct *vma,
struct anon_vma_name *anon_name)
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
{
2022-03-05 04:28:51 +00:00
if (anon_name)
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
return -EINVAL;
return 0;
}
#endif /* CONFIG_ANON_VMA_NAME */
/*
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
* Update the vm_flags on region of a vma, splitting it or merging it as
* necessary. Must be called with mmap_lock held for writing;
mm: fix use-after-free when anon vma name is used after vma is freed When adjacent vmas are being merged it can result in the vma that was originally passed to madvise_update_vma being destroyed. In the current implementation, the name parameter passed to madvise_update_vma points directly to vma->anon_name and it is used after the call to vma_merge. In the cases when vma_merge merges the original vma and destroys it, this might result in UAF. For that the original vma would have to hold the anon_vma_name with the last reference. The following vma would need to contain a different anon_vma_name object with the same string. Such scenario is shown below: madvise_vma_behavior(vma) madvise_update_vma(vma, ..., anon_name == vma->anon_name) vma_merge(vma) __vma_adjust(vma) <-- merges vma with adjacent one vm_area_free(vma) <-- frees the original vma replace_vma_anon_name(anon_name) <-- UAF of vma->anon_name Fix this by raising the name refcount and stabilizing it. Link: https://lkml.kernel.org/r/20220224231834.1481408-3-surenb@google.com Link: https://lkml.kernel.org/r/20220223153613.835563-3-surenb@google.com Fixes: 9a10064f5625 ("mm: add a field to store names for private anonymous memory") Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reported-by: syzbot+aa7b3d4b35f9dc46a366@syzkaller.appspotmail.com Acked-by: Michal Hocko <mhocko@suse.com> Cc: Alexey Gladkov <legion@kernel.org> Cc: Chris Hyser <chris.hyser@oracle.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Collingbourne <pcc@google.com> Cc: Sasha Levin <sashal@kernel.org> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xiaofeng Cao <caoxiaofeng@yulong.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-05 04:28:58 +00:00
* Caller should ensure anon_name stability by raising its refcount even when
* anon_name belongs to a valid vma because this function might free that vma.
*/
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
static int madvise_update_vma(struct vm_area_struct *vma,
struct vm_area_struct **prev, unsigned long start,
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
unsigned long end, unsigned long new_flags,
2022-03-05 04:28:51 +00:00
struct anon_vma_name *anon_name)
{
struct mm_struct *mm = vma->vm_mm;
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
int error;
VMA_ITERATOR(vmi, mm, start);
2022-03-05 04:28:51 +00:00
if (new_flags == vma->vm_flags && anon_vma_name_eq(anon_vma_name(vma), anon_name)) {
*prev = vma;
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return 0;
}
mm: abstract the vma_merge()/split_vma() pattern for mprotect() et al. mprotect() and other functions which change VMA parameters over a range each employ a pattern of:- 1. Attempt to merge the range with adjacent VMAs. 2. If this fails, and the range spans a subset of the VMA, split it accordingly. This is open-coded and duplicated in each case. Also in each case most of the parameters passed to vma_merge() remain the same. Create a new function, vma_modify(), which abstracts this operation, accepting only those parameters which can be changed. To avoid the mess of invoking each function call with unnecessary parameters, create inline wrapper functions for each of the modify operations, parameterised only by what is required to perform the action. We can also significantly simplify the logic - by returning the VMA if we split (or merged VMA if we do not) we no longer need specific handling for merge/split cases in any of the call sites. Note that the userfaultfd_release() case works even though it does not split VMAs - since start is set to vma->vm_start and end is set to vma->vm_end, the split logic does not trigger. In addition, since we calculate pgoff to be equal to vma->vm_pgoff + (start - vma->vm_start) >> PAGE_SHIFT, and start - vma->vm_start will be 0 in this instance, this invocation will remain unchanged. We eliminate a VM_WARN_ON() in mprotect_fixup() as this simply asserts that vma_merge() correctly ensures that flags remain the same, something that is already checked in is_mergeable_vma() and elsewhere, and in any case is not specific to mprotect(). Link: https://lkml.kernel.org/r/0dfa9368f37199a423674bf0ee312e8ea0619044.1697043508.git.lstoakes@gmail.com Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Christian Brauner <brauner@kernel.org> Cc: Liam R. Howlett <Liam.Howlett@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-11 17:04:28 +00:00
vma = vma_modify_flags_name(&vmi, *prev, vma, start, end, new_flags,
anon_name);
if (IS_ERR(vma))
return PTR_ERR(vma);
*prev = vma;
/* vm_flags is protected by the mmap_lock held in write mode. */
vma_start_write(vma);
mm: replace vma->vm_flags direct modifications with modifier calls Replace direct modifications to vma->vm_flags with calls to modifier functions to be able to track flag changes and to keep vma locking correctness. [akpm@linux-foundation.org: fix drivers/misc/open-dice.c, per Hyeonggon Yoo] Link: https://lkml.kernel.org/r/20230126193752.297968-5-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Mike Rapoport (IBM) <rppt@kernel.org> Acked-by: Sebastian Reichel <sebastian.reichel@collabora.com> Reviewed-by: Liam R. Howlett <Liam.Howlett@Oracle.com> Reviewed-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Cc: Andy Lutomirski <luto@kernel.org> Cc: Arjun Roy <arjunroy@google.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: David Rientjes <rientjes@google.com> Cc: Eric Dumazet <edumazet@google.com> Cc: Greg Thelen <gthelen@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jann Horn <jannh@google.com> Cc: Joel Fernandes <joelaf@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kent Overstreet <kent.overstreet@linux.dev> Cc: Laurent Dufour <ldufour@linux.ibm.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Minchan Kim <minchan@google.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Peter Oskolkov <posk@google.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Punit Agrawal <punit.agrawal@bytedance.com> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Soheil Hassas Yeganeh <soheil@google.com> Cc: Song Liu <songliubraving@fb.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Will Deacon <will@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-01-26 19:37:49 +00:00
vm_flags_reset(vma, new_flags);
mm: anonymous shared memory naming Since commit 9a10064f5625 ("mm: add a field to store names for private anonymous memory"), name for private anonymous memory, but not shared anonymous, can be set. However, naming shared anonymous memory just as useful for tracking purposes. Extend the functionality to be able to set names for shared anon. There are two ways to create anonymous shared memory, using memfd or directly via mmap(): 1. fd = memfd_create(...) mem = mmap(..., MAP_SHARED, fd, ...) 2. mem = mmap(..., MAP_SHARED | MAP_ANONYMOUS, -1, ...) In both cases the anonymous shared memory is created the same way by mapping an unlinked file on tmpfs. The memfd way allows to give a name for anonymous shared memory, but not useful when parts of shared memory require to have distinct names. Example use case: The VMM maps VM memory as anonymous shared memory (not private because VMM is sandboxed and drivers are running in their own processes). However, the VM tells back to the VMM how parts of the memory are actually used by the guest, how each of the segments should be backed (i.e. 4K pages, 2M pages), and some other information about the segments. The naming allows us to monitor the effective memory footprint for each of these segments from the host without looking inside the guest. Sample output: /* Create shared anonymous segmenet */ anon_shmem = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); /* Name the segment: "MY-NAME" */ rv = prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, anon_shmem, SIZE, "MY-NAME"); cat /proc/<pid>/maps (and smaps): 7fc8e2b4c000-7fc8f2b4c000 rw-s 00000000 00:01 1024 [anon_shmem:MY-NAME] If the segment is not named, the output is: 7fc8e2b4c000-7fc8f2b4c000 rw-s 00000000 00:01 1024 /dev/zero (deleted) Link: https://lkml.kernel.org/r/20221115020602.804224-1-pasha.tatashin@soleen.com Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bagas Sanjaya <bagasdotme@gmail.com> Cc: Colin Cross <ccross@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Jonathan Corbet <corbet@lwn.net> Cc: "Kirill A . Shutemov" <kirill@shutemov.name> Cc: Liam Howlett <liam.howlett@oracle.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Rapoport <rppt@kernel.org> Cc: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Peter Xu <peterx@redhat.com> Cc: Sean Christopherson <seanjc@google.com> Cc: Vincent Whitchurch <vincent.whitchurch@axis.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: xu xin <cgel.zte@gmail.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-11-15 02:06:01 +00:00
if (!vma->vm_file || vma_is_anon_shmem(vma)) {
2022-03-05 04:28:51 +00:00
error = replace_anon_vma_name(vma, anon_name);
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
if (error)
return error;
}
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return 0;
}
#ifdef CONFIG_SWAP
static int swapin_walk_pmd_entry(pmd_t *pmd, unsigned long start,
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
unsigned long end, struct mm_walk *walk)
{
struct vm_area_struct *vma = walk->private;
struct swap_iocb *splug = NULL;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
pte_t *ptep = NULL;
spinlock_t *ptl;
unsigned long addr;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
for (addr = start; addr < end; addr += PAGE_SIZE) {
pte_t pte;
swp_entry_t entry;
struct folio *folio;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
if (!ptep++) {
ptep = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
if (!ptep)
break;
}
mm: ptep_get() conversion Convert all instances of direct pte_t* dereferencing to instead use ptep_get() helper. This means that by default, the accesses change from a C dereference to a READ_ONCE(). This is technically the correct thing to do since where pgtables are modified by HW (for access/dirty) they are volatile and therefore we should always ensure READ_ONCE() semantics. But more importantly, by always using the helper, it can be overridden by the architecture to fully encapsulate the contents of the pte. Arch code is deliberately not converted, as the arch code knows best. It is intended that arch code (arm64) will override the default with its own implementation that can (e.g.) hide certain bits from the core code, or determine young/dirty status by mixing in state from another source. Conversion was done using Coccinelle: ---- // $ make coccicheck \ // COCCI=ptepget.cocci \ // SPFLAGS="--include-headers" \ // MODE=patch virtual patch @ depends on patch @ pte_t *v; @@ - *v + ptep_get(v) ---- Then reviewed and hand-edited to avoid multiple unnecessary calls to ptep_get(), instead opting to store the result of a single call in a variable, where it is correct to do so. This aims to negate any cost of READ_ONCE() and will benefit arch-overrides that may be more complex. Included is a fix for an issue in an earlier version of this patch that was pointed out by kernel test robot. The issue arose because config MMU=n elides definition of the ptep helper functions, including ptep_get(). HUGETLB_PAGE=n configs still define a simple huge_ptep_clear_flush() for linking purposes, which dereferences the ptep. So when both configs are disabled, this caused a build error because ptep_get() is not defined. Fix by continuing to do a direct dereference when MMU=n. This is safe because for this config the arch code cannot be trying to virtualize the ptes because none of the ptep helpers are defined. Link: https://lkml.kernel.org/r/20230612151545.3317766-4-ryan.roberts@arm.com Reported-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/oe-kbuild-all/202305120142.yXsNEo6H-lkp@intel.com/ Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Potapenko <glider@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alex Williamson <alex.williamson@redhat.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Andrey Konovalov <andreyknvl@gmail.com> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christoph Hellwig <hch@infradead.org> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Dave Airlie <airlied@gmail.com> Cc: Dimitri Sivanich <dimitri.sivanich@hpe.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Ian Rogers <irogers@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Jérôme Glisse <jglisse@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: SeongJae Park <sj@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Uladzislau Rezki (Sony) <urezki@gmail.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-12 15:15:45 +00:00
pte = ptep_get(ptep);
if (!is_swap_pte(pte))
continue;
entry = pte_to_swp_entry(pte);
if (unlikely(non_swap_entry(entry)))
continue;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
pte_unmap_unlock(ptep, ptl);
ptep = NULL;
folio = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE,
swap: remove remnants of polling from read_swap_cache_async Patch series "Per-VMA lock support for swap and userfaults", v7. When per-VMA locks were introduced in [1] several types of page faults would still fall back to mmap_lock to keep the patchset simple. Among them are swap and userfault pages. The main reason for skipping those cases was the fact that mmap_lock could be dropped while handling these faults and that required additional logic to be implemented. Implement the mechanism to allow per-VMA locks to be dropped for these cases. First, change handle_mm_fault to drop per-VMA locks when returning VM_FAULT_RETRY or VM_FAULT_COMPLETED to be consistent with the way mmap_lock is handled. Then change folio_lock_or_retry to accept vm_fault and return vm_fault_t which simplifies later patches. Finally allow swap and uffd page faults to be handled under per-VMA locks by dropping per-VMA and retrying, the same way it's done under mmap_lock. Naturally, once VMA lock is dropped that VMA should be assumed unstable and can't be used. This patch (of 6): Commit [1] introduced IO polling support duding swapin to reduce swap read latency for block devices that can be polled. However later commit [2] removed polling support. Therefore it seems safe to remove do_poll parameter in read_swap_cache_async and always call swap_readpage with synchronous=false waiting for IO completion in folio_lock_or_retry. [1] commit 23955622ff8d ("swap: add block io poll in swapin path") [2] commit 9650b453a3d4 ("block: ignore RWF_HIPRI hint for sync dio") Link: https://lkml.kernel.org/r/20230630211957.1341547-1-surenb@google.com Link: https://lkml.kernel.org/r/20230630211957.1341547-2-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: Alistair Popple <apopple@nvidia.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christian Brauner <brauner@kernel.org> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Hillf Danton <hdanton@sina.com> Cc: Hugh Dickins <hughd@google.com> Cc: Jan Kara <jack@suse.cz> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Laurent Dufour <ldufour@linux.ibm.com> Cc: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Michel Lespinasse <michel@lespinasse.org> Cc: Minchan Kim <minchan@google.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Punit Agrawal <punit.agrawal@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-30 21:19:52 +00:00
vma, addr, &splug);
if (folio)
folio_put(folio);
}
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
if (ptep)
pte_unmap_unlock(ptep, ptl);
swap_read_unplug(splug);
cond_resched();
return 0;
}
static const struct mm_walk_ops swapin_walk_ops = {
.pmd_entry = swapin_walk_pmd_entry,
mm: enable page walking API to lock vmas during the walk walk_page_range() and friends often operate under write-locked mmap_lock. With introduction of vma locks, the vmas have to be locked as well during such walks to prevent concurrent page faults in these areas. Add an additional member to mm_walk_ops to indicate locking requirements for the walk. The change ensures that page walks which prevent concurrent page faults by write-locking mmap_lock, operate correctly after introduction of per-vma locks. With per-vma locks page faults can be handled under vma lock without taking mmap_lock at all, so write locking mmap_lock would not stop them. The change ensures vmas are properly locked during such walks. A sample issue this solves is do_mbind() performing queue_pages_range() to queue pages for migration. Without this change a concurrent page can be faulted into the area and be left out of migration. Link: https://lkml.kernel.org/r/20230804152724.3090321-2-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org> Suggested-by: Jann Horn <jannh@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Laurent Dufour <ldufour@linux.ibm.com> Cc: Liam Howlett <liam.howlett@oracle.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Michel Lespinasse <michel@lespinasse.org> Cc: Peter Xu <peterx@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-08-04 15:27:19 +00:00
.walk_lock = PGWALK_RDLOCK,
};
mm/madvise: clean up force_shm_swapin_readahead() Some nearby MADV_WILLNEED cleanup unrelated to pte_offset_map_lock(). shmem_swapin_range() is a better name than force_shm_swapin_readahead(). Fix unimportant off-by-one on end_index. Call the swp_entry_t "entry" rather than "swap": either is okay, but entry is the name used elsewhere in mm/madvise.c. Do not assume GFP_HIGHUSER_MOVABLE: that's right for anon swap, but shmem should take gfp from mapping. Pass the actual vma and address to read_swap_cache_async(), in case a NUMA mempolicy applies. lru_add_drain() at outer level, like madvise_willneed()'s other branch. Link: https://lkml.kernel.org/r/67e18875-ffb3-ec27-346-f350e07bed87@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:35:14 +00:00
static void shmem_swapin_range(struct vm_area_struct *vma,
unsigned long start, unsigned long end,
struct address_space *mapping)
{
XA_STATE(xas, &mapping->i_pages, linear_page_index(vma, start));
mm/madvise: clean up force_shm_swapin_readahead() Some nearby MADV_WILLNEED cleanup unrelated to pte_offset_map_lock(). shmem_swapin_range() is a better name than force_shm_swapin_readahead(). Fix unimportant off-by-one on end_index. Call the swp_entry_t "entry" rather than "swap": either is okay, but entry is the name used elsewhere in mm/madvise.c. Do not assume GFP_HIGHUSER_MOVABLE: that's right for anon swap, but shmem should take gfp from mapping. Pass the actual vma and address to read_swap_cache_async(), in case a NUMA mempolicy applies. lru_add_drain() at outer level, like madvise_willneed()'s other branch. Link: https://lkml.kernel.org/r/67e18875-ffb3-ec27-346-f350e07bed87@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:35:14 +00:00
pgoff_t end_index = linear_page_index(vma, end) - 1;
struct folio *folio;
struct swap_iocb *splug = NULL;
rcu_read_lock();
xas_for_each(&xas, folio, end_index) {
mm/madvise: clean up force_shm_swapin_readahead() Some nearby MADV_WILLNEED cleanup unrelated to pte_offset_map_lock(). shmem_swapin_range() is a better name than force_shm_swapin_readahead(). Fix unimportant off-by-one on end_index. Call the swp_entry_t "entry" rather than "swap": either is okay, but entry is the name used elsewhere in mm/madvise.c. Do not assume GFP_HIGHUSER_MOVABLE: that's right for anon swap, but shmem should take gfp from mapping. Pass the actual vma and address to read_swap_cache_async(), in case a NUMA mempolicy applies. lru_add_drain() at outer level, like madvise_willneed()'s other branch. Link: https://lkml.kernel.org/r/67e18875-ffb3-ec27-346-f350e07bed87@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:35:14 +00:00
unsigned long addr;
swp_entry_t entry;
if (!xa_is_value(folio))
continue;
entry = radix_to_swp_entry(folio);
/* There might be swapin error entries in shmem mapping. */
mm/madvise: clean up force_shm_swapin_readahead() Some nearby MADV_WILLNEED cleanup unrelated to pte_offset_map_lock(). shmem_swapin_range() is a better name than force_shm_swapin_readahead(). Fix unimportant off-by-one on end_index. Call the swp_entry_t "entry" rather than "swap": either is okay, but entry is the name used elsewhere in mm/madvise.c. Do not assume GFP_HIGHUSER_MOVABLE: that's right for anon swap, but shmem should take gfp from mapping. Pass the actual vma and address to read_swap_cache_async(), in case a NUMA mempolicy applies. lru_add_drain() at outer level, like madvise_willneed()'s other branch. Link: https://lkml.kernel.org/r/67e18875-ffb3-ec27-346-f350e07bed87@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:35:14 +00:00
if (non_swap_entry(entry))
continue;
mm/madvise: clean up force_shm_swapin_readahead() Some nearby MADV_WILLNEED cleanup unrelated to pte_offset_map_lock(). shmem_swapin_range() is a better name than force_shm_swapin_readahead(). Fix unimportant off-by-one on end_index. Call the swp_entry_t "entry" rather than "swap": either is okay, but entry is the name used elsewhere in mm/madvise.c. Do not assume GFP_HIGHUSER_MOVABLE: that's right for anon swap, but shmem should take gfp from mapping. Pass the actual vma and address to read_swap_cache_async(), in case a NUMA mempolicy applies. lru_add_drain() at outer level, like madvise_willneed()'s other branch. Link: https://lkml.kernel.org/r/67e18875-ffb3-ec27-346-f350e07bed87@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:35:14 +00:00
addr = vma->vm_start +
((xas.xa_index - vma->vm_pgoff) << PAGE_SHIFT);
xas_pause(&xas);
rcu_read_unlock();
folio = read_swap_cache_async(entry, mapping_gfp_mask(mapping),
swap: remove remnants of polling from read_swap_cache_async Patch series "Per-VMA lock support for swap and userfaults", v7. When per-VMA locks were introduced in [1] several types of page faults would still fall back to mmap_lock to keep the patchset simple. Among them are swap and userfault pages. The main reason for skipping those cases was the fact that mmap_lock could be dropped while handling these faults and that required additional logic to be implemented. Implement the mechanism to allow per-VMA locks to be dropped for these cases. First, change handle_mm_fault to drop per-VMA locks when returning VM_FAULT_RETRY or VM_FAULT_COMPLETED to be consistent with the way mmap_lock is handled. Then change folio_lock_or_retry to accept vm_fault and return vm_fault_t which simplifies later patches. Finally allow swap and uffd page faults to be handled under per-VMA locks by dropping per-VMA and retrying, the same way it's done under mmap_lock. Naturally, once VMA lock is dropped that VMA should be assumed unstable and can't be used. This patch (of 6): Commit [1] introduced IO polling support duding swapin to reduce swap read latency for block devices that can be polled. However later commit [2] removed polling support. Therefore it seems safe to remove do_poll parameter in read_swap_cache_async and always call swap_readpage with synchronous=false waiting for IO completion in folio_lock_or_retry. [1] commit 23955622ff8d ("swap: add block io poll in swapin path") [2] commit 9650b453a3d4 ("block: ignore RWF_HIPRI hint for sync dio") Link: https://lkml.kernel.org/r/20230630211957.1341547-1-surenb@google.com Link: https://lkml.kernel.org/r/20230630211957.1341547-2-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: Alistair Popple <apopple@nvidia.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christian Brauner <brauner@kernel.org> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Hillf Danton <hdanton@sina.com> Cc: Hugh Dickins <hughd@google.com> Cc: Jan Kara <jack@suse.cz> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Laurent Dufour <ldufour@linux.ibm.com> Cc: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Michel Lespinasse <michel@lespinasse.org> Cc: Minchan Kim <minchan@google.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Punit Agrawal <punit.agrawal@bytedance.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-30 21:19:52 +00:00
vma, addr, &splug);
if (folio)
folio_put(folio);
rcu_read_lock();
}
rcu_read_unlock();
swap_read_unplug(splug);
}
#endif /* CONFIG_SWAP */
/*
* Schedule all required I/O operations. Do not wait for completion.
*/
static long madvise_willneed(struct vm_area_struct *vma,
struct vm_area_struct **prev,
unsigned long start, unsigned long end)
{
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
struct mm_struct *mm = vma->vm_mm;
struct file *file = vma->vm_file;
loff_t offset;
mm/madvise.c: fix madvise() infinite loop under special circumstances MADVISE_WILLNEED has always been a noop for DAX (formerly XIP) mappings. Unfortunately madvise_willneed() doesn't communicate this information properly to the generic madvise syscall implementation. The calling convention is quite subtle there. madvise_vma() is supposed to either return an error or update &prev otherwise the main loop will never advance to the next vma and it will keep looping for ever without a way to get out of the kernel. It seems this has been broken since introduction. Nobody has noticed because nobody seems to be using MADVISE_WILLNEED on these DAX mappings. [mhocko@suse.com: rewrite changelog] Link: http://lkml.kernel.org/r/20171127115318.911-1-guoxuenan@huawei.com Fixes: fe77ba6f4f97 ("[PATCH] xip: madvice/fadvice: execute in place") Signed-off-by: chenjie <chenjie6@huawei.com> Signed-off-by: guoxuenan <guoxuenan@huawei.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: zhangyi (F) <yi.zhang@huawei.com> Cc: Miao Xie <miaoxie@huawei.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Cc: Shaohua Li <shli@fb.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: David Rientjes <rientjes@google.com> Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com> Cc: Rik van Riel <riel@redhat.com> Cc: Carsten Otte <cotte@de.ibm.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-11-30 00:10:54 +00:00
*prev = vma;
#ifdef CONFIG_SWAP
if (!file) {
walk_page_range(vma->vm_mm, start, end, &swapin_walk_ops, vma);
lru_add_drain(); /* Push any new pages onto the LRU now */
return 0;
}
if (shmem_mapping(file->f_mapping)) {
mm/madvise: clean up force_shm_swapin_readahead() Some nearby MADV_WILLNEED cleanup unrelated to pte_offset_map_lock(). shmem_swapin_range() is a better name than force_shm_swapin_readahead(). Fix unimportant off-by-one on end_index. Call the swp_entry_t "entry" rather than "swap": either is okay, but entry is the name used elsewhere in mm/madvise.c. Do not assume GFP_HIGHUSER_MOVABLE: that's right for anon swap, but shmem should take gfp from mapping. Pass the actual vma and address to read_swap_cache_async(), in case a NUMA mempolicy applies. lru_add_drain() at outer level, like madvise_willneed()'s other branch. Link: https://lkml.kernel.org/r/67e18875-ffb3-ec27-346-f350e07bed87@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:35:14 +00:00
shmem_swapin_range(vma, start, end, file->f_mapping);
lru_add_drain(); /* Push any new pages onto the LRU now */
return 0;
}
#else
if (!file)
return -EBADF;
#endif
if (IS_DAX(file_inode(file))) {
/* no bad return value, but ignore advice */
return 0;
}
/*
* Filesystem's fadvise may need to take various locks. We need to
* explicitly grab a reference because the vma (and hence the
* vma's reference to the file) can go away as soon as we drop
* mmap_lock.
*/
*prev = NULL; /* tell sys_madvise we drop mmap_lock */
get_file(file);
offset = (loff_t)(start - vma->vm_start)
+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_unlock(mm);
vfs_fadvise(file, offset, end - start, POSIX_FADV_WILLNEED);
fput(file);
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_lock(mm);
return 0;
}
static inline bool can_do_file_pageout(struct vm_area_struct *vma)
{
if (!vma->vm_file)
return false;
/*
* paging out pagecache only for non-anonymous mappings that correspond
* to the files the calling process could (if tried) open for writing;
* otherwise we'd be including shared non-exclusive mappings, which
* opens a side channel.
