Re: [PATCH v3 3/3] mm: introduce mmap_lock_speculate_{try_begin|retry}

[Suren Baghdasaryan <surenb@google.com>]

On Sun, Nov 24, 2024 at 4:58 PM Wei Yang <richard.weiyang@gmail.com> wrote:
>
> On Fri, Nov 22, 2024 at 09:44:16AM -0800, Suren Baghdasaryan wrote:
> >Add helper functions to speculatively perform operations without
> >read-locking mmap_lock, expecting that mmap_lock will not be
> >write-locked and mm is not modified from under us.
> >
> >Suggested-by: Peter Zijlstra <peterz@infradead.org>
> >Signed-off-by: Suren Baghdasaryan <surenb@google.com>
> >Reviewed-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
> >---
> >Changes since v2 [1]
> >- Added SOB, per Liam Howlett
> >
> >[1] https://lore.kernel.org/all/20241121162826.987947-3-surenb@google.com/
> >
> > include/linux/mmap_lock.h | 33 +++++++++++++++++++++++++++++++--
> > 1 file changed, 31 insertions(+), 2 deletions(-)
> >
> >diff --git a/include/linux/mmap_lock.h b/include/linux/mmap_lock.h
> >index 9715326f5a85..8ac3041df053 100644
> >--- a/include/linux/mmap_lock.h
> >+++ b/include/linux/mmap_lock.h
> >@@ -71,6 +71,7 @@ static inline void mmap_assert_write_locked(const struct mm_struct *mm)
> > }
> >
> > #ifdef CONFIG_PER_VMA_LOCK
> >+
> > static inline void mm_lock_seqcount_init(struct mm_struct *mm)
> > {
> >       seqcount_init(&mm->mm_lock_seq);
> >@@ -87,11 +88,39 @@ static inline void mm_lock_seqcount_end(struct mm_struct *mm)
> >       do_raw_write_seqcount_end(&mm->mm_lock_seq);
> > }
> >
> >-#else
> >+static inline bool mmap_lock_speculate_try_begin(struct mm_struct *mm, unsigned int *seq)
> >+{
> >+      /*
> >+       * Since mmap_lock is a sleeping lock, and waiting for it to become
> >+       * unlocked is more or less equivalent with taking it ourselves, don't
> >+       * bother with the speculative path if mmap_lock is already write-locked
> >+       * and take the slow path, which takes the lock.
> >+       */
> >+      return raw_seqcount_try_begin(&mm->mm_lock_seq, *seq);
> >+}
> >+
> >+static inline bool mmap_lock_speculate_retry(struct mm_struct *mm, unsigned int seq)
> >+{
> >+      return do_read_seqcount_retry(&mm->mm_lock_seq, seq);
>
> Just curious why we don't use read_seqcount_retry().
>
> Looks this is the only user outside seqlock.h.

Ah, good eye! read_seqcount_retry() would be better.

Peter, do you want me to post a new patchset or you can patch it when
picking it up?

>
> >+}
> >+
> >+#else /* CONFIG_PER_VMA_LOCK */
> >+
> > static inline void mm_lock_seqcount_init(struct mm_struct *mm) {}
> > static inline void mm_lock_seqcount_begin(struct mm_struct *mm) {}
> > static inline void mm_lock_seqcount_end(struct mm_struct *mm) {}
> >-#endif
> >+
> >+static inline bool mmap_lock_speculate_try_begin(struct mm_struct *mm, unsigned int *seq)
> >+{
> >+      return false;
> >+}
> >+
> >+static inline bool mmap_lock_speculate_retry(struct mm_struct *mm, unsigned int seq)
> >+{
> >+      return true;
> >+}
> >+
> >+#endif /* CONFIG_PER_VMA_LOCK */
> >
> > static inline void mmap_init_lock(struct mm_struct *mm)
> > {
> >--
> >2.47.0.371.ga323438b13-goog
>
> --
> Wei Yang
> Help you, Help me