Bug #11096

"Oh no!" / Xorg crash after logging in at the Greeter with Intel 855GM graphics

Added by goupille 2016-02-09 14:49:25 . Updated 2016-08-14 11:04:10 .

Status:
Resolved
Priority:
Normal
Assignee:
Category:
Hardware support
Target version:
Start date:
2016-02-09
Due date:
% Done:

100%

Feature Branch:
emmapeel:doc-11096
Type of work:
End-user documentation
Blueprint:

Starter:
1
Affected tool:
Deliverable for:

Description

a user reported that error, I think it is related to https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=776911 and
https://labs.riseup.net/code/issues/8778.

the user could send us the result of ‘sudo tails-debugging-info’, I found that in the logs :

Feb 08 17:32:19 localhost.localdomain kernel: [drm:i9xx_check_fifo_underruns] *ERROR* pipe B underrun
Feb 08 17:45:35 amnesia kernel: [drm:i9xx_set_fifo_underrun_reporting] *ERROR* pipe A underrun
Feb 08 17:49:25 amnesia kernel: [drm:i8xx_irq_handler] *ERROR* pipe B underrun

and

Feb 08 17:45:21 amnesia gnome-session[3791]: (gnome-shell:4839): Cogl-ERROR **: Failed to create texture 2d due to size/format constraints
Feb 08 17:45:21 amnesia x-session-manager[3791]: WARNING: Application 'gnome-shell-classic.desktop' killed by signal 5
Feb 08 17:45:21 amnesia x-session-manager[3791]: WARNING: App 'gnome-shell-classic.desktop' respawning too quickly
Feb 08 17:45:21 amnesia x-session-manager[3791]: Unrecoverable failure in required component gnome-shell-classic.desktop
Feb 08 17:45:21 amnesia gnome-session[3791]: x-session-manager[3791]: WARNING: Application 'gnome-shell-classic.desktop' killed by signal 5
Feb 08 17:45:21 amnesia gnome-session[3791]: x-session-manager[3791]: WARNING: App 'gnome-shell-classic.desktop' respawning too quickly
Feb 08 17:45:21 amnesia gnome-session[3791]: (gnome-settings-daemon:3881): GLib-GIO-CRITICAL **: g_dbus_proxy_call_internal: assertion 'G_IS_DBUS_PROXY (proxy)' failed

Subtasks


History

#1 Updated by goupille 2016-02-09 15:11:35

#2 Updated by goupille 2016-02-09 15:11:47

  • related to Bug #8778: "Oh no!" / Xorg crash after logging in at the Greeter on Jessie added

#3 Updated by intrigeri 2016-02-10 11:28:56

  • Subject changed from "Oh no!" / Xorg crash after logging in at the Greeter on Tails 2.0 to "Oh no!" / Xorg crash after logging in at the Greeter with Intel 855GM graphics
  • Category set to Hardware support
  • Status changed from New to Confirmed
  • Target version set to Tails_2.3

Thanks for the logs. Please make sure you forward them to me without wrapping text, next time.

I think you identified the right error messages from the logs. I’ll have another good look at the corresponding Debian bug reports, in the hope that there’s a trick that could fix such problems (e.g. rebuilding mesa with llvm 3.4). There is a slight chance that upgrading the kernel would fix the problem, but I’m not holding my breath. If we’re unlucky this will be another instance of “upgrading to a new Debian improves things on some hardware, but introduces regressions on some other hardware”, which is not something we can do much about in general.

FTR the affected hardware is “00:02.1 Display controller: Intel Corporation 82852/855GM Integrated Graphics Device (rev 02)”.

#4 Updated by elouann 2016-02-19 12:19:41

  • related to Bug #11014: Xorg/EQ overflow/Intel GPU lockup added

#5 Updated by intrigeri 2016-04-14 20:54:38

  • Target version changed from Tails_2.3 to Tails_2.4

#6 Updated by dave_cass 2016-04-16 08:45:17

I spotted that a previous reporter now has difficulty finding test machines for this bug.

I have a Dell Inspiron 510m still in regular use mainly because of its hardware serial I/O port. This one has the relevant GPU and I am also aware of a number of machines recently liberated to Linux from Win XP thanks to the MS end of life policy.

This bug is a Tails show stopper for any machine with the Intel 855GM graphics so as there do seem to be a fair few of them about. I’m happy to do guided testing for as long as I can get access to one and I’ll report back as directed if that helps.

While I appreciate it may not be easy to influence events and changes upstream, it does seem a shame to lose a whole raft of new Linux users because of a regression like this when the intel driver has been getting better and better.

Regards - David C

#7 Updated by intrigeri 2016-04-29 13:37:33

  • related to deleted (Bug #11014: Xorg/EQ overflow/Intel GPU lockup)

#8 Updated by intrigeri 2016-04-29 13:38:04

  • related to deleted (Bug #8778: "Oh no!" / Xorg crash after logging in at the Greeter on Jessie)

#9 Updated by intrigeri 2016-04-29 13:38:39

(Dropped “Related” relationships since these other tickets were about totally different hardware.)

