Created attachment 139245 [details] GPU crash dump With Linux-4.16.6 on a Lenovo X1 laptop (Skylake GT2 [HD Graphics 520]) I just hit the following: [ 24.801885] [drm] GPU HANG: ecode 9:1:0xfefffffe, in Xorg [600], reason: Hang on bcs0, action: reset [ 24.801886] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. [ 24.801886] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel [ 24.801886] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. [ 24.801887] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. [ 24.801887] [drm] GPU crash dump saved to /sys/class/drm/card0/error [ 24.801927] i915 0000:00:02.0: Resetting bcs0 after gpu hang [ 27.868430] asynchronous wait on fence i915:clflush:0 timed out [ 27.868450] asynchronous wait on fence i915:Xorg[600]/0:3 timed out [ 32.796660] i915 0000:00:02.0: Resetting chip after gpu hang [ 32.796739] i915 0000:00:02.0: GPU recovery failed
bcs completed with HEAD==TAIL; ELSP is loaded with the next request. Looks like it never switched and started the new request. No other clues in the register state afaics. Relevance of it being an early request?
Seems to happen right the moment the X server starts and is easily reproducible here.
Are you able to try using latest drm-tip: https://cgit.freedesktop.org/drm-tip and send dmesg with drm.debug=0x1e log_buf_len=4M? Does it help Chris?
Mika, Chris, any help here?
Udo, have you been able to test latest drm-tip?
I just tried drm-tip commit 8e1dab6e and with that kernel the hang no longer occurs. With 4.16.8 it still reliably hangs.
OK, thanks. Chris, Mika, Jani, an idea what has changed?
Based on comment resolving as fixed in drm-tip.
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.