| Summary: | GPU HANG: ecode 9:0:0x85dffffb, in Borderlands2 [3497], reason: Engine(s) hung, action: reset | ||
|---|---|---|---|
| Product: | Mesa | Reporter: | marco.grimaldi |
| Component: | Drivers/DRI/i965 | Assignee: | Intel 3D Bugs Mailing List <intel-3d-bugs> |
| Status: | RESOLVED MOVED | QA Contact: | Intel 3D Bugs Mailing List <intel-3d-bugs> |
| Severity: | blocker | ||
| Priority: | medium | CC: | david, intel-gfx-bugs |
| Version: | unspecified | ||
| Hardware: | x86-64 (AMD64) | ||
| OS: | Linux (All) | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: |
cat /sys/class/drm/card0/error | gzip > error.gz
cat /sys/class/drm/card0/error | gzip > error_BL2_update01.gz cat /sys/class/drm/card0/error | gzip > error_BL2_update02.gz Bisect between 12.0.3 and master to find the patch which fixed the hangs |
||
|
Description
marco.grimaldi
2016-09-09 13:41:35 UTC
Created attachment 126374 [details]
cat /sys/class/drm/card0/error | gzip > error.gz
i5 6600K
i530
What version of mesa do you have? (In reply to Chris Wilson from comment #2) > What version of mesa do you have? Hi Chris, the default mesa from Manjaro: mesa 12.0.1-7 Hi there, today manjaro updated a few component: Kernel 4.7.3 Xorg 1.18.4 Mesa 12.0.2 Tested Borderlands2 again, and the system crashes in the exct same point: 10 mins into the game, after the initial intro, folowing the bot up to the 'house' Before been able to get into it, the game freazes and then crashes All the times. attached the updated card crash dmesg is as before: [ 527.325544] [drm] stuck on render ring [ 527.328566] [drm] GPU HANG: ecode 9:0:0x85dffffb, in Borderlands2 [1189], reason: Engine(s) hung, action: reset [ 527.328570] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 527.328573] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 527.328576] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 527.328578] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 527.328581] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 527.331700] drm/i915: Resetting chip after gpu hang [ 529.325647] [drm] RC6 on [ 539.359244] [drm] stuck on render ring [ 539.362504] [drm] GPU HANG: ecode 9:0:0x85dffffb, in Borderlands2 [1189], reason: Engine(s) hung, action: reset [ 539.364389] drm/i915: Resetting chip after gpu hang [ 541.359305] [drm] RC6 on Created attachment 126478 [details]
cat /sys/class/drm/card0/error | gzip > error_BL2_update01.gz
Hi, after updating to manjaro testing, I still get into the same error/crash uname -a Linux moby 4.7.4-1-MANJARO #1 SMP PREEMPT Thu Sep 15 20:11:38 UTC 2016 x86_64 GNU/Linux Mesa 12.0.3 X.Org X Server 1.18.4 Attached the updated crash dump Cheers, M. Created attachment 126632 [details]
cat /sys/class/drm/card0/error | gzip > error_BL2_update02.gz
Hi, I forgot to add the dmesg output in the last update: [ 572.424996] [drm] stuck on render ring [ 572.428208] [drm] GPU HANG: ecode 9:0:0x85dffffb, in Borderlands2 [1288], reason: Engine(s) hung, action: reset [ 572.428213] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 572.428216] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 572.428218] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 572.428220] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 572.428223] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 572.431747] drm/i915: Resetting chip after gpu hang [ 574.425058] [drm] RC6 on [ 582.428673] [drm] stuck on render ring [ 582.431709] [drm] GPU HANG: ecode 9:0:0x85dffffb, in Borderlands2 [1288], reason: Engine(s) hung, action: reset [ 582.434741] drm/i915: Resetting chip after gpu hang [ 584.405443] [drm] RC6 on M. I've been having the exact same issue at the same point in the game. Running a Dell XPS 13 9350 Ubuntu 16.04 w/ kernels up to the latest mainline 4.8RC6. All have same issue. dmesg and gpu crash info below: DMESG: [ 761.526619] [drm] GPU HANG: ecode 9:0:0x85dffffb, in Borderlands2 [2831], reason: Hang on render ring, action: reset [ 761.526622] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 761.526623] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 761.526624] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 761.526625] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 761.526627] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 761.534566] drm/i915: Resetting chip after gpu hang [ 761.534955] [drm] GuC firmware load skipped [ 763.499353] [drm] RC6 on [ 777.523471] drm/i915: Resetting chip after gpu hang [ 777.523951] [drm] GuC firmware load skipped [ 779.530766] [drm] RC6 on [ 787.501268] drm/i915: Resetting chip after gpu hang [ 787.501702] [drm] GuC firmware load skipped [ 787.550160] show_signal_msg: 36 callbacks suppressed [ 787.550165] Borderlands2[2831]: segfault at 72a5c ip 00000 I tried Borderlands 2 on my Skylake GT4 today, and I was able to reproduce your hang with Mesa 12.0.3. Exact same spot - as soon as you and Claptrap reach the door, it tanks. Error state looks identical - a PIPE_CONTROL just before 3DSTATE_DEPTH_BUFFER, on the normal drawing path (notably not BLORP). I was unable to reproduce the hang with Mesa master, however. So, the good news is that this may be fixed already. Unfortunately I'm not sure what fixed it. I suppose I'll have to bisect... Created attachment 126774 [details] Bisect between 12.0.3 and master to find the patch which fixed the hangs Annoyingly, the fix is apparently: commit 1eef0b73aa323d94d5a080cd1efa81ccacdbd0d2 Author: Kenneth Graunke <kenneth@whitecape.org> Date: Tue Jul 12 03:57:25 2016 -0700 i965: Rewrite FS input handling to use the new NIR intrinsics. which wasn't supposed to fix anything, but changes a lot of fragment shaders. I suppose that tells us that it's a fragment shader that is hanging the GPU, which is at least something to go on... I tried to take an apitrace to reproduce the hang, so I could narrow down which draw call is going awry, but...that doesn't hang. :( *** Bug 93528 has been marked as a duplicate of this bug. *** In the bug I marked as a duplicate, the reporter said "mesa 10.5.7 works, while 10.6.0 causes the game to crash while loading." So, I think this is more evidence that whatever the problem is -- it's something not directly related to the bisected commit. -- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/1537. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.