freebsd-src/sys/vm/vm_page.h

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1025 lines
37 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: (BSD-3-Clause AND MIT-CMU)
*
1994-05-24 10:09:53 +00:00
* Copyright (c) 1991, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* The Mach Operating System project at Carnegie-Mellon University.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
1994-05-24 10:09:53 +00:00
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
*
* Copyright (c) 1987, 1990 Carnegie-Mellon University.
* All rights reserved.
*
* Authors: Avadis Tevanian, Jr., Michael Wayne Young
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
*
1994-05-24 10:09:53 +00:00
* Permission to use, copy, modify and distribute this software and
* its documentation is hereby granted, provided that both the copyright
* notice and this permission notice appear in all copies of the
* software, derivative works or modified versions, and any portions
* thereof, and that both notices appear in supporting documentation.
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
*
* CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
* CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
1994-05-24 10:09:53 +00:00
* FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
*
1994-05-24 10:09:53 +00:00
* Carnegie Mellon requests users of this software to return to
*
* Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
* School of Computer Science
* Carnegie Mellon University
* Pittsburgh PA 15213-3890
*
* any improvements or extensions that they make and grant Carnegie the
* rights to redistribute these changes.
*/
/*
* Resident memory system definitions.
*/
#ifndef _VM_PAGE_
#define _VM_PAGE_
#include <vm/pmap.h>
#include <vm/_vm_phys.h>
1994-05-24 10:09:53 +00:00
/*
* Management of resident (logical) pages.
*
* A small structure is kept for each resident
* page, indexed by page number. Each structure
2013-06-04 05:44:52 +00:00
* is an element of several collections:
1994-05-24 10:09:53 +00:00
*
2013-06-04 05:44:52 +00:00
* A radix tree used to quickly
1994-05-24 10:09:53 +00:00
* perform object/offset lookups
*
* A list of all pages for a given object,
* so they can be quickly deactivated at
* time of deallocation.
*
* An ordered list of pages due for pageout.
*
* In addition, the structure contains the object
* and offset to which this page belongs (for pageout),
* and sundry status bits.
*
* In general, operations on this structure's mutable fields are
* synchronized using either one of or a combination of locks. If a
* field is annotated with two of these locks then holding either is
* sufficient for read access but both are required for write access.
* The queue lock for a page depends on the value of its queue field and is
* described in detail below.
*
* The following annotations are possible:
* (A) the field must be accessed using atomic(9) and may require
* additional synchronization.
* (B) the page busy lock.
* (C) the field is immutable.
* (F) the per-domain lock for the free queues.
* (M) Machine dependent, defined by pmap layer.
* (O) the object that the page belongs to.
* (Q) the page's queue lock.
*
* The busy lock is an embedded reader-writer lock that protects the
* page's contents and identity (i.e., its <object, pindex> tuple) as
* well as certain valid/dirty modifications. To avoid bloating the
* the page structure, the busy lock lacks some of the features available
* the kernel's general-purpose synchronization primitives. As a result,
* busy lock ordering rules are not verified, lock recursion is not
* detected, and an attempt to xbusy a busy page or sbusy an xbusy page
* results will trigger a panic rather than causing the thread to block.
* vm_page_sleep_if_busy() can be used to sleep until the page's busy
* state changes, after which the caller must re-lookup the page and
* re-evaluate its state. vm_page_busy_acquire() will block until
* the lock is acquired.
*
* The valid field is protected by the page busy lock (B) and object
* lock (O). Transitions from invalid to valid are generally done
* via I/O or zero filling and do not require the object lock.
* These must be protected with the busy lock to prevent page-in or
* creation races. Page invalidation generally happens as a result
* of truncate or msync. When invalidated, pages must not be present
* in pmap and must hold the object lock to prevent concurrent
* speculative read-only mappings that do not require busy. I/O
* routines may check for validity without a lock if they are prepared
* to handle invalidation races with higher level locks (vnode) or are
* unconcerned with races so long as they hold a reference to prevent
* recycling. When a valid bit is set while holding a shared busy
* lock (A) atomic operations are used to protect against concurrent
* modification.
*
* In contrast, the synchronization of accesses to the page's
* dirty field is a mix of machine dependent (M) and busy (B). In
* the machine-independent layer, the page busy must be held to
* operate on the field. However, the pmap layer is permitted to
* set all bits within the field without holding that lock. If the
* underlying architecture does not support atomic read-modify-write
* operations on the field's type, then the machine-independent
* layer uses a 32-bit atomic on the aligned 32-bit word that
* contains the dirty field. In the machine-independent layer,
* the implementation of read-modify-write operations on the
* field is encapsulated in vm_page_clear_dirty_mask(). An
* exclusive busy lock combined with pmap_remove_{write/all}() is the
* only way to ensure a page can not become dirty. I/O generally
* removes the page from pmap to ensure exclusive access and atomic
* writes.
*
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
* The ref_count field tracks references to the page. References that
* prevent the page from being reclaimable are called wirings and are
* counted in the low bits of ref_count. The containing object's
* reference, if one exists, is counted using the VPRC_OBJREF bit in the
* ref_count field. Additionally, the VPRC_BLOCKED bit is used to
* atomically check for wirings and prevent new wirings via
* pmap_extract_and_hold(). When a page belongs to an object, it may be
* wired only when the object is locked, or the page is busy, or by
* pmap_extract_and_hold(). As a result, if the object is locked and the
* page is not busy (or is exclusively busied by the current thread), and
* the page is unmapped, its wire count will not increase. The ref_count
* field is updated using atomic operations in most cases, except when it
* is known that no other references to the page exist, such as in the page
* allocator. A page may be present in the page queues, or even actively
* scanned by the page daemon, without an explicitly counted referenced.
* The page daemon must therefore handle the possibility of a concurrent
* free of the page.
*
* The queue state of a page consists of the queue and act_count fields of
* its atomically updated state, and the subset of atomic flags specified
* by PGA_QUEUE_STATE_MASK. The queue field contains the page's page queue
* index, or PQ_NONE if it does not belong to a page queue. To modify the
* queue field, the page queue lock corresponding to the old value must be
* held, unless that value is PQ_NONE, in which case the queue index must
* be updated using an atomic RMW operation. There is one exception to
* this rule: the page daemon may transition the queue field from
* PQ_INACTIVE to PQ_NONE immediately prior to freeing the page during an
* inactive queue scan. At that point the page is already dequeued and no
* other references to that vm_page structure can exist. The PGA_ENQUEUED
* flag, when set, indicates that the page structure is physically inserted
* into the queue corresponding to the page's queue index, and may only be
* set or cleared with the corresponding page queue lock held.
*
* To avoid contention on page queue locks, page queue operations (enqueue,
* dequeue, requeue) are batched using fixed-size per-CPU queues. A
* deferred operation is requested by setting one of the flags in
* PGA_QUEUE_OP_MASK and inserting an entry into a batch queue. When a
* queue is full, an attempt to insert a new entry will lock the page
* queues and trigger processing of the pending entries. The
* type-stability of vm_page structures is crucial to this scheme since the
* processing of entries in a given batch queue may be deferred
* indefinitely. In particular, a page may be freed with pending batch
* queue entries. The page queue operation flags must be set using atomic
* RWM operations.
1994-05-24 10:09:53 +00:00
*/
#if PAGE_SIZE == 4096
#define VM_PAGE_BITS_ALL 0xffu
typedef uint8_t vm_page_bits_t;
#elif PAGE_SIZE == 8192
#define VM_PAGE_BITS_ALL 0xffffu
typedef uint16_t vm_page_bits_t;
#elif PAGE_SIZE == 16384
#define VM_PAGE_BITS_ALL 0xffffffffu
typedef uint32_t vm_page_bits_t;
#elif PAGE_SIZE == 32768
#define VM_PAGE_BITS_ALL 0xfffffffffffffffflu
typedef uint64_t vm_page_bits_t;
#endif
typedef union vm_page_astate {
struct {
uint16_t flags;
uint8_t queue;
uint8_t act_count;
};
uint32_t _bits;
} vm_page_astate_t;
1994-05-24 10:09:53 +00:00
struct vm_page {
union {
TAILQ_ENTRY(vm_page) q; /* page queue or free list (Q) */
struct {
SLIST_ENTRY(vm_page) ss; /* private slists */
} s;
struct {
u_long p;
u_long v;
} memguard;
struct {
void *slab;
void *zone;
} uma;
} plinks;
TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
vm_object_t object; /* which object am I in (O) */
vm_pindex_t pindex; /* offset into object (O,P) */
vm_paddr_t phys_addr; /* physical address of page (C) */
struct md_page md; /* machine dependent stuff */
u_int ref_count; /* page references (A) */
u_int busy_lock; /* busy owners lock (A) */
union vm_page_astate a; /* state accessed atomically (A) */
uint8_t order; /* index of the buddy queue (F) */
uint8_t pool; /* vm_phys freepool index (F) */
uint8_t flags; /* page PG_* flags (P) */
uint8_t oflags; /* page VPO_* flags (O) */
int8_t psind; /* pagesizes[] index (O) */
int8_t segind; /* vm_phys segment index (C) */
/* NOTE that these must support one bit per DEV_BSIZE in a page */
/* so, on normal X86 kernels, they must be at least 8 bits wide */
vm_page_bits_t valid; /* valid DEV_BSIZE chunk map (O,B) */
vm_page_bits_t dirty; /* dirty DEV_BSIZE chunk map (M,B) */
1994-05-24 10:09:53 +00:00
};
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
/*
* Special bits used in the ref_count field.
