From: Lorenzo Stoakes <ljs@kernel•org>
To: "David Hildenbrand (Arm)" <david@kernel•org>
Cc: Barry Song <baohua@kernel•org>,
Matthew Wilcox <willy@infradead•org>,
surenb@google•com, akpm@linux-foundation•org,
linux-mm@kvack•org, liam@infradead•org, vbabka@kernel•org,
rppt@kernel•org, mhocko@suse•com, jack@suse•cz,
pfalcato@suse•de, wanglian@kylinos•cn, chentao@kylinos•cn,
lianux.mm@gmail•com, kunwu.chan@gmail•com,
liyangouwen1@oppo•com, chrisl@kernel•org, kasong@tencent•com,
shikemeng@huaweicloud•com, nphamcs@gmail•com, bhe@redhat•com,
youngjun.park@lge•com, linux-arm-kernel@lists•infradead.org,
linux-kernel@vger•kernel.org, loongarch@lists•linux.dev,
linuxppc-dev@lists•ozlabs.org, linux-riscv@lists•infradead.org,
linux-s390@vger•kernel.org, Nanzhe Zhao <nzzhao@126•com>
Subject: Re: [PATCH v2 0/5] mm: reduce mmap_lock contention and improve page fault performance
Date: Tue, 19 May 2026 14:42:11 +0100 [thread overview]
Message-ID: <agxok1UzYWhLu06G@lucifer> (raw)
In-Reply-To: <b65722ee-6476-4038-bfbb-44a32b3544fd@kernel.org>
On Mon, May 18, 2026 at 11:53:37AM +0200, David Hildenbrand (Arm) wrote:
> On 5/17/26 10:45, Barry Song wrote:
> > On Sat, May 2, 2026 at 1:58 AM Matthew Wilcox <willy@infradead•org> wrote:
> >>
> >> On Sat, May 02, 2026 at 01:44:34AM +0800, Barry Song wrote:
> >>>
> >>> It doesn’t have to involve unmapping or applying mprotect to
> >>> the entire VMA—just a portion of it is sufficient.
> >>
> >> Yes, but that still fails to answer "does this actually happen". How much
> >> performance is all this complexity in the page fault handler buying us?
> >> If you don't answer this question, I'm just going to go in and rip it
> >> all out.
> >>
> >
> > Hi Matthew (and Lorenzo, Jan, and anyone else who may be
> > waiting for answers),
> >
> > As promised during LSF/MM/BPF, we conducted thorough
> > testing on Android phones to determine whether performing
> > I/O in `filemap_fault()` can block `vma_start_write()`.
> > I wanted to give a quick update on this question.
> >
> > Nanzhe at Xiaomi created tracing scripts and ran various
> > applications on Android devices with I/O performed under
> > the VMA lock in `filemap_fault()`. We found that:
> >
> > 1. There are very few cases where unmap() is blocked by
> > page faults. I assume this is due to buggy user code
> > or poor synchronization between reads and unmap().
> > So I assume it is not a problem.
> >
> > 2. We observed many cases where `vma_start_write()`
> > is blocked by page-fault I/O in some applications.
> > The blocking occurs in the `dup_mmap()` path during
> > fork().
> >
> > With Suren's commit fb49c455323ff ("fork: lock VMAs of
> > the parent process when forking"), we now always hold
> > `vma_write_lock()` for each VMA. Note that the
> > `mmap_lock` write lock is also held, which could lead to
> > chained waiting if page-fault I/O is performed without
> > releasing the VMA lock.
> >
> > My gut feeling is that Suren's commit may be overshooting,
> > so my rough idea is that we might want to do something like
> > the following (we haven't tested it yet and it might be
> > wrong):
> >
> > diff --git a/mm/mmap.c b/mm/mmap.c
> > index 2311ae7c2ff4..5ddaf297f31a 100644
> > --- a/mm/mmap.c
> > +++ b/mm/mmap.c
> > @@ -1762,7 +1762,13 @@ __latent_entropy int dup_mmap(struct mm_struct
> > *mm, struct mm_struct *oldmm)
> > for_each_vma(vmi, mpnt) {
> > struct file *file;
> >
> > - retval = vma_start_write_killable(mpnt);
> > + /*
> > + * For anonymous or writable private VMAs, prevent
> > + * concurrent CoW faults.
> > + */
> > + if (!mpnt->vm_file || (!(mpnt->vm_flags & VM_SHARED) &&
> > + (mpnt->vm_flags & VM_WRITE)))
> > + retval = vma_start_write_killable(mpnt);
>
> Likely is_cow_mapping() is what you would want to check to handle VMAs that
> could have anonymous pages in them.
