Xorg fails with "NVIDIA(0): Failed to allocate primary buffer: out of memory."

Bug #607490 reported by Søren Holm
54
This bug affects 9 people
Affects Status Importance Assigned to Milestone
nvidia-graphics-drivers (Ubuntu)
Fix Released
Medium
Canonical Desktop Team
Maverick
Fix Released
Medium
Canonical Desktop Team

Bug Description

Something happenend in linux-image-generic since linux-image-2.6.35-7-generic. in 2.6.35-7 the proprietery NVIDIA-drivers was working, but in 2.6.35-8 Xorg fails with "(EE) NVIDIA(0): Failed to allocate primary buffer: out of memory."

My only solution is to add "nopat" to the kernel command-line. I'm unsure of what PAT actualy is, but works around the error for me. I do not know which of the 2.6.35-8 changes that caused this.

ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: linux-image-2.6.35-9-generic 2.6.35-9.14
Regression: Yes
Reproducible: Yes
ProcVersionSignature: Ubuntu 2.6.35-9.14-generic 2.6.35-rc5
Uname: Linux 2.6.35-9-generic x86_64
NonfreeKernelModules: nvidia
AcpiTables: Error: command ['kdesudo', '--', '/usr/share/apport/dump_acpi_tables.py'] failed with exit code 1:
AlsaVersion: Advanced Linux Sound Architecture Driver Version 1.0.23.
Architecture: amd64
ArecordDevices:
 **** List of CAPTURE Hardware Devices ****
 card 0: NVidia [HDA NVidia], device 0: ALC660-VD Analog [ALC660-VD Analog]
   Subdevices: 1/1
   Subdevice #0: subdevice #0
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: sgh 1691 F.... kmix
CRDA: Error: [Errno 2] No such file or directory
Card0.Amixer.info:
 Card hw:0 'NVidia'/'HDA NVidia at 0xfecd8000 irq 22'
   Mixer name : 'Realtek ALC660-VD'
   Components : 'HDA:10ec0660,10430000,00100001'
   Controls : 18
   Simple ctrls : 11
Date: Mon Jul 19 23:52:45 2010
HibernationDevice: RESUME=UUID=72b0f80c-0773-4f9a-977f-e86c85be3593
InstallationMedia: Kubuntu 10.10 "Maverick Meerkat" - Alpha amd64 (20100616)
Lsusb:
 Bus 002 Device 002: ID 046d:c03d Logitech, Inc. M-BT96a Pilot Optical Mouse
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 001 Device 003: ID 174f:a821 Syntek Web Cam - Packard Bell BU45, PB Easynote MX66-208W
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
MachineType: Packard Bell BV Easy Note_MX51
ProcCmdLine: BOOT_IMAGE=/boot/vmlinuz-2.6.35-9-generic root=UUID=1269b5ff-f739-45f7-912f-def9cfd5b455 ro quiet splash nopat
ProcEnviron:
 SHELL=/bin/bash
 PATH=(custom, user)
 LANG=da_DK.UTF-8
 LANGUAGE=
RelatedPackageVersions: linux-firmware 1.37
RfKill:
 0: phy0: Wireless LAN
  Soft blocked: no
  Hard blocked: no
SourcePackage: linux
dmi.bios.date: 10/11/2006
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: T12MvNC.204
dmi.board.asset.tag: To Be Filled By O.E.M.
dmi.board.name: Easy Note_MX51
dmi.board.vendor: Packard Bell BV
dmi.board.version: 1.00
dmi.chassis.asset.tag: 0x00000000
dmi.chassis.type: 10
dmi.chassis.vendor: Packard Bell BV
dmi.chassis.version: 1.0
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrT12MvNC.204:bd10/11/2006:svnPackardBellBV:pnEasyNote_MX51:pvrPB71HD0091:rvnPackardBellBV:rnEasyNote_MX51:rvr1.00:cvnPackardBellBV:ct10:cvr1.0:
dmi.product.name: Easy Note_MX51
dmi.product.version: PB71HD0091
dmi.sys.vendor: Packard Bell BV

