From: Michal Hocko <mhocko@kernel.org>
To: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: rientjes@google.com, akpm@linux-foundation.org,
aarcange@redhat.com, linux-kernel@vger.kernel.org,
linux-mm@kvack.org
Subject: Re: Multiple oom_reaper BUGs: unmap_page_range racing with exit_mmap
Date: Thu, 7 Dec 2017 17:30:03 +0100 [thread overview]
Message-ID: <20171207163003.GM20234@dhcp22.suse.cz> (raw)
In-Reply-To: <201712080044.BID56711.FFVOLMStJOQHOF@I-love.SAKURA.ne.jp>
On Fri 08-12-17 00:44:11, Tetsuo Handa wrote:
> Michal Hocko wrote:
> > David, could you test with this patch please?
>
> Even if this patch solved David's case, you need to update
>
> * tsk_is_oom_victim() cannot be set from under us
> * either because current->mm is already set to NULL
> * under task_lock before calling mmput and oom_mm is
> * set not NULL by the OOM killer only if current->mm
> * is found not NULL while holding the task_lock.
>
> part as well, for it is the explanation of why
> tsk_is_oom_victim() test was expected to work.
Yes, the same applies for mm_is_oom_victim. I will fixup s@tsk_@mm_@
here.
> Also, do we need to do
>
> set_bit(MMF_OOM_SKIP, &mm->flags);
>
> if mm_is_oom_victim(mm) == false?
I do not think we really need to set MMF_OOM_SKIP if we are not going to
synchronize.
> exit_mmap() is called means that nobody can reach this mm
> except ->signal->oom_mm, and mm_is_oom_victim(mm) == false
> means that this mm cannot be reached by ->signal->oom_mm .
>
> Then, I think we do not need to set MMF_OOM_SKIP on this mm
> at exit_mmap() if mm_is_oom_victim(mm) == false.
yes. I will fold the following in if this turned out to really address
David's issue. But I suspect this will be the case considering the NULL
pmd in the report which would suggest racing with free_pgtable...
Thanks for the review!
---
diff --git a/mm/mmap.c b/mm/mmap.c
index d00a06248ef1..e63b7a576670 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -3004,7 +3004,6 @@ void exit_mmap(struct mm_struct *mm)
/* Use -1 here to ensure all VMAs in the mm are unmapped */
unmap_vmas(&tlb, vma, 0, -1);
- set_bit(MMF_OOM_SKIP, &mm->flags);
if (unlikely(mm_is_oom_victim(mm))) {
/*
* Wait for oom_reap_task() to stop working on this
@@ -3012,12 +3011,13 @@ void exit_mmap(struct mm_struct *mm)
* calling down_read(), oom_reap_task() will not run
* on this "mm" post up_write().
*
- * tsk_is_oom_victim() cannot be set from under us
+ * mm_is_oom_victim() cannot be set from under us
* either because current->mm is already set to NULL
* under task_lock before calling mmput and oom_mm is
* set not NULL by the OOM killer only if current->mm
* is found not NULL while holding the task_lock.
*/
+ set_bit(MMF_OOM_SKIP, &mm->flags);
down_write(&mm->mmap_sem);
up_write(&mm->mmap_sem);
}
--
Michal Hocko
SUSE Labs
next prev parent reply other threads:[~2017-12-07 16:30 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-12-06 2:43 Multiple oom_reaper BUGs: unmap_page_range racing with exit_mmap David Rientjes
2017-12-06 2:58 ` David Rientjes
[not found] ` <201712060328.vB63SrDK069830@www262.sakura.ne.jp>
2017-12-06 7:48 ` David Rientjes
2017-12-06 9:00 ` Michal Hocko
[not found] ` <201712070720.vB77KlBQ009754@www262.sakura.ne.jp>
2017-12-07 8:28 ` Michal Hocko
2017-12-07 8:44 ` Michal Hocko
2017-12-07 11:00 ` Tetsuo Handa
2017-12-07 21:22 ` David Rientjes
2017-12-08 7:50 ` Michal Hocko
2017-12-06 11:37 ` Tetsuo Handa
2017-12-06 8:50 ` Michal Hocko
2017-12-06 8:31 ` Michal Hocko
2017-12-07 11:35 ` Michal Hocko
2017-12-07 15:44 ` Tetsuo Handa
2017-12-07 16:30 ` Michal Hocko [this message]
2017-12-07 21:55 ` David Rientjes
2017-12-08 9:26 ` David Rientjes
2017-12-08 11:27 ` Michal Hocko
2017-12-08 10:11 ` Tetsuo Handa
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20171207163003.GM20234@dhcp22.suse.cz \
--to=mhocko@kernel.org \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=penguin-kernel@I-love.SAKURA.ne.jp \
--cc=rientjes@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).