Re: [PATCH v5] mm: shrink skip folio mapped by an exiting process

[zhiguojiang <justinjiang@vivo.com>]

在 2024/7/8 20:17, zhiguojiang 写道:
>
>
> 在 2024/7/8 19:02, Barry Song 写道:
>> On Mon, Jul 8, 2024 at 9:04 PM Zhiguo Jiang <justinjiang@vivo.com> 
>> wrote:
>>> The releasing process of the non-shared anonymous folio mapped 
>>> solely by
>>> an exiting process may go through two flows: 1) the anonymous folio is
>>> firstly is swaped-out into swapspace and transformed into a swp_entry
>>> in shrink_folio_list; 2) then the swp_entry is released in the process
>>> exiting flow. This will increase the cpu load of releasing a non-shared
>>> anonymous folio mapped solely by an exiting process, because the folio
>>> go through swap-out and the releasing the swapspace and swp_entry.
>>>
>>> When system is low memory, it is more likely to occur, because more
>>> backend applidatuions will be killed.
>>>
>>> The modification is that shrink skips the non-shared anonymous folio
>>> solely mapped by an exting process and the folio is only released
>>> directly in the process exiting flow, which will save swap-out time
>>> and alleviate the load of the process exiting.
>>>
>>> Signed-off-by: Zhiguo Jiang <justinjiang@vivo.com>
>>> ---
>>>
>>> Change log:
>>> v4->v5:
>>> 1.Modify to skip non-shared anonymous folio only.
>>> 2.Update comments for pra->referenced = -1.
>>> v3->v4:
>>> 1.Modify that the unshared folios mapped only in exiting task are skip.
>>> v2->v3:
>>> Nothing.
>>> v1->v2:
>>> 1.The VM_EXITING added in v1 patch is removed, because it will fail
>>> to compile in 32-bit system.
>>>
>>>   mm/rmap.c   | 13 +++++++++++++
>>>   mm/vmscan.c |  7 ++++++-
>>>   2 files changed, 19 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/mm/rmap.c b/mm/rmap.c
>>> index 26806b49a86f..5b5281d71dbb
>>> --- a/mm/rmap.c
>>> +++ b/mm/rmap.c
>>> @@ -843,6 +843,19 @@ static bool folio_referenced_one(struct folio 
>>> *folio,
>>>          int referenced = 0;
>>>          unsigned long start = address, ptes = 0;
>>>
>>> +       /*
>>> +        * Skip the non-shared anonymous folio mapped solely by
>>> +        * the single exiting process, and release it directly
>>> +        * in the process exiting.
>>> +        */
>>> +       if ((!atomic_read(&vma->vm_mm->mm_users) ||
>>> +               test_bit(MMF_OOM_SKIP, &vma->vm_mm->flags)) &&
>>> +               folio_test_anon(folio) && 
>>> folio_test_swapbacked(folio) &&
>>> +               !folio_likely_mapped_shared(folio)) {
>>> +               pra->referenced = -1;
>>> +               return false;
>>> +       }
>>> +
>>>          while (page_vma_mapped_walk(&pvmw)) {
>>>                  address = pvmw.address;
> Sure, I agree with your modification suggestions. This way, using PTL 
> indeed sure
> that the folio is mapped by this process.
> Thanks
>> As David suggested, what about the below?
>>
>> @@ -883,6 +870,21 @@ static bool folio_referenced_one(struct folio 
>> *folio,
>>                          continue;
>>                  }
>>
>> +               /*
>> +                * Skip the non-shared anonymous folio mapped solely by
>> +                * the single exiting process, and release it directly
>> +                * in the process exiting.
>> +                */
>> +               if ((!atomic_read(&vma->vm_mm->mm_users) ||
>> +                                       test_bit(MMF_OOM_SKIP,
>> &vma->vm_mm->flags)) &&
>> +                               folio_test_anon(folio) &&
>> folio_test_swapbacked(folio) &&
>> + !folio_likely_mapped_shared(folio)) {
>> +                       pra->referenced = -1;
>> +                       page_vma_mapped_walk_done(&pvmw);
>> +                       return false;
>> +               }
>> +
>>                  if (pvmw.pte) {
>>                          if (lru_gen_enabled() &&
>>                              pte_young(ptep_get(pvmw.pte))) {
>>
>>
>> By the way, I am not convinced that using test_bit(MMF_OOM_SKIP,
>> &vma->vm_mm->flags) is
>> correct (I think it is wrong).   For example, global_init can 
>> directly have it:
>>                  if (is_global_init(p)) {
>>                          can_oom_reap = false;
>>                          set_bit(MMF_OOM_SKIP, &mm->flags);
>>                          pr_info("oom killer %d (%s) has mm pinned by 
>> %d (%s)\n",
>>                                          task_pid_nr(victim), 
>> victim->comm,
>>                                          task_pid_nr(p), p->comm);
>>                          continue;
>>                  }
>>
>> And exit_mmap() automatically has MMF_OOM_SKIP.
>>
>> What is the purpose of this check? Is there a better way to determine
>> if a process is an
>> OOM target? What about check_stable_address_space() ?
> 1.Sorry, I overlook the situation with if (is_global_init(p)), 
> MMF_OOM_SKIP is indeed not suitable.
>
> 2.check_stable_address_space() can indicate oom_reaper, but it seems 
> unable to identify the situation where the process exits normally. 
> What about task_is_dying()? static inline bool task_is_dying(void) { 
> return tsk_is_oom_victim(current) || fatal_signal_pending(current) || 
> (current->flags & PF_EXITING); } Thanks
We can migrate task_is_dying() from mm/memcontrol.c to include/linux/oom.h
> static inline bool task_is_dying(void)
> {
>     return tsk_is_oom_victim(current) || fatal_signal_pending(current) ||
>         (current->flags & PF_EXITING);
> }

>>
>>
>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>> index 0761f91b407f..bae7a8bf6b3d
>>> --- a/mm/vmscan.c
>>> +++ b/mm/vmscan.c
>>> @@ -863,7 +863,12 @@ static enum folio_references 
>>> folio_check_references(struct folio *folio,
>>>          if (vm_flags & VM_LOCKED)
>>>                  return FOLIOREF_ACTIVATE;
>>>
>>> -       /* rmap lock contention: rotate */
>>> +       /*
>>> +        * There are two cases to consider.
>>> +        * 1) Rmap lock contention: rotate.
>>> +        * 2) Skip the non-shared anonymous folio mapped solely by
>>> +        *    the single exiting process.
>>> +        */
>>>          if (referenced_ptes == -1)
>>>                  return FOLIOREF_KEEP;
>>>
>>> -- 
>>> 2.39.0
>>>
>> Thanks
>> Barry
>