*
* ref_count is normally used to count wirings that prevent the page from being
* reclaimed, but also supports several special types of references that do not
* prevent reclamation. Accesses to the ref_count field must be atomic unless
* the page is unallocated.
*
* VPRC_OBJREF is the reference held by the containing object. It can set or
* cleared only when the corresponding object's write lock is held.
*
* VPRC_BLOCKED is used to atomically block wirings via pmap lookups while
* attempting to tear down all mappings of a given page. The page busy lock and
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
* object write lock must both be held in order to set or clear this bit.
*/
#define VPRC_BLOCKED 0x40000000u /* mappings are being removed */
#define VPRC_OBJREF 0x80000000u /* object reference, cleared with (O) */
#define VPRC_WIRE_COUNT(c) ((c) & ~(VPRC_BLOCKED | VPRC_OBJREF))
#define VPRC_WIRE_COUNT_MAX (~(VPRC_BLOCKED | VPRC_OBJREF))
/*
* Page flags stored in oflags:
*
* Access to these page flags is synchronized by the lock on the object
* containing the page (O).
*
* Note: VPO_UNMANAGED (used by OBJT_DEVICE, OBJT_PHYS and OBJT_SG)
* indicates that the page is not under PV management but
* otherwise should be treated as a normal page. Pages not
* under PV management cannot be paged out via the
* object/vm_page_t because there is no knowledge of their pte
* mappings, and such pages are also not on any PQ queue.
*
*/
#define VPO_KMEM_EXEC 0x01 /* kmem mapping allows execution */
#define VPO_SWAPSLEEP 0x02 /* waiting for swap to finish */
#define VPO_UNMANAGED 0x04 /* no PV management for page */
#define VPO_SWAPINPROG 0x08 /* swap I/O in progress on page */
/*
* Busy page implementation details.
* The algorithm is taken mostly by rwlock(9) and sx(9) locks implementation,
* even if the support for owner identity is removed because of size
* constraints. Checks on lock recursion are then not possible, while the
* lock assertions effectiveness is someway reduced.
*/
#define VPB_BIT_SHARED 0x01
#define VPB_BIT_EXCLUSIVE 0x02
#define VPB_BIT_WAITERS 0x04
#define VPB_BIT_FLAGMASK \
(VPB_BIT_SHARED | VPB_BIT_EXCLUSIVE | VPB_BIT_WAITERS)
#define VPB_SHARERS_SHIFT 3
#define VPB_SHARERS(x) \
(((x) & ~VPB_BIT_FLAGMASK) >> VPB_SHARERS_SHIFT)
#define VPB_SHARERS_WORD(x) ((x) << VPB_SHARERS_SHIFT | VPB_BIT_SHARED)
#define VPB_ONE_SHARER (1 << VPB_SHARERS_SHIFT)
#define VPB_SINGLE_EXCLUSIVE VPB_BIT_EXCLUSIVE
#ifdef INVARIANTS
#define VPB_CURTHREAD_EXCLUSIVE \
(VPB_BIT_EXCLUSIVE | ((u_int)(uintptr_t)curthread & ~VPB_BIT_FLAGMASK))
#else
#define VPB_CURTHREAD_EXCLUSIVE VPB_SINGLE_EXCLUSIVE
#endif
#define VPB_UNBUSIED VPB_SHARERS_WORD(0)
/* Freed lock blocks both shared and exclusive. */
#define VPB_FREED (0xffffffff - VPB_BIT_SHARED)
#define PQ_NONE 255
#define PQ_INACTIVE 0
#define PQ_ACTIVE 1
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
#define PQ_LAUNDRY 2
#define PQ_UNSWAPPABLE 3
#define PQ_COUNT 4
#ifndef VM_PAGE_HAVE_PGLIST
TAILQ_HEAD(pglist, vm_page);
#define VM_PAGE_HAVE_PGLIST
#endif
SLIST_HEAD(spglist, vm_page);
#ifdef _KERNEL
extern vm_page_t bogus_page;
#endif /* _KERNEL */
extern struct mtx_padalign pa_lock[];
#if defined(__arm__)
#define PDRSHIFT PDR_SHIFT
#elif !defined(PDRSHIFT)
#define PDRSHIFT 21
#endif
#define pa_index(pa) ((pa) >> PDRSHIFT)
#define PA_LOCKPTR(pa) ((struct mtx *)(&pa_lock[pa_index(pa) % PA_LOCK_COUNT]))
#define PA_LOCKOBJPTR(pa) ((struct lock_object *)PA_LOCKPTR((pa)))
#define PA_LOCK(pa) mtx_lock(PA_LOCKPTR(pa))
#define PA_TRYLOCK(pa) mtx_trylock(PA_LOCKPTR(pa))
#define PA_UNLOCK(pa) mtx_unlock(PA_LOCKPTR(pa))
#define PA_UNLOCK_COND(pa) \
do { \
if ((pa) != 0) { \
PA_UNLOCK((pa)); \
(pa) = 0; \
} \
} while (0)
#define PA_LOCK_ASSERT(pa, a) mtx_assert(PA_LOCKPTR(pa), (a))
#if defined(KLD_MODULE) && !defined(KLD_TIED)
#define vm_page_lock(m) vm_page_lock_KBI((m), LOCK_FILE, LOCK_LINE)
#define vm_page_unlock(m) vm_page_unlock_KBI((m), LOCK_FILE, LOCK_LINE)
#define vm_page_trylock(m) vm_page_trylock_KBI((m), LOCK_FILE, LOCK_LINE)
#else /* !KLD_MODULE */
#define vm_page_lockptr(m) (PA_LOCKPTR(VM_PAGE_TO_PHYS((m))))
#define vm_page_lock(m) mtx_lock(vm_page_lockptr((m)))
#define vm_page_unlock(m) mtx_unlock(vm_page_lockptr((m)))
#define vm_page_trylock(m) mtx_trylock(vm_page_lockptr((m)))
#endif
#if defined(INVARIANTS)
#define vm_page_assert_locked(m) \
vm_page_assert_locked_KBI((m), __FILE__, __LINE__)
#define vm_page_lock_assert(m, a) \
vm_page_lock_assert_KBI((m), (a), __FILE__, __LINE__)
#else
#define vm_page_assert_locked(m)
#define vm_page_lock_assert(m, a)
#endif
1994-05-24 10:09:53 +00:00
/*
* The vm_page's aflags are updated using atomic operations. To set or clear
* these flags, the functions vm_page_aflag_set() and vm_page_aflag_clear()
* must be used. Neither these flags nor these functions are part of the KBI.
*
* PGA_REFERENCED may be cleared only if the page is locked. It is set by
* both the MI and MD VM layers. However, kernel loadable modules should not
* directly set this flag. They should call vm_page_reference() instead.
