[drm] nouveau 0000:02:00.0: PFIFO_CACHE_ERROR - Ch 2/2 Mthd 0x0800 Data 0x00000000

[Bowmore]

Screen corruption occurs at start up and reboot, for
- login screen
- desktop screen

Prerequisities:
- running the nouveau driver
- running the 3.5.0-10 and -11 kernels

At non-autologin one can type the login password blindly but just to enter a corrupted desktop screen. At autologin a corrupted desktop screen is presented.

The CPU-load is very high and caused by millions of loggings of messages like:
[drm] nouveau 0000:02:00.0: PFIFO_CACHE_ERROR - Ch 2/2 Mthd 0x0800 Data 0x00000000

so it seems to be related to DRM and the PFIFO cache.

Workaround 1:
Use an older kernel, e.g. 3.5.0-9 in my case.

Workaround 2:
Activate Ctrl-Alt-Backspace (as Alt-SysRq-K has been disabled) and use that shortcut to log out from the corrupted desktop screen. Then back to a normal login screen the login works as normal both for the 3.5.0-10 and -11 kernels and even for the 3.6.0 rc1 and rc2 kernels and normal CPU-load.

Instead of Ctrl-Alt-Backspace one can do the following:
- Press Ctrl-Alt-F2
- login with your username and password
- Restart lightdm with: sudo service lightdm restart

Workaround 3:
Remove the grub splash option.

Not sure it's the libdrm-nouveau2 package but remember that one was introduced for DRM in the time frame for this bug.

ProblemType: Bug
DistroRelease: Ubuntu 12.10
Package: libdrm-nouveau2 2.4.38-0ubuntu1
ProcVersionSignature: Ubuntu 3.5.0-11.11-generic 3.5.2
Uname: Linux 3.5.0-11-generic x86_64
.tmp.unity.support.test.0:

ApportVersion: 2.4-0ubuntu8
Architecture: amd64
CompizPlugins: [core,composite,opengl,compiztoolbox,decor,vpswitch,mousepoll,wall,place,snap,move,regex,imgpng,resize,widget,gnomecompat,session,unitymtgrabhandles,animation,workarounds,fade,expo,scale,ezoom,unityshell]
CompositorRunning: compiz
Date: Mon Aug 20 18:40:48 2012
DistUpgraded: Fresh install
DistroCodename: quantal
DistroVariant: ubuntu
DkmsStatus:
 virtualbox, 4.1.18, 3.5.0-10-generic, x86_64: installed
 virtualbox, 4.1.18, 3.5.0-11-generic, x86_64: installed
 virtualbox, 4.1.18, 3.6.0-030600rc2-generic, x86_64: installed
 virtualbox-guest, 4.1.18, 3.5.0-11-generic, x86_64: installed
EcryptfsInUse: Yes
GraphicsCard:
 NVIDIA Corporation GT216 [GeForce GT 330M] [10de:0a29] (rev a2) (prog-if 00 [VGA controller])
   Subsystem: Samsung Electronics Co Ltd Device [144d:c06a]
InstallationMedia: Ubuntu 12.04 LTS "Precise Pangolin" - Beta amd64 (20120421)
MachineType: SAMSUNG ELECTRONICS CO., LTD. R580/R590
ProcEnviron:
 TERM=xterm
 PATH=(custom, no user)
 LANG=sv_SE.UTF-8
 SHELL=/bin/bash
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-3.5.0-11-generic root=UUID=d290b749-dff5-4155-936a-23e3effc21a8 ro quiet splash ipv6.disable=1
Renderer: Software
SourcePackage: libdrm
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 06/22/2010
dmi.bios.vendor: Phoenix Technologies Ltd.
dmi.bios.version: 11JB.M044.20100622.hkk
dmi.board.asset.tag: Tag 12345
dmi.board.name: R580/R590
dmi.board.vendor: SAMSUNG ELECTRONICS CO., LTD.
dmi.board.version: Not Applicable
dmi.chassis.asset.tag: No Asset Tag
dmi.chassis.type: 9
dmi.chassis.vendor: SAMSUNG ELECTRONICS CO., LTD.
dmi.chassis.version: N/A
dmi.modalias: dmi:bvnPhoenixTechnologiesLtd.:bvr11JB.M044.20100622.hkk:bd06/22/2010:svnSAMSUNGELECTRONICSCO.,LTD.:pnR580/R590:pvrNotApplicable:rvnSAMSUNGELECTRONICSCO.,LTD.:rnR580/R590:rvrNotApplicable:cvnSAMSUNGELECTRONICSCO.,LTD.:ct9:cvrN/A:
dmi.product.name: R580/R590
dmi.product.version: Not Applicable
dmi.sys.vendor: SAMSUNG ELECTRONICS CO., LTD.
version.compiz: compiz 1:0.9.8+bzr3249-0ubuntu4
version.ia32-libs: ia32-libs N/A
version.libdrm2: libdrm2 2.4.38-0ubuntu1
version.libgl1-mesa-dri: libgl1-mesa-dri 8.0.4-1ubuntu1
version.libgl1-mesa-dri-experimental: libgl1-mesa-dri-experimental N/A
version.libgl1-mesa-glx: libgl1-mesa-glx 8.0.4-1ubuntu1
version.xserver-xorg-core: xserver-xorg-core 2:1.12.99.904-0ubuntu1
version.xserver-xorg-input-evdev: xserver-xorg-input-evdev 1:2.7.3-0ubuntu1
version.xserver-xorg-video-ati: xserver-xorg-video-ati 1:6.99.99~git20120713.6ef1ad6a-0ubuntu1
version.xserver-xorg-video-intel: xserver-xorg-video-intel 2:2.20.3-0ubuntu1
version.xserver-xorg-video-nouveau: xserver-xorg-video-nouveau 1:1.0.1-4~ubuntu1