*/
return inode_owner_or_capable(&nop_mnt_idmap,
file_inode(vma->vm_file)) ||
file_permission(vma->vm_file, MAY_WRITE) == 0;
}
static inline int madvise_folio_pte_batch(unsigned long addr, unsigned long end,
struct folio *folio, pte_t *ptep,
pte_t pte, bool *any_young,
bool *any_dirty)
{
const fpb_t fpb_flags = FPB_IGNORE_DIRTY | FPB_IGNORE_SOFT_DIRTY;
int max_nr = (end - addr) / PAGE_SIZE;
return folio_pte_batch(folio, addr, ptep, pte, max_nr, fpb_flags, NULL,
any_young, any_dirty);
}
static int madvise_cold_or_pageout_pte_range(pmd_t *pmd,
unsigned long addr, unsigned long end,
struct mm_walk *walk)
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
{
struct madvise_walk_private *private = walk->private;
struct mmu_gather *tlb = private->tlb;
bool pageout = private->pageout;
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
struct mm_struct *mm = tlb->mm;
struct vm_area_struct *vma = walk->vma;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
pte_t *start_pte, *pte, ptent;
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
spinlock_t *ptl;
struct folio *folio = NULL;
LIST_HEAD(folio_list);
bool pageout_anon_only_filter;
mm/madvise: add cond_resched() in madvise_cold_or_pageout_pte_range() I conducted real-time testing and observed that madvise_cold_or_pageout_pte_range() causes significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. I tested on the LicheePi 4A board using Cylictest for latency testing and Ftrace for latency tracing. The board uses TH1520 processor and has a memory size of 8GB. The kernel version is 6.5.0 with the PREEMPT_RT patch applied. The script I tested is as follows: echo wakeup_rt > /sys/kernel/tracing/current_tracer echo 1 > /sys/kernel/tracing/tracing_on echo 0 > /sys/kernel/tracing/tracing_max_latency stress-ng --vm 8 --vm-bytes 2G & cyclictest --mlockall --smp --priority=99 --distance=0 --duration=30m echo 0 > /sys/kernel/tracing/tracing_on cat /sys/kernel/tracing/trace The tracing results before modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00003-g999d221864bf # -------------------------------------------------------------------- # latency: 2552 us, #6/6, CPU#3 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-196 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-206 3dn.h512 2us : 206:120:R + [003] 196: 0:R cyclictest stress-n-206 3dn.h512 7us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-206 3dn.h512 9us#: 0 stress-n-206 3d...3.. 2544us : __schedule stress-n-206 3d...3.. 2545us : 206:120:R ==> [003] 196: 0:R cyclictest stress-n-206 3d...3.. 2551us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => rt_spin_unlock => madvise_cold_or_pageout_pte_range => walk_pgd_range => __walk_page_range => walk_page_range => madvise_pageout => madvise_vma_behavior => do_madvise => sys_madvise => do_trap_ecall_u => ret_from_exception The tracing results after modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00004-gca3876fc69a6-dirty # -------------------------------------------------------------------- # latency: 1689 us, #6/6, CPU#0 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-217 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-232 0dn.h413 1us+: 232:120:R + [000] 217: 0:R cyclictest stress-n-232 0dn.h413 12us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-232 0dn.h413 19us#: 0 stress-n-232 0d...3.. 1671us : __schedule stress-n-232 0d...3.. 1676us+: 232:120:R ==> [000] 217: 0:R cyclictest stress-n-232 0d...3.. 1687us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => free_unref_page_list => release_pages => free_pages_and_swap_cache => tlb_batch_pages_flush => tlb_flush_mmu => unmap_page_range => unmap_vmas => unmap_region => do_vmi_align_munmap.constprop.0 => do_vmi_munmap => __vm_munmap => sys_munmap => do_trap_ecall_u => ret_from_exception After the modification, the cause of maximum latency is no longer madvise_cold_or_pageout_pte_range(), so this modification can reduce the latency caused by madvise_cold_or_pageout_pte_range(). Currently the madvise_cold_or_pageout_pte_range() function exhibits significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. When the batch_count reaches SWAP_CLUSTER_MAX, we reschedule the task to ensure fairness and avoid long lock holding times. Link: https://lkml.kernel.org/r/85363861af65fac66c7a98c251906afc0d9c8098.1695291046.git.wangjiexun@tinylab.org Signed-off-by: Jiexun Wang <wangjiexun@tinylab.org> Cc: Zhangjin Wu <falcon@tinylab.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-21 12:27:51 +00:00
unsigned int batch_count = 0;
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
int nr;
if (fatal_signal_pending(current))
return -EINTR;
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
pageout_anon_only_filter = pageout && !vma_is_anonymous(vma) &&
!can_do_file_pageout(vma);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
if (pmd_trans_huge(*pmd)) {
pmd_t orig_pmd;
unsigned long next = pmd_addr_end(addr, end);
tlb_change_page_size(tlb, HPAGE_PMD_SIZE);
ptl = pmd_trans_huge_lock(pmd, vma);
if (!ptl)
return 0;
orig_pmd = *pmd;
if (is_huge_zero_pmd(orig_pmd))
goto huge_unlock;
if (unlikely(!pmd_present(orig_pmd))) {
VM_BUG_ON(thp_migration_supported() &&
!is_pmd_migration_entry(orig_pmd));
goto huge_unlock;
}
folio = pmd_folio(orig_pmd);
mm: do not allow MADV_PAGEOUT for CoW pages Jann has brought up a very interesting point [1]. While shared pages are excluded from MADV_PAGEOUT normally, CoW pages can be easily reclaimed that way. This can lead to all sorts of hard to debug problems. E.g. performance problems outlined by Daniel [2]. There are runtime environments where there is a substantial memory shared among security domains via CoW memory and a easy to reclaim way of that memory, which MADV_{COLD,PAGEOUT} offers, can lead to either performance degradation in for the parent process which might be more privileged or even open side channel attacks. The feasibility of the latter is not really clear to me TBH but there is no real reason for exposure at this stage. It seems there is no real use case to depend on reclaiming CoW memory via madvise at this stage so it is much easier to simply disallow it and this is what this patch does. Put it simply MADV_{PAGEOUT,COLD} can operate only on the exclusively owned memory which is a straightforward semantic. [1] http://lkml.kernel.org/r/CAG48ez0G3JkMq61gUmyQAaCq=_TwHbi1XKzWRooxZkv08PQKuw@mail.gmail.com [2] http://lkml.kernel.org/r/CAKOZueua_v8jHCpmEtTB6f3i9e2YnmX4mqdYVWhV4E=Z-n+zRQ@mail.gmail.com Fixes: 9c276cc65a58 ("mm: introduce MADV_COLD") Reported-by: Jann Horn <jannh@google.com> Signed-off-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: "Joel Fernandes (Google)" <joel@joelfernandes.org> Cc: <stable@vger.kernel.org> Link: http://lkml.kernel.org/r/20200312082248.GS23944@dhcp22.suse.cz Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-03-22 01:22:26 +00:00
/* Do not interfere with other mappings of this folio */
mm: convert folio_estimated_sharers() to folio_likely_mapped_shared() Callers of folio_estimated_sharers() only care about "mapped shared vs. mapped exclusively", not the exact estimate of sharers. Let's consolidate and unify the condition users are checking. While at it clarify the semantics and extend the discussion on the fuzziness. Use the "likely mapped shared" terminology to better express what the (adjusted) function actually checks. Whether a partially-mappable folio is more likely to not be partially mapped than partially mapped is debatable. In the future, we might be able to improve our estimate for partially-mappable folios, though. Note that we will now consistently detect "mapped shared" only if the first subpage is actually mapped multiple times. When the first subpage is not mapped, we will consistently detect it as "mapped exclusively". This change should currently only affect the usage in madvise_free_pte_range() and queue_folios_pte_range() for large folios: if the first page was already unmapped, we would have skipped the folio. [david@redhat.com: folio_likely_mapped_shared() kerneldoc fixup] Link: https://lkml.kernel.org/r/dd0ad9f2-2d7a-45f3-9ba3-979488c7dd27@redhat.com Link: https://lkml.kernel.org/r/20240227201548.857831-1-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Acked-by: Barry Song <v-songbaohua@oppo.com> Reviewed-by: Vishal Moola (Oracle) <vishal.moola@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Zi Yan <ziy@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-02-27 20:15:48 +00:00
if (folio_likely_mapped_shared(folio))
mm: do not allow MADV_PAGEOUT for CoW pages Jann has brought up a very interesting point [1]. While shared pages are excluded from MADV_PAGEOUT normally, CoW pages can be easily reclaimed that way. This can lead to all sorts of hard to debug problems. E.g. performance problems outlined by Daniel [2]. There are runtime environments where there is a substantial memory shared among security domains via CoW memory and a easy to reclaim way of that memory, which MADV_{COLD,PAGEOUT} offers, can lead to either performance degradation in for the parent process which might be more privileged or even open side channel attacks. The feasibility of the latter is not really clear to me TBH but there is no real reason for exposure at this stage. It seems there is no real use case to depend on reclaiming CoW memory via madvise at this stage so it is much easier to simply disallow it and this is what this patch does. Put it simply MADV_{PAGEOUT,COLD} can operate only on the exclusively owned memory which is a straightforward semantic. [1] http://lkml.kernel.org/r/CAG48ez0G3JkMq61gUmyQAaCq=_TwHbi1XKzWRooxZkv08PQKuw@mail.gmail.com [2] http://lkml.kernel.org/r/CAKOZueua_v8jHCpmEtTB6f3i9e2YnmX4mqdYVWhV4E=Z-n+zRQ@mail.gmail.com Fixes: 9c276cc65a58 ("mm: introduce MADV_COLD") Reported-by: Jann Horn <jannh@google.com> Signed-off-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: "Joel Fernandes (Google)" <joel@joelfernandes.org> Cc: <stable@vger.kernel.org> Link: http://lkml.kernel.org/r/20200312082248.GS23944@dhcp22.suse.cz Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-03-22 01:22:26 +00:00
goto huge_unlock;
if (pageout_anon_only_filter && !folio_test_anon(folio))
goto huge_unlock;
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
if (next - addr != HPAGE_PMD_SIZE) {
int err;
folio_get(folio);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
spin_unlock(ptl);
folio_lock(folio);
err = split_folio(folio);
folio_unlock(folio);
folio_put(folio);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
if (!err)
goto regular_folio;
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
return 0;
}
mm: madvise: pageout: ignore references rather than clearing young While doing MADV_PAGEOUT, the current code will clear PTE young so that vmscan won't read young flags to allow the reclamation of madvised folios to go ahead. It seems we can do it by directly ignoring references, thus we can remove tlb flush in madvise and rmap overhead in vmscan. Regarding the side effect, in the original code, if a parallel thread runs side by side to access the madvised memory with the thread doing madvise, folios will get a chance to be re-activated by vmscan (though the time gap is actually quite small since checking PTEs is done immediately after clearing PTEs young). But with this patch, they will still be reclaimed. But this behaviour doing PAGEOUT and doing access at the same time is quite silly like DoS. So probably, we don't need to care. Or ignoring the new access during the quite small time gap is even better. For DAMON's DAMOS_PAGEOUT based on physical address region, we still keep its behaviour as is since a physical address might be mapped by multiple processes. MADV_PAGEOUT based on virtual address is actually much more aggressive on reclamation. To untouch paddr's DAMOS_PAGEOUT, we simply pass ignore_references as false in reclaim_pages(). A microbench as below has shown 6% decrement on the latency of MADV_PAGEOUT, #define PGSIZE 4096 main() { int i; #define SIZE 512*1024*1024 volatile long *p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); for (i = 0; i < SIZE/sizeof(long); i += PGSIZE / sizeof(long)) p[i] = 0x11; madvise(p, SIZE, MADV_PAGEOUT); } w/o patch w/ patch root@10:~# time ./a.out root@10:~# time ./a.out real 0m49.634s real 0m46.334s user 0m0.637s user 0m0.648s sys 0m47.434s sys 0m44.265s Link: https://lkml.kernel.org/r/20240226005739.24350-1-21cnbao@gmail.com Signed-off-by: Barry Song <v-songbaohua@oppo.com> Acked-by: Minchan Kim <minchan@kernel.org> Cc: SeongJae Park <sj@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-02-26 00:57:39 +00:00
if (!pageout && pmd_young(orig_pmd)) {
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
pmdp_invalidate(vma, addr, pmd);
orig_pmd = pmd_mkold(orig_pmd);
set_pmd_at(mm, addr, pmd, orig_pmd);
tlb_remove_pmd_tlb_entry(tlb, pmd, addr);
}
folio_clear_referenced(folio);
folio_test_clear_young(folio);
mm: madvise: fix uneven accounting of psi A folio turns into a Workingset during: 1) shrink_active_list() placing the folio from active to inactive list. 2) When a workingset transition is happening during the folio refault. And when Workingset is set on a folio, PSI for memory can be accounted during a) That folio is being reclaimed and b) Refault of that folio, for usual reclaims. This accounting of PSI for memory is not consistent for reclaim + refault operation between usual reclaim and madvise(COLD/PAGEOUT) which deactivate or proactively reclaim a folio: a) A folio started at inactive and moved to active as part of accesses. Workingset is absent on the folio thus refault of it when reclaimed through MADV_PAGEOUT operation doesn't account for PSI. b) When the same folio transition from inactive->active and then to inactive through shrink_active_list(). Workingset is set on the folio thus refault of it when reclaimed through MADV_PAGEOUT operation accounts for PSI. c) When the same folio is part of active list directly as a result of folio refault and this was a workingset folio prior to eviction. Workingset is set on the folio thus the refault of it when reclaimed through MADV_PAGEOUT/MADV_COLD operation accounts for PSI. d) MADV_COLD transfers the folio from active list to inactive list. Such folios may not have the Workingset thus refault operation on such folio doesn't account for PSI. As said above, refault operation caused because of MADV_PAGEOUT on a folio is accounts for memory PSI in b) and c) but not in a). Refault caused by the reclaim of a folio on which MADV_COLD is performed accounts memory PSI in c) but not in d). These behaviours are inconsistent w.r.t usual reclaim + refault operation. Make this PSI accounting always consistent by turning a folio into a workingset one whenever it is leaving the active list. Also, accounting of PSI on a folio whenever it leaves the active list as part of the MADV_COLD/PAGEOUT operation helps the users whether they are operating on proper folios[1]. [1] https://lore.kernel.org/all/20230605180013.GD221380@cmpxchg.org/ Link: https://lkml.kernel.org/r/1688393201-11135-1-git-send-email-quic_charante@quicinc.com Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com> Suggested-by: Suren Baghdasaryan <surenb@google.com> Reported-by: Sai Manobhiram Manapragada <quic_smanapra@quicinc.com> Reported-by: Pavan Kondeti <quic_pkondeti@quicinc.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Pavankumar Kondeti <quic_pkondeti@quicinc.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-07-03 14:06:41 +00:00
if (folio_test_active(folio))
folio_set_workingset(folio);
if (pageout) {
mm: change to return bool for folio_isolate_lru() Patch series "Change the return value for page isolation functions", v3. Now the page isolation functions did not return a boolean to indicate success or not, instead it will return a negative error when failed to isolate a page. So below code used in most places seem a boolean success/failure thing, which can confuse people whether the isolation is successful. if (folio_isolate_lru(folio)) continue; Moreover the page isolation functions only return 0 or -EBUSY, and most users did not care about the negative error except for few users, thus we can convert all page isolation functions to return a boolean value, which can remove the confusion to make code more clear. No functional changes intended in this patch series. This patch (of 4): Now the folio_isolate_lru() did not return a boolean value to indicate isolation success or not, however below code checking the return value can make people think that it was a boolean success/failure thing, which makes people easy to make mistakes (see the fix patch[1]). if (folio_isolate_lru(folio)) continue; Thus it's better to check the negative error value expilictly returned by folio_isolate_lru(), which makes code more clear per Linus's suggestion[2]. Moreover Matthew suggested we can convert the isolation functions to return a boolean[3], since most users did not care about the negative error value, and can also remove the confusing of checking return value. So this patch converts the folio_isolate_lru() to return a boolean value, which means return 'true' to indicate the folio isolation is successful, and 'false' means a failure to isolation. Meanwhile changing all users' logic of checking the isolation state. No functional changes intended. [1] https://lore.kernel.org/all/20230131063206.28820-1-Kuan-Ying.Lee@mediatek.com/T/#u [2] https://lore.kernel.org/all/CAHk-=wiBrY+O-4=2mrbVyxR+hOqfdJ=Do6xoucfJ9_5az01L4Q@mail.gmail.com/ [3] https://lore.kernel.org/all/Y+sTFqwMNAjDvxw3@casper.infradead.org/ Link: https://lkml.kernel.org/r/cover.1676424378.git.baolin.wang@linux.alibaba.com Link: https://lkml.kernel.org/r/8a4e3679ed4196168efadf7ea36c038f2f7d5aa9.1676424378.git.baolin.wang@linux.alibaba.com Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by: SeongJae Park <sj@kernel.org> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <muchun.song@linux.dev> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Shakeel Butt <shakeelb@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-02-15 10:39:34 +00:00
if (folio_isolate_lru(folio)) {
if (folio_test_unevictable(folio))
folio_putback_lru(folio);
mm: fix trying to reclaim unevictable lru page when calling madvise_pageout Recently, I hit the following issue when running upstream. kernel BUG at mm/vmscan.c:1521! invalid opcode: 0000 [#1] SMP KASAN PTI CPU: 0 PID: 23385 Comm: syz-executor.6 Not tainted 5.4.0-rc4+ #1 RIP: 0010:shrink_page_list+0x12b6/0x3530 mm/vmscan.c:1521 Call Trace: reclaim_pages+0x499/0x800 mm/vmscan.c:2188 madvise_cold_or_pageout_pte_range+0x58a/0x710 mm/madvise.c:453 walk_pmd_range mm/pagewalk.c:53 [inline] walk_pud_range mm/pagewalk.c:112 [inline] walk_p4d_range mm/pagewalk.c:139 [inline] walk_pgd_range mm/pagewalk.c:166 [inline] __walk_page_range+0x45a/0xc20 mm/pagewalk.c:261 walk_page_range+0x179/0x310 mm/pagewalk.c:349 madvise_pageout_page_range mm/madvise.c:506 [inline] madvise_pageout+0x1f0/0x330 mm/madvise.c:542 madvise_vma mm/madvise.c:931 [inline] __do_sys_madvise+0x7d2/0x1600 mm/madvise.c:1113 do_syscall_64+0x9f/0x4c0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe madvise_pageout() accesses the specified range of the vma and isolates them, then runs shrink_page_list() to reclaim its memory. But it also isolates the unevictable pages to reclaim. Hence, we can catch the cases in shrink_page_list(). The root cause is that we scan the page tables instead of specific LRU list. and so we need to filter out the unevictable lru pages from our end. Link: http://lkml.kernel.org/r/1572616245-18946-1-git-send-email-zhongjiang@huawei.com Fixes: 1a4e58cce84e ("mm: introduce MADV_PAGEOUT") Signed-off-by: zhong jiang <zhongjiang@huawei.com> Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-11-16 01:34:36 +00:00
else
list_add(&folio->lru, &folio_list);
mm: fix trying to reclaim unevictable lru page when calling madvise_pageout Recently, I hit the following issue when running upstream. kernel BUG at mm/vmscan.c:1521! invalid opcode: 0000 [#1] SMP KASAN PTI CPU: 0 PID: 23385 Comm: syz-executor.6 Not tainted 5.4.0-rc4+ #1 RIP: 0010:shrink_page_list+0x12b6/0x3530 mm/vmscan.c:1521 Call Trace: reclaim_pages+0x499/0x800 mm/vmscan.c:2188 madvise_cold_or_pageout_pte_range+0x58a/0x710 mm/madvise.c:453 walk_pmd_range mm/pagewalk.c:53 [inline] walk_pud_range mm/pagewalk.c:112 [inline] walk_p4d_range mm/pagewalk.c:139 [inline] walk_pgd_range mm/pagewalk.c:166 [inline] __walk_page_range+0x45a/0xc20 mm/pagewalk.c:261 walk_page_range+0x179/0x310 mm/pagewalk.c:349 madvise_pageout_page_range mm/madvise.c:506 [inline] madvise_pageout+0x1f0/0x330 mm/madvise.c:542 madvise_vma mm/madvise.c:931 [inline] __do_sys_madvise+0x7d2/0x1600 mm/madvise.c:1113 do_syscall_64+0x9f/0x4c0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe madvise_pageout() accesses the specified range of the vma and isolates them, then runs shrink_page_list() to reclaim its memory. But it also isolates the unevictable pages to reclaim. Hence, we can catch the cases in shrink_page_list(). The root cause is that we scan the page tables instead of specific LRU list. and so we need to filter out the unevictable lru pages from our end. Link: http://lkml.kernel.org/r/1572616245-18946-1-git-send-email-zhongjiang@huawei.com Fixes: 1a4e58cce84e ("mm: introduce MADV_PAGEOUT") Signed-off-by: zhong jiang <zhongjiang@huawei.com> Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-11-16 01:34:36 +00:00
}
} else
folio_deactivate(folio);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
huge_unlock:
spin_unlock(ptl);
if (pageout)
reclaim_pages(&folio_list);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
return 0;
}
regular_folio:
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
#endif
tlb_change_page_size(tlb, PAGE_SIZE);
mm/madvise: add cond_resched() in madvise_cold_or_pageout_pte_range() I conducted real-time testing and observed that madvise_cold_or_pageout_pte_range() causes significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. I tested on the LicheePi 4A board using Cylictest for latency testing and Ftrace for latency tracing. The board uses TH1520 processor and has a memory size of 8GB. The kernel version is 6.5.0 with the PREEMPT_RT patch applied. The script I tested is as follows: echo wakeup_rt > /sys/kernel/tracing/current_tracer echo 1 > /sys/kernel/tracing/tracing_on echo 0 > /sys/kernel/tracing/tracing_max_latency stress-ng --vm 8 --vm-bytes 2G & cyclictest --mlockall --smp --priority=99 --distance=0 --duration=30m echo 0 > /sys/kernel/tracing/tracing_on cat /sys/kernel/tracing/trace The tracing results before modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00003-g999d221864bf # -------------------------------------------------------------------- # latency: 2552 us, #6/6, CPU#3 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-196 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-206 3dn.h512 2us : 206:120:R + [003] 196: 0:R cyclictest stress-n-206 3dn.h512 7us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-206 3dn.h512 9us#: 0 stress-n-206 3d...3.. 2544us : __schedule stress-n-206 3d...3.. 2545us : 206:120:R ==> [003] 196: 0:R cyclictest stress-n-206 3d...3.. 2551us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => rt_spin_unlock => madvise_cold_or_pageout_pte_range => walk_pgd_range => __walk_page_range => walk_page_range => madvise_pageout => madvise_vma_behavior => do_madvise => sys_madvise => do_trap_ecall_u => ret_from_exception The tracing results after modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00004-gca3876fc69a6-dirty # -------------------------------------------------------------------- # latency: 1689 us, #6/6, CPU#0 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-217 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-232 0dn.h413 1us+: 232:120:R + [000] 217: 0:R cyclictest stress-n-232 0dn.h413 12us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-232 0dn.h413 19us#: 0 stress-n-232 0d...3.. 1671us : __schedule stress-n-232 0d...3.. 1676us+: 232:120:R ==> [000] 217: 0:R cyclictest stress-n-232 0d...3.. 1687us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => free_unref_page_list => release_pages => free_pages_and_swap_cache => tlb_batch_pages_flush => tlb_flush_mmu => unmap_page_range => unmap_vmas => unmap_region => do_vmi_align_munmap.constprop.0 => do_vmi_munmap => __vm_munmap => sys_munmap => do_trap_ecall_u => ret_from_exception After the modification, the cause of maximum latency is no longer madvise_cold_or_pageout_pte_range(), so this modification can reduce the latency caused by madvise_cold_or_pageout_pte_range(). Currently the madvise_cold_or_pageout_pte_range() function exhibits significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. When the batch_count reaches SWAP_CLUSTER_MAX, we reschedule the task to ensure fairness and avoid long lock holding times. Link: https://lkml.kernel.org/r/85363861af65fac66c7a98c251906afc0d9c8098.1695291046.git.wangjiexun@tinylab.org Signed-off-by: Jiexun Wang <wangjiexun@tinylab.org> Cc: Zhangjin Wu <falcon@tinylab.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-21 12:27:51 +00:00
restart:
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
start_pte = pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
if (!start_pte)
return 0;
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
flush_tlb_batched_pending(mm);
arch_enter_lazy_mmu_mode();
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
for (; addr < end; pte += nr, addr += nr * PAGE_SIZE) {
nr = 1;
mm: ptep_get() conversion Convert all instances of direct pte_t* dereferencing to instead use ptep_get() helper. This means that by default, the accesses change from a C dereference to a READ_ONCE(). This is technically the correct thing to do since where pgtables are modified by HW (for access/dirty) they are volatile and therefore we should always ensure READ_ONCE() semantics. But more importantly, by always using the helper, it can be overridden by the architecture to fully encapsulate the contents of the pte. Arch code is deliberately not converted, as the arch code knows best. It is intended that arch code (arm64) will override the default with its own implementation that can (e.g.) hide certain bits from the core code, or determine young/dirty status by mixing in state from another source. Conversion was done using Coccinelle: ---- // $ make coccicheck \ // COCCI=ptepget.cocci \ // SPFLAGS="--include-headers" \ // MODE=patch virtual patch @ depends on patch @ pte_t *v; @@ - *v + ptep_get(v) ---- Then reviewed and hand-edited to avoid multiple unnecessary calls to ptep_get(), instead opting to store the result of a single call in a variable, where it is correct to do so. This aims to negate any cost of READ_ONCE() and will benefit arch-overrides that may be more complex. Included is a fix for an issue in an earlier version of this patch that was pointed out by kernel test robot. The issue arose because config MMU=n elides definition of the ptep helper functions, including ptep_get(). HUGETLB_PAGE=n configs still define a simple huge_ptep_clear_flush() for linking purposes, which dereferences the ptep. So when both configs are disabled, this caused a build error because ptep_get() is not defined. Fix by continuing to do a direct dereference when MMU=n. This is safe because for this config the arch code cannot be trying to virtualize the ptes because none of the ptep helpers are defined. Link: https://lkml.kernel.org/r/20230612151545.3317766-4-ryan.roberts@arm.com Reported-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/oe-kbuild-all/202305120142.yXsNEo6H-lkp@intel.com/ Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Potapenko <glider@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alex Williamson <alex.williamson@redhat.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Andrey Konovalov <andreyknvl@gmail.com> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christoph Hellwig <hch@infradead.org> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Dave Airlie <airlied@gmail.com> Cc: Dimitri Sivanich <dimitri.sivanich@hpe.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Ian Rogers <irogers@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Jérôme Glisse <jglisse@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: SeongJae Park <sj@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Uladzislau Rezki (Sony) <urezki@gmail.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-12 15:15:45 +00:00
ptent = ptep_get(pte);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
mm/madvise: add cond_resched() in madvise_cold_or_pageout_pte_range() I conducted real-time testing and observed that madvise_cold_or_pageout_pte_range() causes significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. I tested on the LicheePi 4A board using Cylictest for latency testing and Ftrace for latency tracing. The board uses TH1520 processor and has a memory size of 8GB. The kernel version is 6.5.0 with the PREEMPT_RT patch applied. The script I tested is as follows: echo wakeup_rt > /sys/kernel/tracing/current_tracer echo 1 > /sys/kernel/tracing/tracing_on echo 0 > /sys/kernel/tracing/tracing_max_latency stress-ng --vm 8 --vm-bytes 2G & cyclictest --mlockall --smp --priority=99 --distance=0 --duration=30m echo 0 > /sys/kernel/tracing/tracing_on cat /sys/kernel/tracing/trace The tracing results before modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00003-g999d221864bf # -------------------------------------------------------------------- # latency: 2552 us, #6/6, CPU#3 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-196 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-206 3dn.h512 2us : 206:120:R + [003] 196: 0:R cyclictest stress-n-206 3dn.h512 7us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-206 3dn.h512 9us#: 0 stress-n-206 3d...3.. 2544us : __schedule stress-n-206 3d...3.. 2545us : 206:120:R ==> [003] 196: 0:R cyclictest stress-n-206 3d...3.. 2551us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => rt_spin_unlock => madvise_cold_or_pageout_pte_range => walk_pgd_range => __walk_page_range => walk_page_range => madvise_pageout => madvise_vma_behavior => do_madvise => sys_madvise => do_trap_ecall_u => ret_from_exception The tracing results after modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00004-gca3876fc69a6-dirty # -------------------------------------------------------------------- # latency: 1689 us, #6/6, CPU#0 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-217 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-232 0dn.h413 1us+: 232:120:R + [000] 217: 0:R cyclictest stress-n-232 0dn.h413 12us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-232 0dn.h413 19us#: 0 stress-n-232 0d...3.. 1671us : __schedule stress-n-232 0d...3.. 1676us+: 232:120:R ==> [000] 217: 0:R cyclictest stress-n-232 0d...3.. 1687us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => free_unref_page_list => release_pages => free_pages_and_swap_cache => tlb_batch_pages_flush => tlb_flush_mmu => unmap_page_range => unmap_vmas => unmap_region => do_vmi_align_munmap.constprop.0 => do_vmi_munmap => __vm_munmap => sys_munmap => do_trap_ecall_u => ret_from_exception After the modification, the cause of maximum latency is no longer madvise_cold_or_pageout_pte_range(), so this modification can reduce the latency caused by madvise_cold_or_pageout_pte_range(). Currently the madvise_cold_or_pageout_pte_range() function exhibits significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. When the batch_count reaches SWAP_CLUSTER_MAX, we reschedule the task to ensure fairness and avoid long lock holding times. Link: https://lkml.kernel.org/r/85363861af65fac66c7a98c251906afc0d9c8098.1695291046.git.wangjiexun@tinylab.org Signed-off-by: Jiexun Wang <wangjiexun@tinylab.org> Cc: Zhangjin Wu <falcon@tinylab.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-21 12:27:51 +00:00
if (++batch_count == SWAP_CLUSTER_MAX) {
batch_count = 0;
if (need_resched()) {
arch_leave_lazy_mmu_mode();
mm/madvise: add cond_resched() in madvise_cold_or_pageout_pte_range() I conducted real-time testing and observed that madvise_cold_or_pageout_pte_range() causes significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. I tested on the LicheePi 4A board using Cylictest for latency testing and Ftrace for latency tracing. The board uses TH1520 processor and has a memory size of 8GB. The kernel version is 6.5.0 with the PREEMPT_RT patch applied. The script I tested is as follows: echo wakeup_rt > /sys/kernel/tracing/current_tracer echo 1 > /sys/kernel/tracing/tracing_on echo 0 > /sys/kernel/tracing/tracing_max_latency stress-ng --vm 8 --vm-bytes 2G & cyclictest --mlockall --smp --priority=99 --distance=0 --duration=30m echo 0 > /sys/kernel/tracing/tracing_on cat /sys/kernel/tracing/trace The tracing results before modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00003-g999d221864bf # -------------------------------------------------------------------- # latency: 2552 us, #6/6, CPU#3 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-196 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-206 3dn.h512 2us : 206:120:R + [003] 196: 0:R cyclictest stress-n-206 3dn.h512 7us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-206 3dn.h512 9us#: 0 stress-n-206 3d...3.. 2544us : __schedule stress-n-206 3d...3.. 2545us : 206:120:R ==> [003] 196: 0:R cyclictest stress-n-206 3d...3.. 2551us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => rt_spin_unlock => madvise_cold_or_pageout_pte_range => walk_pgd_range => __walk_page_range => walk_page_range => madvise_pageout => madvise_vma_behavior => do_madvise => sys_madvise => do_trap_ecall_u => ret_from_exception The tracing results after modification are as follows: # tracer: wakeup_rt # # wakeup_rt latency trace v1.1.5 on 6.5.0-rt6-r1208-00004-gca3876fc69a6-dirty # -------------------------------------------------------------------- # latency: 1689 us, #6/6, CPU#0 | (M:preempt_rt VP:0, KP:0, SP:0 HP:0 #P:4) # ----------------- # | task: cyclictest-217 (uid:0 nice:0 policy:1 rt_prio:99) # ----------------- # # _--------=> CPU# # / _-------=> irqs-off/BH-disabled # | / _------=> need-resched # || / _-----=> need-resched-lazy # ||| / _----=> hardirq/softirq # |||| / _---=> preempt-depth # ||||| / _--=> preempt-lazy-depth # |||||| / _-=> migrate-disable # ||||||| / delay # cmd pid |||||||| time | caller # \ / |||||||| \ | / stress-n-232 0dn.h413 1us+: 232:120:R + [000] 217: 0:R cyclictest stress-n-232 0dn.h413 12us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup => ttwu_do_activate => try_to_wake_up => wake_up_process => hrtimer_wakeup => __hrtimer_run_queues => hrtimer_interrupt => riscv_timer_interrupt => handle_percpu_devid_irq => generic_handle_domain_irq => riscv_intc_irq => handle_riscv_irq => do_irq stress-n-232 0dn.h413 19us#: 0 stress-n-232 0d...3.. 1671us : __schedule stress-n-232 0d...3.. 1676us+: 232:120:R ==> [000] 217: 0:R cyclictest stress-n-232 0d...3.. 1687us : <stack trace> => __ftrace_trace_stack => __trace_stack => probe_wakeup_sched_switch => __schedule => preempt_schedule => migrate_enable => free_unref_page_list => release_pages => free_pages_and_swap_cache => tlb_batch_pages_flush => tlb_flush_mmu => unmap_page_range => unmap_vmas => unmap_region => do_vmi_align_munmap.constprop.0 => do_vmi_munmap => __vm_munmap => sys_munmap => do_trap_ecall_u => ret_from_exception After the modification, the cause of maximum latency is no longer madvise_cold_or_pageout_pte_range(), so this modification can reduce the latency caused by madvise_cold_or_pageout_pte_range(). Currently the madvise_cold_or_pageout_pte_range() function exhibits significant latency under memory pressure, which can be effectively reduced by adding cond_resched() within the loop. When the batch_count reaches SWAP_CLUSTER_MAX, we reschedule the task to ensure fairness and avoid long lock holding times. Link: https://lkml.kernel.org/r/85363861af65fac66c7a98c251906afc0d9c8098.1695291046.git.wangjiexun@tinylab.org Signed-off-by: Jiexun Wang <wangjiexun@tinylab.org> Cc: Zhangjin Wu <falcon@tinylab.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-21 12:27:51 +00:00
pte_unmap_unlock(start_pte, ptl);
cond_resched();
goto restart;
}
}
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
if (pte_none(ptent))
continue;
if (!pte_present(ptent))
continue;
folio = vm_normal_folio(vma, addr, ptent);
if (!folio || folio_is_zone_device(folio))
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
continue;
/*
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
* If we encounter a large folio, only split it if it is not
* fully mapped within the range we are operating on. Otherwise
* leave it as is so that it can be swapped out whole. If we
* fail to split a folio, leave it in place and advance to the
* next pte in the range.
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
*/
if (folio_test_large(folio)) {
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
bool any_young;
nr = madvise_folio_pte_batch(addr, end, folio, pte,
ptent, &any_young, NULL);
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
if (any_young)
ptent = pte_mkyoung(ptent);
if (nr < folio_nr_pages(folio)) {
int err;
if (folio_likely_mapped_shared(folio))
continue;
if (pageout_anon_only_filter && !folio_test_anon(folio))
continue;
if (!folio_trylock(folio))
continue;
folio_get(folio);
arch_leave_lazy_mmu_mode();
pte_unmap_unlock(start_pte, ptl);
start_pte = NULL;
err = split_folio(folio);
folio_unlock(folio);
folio_put(folio);
start_pte = pte =
pte_offset_map_lock(mm, pmd, addr, &ptl);
if (!start_pte)
break;
arch_enter_lazy_mmu_mode();
if (!err)
nr = 0;
continue;
}
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
}
/*
* Do not interfere with other mappings of this folio and
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
* non-LRU folio. If we have a large folio at this point, we
* know it is fully mapped so if its mapcount is the same as its
* number of pages, it must be exclusive.
*/
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
if (!folio_test_lru(folio) ||
folio_mapcount(folio) != folio_nr_pages(folio))
mm: do not allow MADV_PAGEOUT for CoW pages Jann has brought up a very interesting point [1]. While shared pages are excluded from MADV_PAGEOUT normally, CoW pages can be easily reclaimed that way. This can lead to all sorts of hard to debug problems. E.g. performance problems outlined by Daniel [2]. There are runtime environments where there is a substantial memory shared among security domains via CoW memory and a easy to reclaim way of that memory, which MADV_{COLD,PAGEOUT} offers, can lead to either performance degradation in for the parent process which might be more privileged or even open side channel attacks. The feasibility of the latter is not really clear to me TBH but there is no real reason for exposure at this stage. It seems there is no real use case to depend on reclaiming CoW memory via madvise at this stage so it is much easier to simply disallow it and this is what this patch does. Put it simply MADV_{PAGEOUT,COLD} can operate only on the exclusively owned memory which is a straightforward semantic. [1] http://lkml.kernel.org/r/CAG48ez0G3JkMq61gUmyQAaCq=_TwHbi1XKzWRooxZkv08PQKuw@mail.gmail.com [2] http://lkml.kernel.org/r/CAKOZueua_v8jHCpmEtTB6f3i9e2YnmX4mqdYVWhV4E=Z-n+zRQ@mail.gmail.com Fixes: 9c276cc65a58 ("mm: introduce MADV_COLD") Reported-by: Jann Horn <jannh@google.com> Signed-off-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Minchan Kim <minchan@kernel.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: "Joel Fernandes (Google)" <joel@joelfernandes.org> Cc: <stable@vger.kernel.org> Link: http://lkml.kernel.org/r/20200312082248.GS23944@dhcp22.suse.cz Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-03-22 01:22:26 +00:00
continue;
if (pageout_anon_only_filter && !folio_test_anon(folio))
continue;
mm: madvise: pageout: ignore references rather than clearing young While doing MADV_PAGEOUT, the current code will clear PTE young so that vmscan won't read young flags to allow the reclamation of madvised folios to go ahead. It seems we can do it by directly ignoring references, thus we can remove tlb flush in madvise and rmap overhead in vmscan. Regarding the side effect, in the original code, if a parallel thread runs side by side to access the madvised memory with the thread doing madvise, folios will get a chance to be re-activated by vmscan (though the time gap is actually quite small since checking PTEs is done immediately after clearing PTEs young). But with this patch, they will still be reclaimed. But this behaviour doing PAGEOUT and doing access at the same time is quite silly like DoS. So probably, we don't need to care. Or ignoring the new access during the quite small time gap is even better. For DAMON's DAMOS_PAGEOUT based on physical address region, we still keep its behaviour as is since a physical address might be mapped by multiple processes. MADV_PAGEOUT based on virtual address is actually much more aggressive on reclamation. To untouch paddr's DAMOS_PAGEOUT, we simply pass ignore_references as false in reclaim_pages(). A microbench as below has shown 6% decrement on the latency of MADV_PAGEOUT, #define PGSIZE 4096 main() { int i; #define SIZE 512*1024*1024 volatile long *p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); for (i = 0; i < SIZE/sizeof(long); i += PGSIZE / sizeof(long)) p[i] = 0x11; madvise(p, SIZE, MADV_PAGEOUT); } w/o patch w/ patch root@10:~# time ./a.out root@10:~# time ./a.out real 0m49.634s real 0m46.334s user 0m0.637s user 0m0.648s sys 0m47.434s sys 0m44.265s Link: https://lkml.kernel.org/r/20240226005739.24350-1-21cnbao@gmail.com Signed-off-by: Barry Song <v-songbaohua@oppo.com> Acked-by: Minchan Kim <minchan@kernel.org> Cc: SeongJae Park <sj@kernel.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-02-26 00:57:39 +00:00
if (!pageout && pte_young(ptent)) {
mm/madvise: introduce clear_young_dirty_ptes() batch helper Patch series "mm/madvise: enhance lazyfreeing with mTHP in madvise_free", v10. This patchset adds support for lazyfreeing multi-size THP (mTHP) without needing to first split the large folio via split_folio(). However, we still need to split a large folio that is not fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. Performance Testing =================== On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% This patch (of 4): This commit introduces clear_young_dirty_ptes() to replace mkold_ptes(). By doing so, we can use the same function for both use cases (madvise_pageout and madvise_free), and it also provides the flexibility to only clear the dirty flag in the future if needed. Link: https://lkml.kernel.org/r/20240418134435.6092-1-ioworker0@gmail.com Link: https://lkml.kernel.org/r/20240418134435.6092-2-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Suggested-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:32 +00:00
clear_young_dirty_ptes(vma, addr, pte, nr,
CYDP_CLEAR_YOUNG);
mm: madvise: avoid split during MADV_PAGEOUT and MADV_COLD Rework madvise_cold_or_pageout_pte_range() to avoid splitting any large folio that is fully and contiguously mapped in the pageout/cold vm range. This change means that large folios will be maintained all the way to swap storage. This both improves performance during swap-out, by eliding the cost of splitting the folio, and sets us up nicely for maintaining the large folio when it is swapped back in (to be covered in a separate series). Folios that are not fully mapped in the target range are still split, but note that behavior is changed so that if the split fails for any reason (folio locked, shared, etc) we now leave it as is and move to the next pte in the range and continue work on the proceeding folios. Previously any failure of this sort would cause the entire operation to give up and no folios mapped at higher addresses were paged out or made cold. Given large folios are becoming more common, this old behavior would have likely lead to wasted opportunities. While we are at it, change the code that clears young from the ptes to use ptep_test_and_clear_young(), via the new mkold_ptes() batch helper function. This is more efficent than get_and_clear/modify/set, especially for contpte mappings on arm64, where the old approach would require unfolding/refolding and the new approach can be done in place. Link: https://lkml.kernel.org/r/20240408183946.2991168-8-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Reviewed-by: Barry Song <v-songbaohua@oppo.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:46 +00:00
tlb_remove_tlb_entries(tlb, pte, nr, addr);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
}
/*
* We are deactivating a folio for accelerating reclaiming.
* VM couldn't reclaim the folio unless we clear PG_young.
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
* As a side effect, it makes confuse idle-page tracking
* because they will miss recent referenced history.