Reduce the scope of the page queues lock and the number of PG_REFERENCED changes in vm_pageout_object_deactivate_pages(). Simplify this function's inner loop using TAILQ_FOREACH(), and shorten some of its overly long lines. Update a stale comment. Assert that PG_REFERENCED may be cleared only if the object containing the page is locked. Add a comment documenting this. Assert that a caller to vm_page_requeue() holds the page queues lock, and assert that the page is on a page queue. Push down the page queues lock into pmap_ts_referenced() and pmap_page_exists_quick(). (As of now, there are no longer any pmap functions that expect to be called with the page queues lock held.) Neither pmap_ts_referenced() nor pmap_page_exists_quick() should ever be passed an unmanaged page. Assert this rather than returning "0" and "FALSE" respectively. ARM: Simplify pmap_page_exists_quick() by switching to TAILQ_FOREACH(). Push down the page queues lock inside of pmap_clearbit(), simplifying pmap_clear_modify(), pmap_clear_reference(), and pmap_remove_write(). Additionally, this allows for avoiding the acquisition of the page queues lock in some cases. PowerPC/AIM: moea*_page_exits_quick() and moea*_page_wired_mappings() will never be called before pmap initialization is complete. Therefore, the check for moea_initialized can be eliminated. Push down the page queues lock inside of moea*_clear_bit(), simplifying moea*_clear_modify() and moea*_clear_reference(). The last parameter to moea*_clear_bit() is never used. Eliminate it. PowerPC/BookE: Simplify mmu_booke_page_exists_quick()'s control flow. Reviewed by: kib@
2010-06-10 16:56:35 +00:00
*
* PGA_WRITEABLE is set exclusively on managed pages by pmap_enter().
* When it does so, the object must be locked, or the page must be
* exclusive busied. The MI VM layer must never access this flag
* directly. Instead, it should call pmap_page_is_write_mapped().
*
* PGA_EXECUTABLE may be set by pmap routines, and indicates that a page has
* at least one executable mapping. It is not consumed by the MI VM layer.
*
* PGA_NOSYNC must be set and cleared with the page busy lock held.
*
* PGA_ENQUEUED is set and cleared when a page is inserted into or removed
* from a page queue, respectively. It determines whether the plinks.q field
* of the page is valid. To set or clear this flag, page's "queue" field must
* be a valid queue index, and the corresponding page queue lock must be held.
*
* PGA_DEQUEUE is set when the page is scheduled to be dequeued from a page
* queue, and cleared when the dequeue request is processed. A page may
* have PGA_DEQUEUE set and PGA_ENQUEUED cleared, for instance if a dequeue
* is requested after the page is scheduled to be enqueued but before it is
* actually inserted into the page queue.
*
* PGA_REQUEUE is set when the page is scheduled to be enqueued or requeued
* in its page queue.
*
* PGA_REQUEUE_HEAD is a special flag for enqueuing pages near the head of
* the inactive queue, thus bypassing LRU.
*
* The PGA_DEQUEUE, PGA_REQUEUE and PGA_REQUEUE_HEAD flags must be set using an
* atomic RMW operation to ensure that the "queue" field is a valid queue index,
* and the corresponding page queue lock must be held when clearing any of the
* flags.
*
* PGA_SWAP_FREE is used to defer freeing swap space to the pageout daemon
* when the context that dirties the page does not have the object write lock
* held.
1994-05-24 10:09:53 +00:00
*/
#define PGA_WRITEABLE 0x0001 /* page may be mapped writeable */
#define PGA_REFERENCED 0x0002 /* page has been referenced */
#define PGA_EXECUTABLE 0x0004 /* page may be mapped executable */
#define PGA_ENQUEUED 0x0008 /* page is enqueued in a page queue */
#define PGA_DEQUEUE 0x0010 /* page is due to be dequeued */
#define PGA_REQUEUE 0x0020 /* page is due to be requeued */
#define PGA_REQUEUE_HEAD 0x0040 /* page requeue should bypass LRU */
#define PGA_NOSYNC 0x0080 /* do not collect for syncer */
#define PGA_SWAP_FREE 0x0100 /* page with swap space was dirtied */
#define PGA_SWAP_SPACE 0x0200 /* page has allocated swap space */
#define PGA_QUEUE_OP_MASK (PGA_DEQUEUE | PGA_REQUEUE | PGA_REQUEUE_HEAD)
#define PGA_QUEUE_STATE_MASK (PGA_ENQUEUED | PGA_QUEUE_OP_MASK)
/*
* Page flags. Updates to these flags are not synchronized, and thus they must
* be set during page allocation or free to avoid races.
*
* The PG_PCPU_CACHE flag is set at allocation time if the page was
* allocated from a per-CPU cache. It is cleared the next time that the
* page is allocated from the physical memory allocator.
*/
#define PG_PCPU_CACHE 0x01 /* was allocated from per-CPU caches */
#define PG_FICTITIOUS 0x02 /* physical page doesn't exist */
#define PG_ZERO 0x04 /* page is zeroed */
#define PG_MARKER 0x08 /* special queue marker page */
#define PG_NODUMP 0x10 /* don't include this page in a dump */
1994-05-24 10:09:53 +00:00
NOTE: libkvm, w, ps, 'top', and any other utility which depends on struct proc or any VM system structure will have to be rebuilt!!! Much needed overhaul of the VM system. Included in this first round of changes: 1) Improved pager interfaces: init, alloc, dealloc, getpages, putpages, haspage, and sync operations are supported. The haspage interface now provides information about clusterability. All pager routines now take struct vm_object's instead of "pagers". 2) Improved data structures. In the previous paradigm, there is constant confusion caused by pagers being both a data structure ("allocate a pager") and a collection of routines. The idea of a pager structure has escentially been eliminated. Objects now have types, and this type is used to index the appropriate pager. In most cases, items in the pager structure were duplicated in the object data structure and thus were unnecessary. In the few cases that remained, a un_pager structure union was created in the object to contain these items. 3) Because of the cleanup of #1 & #2, a lot of unnecessary layering can now be removed. For instance, vm_object_enter(), vm_object_lookup(), vm_object_remove(), and the associated object hash list were some of the things that were removed. 4) simple_lock's removed. Discussion with several people reveals that the SMP locking primitives used in the VM system aren't likely the mechanism that we'll be adopting. Even if it were, the locking that was in the code was very inadequate and would have to be mostly re-done anyway. The locking in a uni-processor kernel was a no-op but went a long way toward making the code difficult to read and debug. 5) Places that attempted to kludge-up the fact that we don't have kernel thread support have been fixed to reflect the reality that we are really dealing with processes, not threads. The VM system didn't have complete thread support, so the comments and mis-named routines were just wrong. We now use tsleep and wakeup directly in the lock routines, for instance. 6) Where appropriate, the pagers have been improved, especially in the pager_alloc routines. Most of the pager_allocs have been rewritten and are now faster and easier to maintain. 7) The pagedaemon pageout clustering algorithm has been rewritten and now tries harder to output an even number of pages before and after the requested page. This is sort of the reverse of the ideal pagein algorithm and should provide better overall performance. 8) Unnecessary (incorrect) casts to caddr_t in calls to tsleep & wakeup have been removed. Some other unnecessary casts have also been removed. 9) Some almost useless debugging code removed. 10) Terminology of shadow objects vs. backing objects straightened out. The fact that the vm_object data structure escentially had this backwards really confused things. The use of "shadow" and "backing object" throughout the code is now internally consistent and correct in the Mach terminology. 11) Several minor bug fixes, including one in the vm daemon that caused 0 RSS objects to not get purged as intended. 12) A "default pager" has now been created which cleans up the transition of objects to the "swap" type. The previous checks throughout the code for swp->pg_data != NULL were really ugly. This change also provides the rudiments for future backing of "anonymous" memory by something other than the swap pager (via the vnode pager, for example), and it allows the decision about which of these pagers to use to be made dynamically (although will need some additional decision code to do this, of course). 13) (dyson) MAP_COPY has been deprecated and the corresponding "copy object" code has been removed. MAP_COPY was undocumented and non- standard. It was furthermore broken in several ways which caused its behavior to degrade to MAP_PRIVATE. Binaries that use MAP_COPY will continue to work correctly, but via the slightly different semantics of MAP_PRIVATE. 14) (dyson) Sharing maps have been removed. It's marginal usefulness in a threads design can be worked around in other ways. Both #12 and #13 were done to simplify the code and improve readability and maintain- ability. (As were most all of these changes) TODO: 1) Rewrite most of the vnode pager to use VOP_GETPAGES/PUTPAGES. Doing this will reduce the vnode pager to a mere fraction of its current size. 2) Rewrite vm_fault and the swap/vnode pagers to use the clustering information provided by the new haspage pager interface. This will substantially reduce the overhead by eliminating a large number of VOP_BMAP() calls. The VOP_BMAP() filesystem interface should be improved to provide both a "behind" and "ahead" indication of contiguousness. 3) Implement the extended features of pager_haspage in swap_pager_haspage(). It currently just says 0 pages ahead/behind. 4) Re-implement the swap device (swstrategy) in a more elegant way, perhaps via a much more general mechanism that could also be used for disk striping of regular filesystems. 5) Do something to improve the architecture of vm_object_collapse(). The fact that it makes calls into the swap pager and knows too much about how the swap pager operates really bothers me. It also doesn't allow for collapsing of non-swap pager objects ("unnamed" objects backed by other pagers).
