Bug 34429

Summary:

[PATCH]Failure during resume on NV50: INIT_AUXCH: rd auxch fail -121

Product:

xorg

Reporter:

Fabian Moser <e-mail+freedesktop>

Component:

Driver/nouveau

Assignee:

Nouveau Project <nouveau>

Status:

RESOLVED FIXED

QA Contact:

Xorg Project Team <xorg-team>

Severity:

major

Priority:

medium

CC:

homyur, maciej

Version:

git

Keywords:

patch

Hardware:

x86-64 (AMD64)

OS:

Linux (All)

See Also:

https://bugzilla.redhat.com/show_bug.cgi?id=646540
https://bugzilla.redhat.com/show_bug.cgi?id=598162

Whiteboard:

i915 platform:

i915 features:

Bug Depends on:

27398, 31676

Bug Blocks:

Attachments:

Description	Flags
Kernel messages from boot thorugh suspend to resume	none
Power manager activities during suspend/resume	none
Xorg messages from boot to resume	none
dmesg output with "drm.debug=14"	none
dmesg from the nouveau git	none
Apply power to the panel before enabling hotplug interrupts and before communicating with the panel on the I2C bus	none

Description Fabian Moser 2011-02-18 00:55:56 UTC

Created attachment 43511 [details]
Kernel messages from boot thorugh suspend to resume

Upon return from suspend, the display turns white and stays that way. The X server is unusable. Other VTs as well. The problem is 100% reproducible.

I'm using openSUSE 11.4 RC 1 on an HP 8440p with a NVS 3100m card (GT218M, thus NV50-type). I reported first at https://bugzilla.novell.com/show_bug.cgi?id=669932 and was asked to report here. 

I'm using packages dating git20110210 from http://download.opensuse.org/repositories/home:/jobermayr/openSUSE_Factory/x86_64/ 

Since the problem is reproduced every time, I would be happy to help if I can provide debugging information. The installation currently is dedicated for this sole purpose (I need suspend-resume in a would-be production system).

Bugs I found related but IMO still different enough to report this one:
https://bugs.freedesktop.org/show_bug.cgi?id=25762
https://bugs.freedesktop.org/show_bug.cgi?id=25410

Comment 1 Fabian Moser 2011-02-18 00:56:48 UTC

Created attachment 43512 [details]
Power manager activities during suspend/resume

Comment 2 Fabian Moser 2011-02-18 00:57:22 UTC

Created attachment 43513 [details]
Xorg messages from boot to resume

Comment 3 Fabian Moser 2011-02-22 07:57:49 UTC

An apparently similar issue has been reproduced during Fedora 15 nouveau Test Day with the provided Live ISO.

Comment 4 Yuriy Khomchik 2011-02-25 12:41:19 UTC

I have a similar problem on the same hardware with Ubuntu 11.04 Alpha 2.
dmesg output contains the same lines:
  ...
  [drm] nouveau 0000:01:00.0: Restoring mode...
  [drm] nouveau 0000:01:00.0: INIT_AUXCH: rd auxch fail -121
  [drm] nouveau 0000:01:00.0: INIT_AUXCH: rd auxch fail -121
  [drm] nouveau 0000:01:00.0: 0xBDCC: auxch rd fail: -121
  [drm] nouveau 0000:01:00.0: INIT_AUXCH: rd auxch fail -121
  ...

Comment 5 Yuriy Khomchik 2011-02-25 14:30:07 UTC

Created attachment 43841 [details]
dmesg output with "drm.debug=14"

Comment 6 Yuriy Khomchik 2011-04-25 15:05:58 UTC

Probably this bug is similar to bug 27398. Common features:
1. Identical lines in dmesg output (INIT_AUXCH: rd auxch fail -121).
2. Displays are connected via DisplayPort (bug 34429 - eDP, bug 27398 - DP).
3. Displays have not been initialized by BIOS (bug 34429 - after resume, bug 27398 - external display).

Comment 7 Yuriy Khomchik 2011-07-15 15:03:14 UTC

Similar bug: bug 31676.

Comment 8 Yuriy Khomchik 2011-09-09 06:09:20 UTC

Created attachment 50998 [details]
dmesg from the nouveau git

Comment 9 Yuriy Khomchik 2011-09-09 06:13:28 UTC

Bug still reproduced on the latest git version. See attached dmesg log.

Comment 10 Yuriy Khomchik 2012-01-29 00:51:57 UTC

Created attachment 56275 [details] [review]
Apply power to the panel before enabling hotplug interrupts and before communicating with the panel on the I2C bus

Comment 11 Ben Skeggs 2012-01-30 20:21:30 UTC

(In reply to comment #10)
> Created attachment 56275 [details] [review] [review]
> Apply power to the panel before enabling hotplug interrupts and before
> communicating with the panel on the I2C bus

I just pushed a slightly different version of this patch to nouveau git.

Comment 12 Yuriy Khomchik 2012-01-30 23:48:45 UTC

(In reply to comment #11)
> (In reply to comment #10)
> > Created attachment 56275 [details] [review] [review] [review]
> > Apply power to the panel before enabling hotplug interrupts and before
> > communicating with the panel on the I2C bus
> 
> I just pushed a slightly different version of this patch to nouveau git.

Thanks, Ben.

Comment 13 Ben Skeggs 2012-01-31 00:46:10 UTC

(In reply to comment #12)
> (In reply to comment #11)
> > (In reply to comment #10)
> > > Created attachment 56275 [details] [review] [review] [review] [review]
> > > Apply power to the panel before enabling hotplug interrupts and before
> > > communicating with the panel on the I2C bus
> > 
> > I just pushed a slightly different version of this patch to nouveau git.
> 
> Thanks, Ben.

No worries.  Thanks for tracking down the root cause of this.  I was aware of this GPIO function previously, but didn't honestly expect any VBIOS init table to default it to off when there was a panel attached :)

Comment 14 Emil Velikov 2012-06-10 02:09:29 UTC

As mentioned, the fix has been pushed. Please reopen if the issue persist

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.