linux/kernel/profile.c

506 lines
14 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* linux/kernel/profile.c
* Simple profiling. Manages a direct-mapped profile hit count buffer,
* with configurable resolution, support for restricting the cpus on
* which profiling is done, and switching between cpu time and
* schedule() calls via kernel command line parameters passed at boot.
*
* Scheduler profiling support, Arjan van de Ven and Ingo Molnar,
* Red Hat, July 2004
* Consolidation of architecture support code for profiling,
* Nadia Yvette Chambers, Oracle, July 2004
* Amortized hit count accounting via per-cpu open-addressed hashtables
* to resolve timer interrupt livelocks, Nadia Yvette Chambers,
* Oracle, 2004
*/
#include <linux/export.h>
#include <linux/profile.h>
mm: remove include/linux/bootmem.h Move remaining definitions and declarations from include/linux/bootmem.h into include/linux/memblock.h and remove the redundant header. The includes were replaced with the semantic patch below and then semi-automated removal of duplicated '#include <linux/memblock.h> @@ @@ - #include <linux/bootmem.h> + #include <linux/memblock.h> [sfr@canb.auug.org.au: dma-direct: fix up for the removal of linux/bootmem.h] Link: http://lkml.kernel.org/r/20181002185342.133d1680@canb.auug.org.au [sfr@canb.auug.org.au: powerpc: fix up for removal of linux/bootmem.h] Link: http://lkml.kernel.org/r/20181005161406.73ef8727@canb.auug.org.au [sfr@canb.auug.org.au: x86/kaslr, ACPI/NUMA: fix for linux/bootmem.h removal] Link: http://lkml.kernel.org/r/20181008190341.5e396491@canb.auug.org.au Link: http://lkml.kernel.org/r/1536927045-23536-30-git-send-email-rppt@linux.vnet.ibm.com Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Chris Zankel <chris@zankel.net> Cc: "David S. Miller" <davem@davemloft.net> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Greentime Hu <green.hu@gmail.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Guan Xuetao <gxt@pku.edu.cn> Cc: Ingo Molnar <mingo@redhat.com> Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> Cc: Jonas Bonn <jonas@southpole.se> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Ley Foon Tan <lftan@altera.com> Cc: Mark Salter <msalter@redhat.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Michal Simek <monstr@monstr.eu> Cc: Palmer Dabbelt <palmer@sifive.com> Cc: Paul Burton <paul.burton@mips.com> Cc: Richard Kuo <rkuo@codeaurora.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rich Felker <dalias@libc.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Serge Semin <fancer.lancer@gmail.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tony Luck <tony.luck@intel.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-30 22:09:49 +00:00
#include <linux/memblock.h>
#include <linux/notifier.h>
#include <linux/mm.h>
#include <linux/cpumask.h>
#include <linux/cpu.h>
#include <linux/highmem.h>
#include <linux/mutex.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/sched/stat.h>
#include <asm/sections.h>
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 13:55:46 +00:00
#include <asm/irq_regs.h>
#include <asm/ptrace.h>
struct profile_hit {
u32 pc, hits;
};
#define PROFILE_GRPSHIFT 3
#define PROFILE_GRPSZ (1 << PROFILE_GRPSHIFT)
#define NR_PROFILE_HIT (PAGE_SIZE/sizeof(struct profile_hit))
#define NR_PROFILE_GRP (NR_PROFILE_HIT/PROFILE_GRPSZ)
static atomic_t *prof_buffer;
static unsigned long prof_len;
static unsigned short int prof_shift;
int prof_on __read_mostly;
EXPORT_SYMBOL_GPL(prof_on);
