3d session fails to initialize with nouveau on many GeForce cards

[Doug McMahon]

With the current 3.7.0-2-generic #8-Ubuntu SMP & also the mainline 3.7.0-030700rc6 both unity & gnome-shell fail to start.
The greeter screen comes up fine then goes to just the background with a cursor. No actions are possible inc. going to a VT.
After a period of time system logs back out to the greeter. (or a totally corrupted greeter

A Classic (no effects) session loads fine, am filing bug from that
Installing nvidia-* & 3d sessions work fine
3d sessions also worked fine with nouveau with the previous 3.7.0.0 kernel

On todays current live image the same is seen, attempting "Try Me" goes to a blank background image with a cursor, nothing can be done at that point

Quite possibly related to this reported kernel bug
https://lkml.org/lkml/2012/11/17/108

Confirmed affected cards so far -
8400m GS
GT-210
8500 GT
7600gs
GTX 660
GTS 250

ProblemType: Bug
DistroRelease: Ubuntu 13.04
Package: linux-image-3.7.0-2-generic 3.7.0-2.8
ProcVersionSignature: Ubuntu 3.7.0-2.8-generic 3.7.0-rc5
Uname: Linux 3.7.0-2-generic x86_64
ApportVersion: 2.6.2-0ubuntu3
Architecture: amd64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: doug 2529 F.... pulseaudio
Date: Sun Nov 18 11:30:40 2012
HibernationDevice: RESUME=UUID=aa193222-44ce-4f0e-9c9f-83758886b5c8
InstallationDate: Installed on 2012-10-23 (26 days ago)
InstallationMedia: Ubuntu 12.10 "Quantal Quetzal" - Release amd64 (20121017)
MachineType: Dell Inc. XPS M1330
MarkForUpload: True
ProcEnviron:
 TERM=xterm
 PATH=(custom, no user)
 XDG_RUNTIME_DIR=<set>
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcFB: 0 nouveaufb
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-3.7.0-2-generic root=UUID=7276fdbb-8025-4feb-aff4-f2da9732a928 ro quiet splash
RelatedPackageVersions:
 linux-restricted-modules-3.7.0-2-generic N/A
 linux-backports-modules-3.7.0-2-generic N/A
 linux-firmware 1.97
SourcePackage: linux
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 12/26/2008
dmi.bios.vendor: Dell Inc.
dmi.bios.version: A15
dmi.board.vendor: Dell Inc.
dmi.chassis.type: 8
dmi.chassis.vendor: Dell Inc.
dmi.modalias: dmi:bvnDellInc.:bvrA15:bd12/26/2008:svnDellInc.:pnXPSM1330:pvr:rvnDellInc.:rn:rvr:cvnDellInc.:ct8:cvr:
dmi.product.name: XPS M1330
dmi.sys.vendor: Dell Inc.

[Brad Figg (brad-figg)]

This change was made by a bot.

[Doug McMahon (mc3man)]

Once this shows guess I'll know if the cause
http://cgit.freedesktop.org/nouveau/linux-2.6/commit/?id=7a259e65569bd7593ad541c84982027969ec9c45

[Joseph Salisbury (jsalisbury)]

@Doug

It would be great if you can post back where or not that commit fixes this bug, once it's in mainline.

[Doug McMahon (mc3man)]

On 11/19/2012 11:26 AM, Joseph Salisbury wrote:
> @Doug
>
> It would be great if you can post back where or not that commit fixes
> this bug, once it's in mainline.
>
> ** Changed in: linux (Ubuntu)
> Importance: Undecided => Medium
>
> ** Tags added: kernel-bug-exists-upstream
>
> ** Changed in: linux (Ubuntu)
> Status: Confirmed => Triaged
>
Will do - is it safe to assume it will be in rc7?

[Joseph Salisbury (jsalisbury)]

I wouldn't assume it will be in rc7. I don't see that it is in linux-next yet, so it might not make into Linus' tree until 3.8.

[Doug McMahon (mc3man)]

3.8 seems aways off so I -
Apt getted the current 13.04 kernel source (linux_3.7.0-2.8.tar.gz
extracted & patched it with patch from link in prior comment
debian/rules updateconfigs
fakeroot debian/rules clean
AUTOBUILD=1 fakeroot debian/rules binary-debs
Re-installed with the new .debs & rebooted

Booted fine & is now able to log in to 3d sessions (unity & Gs

[Joseph Salisbury (jsalisbury)]

@Doug,

I'm glad to hear you have a work around for now. The final release of raring will probably be based on the 3.8 kernel.

[Doug McMahon (mc3man)]

Confirmed affected adapters so far - 8400m GS; GT-210; 8500 GT; 7600gs
So it would appear it may be a bit more widespread than just 8400*

The biggest current 'issue' would be on new installs on daily images, affected hardware would log into basically nothing. While there are workarounds to revive theses installs not all will know how.

If it's going to be some time before officially showing then some consideration could be given to patching Ubuntu builds, haven't had any issues with the patched kernel here.

[cariboo (cariboo)]

Ass my card to the list:

02:00.0 VGA compatible controller: NVIDIA Corporation GT218 [GeForce 210] (rev a2)

[Joseph Salisbury (jsalisbury)]

I created a raring test kernel with the commit mentioned in comment #4 applied. The test kernel can be downloaded from:

http://people.canonical.com/~jsalisbury/lp1080389/

Can you test that kernel and report back if it resolves this bug?

I queried upstream and it seems the fix in comment #4 should make it into the 3.7 kernel. If it does not, I can submit a request for it to be included in raring.

Not updated yet, but the response to my query will appear here:
https://lkml.org/lkml/2012/11/17/108

[Bowmore (bowmore)]

Thanks, your test kernel works great here so far ;)

$ uname -a
Linux ltop1 3.7.0-3-generic #9~lp1080389 SMP Wed Nov 21 17:07:50 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux

[Doug McMahon (mc3man)]

On 11/21/2012 02:00 PM, Joseph Salisbury wrote:
> I created a raring test kernel with the commit mentioned in comment #4
> applied. The test kernel can be downloaded from:
>
> http://people.canonical.com/~jsalisbury/lp1080389/
>
> Can you test that kernel and report back if it resolves this bug?
>
> I queried upstream and it seems the fix in comment #4 should make it
> into the 3.7 kernel. If it does not, I can submit a request for it to
> be included in raring.
>
> Not updated yet, but the response to my query will appear here:
> https://lkml.org/lkml/2012/11/17/108
>
No issue here with the testing kernel, logs in fine to unity

[Joseph Salisbury (jsalisbury)]

Thanks for testing! I submitted this patch for inclusion in the raring kernel, until it is official merged in Linus' tree.

[Doug McMahon (mc3man)]

Seems fix released with today's linux-* (3.7.0-4.12) upgrades
Also seen in 11/26 mainline -
"Kelly Doran (1):
      drm/nvc0/disp: fix thinko in vblank regression fix.."