1995-07-13 08:48:48 +00:00
/*
* Misc constants.
*/
#define ACT_DECLINE 1
#define ACT_ADVANCE 3
#define ACT_INIT 5
#define ACT_MAX 64
1994-05-24 10:09:53 +00:00
#ifdef _KERNEL
#include <sys/kassert.h>
#include <machine/atomic.h>
1994-05-24 10:09:53 +00:00
/*
* Each pageable resident page falls into one of five lists:
1994-05-24 10:09:53 +00:00
*
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
* free
1994-05-24 10:09:53 +00:00
* Available for allocation now.
NOTE: libkvm, w, ps, 'top', and any other utility which depends on struct proc or any VM system structure will have to be rebuilt!!! Much needed overhaul of the VM system. Included in this first round of changes: 1) Improved pager interfaces: init, alloc, dealloc, getpages, putpages, haspage, and sync operations are supported. The haspage interface now provides information about clusterability. All pager routines now take struct vm_object's instead of "pagers". 2) Improved data structures. In the previous paradigm, there is constant confusion caused by pagers being both a data structure ("allocate a pager") and a collection of routines. The idea of a pager structure has escentially been eliminated. Objects now have types, and this type is used to index the appropriate pager. In most cases, items in the pager structure were duplicated in the object data structure and thus were unnecessary. In the few cases that remained, a un_pager structure union was created in the object to contain these items. 3) Because of the cleanup of #1 & #2, a lot of unnecessary layering can now be removed. For instance, vm_object_enter(), vm_object_lookup(), vm_object_remove(), and the associated object hash list were some of the things that were removed. 4) simple_lock's removed. Discussion with several people reveals that the SMP locking primitives used in the VM system aren't likely the mechanism that we'll be adopting. Even if it were, the locking that was in the code was very inadequate and would have to be mostly re-done anyway. The locking in a uni-processor kernel was a no-op but went a long way toward making the code difficult to read and debug. 5) Places that attempted to kludge-up the fact that we don't have kernel thread support have been fixed to reflect the reality that we are really dealing with processes, not threads. The VM system didn't have complete thread support, so the comments and mis-named routines were just wrong. We now use tsleep and wakeup directly in the lock routines, for instance. 6) Where appropriate, the pagers have been improved, especially in the pager_alloc routines. Most of the pager_allocs have been rewritten and are now faster and easier to maintain. 7) The pagedaemon pageout clustering algorithm has been rewritten and now tries harder to output an even number of pages before and after the requested page. This is sort of the reverse of the ideal pagein algorithm and should provide better overall performance. 8) Unnecessary (incorrect) casts to caddr_t in calls to tsleep & wakeup have been removed. Some other unnecessary casts have also been removed. 9) Some almost useless debugging code removed. 10) Terminology of shadow objects vs. backing objects straightened out. The fact that the vm_object data structure escentially had this backwards really confused things. The use of "shadow" and "backing object" throughout the code is now internally consistent and correct in the Mach terminology. 11) Several minor bug fixes, including one in the vm daemon that caused 0 RSS objects to not get purged as intended. 12) A "default pager" has now been created which cleans up the transition of objects to the "swap" type. The previous checks throughout the code for swp->pg_data != NULL were really ugly. This change also provides the rudiments for future backing of "anonymous" memory by something other than the swap pager (via the vnode pager, for example), and it allows the decision about which of these pagers to use to be made dynamically (although will need some additional decision code to do this, of course). 13) (dyson) MAP_COPY has been deprecated and the corresponding "copy object" code has been removed. MAP_COPY was undocumented and non- standard. It was furthermore broken in several ways which caused its behavior to degrade to MAP_PRIVATE. Binaries that use MAP_COPY will continue to work correctly, but via the slightly different semantics of MAP_PRIVATE. 14) (dyson) Sharing maps have been removed. It's marginal usefulness in a threads design can be worked around in other ways. Both #12 and #13 were done to simplify the code and improve readability and maintain- ability. (As were most all of these changes) TODO: 1) Rewrite most of the vnode pager to use VOP_GETPAGES/PUTPAGES. Doing this will reduce the vnode pager to a mere fraction of its current size. 2) Rewrite vm_fault and the swap/vnode pagers to use the clustering information provided by the new haspage pager interface. This will substantially reduce the overhead by eliminating a large number of VOP_BMAP() calls. The VOP_BMAP() filesystem interface should be improved to provide both a "behind" and "ahead" indication of contiguousness. 3) Implement the extended features of pager_haspage in swap_pager_haspage(). It currently just says 0 pages ahead/behind. 4) Re-implement the swap device (swstrategy) in a more elegant way, perhaps via a much more general mechanism that could also be used for disk striping of regular filesystems. 5) Do something to improve the architecture of vm_object_collapse(). The fact that it makes calls into the swap pager and knows too much about how the swap pager operates really bothers me. It also doesn't allow for collapsing of non-swap pager objects ("unnamed" objects backed by other pagers).
1995-07-13 08:48:48 +00:00
*
1994-05-24 10:09:53 +00:00
* inactive
* Low activity, candidates for reclamation.
* This list is approximately LRU ordered.
*
* laundry
1994-05-24 10:09:53 +00:00
* This is the list of pages that should be
* paged out next.
NOTE: libkvm, w, ps, 'top', and any other utility which depends on struct proc or any VM system structure will have to be rebuilt!!! Much needed overhaul of the VM system. Included in this first round of changes: 1) Improved pager interfaces: init, alloc, dealloc, getpages, putpages, haspage, and sync operations are supported. The haspage interface now provides information about clusterability. All pager routines now take struct vm_object's instead of "pagers". 2) Improved data structures. In the previous paradigm, there is constant confusion caused by pagers being both a data structure ("allocate a pager") and a collection of routines. The idea of a pager structure has escentially been eliminated. Objects now have types, and this type is used to index the appropriate pager. In most cases, items in the pager structure were duplicated in the object data structure and thus were unnecessary. In the few cases that remained, a un_pager structure union was created in the object to contain these items. 3) Because of the cleanup of #1 & #2, a lot of unnecessary layering can now be removed. For instance, vm_object_enter(), vm_object_lookup(), vm_object_remove(), and the associated object hash list were some of the things that were removed. 4) simple_lock's removed. Discussion with several people reveals that the SMP locking primitives used in the VM system aren't likely the mechanism that we'll be adopting. Even if it were, the locking that was in the code was very inadequate and would have to be mostly re-done anyway. The locking in a uni-processor kernel was a no-op but went a long way toward making the code difficult to read and debug. 5) Places that attempted to kludge-up the fact that we don't have kernel thread support have been fixed to reflect the reality that we are really dealing with processes, not threads. The VM system didn't have complete thread support, so the comments and mis-named routines were just wrong. We now use tsleep and wakeup directly in the lock routines, for instance. 6) Where appropriate, the pagers have been improved, especially in the pager_alloc routines. Most of the pager_allocs have been rewritten and are now faster and easier to maintain. 7) The pagedaemon pageout clustering algorithm has been rewritten and now tries harder to output an even number of pages before and after the requested page. This is sort of the reverse of the ideal pagein algorithm and should provide better overall performance. 8) Unnecessary (incorrect) casts to caddr_t in calls to tsleep & wakeup have been removed. Some other unnecessary casts have also been removed. 9) Some almost useless debugging code removed. 10) Terminology of shadow objects vs. backing objects straightened out. The fact that the vm_object data structure escentially had this backwards really confused things. The use of "shadow" and "backing object" throughout the code is now internally consistent and correct in the Mach terminology. 11) Several minor bug fixes, including one in the vm daemon that caused 0 RSS objects to not get purged as intended. 12) A "default pager" has now been created which cleans up the transition of objects to the "swap" type. The previous checks throughout the code for swp->pg_data != NULL were really ugly. This change also provides the rudiments for future backing of "anonymous" memory by something other than the swap pager (via the vnode pager, for example), and it allows the decision about which of these pagers to use to be made dynamically (although will need some additional decision code to do this, of course). 13) (dyson) MAP_COPY has been deprecated and the corresponding "copy object" code has been removed. MAP_COPY was undocumented and non- standard. It was furthermore broken in several ways which caused its behavior to degrade to MAP_PRIVATE. Binaries that use MAP_COPY will continue to work correctly, but via the slightly different semantics of MAP_PRIVATE. 14) (dyson) Sharing maps have been removed. It's marginal usefulness in a threads design can be worked around in other ways. Both #12 and #13 were done to simplify the code and improve readability and maintain- ability. (As were most all of these changes) TODO: 1) Rewrite most of the vnode pager to use VOP_GETPAGES/PUTPAGES. Doing this will reduce the vnode pager to a mere fraction of its current size. 2) Rewrite vm_fault and the swap/vnode pagers to use the clustering information provided by the new haspage pager interface. This will substantially reduce the overhead by eliminating a large number of VOP_BMAP() calls. The VOP_BMAP() filesystem interface should be improved to provide both a "behind" and "ahead" indication of contiguousness. 3) Implement the extended features of pager_haspage in swap_pager_haspage(). It currently just says 0 pages ahead/behind. 4) Re-implement the swap device (swstrategy) in a more elegant way, perhaps via a much more general mechanism that could also be used for disk striping of regular filesystems. 5) Do something to improve the architecture of vm_object_collapse(). The fact that it makes calls into the swap pager and knows too much about how the swap pager operates really bothers me. It also doesn't allow for collapsing of non-swap pager objects ("unnamed" objects backed by other pagers).