*/
folio_clear_referenced(folio);
folio_test_clear_young(folio);
mm: madvise: fix uneven accounting of psi A folio turns into a Workingset during: 1) shrink_active_list() placing the folio from active to inactive list. 2) When a workingset transition is happening during the folio refault. And when Workingset is set on a folio, PSI for memory can be accounted during a) That folio is being reclaimed and b) Refault of that folio, for usual reclaims. This accounting of PSI for memory is not consistent for reclaim + refault operation between usual reclaim and madvise(COLD/PAGEOUT) which deactivate or proactively reclaim a folio: a) A folio started at inactive and moved to active as part of accesses. Workingset is absent on the folio thus refault of it when reclaimed through MADV_PAGEOUT operation doesn't account for PSI. b) When the same folio transition from inactive->active and then to inactive through shrink_active_list(). Workingset is set on the folio thus refault of it when reclaimed through MADV_PAGEOUT operation accounts for PSI. c) When the same folio is part of active list directly as a result of folio refault and this was a workingset folio prior to eviction. Workingset is set on the folio thus the refault of it when reclaimed through MADV_PAGEOUT/MADV_COLD operation accounts for PSI. d) MADV_COLD transfers the folio from active list to inactive list. Such folios may not have the Workingset thus refault operation on such folio doesn't account for PSI. As said above, refault operation caused because of MADV_PAGEOUT on a folio is accounts for memory PSI in b) and c) but not in a). Refault caused by the reclaim of a folio on which MADV_COLD is performed accounts memory PSI in c) but not in d). These behaviours are inconsistent w.r.t usual reclaim + refault operation. Make this PSI accounting always consistent by turning a folio into a workingset one whenever it is leaving the active list. Also, accounting of PSI on a folio whenever it leaves the active list as part of the MADV_COLD/PAGEOUT operation helps the users whether they are operating on proper folios[1]. [1] https://lore.kernel.org/all/20230605180013.GD221380@cmpxchg.org/ Link: https://lkml.kernel.org/r/1688393201-11135-1-git-send-email-quic_charante@quicinc.com Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com> Suggested-by: Suren Baghdasaryan <surenb@google.com> Reported-by: Sai Manobhiram Manapragada <quic_smanapra@quicinc.com> Reported-by: Pavan Kondeti <quic_pkondeti@quicinc.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Pavankumar Kondeti <quic_pkondeti@quicinc.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-07-03 14:06:41 +00:00
if (folio_test_active(folio))
folio_set_workingset(folio);
if (pageout) {
mm: change to return bool for folio_isolate_lru() Patch series "Change the return value for page isolation functions", v3. Now the page isolation functions did not return a boolean to indicate success or not, instead it will return a negative error when failed to isolate a page. So below code used in most places seem a boolean success/failure thing, which can confuse people whether the isolation is successful. if (folio_isolate_lru(folio)) continue; Moreover the page isolation functions only return 0 or -EBUSY, and most users did not care about the negative error except for few users, thus we can convert all page isolation functions to return a boolean value, which can remove the confusion to make code more clear. No functional changes intended in this patch series. This patch (of 4): Now the folio_isolate_lru() did not return a boolean value to indicate isolation success or not, however below code checking the return value can make people think that it was a boolean success/failure thing, which makes people easy to make mistakes (see the fix patch[1]). if (folio_isolate_lru(folio)) continue; Thus it's better to check the negative error value expilictly returned by folio_isolate_lru(), which makes code more clear per Linus's suggestion[2]. Moreover Matthew suggested we can convert the isolation functions to return a boolean[3], since most users did not care about the negative error value, and can also remove the confusing of checking return value. So this patch converts the folio_isolate_lru() to return a boolean value, which means return 'true' to indicate the folio isolation is successful, and 'false' means a failure to isolation. Meanwhile changing all users' logic of checking the isolation state. No functional changes intended. [1] https://lore.kernel.org/all/20230131063206.28820-1-Kuan-Ying.Lee@mediatek.com/T/#u [2] https://lore.kernel.org/all/CAHk-=wiBrY+O-4=2mrbVyxR+hOqfdJ=Do6xoucfJ9_5az01L4Q@mail.gmail.com/ [3] https://lore.kernel.org/all/Y+sTFqwMNAjDvxw3@casper.infradead.org/ Link: https://lkml.kernel.org/r/cover.1676424378.git.baolin.wang@linux.alibaba.com Link: https://lkml.kernel.org/r/8a4e3679ed4196168efadf7ea36c038f2f7d5aa9.1676424378.git.baolin.wang@linux.alibaba.com Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by: SeongJae Park <sj@kernel.org> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <muchun.song@linux.dev> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Shakeel Butt <shakeelb@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-02-15 10:39:34 +00:00
if (folio_isolate_lru(folio)) {
if (folio_test_unevictable(folio))
folio_putback_lru(folio);
mm: fix trying to reclaim unevictable lru page when calling madvise_pageout Recently, I hit the following issue when running upstream. kernel BUG at mm/vmscan.c:1521! invalid opcode: 0000 [#1] SMP KASAN PTI CPU: 0 PID: 23385 Comm: syz-executor.6 Not tainted 5.4.0-rc4+ #1 RIP: 0010:shrink_page_list+0x12b6/0x3530 mm/vmscan.c:1521 Call Trace: reclaim_pages+0x499/0x800 mm/vmscan.c:2188 madvise_cold_or_pageout_pte_range+0x58a/0x710 mm/madvise.c:453 walk_pmd_range mm/pagewalk.c:53 [inline] walk_pud_range mm/pagewalk.c:112 [inline] walk_p4d_range mm/pagewalk.c:139 [inline] walk_pgd_range mm/pagewalk.c:166 [inline] __walk_page_range+0x45a/0xc20 mm/pagewalk.c:261 walk_page_range+0x179/0x310 mm/pagewalk.c:349 madvise_pageout_page_range mm/madvise.c:506 [inline] madvise_pageout+0x1f0/0x330 mm/madvise.c:542 madvise_vma mm/madvise.c:931 [inline] __do_sys_madvise+0x7d2/0x1600 mm/madvise.c:1113 do_syscall_64+0x9f/0x4c0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe madvise_pageout() accesses the specified range of the vma and isolates them, then runs shrink_page_list() to reclaim its memory. But it also isolates the unevictable pages to reclaim. Hence, we can catch the cases in shrink_page_list(). The root cause is that we scan the page tables instead of specific LRU list. and so we need to filter out the unevictable lru pages from our end. Link: http://lkml.kernel.org/r/1572616245-18946-1-git-send-email-zhongjiang@huawei.com Fixes: 1a4e58cce84e ("mm: introduce MADV_PAGEOUT") Signed-off-by: zhong jiang <zhongjiang@huawei.com> Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-11-16 01:34:36 +00:00
else
list_add(&folio->lru, &folio_list);
mm: fix trying to reclaim unevictable lru page when calling madvise_pageout Recently, I hit the following issue when running upstream. kernel BUG at mm/vmscan.c:1521! invalid opcode: 0000 [#1] SMP KASAN PTI CPU: 0 PID: 23385 Comm: syz-executor.6 Not tainted 5.4.0-rc4+ #1 RIP: 0010:shrink_page_list+0x12b6/0x3530 mm/vmscan.c:1521 Call Trace: reclaim_pages+0x499/0x800 mm/vmscan.c:2188 madvise_cold_or_pageout_pte_range+0x58a/0x710 mm/madvise.c:453 walk_pmd_range mm/pagewalk.c:53 [inline] walk_pud_range mm/pagewalk.c:112 [inline] walk_p4d_range mm/pagewalk.c:139 [inline] walk_pgd_range mm/pagewalk.c:166 [inline] __walk_page_range+0x45a/0xc20 mm/pagewalk.c:261 walk_page_range+0x179/0x310 mm/pagewalk.c:349 madvise_pageout_page_range mm/madvise.c:506 [inline] madvise_pageout+0x1f0/0x330 mm/madvise.c:542 madvise_vma mm/madvise.c:931 [inline] __do_sys_madvise+0x7d2/0x1600 mm/madvise.c:1113 do_syscall_64+0x9f/0x4c0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe madvise_pageout() accesses the specified range of the vma and isolates them, then runs shrink_page_list() to reclaim its memory. But it also isolates the unevictable pages to reclaim. Hence, we can catch the cases in shrink_page_list(). The root cause is that we scan the page tables instead of specific LRU list. and so we need to filter out the unevictable lru pages from our end. Link: http://lkml.kernel.org/r/1572616245-18946-1-git-send-email-zhongjiang@huawei.com Fixes: 1a4e58cce84e ("mm: introduce MADV_PAGEOUT") Signed-off-by: zhong jiang <zhongjiang@huawei.com> Suggested-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-11-16 01:34:36 +00:00
}
} else
folio_deactivate(folio);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
}
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
if (start_pte) {
arch_leave_lazy_mmu_mode();
pte_unmap_unlock(start_pte, ptl);
}
if (pageout)
reclaim_pages(&folio_list);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
cond_resched();
return 0;
}
static const struct mm_walk_ops cold_walk_ops = {
.pmd_entry = madvise_cold_or_pageout_pte_range,
mm: enable page walking API to lock vmas during the walk walk_page_range() and friends often operate under write-locked mmap_lock. With introduction of vma locks, the vmas have to be locked as well during such walks to prevent concurrent page faults in these areas. Add an additional member to mm_walk_ops to indicate locking requirements for the walk. The change ensures that page walks which prevent concurrent page faults by write-locking mmap_lock, operate correctly after introduction of per-vma locks. With per-vma locks page faults can be handled under vma lock without taking mmap_lock at all, so write locking mmap_lock would not stop them. The change ensures vmas are properly locked during such walks. A sample issue this solves is do_mbind() performing queue_pages_range() to queue pages for migration. Without this change a concurrent page can be faulted into the area and be left out of migration. Link: https://lkml.kernel.org/r/20230804152724.3090321-2-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org> Suggested-by: Jann Horn <jannh@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Laurent Dufour <ldufour@linux.ibm.com> Cc: Liam Howlett <liam.howlett@oracle.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Michel Lespinasse <michel@lespinasse.org> Cc: Peter Xu <peterx@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-08-04 15:27:19 +00:00
.walk_lock = PGWALK_RDLOCK,
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
};
static void madvise_cold_page_range(struct mmu_gather *tlb,
struct vm_area_struct *vma,
unsigned long addr, unsigned long end)
{
struct madvise_walk_private walk_private = {
.pageout = false,
.tlb = tlb,
};
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
tlb_start_vma(tlb, vma);
walk_page_range(vma->vm_mm, addr, end, &cold_walk_ops, &walk_private);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
tlb_end_vma(tlb, vma);
}
static inline bool can_madv_lru_vma(struct vm_area_struct *vma)
{
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
return !(vma->vm_flags & (VM_LOCKED|VM_PFNMAP|VM_HUGETLB));
}
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
static long madvise_cold(struct vm_area_struct *vma,
struct vm_area_struct **prev,
unsigned long start_addr, unsigned long end_addr)
{
struct mm_struct *mm = vma->vm_mm;
struct mmu_gather tlb;
*prev = vma;
if (!can_madv_lru_vma(vma))
return -EINVAL;
lru_add_drain();
tlb_gather_mmu(&tlb, mm);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
madvise_cold_page_range(&tlb, vma, start_addr, end_addr);
tlb_finish_mmu(&tlb);
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
return 0;
}
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
static void madvise_pageout_page_range(struct mmu_gather *tlb,
struct vm_area_struct *vma,
unsigned long addr, unsigned long end)
{
struct madvise_walk_private walk_private = {
.pageout = true,
.tlb = tlb,
};
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
tlb_start_vma(tlb, vma);
walk_page_range(vma->vm_mm, addr, end, &cold_walk_ops, &walk_private);
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
tlb_end_vma(tlb, vma);
}
static long madvise_pageout(struct vm_area_struct *vma,
struct vm_area_struct **prev,
unsigned long start_addr, unsigned long end_addr)
{
struct mm_struct *mm = vma->vm_mm;
struct mmu_gather tlb;
*prev = vma;
if (!can_madv_lru_vma(vma))
return -EINVAL;
/*
* If the VMA belongs to a private file mapping, there can be private
* dirty pages which can be paged out if even this process is neither
* owner nor write capable of the file. We allow private file mappings
* further to pageout dirty anon pages.
*/
if (!vma_is_anonymous(vma) && (!can_do_file_pageout(vma) &&
(vma->vm_flags & VM_MAYSHARE)))
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
return 0;
lru_add_drain();
tlb_gather_mmu(&tlb, mm);
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
madvise_pageout_page_range(&tlb, vma, start_addr, end_addr);
tlb_finish_mmu(&tlb);
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
return 0;
}
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
static int madvise_free_pte_range(pmd_t *pmd, unsigned long addr,
unsigned long end, struct mm_walk *walk)
{
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
const cydp_t cydp_flags = CYDP_CLEAR_YOUNG | CYDP_CLEAR_DIRTY;
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
struct mmu_gather *tlb = walk->private;
struct mm_struct *mm = tlb->mm;
struct vm_area_struct *vma = walk->vma;
spinlock_t *ptl;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
pte_t *start_pte, *pte, ptent;
struct folio *folio;
mm/madvise.c: free swp_entry in madvise_free When I test below piece of code with 12 processes(ie, 512M * 12 = 6G consume) on my (3G ram + 12 cpu + 8G swap, the madvise_free is siginficat slower (ie, 2x times) than madvise_dontneed. loop = 5; mmap(512M); while (loop--) { memset(512M); madvise(MADV_FREE or MADV_DONTNEED); } The reason is lots of swapin. 1) dontneed: 1,612 swapin 2) madvfree: 879,585 swapin If we find hinted pages were already swapped out when syscall is called, it's pointless to keep the swapped-out pages in pte. Instead, let's free the cold page because swapin is more expensive than (alloc page + zeroing). With this patch, it reduced swapin from 879,585 to 1,878 so elapsed time 1) dontneed: 6.10user 233.50system 0:50.44elapsed 2) madvfree: 6.03user 401.17system 1:30.67elapsed 2) madvfree + below patch: 6.70user 339.14system 1:04.45elapsed Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:06 +00:00
int nr_swap = 0;
mm/huge_memory.c: don't split THP page when MADV_FREE syscall is called We don't need to split THP page when MADV_FREE syscall is called if [start, len] is aligned with THP size. The split could be done when VM decide to free it in reclaim path if memory pressure is heavy. With that, we could avoid unnecessary THP split. For the feature, this patch changes pte dirtness marking logic of THP. Now, it marks every ptes of pages dirty unconditionally in splitting, which makes MADV_FREE void. So, instead, this patch propagates pmd dirtiness to all pages via PG_dirty and restores pte dirtiness from PG_dirty. With this, if pmd is clean(ie, MADV_FREEed) when split happens(e,g, shrink_page_list), all of pages are clean too so we could discard them. Signed-off-by: Minchan Kim <minchan@kernel.org> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Hugh Dickins <hughd@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:42 +00:00
unsigned long next;
mm: swap: free_swap_and_cache_nr() as batched free_swap_and_cache() Now that we no longer have a convenient flag in the cluster to determine if a folio is large, free_swap_and_cache() will take a reference and lock a large folio much more often, which could lead to contention and (e.g.) failure to split large folios, etc. Let's solve that problem by batch freeing swap and cache with a new function, free_swap_and_cache_nr(), to free a contiguous range of swap entries together. This allows us to first drop a reference to each swap slot before we try to release the cache folio. This means we only try to release the folio once, only taking the reference and lock once - much better than the previous 512 times for the 2M THP case. Contiguous swap entries are gathered in zap_pte_range() and madvise_free_pte_range() in a similar way to how present ptes are already gathered in zap_pte_range(). While we are at it, let's simplify by converting the return type of both functions to void. The return value was used only by zap_pte_range() to print a bad pte, and was ignored by everyone else, so the extra reporting wasn't exactly guaranteed. We will still get the warning with most of the information from get_swap_device(). With the batch version, we wouldn't know which pte was bad anyway so could print the wrong one. [ryan.roberts@arm.com: fix a build warning on parisc] Link: https://lkml.kernel.org/r/20240409111840.3173122-1-ryan.roberts@arm.com Link: https://lkml.kernel.org/r/20240408183946.2991168-3-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Barry Song <v-songbaohua@oppo.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:41 +00:00
int nr, max_nr;
mm/huge_memory.c: don't split THP page when MADV_FREE syscall is called We don't need to split THP page when MADV_FREE syscall is called if [start, len] is aligned with THP size. The split could be done when VM decide to free it in reclaim path if memory pressure is heavy. With that, we could avoid unnecessary THP split. For the feature, this patch changes pte dirtness marking logic of THP. Now, it marks every ptes of pages dirty unconditionally in splitting, which makes MADV_FREE void. So, instead, this patch propagates pmd dirtiness to all pages via PG_dirty and restores pte dirtiness from PG_dirty. With this, if pmd is clean(ie, MADV_FREEed) when split happens(e,g, shrink_page_list), all of pages are clean too so we could discard them. Signed-off-by: Minchan Kim <minchan@kernel.org> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Hugh Dickins <hughd@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:42 +00:00
next = pmd_addr_end(addr, end);
if (pmd_trans_huge(*pmd))
if (madvise_free_huge_pmd(tlb, vma, pmd, addr, next))
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
return 0;
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
tlb_change_page_size(tlb, PAGE_SIZE);
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
start_pte = pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
if (!start_pte)
return 0;
mm, mprotect: flush TLB if potentially racing with a parallel reclaim leaving stale TLB entries Nadav Amit identified a theoritical race between page reclaim and mprotect due to TLB flushes being batched outside of the PTL being held. He described the race as follows: CPU0 CPU1 ---- ---- user accesses memory using RW PTE [PTE now cached in TLB] try_to_unmap_one() ==> ptep_get_and_clear() ==> set_tlb_ubc_flush_pending() mprotect(addr, PROT_READ) ==> change_pte_range() ==> [ PTE non-present - no flush ] user writes using cached RW PTE ... try_to_unmap_flush() The same type of race exists for reads when protecting for PROT_NONE and also exists for operations that can leave an old TLB entry behind such as munmap, mremap and madvise. For some operations like mprotect, it's not necessarily a data integrity issue but it is a correctness issue as there is a window where an mprotect that limits access still allows access. For munmap, it's potentially a data integrity issue although the race is massive as an munmap, mmap and return to userspace must all complete between the window when reclaim drops the PTL and flushes the TLB. However, it's theoritically possible so handle this issue by flushing the mm if reclaim is potentially currently batching TLB flushes. Other instances where a flush is required for a present pte should be ok as either the page lock is held preventing parallel reclaim or a page reference count is elevated preventing a parallel free leading to corruption. In the case of page_mkclean there isn't an obvious path that userspace could take advantage of without using the operations that are guarded by this patch. Other users such as gup as a race with reclaim looks just at PTEs. huge page variants should be ok as they don't race with reclaim. mincore only looks at PTEs. userfault also should be ok as if a parallel reclaim takes place, it will either fault the page back in or read some of the data before the flush occurs triggering a fault. Note that a variant of this patch was acked by Andy Lutomirski but this was for the x86 parts on top of his PCID work which didn't make the 4.13 merge window as expected. His ack is dropped from this version and there will be a follow-on patch on top of PCID that will include his ack. [akpm@linux-foundation.org: tweak comments] [akpm@linux-foundation.org: fix spello] Link: http://lkml.kernel.org/r/20170717155523.emckq2esjro6hf3z@suse.de Reported-by: Nadav Amit <nadav.amit@gmail.com> Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Andy Lutomirski <luto@kernel.org> Cc: <stable@vger.kernel.org> [v4.4+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-08-02 20:31:52 +00:00
flush_tlb_batched_pending(mm);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
arch_enter_lazy_mmu_mode();
mm: swap: free_swap_and_cache_nr() as batched free_swap_and_cache() Now that we no longer have a convenient flag in the cluster to determine if a folio is large, free_swap_and_cache() will take a reference and lock a large folio much more often, which could lead to contention and (e.g.) failure to split large folios, etc. Let's solve that problem by batch freeing swap and cache with a new function, free_swap_and_cache_nr(), to free a contiguous range of swap entries together. This allows us to first drop a reference to each swap slot before we try to release the cache folio. This means we only try to release the folio once, only taking the reference and lock once - much better than the previous 512 times for the 2M THP case. Contiguous swap entries are gathered in zap_pte_range() and madvise_free_pte_range() in a similar way to how present ptes are already gathered in zap_pte_range(). While we are at it, let's simplify by converting the return type of both functions to void. The return value was used only by zap_pte_range() to print a bad pte, and was ignored by everyone else, so the extra reporting wasn't exactly guaranteed. We will still get the warning with most of the information from get_swap_device(). With the batch version, we wouldn't know which pte was bad anyway so could print the wrong one. [ryan.roberts@arm.com: fix a build warning on parisc] Link: https://lkml.kernel.org/r/20240409111840.3173122-1-ryan.roberts@arm.com Link: https://lkml.kernel.org/r/20240408183946.2991168-3-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Barry Song <v-songbaohua@oppo.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:41 +00:00
for (; addr != end; pte += nr, addr += PAGE_SIZE * nr) {
nr = 1;
mm: ptep_get() conversion Convert all instances of direct pte_t* dereferencing to instead use ptep_get() helper. This means that by default, the accesses change from a C dereference to a READ_ONCE(). This is technically the correct thing to do since where pgtables are modified by HW (for access/dirty) they are volatile and therefore we should always ensure READ_ONCE() semantics. But more importantly, by always using the helper, it can be overridden by the architecture to fully encapsulate the contents of the pte. Arch code is deliberately not converted, as the arch code knows best. It is intended that arch code (arm64) will override the default with its own implementation that can (e.g.) hide certain bits from the core code, or determine young/dirty status by mixing in state from another source. Conversion was done using Coccinelle: ---- // $ make coccicheck \ // COCCI=ptepget.cocci \ // SPFLAGS="--include-headers" \ // MODE=patch virtual patch @ depends on patch @ pte_t *v; @@ - *v + ptep_get(v) ---- Then reviewed and hand-edited to avoid multiple unnecessary calls to ptep_get(), instead opting to store the result of a single call in a variable, where it is correct to do so. This aims to negate any cost of READ_ONCE() and will benefit arch-overrides that may be more complex. Included is a fix for an issue in an earlier version of this patch that was pointed out by kernel test robot. The issue arose because config MMU=n elides definition of the ptep helper functions, including ptep_get(). HUGETLB_PAGE=n configs still define a simple huge_ptep_clear_flush() for linking purposes, which dereferences the ptep. So when both configs are disabled, this caused a build error because ptep_get() is not defined. Fix by continuing to do a direct dereference when MMU=n. This is safe because for this config the arch code cannot be trying to virtualize the ptes because none of the ptep helpers are defined. Link: https://lkml.kernel.org/r/20230612151545.3317766-4-ryan.roberts@arm.com Reported-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/oe-kbuild-all/202305120142.yXsNEo6H-lkp@intel.com/ Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Potapenko <glider@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alex Williamson <alex.williamson@redhat.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Andrey Konovalov <andreyknvl@gmail.com> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christoph Hellwig <hch@infradead.org> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Dave Airlie <airlied@gmail.com> Cc: Dimitri Sivanich <dimitri.sivanich@hpe.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Ian Rogers <irogers@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Jérôme Glisse <jglisse@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: SeongJae Park <sj@kernel.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Uladzislau Rezki (Sony) <urezki@gmail.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-12 15:15:45 +00:00
ptent = ptep_get(pte);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
mm/madvise.c: free swp_entry in madvise_free When I test below piece of code with 12 processes(ie, 512M * 12 = 6G consume) on my (3G ram + 12 cpu + 8G swap, the madvise_free is siginficat slower (ie, 2x times) than madvise_dontneed. loop = 5; mmap(512M); while (loop--) { memset(512M); madvise(MADV_FREE or MADV_DONTNEED); } The reason is lots of swapin. 1) dontneed: 1,612 swapin 2) madvfree: 879,585 swapin If we find hinted pages were already swapped out when syscall is called, it's pointless to keep the swapped-out pages in pte. Instead, let's free the cold page because swapin is more expensive than (alloc page + zeroing). With this patch, it reduced swapin from 879,585 to 1,878 so elapsed time 1) dontneed: 6.10user 233.50system 0:50.44elapsed 2) madvfree: 6.03user 401.17system 1:30.67elapsed 2) madvfree + below patch: 6.70user 339.14system 1:04.45elapsed Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:06 +00:00
if (pte_none(ptent))
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
continue;
mm/madvise.c: free swp_entry in madvise_free When I test below piece of code with 12 processes(ie, 512M * 12 = 6G consume) on my (3G ram + 12 cpu + 8G swap, the madvise_free is siginficat slower (ie, 2x times) than madvise_dontneed. loop = 5; mmap(512M); while (loop--) { memset(512M); madvise(MADV_FREE or MADV_DONTNEED); } The reason is lots of swapin. 1) dontneed: 1,612 swapin 2) madvfree: 879,585 swapin If we find hinted pages were already swapped out when syscall is called, it's pointless to keep the swapped-out pages in pte. Instead, let's free the cold page because swapin is more expensive than (alloc page + zeroing). With this patch, it reduced swapin from 879,585 to 1,878 so elapsed time 1) dontneed: 6.10user 233.50system 0:50.44elapsed 2) madvfree: 6.03user 401.17system 1:30.67elapsed 2) madvfree + below patch: 6.70user 339.14system 1:04.45elapsed Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:06 +00:00
/*
* If the pte has swp_entry, just clear page table to
* prevent swap-in which is more expensive rather than
* (page allocation + zeroing).
*/
if (!pte_present(ptent)) {
swp_entry_t entry;
entry = pte_to_swp_entry(ptent);
if (!non_swap_entry(entry)) {
mm: swap: free_swap_and_cache_nr() as batched free_swap_and_cache() Now that we no longer have a convenient flag in the cluster to determine if a folio is large, free_swap_and_cache() will take a reference and lock a large folio much more often, which could lead to contention and (e.g.) failure to split large folios, etc. Let's solve that problem by batch freeing swap and cache with a new function, free_swap_and_cache_nr(), to free a contiguous range of swap entries together. This allows us to first drop a reference to each swap slot before we try to release the cache folio. This means we only try to release the folio once, only taking the reference and lock once - much better than the previous 512 times for the 2M THP case. Contiguous swap entries are gathered in zap_pte_range() and madvise_free_pte_range() in a similar way to how present ptes are already gathered in zap_pte_range(). While we are at it, let's simplify by converting the return type of both functions to void. The return value was used only by zap_pte_range() to print a bad pte, and was ignored by everyone else, so the extra reporting wasn't exactly guaranteed. We will still get the warning with most of the information from get_swap_device(). With the batch version, we wouldn't know which pte was bad anyway so could print the wrong one. [ryan.roberts@arm.com: fix a build warning on parisc] Link: https://lkml.kernel.org/r/20240409111840.3173122-1-ryan.roberts@arm.com Link: https://lkml.kernel.org/r/20240408183946.2991168-3-ryan.roberts@arm.com Signed-off-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Barry Song <v-songbaohua@oppo.com> Cc: Chris Li <chrisl@kernel.org> Cc: Gao Xiang <xiang@kernel.org> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Lance Yang <ioworker0@gmail.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-08 18:39:41 +00:00
max_nr = (end - addr) / PAGE_SIZE;
nr = swap_pte_batch(pte, max_nr, ptent);
nr_swap -= nr;
free_swap_and_cache_nr(entry, nr);
clear_not_present_full_ptes(mm, addr, pte, nr, tlb->fullmm);
} else if (is_hwpoison_entry(entry) ||
mm: make PTE_MARKER_SWAPIN_ERROR more general Patch series "add UFFDIO_POISON to simulate memory poisoning with UFFD", v4. This series adds a new userfaultfd feature, UFFDIO_POISON. See commit 4 for a detailed description of the feature. This patch (of 8): Future patches will reuse PTE_MARKER_SWAPIN_ERROR to implement UFFDIO_POISON, so make some various preparations for that: First, rename it to just PTE_MARKER_POISONED. The "SWAPIN" can be confusing since we're going to re-use it for something not really related to swap. This can be particularly confusing for things like hugetlbfs, which doesn't support swap whatsoever. Also rename some various helper functions. Next, fix pte marker copying for hugetlbfs. Previously, it would WARN on seeing a PTE_MARKER_SWAPIN_ERROR, since hugetlbfs doesn't support swap. But, since we're going to re-use it, we want it to go ahead and copy it just like non-hugetlbfs memory does today. Since the code to do this is more complicated now, pull it out into a helper which can be re-used in both places. While we're at it, also make it slightly more explicit in its handling of e.g. uffd wp markers. For non-hugetlbfs page faults, instead of returning VM_FAULT_SIGBUS for an error entry, return VM_FAULT_HWPOISON. For most cases this change doesn't matter, e.g. a userspace program would receive a SIGBUS either way. But for UFFDIO_POISON, this change will let KVM guests get an MCE out of the box, instead of giving a SIGBUS to the hypervisor and requiring it to somehow inject an MCE. Finally, for hugetlbfs faults, handle PTE_MARKER_POISONED, and return VM_FAULT_HWPOISON_LARGE in such cases. Note that this can't happen today because the lack of swap support means we'll never end up with such a PTE anyway, but this behavior will be needed once such entries *can* show up via UFFDIO_POISON. Link: https://lkml.kernel.org/r/20230707215540.2324998-1-axelrasmussen@google.com Link: https://lkml.kernel.org/r/20230707215540.2324998-2-axelrasmussen@google.com Signed-off-by: Axel Rasmussen <axelrasmussen@google.com> Acked-by: Peter Xu <peterx@redhat.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <brauner@kernel.org> Cc: David Hildenbrand <david@redhat.com> Cc: Gaosheng Cui <cuigaosheng1@huawei.com> Cc: Huang, Ying <ying.huang@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: James Houghton <jthoughton@google.com> Cc: Jan Alexander Steffens (heftig) <heftig@archlinux.org> Cc: Jiaqi Yan <jiaqiyan@google.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: Shuah Khan <shuah@kernel.org> Cc: Suleiman Souhlal <suleiman@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: T.J. Alumbaugh <talumbau@google.com> Cc: Yu Zhao <yuzhao@google.com> Cc: ZhangPeng <zhangpeng362@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-07-07 21:55:33 +00:00
is_poisoned_swp_entry(entry)) {
pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
}
mm/madvise.c: free swp_entry in madvise_free When I test below piece of code with 12 processes(ie, 512M * 12 = 6G consume) on my (3G ram + 12 cpu + 8G swap, the madvise_free is siginficat slower (ie, 2x times) than madvise_dontneed. loop = 5; mmap(512M); while (loop--) { memset(512M); madvise(MADV_FREE or MADV_DONTNEED); } The reason is lots of swapin. 1) dontneed: 1,612 swapin 2) madvfree: 879,585 swapin If we find hinted pages were already swapped out when syscall is called, it's pointless to keep the swapped-out pages in pte. Instead, let's free the cold page because swapin is more expensive than (alloc page + zeroing). With this patch, it reduced swapin from 879,585 to 1,878 so elapsed time 1) dontneed: 6.10user 233.50system 0:50.44elapsed 2) madvfree: 6.03user 401.17system 1:30.67elapsed 2) madvfree + below patch: 6.70user 339.14system 1:04.45elapsed Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:06 +00:00
continue;
}
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
folio = vm_normal_folio(vma, addr, ptent);
if (!folio || folio_is_zone_device(folio))
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
continue;
/*
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
* If we encounter a large folio, only split it if it is not
* fully mapped within the range we are operating on. Otherwise
* leave it as is so that it can be marked as lazyfree. If we
* fail to split a folio, leave it in place and advance to the
* next pte in the range.
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
*/
if (folio_test_large(folio)) {
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
bool any_young, any_dirty;
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
nr = madvise_folio_pte_batch(addr, end, folio, pte,
ptent, &any_young, &any_dirty);
if (nr < folio_nr_pages(folio)) {
int err;
if (folio_likely_mapped_shared(folio))
continue;
if (!folio_trylock(folio))
continue;
folio_get(folio);
arch_leave_lazy_mmu_mode();
pte_unmap_unlock(start_pte, ptl);
start_pte = NULL;
err = split_folio(folio);
folio_unlock(folio);
folio_put(folio);
pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
start_pte = pte;
if (!start_pte)
break;
arch_enter_lazy_mmu_mode();
if (!err)
nr = 0;
continue;
}
if (any_young)
ptent = pte_mkyoung(ptent);
if (any_dirty)
ptent = pte_mkdirty(ptent);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
}
if (folio_test_swapcache(folio) || folio_test_dirty(folio)) {
if (!folio_trylock(folio))
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
continue;
/*
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
* If we have a large folio at this point, we know it is
* fully mapped so if its mapcount is the same as its
* number of pages, it must be exclusive.
