diff options
author | KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> | 2010-01-08 17:42:56 -0500 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2010-01-11 12:34:06 -0500 |
commit | 1306d603fcf1f6682f8575d1ff23631a24184b21 (patch) | |
tree | 8e2278cfabf874cb16bf03bd182ae75b87055741 | |
parent | f146aabfe921006b98dfa4a78506763aedfd3206 (diff) |
proc: partially revert "procfs: provide stack information for threads"
Commit d899bf7b (procfs: provide stack information for threads) introduced
to show stack information in /proc/{pid}/status. But it cause large
performance regression. Unfortunately /proc/{pid}/status is used ps
command too and ps is one of most important component. Because both to
take mmap_sem and page table walk are heavily operation.
If many process run, the ps performance is,
[before d899bf7b]
% perf stat ps >/dev/null
Performance counter stats for 'ps':
4090.435806 task-clock-msecs # 0.032 CPUs
229 context-switches # 0.000 M/sec
0 CPU-migrations # 0.000 M/sec
234 page-faults # 0.000 M/sec
8587565207 cycles # 2099.425 M/sec
9866662403 instructions # 1.149 IPC
3789415411 cache-references # 926.409 M/sec
30419509 cache-misses # 7.437 M/sec
128.859521955 seconds time elapsed
[after d899bf7b]
% perf stat ps > /dev/null
Performance counter stats for 'ps':
4305.081146 task-clock-msecs # 0.028 CPUs
480 context-switches # 0.000 M/sec
2 CPU-migrations # 0.000 M/sec
237 page-faults # 0.000 M/sec
9021211334 cycles # 2095.480 M/sec
10605887536 instructions # 1.176 IPC
3612650999 cache-references # 839.160 M/sec
23917502 cache-misses # 5.556 M/sec
152.277819582 seconds time elapsed
Thus, this patch revert it. Fortunately /proc/{pid}/task/{tid}/smaps
provide almost same information. we can use it.
Commit d899bf7b introduced two features:
1) Add the annotattion of [thread stack: xxxx] mark to
/proc/{pid}/task/{tid}/maps.
2) Add StackUsage field to /proc/{pid}/status.
I only revert (2), because I haven't seen (1) cause regression.
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Stefani Seibold <stefani@seibold.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-rw-r--r-- | Documentation/filesystems/proc.txt | 2 | ||||
-rw-r--r-- | fs/proc/array.c | 89 |
2 files changed, 0 insertions, 91 deletions
diff --git a/Documentation/filesystems/proc.txt b/Documentation/filesystems/proc.txt index 220cc6376ef8..0d07513a67a6 100644 --- a/Documentation/filesystems/proc.txt +++ b/Documentation/filesystems/proc.txt | |||
@@ -177,7 +177,6 @@ read the file /proc/PID/status: | |||
177 | CapBnd: ffffffffffffffff | 177 | CapBnd: ffffffffffffffff |
178 | voluntary_ctxt_switches: 0 | 178 | voluntary_ctxt_switches: 0 |
179 | nonvoluntary_ctxt_switches: 1 | 179 | nonvoluntary_ctxt_switches: 1 |
180 | Stack usage: 12 kB | ||
181 | 180 | ||
182 | This shows you nearly the same information you would get if you viewed it with | 181 | This shows you nearly the same information you would get if you viewed it with |
183 | the ps command. In fact, ps uses the proc file system to obtain its | 182 | the ps command. In fact, ps uses the proc file system to obtain its |
@@ -231,7 +230,6 @@ Table 1-2: Contents of the statm files (as of 2.6.30-rc7) | |||
231 | Mems_allowed_list Same as previous, but in "list format" | 230 | Mems_allowed_list Same as previous, but in "list format" |
232 | voluntary_ctxt_switches number of voluntary context switches | 231 | voluntary_ctxt_switches number of voluntary context switches |
233 | nonvoluntary_ctxt_switches number of non voluntary context switches | 232 | nonvoluntary_ctxt_switches number of non voluntary context switches |
234 | Stack usage: stack usage high water mark (round up to page size) | ||
235 | .............................................................................. | 233 | .............................................................................. |
236 | 234 | ||