Yes :) I made pretty much the same comment though I forgot the correct helper :P
>
> --
> Cheers,
>
> David
Cheers, Lorenzo
next prev parent reply other threads:[~2026-05-19 13:42 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-30 4:04 [PATCH v2 0/5] mm: reduce mmap_lock contention and improve page fault performance Barry Song (Xiaomi)
2026-04-30 4:04 ` [PATCH v2 1/5] mm/filemap: Retry fault by VMA lock if the lock was released for I/O Barry Song (Xiaomi)
2026-04-30 4:04 ` [PATCH v2 2/5] mm/swapin: Retry swapin " Barry Song (Xiaomi)
2026-04-30 4:04 ` [PATCH v2 3/5] mm: Move folio_lock_or_retry() and drop __folio_lock_or_retry() Barry Song (Xiaomi)
2026-04-30 4:04 ` [PATCH v2 4/5] mm: Don't retry page fault if folio is uptodate during swap-in Barry Song (Xiaomi)
2026-04-30 12:35 ` Matthew Wilcox
2026-05-01 16:11 ` Matthew Wilcox
2026-04-30 4:04 ` [PATCH v2 5/5] mm/filemap: Avoid retrying page faults on uptodate folios in filemap faults Barry Song (Xiaomi)
2026-04-30 12:37 ` [PATCH v2 0/5] mm: reduce mmap_lock contention and improve page fault performance Matthew Wilcox
2026-04-30 22:49 ` Barry Song
2026-05-01 14:56 ` Matthew Wilcox
2026-05-01 17:44 ` Barry Song
2026-05-01 17:57 ` Matthew Wilcox
2026-05-01 18:25 ` Barry Song
2026-05-01 19:39 ` Matthew Wilcox
2026-05-03 20:39 ` Barry Song
2026-05-03 13:13 ` Jan Kara
2026-05-03 19:55 ` Barry Song
2026-05-04 13:03 ` Jan Kara
2026-05-04 13:35 ` Barry Song
2026-05-04 14:15 ` Barry Song
2026-05-17 8:45 ` Barry Song
2026-05-18 9:46 ` Lorenzo Stoakes
2026-05-18 11:25 ` Barry Song
2026-05-18 16:17 ` Matthew Wilcox
2026-05-18 20:50 ` Barry Song
2026-05-18 19:56 ` Suren Baghdasaryan
2026-05-18 21:14 ` Barry Song
2026-05-19 12:45 ` Lorenzo Stoakes
2026-05-19 14:17 ` Liam R. Howlett
2026-05-19 22:01 ` Barry Song
2026-05-20 21:04 ` Matthew Wilcox
2026-05-20 21:14 ` Barry Song
2026-05-20 21:15 ` Matthew Wilcox
2026-05-20 21:35 ` David Hildenbrand (Arm)
2026-05-20 23:37 ` Barry Song
2026-05-22 15:53 ` Lorenzo Stoakes
2026-05-22 21:31 ` Barry Song
2026-05-22 2:33 ` Barry Song (Xiaomi)
2026-05-22 13:09 ` Matthew Wilcox
2026-05-22 13:36 ` Barry Song
2026-05-22 13:48 ` Barry Song
2026-05-22 15:42 ` Lorenzo Stoakes
2026-05-19 12:53 ` Lorenzo Stoakes
2026-05-19 21:18 ` Barry Song
2026-05-20 7:50 ` Lorenzo Stoakes
2026-05-20 9:07 ` Barry Song
2026-05-20 10:07 ` Lorenzo Stoakes
2026-05-20 16:20 ` Suren Baghdasaryan
2026-05-20 5:51 ` Suren Baghdasaryan
2026-05-22 15:39 ` Lorenzo Stoakes
2026-05-20 10:33 ` David Hildenbrand (Arm)
2026-05-20 12:55 ` Lorenzo Stoakes
2026-05-20 21:39 ` Yang Shi
2026-05-22 15:37 ` Lorenzo Stoakes
2026-05-19 12:43 ` Lorenzo Stoakes
2026-05-18 9:53 ` David Hildenbrand (Arm)
2026-05-19 13:42 ` Lorenzo Stoakes [this message]
2026-05-18 21:21 ` Yang Shi
2026-05-19 11:07 ` Barry Song
2026-05-19 13:34 ` Lorenzo Stoakes
2026-05-19 18:50 ` Yang Shi
2026-05-19 20:53 ` Yang Shi
2026-05-19 13:12 ` Lorenzo Stoakes
2026-05-19 13:39 ` Lorenzo Stoakes
2026-05-19 18:41 ` Yang Shi
2026-05-19 21:02 ` Yang Shi
2026-05-20 8:11 ` Lorenzo Stoakes
2026-05-01 15:52 ` Lorenzo Stoakes
2026-05-01 16:06 ` Matthew Wilcox
2026-05-01 17:09 ` Lorenzo Stoakes
2026-05-01 17:59 ` Barry Song
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=agxok1UzYWhLu06G@lucifer \
--to=ljs@kernel$(echo .)org \
--cc=akpm@linux-foundation$(echo .)org \
--cc=baohua@kernel$(echo .)org \
--cc=bhe@redhat$(echo .)com \
--cc=chentao@kylinos$(echo .)cn \
--cc=chrisl@kernel$(echo .)org \
--cc=david@kernel$(echo .)org \
--cc=jack@suse$(echo .)cz \
--cc=kasong@tencent$(echo .)com \
--cc=kunwu.chan@gmail$(echo .)com \
--cc=liam@infradead$(echo .)org \
--cc=lianux.mm@gmail$(echo .)com \
--cc=linux-arm-kernel@lists$(echo .)infradead.org \
--cc=linux-kernel@vger$(echo .)kernel.org \
--cc=linux-mm@kvack$(echo .)org \
--cc=linux-riscv@lists$(echo .)infradead.org \
--cc=linux-s390@vger$(echo .)kernel.org \
--cc=linuxppc-dev@lists$(echo .)ozlabs.org \
--cc=liyangouwen1@oppo$(echo .)com \
--cc=loongarch@lists$(echo .)linux.dev \
--cc=mhocko@suse$(echo .)com \
--cc=nphamcs@gmail$(echo .)com \
--cc=nzzhao@126$(echo .)com \
--cc=pfalcato@suse$(echo .)de \
--cc=rppt@kernel$(echo .)org \
--cc=shikemeng@huaweicloud$(echo .)com \
--cc=surenb@google$(echo .)com \
--cc=vbabka@kernel$(echo .)org \
--cc=wanglian@kylinos$(echo .)cn \
--cc=willy@infradead$(echo .)org \
--cc=youngjun.park@lge$(echo .)com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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