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
*/
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
if (folio_mapcount(folio) != folio_nr_pages(folio)) {
folio_unlock(folio);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
continue;
}
if (folio_test_swapcache(folio) &&
!folio_free_swap(folio)) {
folio_unlock(folio);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
continue;
}
folio_clear_dirty(folio);
folio_unlock(folio);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
}
if (pte_young(ptent) || pte_dirty(ptent)) {
mm/madvise: optimize lazyfreeing with mTHP in madvise_free This patch optimizes lazyfreeing with PTE-mapped mTHP[1] (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio splitting if the large folio is fully mapped within the target range. If a large folio is locked or shared, or if we fail to split it, we just leave it in place and advance to the next PTE in the range. But note that the behavior is changed; previously, any failure of this sort would cause the entire operation to give up. As large folios become more common, sticking to the old way could result in wasted opportunities. On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of the same size results in the following runtimes for madvise(MADV_FREE) in seconds (shorter is better): Folio Size | Old | New | Change ------------------------------------------ 4KiB | 0.590251 | 0.590259 | 0% 16KiB | 2.990447 | 0.185655 | -94% 32KiB | 2.547831 | 0.104870 | -95% 64KiB | 2.457796 | 0.052812 | -97% 128KiB | 2.281034 | 0.032777 | -99% 256KiB | 2.230387 | 0.017496 | -99% 512KiB | 2.189106 | 0.010781 | -99% 1024KiB | 2.183949 | 0.007753 | -99% 2048KiB | 0.002799 | 0.002804 | 0% [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com Link: https://lkml.kernel.org/r/20240418134435.6092-5-ioworker0@gmail.com Signed-off-by: Lance Yang <ioworker0@gmail.com> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Barry Song <21cnbao@gmail.com> Cc: Jeff Xie <xiehuan09@gmail.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Peter Xu <peterx@redhat.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yin Fengwei <fengwei.yin@intel.com> Cc: Zach O'Keefe <zokeefe@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-18 13:44:35 +00:00
clear_young_dirty_ptes(vma, addr, pte, nr, cydp_flags);
tlb_remove_tlb_entries(tlb, pte, nr, addr);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
}
folio_mark_lazyfree(folio);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
}
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
if (nr_swap)
mm/madvise.c: free swp_entry in madvise_free When I test below piece of code with 12 processes(ie, 512M * 12 = 6G consume) on my (3G ram + 12 cpu + 8G swap, the madvise_free is siginficat slower (ie, 2x times) than madvise_dontneed. loop = 5; mmap(512M); while (loop--) { memset(512M); madvise(MADV_FREE or MADV_DONTNEED); } The reason is lots of swapin. 1) dontneed: 1,612 swapin 2) madvfree: 879,585 swapin If we find hinted pages were already swapped out when syscall is called, it's pointless to keep the swapped-out pages in pte. Instead, let's free the cold page because swapin is more expensive than (alloc page + zeroing). With this patch, it reduced swapin from 879,585 to 1,878 so elapsed time 1) dontneed: 6.10user 233.50system 0:50.44elapsed 2) madvfree: 6.03user 401.17system 1:30.67elapsed 2) madvfree + below patch: 6.70user 339.14system 1:04.45elapsed Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Micay <danielmicay@gmail.com> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Jason Evans <je@fb.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Kirill A. Shutemov <kirill@shutemov.name> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rik van Riel <riel@redhat.com> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:55:06 +00:00
add_mm_counter(mm, MM_SWAPENTS, nr_swap);
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
if (start_pte) {
arch_leave_lazy_mmu_mode();
pte_unmap_unlock(start_pte, ptl);
}
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
cond_resched();
mm/madvise: clean up pte_offset_map_lock() scans Came here to make madvise's several pte_offset_map_lock() scans advance to next extent on failure, and remove superfluous pmd_trans_unstable() and pmd_none_or_trans_huge_or_clear_bad() calls. But also did some nearby cleanup. swapin_walk_pmd_entry(): don't name an address "index"; don't drop the lock after every pte, only when calling out to read_swap_cache_async(). madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(): prefer "start_pte" for pointer, orig_pte usually denotes a saved pte value; leave lazy MMU mode before unlocking; merge the success and failure paths after split_folio(). Link: https://lkml.kernel.org/r/cc4d9a88-9da6-362-50d9-6735c2b125c6@google.com Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Christophe Leroy <christophe.leroy@csgroup.eu> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Xu <peterx@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Qi Zheng <zhengqi.arch@bytedance.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Ryan Roberts <ryan.roberts@arm.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <song@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Will Deacon <will@kernel.org> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Cc: Zack Rusin <zackr@vmware.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-09 01:34:03 +00:00
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
return 0;
}
static const struct mm_walk_ops madvise_free_walk_ops = {
.pmd_entry = madvise_free_pte_range,
mm: enable page walking API to lock vmas during the walk walk_page_range() and friends often operate under write-locked mmap_lock. With introduction of vma locks, the vmas have to be locked as well during such walks to prevent concurrent page faults in these areas. Add an additional member to mm_walk_ops to indicate locking requirements for the walk. The change ensures that page walks which prevent concurrent page faults by write-locking mmap_lock, operate correctly after introduction of per-vma locks. With per-vma locks page faults can be handled under vma lock without taking mmap_lock at all, so write locking mmap_lock would not stop them. The change ensures vmas are properly locked during such walks. A sample issue this solves is do_mbind() performing queue_pages_range() to queue pages for migration. Without this change a concurrent page can be faulted into the area and be left out of migration. Link: https://lkml.kernel.org/r/20230804152724.3090321-2-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org> Suggested-by: Jann Horn <jannh@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Laurent Dufour <ldufour@linux.ibm.com> Cc: Liam Howlett <liam.howlett@oracle.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Michel Lespinasse <michel@lespinasse.org> Cc: Peter Xu <peterx@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-08-04 15:27:19 +00:00
.walk_lock = PGWALK_RDLOCK,
};
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
static int madvise_free_single_vma(struct vm_area_struct *vma,
unsigned long start_addr, unsigned long end_addr)
{
struct mm_struct *mm = vma->vm_mm;
struct mmu_notifier_range range;
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
struct mmu_gather tlb;
/* MADV_FREE works for only anon vma at the moment */
if (!vma_is_anonymous(vma))
return -EINVAL;
range.start = max(vma->vm_start, start_addr);
if (range.start >= vma->vm_end)
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
return -EINVAL;
range.end = min(vma->vm_end, end_addr);
if (range.end <= vma->vm_start)
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
return -EINVAL;
mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, mm,
mm/mmu_notifier: contextual information for event triggering invalidation CPU page table update can happens for many reasons, not only as a result of a syscall (munmap(), mprotect(), mremap(), madvise(), ...) but also as a result of kernel activities (memory compression, reclaim, migration, ...). Users of mmu notifier API track changes to the CPU page table and take specific action for them. While current API only provide range of virtual address affected by the change, not why the changes is happening. This patchset do the initial mechanical convertion of all the places that calls mmu_notifier_range_init to also provide the default MMU_NOTIFY_UNMAP event as well as the vma if it is know (most invalidation happens against a given vma). Passing down the vma allows the users of mmu notifier to inspect the new vma page protection. The MMU_NOTIFY_UNMAP is always the safe default as users of mmu notifier should assume that every for the range is going away when that event happens. A latter patch do convert mm call path to use a more appropriate events for each call. This is done as 2 patches so that no call site is forgotten especialy as it uses this following coccinelle patch: %<---------------------------------------------------------------------- @@ identifier I1, I2, I3, I4; @@ static inline void mmu_notifier_range_init(struct mmu_notifier_range *I1, +enum mmu_notifier_event event, +unsigned flags, +struct vm_area_struct *vma, struct mm_struct *I2, unsigned long I3, unsigned long I4) { ... } @@ @@ -#define mmu_notifier_range_init(range, mm, start, end) +#define mmu_notifier_range_init(range, event, flags, vma, mm, start, end) @@ expression E1, E3, E4; identifier I1; @@ <... mmu_notifier_range_init(E1, +MMU_NOTIFY_UNMAP, 0, I1, I1->vm_mm, E3, E4) ...> @@ expression E1, E2, E3, E4; identifier FN, VMA; @@ FN(..., struct vm_area_struct *VMA, ...) { <... mmu_notifier_range_init(E1, +MMU_NOTIFY_UNMAP, 0, VMA, E2, E3, E4) ...> } @@ expression E1, E2, E3, E4; identifier FN, VMA; @@ FN(...) { struct vm_area_struct *VMA; <... mmu_notifier_range_init(E1, +MMU_NOTIFY_UNMAP, 0, VMA, E2, E3, E4) ...> } @@ expression E1, E2, E3, E4; identifier FN; @@ FN(...) { <... mmu_notifier_range_init(E1, +MMU_NOTIFY_UNMAP, 0, NULL, E2, E3, E4) ...> } ---------------------------------------------------------------------->% Applied with: spatch --all-includes --sp-file mmu-notifier.spatch fs/proc/task_mmu.c --in-place spatch --sp-file mmu-notifier.spatch --dir kernel/events/ --in-place spatch --sp-file mmu-notifier.spatch --dir mm --in-place Link: http://lkml.kernel.org/r/20190326164747.24405-6-jglisse@redhat.com Signed-off-by: Jérôme Glisse <jglisse@redhat.com> Reviewed-by: Ralph Campbell <rcampbell@nvidia.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Cc: Christian König <christian.koenig@amd.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Jan Kara <jack@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Peter Xu <peterx@redhat.com> Cc: Felix Kuehling <Felix.Kuehling@amd.com> Cc: Jason Gunthorpe <jgg@mellanox.com> Cc: Ross Zwisler <zwisler@kernel.org> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krcmar <rkrcmar@redhat.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-14 00:20:49 +00:00
range.start, range.end);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
lru_add_drain();
tlb_gather_mmu(&tlb, mm);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
update_hiwater_rss(mm);
mmu_notifier_invalidate_range_start(&range);
tlb_start_vma(&tlb, vma);
walk_page_range(vma->vm_mm, range.start, range.end,
&madvise_free_walk_ops, &tlb);
tlb_end_vma(&tlb, vma);
mmu_notifier_invalidate_range_end(&range);
tlb_finish_mmu(&tlb);
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
return 0;
}
/*
* Application no longer needs these pages. If the pages are dirty,
* it's OK to just throw them away. The app will be more careful about
* data it wants to keep. Be sure to free swap resources too. The
madvise: use zap_page_range_single for madvise dontneed This series addresses the issue first reported in [1], and fully described in patch 2. Patches 1 and 2 address the user visible issue and are tagged for stable backports. While exploring solutions to this issue, related problems with mmu notification calls were discovered. This is addressed in the patch "hugetlb: remove duplicate mmu notifications:". Since there are no user visible effects, this third is not tagged for stable backports. Previous discussions suggested further cleanup by removing the routine zap_page_range. This is possible because zap_page_range_single is now exported, and all callers of zap_page_range pass ranges entirely within a single vma. This work will be done in a later patch so as not to distract from this bug fix. [1] https://lore.kernel.org/lkml/CAO4mrfdLMXsao9RF4fUE8-Wfde8xmjsKrTNMNC9wjUb6JudD0g@mail.gmail.com/ This patch (of 2): Expose the routine zap_page_range_single to zap a range within a single vma. The madvise routine madvise_dontneed_single_vma can use this routine as it explicitly operates on a single vma. Also, update the mmu notification range in zap_page_range_single to take hugetlb pmd sharing into account. This is required as MADV_DONTNEED supports hugetlb vmas. Link: https://lkml.kernel.org/r/20221114235507.294320-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20221114235507.294320-2-mike.kravetz@oracle.com Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Reported-by: Wei Chen <harperchen1110@gmail.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mina Almasry <almasrymina@google.com> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: Peter Xu <peterx@redhat.com> Cc: Rik van Riel <riel@surriel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-11-14 23:55:05 +00:00
* zap_page_range_single call sets things up for shrink_active_list to actually
* free these pages later if no one else has touched them in the meantime,
* although we could add these pages to a global reuse list for
* shrink_active_list to pick up before reclaiming other pages.
*
* NB: This interface discards data rather than pushes it out to swap,
* as some implementations do. This has performance implications for
* applications like large transactional databases which want to discard
* pages in anonymous maps after committing to backing store the data
* that was kept in them. There is no reason to write this data out to
* the swap area if the application is discarding it.
*
* An interface that causes the system to free clean pages and flush
* dirty pages is already available as msync(MS_INVALIDATE).
*/
static long madvise_dontneed_single_vma(struct vm_area_struct *vma,
unsigned long start, unsigned long end)
{
madvise: use zap_page_range_single for madvise dontneed This series addresses the issue first reported in [1], and fully described in patch 2. Patches 1 and 2 address the user visible issue and are tagged for stable backports. While exploring solutions to this issue, related problems with mmu notification calls were discovered. This is addressed in the patch "hugetlb: remove duplicate mmu notifications:". Since there are no user visible effects, this third is not tagged for stable backports. Previous discussions suggested further cleanup by removing the routine zap_page_range. This is possible because zap_page_range_single is now exported, and all callers of zap_page_range pass ranges entirely within a single vma. This work will be done in a later patch so as not to distract from this bug fix. [1] https://lore.kernel.org/lkml/CAO4mrfdLMXsao9RF4fUE8-Wfde8xmjsKrTNMNC9wjUb6JudD0g@mail.gmail.com/ This patch (of 2): Expose the routine zap_page_range_single to zap a range within a single vma. The madvise routine madvise_dontneed_single_vma can use this routine as it explicitly operates on a single vma. Also, update the mmu notification range in zap_page_range_single to take hugetlb pmd sharing into account. This is required as MADV_DONTNEED supports hugetlb vmas. Link: https://lkml.kernel.org/r/20221114235507.294320-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20221114235507.294320-2-mike.kravetz@oracle.com Fixes: 90e7e7f5ef3f ("mm: enable MADV_DONTNEED for hugetlb mappings") Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Reported-by: Wei Chen <harperchen1110@gmail.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mina Almasry <almasrymina@google.com> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: Peter Xu <peterx@redhat.com> Cc: Rik van Riel <riel@surriel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-11-14 23:55:05 +00:00
zap_page_range_single(vma, start, end - start, NULL);
return 0;
}
mm: enable MADV_DONTNEED for hugetlb mappings Patch series "Add hugetlb MADV_DONTNEED support", v3. Userfaultfd selftests for hugetlb does not perform UFFD_EVENT_REMAP testing. However, mremap support was recently added in commit 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma"). While attempting to enable mremap support in the test, it was discovered that the mremap test indirectly depends on MADV_DONTNEED. madvise does not allow MADV_DONTNEED for hugetlb mappings. However, that is primarily due to the check in can_madv_lru_vma(). By simply removing the check and adding huge page alignment, MADV_DONTNEED can be made to work for hugetlb mappings. Do note that there is no compelling use case for adding this support. This was discussed in the RFC [1]. However, adding support makes sense as it is fairly trivial and brings hugetlb functionality more in line with 'normal' memory. After enabling support, add selftest for MADV_DONTNEED as well as MADV_REMOVE. Then update userfaultfd selftest. If new functionality is accepted, then madvise man page will be updated to indicate hugetlb is supported. It will also be updated to clarify what happens to the passed length argument. This patch (of 3): MADV_DONTNEED is currently disabled for hugetlb mappings. This certainly makes sense in shared file mappings as the pagecache maintains a reference to the page and it will never be freed. However, it could be useful to unmap and free pages in private mappings. In addition, userfaultfd minor fault users may be able to simplify code by using MADV_DONTNEED. The primary thing preventing MADV_DONTNEED from working on hugetlb mappings is a check in can_madv_lru_vma(). To allow support for hugetlb mappings create and use a new routine madvise_dontneed_free_valid_vma() that allows hugetlb mappings in this specific case. For normal mappings, madvise requires the start address be PAGE aligned and rounds up length to the next multiple of PAGE_SIZE. Do similarly for hugetlb mappings: require start address be huge page size aligned and round up length to the next multiple of huge page size. Use the new madvise_dontneed_free_valid_vma routine to check alignment and round up length/end. zap_page_range requires this alignment for hugetlb vmas otherwise we will hit BUGs. Link: https://lkml.kernel.org/r/20220215002348.128823-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20220215002348.128823-2-mike.kravetz@oracle.com Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: David Hildenbrand <david@redhat.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mina Almasry <almasrymina@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Shuah Khan <skhan@linuxfoundation.org> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:13:18 +00:00
static bool madvise_dontneed_free_valid_vma(struct vm_area_struct *vma,
unsigned long start,
unsigned long *end,
int behavior)
{
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
if (!is_vm_hugetlb_page(vma)) {
unsigned int forbidden = VM_PFNMAP;
if (behavior != MADV_DONTNEED_LOCKED)
forbidden |= VM_LOCKED;
return !(vma->vm_flags & forbidden);
}
mm: enable MADV_DONTNEED for hugetlb mappings Patch series "Add hugetlb MADV_DONTNEED support", v3. Userfaultfd selftests for hugetlb does not perform UFFD_EVENT_REMAP testing. However, mremap support was recently added in commit 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma"). While attempting to enable mremap support in the test, it was discovered that the mremap test indirectly depends on MADV_DONTNEED. madvise does not allow MADV_DONTNEED for hugetlb mappings. However, that is primarily due to the check in can_madv_lru_vma(). By simply removing the check and adding huge page alignment, MADV_DONTNEED can be made to work for hugetlb mappings. Do note that there is no compelling use case for adding this support. This was discussed in the RFC [1]. However, adding support makes sense as it is fairly trivial and brings hugetlb functionality more in line with 'normal' memory. After enabling support, add selftest for MADV_DONTNEED as well as MADV_REMOVE. Then update userfaultfd selftest. If new functionality is accepted, then madvise man page will be updated to indicate hugetlb is supported. It will also be updated to clarify what happens to the passed length argument. This patch (of 3): MADV_DONTNEED is currently disabled for hugetlb mappings. This certainly makes sense in shared file mappings as the pagecache maintains a reference to the page and it will never be freed. However, it could be useful to unmap and free pages in private mappings. In addition, userfaultfd minor fault users may be able to simplify code by using MADV_DONTNEED. The primary thing preventing MADV_DONTNEED from working on hugetlb mappings is a check in can_madv_lru_vma(). To allow support for hugetlb mappings create and use a new routine madvise_dontneed_free_valid_vma() that allows hugetlb mappings in this specific case. For normal mappings, madvise requires the start address be PAGE aligned and rounds up length to the next multiple of PAGE_SIZE. Do similarly for hugetlb mappings: require start address be huge page size aligned and round up length to the next multiple of huge page size. Use the new madvise_dontneed_free_valid_vma routine to check alignment and round up length/end. zap_page_range requires this alignment for hugetlb vmas otherwise we will hit BUGs. Link: https://lkml.kernel.org/r/20220215002348.128823-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20220215002348.128823-2-mike.kravetz@oracle.com Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: David Hildenbrand <david@redhat.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mina Almasry <almasrymina@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Shuah Khan <skhan@linuxfoundation.org> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:13:18 +00:00
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
if (behavior != MADV_DONTNEED && behavior != MADV_DONTNEED_LOCKED)
mm: enable MADV_DONTNEED for hugetlb mappings Patch series "Add hugetlb MADV_DONTNEED support", v3. Userfaultfd selftests for hugetlb does not perform UFFD_EVENT_REMAP testing. However, mremap support was recently added in commit 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma"). While attempting to enable mremap support in the test, it was discovered that the mremap test indirectly depends on MADV_DONTNEED. madvise does not allow MADV_DONTNEED for hugetlb mappings. However, that is primarily due to the check in can_madv_lru_vma(). By simply removing the check and adding huge page alignment, MADV_DONTNEED can be made to work for hugetlb mappings. Do note that there is no compelling use case for adding this support. This was discussed in the RFC [1]. However, adding support makes sense as it is fairly trivial and brings hugetlb functionality more in line with 'normal' memory. After enabling support, add selftest for MADV_DONTNEED as well as MADV_REMOVE. Then update userfaultfd selftest. If new functionality is accepted, then madvise man page will be updated to indicate hugetlb is supported. It will also be updated to clarify what happens to the passed length argument. This patch (of 3): MADV_DONTNEED is currently disabled for hugetlb mappings. This certainly makes sense in shared file mappings as the pagecache maintains a reference to the page and it will never be freed. However, it could be useful to unmap and free pages in private mappings. In addition, userfaultfd minor fault users may be able to simplify code by using MADV_DONTNEED. The primary thing preventing MADV_DONTNEED from working on hugetlb mappings is a check in can_madv_lru_vma(). To allow support for hugetlb mappings create and use a new routine madvise_dontneed_free_valid_vma() that allows hugetlb mappings in this specific case. For normal mappings, madvise requires the start address be PAGE aligned and rounds up length to the next multiple of PAGE_SIZE. Do similarly for hugetlb mappings: require start address be huge page size aligned and round up length to the next multiple of huge page size. Use the new madvise_dontneed_free_valid_vma routine to check alignment and round up length/end. zap_page_range requires this alignment for hugetlb vmas otherwise we will hit BUGs. Link: https://lkml.kernel.org/r/20220215002348.128823-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20220215002348.128823-2-mike.kravetz@oracle.com Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: David Hildenbrand <david@redhat.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mina Almasry <almasrymina@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Shuah Khan <skhan@linuxfoundation.org> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:13:18 +00:00
return false;
if (start & ~huge_page_mask(hstate_vma(vma)))
return false;
/*
* Madvise callers expect the length to be rounded up to PAGE_SIZE
* boundaries, and may be unaware that this VMA uses huge pages.
* Avoid unexpected data loss by rounding down the number of
* huge pages freed.
*/
*end = ALIGN_DOWN(*end, huge_page_size(hstate_vma(vma)));
mm: enable MADV_DONTNEED for hugetlb mappings Patch series "Add hugetlb MADV_DONTNEED support", v3. Userfaultfd selftests for hugetlb does not perform UFFD_EVENT_REMAP testing. However, mremap support was recently added in commit 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma"). While attempting to enable mremap support in the test, it was discovered that the mremap test indirectly depends on MADV_DONTNEED. madvise does not allow MADV_DONTNEED for hugetlb mappings. However, that is primarily due to the check in can_madv_lru_vma(). By simply removing the check and adding huge page alignment, MADV_DONTNEED can be made to work for hugetlb mappings. Do note that there is no compelling use case for adding this support. This was discussed in the RFC [1]. However, adding support makes sense as it is fairly trivial and brings hugetlb functionality more in line with 'normal' memory. After enabling support, add selftest for MADV_DONTNEED as well as MADV_REMOVE. Then update userfaultfd selftest. If new functionality is accepted, then madvise man page will be updated to indicate hugetlb is supported. It will also be updated to clarify what happens to the passed length argument. This patch (of 3): MADV_DONTNEED is currently disabled for hugetlb mappings. This certainly makes sense in shared file mappings as the pagecache maintains a reference to the page and it will never be freed. However, it could be useful to unmap and free pages in private mappings. In addition, userfaultfd minor fault users may be able to simplify code by using MADV_DONTNEED. The primary thing preventing MADV_DONTNEED from working on hugetlb mappings is a check in can_madv_lru_vma(). To allow support for hugetlb mappings create and use a new routine madvise_dontneed_free_valid_vma() that allows hugetlb mappings in this specific case. For normal mappings, madvise requires the start address be PAGE aligned and rounds up length to the next multiple of PAGE_SIZE. Do similarly for hugetlb mappings: require start address be huge page size aligned and round up length to the next multiple of huge page size. Use the new madvise_dontneed_free_valid_vma routine to check alignment and round up length/end. zap_page_range requires this alignment for hugetlb vmas otherwise we will hit BUGs. Link: https://lkml.kernel.org/r/20220215002348.128823-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20220215002348.128823-2-mike.kravetz@oracle.com Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: David Hildenbrand <david@redhat.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mina Almasry <almasrymina@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Shuah Khan <skhan@linuxfoundation.org> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:13:18 +00:00
return true;
}
static long madvise_dontneed_free(struct vm_area_struct *vma,
struct vm_area_struct **prev,
unsigned long start, unsigned long end,
int behavior)
{
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
struct mm_struct *mm = vma->vm_mm;
*prev = vma;
mm: enable MADV_DONTNEED for hugetlb mappings Patch series "Add hugetlb MADV_DONTNEED support", v3. Userfaultfd selftests for hugetlb does not perform UFFD_EVENT_REMAP testing. However, mremap support was recently added in commit 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma"). While attempting to enable mremap support in the test, it was discovered that the mremap test indirectly depends on MADV_DONTNEED. madvise does not allow MADV_DONTNEED for hugetlb mappings. However, that is primarily due to the check in can_madv_lru_vma(). By simply removing the check and adding huge page alignment, MADV_DONTNEED can be made to work for hugetlb mappings. Do note that there is no compelling use case for adding this support. This was discussed in the RFC [1]. However, adding support makes sense as it is fairly trivial and brings hugetlb functionality more in line with 'normal' memory. After enabling support, add selftest for MADV_DONTNEED as well as MADV_REMOVE. Then update userfaultfd selftest. If new functionality is accepted, then madvise man page will be updated to indicate hugetlb is supported. It will also be updated to clarify what happens to the passed length argument. This patch (of 3): MADV_DONTNEED is currently disabled for hugetlb mappings. This certainly makes sense in shared file mappings as the pagecache maintains a reference to the page and it will never be freed. However, it could be useful to unmap and free pages in private mappings. In addition, userfaultfd minor fault users may be able to simplify code by using MADV_DONTNEED. The primary thing preventing MADV_DONTNEED from working on hugetlb mappings is a check in can_madv_lru_vma(). To allow support for hugetlb mappings create and use a new routine madvise_dontneed_free_valid_vma() that allows hugetlb mappings in this specific case. For normal mappings, madvise requires the start address be PAGE aligned and rounds up length to the next multiple of PAGE_SIZE. Do similarly for hugetlb mappings: require start address be huge page size aligned and round up length to the next multiple of huge page size. Use the new madvise_dontneed_free_valid_vma routine to check alignment and round up length/end. zap_page_range requires this alignment for hugetlb vmas otherwise we will hit BUGs. Link: https://lkml.kernel.org/r/20220215002348.128823-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20220215002348.128823-2-mike.kravetz@oracle.com Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: David Hildenbrand <david@redhat.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mina Almasry <almasrymina@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Shuah Khan <skhan@linuxfoundation.org> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:13:18 +00:00
if (!madvise_dontneed_free_valid_vma(vma, start, &end, behavior))
return -EINVAL;
if (start == end)
return 0;
if (!userfaultfd_remove(vma, start, end)) {
*prev = NULL; /* mmap_lock has been dropped, prev is stale */
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_lock(mm);
vma = vma_lookup(mm, start);
if (!vma)
return -ENOMEM;
mm: enable MADV_DONTNEED for hugetlb mappings Patch series "Add hugetlb MADV_DONTNEED support", v3. Userfaultfd selftests for hugetlb does not perform UFFD_EVENT_REMAP testing. However, mremap support was recently added in commit 550a7d60bd5e ("mm, hugepages: add mremap() support for hugepage backed vma"). While attempting to enable mremap support in the test, it was discovered that the mremap test indirectly depends on MADV_DONTNEED. madvise does not allow MADV_DONTNEED for hugetlb mappings. However, that is primarily due to the check in can_madv_lru_vma(). By simply removing the check and adding huge page alignment, MADV_DONTNEED can be made to work for hugetlb mappings. Do note that there is no compelling use case for adding this support. This was discussed in the RFC [1]. However, adding support makes sense as it is fairly trivial and brings hugetlb functionality more in line with 'normal' memory. After enabling support, add selftest for MADV_DONTNEED as well as MADV_REMOVE. Then update userfaultfd selftest. If new functionality is accepted, then madvise man page will be updated to indicate hugetlb is supported. It will also be updated to clarify what happens to the passed length argument. This patch (of 3): MADV_DONTNEED is currently disabled for hugetlb mappings. This certainly makes sense in shared file mappings as the pagecache maintains a reference to the page and it will never be freed. However, it could be useful to unmap and free pages in private mappings. In addition, userfaultfd minor fault users may be able to simplify code by using MADV_DONTNEED. The primary thing preventing MADV_DONTNEED from working on hugetlb mappings is a check in can_madv_lru_vma(). To allow support for hugetlb mappings create and use a new routine madvise_dontneed_free_valid_vma() that allows hugetlb mappings in this specific case. For normal mappings, madvise requires the start address be PAGE aligned and rounds up length to the next multiple of PAGE_SIZE. Do similarly for hugetlb mappings: require start address be huge page size aligned and round up length to the next multiple of huge page size. Use the new madvise_dontneed_free_valid_vma routine to check alignment and round up length/end. zap_page_range requires this alignment for hugetlb vmas otherwise we will hit BUGs. Link: https://lkml.kernel.org/r/20220215002348.128823-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20220215002348.128823-2-mike.kravetz@oracle.com Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: David Hildenbrand <david@redhat.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mina Almasry <almasrymina@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Xu <peterx@redhat.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Shuah Khan <skhan@linuxfoundation.org> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:13:18 +00:00
/*
* Potential end adjustment for hugetlb vma is OK as
* the check below keeps end within vma.
*/
if (!madvise_dontneed_free_valid_vma(vma, start, &end,
behavior))
return -EINVAL;
if (end > vma->vm_end) {
/*
* Don't fail if end > vma->vm_end. If the old
* vma was split while the mmap_lock was
* released the effect of the concurrent
* operation may not cause madvise() to
* have an undefined result. There may be an
* adjacent next vma that we'll walk
* next. userfaultfd_remove() will generate an
* UFFD_EVENT_REMOVE repetition on the
* end-vma->vm_end range, but the manager can
* handle a repetition fine.
*/
end = vma->vm_end;
}
VM_WARN_ON(start >= end);
}
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
if (behavior == MADV_DONTNEED || behavior == MADV_DONTNEED_LOCKED)
return madvise_dontneed_single_vma(vma, start, end);
else if (behavior == MADV_FREE)
return madvise_free_single_vma(vma, start, end);
else
return -EINVAL;
}
mm/madvise: don't perform madvise VMA walk for MADV_POPULATE_(READ|WRITE) We changed faultin_page_range() to no longer consume a VMA, because faultin_page_range() might internally release the mm lock to lookup the VMA again -- required to cleanly handle VM_FAULT_RETRY. But independent of that, __get_user_pages() will always lookup the VMA itself. Now that we let __get_user_pages() just handle VMA checks in a way that is suitable for MADV_POPULATE_(READ|WRITE), the VMA walk in madvise() is just overhead. So let's just call madvise_populate() on the full range instead. There is one change in behavior: madvise_walk_vmas() would skip any VMA holes, and if everything succeeded, it would return -ENOMEM after processing all VMAs. However, for MADV_POPULATE_(READ|WRITE) it's unlikely for the caller to notice any difference: -ENOMEM might either indicate that there were VMA holes or that populating page tables failed because there was not enough memory. So it's unlikely that user space will notice the difference, and that special handling likely only makes sense for some other madvise() actions. Further, we'd already fail with -ENOMEM early in the past if looking up the VMA after dropping the MM lock failed because of concurrent VMA modifications. So let's just keep it simple and avoid the madvise VMA walk, and consistently fail early if we find a VMA hole. Link: https://lkml.kernel.org/r/20240314161300.382526-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Darrick J. Wong <djwong@kernel.org> Cc: Hugh Dickins <hughd@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-14 16:13:00 +00:00
static long madvise_populate(struct mm_struct *mm, unsigned long start,
unsigned long end, int behavior)
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
{
const bool write = behavior == MADV_POPULATE_WRITE;
int locked = 1;
long pages;
while (start < end) {
/* Populate (prefault) page tables readable/writable. */
mm/madvise: make MADV_POPULATE_(READ|WRITE) handle VM_FAULT_RETRY properly Darrick reports that in some cases where pread() would fail with -EIO and mmap()+access would generate a SIGBUS signal, MADV_POPULATE_READ / MADV_POPULATE_WRITE will keep retrying forever and not fail with -EFAULT. While the madvise() call can be interrupted by a signal, this is not the desired behavior. MADV_POPULATE_READ / MADV_POPULATE_WRITE should behave like page faults in that case: fail and not retry forever. A reproducer can be found at [1]. The reason is that __get_user_pages(), as called by faultin_vma_page_range(), will not handle VM_FAULT_RETRY in a proper way: it will simply return 0 when VM_FAULT_RETRY happened, making madvise_populate()->faultin_vma_page_range() retry again and again, never setting FOLL_TRIED->FAULT_FLAG_TRIED for __get_user_pages(). __get_user_pages_locked() does what we want, but duplicating that logic in faultin_vma_page_range() feels wrong. So let's use __get_user_pages_locked() instead, that will detect VM_FAULT_RETRY and set FOLL_TRIED when retrying, making the fault handler return VM_FAULT_SIGBUS (VM_FAULT_ERROR) at some point, propagating -EFAULT from faultin_page() to __get_user_pages(), all the way to madvise_populate(). But, there is an issue: __get_user_pages_locked() will end up re-taking the MM lock and then __get_user_pages() will do another VMA lookup. In the meantime, the VMA layout could have changed and we'd fail with different error codes than we'd want to. As __get_user_pages() will currently do a new VMA lookup either way, let it do the VMA handling in a different way, controlled by a new FOLL_MADV_POPULATE flag, effectively moving these checks from madvise_populate() + faultin_page_range() in there. With this change, Darricks reproducer properly fails with -EFAULT, as documented for MADV_POPULATE_READ / MADV_POPULATE_WRITE. [1] https://lore.kernel.org/all/20240313171936.GN1927156@frogsfrogsfrogs/ Link: https://lkml.kernel.org/r/20240314161300.382526-1-david@redhat.com Link: https://lkml.kernel.org/r/20240314161300.382526-2-david@redhat.com Fixes: 4ca9b3859dac ("mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables") Signed-off-by: David Hildenbrand <david@redhat.com> Reported-by: Darrick J. Wong <djwong@kernel.org> Closes: https://lore.kernel.org/all/20240311223815.GW1927156@frogsfrogsfrogs/ Cc: Darrick J. Wong <djwong@kernel.org> Cc: Hugh Dickins <hughd@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-14 16:12:59 +00:00
pages = faultin_page_range(mm, start, end, write, &locked);
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
if (!locked) {
mmap_read_lock(mm);
locked = 1;
}
if (pages < 0) {
switch (pages) {
case -EINTR:
return -EINTR;
mm/madvise: report SIGBUS as -EFAULT for MADV_POPULATE_(READ|WRITE) Doing some extended tests and polishing the man page update for MADV_POPULATE_(READ|WRITE), I realized that we end up converting also SIGBUS (via -EFAULT) to -EINVAL, making it look like yet another madvise() user error. We want to report only problematic mappings and permission problems that the user could have know as -EINVAL. Let's not convert -EFAULT arising due to SIGBUS (or SIGSEGV) to -EINVAL, but instead indicate -EFAULT to user space. While we could also convert it to -ENOMEM, using -EFAULT looks more helpful when user space might want to troubleshoot what's going wrong: MADV_POPULATE_(READ|WRITE) is not part of an final Linux release and we can still adjust the behavior. Link: https://lkml.kernel.org/r/20210726154932.102880-1-david@redhat.com Fixes: 4ca9b3859dac ("mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables") Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-08-13 23:54:37 +00:00
case -EINVAL: /* Incompatible mappings / permissions. */
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
return -EINVAL;
case -EHWPOISON:
return -EHWPOISON;
mm/madvise: report SIGBUS as -EFAULT for MADV_POPULATE_(READ|WRITE) Doing some extended tests and polishing the man page update for MADV_POPULATE_(READ|WRITE), I realized that we end up converting also SIGBUS (via -EFAULT) to -EINVAL, making it look like yet another madvise() user error. We want to report only problematic mappings and permission problems that the user could have know as -EINVAL. Let's not convert -EFAULT arising due to SIGBUS (or SIGSEGV) to -EINVAL, but instead indicate -EFAULT to user space. While we could also convert it to -ENOMEM, using -EFAULT looks more helpful when user space might want to troubleshoot what's going wrong: MADV_POPULATE_(READ|WRITE) is not part of an final Linux release and we can still adjust the behavior. Link: https://lkml.kernel.org/r/20210726154932.102880-1-david@redhat.com Fixes: 4ca9b3859dac ("mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables") Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-08-13 23:54:37 +00:00
case -EFAULT: /* VM_FAULT_SIGBUS or VM_FAULT_SIGSEGV */
return -EFAULT;
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
default:
pr_warn_once("%s: unhandled return value: %ld\n",
__func__, pages);
fallthrough;
mm/madvise: make MADV_POPULATE_(READ|WRITE) handle VM_FAULT_RETRY properly Darrick reports that in some cases where pread() would fail with -EIO and mmap()+access would generate a SIGBUS signal, MADV_POPULATE_READ / MADV_POPULATE_WRITE will keep retrying forever and not fail with -EFAULT. While the madvise() call can be interrupted by a signal, this is not the desired behavior. MADV_POPULATE_READ / MADV_POPULATE_WRITE should behave like page faults in that case: fail and not retry forever. A reproducer can be found at [1]. The reason is that __get_user_pages(), as called by faultin_vma_page_range(), will not handle VM_FAULT_RETRY in a proper way: it will simply return 0 when VM_FAULT_RETRY happened, making madvise_populate()->faultin_vma_page_range() retry again and again, never setting FOLL_TRIED->FAULT_FLAG_TRIED for __get_user_pages(). __get_user_pages_locked() does what we want, but duplicating that logic in faultin_vma_page_range() feels wrong. So let's use __get_user_pages_locked() instead, that will detect VM_FAULT_RETRY and set FOLL_TRIED when retrying, making the fault handler return VM_FAULT_SIGBUS (VM_FAULT_ERROR) at some point, propagating -EFAULT from faultin_page() to __get_user_pages(), all the way to madvise_populate(). But, there is an issue: __get_user_pages_locked() will end up re-taking the MM lock and then __get_user_pages() will do another VMA lookup. In the meantime, the VMA layout could have changed and we'd fail with different error codes than we'd want to. As __get_user_pages() will currently do a new VMA lookup either way, let it do the VMA handling in a different way, controlled by a new FOLL_MADV_POPULATE flag, effectively moving these checks from madvise_populate() + faultin_page_range() in there. With this change, Darricks reproducer properly fails with -EFAULT, as documented for MADV_POPULATE_READ / MADV_POPULATE_WRITE. [1] https://lore.kernel.org/all/20240313171936.GN1927156@frogsfrogsfrogs/ Link: https://lkml.kernel.org/r/20240314161300.382526-1-david@redhat.com Link: https://lkml.kernel.org/r/20240314161300.382526-2-david@redhat.com Fixes: 4ca9b3859dac ("mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables") Signed-off-by: David Hildenbrand <david@redhat.com> Reported-by: Darrick J. Wong <djwong@kernel.org> Closes: https://lore.kernel.org/all/20240311223815.GW1927156@frogsfrogsfrogs/ Cc: Darrick J. Wong <djwong@kernel.org> Cc: Hugh Dickins <hughd@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-14 16:12:59 +00:00
case -ENOMEM: /* No VMA or out of memory. */
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
return -ENOMEM;
}
}
start += pages * PAGE_SIZE;
}
return 0;
}
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
/*
* Application wants to free up the pages and associated backing store.
