Re: [PATCH v4 5/5] cgroup: use subsystem-specific rstat locks to avoid contention

[JP Kobryn <inwardvessel@gmail.com>]

On 4/16/25 2:50 AM, Michal Koutný wrote:
> On Tue, Apr 15, 2025 at 09:30:35AM -1000, Tejun Heo <tj@kernel.org> wrote:
>> I don't know whether this is a controversial opinion but I personally would
>> prefer to not have __acquires/__releases() notations at all.
> 
> Thanks for pointing it out.
> 
>> They add unverifiable clutter
> 
> IIUC, `make C=1` should verify them but they're admittedly in more sorry
> state than selftests.
> 
>> and don't really add anything valuable on top of lockdep.
> 
> I also find lockdep macros better at documenting (cf `__releases(lock)
> __acquires(lock)` annotations) and more functional (part of running
> other tests with lockdep).
> 
> So I'd say cleanup with review of lockdep annotations is not a bad
> (separate) idea.

Maybe we can keep these intact for now while updating the text within
to something more specific like Michal is suggesting?

Tejun, let us know if you have a strong feeling on removing them at
this point.

> 
> Michal