static cpumask_var_t prof_cpu_mask;
#if defined(CONFIG_SMP) && defined(CONFIG_PROC_FS)
static DEFINE_PER_CPU(struct profile_hit *[2], cpu_profile_hits);
static DEFINE_PER_CPU(int, cpu_profile_flip);
static DEFINE_MUTEX(profile_flip_mutex);
#endif /* CONFIG_SMP */
int profile_setup(char *str)
{
static const char schedstr[] = "schedule";
static const char sleepstr[] = "sleep";
static const char kvmstr[] = "kvm";
int par;
if (!strncmp(str, sleepstr, strlen(sleepstr))) {
#ifdef CONFIG_SCHEDSTATS
sched/debug: Make schedstats a runtime tunable that is disabled by default schedstats is very useful during debugging and performance tuning but it incurs overhead to calculate the stats. As such, even though it can be disabled at build time, it is often enabled as the information is useful. This patch adds a kernel command-line and sysctl tunable to enable or disable schedstats on demand (when it's built in). It is disabled by default as someone who knows they need it can also learn to enable it when necessary. The benefits are dependent on how scheduler-intensive the workload is. If it is then the patch reduces the number of cycles spent calculating the stats with a small benefit from reducing the cache footprint of the scheduler. These measurements were taken from a 48-core 2-socket machine with Xeon(R) E5-2670 v3 cpus although they were also tested on a single socket machine 8-core machine with Intel i7-3770 processors. netperf-tcp 4.5.0-rc1 4.5.0-rc1 vanilla nostats-v3r1 Hmean 64 560.45 ( 0.00%) 575.98 ( 2.77%) Hmean 128 766.66 ( 0.00%) 795.79 ( 3.80%) Hmean 256 950.51 ( 0.00%) 981.50 ( 3.26%) Hmean 1024 1433.25 ( 0.00%) 1466.51 ( 2.32%) Hmean 2048 2810.54 ( 0.00%) 2879.75 ( 2.46%) Hmean 3312 4618.18 ( 0.00%) 4682.09 ( 1.38%) Hmean 4096 5306.42 ( 0.00%) 5346.39 ( 0.75%) Hmean 8192 10581.44 ( 0.00%) 10698.15 ( 1.10%) Hmean 16384 18857.70 ( 0.00%) 18937.61 ( 0.42%) Small gains here, UDP_STREAM showed nothing intresting and neither did the TCP_RR tests. The gains on the 8-core machine were very similar. tbench4 4.5.0-rc1 4.5.0-rc1 vanilla nostats-v3r1 Hmean mb/sec-1 500.85 ( 0.00%) 522.43 ( 4.31%) Hmean mb/sec-2 984.66 ( 0.00%) 1018.19 ( 3.41%) Hmean mb/sec-4 1827.91 ( 0.00%) 1847.78 ( 1.09%) Hmean mb/sec-8 3561.36 ( 0.00%) 3611.28 ( 1.40%) Hmean mb/sec-16 5824.52 ( 0.00%) 5929.03 ( 1.79%) Hmean mb/sec-32 10943.10 ( 0.00%) 10802.83 ( -1.28%) Hmean mb/sec-64 15950.81 ( 0.00%) 16211.31 ( 1.63%) Hmean mb/sec-128 15302.17 ( 0.00%) 15445.11 ( 0.93%) Hmean mb/sec-256 14866.18 ( 0.00%) 15088.73 ( 1.50%) Hmean mb/sec-512 15223.31 ( 0.00%) 15373.69 ( 0.99%) Hmean mb/sec-1024 14574.25 ( 0.00%) 14598.02 ( 0.16%) Hmean mb/sec-2048 13569.02 ( 0.00%) 13733.86 ( 1.21%) Hmean mb/sec-3072 12865.98 ( 0.00%) 13209.23 ( 2.67%) Small gains of 2-4% at low thread counts and otherwise flat. The gains on the 8-core machine were slightly different tbench4 on 8-core i7-3770 single socket machine Hmean mb/sec-1 442.59 ( 0.00%) 448.73 ( 1.39%) Hmean mb/sec-2 796.68 ( 0.00%) 794.39 ( -0.29%) Hmean mb/sec-4 1322.52 ( 0.00%) 1343.66 ( 1.60%) Hmean mb/sec-8 2611.65 ( 0.00%) 2694.86 ( 3.19%) Hmean mb/sec-16 2537.07 ( 0.00%) 2609.34 ( 2.85%) Hmean mb/sec-32 2506.02 ( 0.00%) 2578.18 ( 2.88%) Hmean mb/sec-64 2511.06 ( 0.00%) 2569.16 ( 2.31%) Hmean mb/sec-128 2313.38 ( 0.00%) 2395.50 ( 3.55%) Hmean mb/sec-256 2110.04 ( 0.00%) 2177.45 ( 3.19%) Hmean mb/sec-512 2072.51 ( 0.00%) 2053.97 ( -0.89%) In constract, this shows a relatively steady 2-3% gain at higher thread counts. Due to the nature of the patch and the type