memcg: deuglify mm_update_next_owner()

[PATCH -mm 0/2] memcg: deuglify mm_update_next_owner()

Posted by Oleg Nesterov 1 year, 5 months ago

On top of mm-optimize-the-redundant-loop-of-mm_update_owner_next.patch

Michal, et al, could you review ? Compile tested only.

Oleg.

 kernel/exit.c | 81 +++++++++++++++++++++++++++++++----------------------------
 1 file changed, 43 insertions(+), 38 deletions(-)

Re: [PATCH -mm 0/2] memcg: deuglify mm_update_next_owner()

Posted by Andrew Morton 1 year, 5 months ago

On Wed, 26 Jun 2024 17:28:35 +0200 Oleg Nesterov <oleg@redhat.com> wrote:

> On top of mm-optimize-the-redundant-loop-of-mm_update_owner_next.patch

What should we do with
mm-optimize-the-redundant-loop-of-mm_update_owner_next.patch?  It
prevents a hard lockup splat so I'm inclined to merge it into 6.10-rcX
and cc:stable.  Further improvements (this series) can be made in the
normal fashion?

Re: [PATCH -mm 0/2] memcg: deuglify mm_update_next_owner()

Posted by Oleg Nesterov 1 year, 5 months ago

On 06/26, Andrew Morton wrote:
>
> On Wed, 26 Jun 2024 17:28:35 +0200 Oleg Nesterov <oleg@redhat.com> wrote:
>
> > On top of mm-optimize-the-redundant-loop-of-mm_update_owner_next.patch
>
> What should we do with
> mm-optimize-the-redundant-loop-of-mm_update_owner_next.patch?

Well, if you ask me (although you shoudn't ;) I think this patch makes
sense even if it can't fix all the problems.

> Further improvements (this series) can be made in the
> normal fashion?

Yes, yes. And just in case, this series mostly tries to cleanup this code
and it doesn't really depend on that patch from Jinliang.

Oleg.