237 | Table 1-3: Contents of the statm files (as of 2.6.8-rc3) | 235 | Table 1-3: Contents of the statm files (as of 2.6.8-rc3) |
diff --git a/fs/proc/array.c b/fs/proc/array.c index f560325c444f..13b5d0708175 100644 --- a/fs/proc/array.c +++ b/fs/proc/array.c | |||
@@ -327,94 +327,6 @@ static inline void task_context_switch_counts(struct seq_file *m, | |||
327 | p->nivcsw); | 327 | p->nivcsw); |
328 | } | 328 | } |
329 | 329 | ||
330 | #ifdef CONFIG_MMU | ||
331 | |||
332 | struct stack_stats { | ||
333 | struct vm_area_struct *vma; | ||
334 | unsigned long startpage; | ||
335 | unsigned long usage; | ||
336 | }; | ||
337 | |||
338 | static int stack_usage_pte_range(pmd_t *pmd, unsigned long addr, | ||
339 | unsigned long end, struct mm_walk *walk) | ||
340 | { | ||
341 | struct stack_stats *ss = walk->private; | ||
342 | struct vm_area_struct *vma = ss->vma; | ||
343 | pte_t *pte, ptent; | ||
344 | spinlock_t *ptl; | ||
345 | int ret = 0; | ||
346 | |||
347 | pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); | ||
348 | for (; addr != end; pte++, addr += PAGE_SIZE) { | ||
349 | ptent = *pte; | ||
350 | |||
351 | #ifdef CONFIG_STACK_GROWSUP | ||
352 | if (pte_present(ptent) || is_swap_pte(ptent)) | ||
353 | ss->usage = addr - ss->startpage + PAGE_SIZE; | ||
354 | #else | ||
355 | if (pte_present(ptent) || is_swap_pte(ptent)) { | ||
356 | ss->usage = ss->startpage - addr + PAGE_SIZE; | ||
357 | pte++; | ||
358 | ret = 1; | ||
359 | break; | ||
360 | } | ||
361 | #endif | ||
362 | } | ||
363 | pte_unmap_unlock(pte - 1, ptl); | ||
364 | cond_resched(); | ||
365 | return ret; | ||
366 | } | ||
367 | |||
368 | static inline unsigned long get_stack_usage_in_bytes(struct vm_area_struct *vma, | ||
369 | struct task_struct *task) | ||
370 | { | ||
371 | struct stack_stats ss; | ||
372 | struct mm_walk stack_walk = { | ||
373 | .pmd_entry = stack_usage_pte_range, | ||
374 | .mm = vma->vm_mm, | ||
375 | .private = &ss, | ||
376 | }; | ||
377 | |||
378 | if (!vma->vm_mm || is_vm_hugetlb_page(vma)) | ||
379 | return 0; | ||
380 | |||
381 | ss.vma = vma; | ||
382 | ss.startpage = task->stack_start & PAGE_MASK; | ||
383 | ss.usage = 0; | ||
384 | |||
385 | #ifdef CONFIG_STACK_GROWSUP | ||
386 | walk_page_range(KSTK_ESP(task) & PAGE_MASK, vma->vm_end, | ||
387 | &stack_walk); | ||
388 | #else | ||
389 | walk_page_range(vma->vm_start, (KSTK_ESP(task) & PAGE_MASK) + PAGE_SIZE, | ||
390 | &stack_walk); | ||
391 | #endif | ||
392 | return ss.usage; | ||
393 | } | ||
394 | |||
395 | static inline void task_show_stack_usage(struct seq_file *m, | ||
396 | struct task_struct *task) | ||
397 | { | ||
398 | struct vm_area_struct *vma; | ||
399 | struct mm_struct *mm = get_task_mm(task); | ||
400 | |||
401 | if (mm) { | ||
402 | down_read(&mm->mmap_sem); | ||
403 | vma = find_vma(mm, task->stack_start); | ||
404 | if (vma) | ||
405 | seq_printf(m, "Stack usage:\t%lu kB\n", | ||
406 | get_stack_usage_in_bytes(vma, task) >> 10); | ||
407 | |||
408 | up_read(&mm->mmap_sem); | ||
409 | mmput(mm); | ||
410 | } | ||
411 | } | ||
412 | #else | ||
413 | static void task_show_stack_usage(struct seq_file *m, struct task_struct *task) | ||
414 | { | ||
415 | } | ||
416 | #endif /* CONFIG_MMU */ | ||
417 | |||
418 | static void task_cpus_allowed(struct seq_file *m, struct task_struct *task) | 330 | static void task_cpus_allowed(struct seq_file *m, struct task_struct *task) |
419 | { | 331 | { |
420 | seq_printf(m, "Cpus_allowed:\t"); | 332 | seq_printf(m, "Cpus_allowed:\t"); |
@@ -445,7 +357,6 @@ int proc_pid_status(struct seq_file *m, struct pid_namespace *ns, | |||
445 | task_show_regs(m, task); | 357 | task_show_regs(m, task); |
446 | #endif | 358 | #endif |
447 | task_context_switch_counts(m, task); | 359 | task_context_switch_counts(m, task); |
448 | task_show_stack_usage(m, task); | ||
449 | return 0; | 360 | return 0; |
450 | } | 361 | } |
451 | 362 | ||