of workload, it's not a surprise that the result will depend on the CPU used. hackbench-pipes 4.5.0-rc1 4.5.0-rc1 vanilla nostats-v3r1 Amean 1 0.0637 ( 0.00%) 0.0660 ( -3.59%) Amean 4 0.1229 ( 0.00%) 0.1181 ( 3.84%) Amean 7 0.1921 ( 0.00%) 0.1911 ( 0.52%) Amean 12 0.3117 ( 0.00%) 0.2923 ( 6.23%) Amean 21 0.4050 ( 0.00%) 0.3899 ( 3.74%) Amean 30 0.4586 ( 0.00%) 0.4433 ( 3.33%) Amean 48 0.5910 ( 0.00%) 0.5694 ( 3.65%) Amean 79 0.8663 ( 0.00%) 0.8626 ( 0.43%) Amean 110 1.1543 ( 0.00%) 1.1517 ( 0.22%) Amean 141 1.4457 ( 0.00%) 1.4290 ( 1.16%) Amean 172 1.7090 ( 0.00%) 1.6924 ( 0.97%) Amean 192 1.9126 ( 0.00%) 1.9089 ( 0.19%) Some small gains and losses and while the variance data is not included, it's close to the noise. The UMA machine did not show anything particularly different pipetest 4.5.0-rc1 4.5.0-rc1 vanilla nostats-v2r2 Min Time 4.13 ( 0.00%) 3.99 ( 3.39%) 1st-qrtle Time 4.38 ( 0.00%) 4.27 ( 2.51%) 2nd-qrtle Time 4.46 ( 0.00%) 4.39 ( 1.57%) 3rd-qrtle Time 4.56 ( 0.00%) 4.51 ( 1.10%) Max-90% Time 4.67 ( 0.00%) 4.60 ( 1.50%) Max-93% Time 4.71 ( 0.00%) 4.65 ( 1.27%) Max-95% Time 4.74 ( 0.00%) 4.71 ( 0.63%) Max-99% Time 4.88 ( 0.00%) 4.79 ( 1.84%) Max Time 4.93 ( 0.00%) 4.83 ( 2.03%) Mean Time 4.48 ( 0.00%) 4.39 ( 1.91%) Best99%Mean Time 4.47 ( 0.00%) 4.39 ( 1.91%) Best95%Mean Time 4.46 ( 0.00%) 4.38 ( 1.93%) Best90%Mean Time 4.45 ( 0.00%) 4.36 ( 1.98%) Best50%Mean Time 4.36 ( 0.00%) 4.25 ( 2.49%) Best10%Mean Time 4.23 ( 0.00%) 4.10 ( 3.13%) Best5%Mean Time 4.19 ( 0.00%) 4.06 ( 3.20%) Best1%Mean Time 4.13 ( 0.00%) 4.00 ( 3.39%) Small improvement and similar gains were seen on the UMA machine. The gain is small but it stands to reason that doing less work in the scheduler is a good thing. The downside is that the lack of schedstats and tracepoints may be surprising to experts doing performance analysis until they find the existence of the schedstats= parameter or schedstats sysctl. It will be automatically activated for latencytop and sleep profiling to alleviate the problem. For tracepoints, there is a simple warning as it's not safe to activate schedstats in the context when it's known the tracepoint may be wanted but is unavailable. Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Reviewed-by: Matt Fleming <matt@codeblueprint.co.uk> Reviewed-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <mgalbraith@suse.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1454663316-22048-1-git-send-email-mgorman@techsingularity.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-02-05 09:08:36 +00:00
force_schedstat_enabled();
prof_on = SLEEP_PROFILING;
if (str[strlen(sleepstr)] == ',')
str += strlen(sleepstr) + 1;
if (get_option(&str, &par))
prof_shift = clamp(par, 0, BITS_PER_LONG - 1);
pr_info("kernel sleep profiling enabled (shift: %u)\n",
prof_shift);
#else
pr_warn("kernel sleep profiling requires CONFIG_SCHEDSTATS\n");
#endif /* CONFIG_SCHEDSTATS */
} else if (!strncmp(str, schedstr, strlen(schedstr))) {
prof_on = SCHED_PROFILING;
if (str[strlen(schedstr)] == ',')
str += strlen(schedstr) + 1;
if (get_option(&str, &par))
prof_shift = clamp(par, 0, BITS_PER_LONG - 1);
pr_info("kernel schedule profiling enabled (shift: %u)\n",
prof_shift);
} else if (!strncmp(str, kvmstr, strlen(kvmstr))) {
prof_on = KVM_PROFILING;
if (str[strlen(kvmstr)] == ',')
str += strlen(kvmstr) + 1;
if (get_option(&str, &par))
prof_shift = clamp(par, 0, BITS_PER_LONG - 1);
pr_info("kernel KVM profiling enabled (shift: %u)\n",
prof_shift);
} else if (get_option(&str, &par)) {
