Re: [PATCH v2] mm/page_alloc: fix potential deadlock on zonelist_update_seq seqlock

[Petr Mladek <pmladek@suse.com>]

On Wed 2023-04-05 10:28:07, Michal Hocko wrote:
> On Tue 04-04-23 14:25:28, Andrew Morton wrote:
> > On Tue, 4 Apr 2023 23:31:58 +0900 Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> wrote:
> > 
> > > syzbot is reporting circular locking dependency which involves
> > > zonelist_update_seq seqlock [1], for this lock is checked by memory
> > > allocation requests which do not need to be retried.
> > 
> > I queued this, along with a note that an updated changelog is likely.
> > 
> > Do we feel that a -stable backport is warranted?  I think so, from your
> > earlier comments.  Please add the cc:stable to the changelog in this
> > situation.
> 
> Memory hotplug is pretty rare event so the deadlock is quite unlikely.
> On the other hand the fix is pretty easy so it shouldn't hurt to have it
> in stable kernels.

Note that printk_deferred_enter()/exit() has been added in v5.15-rc1
by the commit 85e3e7fbbb720b9897 ("printk: remove NMI tracking").

The commit has non-trivial dependencies. Any backport for older stable
kernel would need a custom patch just adding these two wrappers.

Best Regards,
Petr