Revision history for this message
Søren Holm (sgh) wrote :
Revision history for this message
Søren Holm (sgh) wrote :

Apparently disabling boot splash by removing "splash" from the kernel commandline is also a workaround for this issue. So I to not thnik it is a kernel-issue anymore. So I changed it to nvidia-current because both linux-image and nvidia-current had been was upgraded prior.

affects: linux (Ubuntu) → nvidia-graphics-drivers (Ubuntu)
Revision history for this message
Søren Holm (sgh) wrote :

2.6.35-10-generic fixes this issue. So I guess it was a kernel thing afterall. :D

Changed in nvidia-graphics-drivers (Ubuntu):
status: New → Fix Released
Revision history for this message
Søren Holm (sgh) wrote :

I was too fast on fast. On most boots it fails, but on some it doesn't. When booting with 2.6.35-10 that was the case. Anyway, this nevertheless not fixed.

Changed in nvidia-graphics-drivers (Ubuntu):
status: Fix Released → New
Bryce Harrington (bryce)
tags: added: kubuntu
Revision history for this message
Søren Holm (sgh) wrote :

Even though removing "splash" from the kernel commandline it only works when doing it manualy from grub during boot. Removing "splash" from menu.cfg does not work. So I'm fearing it might be related to something that grub does when it displays the menu.

Do you guys have any ideas as to what I could try ?

Revision history for this message
Søren Holm (sgh) wrote :

To sum up.

My system always boots on the low graphics mode,

When in low graphics mode and I press CTRL-ALT-F1 to get to the console, there is usualy no text.

BUT if there are text I can restart X and get the nvidia dirver up and running.

If I during boot enter grub and press 'e' to edit,and CTRL-x to boot it is more likely that en login prompt will be visible - actually just booting without entering grub has not worked for me since 2010-07-12

Revision history for this message
Søren Holm (sgh) wrote :

I have found that my work-laptop containing an intel chip also has issues. It sometimes end up in low-graphics mode. But on that machine it always works to just restart Xorg.

So something races during boot in relation to graphics. How is it decided that low graphics-mode should be initiated?

Revision history for this message
Christian Mangold (neversfelde) wrote :

I can confirm these problems.

Changed in nvidia-graphics-drivers (Ubuntu):
status: New → Confirmed
Revision history for this message
Søren Holm (sgh) wrote :

Great ...... do you have any ideas as to what to try to hunt it down.

Revision history for this message
Christian Mangold (neversfelde) wrote :

Unfortunately not at the moment, but I can also confirm that setting nopat seems to solve the problem. Disabling the splash, did not work for me.

Revision history for this message
Søren Holm (sgh) wrote : Re: [Bug 607490] Re: Xorg fails with "NVIDIA(0): Failed to allocate primary buffer: out of memory."

Read further down ..... for me the real difference is weither I enter grub or
not.

Revision history for this message
jolo (joop-loep) wrote :

I have Ubuntu Maverick on an old AthlonXP PC, with an Asus A7V266-E motherboard and a 7600GS video card from Club 3D. I can confirm the nvidia error, the NOPAT and splash workaround.

jolo@AthlonXP:~$ uname -a
Linux AthlonXP 2.6.35-13-generic-pae #18-Ubuntu SMP Sat Jul 31 03:07:03 UTC 2010 i686

jolo@AthlonXP:~$ lspci -tv
-[0000:00]-+-00.0 VIA Technologies, Inc. VT8366/A/7 [Apollo KT266/A/333]
           +-01.0-[01]----00.0 nVidia Corporation G73 [GeForce 7600 GS]
           +-0d.0 VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
           +-0d.1 VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
           +-0d.2 VIA Technologies, Inc. USB 2.0
           +-0d.3 VIA Technologies, Inc. VT6306/7/8 [Fire II(M)] IEEE 1394 OHCI Controller
           +-0e.0 Brooktree Corporation Bt848 Video Capture
           +-0f.0 3Com Corporation 3c905C-TX/TX-M [Tornado]
           +-10.0 Creative Labs SB Live! EMU10k1
           +-10.1 Creative Labs SB Live! Game Port
           +-11.0 VIA Technologies, Inc. VT8233 PCI to ISA Bridge
           +-11.1 VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE
           +-11.2 VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
           +-11.3 VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
           \-11.4 VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller

Revision history for this message
Daniel Hahler (blueyed) wrote :

@Søren: is it enough to get the list of kernels/options displayed already for you? Or do you need to edit an entry or select another one (stopping the countdown)?
I'm asking since the screen resolution does not change when editing entries or something similar (AFAICS).

Revision history for this message
Søren Holm (sgh) wrote :

Yes. Just entering grub and "looking" at the options is enough. But the laster
couple of day since the Xorg upgrade the problems have progressed. It seems
like the proprietary driver is no longer compatible wilth Xorg. So it is not
100% improsible to use anything else than nouveau.

Daniel Hahler (blueyed)
Changed in nvidia-graphics-drivers (Ubuntu):
status: Confirmed → Triaged
importance: Undecided → Medium
Revision history for this message
Daniel Hahler (blueyed) wrote :

@Søren: bug 616023 is about nvidia-current not working with current xserver - my workaround to this was adding/using the xorg-edgers ppa, but nvidia-current 256.44 will land in Maverick soon, too.
I would have expected the dependencies to hold the xserver upgrade back though.

So, you're currently not seeing grub at all? Only the "Press <something> to get the grub menu..." during boot? I'm asking since I get the list of kernels for $timeout seconds.

For me, adding "nopat" (via /etc/default/grub and update-grub) seems to do the trick, but keep in mind that I seem to see automatically what you call manually for working around this (the grub menu).

Revision history for this message
Søren Holm (sgh) wrote :

I'll try xorg-edgers-ppa and report back how it behaves.

Revision history for this message
Søren Holm (sgh) wrote :

I have nvidia-current 256.44 installed now. But it behaved the same.

If I let the system boot normally I end up in low graphics mode.
If I interrupt grub and press 'e' to edit the item it will boot, and press CTRL-x to actually boot it everything is ok. My system starts up perfectly X starts up using nvidia-current.

This is really strange.

Revision history for this message
jolo (joop-loep) wrote :

Referring back at my Ubuntu Maverick on an old AthlonXP PC, with an Asus A7V266-E motherboard and a 7600GS video card from Club 3D.

Today I downloaded the maverick-desktop-i386.iso daily build. Because is was 734,0 MB I burned it on a DVD-RW.

After the clean installation I updated it with

sudo apt-get install aptitude
sudo aptitude update && sudo aptitude safe-upgrade

And I installed the current nvidia drivers version 256.44
It seems the "out of memory" error is fixed, without using any workarounds.

Revision history for this message
Krzysztof Matusik (krzysztof-matusik) wrote :

This bug affects me exactly the same way as described by Søren Holm.
The card is geforce 7600GS.
Kernel- 2.6.35-17-generic x86_64
I'm not having this problem when using old kernel from lucid (2.6.32) with new nvidia driver from maverick.

I'm attaching some logs.

Revision history for this message
jolo (joop-loep) wrote :
Download full text (4.2 KiB)

My Ubuntu Maverick on an old AthlonXP PC, with an Asus A7V266-E motherboard and a 7600GS video card from Club 3D worked with the current nvidia drivers version 256.44 until
2010-08-19 20:29:40. After the updates of that date, I had a different nvidia problem then Søren Holm. These are the updates