1995-07-13 08:48:48 +00:00
*
* unswappable
* Dirty anonymous pages that cannot be paged
* out because no swap device is configured.
*
1994-05-24 10:09:53 +00:00
* active
* Pages that are "active", i.e., they have been
NOTE: libkvm, w, ps, 'top', and any other utility which depends on struct proc or any VM system structure will have to be rebuilt!!! Much needed overhaul of the VM system. Included in this first round of changes: 1) Improved pager interfaces: init, alloc, dealloc, getpages, putpages, haspage, and sync operations are supported. The haspage interface now provides information about clusterability. All pager routines now take struct vm_object's instead of "pagers". 2) Improved data structures. In the previous paradigm, there is constant confusion caused by pagers being both a data structure ("allocate a pager") and a collection of routines. The idea of a pager structure has escentially been eliminated. Objects now have types, and this type is used to index the appropriate pager. In most cases, items in the pager structure were duplicated in the object data structure and thus were unnecessary. In the few cases that remained, a un_pager structure union was created in the object to contain these items. 3) Because of the cleanup of #1 & #2, a lot of unnecessary layering can now be removed. For instance, vm_object_enter(), vm_object_lookup(), vm_object_remove(), and the associated object hash list were some of the things that were removed. 4) simple_lock's removed. Discussion with several people reveals that the SMP locking primitives used in the VM system aren't likely the mechanism that we'll be adopting. Even if it were, the locking that was in the code was very inadequate and would have to be mostly re-done anyway. The locking in a uni-processor kernel was a no-op but went a long way toward making the code difficult to read and debug. 5) Places that attempted to kludge-up the fact that we don't have kernel thread support have been fixed to reflect the reality that we are really dealing with processes, not threads. The VM system didn't have complete thread support, so the comments and mis-named routines were just wrong. We now use tsleep and wakeup directly in the lock routines, for instance. 6) Where appropriate, the pagers have been improved, especially in the pager_alloc routines. Most of the pager_allocs have been rewritten and are now faster and easier to maintain. 7) The pagedaemon pageout clustering algorithm has been rewritten and now tries harder to output an even number of pages before and after the requested page. This is sort of the reverse of the ideal pagein algorithm and should provide better overall performance. 8) Unnecessary (incorrect) casts to caddr_t in calls to tsleep & wakeup have been removed. Some other unnecessary casts have also been removed. 9) Some almost useless debugging code removed. 10) Terminology of shadow objects vs. backing objects straightened out. The fact that the vm_object data structure escentially had this backwards really confused things. The use of "shadow" and "backing object" throughout the code is now internally consistent and correct in the Mach terminology. 11) Several minor bug fixes, including one in the vm daemon that caused 0 RSS objects to not get purged as intended. 12) A "default pager" has now been created which cleans up the transition of objects to the "swap" type. The previous checks throughout the code for swp->pg_data != NULL were really ugly. This change also provides the rudiments for future backing of "anonymous" memory by something other than the swap pager (via the vnode pager, for example), and it allows the decision about which of these pagers to use to be made dynamically (although will need some additional decision code to do this, of course). 13) (dyson) MAP_COPY has been deprecated and the corresponding "copy object" code has been removed. MAP_COPY was undocumented and non- standard. It was furthermore broken in several ways which caused its behavior to degrade to MAP_PRIVATE. Binaries that use MAP_COPY will continue to work correctly, but via the slightly different semantics of MAP_PRIVATE. 14) (dyson) Sharing maps have been removed. It's marginal usefulness in a threads design can be worked around in other ways. Both #12 and #13 were done to simplify the code and improve readability and maintain- ability. (As were most all of these changes) TODO: 1) Rewrite most of the vnode pager to use VOP_GETPAGES/PUTPAGES. Doing this will reduce the vnode pager to a mere fraction of its current size. 2) Rewrite vm_fault and the swap/vnode pagers to use the clustering information provided by the new haspage pager interface. This will substantially reduce the overhead by eliminating a large number of VOP_BMAP() calls. The VOP_BMAP() filesystem interface should be improved to provide both a "behind" and "ahead" indication of contiguousness. 3) Implement the extended features of pager_haspage in swap_pager_haspage(). It currently just says 0 pages ahead/behind. 4) Re-implement the swap device (swstrategy) in a more elegant way, perhaps via a much more general mechanism that could also be used for disk striping of regular filesystems. 5) Do something to improve the architecture of vm_object_collapse(). The fact that it makes calls into the swap pager and knows too much about how the swap pager operates really bothers me. It also doesn't allow for collapsing of non-swap pager objects ("unnamed" objects backed by other pagers).
1995-07-13 08:48:48 +00:00
* recently referenced.
*
1994-05-24 10:09:53 +00:00
*/
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
extern vm_page_t vm_page_array; /* First resident page in table */
extern long vm_page_array_size; /* number of vm_page_t's */
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
extern long first_page; /* first physical page number */
1994-05-24 10:09:53 +00:00
#define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr)
/*
2015-10-23 12:06:06 +00:00
* PHYS_TO_VM_PAGE() returns the vm_page_t object that represents a memory
* page to which the given physical address belongs. The correct vm_page_t
* object is returned for addresses that are not page-aligned.
*/
vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa);
1994-05-24 10:09:53 +00:00
/*
* Page allocation parameters for vm_page for the functions
* vm_page_alloc(), vm_page_grab(), vm_page_alloc_contig() and
* vm_page_alloc_freelist(). Some functions support only a subset
* of the flags, and ignore others, see the flags legend.
*
* The meaning of VM_ALLOC_ZERO differs slightly between the vm_page_alloc*()
* and the vm_page_grab*() functions. See these functions for details.
*
* Bits 0 - 1 define class.
* Bits 2 - 15 dedicated for flags.
* Legend:
* (a) - vm_page_alloc() supports the flag.
* (c) - vm_page_alloc_contig() supports the flag.
* (g) - vm_page_grab() supports the flag.
* (n) - vm_page_alloc_noobj() and vm_page_alloc_freelist() support the flag.
* (p) - vm_page_grab_pages() supports the flag.
* Bits above 15 define the count of additional pages that the caller
* intends to allocate.