* This is effectively punching a hole into the middle of a file.
*/
static long madvise_remove(struct vm_area_struct *vma,
struct vm_area_struct **prev,
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
unsigned long start, unsigned long end)
{
loff_t offset;
int error;
struct file *f;
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
struct mm_struct *mm = vma->vm_mm;
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
*prev = NULL; /* tell sys_madvise we drop mmap_lock */
if (vma->vm_flags & VM_LOCKED)
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
return -EINVAL;
f = vma->vm_file;
if (!f || !f->f_mapping || !f->f_mapping->host) {
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
return -EINVAL;
}
mm: drop the assumption that VM_SHARED always implies writable Patch series "permit write-sealed memfd read-only shared mappings", v4. The man page for fcntl() describing memfd file seals states the following about F_SEAL_WRITE:- Furthermore, trying to create new shared, writable memory-mappings via mmap(2) will also fail with EPERM. With emphasis on 'writable'. In turns out in fact that currently the kernel simply disallows all new shared memory mappings for a memfd with F_SEAL_WRITE applied, rendering this documentation inaccurate. This matters because users are therefore unable to obtain a shared mapping to a memfd after write sealing altogether, which limits their usefulness. This was reported in the discussion thread [1] originating from a bug report [2]. This is a product of both using the struct address_space->i_mmap_writable atomic counter to determine whether writing may be permitted, and the kernel adjusting this counter when any VM_SHARED mapping is performed and more generally implicitly assuming VM_SHARED implies writable. It seems sensible that we should only update this mapping if VM_MAYWRITE is specified, i.e. whether it is possible that this mapping could at any point be written to. If we do so then all we need to do to permit write seals to function as documented is to clear VM_MAYWRITE when mapping read-only. It turns out this functionality already exists for F_SEAL_FUTURE_WRITE - we can therefore simply adapt this logic to do the same for F_SEAL_WRITE. We then hit a chicken and egg situation in mmap_region() where the check for VM_MAYWRITE occurs before we are able to clear this flag. To work around this, perform this check after we invoke call_mmap(), with careful consideration of error paths. Thanks to Andy Lutomirski for the suggestion! [1]:https://lore.kernel.org/all/20230324133646.16101dfa666f253c4715d965@linux-foundation.org/ [2]:https://bugzilla.kernel.org/show_bug.cgi?id=217238 This patch (of 3): There is a general assumption that VMAs with the VM_SHARED flag set are writable. If the VM_MAYWRITE flag is not set, then this is simply not the case. Update those checks which affect the struct address_space->i_mmap_writable field to explicitly test for this by introducing [vma_]is_shared_maywrite() helper functions. This remains entirely conservative, as the lack of VM_MAYWRITE guarantees that the VMA cannot be written to. Link: https://lkml.kernel.org/r/cover.1697116581.git.lstoakes@gmail.com Link: https://lkml.kernel.org/r/d978aefefa83ec42d18dfa964ad180dbcde34795.1697116581.git.lstoakes@gmail.com Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com> Suggested-by: Andy Lutomirski <luto@kernel.org> Reviewed-by: Jan Kara <jack@suse.cz> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Christian Brauner <brauner@kernel.org> Cc: Hugh Dickins <hughd@google.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <muchun.song@linux.dev> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-12 17:04:28 +00:00
if (!vma_is_shared_maywrite(vma))
return -EACCES;
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
offset = (loff_t)(start - vma->vm_start)
+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
/*
* Filesystem's fallocate may need to take i_rwsem. We need to
* explicitly grab a reference because the vma (and hence the
* vma's reference to the file) can go away as soon as we drop
* mmap_lock.
*/
get_file(f);
if (userfaultfd_remove(vma, start, end)) {
/* mmap_lock was not released by userfaultfd_remove() */
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_unlock(mm);
}
error = vfs_fallocate(f,
FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
offset, end - start);
fput(f);
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_lock(mm);
return error;
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
}
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
/*
* Apply an madvise behavior to a region of a vma. madvise_update_vma
* will handle splitting a vm area into separate areas, each area with its own
* behavior.
*/
static int madvise_vma_behavior(struct vm_area_struct *vma,
struct vm_area_struct **prev,
unsigned long start, unsigned long end,
unsigned long behavior)
{
int error;
mm: fix use-after-free when anon vma name is used after vma is freed When adjacent vmas are being merged it can result in the vma that was originally passed to madvise_update_vma being destroyed. In the current implementation, the name parameter passed to madvise_update_vma points directly to vma->anon_name and it is used after the call to vma_merge. In the cases when vma_merge merges the original vma and destroys it, this might result in UAF. For that the original vma would have to hold the anon_vma_name with the last reference. The following vma would need to contain a different anon_vma_name object with the same string. Such scenario is shown below: madvise_vma_behavior(vma) madvise_update_vma(vma, ..., anon_name == vma->anon_name) vma_merge(vma) __vma_adjust(vma) <-- merges vma with adjacent one vm_area_free(vma) <-- frees the original vma replace_vma_anon_name(anon_name) <-- UAF of vma->anon_name Fix this by raising the name refcount and stabilizing it. Link: https://lkml.kernel.org/r/20220224231834.1481408-3-surenb@google.com Link: https://lkml.kernel.org/r/20220223153613.835563-3-surenb@google.com Fixes: 9a10064f5625 ("mm: add a field to store names for private anonymous memory") Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reported-by: syzbot+aa7b3d4b35f9dc46a366@syzkaller.appspotmail.com Acked-by: Michal Hocko <mhocko@suse.com> Cc: Alexey Gladkov <legion@kernel.org> Cc: Chris Hyser <chris.hyser@oracle.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Collingbourne <pcc@google.com> Cc: Sasha Levin <sashal@kernel.org> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xiaofeng Cao <caoxiaofeng@yulong.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-05 04:28:58 +00:00
struct anon_vma_name *anon_name;
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
unsigned long new_flags = vma->vm_flags;
switch (behavior) {
case MADV_REMOVE:
return madvise_remove(vma, prev, start, end);
case MADV_WILLNEED:
return madvise_willneed(vma, prev, start, end);
case MADV_COLD:
return madvise_cold(vma, prev, start, end);
case MADV_PAGEOUT:
return madvise_pageout(vma, prev, start, end);
case MADV_FREE:
case MADV_DONTNEED:
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
case MADV_DONTNEED_LOCKED:
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return madvise_dontneed_free(vma, prev, start, end, behavior);
case MADV_NORMAL:
new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
break;
case MADV_SEQUENTIAL:
new_flags = (new_flags & ~VM_RAND_READ) | VM_SEQ_READ;
break;
case MADV_RANDOM:
new_flags = (new_flags & ~VM_SEQ_READ) | VM_RAND_READ;
break;
case MADV_DONTFORK:
new_flags |= VM_DONTCOPY;
break;
case MADV_DOFORK:
if (vma->vm_flags & VM_IO)
return -EINVAL;
new_flags &= ~VM_DONTCOPY;
break;
case MADV_WIPEONFORK:
/* MADV_WIPEONFORK is only supported on anonymous memory. */
if (vma->vm_file || vma->vm_flags & VM_SHARED)
return -EINVAL;
new_flags |= VM_WIPEONFORK;
break;
case MADV_KEEPONFORK:
mm: add MAP_DROPPABLE for designating always lazily freeable mappings The vDSO getrandom() implementation works with a buffer allocated with a new system call that has certain requirements: - It shouldn't be written to core dumps. * Easy: VM_DONTDUMP. - It should be zeroed on fork. * Easy: VM_WIPEONFORK. - It shouldn't be written to swap. * Uh-oh: mlock is rlimited. * Uh-oh: mlock isn't inherited by forks. - It shouldn't reserve actual memory, but it also shouldn't crash when page faulting in memory if none is available * Uh-oh: VM_NORESERVE means segfaults. It turns out that the vDSO getrandom() function has three really nice characteristics that we can exploit to solve this problem: 1) Due to being wiped during fork(), the vDSO code is already robust to having the contents of the pages it reads zeroed out midway through the function's execution. 2) In the absolute worst case of whatever contingency we're coding for, we have the option to fallback to the getrandom() syscall, and everything is fine. 3) The buffers the function uses are only ever useful for a maximum of 60 seconds -- a sort of cache, rather than a long term allocation. These characteristics mean that we can introduce VM_DROPPABLE, which has the following semantics: a) It never is written out to swap. b) Under memory pressure, mm can just drop the pages (so that they're zero when read back again). c) It is inherited by fork. d) It doesn't count against the mlock budget, since nothing is locked. e) If there's not enough memory to service a page fault, it's not fatal, and no signal is sent. This way, allocations used by vDSO getrandom() can use: VM_DROPPABLE | VM_DONTDUMP | VM_WIPEONFORK | VM_NORESERVE And there will be no problem with OOMing, crashing on overcommitment, using memory when not in use, not wiping on fork(), coredumps, or writing out to swap. In order to let vDSO getrandom() use this, expose these via mmap(2) as MAP_DROPPABLE. Note that this involves removing the MADV_FREE special case from sort_folio(), which according to Yu Zhao is unnecessary and will simply result in an extra call to shrink_folio_list() in the worst case. The chunk removed reenables the swapbacked flag, which we don't want for VM_DROPPABLE, and we can't conditionalize it here because there isn't a vma reference available. Finally, the provided self test ensures that this is working as desired. Cc: linux-mm@kvack.org Acked-by: David Hildenbrand <david@redhat.com> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
2022-12-08 16:55:04 +00:00
if (vma->vm_flags & VM_DROPPABLE)
return -EINVAL;
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
new_flags &= ~VM_WIPEONFORK;
break;
case MADV_DONTDUMP:
new_flags |= VM_DONTDUMP;
break;
case MADV_DODUMP:
mm: add MAP_DROPPABLE for designating always lazily freeable mappings The vDSO getrandom() implementation works with a buffer allocated with a new system call that has certain requirements: - It shouldn't be written to core dumps. * Easy: VM_DONTDUMP. - It should be zeroed on fork. * Easy: VM_WIPEONFORK. - It shouldn't be written to swap. * Uh-oh: mlock is rlimited. * Uh-oh: mlock isn't inherited by forks. - It shouldn't reserve actual memory, but it also shouldn't crash when page faulting in memory if none is available * Uh-oh: VM_NORESERVE means segfaults. It turns out that the vDSO getrandom() function has three really nice characteristics that we can exploit to solve this problem: 1) Due to being wiped during fork(), the vDSO code is already robust to having the contents of the pages it reads zeroed out midway through the function's execution. 2) In the absolute worst case of whatever contingency we're coding for, we have the option to fallback to the getrandom() syscall, and everything is fine. 3) The buffers the function uses are only ever useful for a maximum of 60 seconds -- a sort of cache, rather than a long term allocation. These characteristics mean that we can introduce VM_DROPPABLE, which has the following semantics: a) It never is written out to swap. b) Under memory pressure, mm can just drop the pages (so that they're zero when read back again). c) It is inherited by fork. d) It doesn't count against the mlock budget, since nothing is locked. e) If there's not enough memory to service a page fault, it's not fatal, and no signal is sent. This way, allocations used by vDSO getrandom() can use: VM_DROPPABLE | VM_DONTDUMP | VM_WIPEONFORK | VM_NORESERVE And there will be no problem with OOMing, crashing on overcommitment, using memory when not in use, not wiping on fork(), coredumps, or writing out to swap. In order to let vDSO getrandom() use this, expose these via mmap(2) as MAP_DROPPABLE. Note that this involves removing the MADV_FREE special case from sort_folio(), which according to Yu Zhao is unnecessary and will simply result in an extra call to shrink_folio_list() in the worst case. The chunk removed reenables the swapbacked flag, which we don't want for VM_DROPPABLE, and we can't conditionalize it here because there isn't a vma reference available. Finally, the provided self test ensures that this is working as desired. Cc: linux-mm@kvack.org Acked-by: David Hildenbrand <david@redhat.com> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
2022-12-08 16:55:04 +00:00
if ((!is_vm_hugetlb_page(vma) && new_flags & VM_SPECIAL) ||
(vma->vm_flags & VM_DROPPABLE))
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return -EINVAL;
new_flags &= ~VM_DONTDUMP;
break;
case MADV_MERGEABLE:
case MADV_UNMERGEABLE:
error = ksm_madvise(vma, start, end, behavior, &new_flags);
if (error)
goto out;
break;
case MADV_HUGEPAGE:
case MADV_NOHUGEPAGE:
error = hugepage_madvise(vma, &new_flags, behavior);
if (error)
goto out;
break;
mm/madvise: introduce MADV_COLLAPSE sync hugepage collapse This idea was introduced by David Rientjes[1]. Introduce a new madvise mode, MADV_COLLAPSE, that allows users to request a synchronous collapse of memory at their own expense. The benefits of this approach are: * CPU is charged to the process that wants to spend the cycles for the THP * Avoid unpredictable timing of khugepaged collapse Semantics This call is independent of the system-wide THP sysfs settings, but will fail for memory marked VM_NOHUGEPAGE. If the ranges provided span multiple VMAs, the semantics of the collapse over each VMA is independent from the others. This implies a hugepage cannot cross a VMA boundary. If collapse of a given hugepage-aligned/sized region fails, the operation may continue to attempt collapsing the remainder of memory specified. The memory ranges provided must be page-aligned, but are not required to be hugepage-aligned. If the memory ranges are not hugepage-aligned, the start/end of the range will be clamped to the first/last hugepage-aligned address covered by said range. The memory ranges must span at least one hugepage-sized region. All non-resident pages covered by the range will first be swapped/faulted-in, before being internally copied onto a freshly allocated hugepage. Unmapped pages will have their data directly initialized to 0 in the new hugepage. However, for every eligible hugepage aligned/sized region to-be collapsed, at least one page must currently be backed by memory (a PMD covering the address range must already exist). Allocation for the new hugepage may enter direct reclaim and/or compaction, regardless of VMA flags. When the system has multiple NUMA nodes, the hugepage will be allocated from the node providing the most native pages. This operation operates on the current state of the specified process and makes no persistent changes or guarantees on how pages will be mapped, constructed, or faulted in the future Return Value If all hugepage-sized/aligned regions covered by the provided range were either successfully collapsed, or were already PMD-mapped THPs, this operation will be deemed successful. On success, process_madvise(2) returns the number of bytes advised, and madvise(2) returns 0. Else, -1 is returned and errno is set to indicate the error for the most-recently attempted hugepage collapse. Note that many failures might have occurred, since the operation may continue to collapse in the event a single hugepage-sized/aligned region fails. ENOMEM Memory allocation failed or VMA not found EBUSY Memcg charging failed EAGAIN Required resource temporarily unavailable. Try again might succeed. EINVAL Other error: No PMD found, subpage doesn't have Present bit set, "Special" page no backed by struct page, VMA incorrectly sized, address not page-aligned, ... Most notable here is ENOMEM and EBUSY (new to madvise) which are intended to provide the caller with actionable feedback so they may take an appropriate fallback measure. Use Cases An immediate user of this new functionality are malloc() implementations that manage memory in hugepage-sized chunks, but sometimes subrelease memory back to the system in native-sized chunks via MADV_DONTNEED; zapping the pmd. Later, when the memory is hot, the implementation could madvise(MADV_COLLAPSE) to re-back the memory by THPs to regain hugepage coverage and dTLB performance. TCMalloc is such an implementation that could benefit from this[2]. Only privately-mapped anon memory is supported for now, but additional support for file, shmem, and HugeTLB high-granularity mappings[2] is expected. File and tmpfs/shmem support would permit: * Backing executable text by THPs. Current support provided by CONFIG_READ_ONLY_THP_FOR_FS may take a long time on a large system which might impair services from serving at their full rated load after (re)starting. Tricks like mremap(2)'ing text onto anonymous memory to immediately realize iTLB performance prevents page sharing and demand paging, both of which increase steady state memory footprint. With MADV_COLLAPSE, we get the best of both worlds: Peak upfront performance and lower RAM footprints. * Backing guest memory by hugapages after the memory contents have been migrated in native-page-sized chunks to a new host, in a userfaultfd-based live-migration stack. [1] https://lore.kernel.org/linux-mm/d098c392-273a-36a4-1a29-59731cdf5d3d@google.com/ [2] https://github.com/google/tcmalloc/tree/master/tcmalloc [jrdr.linux@gmail.com: avoid possible memory leak in failure path] Link: https://lkml.kernel.org/r/20220713024109.62810-1-jrdr.linux@gmail.com [zokeefe@google.com add missing kfree() to madvise_collapse()] Link: https://lore.kernel.org/linux-mm/20220713024109.62810-1-jrdr.linux@gmail.com/ Link: https://lkml.kernel.org/r/20220713161851.1879439-1-zokeefe@google.com [zokeefe@google.com: delay computation of hpage boundaries until use]] Link: https://lkml.kernel.org/r/20220720140603.1958773-4-zokeefe@google.com Link: https://lkml.kernel.org/r/20220706235936.2197195-10-zokeefe@google.com Signed-off-by: Zach O'Keefe <zokeefe@google.com> Signed-off-by: "Souptick Joarder (HPE)" <jrdr.linux@gmail.com> Suggested-by: David Rientjes <rientjes@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Chris Kennelly <ckennelly@google.com> Cc: Chris Zankel <chris@zankel.net> Cc: David Hildenbrand <david@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: Hugh Dickins <hughd@google.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-07-06 23:59:27 +00:00
case MADV_COLLAPSE:
return madvise_collapse(vma, prev, start, end);
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
}
mm: fix use-after-free when anon vma name is used after vma is freed When adjacent vmas are being merged it can result in the vma that was originally passed to madvise_update_vma being destroyed. In the current implementation, the name parameter passed to madvise_update_vma points directly to vma->anon_name and it is used after the call to vma_merge. In the cases when vma_merge merges the original vma and destroys it, this might result in UAF. For that the original vma would have to hold the anon_vma_name with the last reference. The following vma would need to contain a different anon_vma_name object with the same string. Such scenario is shown below: madvise_vma_behavior(vma) madvise_update_vma(vma, ..., anon_name == vma->anon_name) vma_merge(vma) __vma_adjust(vma) <-- merges vma with adjacent one vm_area_free(vma) <-- frees the original vma replace_vma_anon_name(anon_name) <-- UAF of vma->anon_name Fix this by raising the name refcount and stabilizing it. Link: https://lkml.kernel.org/r/20220224231834.1481408-3-surenb@google.com Link: https://lkml.kernel.org/r/20220223153613.835563-3-surenb@google.com Fixes: 9a10064f5625 ("mm: add a field to store names for private anonymous memory") Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reported-by: syzbot+aa7b3d4b35f9dc46a366@syzkaller.appspotmail.com Acked-by: Michal Hocko <mhocko@suse.com> Cc: Alexey Gladkov <legion@kernel.org> Cc: Chris Hyser <chris.hyser@oracle.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Collingbourne <pcc@google.com> Cc: Sasha Levin <sashal@kernel.org> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xiaofeng Cao <caoxiaofeng@yulong.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-05 04:28:58 +00:00
anon_name = anon_vma_name(vma);
anon_vma_name_get(anon_name);
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
error = madvise_update_vma(vma, prev, start, end, new_flags,
mm: fix use-after-free when anon vma name is used after vma is freed When adjacent vmas are being merged it can result in the vma that was originally passed to madvise_update_vma being destroyed. In the current implementation, the name parameter passed to madvise_update_vma points directly to vma->anon_name and it is used after the call to vma_merge. In the cases when vma_merge merges the original vma and destroys it, this might result in UAF. For that the original vma would have to hold the anon_vma_name with the last reference. The following vma would need to contain a different anon_vma_name object with the same string. Such scenario is shown below: madvise_vma_behavior(vma) madvise_update_vma(vma, ..., anon_name == vma->anon_name) vma_merge(vma) __vma_adjust(vma) <-- merges vma with adjacent one vm_area_free(vma) <-- frees the original vma replace_vma_anon_name(anon_name) <-- UAF of vma->anon_name Fix this by raising the name refcount and stabilizing it. Link: https://lkml.kernel.org/r/20220224231834.1481408-3-surenb@google.com Link: https://lkml.kernel.org/r/20220223153613.835563-3-surenb@google.com Fixes: 9a10064f5625 ("mm: add a field to store names for private anonymous memory") Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reported-by: syzbot+aa7b3d4b35f9dc46a366@syzkaller.appspotmail.com Acked-by: Michal Hocko <mhocko@suse.com> Cc: Alexey Gladkov <legion@kernel.org> Cc: Chris Hyser <chris.hyser@oracle.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Colin Cross <ccross@google.com> Cc: Cyrill Gorcunov <gorcunov@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Peter Collingbourne <pcc@google.com> Cc: Sasha Levin <sashal@kernel.org> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xiaofeng Cao <caoxiaofeng@yulong.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-05 04:28:58 +00:00
anon_name);
anon_vma_name_put(anon_name);
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
out:
/*
* madvise() returns EAGAIN if kernel resources, such as
* slab, are temporarily unavailable.
*/
if (error == -ENOMEM)
error = -EAGAIN;
return error;
}
#ifdef CONFIG_MEMORY_FAILURE
/*
* Error injection support for memory error handling.
*/
mm/madvise.c: clean up MADV_SOFT_OFFLINE and MADV_HWPOISON This cleans up handling MADV_SOFT_OFFLINE and MADV_HWPOISON called through madvise() system call. * madvise_memory_failure() was misleading to accommodate handling of both memory_failure() as well as soft_offline_page() functions. Basically it handles memory error injection from user space which can go either way as memory failure or soft offline. Renamed as madvise_inject_error() instead. * Renamed struct page pointer 'p' to 'page'. * pr_info() was essentially printing PFN value but it said 'page' which was misleading. Made the process virtual address explicit. Before the patch: Soft offlining page 0x15e3e at 0x3fff8c230000 Soft offlining page 0x1f3 at 0x3fffa0da0000 Soft offlining page 0x744 at 0x3fff7d200000 Soft offlining page 0x1634d at 0x3fff95e20000 Soft offlining page 0x16349 at 0x3fff95e30000 Soft offlining page 0x1d6 at 0x3fff9e8b0000 Soft offlining page 0x5f3 at 0x3fff91bd0000 Injecting memory failure for page 0x15c8b at 0x3fff83280000 Injecting memory failure for page 0x16190 at 0x3fff83290000 Injecting memory failure for page 0x740 at 0x3fff9a2e0000 Injecting memory failure for page 0x741 at 0x3fff9a2f0000 After the patch: Soft offlining pfn 0x1484e at process virtual address 0x3fff883c0000 Soft offlining pfn 0x1484f at process virtual address 0x3fff883d0000 Soft offlining pfn 0x14850 at process virtual address 0x3fff883e0000 Soft offlining pfn 0x14851 at process virtual address 0x3fff883f0000 Soft offlining pfn 0x14852 at process virtual address 0x3fff88400000 Soft offlining pfn 0x14853 at process virtual address 0x3fff88410000 Soft offlining pfn 0x14854 at process virtual address 0x3fff88420000 Soft offlining pfn 0x1521c at process virtual address 0x3fff6bc70000 Injecting memory failure for pfn 0x10fcf at process virtual address 0x3fff86310000 Injecting memory failure for pfn 0x10fd0 at process virtual address 0x3fff86320000 Injecting memory failure for pfn 0x10fd1 at process virtual address 0x3fff86330000 Injecting memory failure for pfn 0x10fd2 at process virtual address 0x3fff86340000 Injecting memory failure for pfn 0x10fd3 at process virtual address 0x3fff86350000 Injecting memory failure for pfn 0x10fd4 at process virtual address 0x3fff86360000 Injecting memory failure for pfn 0x10fd5 at process virtual address 0x3fff86370000 Link: http://lkml.kernel.org/r/20170410084701.11248-1-khandual@linux.vnet.ibm.com Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 21:55:25 +00:00
static int madvise_inject_error(int behavior,
unsigned long start, unsigned long end)
{
unsigned long size;
mm/madvise.c: clean up MADV_SOFT_OFFLINE and MADV_HWPOISON This cleans up handling MADV_SOFT_OFFLINE and MADV_HWPOISON called through madvise() system call. * madvise_memory_failure() was misleading to accommodate handling of both memory_failure() as well as soft_offline_page() functions. Basically it handles memory error injection from user space which can go either way as memory failure or soft offline. Renamed as madvise_inject_error() instead. * Renamed struct page pointer 'p' to 'page'. * pr_info() was essentially printing PFN value but it said 'page' which was misleading. Made the process virtual address explicit. Before the patch: Soft offlining page 0x15e3e at 0x3fff8c230000 Soft offlining page 0x1f3 at 0x3fffa0da0000 Soft offlining page 0x744 at 0x3fff7d200000 Soft offlining page 0x1634d at 0x3fff95e20000 Soft offlining page 0x16349 at 0x3fff95e30000 Soft offlining page 0x1d6 at 0x3fff9e8b0000 Soft offlining page 0x5f3 at 0x3fff91bd0000 Injecting memory failure for page 0x15c8b at 0x3fff83280000 Injecting memory failure for page 0x16190 at 0x3fff83290000 Injecting memory failure for page 0x740 at 0x3fff9a2e0000 Injecting memory failure for page 0x741 at 0x3fff9a2f0000 After the patch: Soft offlining pfn 0x1484e at process virtual address 0x3fff883c0000 Soft offlining pfn 0x1484f at process virtual address 0x3fff883d0000 Soft offlining pfn 0x14850 at process virtual address 0x3fff883e0000 Soft offlining pfn 0x14851 at process virtual address 0x3fff883f0000 Soft offlining pfn 0x14852 at process virtual address 0x3fff88400000 Soft offlining pfn 0x14853 at process virtual address 0x3fff88410000 Soft offlining pfn 0x14854 at process virtual address 0x3fff88420000 Soft offlining pfn 0x1521c at process virtual address 0x3fff6bc70000 Injecting memory failure for pfn 0x10fcf at process virtual address 0x3fff86310000 Injecting memory failure for pfn 0x10fd0 at process virtual address 0x3fff86320000 Injecting memory failure for pfn 0x10fd1 at process virtual address 0x3fff86330000 Injecting memory failure for pfn 0x10fd2 at process virtual address 0x3fff86340000 Injecting memory failure for pfn 0x10fd3 at process virtual address 0x3fff86350000 Injecting memory failure for pfn 0x10fd4 at process virtual address 0x3fff86360000 Injecting memory failure for pfn 0x10fd5 at process virtual address 0x3fff86370000 Link: http://lkml.kernel.org/r/20170410084701.11248-1-khandual@linux.vnet.ibm.com Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 21:55:25 +00:00
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
mm/madvise.c: clean up MADV_SOFT_OFFLINE and MADV_HWPOISON This cleans up handling MADV_SOFT_OFFLINE and MADV_HWPOISON called through madvise() system call. * madvise_memory_failure() was misleading to accommodate handling of both memory_failure() as well as soft_offline_page() functions. Basically it handles memory error injection from user space which can go either way as memory failure or soft offline. Renamed as madvise_inject_error() instead. * Renamed struct page pointer 'p' to 'page'. * pr_info() was essentially printing PFN value but it said 'page' which was misleading. Made the process virtual address explicit. Before the patch: Soft offlining page 0x15e3e at 0x3fff8c230000 Soft offlining page 0x1f3 at 0x3fffa0da0000 Soft offlining page 0x744 at 0x3fff7d200000 Soft offlining page 0x1634d at 0x3fff95e20000 Soft offlining page 0x16349 at 0x3fff95e30000 Soft offlining page 0x1d6 at 0x3fff9e8b0000 Soft offlining page 0x5f3 at 0x3fff91bd0000 Injecting memory failure for page 0x15c8b at 0x3fff83280000 Injecting memory failure for page 0x16190 at 0x3fff83290000 Injecting memory failure for page 0x740 at 0x3fff9a2e0000 Injecting memory failure for page 0x741 at 0x3fff9a2f0000 After the patch: Soft offlining pfn 0x1484e at process virtual address 0x3fff883c0000 Soft offlining pfn 0x1484f at process virtual address 0x3fff883d0000 Soft offlining pfn 0x14850 at process virtual address 0x3fff883e0000 Soft offlining pfn 0x14851 at process virtual address 0x3fff883f0000 Soft offlining pfn 0x14852 at process virtual address 0x3fff88400000 Soft offlining pfn 0x14853 at process virtual address 0x3fff88410000 Soft offlining pfn 0x14854 at process virtual address 0x3fff88420000 Soft offlining pfn 0x1521c at process virtual address 0x3fff6bc70000 Injecting memory failure for pfn 0x10fcf at process virtual address 0x3fff86310000 Injecting memory failure for pfn 0x10fd0 at process virtual address 0x3fff86320000 Injecting memory failure for pfn 0x10fd1 at process virtual address 0x3fff86330000 Injecting memory failure for pfn 0x10fd2 at process virtual address 0x3fff86340000 Injecting memory failure for pfn 0x10fd3 at process virtual address 0x3fff86350000 Injecting memory failure for pfn 0x10fd4 at process virtual address 0x3fff86360000 Injecting memory failure for pfn 0x10fd5 at process virtual address 0x3fff86370000 Link: http://lkml.kernel.org/r/20170410084701.11248-1-khandual@linux.vnet.ibm.com Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 21:55:25 +00:00
for (; start < end; start += size) {
unsigned long pfn;
struct page *page;
int ret;
mm/madvise.c: clean up MADV_SOFT_OFFLINE and MADV_HWPOISON This cleans up handling MADV_SOFT_OFFLINE and MADV_HWPOISON called through madvise() system call. * madvise_memory_failure() was misleading to accommodate handling of both memory_failure() as well as soft_offline_page() functions. Basically it handles memory error injection from user space which can go either way as memory failure or soft offline. Renamed as madvise_inject_error() instead. * Renamed struct page pointer 'p' to 'page'. * pr_info() was essentially printing PFN value but it said 'page' which was misleading. Made the process virtual address explicit. Before the patch: Soft offlining page 0x15e3e at 0x3fff8c230000 Soft offlining page 0x1f3 at 0x3fffa0da0000 Soft offlining page 0x744 at 0x3fff7d200000 Soft offlining page 0x1634d at 0x3fff95e20000 Soft offlining page 0x16349 at 0x3fff95e30000 Soft offlining page 0x1d6 at 0x3fff9e8b0000 Soft offlining page 0x5f3 at 0x3fff91bd0000 Injecting memory failure for page 0x15c8b at 0x3fff83280000 Injecting memory failure for page 0x16190 at 0x3fff83290000 Injecting memory failure for page 0x740 at 0x3fff9a2e0000 Injecting memory failure for page 0x741 at 0x3fff9a2f0000 After the patch: Soft offlining pfn 0x1484e at process virtual address 0x3fff883c0000 Soft offlining pfn 0x1484f at process virtual address 0x3fff883d0000 Soft offlining pfn 0x14850 at process virtual address 0x3fff883e0000 Soft offlining pfn 0x14851 at process virtual address 0x3fff883f0000 Soft offlining pfn 0x14852 at process virtual address 0x3fff88400000 Soft offlining pfn 0x14853 at process virtual address 0x3fff88410000 Soft offlining pfn 0x14854 at process virtual address 0x3fff88420000 Soft offlining pfn 0x1521c at process virtual address 0x3fff6bc70000 Injecting memory failure for pfn 0x10fcf at process virtual address 0x3fff86310000 Injecting memory failure for pfn 0x10fd0 at process virtual address 0x3fff86320000 Injecting memory failure for pfn 0x10fd1 at process virtual address 0x3fff86330000 Injecting memory failure for pfn 0x10fd2 at process virtual address 0x3fff86340000 Injecting memory failure for pfn 0x10fd3 at process virtual address 0x3fff86350000 Injecting memory failure for pfn 0x10fd4 at process virtual address 0x3fff86360000 Injecting memory failure for pfn 0x10fd5 at process virtual address 0x3fff86370000 Link: http://lkml.kernel.org/r/20170410084701.11248-1-khandual@linux.vnet.ibm.com Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 21:55:25 +00:00
ret = get_user_pages_fast(start, 1, 0, &page);
if (ret != 1)
return ret;
pfn = page_to_pfn(page);
/*
* When soft offlining hugepages, after migrating the page
* we dissolve it, therefore in the second loop "page" will
* no longer be a compound page.
