Re: [PATCH v7 3/4] mm: add vma_assert_locked() for !CONFIG_PER_VMA_LOCK

From: Suren Baghdasaryan
Date: Thu Feb 15 2024 - 17:20:07 EST


On Thu, Feb 15, 2024 at 10:28 AM Lokesh Gidra <lokeshgidra@xxxxxxxxxx> wrote:
>
> vma_assert_locked() is needed to replace mmap_assert_locked() once we
> start using per-vma locks in userfaultfd operations.
>
> In !CONFIG_PER_VMA_LOCK case when mm is locked, it implies that the
> given VMA is locked.

Yes, makes sense. With per-vma locks used in more places, this makes
replacing mmap_assert_locked() with vma_assert_locked() very
straight-forward.

>
> Signed-off-by: Lokesh Gidra <lokeshgidra@xxxxxxxxxx>

Reviewed-by: Suren Baghdasaryan <surenb@xxxxxxxxxx>

> ---
> include/linux/mm.h | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 3c85634b186c..5ece3ad34ef8 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -781,6 +781,11 @@ static inline struct vm_area_struct *lock_vma_under_rcu(struct mm_struct *mm,
> return NULL;
> }
>
> +static inline void vma_assert_locked(struct vm_area_struct *vma)
> +{
> + mmap_assert_locked(vma->vm_mm);
> +}
> +
> static inline void release_fault_lock(struct vm_fault *vmf)
> {
> mmap_read_unlock(vmf->vma->vm_mm);
> --
> 2.43.0.687.g38aa6559b0-goog
>