*/
#define VM_ALLOC_NORMAL 0
#define VM_ALLOC_INTERRUPT 1
#define VM_ALLOC_SYSTEM 2
#define VM_ALLOC_CLASS_MASK 3
#define VM_ALLOC_WAITOK 0x0008 /* (acn) Sleep and retry */
#define VM_ALLOC_WAITFAIL 0x0010 /* (acn) Sleep and return error */
#define VM_ALLOC_WIRED 0x0020 /* (acgnp) Allocate a wired page */
#define VM_ALLOC_ZERO 0x0040 /* (acgnp) Allocate a zeroed page */
#define VM_ALLOC_NORECLAIM 0x0080 /* (c) Do not reclaim after failure */
#define VM_ALLOC_AVAIL0 0x0100
#define VM_ALLOC_NOBUSY 0x0200 /* (acgp) Do not excl busy the page */
#define VM_ALLOC_NOCREAT 0x0400 /* (gp) Don't create a page */
#define VM_ALLOC_AVAIL1 0x0800
#define VM_ALLOC_IGN_SBUSY 0x1000 /* (gp) Ignore shared busy flag */
#define VM_ALLOC_NODUMP 0x2000 /* (ag) don't include in dump */
#define VM_ALLOC_SBUSY 0x4000 /* (acgp) Shared busy the page */
#define VM_ALLOC_NOWAIT 0x8000 /* (acgnp) Do not sleep */
#define VM_ALLOC_COUNT_MAX 0xffff
#define VM_ALLOC_COUNT_SHIFT 16
#define VM_ALLOC_COUNT_MASK (VM_ALLOC_COUNT(VM_ALLOC_COUNT_MAX))
#define VM_ALLOC_COUNT(count) ({ \
KASSERT((count) <= VM_ALLOC_COUNT_MAX, \
("%s: invalid VM_ALLOC_COUNT value", __func__)); \
(count) << VM_ALLOC_COUNT_SHIFT; \
})
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
#ifdef M_NOWAIT
static inline int
malloc2vm_flags(int malloc_flags)
{
int pflags;
KASSERT((malloc_flags & M_USE_RESERVE) == 0 ||
(malloc_flags & M_NOWAIT) != 0,
("M_USE_RESERVE requires M_NOWAIT"));
pflags = (malloc_flags & M_USE_RESERVE) != 0 ? VM_ALLOC_INTERRUPT :
VM_ALLOC_SYSTEM;
if ((malloc_flags & M_ZERO) != 0)
pflags |= VM_ALLOC_ZERO;
if ((malloc_flags & M_NODUMP) != 0)
pflags |= VM_ALLOC_NODUMP;
if ((malloc_flags & M_NOWAIT))
pflags |= VM_ALLOC_NOWAIT;
if ((malloc_flags & M_WAITOK))
pflags |= VM_ALLOC_WAITOK;
if ((malloc_flags & M_NORECLAIM))
pflags |= VM_ALLOC_NORECLAIM;
return (pflags);
}
#endif
/*
* Predicates supported by vm_page_ps_test():
*
* PS_ALL_DIRTY is true only if the entire (super)page is dirty.
* However, it can be spuriously false when the (super)page has become
* dirty in the pmap but that information has not been propagated to the
* machine-independent layer.
*/
#define PS_ALL_DIRTY 0x1
#define PS_ALL_VALID 0x2
#define PS_NONE_BUSY 0x4
bool vm_page_busy_acquire(vm_page_t m, int allocflags);
void vm_page_busy_downgrade(vm_page_t m);
int vm_page_busy_tryupgrade(vm_page_t m);
bool vm_page_busy_sleep(vm_page_t m, const char *msg, int allocflags);
void vm_page_busy_sleep_unlocked(vm_object_t obj, vm_page_t m,
vm_pindex_t pindex, const char *wmesg, int allocflags);
void vm_page_free(vm_page_t m);
void vm_page_free_zero(vm_page_t m);
void vm_page_activate (vm_page_t);
void vm_page_advise(vm_page_t m, int advice);
vm_page_t vm_page_alloc(vm_object_t, vm_pindex_t, int);
vm_page_t vm_page_alloc_domain(vm_object_t, vm_pindex_t, int, int);
vm_page_t vm_page_alloc_after(vm_object_t, vm_pindex_t, int, vm_page_t);
vm_page_t vm_page_alloc_domain_after(vm_object_t, vm_pindex_t, int, int,
vm_page_t);
vm_page_t vm_page_alloc_contig(vm_object_t object, vm_pindex_t pindex, int req,
u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment,
vm_paddr_t boundary, vm_memattr_t memattr);
vm_page_t vm_page_alloc_contig_domain(vm_object_t object,
vm_pindex_t pindex, int domain, int req, u_long npages, vm_paddr_t low,
vm_paddr_t high, u_long alignment, vm_paddr_t boundary,
vm_memattr_t memattr);
vm_page_t vm_page_alloc_freelist(int, int);
vm_page_t vm_page_alloc_freelist_domain(int, int, int);
vm_page_t vm_page_alloc_noobj(int);
vm_page_t vm_page_alloc_noobj_domain(int, int);
vm_page_t vm_page_alloc_noobj_contig(int req, u_long npages, vm_paddr_t low,
vm_paddr_t high, u_long alignment, vm_paddr_t boundary,
vm_memattr_t memattr);
vm_page_t vm_page_alloc_noobj_contig_domain(int domain, int req, u_long npages,
vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary,
vm_memattr_t memattr);
void vm_page_bits_set(vm_page_t m, vm_page_bits_t *bits, vm_page_bits_t set);
bool vm_page_blacklist_add(vm_paddr_t pa, bool verbose);
vm_page_t vm_page_grab(vm_object_t, vm_pindex_t, int);
vm_page_t vm_page_grab_unlocked(vm_object_t, vm_pindex_t, int);
int vm_page_grab_pages(vm_object_t object, vm_pindex_t pindex, int allocflags,
vm_page_t *ma, int count);
int vm_page_grab_pages_unlocked(vm_object_t object, vm_pindex_t pindex,
int allocflags, vm_page_t *ma, int count);
int vm_page_grab_valid(vm_page_t *mp, vm_object_t object, vm_pindex_t pindex,
int allocflags);
int vm_page_grab_valid_unlocked(vm_page_t *mp, vm_object_t object,
vm_pindex_t pindex, int allocflags);
void vm_page_deactivate(vm_page_t);
void vm_page_deactivate_noreuse(vm_page_t);
void vm_page_dequeue(vm_page_t m);
void vm_page_dequeue_deferred(vm_page_t m);
vm_page_t vm_page_find_least(vm_object_t, vm_pindex_t);
void vm_page_free_invalid(vm_page_t);
vm_page_t vm_page_getfake(vm_paddr_t paddr, vm_memattr_t memattr);
void vm_page_initfake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr);
void vm_page_init_marker(vm_page_t marker, int queue, uint16_t aflags);
void vm_page_init_page(vm_page_t m, vm_paddr_t pa, int segind, int pool);
int vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t);
void vm_page_invalid(vm_page_t m);
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
void vm_page_launder(vm_page_t m);
vm_page_t vm_page_lookup(vm_object_t, vm_pindex_t);
vm_page_t vm_page_lookup_unlocked(vm_object_t, vm_pindex_t);
vm_page_t vm_page_next(vm_page_t m);
void vm_page_pqbatch_drain(void);
void vm_page_pqbatch_submit(vm_page_t m, uint8_t queue);
bool vm_page_pqstate_commit(vm_page_t m, vm_page_astate_t *old,
vm_page_astate_t new);
vm_page_t vm_page_prev(vm_page_t m);
bool vm_page_ps_test(vm_page_t m, int flags, vm_page_t skip_m);
void vm_page_putfake(vm_page_t m);
void vm_page_readahead_finish(vm_page_t m);
int vm_page_reclaim_contig(int req, u_long npages, vm_paddr_t low,
vm_paddr_t high, u_long alignment, vm_paddr_t boundary);
int vm_page_reclaim_contig_domain(int domain, int req, u_long npages,
vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary);
int vm_page_reclaim_contig_domain_ext(int domain, int req, u_long npages,
vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary,
int desired_runs);
void vm_page_reference(vm_page_t m);
#define VPR_TRYFREE 0x01
#define VPR_NOREUSE 0x02
void vm_page_release(vm_page_t m, int flags);
void vm_page_release_locked(vm_page_t m, int flags);
vm_page_t vm_page_relookup(vm_object_t, vm_pindex_t);
bool vm_page_remove(vm_page_t);
bool vm_page_remove_xbusy(vm_page_t);
int vm_page_rename(vm_page_t, vm_object_t, vm_pindex_t);
void vm_page_replace(vm_page_t mnew, vm_object_t object,
vm_pindex_t pindex, vm_page_t mold);
int vm_page_sbusied(vm_page_t m);
vm_page_bits_t vm_page_set_dirty(vm_page_t m);
void vm_page_set_valid_range(vm_page_t m, int base, int size);
vm_offset_t vm_page_startup(vm_offset_t vaddr);
void vm_page_sunbusy(vm_page_t m);
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
bool vm_page_try_remove_all(vm_page_t m);
bool vm_page_try_remove_write(vm_page_t m);