#10 Updated by intrigeri 2016-04-29 13:43:16

dave_cass: great! Can you please test the experimental (no guarantees, blah) ISO image at http://nightly.tails.boum.org/build_Tails_ISO_feature-11303-mesa-backport/lastSuccessful/archive/latest.iso on a machine with Intel 855GM graphics?

I think that’s our last chance at fixing this: given the hardware was released 13 years ago, it’s unlikely that anyone upstream (Mesa, X.Org Intel drivers, whatever) works on fixing this.

#11 Updated by dave_cass 2016-04-29 14:57:24

I’ll grab that and try it first thing tomorrowand let you know how it goes. DC

#12 Updated by dave_cass 2016-04-30 01:28:44

Just tried that and it gets through the greeter and works with language selections and further options etc. but fails with “Ahh No….”at the same point as Tails 2.0 when finally launching the desktop.

Errors on tty01 are just the same as seen previously and quoted in the first Feb 08 quote above, with pipe B underrun followed by pipe A underrun then another pipe B underrun.

If you feel it is worth following further I can run diagnostics and supply logs. Just let me know.
DC

#13 Updated by intrigeri 2016-04-30 05:45:05

  • Assignee changed from intrigeri to emmapeel
  • Type of work changed from Research to End-user documentation

OK, thanks for testing. Too bad :/

There’s not much more that we can do, so I’m reassigning to current frontdesk so that this is added to the list of known issues.

#14 Updated by dave_cass 2016-04-30 23:02:16

OK. I can see the problem. I’ll continue to try if asked. I do know that over the years of free graphics driver development Those machines often managed with driver problems if the rendering method was changed or 3D acceleration turned off completely. That may be a road to a workaround.

DC

#15 Updated by anonym 2016-06-08 01:35:03

  • Target version changed from Tails_2.4 to Tails_2.5

#16 Updated by dave_cass 2016-06-08 05:06:04

Just tried a live CD Ubuntu 16.04 on a laptop with 855GM graphics and that’s working now. It has the same “[drm:i9xx_check_fifo_underruns]” errors as above during the initial startup giving “ERROR pipe B underrun” then “ERROR pipe A underrun”. However it comes out the other side with a successful boot and finally reports:
Gallium 0.4 on llvmpipe (LLVM3.8, 128bits) in the driver info.
I don’t have enough knowledge of the workings of the graphics stack to analyse that much further without some assistance but it does look as though Ubuntu may have a work around for the Debian regression of the intel driver which has hit TAILS users on earlier machines.
I’m still happy to do testing on these machines if required. D.C.

#17 Updated by intrigeri 2016-07-30 08:25:40

emmapeel: ping! (Bug #11096#note-13)

#18 Updated by emmapeel 2016-08-01 03:15:19

  • Status changed from Confirmed to In Progress
  • Assignee deleted (emmapeel)
  • % Done changed from 0 to 40
  • QA Check set to Ready for QA
  • Feature Branch set to emmapeel:doc-11096
  • Starter set to Yes

please review my branch at

https://git-tails.immerda.ch/emmapeel/tails/log/?h=doc-11096

I decided to list it under Dell hardware as one user reported the model, but maybe I should have added a section about graphic cards, as it is likely there are going to be more laptops affected, and maybe listing the laptop models on it.

What does the reviewers think?

#19 Updated by intrigeri 2016-08-01 03:52:56

  • Assignee set to emmapeel
  • Target version changed from Tails_2.5 to Tails_2.6
  • QA Check changed from Ready for QA to Dev Needed

> I decided to list it under Dell hardware as one user reported the model, but maybe I should have added a section about graphic cards, as it is likely there are going to be more laptops affected, and maybe listing the laptop models on it.

Looks like all systems with Intel 855GM are affected, so indeed I think it should not be listed as specific to some specific Dell laptop.

#20 Updated by emmapeel 2016-08-01 09:53:57

  • Assignee deleted (emmapeel)
  • QA Check changed from Dev Needed to Ready for QA

Ok new version on the ‘graphics’ section. please review.

#21 Updated by intrigeri 2016-08-01 13:12:31

  • Assignee set to intrigeri

#22 Updated by intrigeri 2016-08-11 10:51:33

  • Assignee changed from intrigeri to emmapeel
  • % Done changed from 40 to 50
  • QA Check changed from Ready for QA to Dev Needed

commit:331703b62b816fadb0f3982ab309fd60e71c4e26 removes the “Problems starting Tails” header. Is it on purpose?

#23 Updated by emmapeel 2016-08-13 02:32:03

  • Assignee changed from emmapeel to intrigeri
  • QA Check changed from Dev Needed to Ready for QA

intrigeri wrote:
> commit:331703b62b816fadb0f3982ab309fd60e71c4e26 removes the “Problems starting Tails” header. Is it on purpose?

Ooops! No, it was a mistake. Not sure how this happened. Maybe this merge stuff

https://twitter.com/basarat/status/763261426752303108

I tried to search for the commit where the missing line is deleted but I don’t see it on the commits.

I have added it in 4083ac3..b1c42cc please review

#24 Updated by intrigeri 2016-08-14 11:04:11

  • Status changed from In Progress to Resolved
  • Assignee deleted (intrigeri)
  • % Done changed from 50 to 100
  • QA Check changed from Ready for QA to Pass

Merged, thanks!