*/
size = page_size(compound_head(page));
mm/madvise.c: clean up MADV_SOFT_OFFLINE and MADV_HWPOISON This cleans up handling MADV_SOFT_OFFLINE and MADV_HWPOISON called through madvise() system call. * madvise_memory_failure() was misleading to accommodate handling of both memory_failure() as well as soft_offline_page() functions. Basically it handles memory error injection from user space which can go either way as memory failure or soft offline. Renamed as madvise_inject_error() instead. * Renamed struct page pointer 'p' to 'page'. * pr_info() was essentially printing PFN value but it said 'page' which was misleading. Made the process virtual address explicit. Before the patch: Soft offlining page 0x15e3e at 0x3fff8c230000 Soft offlining page 0x1f3 at 0x3fffa0da0000 Soft offlining page 0x744 at 0x3fff7d200000 Soft offlining page 0x1634d at 0x3fff95e20000 Soft offlining page 0x16349 at 0x3fff95e30000 Soft offlining page 0x1d6 at 0x3fff9e8b0000 Soft offlining page 0x5f3 at 0x3fff91bd0000 Injecting memory failure for page 0x15c8b at 0x3fff83280000 Injecting memory failure for page 0x16190 at 0x3fff83290000 Injecting memory failure for page 0x740 at 0x3fff9a2e0000 Injecting memory failure for page 0x741 at 0x3fff9a2f0000 After the patch: Soft offlining pfn 0x1484e at process virtual address 0x3fff883c0000 Soft offlining pfn 0x1484f at process virtual address 0x3fff883d0000 Soft offlining pfn 0x14850 at process virtual address 0x3fff883e0000 Soft offlining pfn 0x14851 at process virtual address 0x3fff883f0000 Soft offlining pfn 0x14852 at process virtual address 0x3fff88400000 Soft offlining pfn 0x14853 at process virtual address 0x3fff88410000 Soft offlining pfn 0x14854 at process virtual address 0x3fff88420000 Soft offlining pfn 0x1521c at process virtual address 0x3fff6bc70000 Injecting memory failure for pfn 0x10fcf at process virtual address 0x3fff86310000 Injecting memory failure for pfn 0x10fd0 at process virtual address 0x3fff86320000 Injecting memory failure for pfn 0x10fd1 at process virtual address 0x3fff86330000 Injecting memory failure for pfn 0x10fd2 at process virtual address 0x3fff86340000 Injecting memory failure for pfn 0x10fd3 at process virtual address 0x3fff86350000 Injecting memory failure for pfn 0x10fd4 at process virtual address 0x3fff86360000 Injecting memory failure for pfn 0x10fd5 at process virtual address 0x3fff86370000 Link: http://lkml.kernel.org/r/20170410084701.11248-1-khandual@linux.vnet.ibm.com Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com> Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 21:55:25 +00:00
if (behavior == MADV_SOFT_OFFLINE) {
pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n",
pfn, start);
ret = soft_offline_page(pfn, MF_COUNT_INCREASED);
} else {
pr_info("Injecting memory failure for pfn %#lx at process virtual address %#lx\n",
pfn, start);
ret = memory_failure(pfn, MF_ACTION_REQUIRED | MF_COUNT_INCREASED | MF_SW_SIMULATED);
if (ret == -EOPNOTSUPP)
ret = 0;
}
if (ret)
return ret;
}
mm, madvise: ensure poisoned pages are removed from per-cpu lists Wendy Wang reported off-list that a RAS HWPOISON-SOFT test case failed and bisected it to the commit 479f854a207c ("mm, page_alloc: defer debugging checks of pages allocated from the PCP"). The problem is that a page that was poisoned with madvise() is reused. The commit removed a check that would trigger if DEBUG_VM was enabled but re-enabling the check only fixes the problem as a side-effect by printing a bad_page warning and recovering. The root of the problem is that an madvise() can leave a poisoned page on the per-cpu list. This patch drains all per-cpu lists after pages are poisoned so that they will not be reused. Wendy reports that the test case in question passes with this patch applied. While this could be done in a targeted fashion, it is over-complicated for such a rare operation. Link: http://lkml.kernel.org/r/20170828133414.7qro57jbepdcyz5x@techsingularity.net Fixes: 479f854a207c ("mm, page_alloc: defer debugging checks of pages allocated from the PCP") Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Reported-by: Wang, Wendy <wendy.wang@intel.com> Tested-by: Wang, Wendy <wendy.wang@intel.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: "Hansen, Dave" <dave.hansen@intel.com> Cc: "Luck, Tony" <tony.luck@intel.com> Cc: Naoya Horiguchi <nao.horiguchi@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-08-31 23:15:30 +00:00
return 0;
}
#endif
static bool
madvise_behavior_valid(int behavior)
{
switch (behavior) {
case MADV_DOFORK:
case MADV_DONTFORK:
case MADV_NORMAL:
case MADV_SEQUENTIAL:
case MADV_RANDOM:
case MADV_REMOVE:
case MADV_WILLNEED:
case MADV_DONTNEED:
mm: madvise: MADV_DONTNEED_LOCKED MADV_DONTNEED historically rejects mlocked ranges, but with MLOCK_ONFAULT and MCL_ONFAULT allowing to mlock without populating, there are valid use cases for depopulating locked ranges as well. Users mlock memory to protect secrets. There are allocators for secure buffers that want in-use memory generally mlocked, but cleared and invalidated memory to give up the physical pages. This could be done with explicit munlock -> mlock calls on free -> alloc of course, but that adds two unnecessary syscalls, heavy mmap_sem write locks, vma splits and re-merges - only to get rid of the backing pages. Users also mlockall(MCL_ONFAULT) to suppress sustained paging, but are okay with on-demand initial population. It seems valid to selectively free some memory during the lifetime of such a process, without having to mess with its overall policy. Why add a separate flag? Isn't this a pretty niche usecase? - MADV_DONTNEED has been bailing on locked vmas forever. It's at least conceivable that someone, somewhere is relying on mlock to protect data from perhaps broader invalidation calls. Changing this behavior now could lead to quiet data corruption. - It also clarifies expectations around MADV_FREE and maybe MADV_REMOVE. It avoids the situation where one quietly behaves different than the others. MADV_FREE_LOCKED can be added later. - The combination of mlock() and madvise() in the first place is probably niche. But where it happens, I'd say that dropping pages from a locked region once they don't contain secrets or won't page anymore is much saner than relying on mlock to protect memory from speculative or errant invalidation calls. It's just that we can't change the default behavior because of the two previous points. Given that, an explicit new flag seems to make the most sense. [hannes@cmpxchg.org: fix mips build] Link: https://lkml.kernel.org/r/20220304171912.305060-1-hannes@cmpxchg.org Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Shakeel Butt <shakeelb@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-25 01:14:12 +00:00
case MADV_DONTNEED_LOCKED:
mm: support madvise(MADV_FREE) Linux doesn't have an ability to free pages lazy while other OS already have been supported that named by madvise(MADV_FREE). The gain is clear that kernel can discard freed pages rather than swapping out or OOM if memory pressure happens. Without memory pressure, freed pages would be reused by userspace without another additional overhead(ex, page fault + allocation + zeroing). Jason Evans said: : Facebook has been using MAP_UNINITIALIZED : (https://lkml.org/lkml/2012/1/18/308) in some of its applications for : several years, but there are operational costs to maintaining this : out-of-tree in our kernel and in jemalloc, and we are anxious to retire it : in favor of MADV_FREE. When we first enabled MAP_UNINITIALIZED it : increased throughput for much of our workload by ~5%, and although the : benefit has decreased using newer hardware and kernels, there is still : enough benefit that we cannot reasonably retire it without a replacement. : : Aside from Facebook operations, there are numerous broadly used : applications that would benefit from MADV_FREE. The ones that immediately : come to mind are redis, varnish, and MariaDB. I don't have much insight : into Android internals and development process, but I would hope to see : MADV_FREE support eventually end up there as well to benefit applications : linked with the integrated jemalloc. : : jemalloc will use MADV_FREE once it becomes available in the Linux kernel. : In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's : available: *BSD, OS X, Windows, and Solaris -- every platform except Linux : (and AIX, but I'm not sure it even compiles on AIX). The lack of : MADV_FREE on Linux forced me down a long series of increasingly : sophisticated heuristics for madvise() volume reduction, and even so this : remains a common performance issue for people using jemalloc on Linux. : Please integrate MADV_FREE; many people will benefit substantially. How it works: When madvise syscall is called, VM clears dirty bit of ptes of the range. If memory pressure happens, VM checks dirty bit of page table and if it found still "clean", it means it's a "lazyfree pages" so VM could discard the page instead of swapping out. Once there was store operation for the page before VM peek a page to reclaim, dirty bit is set so VM can swap out the page instead of discarding. One thing we should notice is that basically, MADV_FREE relies on dirty bit in page table entry to decide whether VM allows to discard the page or not. IOW, if page table entry includes marked dirty bit, VM shouldn't discard the page. However, as a example, if swap-in by read fault happens, page table entry doesn't have dirty bit so MADV_FREE could discard the page wrongly. For avoiding the problem, MADV_FREE did more checks with PageDirty and PageSwapCache. It worked out because swapped-in page lives on swap cache and since it is evicted from the swap cache, the page has PG_dirty flag. So both page flags check effectively prevent wrong discarding by MADV_FREE. However, a problem in above logic is that swapped-in page has PG_dirty still after they are removed from swap cache so VM cannot consider the page as freeable any more even if madvise_free is called in future. Look at below example for detail. ptr = malloc(); memset(ptr); .. .. .. heavy memory pressure so all of pages are swapped out .. .. var = *ptr; -> a page swapped-in and could be removed from swapcache. Then, page table doesn't mark dirty bit and page descriptor includes PG_dirty .. .. madvise_free(ptr); -> It doesn't clear PG_dirty of the page. .. .. .. .. heavy memory pressure again. .. In this time, VM cannot discard the page because the page .. has *PG_dirty* To solve the problem, this patch clears PG_dirty if only the page is owned exclusively by current process when madvise is called because PG_dirty represents ptes's dirtiness in several processes so we could clear it only if we own it exclusively. Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc and hope glibc supports it) and jemalloc/tcmalloc already have supported the feature for other OS(ex, FreeBSD) barrios@blaptop:~/benchmark/ebizzy$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 12 On-line CPU(s) list: 0-11 Thread(s) per core: 1 Core(s) per socket: 1 Socket(s): 12 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 2 Stepping: 3 CPU MHz: 3200.185 BogoMIPS: 6400.53 Virtualization: VT-x Hypervisor vendor: KVM Virtualization type: full L1d cache: 32K L1i cache: 32K L2 cache: 4096K NUMA node0 CPU(s): 0-11 ebizzy benchmark(./ebizzy -S 10 -n 512) Higher avg is better. vanilla-jemalloc MADV_free-jemalloc 1 thread records: 10 records: 10 avg: 2961.90 avg: 12069.70 std: 71.96(2.43%) std: 186.68(1.55%) max: 3070.00 max: 12385.00 min: 2796.00 min: 11746.00 2 thread records: 10 records: 10 avg: 5020.00 avg: 17827.00 std: 264.87(5.28%) std: 358.52(2.01%) max: 5244.00 max: 18760.00 min: 4251.00 min: 17382.00 4 thread records: 10 records: 10 avg: 8988.80 avg: 27930.80 std: 1175.33(13.08%) std: 3317.33(11.88%) max: 9508.00 max: 30879.00 min: 5477.00 min: 21024.00 8 thread records: 10 records: 10 avg: 13036.50 avg: 33739.40 std: 170.67(1.31%) std: 5146.22(15.25%) max: 13371.00 max: 40572.00 min: 12785.00 min: 24088.00 16 thread records: 10 records: 10 avg: 11092.40 avg: 31424.20 std: 710.60(6.41%) std: 3763.89(11.98%) max: 12446.00 max: 36635.00 min: 9949.00 min: 25669.00 32 thread records: 10 records: 10 avg: 11067.00 avg: 34495.80 std: 971.06(8.77%) std: 2721.36(7.89%) max: 12010.00 max: 38598.00 min: 9002.00 min: 30636.00 In summary, MADV_FREE is about much faster than MADV_DONTNEED. This patch (of 12): Add core MADV_FREE implementation. [akpm@linux-foundation.org: small cleanups] Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Hugh Dickins <hughd@google.com> Cc: Mika Penttil <mika.penttila@nextfour.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jason Evans <je@fb.com> Cc: Daniel Micay <danielmicay@gmail.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Shaohua Li <shli@kernel.org> Cc: <yalin.wang2010@gmail.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: "Shaohua Li" <shli@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chen Gang <gang.chen.5i5j@gmail.com> Cc: Chris Zankel <chris@zankel.net> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David S. Miller <davem@davemloft.net> Cc: Helge Deller <deller@gmx.de> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Roland Dreier <roland@kernel.org> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Shaohua Li <shli@kernel.org> Cc: Will Deacon <will.deacon@arm.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:54:53 +00:00
case MADV_FREE:
mm: introduce MADV_COLD Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:08 +00:00
case MADV_COLD:
mm: introduce MADV_PAGEOUT When a process expects no accesses to a certain memory range for a long time, it could hint kernel that the pages can be reclaimed instantly but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_PAGEOUT hint to madvise(2) syscall. MADV_PAGEOUT can be used by a process to mark a memory range as not expected to be used for a long time so that kernel reclaims *any LRU* pages instantly. The hint can help kernel in deciding which pages to evict proactively. A note: It doesn't apply SWAP_CLUSTER_MAX LRU page isolation limit intentionally because it's automatically bounded by PMD size. If PMD size(e.g., 256) makes some trouble, we could fix it later by limit it to SWAP_CLUSTER_MAX[1]. - man-page material MADV_PAGEOUT (since Linux x.x) Do not expect access in the near future so pages in the specified regions could be reclaimed instantly regardless of memory pressure. Thus, access in the range after successful operation could cause major page fault but never lose the up-to-date contents unlike MADV_DONTNEED. Pages belonging to a shared mapping are only processed if a write access is allowed for the calling process. MADV_PAGEOUT cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [1] https://lore.kernel.org/lkml/20190710194719.GS29695@dhcp22.suse.cz/ [minchan@kernel.org: clear PG_active on MADV_PAGEOUT] Link: http://lkml.kernel.org/r/20190802200643.GA181880@google.com [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-5-minchan@kernel.org Signed-off-by: Minchan Kim <minchan@kernel.org> Reported-by: kbuild test robot <lkp@intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Chris Zankel <chris@zankel.net> Cc: Daniel Colascione <dancol@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hillf Danton <hdanton@sina.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-25 23:49:15 +00:00
case MADV_PAGEOUT:
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
case MADV_POPULATE_READ:
case MADV_POPULATE_WRITE:
ksm: the mm interface to ksm This patch presents the mm interface to a dummy version of ksm.c, for better scrutiny of that interface: the real ksm.c follows later. When CONFIG_KSM is not set, madvise(2) reject MADV_MERGEABLE and MADV_UNMERGEABLE with EINVAL, since that seems more helpful than pretending that they can be serviced. But when CONFIG_KSM=y, accept them even if KSM is not currently running, and even on areas which KSM will not touch (e.g. hugetlb or shared file or special driver mappings). Like other madvices, report ENOMEM despite success if any area in the range is unmapped, and use EAGAIN to report out of memory. Define vma flag VM_MERGEABLE to identify an area on which KSM may try merging pages: leave it to ksm_madvise() to decide whether to set it. Define mm flag MMF_VM_MERGEABLE to identify an mm which might contain VM_MERGEABLE areas, to minimize callouts when forking or exiting. Based upon earlier patches by Chris Wright and Izik Eidus. Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Chris Wright <chrisw@redhat.com> Signed-off-by: Izik Eidus <ieidus@redhat.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Rik van Riel <riel@redhat.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Lee Schermerhorn <lee.schermerhorn@hp.com> Cc: Avi Kivity <avi@redhat.com> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-09-22 00:01:57 +00:00
#ifdef CONFIG_KSM
case MADV_MERGEABLE:
case MADV_UNMERGEABLE:
#endif
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
case MADV_HUGEPAGE:
case MADV_NOHUGEPAGE:
mm/madvise: introduce MADV_COLLAPSE sync hugepage collapse This idea was introduced by David Rientjes[1]. Introduce a new madvise mode, MADV_COLLAPSE, that allows users to request a synchronous collapse of memory at their own expense. The benefits of this approach are: * CPU is charged to the process that wants to spend the cycles for the THP * Avoid unpredictable timing of khugepaged collapse Semantics This call is independent of the system-wide THP sysfs settings, but will fail for memory marked VM_NOHUGEPAGE. If the ranges provided span multiple VMAs, the semantics of the collapse over each VMA is independent from the others. This implies a hugepage cannot cross a VMA boundary. If collapse of a given hugepage-aligned/sized region fails, the operation may continue to attempt collapsing the remainder of memory specified. The memory ranges provided must be page-aligned, but are not required to be hugepage-aligned. If the memory ranges are not hugepage-aligned, the start/end of the range will be clamped to the first/last hugepage-aligned address covered by said range. The memory ranges must span at least one hugepage-sized region. All non-resident pages covered by the range will first be swapped/faulted-in, before being internally copied onto a freshly allocated hugepage. Unmapped pages will have their data directly initialized to 0 in the new hugepage. However, for every eligible hugepage aligned/sized region to-be collapsed, at least one page must currently be backed by memory (a PMD covering the address range must already exist). Allocation for the new hugepage may enter direct reclaim and/or compaction, regardless of VMA flags. When the system has multiple NUMA nodes, the hugepage will be allocated from the node providing the most native pages. This operation operates on the current state of the specified process and makes no persistent changes or guarantees on how pages will be mapped, constructed, or faulted in the future Return Value If all hugepage-sized/aligned regions covered by the provided range were either successfully collapsed, or were already PMD-mapped THPs, this operation will be deemed successful. On success, process_madvise(2) returns the number of bytes advised, and madvise(2) returns 0. Else, -1 is returned and errno is set to indicate the error for the most-recently attempted hugepage collapse. Note that many failures might have occurred, since the operation may continue to collapse in the event a single hugepage-sized/aligned region fails. ENOMEM Memory allocation failed or VMA not found EBUSY Memcg charging failed EAGAIN Required resource temporarily unavailable. Try again might succeed. EINVAL Other error: No PMD found, subpage doesn't have Present bit set, "Special" page no backed by struct page, VMA incorrectly sized, address not page-aligned, ... Most notable here is ENOMEM and EBUSY (new to madvise) which are intended to provide the caller with actionable feedback so they may take an appropriate fallback measure. Use Cases An immediate user of this new functionality are malloc() implementations that manage memory in hugepage-sized chunks, but sometimes subrelease memory back to the system in native-sized chunks via MADV_DONTNEED; zapping the pmd. Later, when the memory is hot, the implementation could madvise(MADV_COLLAPSE) to re-back the memory by THPs to regain hugepage coverage and dTLB performance. TCMalloc is such an implementation that could benefit from this[2]. Only privately-mapped anon memory is supported for now, but additional support for file, shmem, and HugeTLB high-granularity mappings[2] is expected. File and tmpfs/shmem support would permit: * Backing executable text by THPs. Current support provided by CONFIG_READ_ONLY_THP_FOR_FS may take a long time on a large system which might impair services from serving at their full rated load after (re)starting. Tricks like mremap(2)'ing text onto anonymous memory to immediately realize iTLB performance prevents page sharing and demand paging, both of which increase steady state memory footprint. With MADV_COLLAPSE, we get the best of both worlds: Peak upfront performance and lower RAM footprints. * Backing guest memory by hugapages after the memory contents have been migrated in native-page-sized chunks to a new host, in a userfaultfd-based live-migration stack. [1] https://lore.kernel.org/linux-mm/d098c392-273a-36a4-1a29-59731cdf5d3d@google.com/ [2] https://github.com/google/tcmalloc/tree/master/tcmalloc [jrdr.linux@gmail.com: avoid possible memory leak in failure path] Link: https://lkml.kernel.org/r/20220713024109.62810-1-jrdr.linux@gmail.com [zokeefe@google.com add missing kfree() to madvise_collapse()] Link: https://lore.kernel.org/linux-mm/20220713024109.62810-1-jrdr.linux@gmail.com/ Link: https://lkml.kernel.org/r/20220713161851.1879439-1-zokeefe@google.com [zokeefe@google.com: delay computation of hpage boundaries until use]] Link: https://lkml.kernel.org/r/20220720140603.1958773-4-zokeefe@google.com Link: https://lkml.kernel.org/r/20220706235936.2197195-10-zokeefe@google.com Signed-off-by: Zach O'Keefe <zokeefe@google.com> Signed-off-by: "Souptick Joarder (HPE)" <jrdr.linux@gmail.com> Suggested-by: David Rientjes <rientjes@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Chris Kennelly <ckennelly@google.com> Cc: Chris Zankel <chris@zankel.net> Cc: David Hildenbrand <david@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: Hugh Dickins <hughd@google.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-07-06 23:59:27 +00:00
case MADV_COLLAPSE:
ksm: the mm interface to ksm This patch presents the mm interface to a dummy version of ksm.c, for better scrutiny of that interface: the real ksm.c follows later. When CONFIG_KSM is not set, madvise(2) reject MADV_MERGEABLE and MADV_UNMERGEABLE with EINVAL, since that seems more helpful than pretending that they can be serviced. But when CONFIG_KSM=y, accept them even if KSM is not currently running, and even on areas which KSM will not touch (e.g. hugetlb or shared file or special driver mappings). Like other madvices, report ENOMEM despite success if any area in the range is unmapped, and use EAGAIN to report out of memory. Define vma flag VM_MERGEABLE to identify an area on which KSM may try merging pages: leave it to ksm_madvise() to decide whether to set it. Define mm flag MMF_VM_MERGEABLE to identify an mm which might contain VM_MERGEABLE areas, to minimize callouts when forking or exiting. Based upon earlier patches by Chris Wright and Izik Eidus. Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Chris Wright <chrisw@redhat.com> Signed-off-by: Izik Eidus <ieidus@redhat.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Rik van Riel <riel@redhat.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Lee Schermerhorn <lee.schermerhorn@hp.com> Cc: Avi Kivity <avi@redhat.com> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-09-22 00:01:57 +00:00
#endif
case MADV_DONTDUMP:
case MADV_DODUMP:
mm,fork: introduce MADV_WIPEONFORK Introduce MADV_WIPEONFORK semantics, which result in a VMA being empty in the child process after fork. This differs from MADV_DONTFORK in one important way. If a child process accesses memory that was MADV_WIPEONFORK, it will get zeroes. The address ranges are still valid, they are just empty. If a child process accesses memory that was MADV_DONTFORK, it will get a segmentation fault, since those address ranges are no longer valid in the child after fork. Since MADV_DONTFORK also seems to be used to allow very large programs to fork in systems with strict memory overcommit restrictions, changing the semantics of MADV_DONTFORK might break existing programs. MADV_WIPEONFORK only works on private, anonymous VMAs. The use case is libraries that store or cache information, and want to know that they need to regenerate it in the child process after fork. Examples of this would be: - systemd/pulseaudio API checks (fail after fork) (replacing a getpid check, which is too slow without a PID cache) - PKCS#11 API reinitialization check (mandated by specification) - glibc's upcoming PRNG (reseed after fork) - OpenSSL PRNG (reseed after fork) The security benefits of a forking server having a re-inialized PRNG in every child process are pretty obvious. However, due to libraries having all kinds of internal state, and programs getting compiled with many different versions of each library, it is unreasonable to expect calling programs to re-initialize everything manually after fork. A further complication is the proliferation of clone flags, programs bypassing glibc's functions to call clone directly, and programs calling unshare, causing the glibc pthread_atfork hook to not get called. It would be better to have the kernel take care of this automatically. The patch also adds MADV_KEEPONFORK, to undo the effects of a prior MADV_WIPEONFORK. This is similar to the OpenBSD minherit syscall with MAP_INHERIT_ZERO: https://man.openbsd.org/minherit.2 [akpm@linux-foundation.org: numerically order arch/parisc/include/uapi/asm/mman.h #defines] Link: http://lkml.kernel.org/r/20170811212829.29186-3-riel@redhat.com Signed-off-by: Rik van Riel <riel@redhat.com> Reported-by: Florian Weimer <fweimer@redhat.com> Reported-by: Colm MacCártaigh <colm@allcosts.net> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: "Kirill A. Shutemov" <kirill@shutemov.name> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Helge Deller <deller@gmx.de> Cc: Kees Cook <keescook@chromium.org> Cc: Matthew Wilcox <willy@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Will Drewry <wad@chromium.org> Cc: <linux-api@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-06 23:25:15 +00:00
case MADV_WIPEONFORK:
case MADV_KEEPONFORK:
#ifdef CONFIG_MEMORY_FAILURE
case MADV_SOFT_OFFLINE:
case MADV_HWPOISON:
#endif
return true;
default:
return false;
}
}
mm/madvise: add MADV_COLLAPSE to process_madvise() Allow MADV_COLLAPSE behavior for process_madvise(2) if caller has CAP_SYS_ADMIN or is requesting collapse of it's own memory. This is useful for the development of userspace agents that seek to optimize THP utilization system-wide by using userspace signals to prioritize what memory is most deserving of being THP-backed. [zokeefe@google.com: remove CAP_SYS_ADMIN requirement for process_madvise(MADV_COLLAPSE)] Link: https://lkml.kernel.org/r/20220801210946.3069083-1-zokeefe@google.com Link: https://lkml.kernel.org/r/20220706235936.2197195-13-zokeefe@google.com Signed-off-by: Zach O'Keefe <zokeefe@google.com> Acked-by: David Rientjes <rientjes@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Chris Kennelly <ckennelly@google.com> Cc: Chris Zankel <chris@zankel.net> Cc: David Hildenbrand <david@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: Hugh Dickins <hughd@google.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: "Souptick Joarder (HPE)" <jrdr.linux@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-07-06 23:59:30 +00:00
static bool process_madvise_behavior_valid(int behavior)
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
{
switch (behavior) {
case MADV_COLD:
case MADV_PAGEOUT:
case MADV_WILLNEED:
mm/madvise: add MADV_COLLAPSE to process_madvise() Allow MADV_COLLAPSE behavior for process_madvise(2) if caller has CAP_SYS_ADMIN or is requesting collapse of it's own memory. This is useful for the development of userspace agents that seek to optimize THP utilization system-wide by using userspace signals to prioritize what memory is most deserving of being THP-backed. [zokeefe@google.com: remove CAP_SYS_ADMIN requirement for process_madvise(MADV_COLLAPSE)] Link: https://lkml.kernel.org/r/20220801210946.3069083-1-zokeefe@google.com Link: https://lkml.kernel.org/r/20220706235936.2197195-13-zokeefe@google.com Signed-off-by: Zach O'Keefe <zokeefe@google.com> Acked-by: David Rientjes <rientjes@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Chris Kennelly <ckennelly@google.com> Cc: Chris Zankel <chris@zankel.net> Cc: David Hildenbrand <david@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: Hugh Dickins <hughd@google.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Cc: "Souptick Joarder (HPE)" <jrdr.linux@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-07-06 23:59:30 +00:00
case MADV_COLLAPSE:
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
return true;
default:
return false;
}
}
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
/*
* Walk the vmas in range [start,end), and call the visit function on each one.
* The visit function will get start and end parameters that cover the overlap
* between the current vma and the original range. Any unmapped regions in the
* original range will result in this function returning -ENOMEM while still
* calling the visit function on all of the existing vmas in the range.
* Must be called with the mmap_lock held for reading or writing.
*/
static
int madvise_walk_vmas(struct mm_struct *mm, unsigned long start,
unsigned long end, unsigned long arg,
int (*visit)(struct vm_area_struct *vma,
struct vm_area_struct **prev, unsigned long start,
unsigned long end, unsigned long arg))
{
struct vm_area_struct *vma;
struct vm_area_struct *prev;
unsigned long tmp;
int unmapped_error = 0;
/*
* If the interval [start,end) covers some unmapped address
* ranges, just ignore them, but return -ENOMEM at the end.
* - different from the way of handling in mlock etc.
*/
vma = find_vma_prev(mm, start, &prev);
if (vma && start > vma->vm_start)
prev = vma;
for (;;) {
int error;
/* Still start < end. */
if (!vma)
return -ENOMEM;
/* Here start < (end|vma->vm_end). */
if (start < vma->vm_start) {
unmapped_error = -ENOMEM;
start = vma->vm_start;
if (start >= end)
break;
}
/* Here vma->vm_start <= start < (end|vma->vm_end) */
tmp = vma->vm_end;
if (end < tmp)
tmp = end;
/* Here vma->vm_start <= start < tmp <= (end|vma->vm_end). */
error = visit(vma, &prev, start, tmp, arg);
if (error)
return error;
start = tmp;
if (prev && start < prev->vm_end)
start = prev->vm_end;
if (start >= end)
break;
if (prev)
vma = find_vma(mm, prev->vm_end);
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
else /* madvise_remove dropped mmap_lock */
vma = find_vma(mm, start);
}
return unmapped_error;
}
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
#ifdef CONFIG_ANON_VMA_NAME
static int madvise_vma_anon_name(struct vm_area_struct *vma,
struct vm_area_struct **prev,
unsigned long start, unsigned long end,
2022-03-05 04:28:51 +00:00
unsigned long anon_name)
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
{
int error;
/* Only anonymous mappings can be named */
mm: anonymous shared memory naming Since commit 9a10064f5625 ("mm: add a field to store names for private anonymous memory"), name for private anonymous memory, but not shared anonymous, can be set. However, naming shared anonymous memory just as useful for tracking purposes. Extend the functionality to be able to set names for shared anon. There are two ways to create anonymous shared memory, using memfd or directly via mmap(): 1. fd = memfd_create(...) mem = mmap(..., MAP_SHARED, fd, ...) 2. mem = mmap(..., MAP_SHARED | MAP_ANONYMOUS, -1, ...) In both cases the anonymous shared memory is created the same way by mapping an unlinked file on tmpfs. The memfd way allows to give a name for anonymous shared memory, but not useful when parts of shared memory require to have distinct names. Example use case: The VMM maps VM memory as anonymous shared memory (not private because VMM is sandboxed and drivers are running in their own processes). However, the VM tells back to the VMM how parts of the memory are actually used by the guest, how each of the segments should be backed (i.e. 4K pages, 2M pages), and some other information about the segments. The naming allows us to monitor the effective memory footprint for each of these segments from the host without looking inside the guest. Sample output: /* Create shared anonymous segmenet */ anon_shmem = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); /* Name the segment: "MY-NAME" */ rv = prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, anon_shmem, SIZE, "MY-NAME"); cat /proc/<pid>/maps (and smaps): 7fc8e2b4c000-7fc8f2b4c000 rw-s 00000000 00:01 1024 [anon_shmem:MY-NAME] If the segment is not named, the output is: 7fc8e2b4c000-7fc8f2b4c000 rw-s 00000000 00:01 1024 /dev/zero (deleted) Link: https://lkml.kernel.org/r/20221115020602.804224-1-pasha.tatashin@soleen.com Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bagas Sanjaya <bagasdotme@gmail.com> Cc: Colin Cross <ccross@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Jonathan Corbet <corbet@lwn.net> Cc: "Kirill A . Shutemov" <kirill@shutemov.name> Cc: Liam Howlett <liam.howlett@oracle.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Rapoport <rppt@kernel.org> Cc: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Peter Xu <peterx@redhat.com> Cc: Sean Christopherson <seanjc@google.com> Cc: Vincent Whitchurch <vincent.whitchurch@axis.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: xu xin <cgel.zte@gmail.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Yu Zhao <yuzhao@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-11-15 02:06:01 +00:00
if (vma->vm_file && !vma_is_anon_shmem(vma))
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
return -EBADF;
error = madvise_update_vma(vma, prev, start, end, vma->vm_flags,
2022-03-05 04:28:51 +00:00
(struct anon_vma_name *)anon_name);
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
/*
* madvise() returns EAGAIN if kernel resources, such as
* slab, are temporarily unavailable.