prof_shift = clamp(par, 0, BITS_PER_LONG - 1);
prof_on = CPU_PROFILING;
pr_info("kernel profiling enabled (shift: %u)\n",
prof_shift);
}
return 1;
}
__setup("profile=", profile_setup);
int __ref profile_init(void)
{
int buffer_bytes;
if (!prof_on)
return 0;
/* only text is profiled */
prof_len = (_etext - _stext) >> prof_shift;
profiling: fix shift too large makes kernel panic 2d186afd04d6 ("profiling: fix shift-out-of-bounds bugs") limits shift value by [0, BITS_PER_LONG -1], which means [0, 63]. However, syzbot found that the max shift value should be the bit number of (_etext - _stext). If shift is outside of this, the "buffer_bytes" will be zero and will cause kzalloc(0). Then the kernel panics due to dereferencing the returned pointer 16. This can be easily reproduced by passing a large number like 60 to enable profiling and then run readprofile. LOGS: BUG: kernel NULL pointer dereference, address: 0000000000000010 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD 6148067 P4D 6148067 PUD 6142067 PMD 0 PREEMPT SMP CPU: 4 PID: 184 Comm: readprofile Not tainted 5.18.0+ #162 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014 RIP: 0010:read_profile+0x104/0x220 RSP: 0018:ffffc900006fbe80 EFLAGS: 00000202 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 RDX: ffff888006150000 RSI: 0000000000000001 RDI: ffffffff82aba4a0 RBP: 000000000188bb60 R08: 0000000000000010 R09: ffff888006151000 R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff82aba4a0 R13: 0000000000000000 R14: ffffc900006fbf08 R15: 0000000000020c30 FS: 000000000188a8c0(0000) GS:ffff88803ed00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000010 CR3: 0000000006144000 CR4: 00000000000006e0 Call Trace: <TASK> proc_reg_read+0x56/0x70 vfs_read+0x9a/0x1b0 ksys_read+0xa1/0xe0 ? fpregs_assert_state_consistent+0x1e/0x40 do_syscall_64+0x3a/0x80 entry_SYSCALL_64_after_hwframe+0x46/0xb0 RIP: 0033:0x4d4b4e RSP: 002b:00007ffebb668d58 EFLAGS: 00000246 ORIG_RAX: 0000000000000000 RAX: ffffffffffffffda RBX: 000000000188a8a0 RCX: 00000000004d4b4e RDX: 0000000000000400 RSI: 000000000188bb60 RDI: 0000000000000003 RBP: 0000000000000003 R08: 000000000000006e R09: 0000000000000000 R10: 0000000000000041 R11: 0000000000000246 R12: 000000000188bb60 R13: 0000000000000400 R14: 0000000000000000 R15: 000000000188bb60 </TASK> Modules linked in: CR2: 0000000000000010 Killed ---[ end trace 0000000000000000 ]--- Check prof_len in profile_init() to prevent it be zero. Link: https://lkml.kernel.org/r/20220531012854.229439-1-chenzhongjin@huawei.com Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Chen Zhongjin <chenzhongjin@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-05-31 01:28:54 +00:00
if (!prof_len) {
pr_warn("profiling shift: %u too large\n", prof_shift);
prof_on = 0;
return -EINVAL;
}
buffer_bytes = prof_len*sizeof(atomic_t);
if (!alloc_cpumask_var(&prof_cpu_mask, GFP_KERNEL))
return -ENOMEM;
cpumask_copy(prof_cpu_mask, cpu_possible_mask);
prof_buffer = kzalloc(buffer_bytes, GFP_KERNEL|__GFP_NOWARN);
if (prof_buffer)
return 0;
prof_buffer = alloc_pages_exact(buffer_bytes,
GFP_KERNEL|__GFP_ZERO|__GFP_NOWARN);
if (prof_buffer)
return 0;
prof_buffer = vzalloc(buffer_bytes);
if (prof_buffer)
return 0;
free_cpumask_var(prof_cpu_mask);
return -ENOMEM;
}
#if defined(CONFIG_SMP) && defined(CONFIG_PROC_FS)
/*
* Each cpu has a pair of open-addressed hashtables for pending
* profile hits. read_profile() IPI's all cpus to request them
* to flip buffers and flushes their contents to prof_buffer itself.
