| Summary: | [845g regression] pipe A assertion failure (expected on, current off) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | DRI | Reporter: | Chris Wilson <chris> | ||||||||||
| Component: | DRM/Intel | Assignee: | Intel GFX Bugs mailing list <intel-gfx-bugs> | ||||||||||
| Status: | CLOSED FIXED | QA Contact: | Intel GFX Bugs mailing list <intel-gfx-bugs> | ||||||||||
| Severity: | normal | ||||||||||||
| Priority: | medium | CC: | mrmazda | ||||||||||
| Version: | unspecified | ||||||||||||
| Hardware: | Other | ||||||||||||
| OS: | All | ||||||||||||
| Whiteboard: | |||||||||||||
| i915 platform: | i915 features: | ||||||||||||
| Attachments: |
|
||||||||||||
|
Description
Chris Wilson
2013-07-01 14:54:31 UTC
Created attachment 81815 [details]
dmesg
The clock mismatch is just Jesse's fallout, but the pipe state warning is strange. We seem to start out with the pipe enabled, but somewhere along the lines it seems to get disabled. Can you please inquire gdb about where exactly that assert_pipe call is? For the record, the clock mismatch is now fixed. We are still forcing pipea quirk on 845g. Not only is that wrong, but it is the cause of the assertion failures. Created attachment 82051 [details] [review] improve pipe A quirk asserts Dunno whether this is really the right approach or whether we should just give up on assert_pipe for pipe A if the quirk is there ... Bikeshed: s/quirk_force_pipe_a_disable/dev_priv->quirks & QUIRK_PIPEA_IGNORE/ Pipe assertion remains. Created attachment 82371 [details]
similar assert, drm.debug=0xe, by different reporter
New to 3.10.0+ kernel git downloaded 7/10/2013. The 7/7/2013 git did not suffer from this. I'm in a good position to do bisects. Just give the word.
The force being discussed may have been there on 845G to fix an issue when the setterm command shut off my monitor. If nobody else had issues like that, you're more than welcome to remove the fix. I'll find some way to work around it.
(In reply to comment #6) > Bikeshed: s/quirk_force_pipe_a_disable/dev_priv->quirks & QUIRK_PIPEA_IGNORE/ > > Pipe assertion remains. I'll try your patch. BTW, I just came here to report the backtrace. So far, nothing is actually wrong yet. As for things being reported as being off, I'm not sure where things are actually off. The screen blanking time is actually a little shorter than usual, the power light stays green. For my backup/emergency partition, the monitor did go into "Auto Adjust in Progress" mode like something had changed. All is well, though. Happening early during every boot in openSUSE 13.1B1's 3.11.1 with libdrm_intel1-2.4.46-117.1.i586, dating back at least to 3.10.x. Created attachment 86326 [details] [review] preserve pipe A quirk Please test the attached patch. commit 67c72a12254101d4e8d9b9f3a02646ba0be84a2d Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Tue Sep 24 11:46:14 2013 +0200 drm/i915: preserve pipe A quirk in i9xx_set_pipeconf |
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.