| Summary: | [i965gm] GPU lockups (3) playing DVD with VLC | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | xorg | Reporter: | Bryce Harrington <bryce> | ||||||||||
| Component: | Driver/intel | Assignee: | Chris Wilson <chris> | ||||||||||
| Status: | RESOLVED DUPLICATE | QA Contact: | Xorg Project Team <xorg-team> | ||||||||||
| Severity: | major | ||||||||||||
| Priority: | high | ||||||||||||
| Version: | 7.6 (2010.12) | ||||||||||||
| Hardware: | x86-64 (AMD64) | ||||||||||||
| OS: | Linux (All) | ||||||||||||
| Whiteboard: | |||||||||||||
| i915 platform: | i915 features: | ||||||||||||
| Attachments: |
|
||||||||||||
|
Description
Bryce Harrington
2011-02-18 13:17:46 UTC
Created attachment 43536 [details]
BootDmesg.txt
Created attachment 43537 [details]
CurrentDmesg.txt
Created attachment 43538 [details]
XorgLog.txt
Created attachment 43539 [details]
i915_error_state.txt
For the most part all the error messages in dmesg are generic "the gpu locked up" type stuff; the one bit that stands out as unusual is: [ 5486.401830] [drm:init_ring_common] *ERROR* render ring initialization failed ctl 00000000 head 00000000 tail 00000000 start 00000000 however even that looks a bit generic. For the most part, the gpu dumps all look pretty much the same, something like this:
ACTHD: 0xffffffff
EIR: 0x00000000
EMR: 0x00000000
ESR: 0x00000000
PGTBL_ER: 0x00000000
IPEHR: 0x00000000
IPEIR: 0x00000000
INSTDONE: 0x00000000
INSTDONE1: 0x00000000
busy: Row 0, EU 0
busy: Row 0, EU 1
busy: Row 0, EU 2
busy: Row 0, EU 3
busy: Row 1, EU 0
busy: Row 1, EU 1
busy: Row 1, EU 2
busy: Row 1, EU 3
busy: Strips and Fans
busy: Setup Engine
busy: Windowizer
busy: Dispatcher
busy: Projection and LOD
busy: Dependent address generator
...
All of them show 0's for all the gpu dump error registers, which seems odd.
We've had other reports where these were all zero's but no other instances where three crash reports got generated simultaneously like this.
Fwiw, this user had a similar bug on the previous release, which was believed fixed by commit 85345517 (drm/i915: Retire any pending operations on the old scanout when switching), but the lockup just mysteriously went away for the user so we could never verify the fix at the time. https://bugs.freedesktop.org/show_bug.cgi?id=28964 https://bugs.launchpad.net/ubuntu/maverick/+source/linux/+bug/603064 Since the user, hardware, use case, and gpu error codes are all basically the same between this report and that one, it suggests that either the earlier bug was fixed and has regressed with the newer kernel, or else that the bug has been there all along and just was masked by something. Not sure. |
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.