Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.kernel > #89298 > unrolled thread

Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5

Back to article view | Back to linux.debian.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5 Uwe Kleine-König <u.kleine-koenig@baylibre.com> - 2025-09-17 17:20 +0200
    Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5 Christoph Anton Mitterer <calestyo@scientia.org> - 2025-09-17 18:00 +0200
      Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5 Uwe Kleine-König <u.kleine-koenig@baylibre.com> - 2025-09-18 00:00 +0200
        Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5 Christoph Anton Mitterer <calestyo@scientia.org> - 2025-09-18 00:10 +0200
        Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5 Christoph Anton Mitterer <calestyo@scientia.org> - 2025-09-19 22:10 +0200

#89298 — Bug#1115002: linux: "serious breakage" in hibernate since v6.16-rc5

[Multipart message — attachments visible in raw view] — view raw

Control: tag -1 + fixed-upstream

Hello Christoph,

On Fri, Sep 12, 2025 at 08:09:43PM +0200, Christoph Anton Mitterer wrote:
> I've just read on Phoronix about some "serious breakage" bug:
> https://www.phoronix.com/news/Linux-6.17-PM-Hibernation-FIxes
> 
> A fix is available:
> https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/commit/?h=fixes&id=449c9c02537a146ac97ef962327a221e21c9cab3
> 
> TBH, the commit message is a bit unclear to me,... the
> bug was apparently inroduced in 6.16 rc5, but only with
> some other commit from 6.17 rc1 it became "visible".
> 
> Whatever "visible" means here... could be that only then
> it started to happen more often or at all.
> 
> 
> >From the original bug report it seems the original hibernate
> image would get somehow currupted, which *could* mean that basically
> anything might happen... from a crash to complete data corruption.
> 
> Since we already have 6.16, I figured it might be worth to give a heads
> up and peraps the fix should be cherry picked ASAP.

I claim that the upstream stable process works fine here. Said patch is
already part of 6.16.8-rc1
(https://lore.kernel.org/all/20250917123351.839989757@linuxfoundation.org/,
patch #65). My gut feeling here is that it's ok to wait until this patch
flows into the Debian kernel via 6.16.8 without the additional effort to
cherry pick it before (with all up- and downsides involved).

Best regards
Uwe

[toc] | [next] | [standalone]

#89299

Hey Uwe.


On Wed, 2025-09-17 at 17:12 +0200, Uwe Kleine-König wrote:
> My gut feeling here is that it's ok to wait until this patch
> flows into the Debian kernel via 6.16.8 without the additional effort
> to
> cherry pick it before (with all up- and downsides involved).

Well, as I've said it isn't really clear to me how serious this even is
(in 6.16) or not. The commit message merely sounded pretty scary with
"serious breakage" and even more so the "This is super-urgent, so I'm
going to fast-track it."[0].

So I thought I heads up might not harm, in order to prevent that maybe
some peoples filesystem superblocks get corrupted by chance, if some
corrupted memory image is loaded back on resume.

Let's just hope the do .8 fast and it ends up just as fast in sid. :-)
Meanwhile I've simply stopped using hibernation, but this won't of
course help anyone else.

Best,
Chris.


[0] https://lore.kernel.org/lkml/5924662.DvuYhMxLoT@rafael.j.wysocki/

[toc] | [prev] | [next] | [standalone]

#89304

[Multipart message — attachments visible in raw view] — view raw

Hello Christoph,

On Wed, Sep 17, 2025 at 05:50:52PM +0200, Christoph Anton Mitterer wrote:
> On Wed, 2025-09-17 at 17:12 +0200, Uwe Kleine-König wrote:
> > My gut feeling here is that it's ok to wait until this patch
> > flows into the Debian kernel via 6.16.8 without the additional effort
> > to
> > cherry pick it before (with all up- and downsides involved).
> 
> Well, as I've said it isn't really clear to me how serious this even is
> (in 6.16) or not. The commit message merely sounded pretty scary with
> "serious breakage" and even more so the "This is super-urgent, so I'm
> going to fast-track it."[0].

My interpretation is that 2640e819474f (i.e. a commit not present in the
Debian kernels yet) is needed to make this problem considerably more
likely to occur. So I'd expect that yes 6.16 has a problem, but it
doesn't trigger often.
 
> So I thought I heads up might not harm, in order to prevent that maybe
> some peoples filesystem superblocks get corrupted by chance, if some
> corrupted memory image is loaded back on resume.
> 
> Let's just hope the do .8 fast and it ends up just as fast in sid. :-)
> Meanwhile I've simply stopped using hibernation, but this won't of
> course help anyone else.

Applying a commit fast to the Debian kernel also has the downside to
skip over a maturation process. While I would consider Rafael's
judgement a strong hint and the patch simple enough to consider the
change safe enough, there is a good reason the Greg Kroah-Hartman sends
out the patches he queued for the upcoming stable release to the mailing
list. There is an armada of people and bots that review and do build and
runtime tests that is very valuable.

Best regards
Uwe

[toc] | [prev] | [next] | [standalone]

#89305

On Wed, 2025-09-17 at 23:50 +0200, Uwe Kleine-König wrote:
> My interpretation is that 2640e819474f (i.e. a commit not present in
> the
> Debian kernels yet) is needed to make this problem considerably more
> likely to occur. So I'd expect that yes 6.16 has a problem, but it
> doesn't trigger often.

Well I can't judge that. I've been running since you guys uploaded 6.16
to sid wiht that... and at least haven't noticed any breakage.

But if someone's writing his thesis and picks that 1 in 10000 chance
where it even hits there, he might not be pleased. ;-)


> Applying a commit fast to the Debian kernel also has the downside to
> skip over a maturation process. While I would consider Rafael's
> judgement a strong hint and the patch simple enough to consider the
> change safe enough, there is a good reason the Greg Kroah-Hartman
> sends
> out the patches he queued for the upcoming stable release to the
> mailing
> list. There is an armada of people and bots that review and do build
> and
> runtime tests that is very valuable.

Sure :-)
Good enough for me. As said I merely meant this as a heads up so that
the kernel team can judge whether it's worth the risks to cherry pick
or not.

Thanks,
Chris.

[toc] | [prev] | [next] | [standalone]

#89329

Hey.

Guess you've seen it already... but .8 is out.
Would be awesome if that could be fast-tracked. :-)

Cheers,
Chris

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.kernel

csiph-web