Re: [PATCH] mm/rmap: synchronize lock and unlock target in anon_vma_clone

On 8/15/26 14:21, Barry Song wrote:

> On Fri, Aug 14, 2026 at 2:33 PM Eric Kim <[email protected]> wrote:
>> Currently, in anon_vma_clone(), src vma's anon_vma is assigned to
>> active_anon_vma and is used when unlocking anon_vma after linking new
>> AVCs. However, the anon_vma is locked using src->anon_vma, instead of
>> active_anon_vma, making the lock and unlock target inconsistent.
> I'm not a native speaker, but I feel the lock and unlock are actually
> consistent and operate on the same anon_vma. The code just looks
> unpaired: one uses src->anon_vma, while the other uses
> active_anon_vma. But aren't they actually the same thing?
You're right. My original commit message does come as confusing
because it sounds like there's an actual lock/unlock target mismatch.
(I'm not a native speaker too) My intent was to match the variable
name for locking and unlocking, and since active_anon_vma is the
implemented alias for src->anon_vma, anon_vma_clone() should
consistently use active_anon_vma. Should I send a v2 to clarify the commit
message?
> Not an objection to the patch; I just find the changelog a bit
> confusing.
>
>> Use active_anon_vma for both locking and unlocking.
>>
>> Signed-off-by: Eric Kim <[email protected]>
> Best Regards
> Barry

Thanks for the feedback.

Eric Kim