Start-Date: 2010-08-19 20:29:40
Install: libvala-0.10-0:i386 (0.9.7-0ubuntu1, automatic)
Upgrade: ubuntuone-client:i386 (1.3.8-0ubuntu2, 1.3.10-0ubuntu1), libgtk2.0-common:i386 (2.21.5-1ubuntu6, 2.21.6-1ubuntu1), espeak:i386 (1.43.03-2ubuntu1, 1.44.03-0ubuntu1), libgail18:i386 (2.21.5-1ubuntu6, 2.21.6-1ubuntu1), gnome-desktop-data:i386 (2.31.6-0ubuntu1, 2.31.90-0ubuntu1), nautilus:i386 (2.31.6-0ubuntu3.2, 2.31.90-0ubuntu1), libindicator0:i386 (0.3.11-0ubuntu2, 0.3.12-0ubuntu1), libcomerr2:i386 (1.41.12-1ubuntu1, 1.41.12-1ubuntu2), indicator-application:i386 (0.2.3-0ubuntu2, 0.2.4-0ubuntu1), libgail-common:i386 (2.21.5-1ubuntu6, 2.21.6-1ubuntu1), gnome-power-manager:i386 (2.31.6-0ubuntu1, 2.31.90-0ubuntu1), rhythmbox:i386 (0.13.0git20100715-0ubuntu6, 0.13.0git20100715-0ubuntu8), gtk2-engines-pixbuf:i386 (2.21.5-1ubuntu6, 2.21.6-1ubuntu1), nautilus-data:i386 (2.31.6-0ubuntu3.2, 2.31.90-0ubuntu1), ubuntuone-client-gnome:i386 (1.3.8-0ubuntu2, 1.3.10-0ubuntu1), e2fsprogs:i386 (1.41.12-1ubuntu1, 1.41.12-1ubuntu2), libsyncdaemon-1.0-1:i386 (1.3.8-0ubuntu2, 1.3.10-0ubuntu1), libgdu-gtk0:i386 (2.30.1-1, 2.30.1-2), gnome-about:i386 (2.31.6-0ubuntu1, 2.31.90-0ubuntu1), rhythmbox-plugin-cdrecorder:i386 (0.13.0git20100715-0ubuntu6, 0.13.0git20100715-0ubuntu8), libgdu0:i386 (2.30.1-1, 2.30.1-2), eog:i386 (2.31.90-0ubuntu1, 2.31.90-0ubuntu2), busybox-static:i386 (1.15.3-1ubuntu1, 1.15.3-1ubuntu2), gnome-disk-utility:i386 (2.30.1-1, 2.30.1-2), libgtk2.0-bin:i386 (2.21.5-1ubuntu6, 2.21.6-1ubuntu1), espeak-data:i386 (1.43.03-2ubuntu1, 1.44.03-0ubuntu1), python-vte:i386 (0.25.90-0ubuntu2, 0.25.90-0ubuntu3), libvte9:i386 (0.25.90-0ubuntu2, 0.25.90-0ubuntu3), libclutter-1.0-0:i386 (1.2.12-0ubuntu4, 1.2.12-0ubuntu5), libglib2.0-data:i386 (2.25.14-1ubuntu1, 2.25.14-1ubuntu2), libespeak1:i386 (1.43.03-2ubuntu1, 1.44.03-0ubuntu1), e2fslibs:i386 (1.41.12-1ubuntu1, 1.41.12-1ubuntu2), rhythmbox-plugins:i386 (0.13.0git20100715-0ubuntu6, 0.13.0git20100715-0ubuntu8), libglib2.0-0:i386 (2.25.14-1ubuntu1, 2.25.14-1ubuntu2), indicator-sound:i386 (0.3.9-0ubuntu2, 0.4.0-0ubuntu1), libss2:i386 (1.41.12-1ubuntu1, 1.41.12-1ubuntu2), python-ubuntuone-client:i386 (1.3.8-0ubuntu2, 1.3.10-0ubuntu1), libgnome-desktop-2-17:i386 (2.31.6-0ubuntu1, 2.31.90-0ubuntu1), busybox-initramfs:i386 (1.15.3-1ubuntu1, 1.15.3-1ubuntu2), libnautilus-extension1:i386 (2.31.6-0ubuntu3.2, 2.31.90-0ubuntu1), libvte-common:i386 (0.25.90-0ubuntu2, 0.25.90-0ubuntu3), libgtk2.0-0:i386 (2.21.5-1ubuntu6, 2.21.6-1ubuntu1)
End-Date: 2010-08-19 20:31:47

