Created attachment 77123 [details]
Kernel logs with drm.debug=2 nouveau.debug=trace
With 3.9-rc kernels X freezes after resume from s2ram. After ~10 seconds X is restarted.
When this happens the following is printed in the kernel logs:
nouveau E[Xorg] failed to idle channel 0xcccc0000 [Xorg]
nouveau E[ PFB][0000:01:00.0] trapped read at 0x002001e020 on channel 0x00007b23 [unknown] SEMAPHORE_BG/PFIFO_READ/00 reason: PAGE_NOT_PRESENT
nouveau E[ PFIFO][0000:01:00.0] CACHE_ERROR - ch 2 [Xorg] subc 7 mthd 0x1280 data 0x00000000
nouveau E[ PGRAPH][0000:01:00.0] TRAP_TPDMA - TP0: Unhandled ustatus 0x00000004
nouveau E[ PGRAPH][0000:01:00.0] TRAP
nouveau E[ PGRAPH][0000:01:00.0] ch 2 [0x0007b23000 Xorg] subc 7 class 0x8297 mthd 0x15e0 data 0x00000000
nouveau E[ PGRAPH][0000:01:00.0] TRAP_TPDMA_2D - TP 0 - Unknown fault at address 0020339000
nouveau E[ PGRAPH][0000:01:00.0] TRAP_TPDMA_2D - TP 0 - e0c: 00000000, e18: 00000000, e1c: 00000000, e20: 00000011, e24: 0c030000
Full log attached.
Hardware is GeForce 8400M GS (10de:0427).
I seem to have the same problem with a GeForce 8800 GT card. Let me know if I can help.
You've mentioned in bug 59057
> Pontus Fuchs 2013-03-11 19:57:13 UTC
> Suspend / Resume working again with Linus' master git.
Assuming that this is the same system, has the issue regressed again ?
I'm assuming that the issue may not have been fixed and was worked around by disabling "desktop effects".
(In reply to comment #2)
> You've mentioned in bug 59057
> > Pontus Fuchs 2013-03-11 19:57:13 UTC
> > Suspend / Resume working again with Linus' master git.
> Assuming that this is the same system, has the issue regressed again ?
> I'm assuming that the issue may not have been fixed and was worked around by
> disabling "desktop effects".
The symptoms are similar to bug 59057 but not identical. In bug 59057 the screen was completely blank after resume.
In this bug the screen is first restored but everything is frozen except for the mouse pointer.
Since the symptms where not identical I decided to file a new bug.
Can you isolate(bisect) the offending commit ?
(In reply to comment #4)
> Can you isolate(bisect) the offending commit ?
It's not really possible :( The complete story goes like this:
1. Back in 3.7 everything was working fine.
2. In 3.8-rc2 resume was not working with the black screen problem
3. After some time the black screen was fixed but I now had the problem described in this bug.
If I bisect to a completely working system I'll end up before 3.8-rc2 as in bug 59057.
Regarding suspend/resume scrambling screen issues, try this patch:
[PATCH] drm/nouveau/fb: fix suspend/resume fbcon
(In reply to comment #6)
> Regarding suspend/resume scrambling screen issues, try this patch:
> [PATCH] drm/nouveau/fb: fix suspend/resume fbcon
I tried your patch but it does not solve my issue :(
The patch doesn't help me either. :-/
I finally managed to bisect this problem. This is the offending commit:
Author: Ben Skeggs <firstname.lastname@example.org>
Date: Fri Nov 16 11:54:31 2012 +1000
drm/nv50-nvc0: switch to common disp impl, removing previous version