* Flip requests are serialized by the profile_flip_mutex. The sole
* use of having a second hashtable is for avoiding cacheline
* contention that would otherwise happen during flushes of pending
* profile hits required for the accuracy of reported profile hits
* and so resurrect the interrupt livelock issue.
*
* The open-addressed hashtables are indexed by profile buffer slot
* and hold the number of pending hits to that profile buffer slot on
* a cpu in an entry. When the hashtable overflows, all pending hits
* are accounted to their corresponding profile buffer slots with
* atomic_add() and the hashtable emptied. As numerous pending hits
* may be accounted to a profile buffer slot in a hashtable entry,
* this amortizes a number of atomic profile buffer increments likely
* to be far larger than the number of entries in the hashtable,
* particularly given that the number of distinct profile buffer
* positions to which hits are accounted during short intervals (e.g.
* several seconds) is usually very small. Exclusion from buffer
* flipping is provided by interrupt disablement (note that for
* SCHED_PROFILING or SLEEP_PROFILING profile_hit() may be called from
* process context).
* The hash function is meant to be lightweight as opposed to strong,
* and was vaguely inspired by ppc64 firmware-supported inverted
* pagetable hash functions, but uses a full hashtable full of finite
* collision chains, not just pairs of them.
*
* -- nyc
*/
static void __profile_flip_buffers(void *unused)
{
int cpu = smp_processor_id();
per_cpu(cpu_profile_flip, cpu) = !per_cpu(cpu_profile_flip, cpu);
}
static void profile_flip_buffers(void)
{
int i, j, cpu;
mutex_lock(&profile_flip_mutex);
j = per_cpu(cpu_profile_flip, get_cpu());
put_cpu();
on_each_cpu(__profile_flip_buffers, NULL, 1);
for_each_online_cpu(cpu) {
struct profile_hit *hits = per_cpu(cpu_profile_hits, cpu)[j];
for (i = 0; i < NR_PROFILE_HIT; ++i) {
if (!hits[i].hits) {
if (hits[i].pc)
hits[i].pc = 0;
continue;
}
atomic_add(hits[i].hits, &prof_buffer[hits[i].pc]);
hits[i].hits = hits[i].pc = 0;
}
}
mutex_unlock(&profile_flip_mutex);
}
static void profile_discard_flip_buffers(void)
{
int i, cpu;
mutex_lock(&profile_flip_mutex);
i = per_cpu(cpu_profile_flip, get_cpu());
put_cpu();
on_each_cpu(__profile_flip_buffers, NULL, 1);
for_each_online_cpu(cpu) {
struct profile_hit *hits = per_cpu(cpu_profile_hits, cpu)[i];
memset(hits, 0, NR_PROFILE_HIT*sizeof(struct profile_hit));
}
mutex_unlock(&profile_flip_mutex);
}
static void do_profile_hits(int type, void *__pc, unsigned int nr_hits)
{
unsigned long primary, secondary, flags, pc = (unsigned long)__pc;
int i, j, cpu;
struct profile_hit *hits;
pc = min((pc - (unsigned long)_stext) >> prof_shift, prof_len - 1);
i = primary = (pc & (NR_PROFILE_GRP - 1)) << PROFILE_GRPSHIFT;
secondary = (~(pc << 1) & (NR_PROFILE_GRP - 1)) << PROFILE_GRPSHIFT;
cpu = get_cpu();
hits = per_cpu(cpu_profile_hits, cpu)[per_cpu(cpu_profile_flip, cpu)];
if (!hits) {
put_cpu();
return;
}
/*
* We buffer the global profiler buffer into a per-CPU
* queue and thus reduce the number of global (and possibly
* NUMA-alien) accesses. The write-queue is self-coalescing:
*/
local_irq_save(flags);
do {
for (j = 0; j < PROFILE_GRPSZ; ++j) {
if (hits[i + j].pc == pc) {
hits[i + j].hits += nr_hits;
goto out;
} else if (!hits[i + j].hits) {
hits[i + j].pc = pc;
hits[i + j].hits = nr_hits;
goto out;
}
}
i = (i + secondary) & (NR_PROFILE_HIT - 1);
} while (i != primary);