This is in the /var/log/Xorg.0.log

[ 17.845] ================ WARNING WARNING WARNING WARNING ================
[ 17.845] This server has a video driver ABI version of 8.0 that this
driver does not officially support. Please check
http://www.nvidia.com/ for driver updates or downgrade to an X
server with a supported driver ABI.
[ 17.845] ===================...

Read more...

Revision history for this message
Krzysztof Matusik (krzysztof-matusik) wrote :

@jolo : It looks like nvidia kernel module is missing in your system, right? try reinstalling nvidia-current.

Revision history for this message
jolo (joop-loep) wrote :

@Krzysztof Matusik I just checked this. There is a nvidia kernel loaded

jolo@jolo-desktop:~$ lsmod | grep nvidia

nvidia 10195198 0

agpgart 32075 2 nvidia,via_agp

jolo@jolo-desktop:~$ dmesg | grep nvidia

[ 14.357302] nvidia: module license 'NVIDIA' taints kernel.

[ 17.315761] nvidia 0000:01:00.0: PCI INT A -> Link[LNKA] -> GSI 11 (level, low) -> IRQ 11

[ 18.244266] nvidia 0000:01:00.0: putting AGP V2 device into 4x mode

[ 20.430668] nvidia 0000:01:00.0: putting AGP V2 device into 4x mode

[ 21.652614] nvidia 0000:01:00.0: putting AGP V2 device into 4x mode

[ 22.818040] nvidia 0000:01:00.0: putting AGP V2 device into 4x mode

[ 23.871694] nvidia 0000:01:00.0: putting AGP V2 device into 4x mode

[ 24.922755] nvidia 0000:01:00.0: putting AGP V2 device into 4x mode

In the meanwhile I found another workaround. Instead of adding "nopat" in /etc/default/grub and update-grub, I added just one space in /etc/default/grub between quiet and splash
and I did the ussally update-grub.

# If you change this file, run 'update-grub' afterwards to update

# /boot/grub/grub.cfg.

GRUB_DEFAULT=0

#GRUB_HIDDEN_TIMEOUT=0

GRUB_HIDDEN_TIMEOUT_QUIET=true

GRUB_TIMEOUT=10

GRUB_DISTRIBUTOR=`lsb_release -i -s 2> /dev/null || echo Debian`

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash"

GRUB_CMDLINE_LINUX=""

Thus with two spaces instead of one space in /etc/default/grub between quiet and splash the nvidia error seemd to be fixed for me. I also added this file as an
attachment.

Revision history for this message
Søren Holm (sgh) wrote :

Yeah .... that does not surprise me. As I wrote ealier I just had to enter the
editing mode in grub while booting, change nothing, and press CTRL-x to boot.
It works everytime. Have you tried that also ?

Revision history for this message
jolo (joop-loep) wrote :

@Søren Holm I just tried your suggestion several times. Thus with only one space between quiet and splash, and editing mode in grub while booting, change nothing, and press CTRL-x to boot. It does not work for me every time. It seems the new version 256.44 of the Xorg nvidia driver sometimes failed to load in Maverick. So I am not sure anymore whether one of the workarounds actually works or not.

Revision history for this message
Krzysztof Matusik (krzysztof-matusik) wrote :

Guys,

I've just tried to start system (into X) several times with or without grub-editing without success (the "out of memory" error). The computer was powered off before starting.

Then I've added "nopat" in grub command line- X system started as it should.

Then I've rebooted without adding "nopat" parameter- and system started correctly again.

No clues what's going on...