int vm_page_trysbusy(vm_page_t m);
int vm_page_tryxbusy(vm_page_t m);
void vm_page_unhold_pages(vm_page_t *ma, int count);
void vm_page_unswappable(vm_page_t m);
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
void vm_page_unwire(vm_page_t m, uint8_t queue);
bool vm_page_unwire_noq(vm_page_t m);
void vm_page_updatefake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr);
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
void vm_page_wire(vm_page_t);
bool vm_page_wire_mapped(vm_page_t m);
void vm_page_xunbusy_hard(vm_page_t m);
void vm_page_xunbusy_hard_unchecked(vm_page_t m);
void vm_page_set_validclean (vm_page_t, int, int);
void vm_page_clear_dirty(vm_page_t, int, int);
void vm_page_set_invalid(vm_page_t, int, int);
void vm_page_valid(vm_page_t m);
int vm_page_is_valid(vm_page_t, int, int);
void vm_page_test_dirty(vm_page_t);
vm_page_bits_t vm_page_bits(int base, int size);
void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid);
void vm_page_free_pages_toq(struct spglist *free, bool update_wire_count);
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
2012-06-20 23:25:47 +00:00
void vm_page_dirty_KBI(vm_page_t m);
void vm_page_lock_KBI(vm_page_t m, const char *file, int line);
void vm_page_unlock_KBI(vm_page_t m, const char *file, int line);
int vm_page_trylock_KBI(vm_page_t m, const char *file, int line);
#if defined(INVARIANTS) || defined(INVARIANT_SUPPORT)
void vm_page_assert_locked_KBI(vm_page_t m, const char *file, int line);
void vm_page_lock_assert_KBI(vm_page_t m, int a, const char *file, int line);
#endif
#define vm_page_busy_fetch(m) atomic_load_int(&(m)->busy_lock)
#define vm_page_assert_busied(m) \
KASSERT(vm_page_busied(m), \
("vm_page_assert_busied: page %p not busy @ %s:%d", \
(m), __FILE__, __LINE__))
#define vm_page_assert_sbusied(m) \
KASSERT(vm_page_sbusied(m), \
("vm_page_assert_sbusied: page %p not shared busy @ %s:%d", \
(m), __FILE__, __LINE__))
#define vm_page_assert_unbusied(m) \
KASSERT((vm_page_busy_fetch(m) & ~VPB_BIT_WAITERS) != \
VPB_CURTHREAD_EXCLUSIVE, \
("vm_page_assert_unbusied: page %p busy_lock %#x owned" \
" by me (%p) @ %s:%d", \
(m), (m)->busy_lock, curthread, __FILE__, __LINE__)); \
#define vm_page_assert_xbusied_unchecked(m) do { \
KASSERT(vm_page_xbusied(m), \
("vm_page_assert_xbusied: page %p not exclusive busy @ %s:%d", \
(m), __FILE__, __LINE__)); \
} while (0)
#define vm_page_assert_xbusied(m) do { \
vm_page_assert_xbusied_unchecked(m); \
KASSERT((vm_page_busy_fetch(m) & ~VPB_BIT_WAITERS) == \
VPB_CURTHREAD_EXCLUSIVE, \
("vm_page_assert_xbusied: page %p busy_lock %#x not owned" \
" by me (%p) @ %s:%d", \
(m), (m)->busy_lock, curthread, __FILE__, __LINE__)); \
} while (0)
#define vm_page_busied(m) \
(vm_page_busy_fetch(m) != VPB_UNBUSIED)
#define vm_page_xbusied(m) \
((vm_page_busy_fetch(m) & VPB_SINGLE_EXCLUSIVE) != 0)
#define vm_page_busy_freed(m) \
(vm_page_busy_fetch(m) == VPB_FREED)
/* Note: page m's lock must not be owned by the caller. */
#define vm_page_xunbusy(m) do { \
if (!atomic_cmpset_rel_int(&(m)->busy_lock, \
VPB_CURTHREAD_EXCLUSIVE, VPB_UNBUSIED)) \
vm_page_xunbusy_hard(m); \
} while (0)
#define vm_page_xunbusy_unchecked(m) do { \
if (!atomic_cmpset_rel_int(&(m)->busy_lock, \
VPB_CURTHREAD_EXCLUSIVE, VPB_UNBUSIED)) \
vm_page_xunbusy_hard_unchecked(m); \
} while (0)
#ifdef INVARIANTS
void vm_page_object_busy_assert(vm_page_t m);
#define VM_PAGE_OBJECT_BUSY_ASSERT(m) vm_page_object_busy_assert(m)
void vm_page_assert_pga_writeable(vm_page_t m, uint16_t bits);
#define VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits) \
vm_page_assert_pga_writeable(m, bits)
/*
* Claim ownership of a page's xbusy state. In non-INVARIANTS kernels this
* operation is a no-op since ownership is not tracked. In particular
* this macro does not provide any synchronization with the previous owner.
*/
#define vm_page_xbusy_claim(m) do { \
u_int _busy_lock; \
\
vm_page_assert_xbusied_unchecked((m)); \
do { \
_busy_lock = vm_page_busy_fetch(m); \
} while (!atomic_cmpset_int(&(m)->busy_lock, _busy_lock, \
(_busy_lock & VPB_BIT_FLAGMASK) | VPB_CURTHREAD_EXCLUSIVE)); \
} while (0)
#else
#define VM_PAGE_OBJECT_BUSY_ASSERT(m) (void)0
#define VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits) (void)0
#define vm_page_xbusy_claim(m)
#endif
#if BYTE_ORDER == BIG_ENDIAN
#define VM_PAGE_AFLAG_SHIFT 16
#else
#define VM_PAGE_AFLAG_SHIFT 0
#endif
/*
* Load a snapshot of a page's 32-bit atomic state.
*/
static inline vm_page_astate_t
vm_page_astate_load(vm_page_t m)
{
vm_page_astate_t a;
a._bits = atomic_load_32(&m->a._bits);
return (a);
}
/*
* Atomically compare and set a page's atomic state.
*/
static inline bool
vm_page_astate_fcmpset(vm_page_t m, vm_page_astate_t *old, vm_page_astate_t new)
{
KASSERT(new.queue == PQ_INACTIVE || (new.flags & PGA_REQUEUE_HEAD) == 0,
("%s: invalid head requeue request for page %p", __func__, m));
KASSERT((new.flags & PGA_ENQUEUED) == 0 || new.queue != PQ_NONE,
("%s: setting PGA_ENQUEUED with PQ_NONE in page %p", __func__, m));
KASSERT(new._bits != old->_bits,
("%s: bits are unchanged", __func__));
return (atomic_fcmpset_32(&m->a._bits, &old->_bits, new._bits) != 0);
}
/*
* Clear the given bits in the specified page.
*/
static inline void
vm_page_aflag_clear(vm_page_t m, uint16_t bits)
{
uint32_t *addr, val;
/*
* Access the whole 32-bit word containing the aflags field with an
* atomic update. Parallel non-atomic updates to the other fields
* within this word are handled properly by the atomic update.
*/
addr = (void *)&m->a;
val = bits << VM_PAGE_AFLAG_SHIFT;
atomic_clear_32(addr, val);
}
/*
* Set the given bits in the specified page.
*/
static inline void
vm_page_aflag_set(vm_page_t m, uint16_t bits)
{
uint32_t *addr, val;
VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits);
/*
* Access the whole 32-bit word containing the aflags field with an
* atomic update. Parallel non-atomic updates to the other fields
* within this word are handled properly by the atomic update.
*/
addr = (void *)&m->a;
val = bits << VM_PAGE_AFLAG_SHIFT;
atomic_set_32(addr, val);
}
2012-06-20 23:25:47 +00:00
/*
* vm_page_dirty:
*
* Set all bits in the page's dirty field.
*
* The object containing the specified page must be locked if the
* call is made from the machine-independent layer.
*
* See vm_page_clear_dirty_mask().
*/
static __inline void
vm_page_dirty(vm_page_t m)
{
/* Use vm_page_dirty_KBI() under INVARIANTS to save memory. */
#if (defined(KLD_MODULE) && !defined(KLD_TIED)) || defined(INVARIANTS)
2012-06-20 23:25:47 +00:00
vm_page_dirty_KBI(m);
#else
m->dirty = VM_PAGE_BITS_ALL;
#endif
}
/*
* vm_page_undirty:
*
* Set page to not be dirty. Note: does not clear pmap modify bits
*/
static __inline void
vm_page_undirty(vm_page_t m)
{
VM_PAGE_OBJECT_BUSY_ASSERT(m);
m->dirty = 0;
}
static inline uint8_t
_vm_page_queue(vm_page_astate_t as)
{
if ((as.flags & PGA_DEQUEUE) != 0)
return (PQ_NONE);
return (as.queue);
}
/*
* vm_page_queue:
*
* Return the index of the queue containing m.
