Re: [PATCH 0/5 cgroup/for-6.16-fixes] harden css_create() for safe placement of call to css_rstat_init()

[JP Kobryn <inwardvessel@gmail.com>]

Thanks for taking a look Michal.

On 7/25/25 10:23 AM, Michal Koutný wrote:
> Hi.
>
> On Mon, Jul 21, 2025 at 06:40:25PM -0700, JP Kobryn <inwardvessel@gmail.com> wrote:
> ...
>
> Thanks for the instructive summary!
>
>> If any of the three goto jumps are taken, async cleanup will begin and
>> css_rstat_exit() will be invoked. But since css_rstat_init() would not have
>> succeeded, the warning will eventually be reached.
> First thought is why not simply add a flag that'd guide whether
> css_rstat_exit() has work to do.
I did consider adding an "initialized" flag to the css but since there 
can be multiple css's per
cgroup it felt like it would be adding overhead. So I went the path of 
getting the call
sequence right. I'm open to feedback on this, though.
>
> This is meant as a fix, so it should have some metadata, I'd consider this one:
> Fixes: 5da3bfa029d68 ("cgroup: use separate rstat trees for each subsystem")
>
> (that's when css_rstat_init was moved to css_create)
>
> and likely this
> Reported-by: syzbot+8d052e8b99e40bc625ed@syzkaller.appspotmail.com
>
> (Sorry for being such a bureaucrat.)
No problem, I overlooked that.
> It's most appropriate in your 4/5 but do you think it'd be possible to
> reshuffle the series to put the fix in front (to ease it for stable
> kernels) and refactorings after?
Let me give that a try. As it is right now, patches 1-3 are pre-reqs for 
4. I can try to get the
actual fix to the front and then add patches to additionally make 
nicer/refactor.