Created attachment 115484 [details] dmesg Trying to log into the system after boot-up (using slim), the system hangs for a one or two seconds. From dmesg: [ 26.704033] [drm] stuck on render ring [ 26.708701] [drm] GPU HANG: ecode 2:0:0x40907fc1, in X [1310], reason: Ring hung, action: reset [ 26.708705] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 26.708707] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 26.708708] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 26.708710] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 26.708712] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 26.708809] drm/i915: Resetting chip after gpu hang [ 26.708823] [drm:i915_reset] *ERROR* Failed to reset chip: -19 This happens with kernel versions 4.0.0 and 4.0.1. With 3.19.x kernels, the hang does not occur. X.Org 1.17.1 DDX 2.99.197 and git (e7016d30f3a0ae817c77ccbd962f776ac3e7e100)
Created attachment 115485 [details] /sys/class/drm/card0/error
commit b400dd22c28a08d7644b4ede076be9d8c2b8ca9d Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Sun May 3 11:58:25 2015 +0100 sna: Clear has-pinned-batches if we can no longer actually pin Insert rant about useful kernel interfaces being removed without justification. References: https://bugs.freedesktop.org/show_bug.cgi?id=88411 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The error is still there: [ 42.704035] [drm] stuck on render ring [ 42.708714] [drm] GPU HANG: ecode 2:0:0x067fffc1, in X [1296], reason: Ring hung, action: reset [ 42.708719] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 42.708720] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 42.708722] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 42.708724] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 42.708726] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 42.708804] drm/i915: Resetting chip after gpu hang [ 42.708817] [drm:i915_reset] *ERROR* Failed to reset chip: -19 And now there are rendering issues after log in. The desktop background is not drawn, the cursor is invisible until an application is opened (I guess, it becomes visible once it changes its form), and some other issues.
Created attachment 115524 [details] New /sys/class/drm/card0/error GPU dump after applying the patch. I can't tell if something important changed.
Something overwrote at least the first 32MiB of the GTT with ~0.
Created attachment 116186 [details] /sys/class/drm/card0/error (In reply to Chris Wilson from comment #5) > Something overwrote at least the first 32MiB of the GTT with ~0. Now I can say that this overwriting happens, but very rarely. Most of the time, this does not happen. The GPU hang however is still present every time. Here is a current /sys/class/drm/card0/error (Linux 4.0.4, xf-86-video-intel git-fb1643f0f904eb258da71cd0b8deb8d3ec6dafed).
Created attachment 116187 [details] dmesg And the corresponding dmesg output
The GPU still hangs with kernel version 4.1.
The GPU still hangs right after logging in with Linux 4.3.0 and the current git version of xf86-video-intel (0340718366d7cb168a46930eb7be22f2d88354d8). Do you need any additional information? If so, let me know. Or is the hardware so old that no one cares anymore?
Created attachment 119684 [details] /sys/class/drm/card0/error [20151115] Here is a fresh GPU crash dump (linux: 4.3.0, xf86-video-intel: git-0340718366d7cb168a46930eb7be22f2d88354d8).
*** Bug 93056 has been marked as a duplicate of this bug. ***
We seem to have neglected the bug a bit, apologies. tka, since There were improvements pushed in kernel & xf86-video-intel that will benefit to your system, so please re-test with latest kernel & xf86-video-intel -alternatively you may use modesetting/glamor with latest mesa version- and mark as REOPENED if you can reproduce (and attach fresh gpu error dump & kernel log) and RESOLVED/* if you cannot reproduce.
(In reply to yann from comment #12) > We seem to have neglected the bug a bit, apologies. > > tka, since There were improvements pushed in kernel & xf86-video-intel that > will benefit to your system, so please re-test with latest kernel & > xf86-video-intel -alternatively you may use modesetting/glamor with latest > mesa version- and mark as REOPENED if you can reproduce (and attach fresh > gpu error dump & kernel log) and RESOLVED/* if you cannot reproduce. Timeout. Assuming that this is not occurring anymore. If this issue happens again, re-test with latest kernel & xf86-video-intel -alternatively you may use modesetting/glamor with latest mesa version- and REOPEN if you can reproduce (and attach fresh gpu error dump & kernel log)
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.