Re: [PATCH mm-unstable v1 1/4] mm: don't hold css->refcnt during traversal

[Johannes Weiner <hannes@cmpxchg.org>]

On Thu, Aug 01, 2024 at 03:32:53PM -0700, Kinsey Ho wrote:
> Sorry, I replied to this email earlier but it had some issues with plain
> text. Please ignore the first reply of mine (the one with HTML). I'm resending
> the email below.
> 
> Thank you Johannes, Roman, and Yosry for reviewing this patch!
> 
> On Thu, Jul 25, 2024 at 3:34 PM Yosry Ahmed <yosryahmed@google.com> wrote:
> > On Thu, Jul 25, 2024 at 1:43 PM Johannes Weiner <hannes@cmpxchg.org> wrote:
> > > What does this buy us? The tryget is cheap.
> >
> > mem_cgroup_iter() is not an easy function to follow, so I personally
> > appreciate the simplicity gains tbh.
> 
> Yes, the main intention here was to simplify the code's readability.
>
> > This reads to me like it is intentional that RCU protection is enough
> > for @pos and @root, and that the sibling linkage is RCU protected by
> > design. Perhaps we could clarify this further (whether at
> > css_next_descendant_pre(), or above the definition of the linkage
> > members).
> 
> Do we want to move forward with Yosry's suggestion to clarify that the
> sibling linkage is RCU-protected by design? Perhaps this clarification
> can be made in the definition of the linkage members so that the
> safety of the css in this function is more clear to users. If this is
> sufficient, I will make the change in a v2 patchset.

Yes, that sounds like a good way forward to me.