Bug #12217
GNOME freezes with Intel GM965/GL960 Integrated Graphics
100%
Description
a user reported this problem :
since upgrading to 2.10, after enabling persistence gnome doesn’t start and the computer freeze. if the persistence is ot enabled, there are no issue.
the user reproduced the error after reconfiguring the persistent volume, and with a brand new installation, and with a brand new USB stick.
the user managed to get some debugging info before the freeze and I found that in the logs :
localhost.localdomain kernel: [drm:drm_atomic_helper_commit_cleanup_done [drm_kms_helper]] *ERROR* [CRTC:26:pipe A] flip_done timed outa bug report with the same error message was posted here :
Subtasks
History
#1 Updated by intrigeri 2017-02-09 13:12:06
- Category set to Hardware support
- Status changed from New to Confirmed
- Target version set to Tails_2.11
- Type of work changed from Research to Code
According to that Red Hat bug report, this is fixed in Linux 4.8.13. Tails 2.10 ships 4.8.11, but jessie-backports now has 4.8.15. I think we should upgrade to the latest 4.8 kernel from jessie-backports, via our freeze exception process, in Tails 2.11 (if we don’t go for 4.9 i.e. Feature #12122, of course, but that’s unlikely so IMO we should not block on it). anonym, what do you think?
#2 Updated by intrigeri 2017-02-09 13:12:28
- Priority changed from Normal to Elevated
(Regression on hardware that was previously supported.)
#3 Updated by intrigeri 2017-03-03 08:02:47
- Subject changed from Gnome freezes with Intel GM965/GL960 Integrated Graphics to GNOME freezes with Intel GM965/GL960 Integrated Graphics
#4 Updated by intrigeri 2017-03-03 08:35:53
I gave up on Feature #12122 for Tails 2.11 so IMO we should upgrade to the latest 4.8.x that’s ever been in jessie-backports, as I proposed above. However, jessie-backports does not carry 4.8.x anymore, so we need to import it either from our APT snapshots (if it’s still somewhere in there) or from snapshots.d.o.
anonym, can you take care of this? If you want me to handle it, please let me know ASAP (ideally I would have wanted to know weeks ago, but I can still make some room for it over the week-end).
#5 Updated by anonym 2017-03-03 11:14:08
I take it!
#6 Updated by anonym 2017-03-04 09:24:00
- Status changed from Confirmed to In Progress
Applied in changeset commit:e4ceebcfdd905b57125872ddb190cefdb24a25e3.
#7 Updated by anonym 2017-03-05 10:12:32
- Assignee changed from anonym to intrigeri
- % Done changed from 0 to 50
- QA Check set to Ready for QA
- Feature Branch set to bugfix/12217-linux-4.8.15
Sadly all Jenkins test runs so far errored out before even starting due to Bug #12259. I did build a “pre-2.11” branch that included all 2.11 branches not merged yet, including this one, and I saw no problems while using it (although I was mostly rebooting to test the issue with the “Learn more” button in our notifications).
#9 Updated by intrigeri 2017-03-05 14:11:19
- Status changed from In Progress to Fix committed
- % Done changed from 60 to 100
Applied in changeset commit:f95a8ea01d1fac27eddbf2720fbea0c472550199.
#10 Updated by intrigeri 2017-03-05 14:12:21
- Assignee deleted (
intrigeri) - QA Check changed from Ready for QA to Pass
Yeah!
#11 Updated by cypherpunks 2017-03-08 12:35:34
This bug is still present in Tails 2.11.
It is not fixed. Please reopen the bug.
Without persistence configured, the startup
and login is still slower, and freezes occur
when starting GNOME control center or Totem.
Many users with GM965/GL960 GPU affected.
More information:
https://bugzilla.kernel.org/show_bug.cgi?id=187001
https://bbs.archlinux.org/viewtopic.php?id=218581
Adding “video=SVIDEO-1:d” to the startup options
works around the bug, everything works then.
However, this will make the SVIDEO port unusable.
I think the SVIDEO port is hardly useful in the
context of Tails, so this may be a reasonable workaround
until a fixed kernel is released (none exists yet).
This work around is taken from the second link above.
#12 Updated by intrigeri 2017-03-08 13:48:49
> Adding “video=SVIDEO-1:d” to the startup options works around the bug, everything works then.
Documented this on the “known issues” page, thanks!
I’m not reopening this ticket as the problem is well tracked upstream already, and it’s definitely not something we’re going to fix ourselves.
#13 Updated by anonym 2017-03-09 13:59:05
- Status changed from Fix committed to Resolved