[Bowmore (bowmore)]

Another workaround to avoid this drm issue is to remove the grub splash option.

[dino99 (9d9)]

i've also got this issue with 3.5 and 3.6 kernels on i386 system.
My workaround is via xorg.conf:

create (or replace what's in a current one) an /etc/X11/xorg.conf with
Code:
Section "Device"
        Identifier "My Graphics"
        Driver "nouveau"
        Option "GLXVBlank" "on"
EndSection

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[dino99 (9d9)]

more comments:
this has happened with the previous libdrm 2.4.38-0ubuntu1. (maybe the next 2.4.38-0ubuntu2 will help)

libdrm (2.4.38-0ubuntu2) quantal; urgency=low

  * Add libdrm2-udeb; Xserver 1.13 wants libdrm for hotplug config, so
    will be needed for the 1.13 xserver-xorg-core-udeb.

 -- Christopher James Halse Rogers <email address hidden> Tue, 21 Aug 2012 14:15:30 +1000

[dino99 (9d9)]

but right now, there is no libdrm2-udeb nor xserver-xorg-core-udeb in the quantal archives.

[Bowmore (bowmore)]

The following messages in the kern.log seem to cause this issue:

Aug 22 00:25:01 ltop1 kernel: [ 32.421438] [drm] nouveau 0000:02:00.0: fail set_domain
Aug 22 00:25:01 ltop1 kernel: [ 32.421444] [drm] nouveau 0000:02:00.0: validate vram_list
Aug 22 00:25:01 ltop1 kernel: [ 32.421447] [drm] nouveau 0000:02:00.0: validate: -22

present from kernel 3.5.0-10 and on and generated by
../drivers/gpu/drm/nouveau/nouveau_gem.c.

[Bowmore (bowmore)]

I'm attaching the kern.log when running the 3.5.0-11 kernel.

[Bowmore (bowmore)]

@dino99
vblank only works for some nvidia graphic cards (not mine) but worth testing as a workaround.

I guess .udeb packages exist in the repos but not visible as those are special stripped ditto packages for special use to save ram. The kernel uses those instead of the more memory consuming .deb packages and I guess xorg does as well.

[Bowmore (bowmore)]

For the record this proprietary driver works just fine:
nvidia-current 304.43-0ubuntu2

[dino99 (9d9)]

Similar issue but with nouveau : bug #1044912 (vbios init table race)

[Bowmore (bowmore)]

I made some further investigations for the nouveau driver today.

Kernels 3.5.0-10 to 3.5.0-12:
The issue is present with a corrupted screen and massive logging of PFIFO_CACHE_ERROR messages.

Kernel 3.5.0-13
The issue seems gone, i.e the nouveau driver again!

It seems that the fixes for kernel 3.5.0-13 (two bugs) also fixed this issue!
My guess is the framebuffer recycling issue being fixed.

The issue also is present in kernel 3.6, but never mind as 3.5 will be shipped for Quantal.

[Bowmore (bowmore)]

Works with kernel 3.5.0-14 too so seems to have been fixed.

[markb (mark-blakeney)]

@bowmore, I am on 3.5.0-13 using xserver-xorg-video-nouveau 1:1.0.1+git20120816.3aa9a296-0ubuntu0sarvatt~precise but I am still seeing small screen corruptions and plenty of nouveau PFIFO_CACHE_ERROR messages.

[Bowmore (bowmore)]

It seems you're running quantal with the wrong nouveau driver.
The nouveau driver you're running is the xorg-edgers version for precise.

So, switch to xorg-edgers for quantal and if that doesn't solve your problem then purge xorg-edgers.

[markb (mark-blakeney)]

I am running precise (with xorg-edgers ppa clearly). I didn't realise this bug was relevant only to systems running quantal? I am seeing the same issue but ignore my comments if not appropriate here.

[Bowmore (bowmore)]

Well, it might be relevant when running a mainstream precise version.

In this case you should report this issue to the xorg crack pushers team as the ppa says:
"For questions and bugs with software in this PPA please contact xorg crack pushers."