Dear Maintainer,

This is on the same machine as #1025659, and the backtrace is similar
(see that bug for the configuration of the laptop).

After updating mesa from 22.2.4-1 to latest testing (22.3.2-1)
earlier today U.S. time as part of a routine upgrade to up-to-date
testing, the same situation happened as #1025659. Even the backtrace is
similar:
  Xorg (OsLookupColor)
  glibc (__sigaction)
  iris_dri.so (nouveau_drm_screen_create+[three addresses])
  iris_dri.so (?+0x0)
  iris_dri.so (__driDriverGetExtensions_d3d12+[three addresses])

We've installed 22.3.1-1 off snapshot.d.o,
and got the same backtrace again.

Both traces attached as cleaned-up photos.

Seeing as, last time, the update in question was 22.2.0-1 to 22.3.0-2,
it would appear that 22.3.x is just fundamentally broken.

Most testing users probably can't do a s.d.o downgrade
(well, I'm writing this on the behalf of the user who couldn't, so.).

Best,
наб

Would you mind explaining what an 's.d.o downgrade' entails? I used the snapshot archive and pinned the offender, then used aptitude and skipped through solutions until I found an acceptable one. (apt wants to kill all of X!). That is a pretty convoluted and unreliable process. 'dontbreakdebian' also says to never mess around like that.
There is no official SOP as far as I'm aware.

The process on Fedora given here https://gitlab.freedesktop.org/mesa/mesa/-/issues/8007 seems a lot more straightforward.

I understand this is 'testing' - but Debian testing > Arch regular and many people even consider sid to be usable everyday (see 'siduction').

Can we please revert this package instantly while we wait for an upstream fix? Testing is so darn stable usually, many 'normal' people use it in prod. This is a very severe bug in my book, right next to that kernel one a few months ago killing screens. Unless you're rather deep into Debian, you're never going to find the workaround, and likely just move on.

Can you check with 22.3.3-1 (now in testing) or 23.0.0~rc1-1 (in
experimental) if the issue is fixed or still not?

22.3.3-1 works here!

But the upstream bug is unanswered still: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8007.
And there's another guy who did the same to no avail. It's not fixed for everyone. If we want to fix it for all, why not just revert it globally now, until the cause has been found? Who knows how long that'll take.

The laptop I can personally test on is rather average. (intel UHD630, some unused nv card, nouveau is blacklisted for energy savings, and the busses removed via udev) If someone has an idea how to debug this that does not eat all day, I'd help possibly diff to find out why it works here, and not for him.


Side note: I received 2 mails, one triggered directly by you, another from being subscribed to the bug. The latter fails DKIM. If this is unintended, we might want to tell the postmaster.

January 18, 2023 10:25 AM, "Fabio Pedretti" <pedretti.fabio@gmail.com> wrote:

As I answered on gitlab I tried 22.3.3-1 with the same backtrace & error
logs.
I rolled back to 22.2.4-1.

Happy to help if I can.

Le mer. 18 janv. 2023 à 15:28, <
deb-bug.oluj0xj403ou050siy9dpk2z@qmail.sunbirdgrove.com> a écrit :

Can you also check with 23.0.0~rc1-1 in experimental?

Il giorno mer 18 gen 2023 alle ore 16:48 Cédric Nisio
<cedric.nisio@malt.com> ha scritto:

Same result with 23.0.0~rc1-1

Le ven. 20 janv. 2023 à 15:18, Fabio Pedretti <pedretti.fabio@gmail.com> a
écrit :

Same error with 23.0.0~rc1-1; log below.

Best,
наб

Since this also happens of Fedora, I think it's best to add every
additional info on the upstream report
https://gitlab.freedesktop.org/mesa/mesa/-/issues/8007 , also click on
the "thumbs up".

Il giorno sab 28 gen 2023 alle ore 13:38 наб
<nabijaczleweli@nabijaczleweli.xyz> ha scritto: