Re: [PATCH 4.14.y] mm/page_alloc: fix potential deadlock on zonelist_update_seq seqlock

[Petr Mladek <pmladek@suse.com>]

On Mon 2023-05-08 19:06:59, Tetsuo Handa wrote:
> On 2023/05/08 15:56, Greg KH wrote:
> > On Sun, May 07, 2023 at 11:56:29PM +0900, Tetsuo Handa wrote:
> >> commit 1007843a91909a4995ee78a538f62d8665705b66 upstream.
> > 
> > For obvious reasons, we can't just apply this to 4.14.y.  Please provide
> > fixes for all other stable trees as well so that you do not have a
> > regression when updating to a newer kernel.
> > 
> > I'll drop this from my review queue for now and wait for all of the
> > backported versions.
> 
> 5.15+ stable kernels already have the upstream patch applied.
> 
> Only 4.14/4.19/5.4/5.10 stable kernels failed to apply the upstream patch
> due to the need to append include/linux/printk.h part. I want to hear
> whether Petr Mladek is happy with this partial printk.h backport
> ( https://lkml.kernel.org/r/ZC0298t3o6+TyASH@alley ) before I spam
> everyone with the same change for 4.19/5.4/5.10.

The partial backport of the printk_deferred_enter/exit definitions
looks fine to me.

I am just not sure what is the preferred way of adding this to
stable trees.

Tetsuo merged the partial backport with the mm fix.
Alternative solution would be to add the partial backport
as a separate patch. Anyway, the commit message should mention
the original commit 85e3e7fbbb720b9897f ("printk: remove
NMI tracking").

Best Regards,
Petr