*/
if (error == -ENOMEM)
error = -EAGAIN;
return error;
}
int madvise_set_anon_name(struct mm_struct *mm, unsigned long start,
2022-03-05 04:28:51 +00:00
unsigned long len_in, struct anon_vma_name *anon_name)
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
{
unsigned long end;
unsigned long len;
if (start & ~PAGE_MASK)
return -EINVAL;
len = (len_in + ~PAGE_MASK) & PAGE_MASK;
/* Check to see whether len was rounded up from small -ve to zero */
if (len_in && !len)
return -EINVAL;
end = start + len;
if (end < start)
return -EINVAL;
if (end == start)
return 0;
2022-03-05 04:28:51 +00:00
return madvise_walk_vmas(mm, start, end, (unsigned long)anon_name,
mm: add a field to store names for private anonymous memory In many userspace applications, and especially in VM based applications like Android uses heavily, there are multiple different allocators in use. At a minimum there is libc malloc and the stack, and in many cases there are libc malloc, the stack, direct syscalls to mmap anonymous memory, and multiple VM heaps (one for small objects, one for big objects, etc.). Each of these layers usually has its own tools to inspect its usage; malloc by compiling a debug version, the VM through heap inspection tools, and for direct syscalls there is usually no way to track them. On Android we heavily use a set of tools that use an extended version of the logic covered in Documentation/vm/pagemap.txt to walk all pages mapped in userspace and slice their usage by process, shared (COW) vs. unique mappings, backing, etc. This can account for real physical memory usage even in cases like fork without exec (which Android uses heavily to share as many private COW pages as possible between processes), Kernel SamePage Merging, and clean zero pages. It produces a measurement of the pages that only exist in that process (USS, for unique), and a measurement of the physical memory usage of that process with the cost of shared pages being evenly split between processes that share them (PSS). If all anonymous memory is indistinguishable then figuring out the real physical memory usage (PSS) of each heap requires either a pagemap walking tool that can understand the heap debugging of every layer, or for every layer's heap debugging tools to implement the pagemap walking logic, in which case it is hard to get a consistent view of memory across the whole system. Tracking the information in userspace leads to all sorts of problems. It either needs to be stored inside the process, which means every process has to have an API to export its current heap information upon request, or it has to be stored externally in a filesystem that somebody needs to clean up on crashes. It needs to be readable while the process is still running, so it has to have some sort of synchronization with every layer of userspace. Efficiently tracking the ranges requires reimplementing something like the kernel vma trees, and linking to it from every layer of userspace. It requires more memory, more syscalls, more runtime cost, and more complexity to separately track regions that the kernel is already tracking. This patch adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown in /proc/pid/maps and /proc/pid/smaps as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name) Setting the name to NULL clears it. The name length limit is 80 bytes including NUL-terminator and is checked to contain only printable ascii characters (including space), except '[',']','\','$' and '`'. Ascii strings are being used to have a descriptive identifiers for vmas, which can be understood by the users reading /proc/pid/maps or /proc/pid/smaps. Names can be standardized for a given system and they can include some variable parts such as the name of the allocator or a library, tid of the thread using it, etc. The name is stored in a pointer in the shared union in vm_area_struct that points to a null terminated string. Anonymous vmas with the same name (equivalent strings) and are otherwise mergeable will be merged. The name pointers are not shared between vmas even if they contain the same name. The name pointer is stored in a union with fields that are only used on file-backed mappings, so it does not increase memory usage. CONFIG_ANON_VMA_NAME kernel configuration is introduced to enable this feature. It keeps the feature disabled by default to prevent any additional memory overhead and to avoid confusing procfs parsers on systems which are not ready to support named anonymous vmas. The patch is based on the original patch developed by Colin Cross, more specifically on its latest version [1] posted upstream by Sumit Semwal. It used a userspace pointer to store vma names. In that design, name pointers could be shared between vmas. However during the last upstreaming attempt, Kees Cook raised concerns [2] about this approach and suggested to copy the name into kernel memory space, perform validity checks [3] and store as a string referenced from vm_area_struct. One big concern is about fork() performance which would need to strdup anonymous vma names. Dave Hansen suggested experimenting with worst-case scenario of forking a process with 64k vmas having longest possible names [4]. I ran this experiment on an ARM64 Android device and recorded a worst-case regression of almost 40% when forking such a process. This regression is addressed in the followup patch which replaces the pointer to a name with a refcounted structure that allows sharing the name pointer between vmas of the same name. Instead of duplicating the string during fork() or when splitting a vma it increments the refcount. [1] https://lore.kernel.org/linux-mm/20200901161459.11772-4-sumit.semwal@linaro.org/ [2] https://lore.kernel.org/linux-mm/202009031031.D32EF57ED@keescook/ [3] https://lore.kernel.org/linux-mm/202009031022.3834F692@keescook/ [4] https://lore.kernel.org/linux-mm/5d0358ab-8c47-2f5f-8e43-23b89d6a8e95@intel.com/ Changes for prctl(2) manual page (in the options section): PR_SET_VMA Sets an attribute specified in arg2 for virtual memory areas starting from the address specified in arg3 and spanning the size specified in arg4. arg5 specifies the value of the attribute to be set. Note that assigning an attribute to a virtual memory area might prevent it from being merged with adjacent virtual memory areas due to the difference in that attribute's value. Currently, arg2 must be one of: PR_SET_VMA_ANON_NAME Set a name for anonymous virtual memory areas. arg5 should be a pointer to a null-terminated string containing the name. The name length including null byte cannot exceed 80 bytes. If arg5 is NULL, the name of the appropriate anonymous virtual memory areas will be reset. The name can contain only printable ascii characters (including space), except '[',']','\','$' and '`'. This feature is available only if the kernel is built with the CONFIG_ANON_VMA_NAME option enabled. [surenb@google.com: docs: proc.rst: /proc/PID/maps: fix malformed table] Link: https://lkml.kernel.org/r/20211123185928.2513763-1-surenb@google.com [surenb: rebased over v5.15-rc6, replaced userpointer with a kernel copy, added input sanitization and CONFIG_ANON_VMA_NAME config. The bulk of the work here was done by Colin Cross, therefore, with his permission, keeping him as the author] Link: https://lkml.kernel.org/r/20211019215511.3771969-2-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Rientjes <rientjes@google.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Stultz <john.stultz@linaro.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rob Landley <rob@landley.net> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: Shaohua Li <shli@fusionio.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:59 +00:00
madvise_vma_anon_name);
}
#endif /* CONFIG_ANON_VMA_NAME */
/*
* The madvise(2) system call.
*
* Applications can use madvise() to advise the kernel how it should
* handle paging I/O in this VM area. The idea is to help the kernel
* use appropriate read-ahead and caching techniques. The information
* provided is advisory only, and can be safely disregarded by the
* kernel without affecting the correct operation of the application.
*
* behavior values:
* MADV_NORMAL - the default behavior is to read clusters. This
* results in some read-ahead and read-behind.
* MADV_RANDOM - the system should read the minimum amount of data
* on any access, since it is unlikely that the appli-
* cation will need more than what it asks for.
* MADV_SEQUENTIAL - pages in the given range will probably be accessed
* once, so they can be aggressively read ahead, and
* can be freed soon after they are accessed.
* MADV_WILLNEED - the application is notifying the system to read
* some pages ahead.
* MADV_DONTNEED - the application is finished with the given range,
* so the kernel can free resources associated with it.
* MADV_FREE - the application marks pages in the given range as lazy free,
* where actual purges are postponed until memory pressure happens.
[PATCH] madvise(MADV_REMOVE): remove pages from tmpfs shm backing store Here is the patch to implement madvise(MADV_REMOVE) - which frees up a given range of pages & its associated backing store. Current implementation supports only shmfs/tmpfs and other filesystems return -ENOSYS. "Some app allocates large tmpfs files, then when some task quits and some client disconnect, some memory can be released. However the only way to release tmpfs-swap is to MADV_REMOVE". - Andrea Arcangeli Databases want to use this feature to drop a section of their bufferpool (shared memory segments) - without writing back to disk/swap space. This feature is also useful for supporting hot-plug memory on UML. Concerns raised by Andrew Morton: - "We have no plan for holepunching! If we _do_ have such a plan (or might in the future) then what would the API look like? I think sys_holepunch(fd, start, len), so we should start out with that." - Using madvise is very weird, because people will ask "why do I need to mmap my file before I can stick a hole in it?" - None of the other madvise operations call into the filesystem in this manner. A broad question is: is this capability an MM operation or a filesytem operation? truncate, for example, is a filesystem operation which sometimes has MM side-effects. madvise is an mm operation and with this patch, it gains FS side-effects, only they're really, really significant ones." Comments: - Andrea suggested the fs operation too but then it's more efficient to have it as a mm operation with fs side effects, because they don't immediatly know fd and physical offset of the range. It's possible to fixup in userland and to use the fs operation but it's more expensive, the vmas are already in the kernel and we can use them. Short term plan & Future Direction: - We seem to need this interface only for shmfs/tmpfs files in the short term. We have to add hooks into the filesystem for correctness and completeness. This is what this patch does. - In the future, plan is to support both fs and mmap apis also. This also involves (other) filesystem specific functions to be implemented. - Current patch doesn't support VM_NONLINEAR - which can be addressed in the future. Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com> Cc: Hugh Dickins <hugh@veritas.com> Cc: Andrea Arcangeli <andrea@suse.de> Cc: Michael Kerrisk <mtk-manpages@gmx.net> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 08:10:38 +00:00
* MADV_REMOVE - the application wants to free up the given range of
* pages and associated backing store.
* MADV_DONTFORK - omit this area from child's address space when forking:
* typically, to avoid COWing pages pinned by get_user_pages().
* MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
* MADV_WIPEONFORK - present the child process with zero-filled memory in this
* range after a fork.
* MADV_KEEPONFORK - undo the effect of MADV_WIPEONFORK
* MADV_HWPOISON - trigger memory error handler as if the given memory range
* were corrupted by unrecoverable hardware memory failure.
* MADV_SOFT_OFFLINE - try to soft-offline the given range of memory.
ksm: the mm interface to ksm This patch presents the mm interface to a dummy version of ksm.c, for better scrutiny of that interface: the real ksm.c follows later. When CONFIG_KSM is not set, madvise(2) reject MADV_MERGEABLE and MADV_UNMERGEABLE with EINVAL, since that seems more helpful than pretending that they can be serviced. But when CONFIG_KSM=y, accept them even if KSM is not currently running, and even on areas which KSM will not touch (e.g. hugetlb or shared file or special driver mappings). Like other madvices, report ENOMEM despite success if any area in the range is unmapped, and use EAGAIN to report out of memory. Define vma flag VM_MERGEABLE to identify an area on which KSM may try merging pages: leave it to ksm_madvise() to decide whether to set it. Define mm flag MMF_VM_MERGEABLE to identify an mm which might contain VM_MERGEABLE areas, to minimize callouts when forking or exiting. Based upon earlier patches by Chris Wright and Izik Eidus. Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Chris Wright <chrisw@redhat.com> Signed-off-by: Izik Eidus <ieidus@redhat.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Rik van Riel <riel@redhat.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Lee Schermerhorn <lee.schermerhorn@hp.com> Cc: Avi Kivity <avi@redhat.com> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-09-22 00:01:57 +00:00
* MADV_MERGEABLE - the application recommends that KSM try to merge pages in
* this area with pages of identical content from other such areas.
* MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
* MADV_HUGEPAGE - the application wants to back the given range by transparent
* huge pages in the future. Existing pages might be coalesced and
* new pages might be allocated as THP.
* MADV_NOHUGEPAGE - mark the given range as not worth being backed by
* transparent huge pages so the existing pages will not be
* coalesced into THP and new pages will not be allocated as THP.
mm/madvise: introduce MADV_COLLAPSE sync hugepage collapse This idea was introduced by David Rientjes[1]. Introduce a new madvise mode, MADV_COLLAPSE, that allows users to request a synchronous collapse of memory at their own expense. The benefits of this approach are: * CPU is charged to the process that wants to spend the cycles for the THP * Avoid unpredictable timing of khugepaged collapse Semantics This call is independent of the system-wide THP sysfs settings, but will fail for memory marked VM_NOHUGEPAGE. If the ranges provided span multiple VMAs, the semantics of the collapse over each VMA is independent from the others. This implies a hugepage cannot cross a VMA boundary. If collapse of a given hugepage-aligned/sized region fails, the operation may continue to attempt collapsing the remainder of memory specified. The memory ranges provided must be page-aligned, but are not required to be hugepage-aligned. If the memory ranges are not hugepage-aligned, the start/end of the range will be clamped to the first/last hugepage-aligned address covered by said range. The memory ranges must span at least one hugepage-sized region. All non-resident pages covered by the range will first be swapped/faulted-in, before being internally copied onto a freshly allocated hugepage. Unmapped pages will have their data directly initialized to 0 in the new hugepage. However, for every eligible hugepage aligned/sized region to-be collapsed, at least one page must currently be backed by memory (a PMD covering the address range must already exist). Allocation for the new hugepage may enter direct reclaim and/or compaction, regardless of VMA flags. When the system has multiple NUMA nodes, the hugepage will be allocated from the node providing the most native pages. This operation operates on the current state of the specified process and makes no persistent changes or guarantees on how pages will be mapped, constructed, or faulted in the future Return Value If all hugepage-sized/aligned regions covered by the provided range were either successfully collapsed, or were already PMD-mapped THPs, this operation will be deemed successful. On success, process_madvise(2) returns the number of bytes advised, and madvise(2) returns 0. Else, -1 is returned and errno is set to indicate the error for the most-recently attempted hugepage collapse. Note that many failures might have occurred, since the operation may continue to collapse in the event a single hugepage-sized/aligned region fails. ENOMEM Memory allocation failed or VMA not found EBUSY Memcg charging failed EAGAIN Required resource temporarily unavailable. Try again might succeed. EINVAL Other error: No PMD found, subpage doesn't have Present bit set, "Special" page no backed by struct page, VMA incorrectly sized, address not page-aligned, ... Most notable here is ENOMEM and EBUSY (new to madvise) which are intended to provide the caller with actionable feedback so they may take an appropriate fallback measure. Use Cases An immediate user of this new functionality are malloc() implementations that manage memory in hugepage-sized chunks, but sometimes subrelease memory back to the system in native-sized chunks via MADV_DONTNEED; zapping the pmd. Later, when the memory is hot, the implementation could madvise(MADV_COLLAPSE) to re-back the memory by THPs to regain hugepage coverage and dTLB performance. TCMalloc is such an implementation that could benefit from this[2]. Only privately-mapped anon memory is supported for now, but additional support for file, shmem, and HugeTLB high-granularity mappings[2] is expected. File and tmpfs/shmem support would permit: * Backing executable text by THPs. Current support provided by CONFIG_READ_ONLY_THP_FOR_FS may take a long time on a large system which might impair services from serving at their full rated load after (re)starting. Tricks like mremap(2)'ing text onto anonymous memory to immediately realize iTLB performance prevents page sharing and demand paging, both of which increase steady state memory footprint. With MADV_COLLAPSE, we get the best of both worlds: Peak upfront performance and lower RAM footprints. * Backing guest memory by hugapages after the memory contents have been migrated in native-page-sized chunks to a new host, in a userfaultfd-based live-migration stack. [1] https://lore.kernel.org/linux-mm/d098c392-273a-36a4-1a29-59731cdf5d3d@google.com/ [2] https://github.com/google/tcmalloc/tree/master/tcmalloc [jrdr.linux@gmail.com: avoid possible memory leak in failure path] Link: https://lkml.kernel.org/r/20220713024109.62810-1-jrdr.linux@gmail.com [zokeefe@google.com add missing kfree() to madvise_collapse()] Link: https://lore.kernel.org/linux-mm/20220713024109.62810-1-jrdr.linux@gmail.com/ Link: https://lkml.kernel.org/r/20220713161851.1879439-1-zokeefe@google.com [zokeefe@google.com: delay computation of hpage boundaries until use]] Link: https://lkml.kernel.org/r/20220720140603.1958773-4-zokeefe@google.com Link: https://lkml.kernel.org/r/20220706235936.2197195-10-zokeefe@google.com Signed-off-by: Zach O'Keefe <zokeefe@google.com> Signed-off-by: "Souptick Joarder (HPE)" <jrdr.linux@gmail.com> Suggested-by: David Rientjes <rientjes@google.com> Cc: Alex Shi <alex.shi@linux.alibaba.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Chris Kennelly <ckennelly@google.com> Cc: Chris Zankel <chris@zankel.net> Cc: David Hildenbrand <david@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: Hugh Dickins <hughd@google.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Pavel Begunkov <asml.silence@gmail.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com> Cc: SeongJae Park <sj@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Yang Shi <shy828301@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-07-06 23:59:27 +00:00
* MADV_COLLAPSE - synchronously coalesce pages into new THP.
* MADV_DONTDUMP - the application wants to prevent pages in the given range
* from being included in its core dump.
* MADV_DODUMP - cancel MADV_DONTDUMP: no longer exclude from core dump.
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
* MADV_COLD - the application is not expected to use this memory soon,
* deactivate pages in this range so that they can be reclaimed
* easily if memory pressure happens.
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
* MADV_PAGEOUT - the application is not expected to use this memory soon,
* page out the pages in this range immediately.
mm/madvise: introduce MADV_POPULATE_(READ|WRITE) to prefault page tables I. Background: Sparse Memory Mappings When we manage sparse memory mappings dynamically in user space - also sometimes involving MAP_NORESERVE - we want to dynamically populate/ discard memory inside such a sparse memory region. Example users are hypervisors (especially implementing memory ballooning or similar technologies like virtio-mem) and memory allocators. In addition, we want to fail in a nice way (instead of generating SIGBUS) if populating does not succeed because we are out of backend memory (which can happen easily with file-based mappings, especially tmpfs and hugetlbfs). While MADV_DONTNEED, MADV_REMOVE and FALLOC_FL_PUNCH_HOLE allow for reliably discarding memory for most mapping types, there is no generic approach to populate page tables and preallocate memory. Although mmap() supports MAP_POPULATE, it is not applicable to the concept of sparse memory mappings, where we want to populate/discard dynamically and avoid expensive/problematic remappings. In addition, we never actually report errors during the final populate phase - it is best-effort only. fallocate() can be used to preallocate file-based memory and fail in a safe way. However, it cannot really be used for any private mappings on anonymous files via memfd due to COW semantics. In addition, fallocate() does not actually populate page tables, so we still always get pagefaults on first access - which is sometimes undesired (i.e., real-time workloads) and requires real prefaulting of page tables, not just a preallocation of backend storage. There might be interesting use cases for sparse memory regions along with mlockall(MCL_ONFAULT) which fallocate() cannot satisfy as it does not prefault page tables. II. On preallcoation/prefaulting from user space Because we don't have a proper interface, what applications (like QEMU and databases) end up doing is touching (i.e., reading+writing one byte to not overwrite existing data) all individual pages. However, that approach 1) Can result in wear on storage backing, because we end up reading/writing each page; this is especially a problem for dax/pmem. 2) Can result in mmap_sem contention when prefaulting via multiple threads. 3) Requires expensive signal handling, especially to catch SIGBUS in case of hugetlbfs/shmem/file-backed memory. For example, this is problematic in hypervisors like QEMU where SIGBUS handlers might already be used by other subsystems concurrently to e.g, handle hardware errors. "Simply" doing preallocation concurrently from other thread is not that easy. III. On MADV_WILLNEED Extending MADV_WILLNEED is not an option because 1. It would change the semantics: "Expect access in the near future." and "might be a good idea to read some pages" vs. "Definitely populate/ preallocate all memory and definitely fail on errors.". 2. Existing users (like virtio-balloon in QEMU when deflating the balloon) don't want populate/prealloc semantics. They treat this rather as a hint to give a little performance boost without too much overhead - and don't expect that a lot of memory might get consumed or a lot of time might be spent. IV. MADV_POPULATE_READ and MADV_POPULATE_WRITE Let's introduce MADV_POPULATE_READ and MADV_POPULATE_WRITE, inspired by MAP_POPULATE, with the following semantics: 1. MADV_POPULATE_READ can be used to prefault page tables just like manually reading each individual page. This will not break any COW mappings. The shared zero page might get mapped and no backend storage might get preallocated -- allocation might be deferred to write-fault time. Especially shared file mappings require an explicit fallocate() upfront to actually preallocate backend memory (blocks in the file system) in case the file might have holes. 2. If MADV_POPULATE_READ succeeds, all page tables have been populated (prefaulted) readable once. 3. MADV_POPULATE_WRITE can be used to preallocate backend memory and prefault page tables just like manually writing (or reading+writing) each individual page. This will break any COW mappings -- e.g., the shared zeropage is never populated. 4. If MADV_POPULATE_WRITE succeeds, all page tables have been populated (prefaulted) writable once. 5. MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot be applied to special mappings marked with VM_PFNMAP and VM_IO. Also, proper access permissions (e.g., PROT_READ, PROT_WRITE) are required. If any such mapping is encountered, madvise() fails with -EINVAL. 6. If MADV_POPULATE_READ or MADV_POPULATE_WRITE fails, some page tables might have been populated. 7. MADV_POPULATE_READ and MADV_POPULATE_WRITE will return -EHWPOISON when encountering a HW poisoned page in the range. 8. Similar to MAP_POPULATE, MADV_POPULATE_READ and MADV_POPULATE_WRITE cannot protect from the OOM (Out Of Memory) handler killing the process. While the use case for MADV_POPULATE_WRITE is fairly obvious (i.e., preallocate memory and prefault page tables for VMs), one issue is that whenever we prefault pages writable, the pages have to be marked dirty, because the CPU could dirty them any time. while not a real problem for hugetlbfs or dax/pmem, it can be a problem for shared file mappings: each page will be marked dirty and has to be written back later when evicting. MADV_POPULATE_READ allows for optimizing this scenario: Pre-read a whole mapping from backend storage without marking it dirty, such that eviction won't have to write it back. As discussed above, shared file mappings might require an explciit fallocate() upfront to achieve preallcoation+prepopulation. Although sparse memory mappings are the primary use case, this will also be useful for other preallocate/prefault use cases where MAP_POPULATE is not desired or the semantics of MAP_POPULATE are not sufficient: as one example, QEMU users can trigger preallocation/prefaulting of guest RAM after the mapping was created -- and don't want errors to be silently suppressed. Looking at the history, MADV_POPULATE was already proposed in 2013 [1], however, the main motivation back than was performance improvements -- which should also still be the case. V. Single-threaded performance comparison I did a short experiment, prefaulting page tables on completely *empty mappings/files* and repeated the experiment 10 times. The results correspond to the shortest execution time. In general, the performance benefit for huge pages is negligible with small mappings. V.1: Private mappings POPULATE_READ and POPULATE_WRITE is fastest. Note that Reading/POPULATE_READ will populate the shared zeropage where applicable -- which result in short population times. The fastest way to allocate backend storage (here: swap or huge pages) and prefault page tables is POPULATE_WRITE. V.2: Shared mappings fallocate() is fastest, however, doesn't prefault page tables. POPULATE_WRITE is faster than simple writes and read/writes. POPULATE_READ is faster than simple reads. Without a fd, the fastest way to allocate backend storage and prefault page tables is POPULATE_WRITE. With an fd, the fastest way is usually FALLOCATE+POPULATE_READ or FALLOCATE+POPULATE_WRITE respectively; one exception are actual files: FALLOCATE+Read is slightly faster than FALLOCATE+POPULATE_READ. The fastest way to allocate backend storage prefault page tables is FALLOCATE+POPULATE_WRITE -- except when dealing with actual files; then, FALLOCATE+POPULATE_READ is fastest and won't directly mark all pages as dirty. v.3: Detailed results ================================================== 2 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 0.119 ms Anon 4 KiB : Write : 0.222 ms Anon 4 KiB : Read/Write : 0.380 ms Anon 4 KiB : POPULATE_READ : 0.060 ms Anon 4 KiB : POPULATE_WRITE : 0.158 ms Memfd 4 KiB : Read : 0.034 ms Memfd 4 KiB : Write : 0.310 ms Memfd 4 KiB : Read/Write : 0.362 ms Memfd 4 KiB : POPULATE_READ : 0.039 ms Memfd 4 KiB : POPULATE_WRITE : 0.229 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.033 ms tmpfs : Write : 0.313 ms tmpfs : Read/Write : 0.406 ms tmpfs : POPULATE_READ : 0.039 ms tmpfs : POPULATE_WRITE : 0.285 ms file : Read : 0.033 ms file : Write : 0.351 ms file : Read/Write : 0.408 ms file : POPULATE_READ : 0.039 ms file : POPULATE_WRITE : 0.290 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_PRIVATE: ************************************************** Anon 4 KiB : Read : 237.940 ms Anon 4 KiB : Write : 708.409 ms Anon 4 KiB : Read/Write : 1054.041 ms Anon 4 KiB : POPULATE_READ : 124.310 ms Anon 4 KiB : POPULATE_WRITE : 572.582 ms Memfd 4 KiB : Read : 136.928 ms Memfd 4 KiB : Write : 963.898 ms Memfd 4 KiB : Read/Write : 1106.561 ms Memfd 4 KiB : POPULATE_READ : 78.450 ms Memfd 4 KiB : POPULATE_WRITE : 805.881 ms Memfd 2 MiB : Read : 357.116 ms Memfd 2 MiB : Write : 357.210 ms Memfd 2 MiB : Read/Write : 357.606 ms Memfd 2 MiB : POPULATE_READ : 356.094 ms Memfd 2 MiB : POPULATE_WRITE : 356.937 ms tmpfs : Read : 137.536 ms tmpfs : Write : 954.362 ms tmpfs : Read/Write : 1105.954 ms tmpfs : POPULATE_READ : 80.289 ms tmpfs : POPULATE_WRITE : 822.826 ms file : Read : 137.874 ms file : Write : 987.025 ms file : Read/Write : 1107.439 ms file : POPULATE_READ : 80.413 ms file : POPULATE_WRITE : 857.622 ms hugetlbfs : Read : 355.607 ms hugetlbfs : Write : 355.729 ms hugetlbfs : Read/Write : 356.127 ms hugetlbfs : POPULATE_READ : 354.585 ms hugetlbfs : POPULATE_WRITE : 355.138 ms ************************************************** 2 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 0.394 ms Anon 4 KiB : Write : 0.348 ms Anon 4 KiB : Read/Write : 0.400 ms Anon 4 KiB : POPULATE_READ : 0.326 ms Anon 4 KiB : POPULATE_WRITE : 0.273 ms Anon 2 MiB : Read : 0.030 ms Anon 2 MiB : Write : 0.030 ms Anon 2 MiB : Read/Write : 0.030 ms Anon 2 MiB : POPULATE_READ : 0.030 ms Anon 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 4 KiB : Read : 0.412 ms Memfd 4 KiB : Write : 0.372 ms Memfd 4 KiB : Read/Write : 0.419 ms Memfd 4 KiB : POPULATE_READ : 0.343 ms Memfd 4 KiB : POPULATE_WRITE : 0.288 ms Memfd 4 KiB : FALLOCATE : 0.137 ms Memfd 4 KiB : FALLOCATE+Read : 0.446 ms Memfd 4 KiB : FALLOCATE+Write : 0.330 ms Memfd 4 KiB : FALLOCATE+Read/Write : 0.454 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 0.379 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 0.268 ms Memfd 2 MiB : Read : 0.030 ms Memfd 2 MiB : Write : 0.030 ms Memfd 2 MiB : Read/Write : 0.030 ms Memfd 2 MiB : POPULATE_READ : 0.030 ms Memfd 2 MiB : POPULATE_WRITE : 0.030 ms Memfd 2 MiB : FALLOCATE : 0.030 ms Memfd 2 MiB : FALLOCATE+Read : 0.031 ms Memfd 2 MiB : FALLOCATE+Write : 0.031 ms Memfd 2 MiB : FALLOCATE+Read/Write : 0.031 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 0.030 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 0.030 ms tmpfs : Read : 0.416 ms tmpfs : Write : 0.369 ms tmpfs : Read/Write : 0.425 ms tmpfs : POPULATE_READ : 0.346 ms tmpfs : POPULATE_WRITE : 0.295 ms tmpfs : FALLOCATE : 0.139 ms tmpfs : FALLOCATE+Read : 0.447 ms tmpfs : FALLOCATE+Write : 0.333 ms tmpfs : FALLOCATE+Read/Write : 0.454 ms tmpfs : FALLOCATE+POPULATE_READ : 0.380 ms tmpfs : FALLOCATE+POPULATE_WRITE : 0.272 ms file : Read : 0.191 ms file : Write : 0.511 ms file : Read/Write : 0.524 ms file : POPULATE_READ : 0.196 ms file : POPULATE_WRITE : 0.434 ms file : FALLOCATE : 0.004 ms file : FALLOCATE+Read : 0.197 ms file : FALLOCATE+Write : 0.554 ms file : FALLOCATE+Read/Write : 0.480 ms file : FALLOCATE+POPULATE_READ : 0.201 ms file : FALLOCATE+POPULATE_WRITE : 0.381 ms hugetlbfs : Read : 0.030 ms hugetlbfs : Write : 0.030 ms hugetlbfs : Read/Write : 0.030 ms hugetlbfs : POPULATE_READ : 0.030 ms hugetlbfs : POPULATE_WRITE : 0.030 ms hugetlbfs : FALLOCATE : 0.030 ms hugetlbfs : FALLOCATE+Read : 0.031 ms hugetlbfs : FALLOCATE+Write : 0.031 ms hugetlbfs : FALLOCATE+Read/Write : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_READ : 0.030 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 0.030 ms ************************************************** 4096 MiB MAP_SHARED: ************************************************** Anon 4 KiB : Read : 1053.090 ms Anon 4 KiB : Write : 913.642 ms Anon 4 KiB : Read/Write : 1060.350 ms Anon 4 KiB : POPULATE_READ : 893.691 ms Anon 4 KiB : POPULATE_WRITE : 782.885 ms Anon 2 MiB : Read : 358.553 ms Anon 2 MiB : Write : 358.419 ms Anon 2 MiB : Read/Write : 357.992 ms Anon 2 MiB : POPULATE_READ : 357.533 ms Anon 2 MiB : POPULATE_WRITE : 357.808 ms Memfd 4 KiB : Read : 1078.144 ms Memfd 4 KiB : Write : 942.036 ms Memfd 4 KiB : Read/Write : 1100.391 ms Memfd 4 KiB : POPULATE_READ : 925.829 ms Memfd 4 KiB : POPULATE_WRITE : 804.394 ms Memfd 4 KiB : FALLOCATE : 304.632 ms Memfd 4 KiB : FALLOCATE+Read : 1163.359 ms Memfd 4 KiB : FALLOCATE+Write : 933.186 ms Memfd 4 KiB : FALLOCATE+Read/Write : 1187.304 ms Memfd 4 KiB : FALLOCATE+POPULATE_READ : 1013.660 ms Memfd 4 KiB : FALLOCATE+POPULATE_WRITE : 794.560 ms Memfd 2 MiB : Read : 358.131 ms Memfd 2 MiB : Write : 358.099 ms Memfd 2 MiB : Read/Write : 358.250 ms Memfd 2 MiB : POPULATE_READ : 357.563 ms Memfd 2 MiB : POPULATE_WRITE : 357.334 ms Memfd 2 MiB : FALLOCATE : 356.735 ms Memfd 2 MiB : FALLOCATE+Read : 358.152 ms Memfd 2 MiB : FALLOCATE+Write : 358.331 ms Memfd 2 MiB : FALLOCATE+Read/Write : 358.018 ms Memfd 2 MiB : FALLOCATE+POPULATE_READ : 357.286 ms Memfd 2 MiB : FALLOCATE+POPULATE_WRITE : 357.523 ms tmpfs : Read : 1087.265 ms tmpfs : Write : 950.840 ms tmpfs : Read/Write : 1107.567 ms tmpfs : POPULATE_READ : 922.605 ms tmpfs : POPULATE_WRITE : 810.094 ms tmpfs : FALLOCATE : 306.320 ms tmpfs : FALLOCATE+Read : 1169.796 ms tmpfs : FALLOCATE+Write : 933.730 ms tmpfs : FALLOCATE+Read/Write : 1191.610 ms tmpfs : FALLOCATE+POPULATE_READ : 1020.474 ms tmpfs : FALLOCATE+POPULATE_WRITE : 798.945 ms file : Read : 654.101 ms file : Write : 1259.142 ms file : Read/Write : 1289.509 ms file : POPULATE_READ : 661.642 ms file : POPULATE_WRITE : 1106.816 ms file : FALLOCATE : 1.864 ms file : FALLOCATE+Read : 656.328 ms file : FALLOCATE+Write : 1153.300 ms file : FALLOCATE+Read/Write : 1180.613 ms file : FALLOCATE+POPULATE_READ : 668.347 ms file : FALLOCATE+POPULATE_WRITE : 996.143 ms hugetlbfs : Read : 357.245 ms hugetlbfs : Write : 357.413 ms hugetlbfs : Read/Write : 357.120 ms hugetlbfs : POPULATE_READ : 356.321 ms hugetlbfs : POPULATE_WRITE : 356.693 ms hugetlbfs : FALLOCATE : 355.927 ms hugetlbfs : FALLOCATE+Read : 357.074 ms hugetlbfs : FALLOCATE+Write : 357.120 ms hugetlbfs : FALLOCATE+Read/Write : 356.983 ms hugetlbfs : FALLOCATE+POPULATE_READ : 356.413 ms hugetlbfs : FALLOCATE+POPULATE_WRITE : 356.266 ms ************************************************** [1] https://lkml.org/lkml/2013/6/27/698 [akpm@linux-foundation.org: coding style fixes] Link: https://lkml.kernel.org/r/20210419135443.12822-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jann Horn <jannh@google.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Matt Turner <mattst88@gmail.com> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Helge Deller <deller@gmx.de> Cc: Chris Zankel <chris@zankel.net> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Cc: Ram Pai <linuxram@us.ibm.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 01:52:28 +00:00
* MADV_POPULATE_READ - populate (prefault) page tables readable by
* triggering read faults if required
* MADV_POPULATE_WRITE - populate (prefault) page tables writable by
* triggering write faults if required
*
* return values:
* zero - success
* -EINVAL - start + len < 0, start is not page-aligned,
* "behavior" is not a valid value, or application
* is attempting to release locked or shared pages,
* or the specified address range includes file, Huge TLB,
* MAP_SHARED or VMPFNMAP range.
* -ENOMEM - addresses in the specified range are not currently
* mapped, or are outside the AS of the process.
* -EIO - an I/O error occurred while paging in data.
* -EBADF - map exists, but area maps something that isn't a file.
* -EAGAIN - a kernel resource was temporarily unavailable.
mseal: add mseal syscall The new mseal() is an syscall on 64 bit CPU, and with following signature: int mseal(void addr, size_t len, unsigned long flags) addr/len: memory range. flags: reserved. mseal() blocks following operations for the given memory range. 1> Unmapping, moving to another location, and shrinking the size, via munmap() and mremap(), can leave an empty space, therefore can be replaced with a VMA with a new set of attributes. 2> Moving or expanding a different VMA into the current location, via mremap(). 3> Modifying a VMA via mmap(MAP_FIXED). 4> Size expansion, via mremap(), does not appear to pose any specific risks to sealed VMAs. It is included anyway because the use case is unclear. In any case, users can rely on merging to expand a sealed VMA. 5> mprotect() and pkey_mprotect(). 6> Some destructive madvice() behaviors (e.g. MADV_DONTNEED) for anonymous memory, when users don't have write permission to the memory. Those behaviors can alter region contents by discarding pages, effectively a memset(0) for anonymous memory. Following input during RFC are incooperated into this patch: Jann Horn: raising awareness and providing valuable insights on the destructive madvise operations. Linus Torvalds: assisting in defining system call signature and scope. Liam R. Howlett: perf optimization. Theo de Raadt: sharing the experiences and insight gained from implementing mimmutable() in OpenBSD. Finally, the idea that inspired this patch comes from Stephen Röttger's work in Chrome V8 CFI. [jeffxu@chromium.org: add branch prediction hint, per Pedro] Link: https://lkml.kernel.org/r/20240423192825.1273679-2-jeffxu@chromium.org Link: https://lkml.kernel.org/r/20240415163527.626541-3-jeffxu@chromium.org Signed-off-by: Jeff Xu <jeffxu@chromium.org> Reviewed-by: Kees Cook <keescook@chromium.org> Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Pedro Falcato <pedro.falcato@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Guenter Roeck <groeck@chromium.org> Cc: Jann Horn <jannh@google.com> Cc: Jeff Xu <jeffxu@google.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Jorge Lucangeli Obes <jorgelo@chromium.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Muhammad Usama Anjum <usama.anjum@collabora.com> Cc: Pedro Falcato <pedro.falcato@gmail.com> Cc: Stephen Röttger <sroettger@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Amer Al Shanawany <amer.shanawany@gmail.com> Cc: Javier Carrasco <javier.carrasco.cruz@gmail.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-15 16:35:21 +00:00
* -EPERM - memory is sealed.