/*
* Add the current hit(s) and flush the write-queue out
* to the global buffer:
*/
atomic_add(nr_hits, &prof_buffer[pc]);
for (i = 0; i < NR_PROFILE_HIT; ++i) {
atomic_add(hits[i].hits, &prof_buffer[hits[i].pc]);
hits[i].pc = hits[i].hits = 0;
}
out:
local_irq_restore(flags);
put_cpu();
}
static int profile_dead_cpu(unsigned int cpu)
{
struct page *page;
int i;
if (cpumask_available(prof_cpu_mask))
cpumask_clear_cpu(cpu, prof_cpu_mask);
for (i = 0; i < 2; i++) {
if (per_cpu(cpu_profile_hits, cpu)[i]) {
page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[i]);
per_cpu(cpu_profile_hits, cpu)[i] = NULL;
__free_page(page);
}
}
return 0;
}
static int profile_prepare_cpu(unsigned int cpu)
{
int i, node = cpu_to_mem(cpu);
struct page *page;
per_cpu(cpu_profile_flip, cpu) = 0;
for (i = 0; i < 2; i++) {
if (per_cpu(cpu_profile_hits, cpu)[i])
continue;
page = __alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO, 0);
if (!page) {
profile_dead_cpu(cpu);
return -ENOMEM;
}
per_cpu(cpu_profile_hits, cpu)[i] = page_address(page);
}
return 0;
}
static int profile_online_cpu(unsigned int cpu)
{
if (cpumask_available(prof_cpu_mask))
cpumask_set_cpu(cpu, prof_cpu_mask);
return 0;
}
#else /* !CONFIG_SMP */
#define profile_flip_buffers() do { } while (0)
#define profile_discard_flip_buffers() do { } while (0)
static void do_profile_hits(int type, void *__pc, unsigned int nr_hits)
{
unsigned long pc;
pc = ((unsigned long)__pc - (unsigned long)_stext) >> prof_shift;
atomic_add(nr_hits, &prof_buffer[min(pc, prof_len - 1)]);
}
#endif /* !CONFIG_SMP */
void profile_hits(int type, void *__pc, unsigned int nr_hits)
{
if (prof_on != type || !prof_buffer)
return;
do_profile_hits(type, __pc, nr_hits);
}
EXPORT_SYMBOL_GPL(profile_hits);
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 13:55:46 +00:00
void profile_tick(int type)
{
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 13:55:46 +00:00
struct pt_regs *regs = get_irq_regs();
if (!user_mode(regs) && cpumask_available(prof_cpu_mask) &&
cpumask_test_cpu(smp_processor_id(), prof_cpu_mask))
profile_hit(type, (void *)profile_pc(regs));
}
#ifdef CONFIG_PROC_FS
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/uaccess.h>
static int prof_cpu_mask_proc_show(struct seq_file *m, void *v)
{
seq_printf(m, "%*pb\n", cpumask_pr_args(prof_cpu_mask));
return 0;
}
static int prof_cpu_mask_proc_open(struct inode *inode, struct file *file)
{
return single_open(file, prof_cpu_mask_proc_show, NULL);
}
static ssize_t prof_cpu_mask_proc_write(struct file *file,
const char __user *buffer, size_t count, loff_t *pos)
{
cpumask_var_t new_value;
int err;
if (!zalloc_cpumask_var(&new_value, GFP_KERNEL))
return -ENOMEM;
err = cpumask_parse_user(buffer, count, new_value);
if (!err) {
cpumask_copy(prof_cpu_mask, new_value);
err = count;
}
free_cpumask_var(new_value);
return err;
}
static const struct proc_ops prof_cpu_mask_proc_ops = {
.proc_open = prof_cpu_mask_proc_open,
.proc_read = seq_read,
.proc_lseek = seq_lseek,
.proc_release = single_release,
.proc_write = prof_cpu_mask_proc_write,
};
void create_prof_cpu_mask(void)
{
/* create /proc/irq/prof_cpu_mask */
proc_create("irq/prof_cpu_mask", 0600, NULL, &prof_cpu_mask_proc_ops);
}
/*
* This function accesses profiling information. The returned data is
* binary: the sampling step and the actual contents of the profile
* buffer. Use of the program readprofile is recommended in order to
* get meaningful info out of these data.
