Created attachment 114950 [details] /sys/class/drm/card0/error System hang for a sec, GPU reset, screen is all garbled now. HW is an ACER Aspire One 00:02.0 0300: 8086:27ae (rev 03) (prog-if 00 [VGA controller]) Subsystem: 1025:015b Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx- Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx- Latency: 0 Interrupt: pin A routed to IRQ 16 Region 0: Memory at 58480000 (32-bit, non-prefetchable) [size=512K] Region 1: I/O ports at 60c0 [size=8] Region 2: Memory at 40000000 (32-bit, prefetchable) [size=256M] Region 3: Memory at 58500000 (32-bit, non-prefetchable) [size=256K] Expansion ROM at <unassigned> [disabled] Capabilities: <access denied> Kernel driver in use: i915 Kernel modules: i915 Kernel is 4.0.0-0.rc5 on Fedora 22.
Created attachment 114951 [details] dmesg
Running xfce if that's relevant, so not a GL based desktop.
Important question here is which ddx? Could you try xf86-video-intel.git? There has been one issue with using mmap(wc) leaving vertex buffers, even batch buffers, as fenced which has been since fixed. I don't think that's happened here, everything looks fairly consistent, but also everything looks benign. However it hung...
(In reply to Chris Wilson from comment #3) > Important question here is which ddx? Could you try xf86-video-intel.git? > There has been one issue with using mmap(wc) leaving vertex buffers, even > batch buffers, as fenced which has been since fixed. I don't think that's > happened here, everything looks fairly consistent, but also everything looks > benign. However it hung... ddx is the one that comes with F22, ie a git snapshot of xf86-video-intel from Feb, 11 2015 but I am not able to identify the exact sha1. I shall try a more recent git snapshot, but this is bug seems very random so I am still unsure I can reproduce.
The commit I am thinking of is: commit cb2269d1308f2a410508aafecc8b0ee51959afcf Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Mon Mar 2 09:08:54 2015 +0000 sna/gen3: Disable fences associated with vertex buffers Could gen3 actually be trying to read these through a fence? References: https://bugs.freedesktop.org/show_bug.cgi?id=89334 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
(In reply to Chris Wilson from comment #5) > The commit I am thinking of is: > > commit cb2269d1308f2a410508aafecc8b0ee51959afcf > [...] Indeed, the F22 package predates that commit and does not contain this fix. I'm now running current git code (as of today) and I'll see if that re-occur. Thanks!
Hi Olivier, Have you managed to reproduce the error with the git ddx?
(In reply to Mika Kuoppala from comment #7) > Have you managed to reproduce the error with the git ddx? Hi Mika, Nope, it's been working fine since I switched to a more recent git snapshot. So either it's fixed or it was a one shot glitch, either way, I think it's safe to resolve this bug for now. I'll reopen if ever it re-occurs, but I think it's fine. Thanks!
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.