*/
static inline uint8_t
vm_page_queue(vm_page_t m)
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
{
return (_vm_page_queue(vm_page_astate_load(m)));
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
}
static inline bool
vm_page_active(vm_page_t m)
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
{
return (vm_page_queue(m) == PQ_ACTIVE);
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
}
static inline bool
vm_page_inactive(vm_page_t m)
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
{
return (vm_page_queue(m) == PQ_INACTIVE);
Introduce a new page queue, PQ_LAUNDRY, for storing unreferenced, dirty pages, specificially, dirty pages that have passed once through the inactive queue. A new, dedicated thread is responsible for both deciding when to launder pages and actually laundering them. The new policy uses the relative sizes of the inactive and laundry queues to determine whether to launder pages at a given point in time. In general, this leads to more intelligent swapping behavior, since the laundry thread will avoid pageouts when the marginal benefit of doing so is low. Previously, without a dedicated queue for dirty pages, the page daemon didn't have the information to determine whether pageout provides any benefit to the system. Thus, the previous policy often resulted in small but steadily increasing amounts of swap usage when the system is under memory pressure, even when the inactive queue consisted mostly of clean pages. This change addresses that issue, and also paves the way for some future virtual memory system improvements by removing the last source of object-cached clean pages, i.e., PG_CACHE pages. The new laundry thread sleeps while waiting for a request from the page daemon thread(s). A request is raised by setting the variable vm_laundry_request and waking the laundry thread. We request launderings for two reasons: to try and balance the inactive and laundry queue sizes ("background laundering"), and to quickly make up for a shortage of free pages and clean inactive pages ("shortfall laundering"). When background laundering is requested, the laundry thread computes the number of page daemon wakeups that have taken place since the last laundering. If this number is large enough relative to the ratio of the laundry and (global) inactive queue sizes, we will launder vm_background_launder_target pages at vm_background_launder_rate KB/s. Otherwise, the laundry thread goes back to sleep without doing any work. When scanning the laundry queue during background laundering, reactivated pages are counted towards the laundry thread's target. In contrast, shortfall laundering is requested when an inactive queue scan fails to meet its target. In this case, the laundry thread attempts to launder enough pages to meet v_free_target within 0.5s, which is the inactive queue scan period. A laundry request can be latched while another is currently being serviced. In particular, a shortfall request will immediately preempt a background laundering. This change also redefines the meaning of vm_cnt.v_reactivated and removes the functions vm_page_cache() and vm_page_try_to_cache(). The new meaning of vm_cnt.v_reactivated now better reflects its name. It represents the number of inactive or laundry pages that are returned to the active queue on account of a reference. In collaboration with: markj Reviewed by: kib Tested by: pho Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8302
2016-11-09 18:48:37 +00:00
}
static inline bool
vm_page_in_laundry(vm_page_t m)
{
uint8_t queue;
queue = vm_page_queue(m);
return (queue == PQ_LAUNDRY || queue == PQ_UNSWAPPABLE);
}
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
/*
* vm_page_drop:
*
* Release a reference to a page and return the old reference count.
*/
static inline u_int
vm_page_drop(vm_page_t m, u_int val)
{
u_int old;
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
/*
* Synchronize with vm_page_free_prep(): ensure that all updates to the
* page structure are visible before it is freed.
*/
atomic_thread_fence_rel();
old = atomic_fetchadd_int(&m->ref_count, -val);
KASSERT(old != VPRC_BLOCKED,
("vm_page_drop: page %p has an invalid refcount value", m));
return (old);
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
}
/*
* vm_page_wired:
*
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
* Perform a racy check to determine whether a reference prevents the page
* from being reclaimable. If the page's object is locked, and the page is
* unmapped and exclusively busied by the current thread, no new wirings
* may be created.
*/
static inline bool
vm_page_wired(vm_page_t m)
{
Change synchonization rules for vm_page reference counting. There are several mechanisms by which a vm_page reference is held, preventing the page from being freed back to the page allocator. In particular, holding the page's object lock is sufficient to prevent the page from being freed; holding the busy lock or a wiring is sufficent as well. These references are protected by the page lock, which must therefore be acquired for many per-page operations. This results in false sharing since the page locks are external to the vm_page structures themselves and each lock protects multiple structures. Transition to using an atomically updated per-page reference counter. The object's reference is counted using a flag bit in the counter. A second flag bit is used to atomically block new references via pmap_extract_and_hold() while removing managed mappings of a page. Thus, the reference count of a page is guaranteed not to increase if the page is unbusied, unmapped, and the object's write lock is held. As a consequence of this, the page lock no longer protects a page's identity; operations which move pages between objects are now synchronized solely by the objects' locks. The vm_page_wire() and vm_page_unwire() KPIs are changed. The former requires that either the object lock or the busy lock is held. The latter no longer has a return value and may free the page if it releases the last reference to that page. vm_page_unwire_noq() behaves the same as before; the caller is responsible for checking its return value and freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is introduced for use in pmap_extract_and_hold(). It fails if the page is concurrently being unmapped, typically triggering a fallback to the fault handler. vm_page_wire() no longer requires the page lock and vm_page_unwire() now internally acquires the page lock when releasing the last wiring of a page (since the page lock still protects a page's queue state). In particular, synchronization details are no longer leaked into the caller. The change excises the page lock from several frequently executed code paths. In particular, vm_object_terminate() no longer bounces between page locks as it releases an object's pages, and direct I/O and sendfile(SF_NOCACHE) completions no longer require the page lock. In these latter cases we now get linear scalability in the common scenario where different threads are operating on different files. __FreeBSD_version is bumped. The DRM ports have been updated to accomodate the KPI changes. Reviewed by: jeff (earlier version) Tested by: gallatin (earlier version), pho Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
return (VPRC_WIRE_COUNT(m->ref_count) > 0);
}
static inline bool
vm_page_all_valid(vm_page_t m)
{
return (m->valid == VM_PAGE_BITS_ALL);
}
static inline bool
vm_page_any_valid(vm_page_t m)
{
return (m->valid != 0);
}
static inline bool
vm_page_none_valid(vm_page_t m)
{
return (m->valid == 0);
}
static inline int
vm_page_domain(vm_page_t m __numa_used)
{
#ifdef NUMA
int domn, segind;
segind = m->segind;
KASSERT(segind < vm_phys_nsegs, ("segind %d m %p", segind, m));
domn = vm_phys_segs[segind].domain;
KASSERT(domn >= 0 && domn < vm_ndomains, ("domain %d m %p", domn, m));
return (domn);
#else
return (0);
#endif
}
#endif /* _KERNEL */
These changes embody the support of the fully coherent merged VM buffer cache, much higher filesystem I/O performance, and much better paging performance. It represents the culmination of over 6 months of R&D. The majority of the merged VM/cache work is by John Dyson. The following highlights the most significant changes. Additionally, there are (mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to support the new VM/buffer scheme. vfs_bio.c: Significant rewrite of most of vfs_bio to support the merged VM buffer cache scheme. The scheme is almost fully compatible with the old filesystem interface. Significant improvement in the number of opportunities for write clustering. vfs_cluster.c, vfs_subr.c Upgrade and performance enhancements in vfs layer code to support merged VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff. vm_object.c: Yet more improvements in the collapse code. Elimination of some windows that can cause list corruption. vm_pageout.c: Fixed it, it really works better now. Somehow in 2.0, some "enhancements" broke the code. This code has been reworked from the ground-up. vm_fault.c, vm_page.c, pmap.c, vm_object.c Support for small-block filesystems with merged VM/buffer cache scheme. pmap.c vm_map.c Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of kernel PTs. vm_glue.c Much simpler and more effective swapping code. No more gratuitous swapping. proc.h Fixed the problem that the p_lock flag was not being cleared on a fork. swap_pager.c, vnode_pager.c Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the code doesn't need it anymore. machdep.c Changes to better support the parameter values for the merged VM/buffer cache scheme. machdep.c, kern_exec.c, vm_glue.c Implemented a seperate submap for temporary exec string space and another one to contain process upages. This eliminates all map fragmentation problems that previously existed. ffs_inode.c, ufs_inode.c, ufs_readwrite.c Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on busy buffers. Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
#endif /* !_VM_PAGE_ */