*/
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
int do_madvise(struct mm_struct *mm, unsigned long start, size_t len_in, int behavior)
{
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
unsigned long end;
int error;
int write;
size_t len;
struct blk_plug plug;
if (!madvise_behavior_valid(behavior))
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return -EINVAL;
if (!PAGE_ALIGNED(start))
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return -EINVAL;
len = PAGE_ALIGN(len_in);
/* Check to see whether len was rounded up from small -ve to zero */
if (len_in && !len)
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return -EINVAL;
end = start + len;
if (end < start)
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return -EINVAL;
if (end == start)
mm: rearrange madvise code to allow for reuse Patch series "mm: rearrange madvise code to allow for reuse", v11. Avoid performance regression of the new anon vma name field refcounting it. I checked the image sizes with allnoconfig builds: unpatched Linus' ToT text data bss dec hex filename 1324759 32 73928 1398719 1557bf vmlinux After the first patch is applied (madvise refactoring) text data bss dec hex filename 1322346 32 73928 1396306 154e52 vmlinux >>> 2413 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=n text data bss dec hex filename 1322337 32 73928 1396297 154e49 vmlinux >>> 2422 bytes decrease vs ToT <<< After all patches applied with CONFIG_ANON_VMA_NAME=y text data bss dec hex filename 1325228 32 73928 1399188 155994 vmlinux >>> 469 bytes increase vs ToT <<< This patch (of 3): Refactor the madvise syscall to allow for parts of it to be reused by a prctl syscall that affects vmas. Move the code that walks vmas in a virtual address range into a function that takes a function pointer as a parameter. The only caller for now is sys_madvise, which uses it to call madvise_vma_behavior on each vma, but the next patch will add an additional caller. Move handling all vma behaviors inside madvise_behavior, and rename it to madvise_vma_behavior. Move the code that updates the flags on a vma, including splitting or merging the vma as necessary, into a new function called madvise_update_vma. The next patch will add support for updating a new anon_name field as well. Link: https://lkml.kernel.org/r/20211019215511.3771969-1-surenb@google.com Signed-off-by: Colin Cross <ccross@google.com> Signed-off-by: Suren Baghdasaryan <surenb@google.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Jan Glauber <jan.glauber@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Rob Landley <rob@landley.net> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Kees Cook <keescook@chromium.org> Cc: "Serge E. Hallyn" <serge.hallyn@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Shaohua Li <shli@fusionio.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-01-14 22:05:55 +00:00
return 0;
#ifdef CONFIG_MEMORY_FAILURE
if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
return madvise_inject_error(behavior, start, start + len_in);
#endif
write = madvise_need_mmap_write(behavior);
mm: make mmap_sem for write waits killable for mm syscalls This is a follow up work for oom_reaper [1]. As the async OOM killing depends on oom_sem for read we would really appreciate if a holder for write didn't stood in the way. This patchset is changing many of down_write calls to be killable to help those cases when the writer is blocked and waiting for readers to release the lock and so help __oom_reap_task to process the oom victim. Most of the patches are really trivial because the lock is help from a shallow syscall paths where we can return EINTR trivially and allow the current task to die (note that EINTR will never get to the userspace as the task has fatal signal pending). Others seem to be easy as well as the callers are already handling fatal errors and bail and return to userspace which should be sufficient to handle the failure gracefully. I am not familiar with all those code paths so a deeper review is really appreciated. As this work is touching more areas which are not directly connected I have tried to keep the CC list as small as possible and people who I believed would be familiar are CCed only to the specific patches (all should have received the cover though). This patchset is based on linux-next and it depends on down_write_killable for rw_semaphores which got merged into tip locking/rwsem branch and it is merged into this next tree. I guess it would be easiest to route these patches via mmotm because of the dependency on the tip tree but if respective maintainers prefer other way I have no objections. I haven't covered all the mmap_write(mm->mmap_sem) instances here $ git grep "down_write(.*\<mmap_sem\>)" next/master | wc -l 98 $ git grep "down_write(.*\<mmap_sem\>)" | wc -l 62 I have tried to cover those which should be relatively easy to review in this series because this alone should be a nice improvement. Other places can be changed on top. [0] http://lkml.kernel.org/r/1456752417-9626-1-git-send-email-mhocko@kernel.org [1] http://lkml.kernel.org/r/1452094975-551-1-git-send-email-mhocko@kernel.org [2] http://lkml.kernel.org/r/1456750705-7141-1-git-send-email-mhocko@kernel.org This patch (of 18): This is the first step in making mmap_sem write waiters killable. It focuses on the trivial ones which are taking the lock early after entering the syscall and they are not changing state before. Therefore it is very easy to change them to use down_write_killable and immediately return with -EINTR. This will allow the waiter to pass away without blocking the mmap_sem which might be required to make a forward progress. E.g. the oom reaper will need the lock for reading to dismantle the OOM victim address space. The only tricky function in this patch is vm_mmap_pgoff which has many call sites via vm_mmap. To reduce the risk keep vm_mmap with the original non-killable semantic for now. vm_munmap callers do not bother checking the return value so open code it into the munmap syscall path for now for simplicity. Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Hugh Dickins <hughd@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-23 23:25:27 +00:00
if (write) {
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
if (mmap_write_lock_killable(mm))
mm: make mmap_sem for write waits killable for mm syscalls This is a follow up work for oom_reaper [1]. As the async OOM killing depends on oom_sem for read we would really appreciate if a holder for write didn't stood in the way. This patchset is changing many of down_write calls to be killable to help those cases when the writer is blocked and waiting for readers to release the lock and so help __oom_reap_task to process the oom victim. Most of the patches are really trivial because the lock is help from a shallow syscall paths where we can return EINTR trivially and allow the current task to die (note that EINTR will never get to the userspace as the task has fatal signal pending). Others seem to be easy as well as the callers are already handling fatal errors and bail and return to userspace which should be sufficient to handle the failure gracefully. I am not familiar with all those code paths so a deeper review is really appreciated. As this work is touching more areas which are not directly connected I have tried to keep the CC list as small as possible and people who I believed would be familiar are CCed only to the specific patches (all should have received the cover though). This patchset is based on linux-next and it depends on down_write_killable for rw_semaphores which got merged into tip locking/rwsem branch and it is merged into this next tree. I guess it would be easiest to route these patches via mmotm because of the dependency on the tip tree but if respective maintainers prefer other way I have no objections. I haven't covered all the mmap_write(mm->mmap_sem) instances here $ git grep "down_write(.*\<mmap_sem\>)" next/master | wc -l 98 $ git grep "down_write(.*\<mmap_sem\>)" | wc -l 62 I have tried to cover those which should be relatively easy to review in this series because this alone should be a nice improvement. Other places can be changed on top. [0] http://lkml.kernel.org/r/1456752417-9626-1-git-send-email-mhocko@kernel.org [1] http://lkml.kernel.org/r/1452094975-551-1-git-send-email-mhocko@kernel.org [2] http://lkml.kernel.org/r/1456750705-7141-1-git-send-email-mhocko@kernel.org This patch (of 18): This is the first step in making mmap_sem write waiters killable. It focuses on the trivial ones which are taking the lock early after entering the syscall and they are not changing state before. Therefore it is very easy to change them to use down_write_killable and immediately return with -EINTR. This will allow the waiter to pass away without blocking the mmap_sem which might be required to make a forward progress. E.g. the oom reaper will need the lock for reading to dismantle the OOM victim address space. The only tricky function in this patch is vm_mmap_pgoff which has many call sites via vm_mmap. To reduce the risk keep vm_mmap with the original non-killable semantic for now. vm_munmap callers do not bother checking the return value so open code it into the munmap syscall path for now for simplicity. Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Hugh Dickins <hughd@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-23 23:25:27 +00:00
return -EINTR;
} else {
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_lock(mm);
mm: make mmap_sem for write waits killable for mm syscalls This is a follow up work for oom_reaper [1]. As the async OOM killing depends on oom_sem for read we would really appreciate if a holder for write didn't stood in the way. This patchset is changing many of down_write calls to be killable to help those cases when the writer is blocked and waiting for readers to release the lock and so help __oom_reap_task to process the oom victim. Most of the patches are really trivial because the lock is help from a shallow syscall paths where we can return EINTR trivially and allow the current task to die (note that EINTR will never get to the userspace as the task has fatal signal pending). Others seem to be easy as well as the callers are already handling fatal errors and bail and return to userspace which should be sufficient to handle the failure gracefully. I am not familiar with all those code paths so a deeper review is really appreciated. As this work is touching more areas which are not directly connected I have tried to keep the CC list as small as possible and people who I believed would be familiar are CCed only to the specific patches (all should have received the cover though). This patchset is based on linux-next and it depends on down_write_killable for rw_semaphores which got merged into tip locking/rwsem branch and it is merged into this next tree. I guess it would be easiest to route these patches via mmotm because of the dependency on the tip tree but if respective maintainers prefer other way I have no objections. I haven't covered all the mmap_write(mm->mmap_sem) instances here $ git grep "down_write(.*\<mmap_sem\>)" next/master | wc -l 98 $ git grep "down_write(.*\<mmap_sem\>)" | wc -l 62 I have tried to cover those which should be relatively easy to review in this series because this alone should be a nice improvement. Other places can be changed on top. [0] http://lkml.kernel.org/r/1456752417-9626-1-git-send-email-mhocko@kernel.org [1] http://lkml.kernel.org/r/1452094975-551-1-git-send-email-mhocko@kernel.org [2] http://lkml.kernel.org/r/1456750705-7141-1-git-send-email-mhocko@kernel.org This patch (of 18): This is the first step in making mmap_sem write waiters killable. It focuses on the trivial ones which are taking the lock early after entering the syscall and they are not changing state before. Therefore it is very easy to change them to use down_write_killable and immediately return with -EINTR. This will allow the waiter to pass away without blocking the mmap_sem which might be required to make a forward progress. E.g. the oom reaper will need the lock for reading to dismantle the OOM victim address space. The only tricky function in this patch is vm_mmap_pgoff which has many call sites via vm_mmap. To reduce the risk keep vm_mmap with the original non-killable semantic for now. vm_munmap callers do not bother checking the return value so open code it into the munmap syscall path for now for simplicity. Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Hugh Dickins <hughd@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-23 23:25:27 +00:00
}
start = untagged_addr_remote(mm, start);
end = start + len;
mseal: add mseal syscall The new mseal() is an syscall on 64 bit CPU, and with following signature: int mseal(void addr, size_t len, unsigned long flags) addr/len: memory range. flags: reserved. mseal() blocks following operations for the given memory range. 1> Unmapping, moving to another location, and shrinking the size, via munmap() and mremap(), can leave an empty space, therefore can be replaced with a VMA with a new set of attributes. 2> Moving or expanding a different VMA into the current location, via mremap(). 3> Modifying a VMA via mmap(MAP_FIXED). 4> Size expansion, via mremap(), does not appear to pose any specific risks to sealed VMAs. It is included anyway because the use case is unclear. In any case, users can rely on merging to expand a sealed VMA. 5> mprotect() and pkey_mprotect(). 6> Some destructive madvice() behaviors (e.g. MADV_DONTNEED) for anonymous memory, when users don't have write permission to the memory. Those behaviors can alter region contents by discarding pages, effectively a memset(0) for anonymous memory. Following input during RFC are incooperated into this patch: Jann Horn: raising awareness and providing valuable insights on the destructive madvise operations. Linus Torvalds: assisting in defining system call signature and scope. Liam R. Howlett: perf optimization. Theo de Raadt: sharing the experiences and insight gained from implementing mimmutable() in OpenBSD. Finally, the idea that inspired this patch comes from Stephen Röttger's work in Chrome V8 CFI. [jeffxu@chromium.org: add branch prediction hint, per Pedro] Link: https://lkml.kernel.org/r/20240423192825.1273679-2-jeffxu@chromium.org Link: https://lkml.kernel.org/r/20240415163527.626541-3-jeffxu@chromium.org Signed-off-by: Jeff Xu <jeffxu@chromium.org> Reviewed-by: Kees Cook <keescook@chromium.org> Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Pedro Falcato <pedro.falcato@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Guenter Roeck <groeck@chromium.org> Cc: Jann Horn <jannh@google.com> Cc: Jeff Xu <jeffxu@google.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Jorge Lucangeli Obes <jorgelo@chromium.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Muhammad Usama Anjum <usama.anjum@collabora.com> Cc: Pedro Falcato <pedro.falcato@gmail.com> Cc: Stephen Röttger <sroettger@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Amer Al Shanawany <amer.shanawany@gmail.com> Cc: Javier Carrasco <javier.carrasco.cruz@gmail.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-15 16:35:21 +00:00
/*
* Check if the address range is sealed for do_madvise().
* can_modify_mm_madv assumes we have acquired the lock on MM.
*/
if (unlikely(!can_modify_mm_madv(mm, start, end, behavior))) {
error = -EPERM;
goto out;
}
blk_start_plug(&plug);
mm/madvise: don't perform madvise VMA walk for MADV_POPULATE_(READ|WRITE) We changed faultin_page_range() to no longer consume a VMA, because faultin_page_range() might internally release the mm lock to lookup the VMA again -- required to cleanly handle VM_FAULT_RETRY. But independent of that, __get_user_pages() will always lookup the VMA itself. Now that we let __get_user_pages() just handle VMA checks in a way that is suitable for MADV_POPULATE_(READ|WRITE), the VMA walk in madvise() is just overhead. So let's just call madvise_populate() on the full range instead. There is one change in behavior: madvise_walk_vmas() would skip any VMA holes, and if everything succeeded, it would return -ENOMEM after processing all VMAs. However, for MADV_POPULATE_(READ|WRITE) it's unlikely for the caller to notice any difference: -ENOMEM might either indicate that there were VMA holes or that populating page tables failed because there was not enough memory. So it's unlikely that user space will notice the difference, and that special handling likely only makes sense for some other madvise() actions. Further, we'd already fail with -ENOMEM early in the past if looking up the VMA after dropping the MM lock failed because of concurrent VMA modifications. So let's just keep it simple and avoid the madvise VMA walk, and consistently fail early if we find a VMA hole. Link: https://lkml.kernel.org/r/20240314161300.382526-3-david@redhat.com Signed-off-by: David Hildenbrand <david@redhat.com> Cc: Darrick J. Wong <djwong@kernel.org> Cc: Hugh Dickins <hughd@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-14 16:13:00 +00:00
switch (behavior) {
case MADV_POPULATE_READ:
case MADV_POPULATE_WRITE:
error = madvise_populate(mm, start, end, behavior);
break;
default:
error = madvise_walk_vmas(mm, start, end, behavior,
madvise_vma_behavior);
break;
}
blk_finish_plug(&plug);
mseal: add mseal syscall The new mseal() is an syscall on 64 bit CPU, and with following signature: int mseal(void addr, size_t len, unsigned long flags) addr/len: memory range. flags: reserved. mseal() blocks following operations for the given memory range. 1> Unmapping, moving to another location, and shrinking the size, via munmap() and mremap(), can leave an empty space, therefore can be replaced with a VMA with a new set of attributes. 2> Moving or expanding a different VMA into the current location, via mremap(). 3> Modifying a VMA via mmap(MAP_FIXED). 4> Size expansion, via mremap(), does not appear to pose any specific risks to sealed VMAs. It is included anyway because the use case is unclear. In any case, users can rely on merging to expand a sealed VMA. 5> mprotect() and pkey_mprotect(). 6> Some destructive madvice() behaviors (e.g. MADV_DONTNEED) for anonymous memory, when users don't have write permission to the memory. Those behaviors can alter region contents by discarding pages, effectively a memset(0) for anonymous memory. Following input during RFC are incooperated into this patch: Jann Horn: raising awareness and providing valuable insights on the destructive madvise operations. Linus Torvalds: assisting in defining system call signature and scope. Liam R. Howlett: perf optimization. Theo de Raadt: sharing the experiences and insight gained from implementing mimmutable() in OpenBSD. Finally, the idea that inspired this patch comes from Stephen Röttger's work in Chrome V8 CFI. [jeffxu@chromium.org: add branch prediction hint, per Pedro] Link: https://lkml.kernel.org/r/20240423192825.1273679-2-jeffxu@chromium.org Link: https://lkml.kernel.org/r/20240415163527.626541-3-jeffxu@chromium.org Signed-off-by: Jeff Xu <jeffxu@chromium.org> Reviewed-by: Kees Cook <keescook@chromium.org> Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Pedro Falcato <pedro.falcato@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Guenter Roeck <groeck@chromium.org> Cc: Jann Horn <jannh@google.com> Cc: Jeff Xu <jeffxu@google.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Jorge Lucangeli Obes <jorgelo@chromium.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Muhammad Usama Anjum <usama.anjum@collabora.com> Cc: Pedro Falcato <pedro.falcato@gmail.com> Cc: Stephen Röttger <sroettger@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Amer Al Shanawany <amer.shanawany@gmail.com> Cc: Javier Carrasco <javier.carrasco.cruz@gmail.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-15 16:35:21 +00:00
out:
if (write)
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_write_unlock(mm);
else
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
mmap_read_unlock(mm);
return error;
}
SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
{
mm/madvise: pass mm to do_madvise Patch series "introduce memory hinting API for external process", v9. Now, we have MADV_PAGEOUT and MADV_COLD as madvise hinting API. With that, application could give hints to kernel what memory range are preferred to be reclaimed. However, in some platform(e.g., Android), the information required to make the hinting decision is not known to the app. Instead, it is known to a centralized userspace daemon(e.g., ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the concern, this patch introduces new syscall - process_madvise(2). Bascially, it's same with madvise(2) syscall but it has some differences. 1. It needs pidfd of target process to provide the hint 2. It supports only MADV_{COLD|PAGEOUT|MERGEABLE|UNMEREABLE} at this moment. Other hints in madvise will be opened when there are explicit requests from community to prevent unexpected bugs we couldn't support. 3. Only privileged processes can do something for other process's address space. For more detail of the new API, please see "mm: introduce external memory hinting API" description in this patchset. This patch (of 3): In upcoming patches, do_madvise will be called from external process context so we shouldn't asssume "current" is always hinted process's task_struct. Furthermore, we must not access mm_struct via task->mm, but obtain it via access_mm() once (in the following patch) and only use that pointer [1], so pass it to do_madvise() as well. Note the vma->vm_mm pointers are safe, so we can use them further down the call stack. And let's pass current->mm as arguments of do_madvise so it shouldn't change existing behavior but prepare next patch to make review easy. [vbabka@suse.cz: changelog tweak] [minchan@kernel.org: use current->mm for io_uring] Link: http://lkml.kernel.org/r/20200423145215.72666-1-minchan@kernel.org [akpm@linux-foundation.org: fix it for upstream changes] [akpm@linux-foundation.org: whoops] [rdunlap@infradead.org: add missing includes] Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Jann Horn <jannh@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Daniel Colascione <dancol@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: John Dias <joaodias@google.com> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: Christian Brauner <christian@brauner.io> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-1-minchan@kernel.org Link: http://lkml.kernel.org/r/20200302193630.68771-2-minchan@kernel.org Link: http://lkml.kernel.org/r/20200622192900.22757-2-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-2-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:50 +00:00
return do_madvise(current->mm, start, len_in, behavior);
}
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
size_t, vlen, int, behavior, unsigned int, flags)
{
ssize_t ret;
struct iovec iovstack[UIO_FASTIOV];
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
struct iovec *iov = iovstack;
struct iov_iter iter;
struct task_struct *task;
struct mm_struct *mm;
size_t total_len;
unsigned int f_flags;
if (flags != 0) {
ret = -EINVAL;
goto out;
}
ret = import_iovec(ITER_DEST, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
if (ret < 0)
goto out;
task = pidfd_get_task(pidfd, &f_flags);
if (IS_ERR(task)) {
ret = PTR_ERR(task);
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
goto free_iov;
}
if (!process_madvise_behavior_valid(behavior)) {
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
ret = -EINVAL;
goto release_task;
}
mm/madvise: replace ptrace attach requirement for process_madvise process_madvise currently requires ptrace attach capability. PTRACE_MODE_ATTACH gives one process complete control over another process. It effectively removes the security boundary between the two processes (in one direction). Granting ptrace attach capability even to a system process is considered dangerous since it creates an attack surface. This severely limits the usage of this API. The operations process_madvise can perform do not affect the correctness of the operation of the target process; they only affect where the data is physically located (and therefore, how fast it can be accessed). What we want is the ability for one process to influence another process in order to optimize performance across the entire system while leaving the security boundary intact. Replace PTRACE_MODE_ATTACH with a combination of PTRACE_MODE_READ and CAP_SYS_NICE. PTRACE_MODE_READ to prevent leaking ASLR metadata and CAP_SYS_NICE for influencing process performance. Link: https://lkml.kernel.org/r/20210303185807.2160264-1-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: David Rientjes <rientjes@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jeff Vander Stoep <jeffv@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Florian Weimer <fweimer@redhat.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: James Morris <jmorris@namei.org> Cc: <stable@vger.kernel.org> [5.10+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-03-13 05:08:06 +00:00
/* Require PTRACE_MODE_READ to avoid leaking ASLR metadata. */
mm = mm_access(task, PTRACE_MODE_READ_FSCREDS);
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
if (IS_ERR_OR_NULL(mm)) {
ret = IS_ERR(mm) ? PTR_ERR(mm) : -ESRCH;
goto release_task;
}
mm/madvise: replace ptrace attach requirement for process_madvise process_madvise currently requires ptrace attach capability. PTRACE_MODE_ATTACH gives one process complete control over another process. It effectively removes the security boundary between the two processes (in one direction). Granting ptrace attach capability even to a system process is considered dangerous since it creates an attack surface. This severely limits the usage of this API. The operations process_madvise can perform do not affect the correctness of the operation of the target process; they only affect where the data is physically located (and therefore, how fast it can be accessed). What we want is the ability for one process to influence another process in order to optimize performance across the entire system while leaving the security boundary intact. Replace PTRACE_MODE_ATTACH with a combination of PTRACE_MODE_READ and CAP_SYS_NICE. PTRACE_MODE_READ to prevent leaking ASLR metadata and CAP_SYS_NICE for influencing process performance. Link: https://lkml.kernel.org/r/20210303185807.2160264-1-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: David Rientjes <rientjes@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jeff Vander Stoep <jeffv@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Florian Weimer <fweimer@redhat.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: James Morris <jmorris@namei.org> Cc: <stable@vger.kernel.org> [5.10+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-03-13 05:08:06 +00:00
/*
* Require CAP_SYS_NICE for influencing process performance. Note that
* only non-destructive hints are currently supported.
*/
if (!capable(CAP_SYS_NICE)) {
ret = -EPERM;
goto release_mm;
}
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
total_len = iov_iter_count(&iter);
while (iov_iter_count(&iter)) {
ret = do_madvise(mm, (unsigned long)iter_iov_addr(&iter),
iter_iov_len(&iter), behavior);
if (ret < 0)
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
break;
iov_iter_advance(&iter, iter_iov_len(&iter));
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
}
mm: madvise: return correct bytes advised with process_madvise Patch series "mm: madvise: return correct bytes processed with process_madvise", v2. With the process_madvise(), always choose to return non zero processed bytes over an error. This can help the user to know on which VMA, passed in the 'struct iovec' vector list, is failed to advise thus can take the decission of retrying/skipping on that VMA. This patch (of 2): The process_madvise() system call returns error even after processing some VMA's passed in the 'struct iovec' vector list which leaves the user confused to know where to restart the advise next. It is also against this syscall man page[1] documentation where it mentions that "return value may be less than the total number of requested bytes, if an error occurred after some iovec elements were already processed.". Consider a user passed 10 VMA's in the 'struct iovec' vector list of which 9 are processed but one. Then it just returns the error caused on that failed VMA despite the first 9 VMA's processed, leaving the user confused about on which VMA it is failed. Returning the number of bytes processed here can help the user to know which VMA it is failed on and thus can retry/skip the advise on that VMA. [1]https://man7.org/linux/man-pages/man2/process_madvise.2.html. Link: https://lkml.kernel.org/r/cover.1647008754.git.quic_charante@quicinc.com Link: https://lkml.kernel.org/r/125b61a0edcee5c2db8658aed9d06a43a19ccafc.1647008754.git.quic_charante@quicinc.com Fixes: ecb8ac8b1f14("mm/madvise: introduce process_madvise() syscall: an external memory hinting API") Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: David Rientjes <rientjes@google.com> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Minchan Kim <minchan@kernel.org> Cc: Nadav Amit <nadav.amit@gmail.com> Cc: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2022-03-22 21:46:44 +00:00
ret = (total_len - iov_iter_count(&iter)) ? : ret;
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
mm/madvise: replace ptrace attach requirement for process_madvise process_madvise currently requires ptrace attach capability. PTRACE_MODE_ATTACH gives one process complete control over another process. It effectively removes the security boundary between the two processes (in one direction). Granting ptrace attach capability even to a system process is considered dangerous since it creates an attack surface. This severely limits the usage of this API. The operations process_madvise can perform do not affect the correctness of the operation of the target process; they only affect where the data is physically located (and therefore, how fast it can be accessed). What we want is the ability for one process to influence another process in order to optimize performance across the entire system while leaving the security boundary intact. Replace PTRACE_MODE_ATTACH with a combination of PTRACE_MODE_READ and CAP_SYS_NICE. PTRACE_MODE_READ to prevent leaking ASLR metadata and CAP_SYS_NICE for influencing process performance. Link: https://lkml.kernel.org/r/20210303185807.2160264-1-surenb@google.com Signed-off-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Kees Cook <keescook@chromium.org> Acked-by: Minchan Kim <minchan@kernel.org> Acked-by: David Rientjes <rientjes@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jeff Vander Stoep <jeffv@google.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Florian Weimer <fweimer@redhat.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: James Morris <jmorris@namei.org> Cc: <stable@vger.kernel.org> [5.10+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-03-13 05:08:06 +00:00
release_mm:
mm/madvise: introduce process_madvise() syscall: an external memory hinting API There is usecase that System Management Software(SMS) want to give a memory hint like MADV_[COLD|PAGEEOUT] to other processes and in the case of Android, it is the ActivityManagerService. The information required to make the reclaim decision is not known to the app. Instead, it is known to the centralized userspace daemon(ActivityManagerService), and that daemon must be able to initiate reclaim on its own without any app involvement. To solve the issue, this patch introduces a new syscall process_madvise(2). It uses pidfd of an external process to give the hint. It also supports vector address range because Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma.(With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment). Another potential use case for the vector range is to amortize the cost ofTLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature. ince it could affect other process's address range, only privileged process(PTRACE_MODE_ATTACH_FSCREDS) or something else(e.g., being the same UID) gives it the right to ptrace the process could use it successfully. The flag argument is reserved for future use if we need to extend the API. I think supporting all hints madvise has/will supported/support to process_madvise is rather risky. Because we are not sure all hints make sense from external process and implementation for the hint may rely on the caller being in the current context so it could be error-prone. Thus, I just limited hints as MADV_[COLD|PAGEOUT] in this patch. If someone want to add other hints, we could hear the usecase and review it for each hint. It's safer for maintenance rather than introducing a buggy syscall but hard to fix it later. So finally, the API is as follows, ssize_t process_madvise(int pidfd, const struct iovec *iovec, unsigned long vlen, int advice, unsigned int flags); DESCRIPTION The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance. The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information) The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as: struct iovec { void *iov_base; /* starting address */ size_t iov_len; /* number of bytes to be advised */ }; The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len). The vlen represents the number of elements in iovec. The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external. MADV_COLD MADV_PAGEOUT Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2). The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges. RETURN VALUE On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred. FAQ: Q.1 - Why does any external entity have better knowledge? Quote from Sandeep "For Android, every application (including the special SystemServer) are forked from Zygote. The reason of course is to share as many libraries and classes between the two as possible to benefit from the preloading during boot. After applications start, (almost) all of the APIs end up calling into this SystemServer process over IPC (binder) and back to the application. In a fully running system, the SystemServer monitors every single process periodically to calculate their PSS / RSS and also decides which process is "important" to the user for interactivity. So, because of how these processes start _and_ the fact that the SystemServer is looping to monitor each process, it does tend to *know* which address range of the application is not used / useful. Besides, we can never rely on applications to clean things up themselves. We've had the "hey app1, the system is low on memory, please trim your memory usage down" notifications for a long time[1]. They rely on applications honoring the broadcasts and very few do. So, if we want to avoid the inevitable killing of the application and restarting it, some way to be able to tell the OS about unimportant memory in these applications will be useful. - ssp Q.2 - How to guarantee the race(i.e., object validation) between when giving a hint from an external process and get the hint from the target process? process_madvise operates on the target process's address space as it exists at the instant that process_madvise is called. If the space target process can run between the time the process_madvise process inspects the target process address space and the time that process_madvise is actually called, process_madvise may operate on memory regions that the calling process does not expect. It's the responsibility of the process calling process_madvise to close this race condition. For example, the calling process can suspend the target process with ptrace, SIGSTOP, or the freezer cgroup so that it doesn't have an opportunity to change its own address space before process_madvise is called. Another option is to operate on memory regions that the caller knows a priori will be unchanged in the target process. Yet another option is to accept the race for certain process_madvise calls after reasoning that mistargeting will do no harm. The suggested API itself does not provide synchronization. It also apply other APIs like move_pages, process_vm_write. The race isn't really a problem though. Why is it so wrong to require that callers do their own synchronization in some manner? Nobody objects to write(2) merely because it's possible for two processes to open the same file and clobber each other's writes --- instead, we tell people to use flock or something. Think about mmap. It never guarantees newly allocated address space is still valid when the user tries to access it because other threads could unmap the memory right before. That's where we need synchronization by using other API or design from userside. It shouldn't be part of API itself. If someone needs more fine-grained synchronization rather than process level, there were two ideas suggested - cookie[2] and anon-fd[3]. Both are applicable via using last reserved argument of the API but I don't think it's necessary right now since we have already ways to prevent the race so don't want to add additional complexity with more fine-grained optimization model. To make the API extend, it reserved an unsigned long as last argument so we could support it in future if someone really needs it. Q.3 - Why doesn't ptrace work? Injecting an madvise in the target process using ptrace would not work for us because such injected madvise would have to be executed by the target process, which means that process would have to be runnable and that creates the risk of the abovementioned race and hinting a wrong VMA. Furthermore, we want to act the hint in caller's context, not the callee's, because the callee is usually limited in cpuset/cgroups or even freezed state so they can't act by themselves quick enough, which causes more thrashing/kill. It doesn't work if the target process are ptraced(e.g., strace, debugger, minidump) because a process can have at most one ptracer. [1] https://developer.android.com/topic/performance/memory" [2] process_getinfo for getting the cookie which is updated whenever vma of process address layout are changed - Daniel Colascione - https://lore.kernel.org/lkml/20190520035254.57579-1-minchan@kernel.org/T/#m7694416fd179b2066a2c62b5b139b14e3894e224 [3] anonymous fd which is used for the object(i.e., address range) validation - Michal Hocko - https://lore.kernel.org/lkml/20200120112722.GY18451@dhcp22.suse.cz/ [minchan@kernel.org: fix process_madvise build break for arm64] Link: http://lkml.kernel.org/r/20200303145756.GA219683@google.com [minchan@kernel.org: fix build error for mips of process_madvise] Link: http://lkml.kernel.org/r/20200508052517.GA197378@google.com [akpm@linux-foundation.org: fix patch ordering issue] [akpm@linux-foundation.org: fix arm64 whoops] [minchan@kernel.org: make process_madvise() vlen arg have type size_t, per Florian] [akpm@linux-foundation.org: fix i386 build] [sfr@canb.auug.org.au: fix syscall numbering] Link: https://lkml.kernel.org/r/20200905142639.49fc3f1a@canb.auug.org.au [sfr@canb.auug.org.au: madvise.c needs compat.h] Link: https://lkml.kernel.org/r/20200908204547.285646b4@canb.auug.org.au [minchan@kernel.org: fix mips build] Link: https://lkml.kernel.org/r/20200909173655.GC2435453@google.com [yuehaibing@huawei.com: remove duplicate header which is included twice] Link: https://lkml.kernel.org/r/20200915121550.30584-1-yuehaibing@huawei.com [minchan@kernel.org: do not use helper functions for process_madvise] Link: https://lkml.kernel.org/r/20200921175539.GB387368@google.com [akpm@linux-foundation.org: pidfd_get_pid() gained an argument] [sfr@canb.auug.org.au: fix up for "iov_iter: transparently handle compat iovecs in import_iovec"] Link: https://lkml.kernel.org/r/20200928212542.468e1fef@canb.auug.org.au Signed-off-by: Minchan Kim <minchan@kernel.org> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Suren Baghdasaryan <surenb@google.com> Reviewed-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: David Rientjes <rientjes@google.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Christian Brauner <christian@brauner.io> Cc: Daniel Colascione <dancol@google.com> Cc: Jann Horn <jannh@google.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joel Fernandes <joel@joelfernandes.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: John Dias <joaodias@google.com> Cc: Kirill Tkhai <ktkhai@virtuozzo.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oleksandr Natalenko <oleksandr@redhat.com> Cc: Sandeep Patil <sspatil@google.com> Cc: SeongJae Park <sj38.park@gmail.com> Cc: SeongJae Park <sjpark@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Sonny Rao <sonnyrao@google.com> Cc: Tim Murray <timmurray@google.com> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fw@deneb.enyo.de> Cc: <linux-man@vger.kernel.org> Link: http://lkml.kernel.org/r/20200302193630.68771-3-minchan@kernel.org Link: http://lkml.kernel.org/r/20200508183320.GA125527@google.com Link: http://lkml.kernel.org/r/20200622192900.22757-4-minchan@kernel.org Link: https://lkml.kernel.org/r/20200901000633.1920247-4-minchan@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-17 23:14:59 +00:00
mmput(mm);
release_task:
put_task_struct(task);
free_iov:
kfree(iov);
out:
return ret;
}