*/
static ssize_t
read_profile(struct file *file, char __user *buf, size_t count, loff_t *ppos)
{
unsigned long p = *ppos;
ssize_t read;
char *pnt;
unsigned long sample_step = 1UL << prof_shift;
profile_flip_buffers();
if (p >= (prof_len+1)*sizeof(unsigned int))
return 0;
if (count > (prof_len+1)*sizeof(unsigned int) - p)
count = (prof_len+1)*sizeof(unsigned int) - p;
read = 0;
while (p < sizeof(unsigned int) && count > 0) {
if (put_user(*((char *)(&sample_step)+p), buf))
return -EFAULT;
buf++; p++; count--; read++;
}
pnt = (char *)prof_buffer + p - sizeof(atomic_t);
if (copy_to_user(buf, (void *)pnt, count))
return -EFAULT;
read += count;
*ppos += read;
return read;
}
/* default is to not implement this call */
int __weak setup_profiling_timer(unsigned mult)
{
return -EINVAL;
}
/*
* Writing to /proc/profile resets the counters
*
* Writing a 'profiling multiplier' value into it also re-sets the profiling
* interrupt frequency, on architectures that support this.
*/
static ssize_t write_profile(struct file *file, const char __user *buf,
size_t count, loff_t *ppos)
{
#ifdef CONFIG_SMP
if (count == sizeof(int)) {
unsigned int multiplier;
if (copy_from_user(&multiplier, buf, sizeof(int)))
return -EFAULT;
if (setup_profiling_timer(multiplier))
return -EINVAL;
}
#endif
profile_discard_flip_buffers();
memset(prof_buffer, 0, prof_len * sizeof(atomic_t));
return count;
}
static const struct proc_ops profile_proc_ops = {
.proc_read = read_profile,
.proc_write = write_profile,
.proc_lseek = default_llseek,
};
int __ref create_proc_profile(void)
{
struct proc_dir_entry *entry;
#ifdef CONFIG_SMP
enum cpuhp_state online_state;
#endif
int err = 0;
if (!prof_on)
return 0;
#ifdef CONFIG_SMP
err = cpuhp_setup_state(CPUHP_PROFILE_PREPARE, "PROFILE_PREPARE",
profile_prepare_cpu, profile_dead_cpu);
if (err)
return err;
err = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "AP_PROFILE_ONLINE",
profile_online_cpu, NULL);
if (err < 0)
goto err_state_prep;
online_state = err;
err = 0;
#endif
entry = proc_create("profile", S_IWUSR | S_IRUGO,
NULL, &profile_proc_ops);
if (!entry)
goto err_state_onl;
proc_set_size(entry, (1 + prof_len) * sizeof(atomic_t));
return err;
err_state_onl:
#ifdef CONFIG_SMP
cpuhp_remove_state(online_state);
err_state_prep:
cpuhp_remove_state(CPUHP_PROFILE_PREPARE);
#endif
return err;
}
2014-04-03 21:48:35 +00:00
subsys_initcall(create_proc_profile);
#endif /* CONFIG_PROC_FS */