Revision history for this message
Søren Holm (sgh) wrote :

I can add that the "add one space"-solution does not work for me ;(

Revision history for this message
jolo (joop-loep) wrote :

Sinds the nvidia driver version 256.44 randomly hangs at Xorg startup, there is no way I can determine whether a workaround works. Could that be caused by the nvidia driver version 256.44. I just read about freezes with nvidia driver version 256.44 in Arch Linux https://bbs.archlinux.org/viewtopic.php?id=102202
http://www.nvnews.net/vbulletin/showthread.php?t=153875

My /var/log/Xorg.0.log in comment #20 looks like at https://bugs.launchpad.net/ubuntu/maverick/+source/nvidia-graphics-drivers/+bug/616023

Revision history for this message
Søren Holm (sgh) wrote :

As of now it is fixed for me now. Problably it is the grub-pc update to version 1.98+20100804-4ubuntu1

The log includes this :
  * Go back to gfxpayload=text by default for Ubuntu 10.10, as the kernel
    needs more work to avoid lockups with some video cards

Seems to be the change that fixes it.

Revision history for this message
Krzysztof Matusik (krzysztof-matusik) wrote :

My problems stopped as discribed by Søren.
Well, this is it. I'm feeling unsatisfied however, because I don't understand how grub is capable of this...

regards

Revision history for this message
jolo (joop-loep) wrote :

Today I updated Maverick. And after several reboots without a problem. I also think the grub-pc update might fixed the problem. And I did not use the "nopat" option anymore.

Only the warnings in my /var/log/Xorg.0.log from comment #20 are still there

[ 17.845] ================ WARNING WARNING WARNING WARNING ================
[ 17.845] This server has a video driver ABI version of 8.0 that this
driver does not officially support. Please check
http://www.nvidia.com/ for driver updates or downgrade to an X
server with a supported driver ABI.
[ 17.845] =================================================================
[ 17.845] (WW) NVIDIA: The driver will continue to load, but may behave strangely.
[ 17.845] (WW) NVIDIA: This driver was compiled against the X.Org server SDK from git commit 2307ab5bc9365ebbe04568edb7c7620a23689b70 and may not be compatible with the final version of this SDK.
[ 17.845] (WW) NVIDIA: This server has an unsupported input driver ABI version (have 11.0, need < 10.0). The driver will continue to load, but may behave strangely.

But no errors anymore.

Changed in nvidia-graphics-drivers (Ubuntu Maverick):
assignee: nobody → Canonical Kernel Team (canonical-kernel-team)
assignee: Canonical Kernel Team (canonical-kernel-team) → Canonical Desktop Team (canonical-desktop-team)
Revision history for this message
Robbie Williamson (robbiew) wrote :

Can anyone confirm that this problem still exists?

Revision history for this message
Søren Holm (sgh) wrote :

It gone from my side.

Revision history for this message
Didier Roche-Tolomelli (didrocks) wrote :

I'm happily running the nvidia proprietary driver there for quite some time in the alpha release.
Still no issue as of today.

Revision history for this message
Didier Roche-Tolomelli (didrocks) wrote :

Looking at recent comments, setting it as fixed release. Please reopen if it still occurs to you with an uptodate maverick

Changed in nvidia-graphics-drivers (Ubuntu Maverick):
status: Triaged → Fix Released
Revision history for this message
udippel (udippel) wrote :

It *does* occur to me in Maverick, *if* I set
GRUB_GFXPAYLOAD_LINUX=keep
, and get the
set gfxpayload=keep
in grub.cfg

Just for completeness. Any good reason to reopen it?

Uwe,
who doesn't mind living with ugly boot screens otherwise.

Revision history for this message
Daniel Hahler (blueyed) wrote :

It was still happening for me, and I investigated into a kernel patch (landed in 2.6.36.2) yesterday.
This did not help on it's own, but I had to change GRUB_GFXPAYLOAD_LINUX=keep to GRUB_GFXPAYLOAD_LINUX=1280x1024, see: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/605837/comments/9

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.