[nvidia-glx] can suspend/resume once, suspending a second time hangs, nvidia-glx-legacy works

Bug #109529 reported by insyzygy
16
This bug affects 1 person
Affects Status Importance Assigned to Milestone
linux-restricted-modules-2.6.20 (Ubuntu)
Won't Fix
Undecided
Unassigned
linux-restricted-modules-2.6.22 (Debian)
New
Undecided
Unassigned

Bug Description

I have a dell inspiron 8500 upgraded to feisty (through edgy) from dapper.
In dapper suspend/resume worked fine (once I set /etc/default/acpi-support
correctly.)

In feisty, after booting up I can suspend and resume successfully one time. If I at any time suspend a second time it tries
to go to sleep then wakes up with the error message on screen

ata1: ACPI get timing mode failed.
ata2: ACPI get timing mode failed.

It hangs with this text on screen. I can sometimes switch to a different virtual terminal but can't actually login as the keyboard unresponsive
I have to hard power down at this point

This may be related to some other ata errors I am getting at boot. During the boot, it drops out of the splash screen and repeatedly displays

ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0
ata2.00: cmd a0/00:00:00:00:20/00:00:00:00:00/a0 tag 0 cdb 0x1b data 0
         res 51/51:03:00:00:20/00:00:00:00:00/a0 Emask 0x1 (device error)
ata2.00: configured for UDMA/33
ata2: EH complete
ata2.00: 12 bytes trailing data
ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen
ata2.00: cmd a0/00:00:00:00:20/00:00:00:00:00/a0 tag 0 cdb 0x1b data 0
              res 51/51:03:00:00:20/00:00:00:00:00/a0 Emask 0x3 (HSM violation)
ata2: soft resetting port
ata2.00: configured for UDMA/33
ata2: EH complete
sr0: CDROM (ioctl) error, command: <6>Start/Stop Unit 1b 00 00 00 03 00
sr: Current [descriptor]: sense key: Medium Error
    Additional sense: Address mark not found for data field

Eventually it gets passed this and goes to the graphica login. I don't detect anything not working properly due to this (other than suspend).

                                                                                                           Josh

Revision history for this message
insyzygy (setthiry) wrote :

This appears to actually be a bug in the newer nvidia video drivers, the ata errors seem to be a separate problem. If I switch the nvidia-legacy drivers than I can suspend multiple times with no problem.

                                                                               Josh

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

insyzygy:
Can you suspend multiple times with the nv open source driver too? If so could you attach the following files when using nvidia-glx:
/var/log/Xorg.0.log
/etc/X11/xorg.conf
dmesg.txt produced by
dmesg > dmesg.txt

Could you add the output produced by the following commands to this bug:
dpkg -l nvidia-\* | grep ii
lspci | grep -i nv
?

Changed in acpi:
status: Unconfirmed → Needs Info
Revision history for this message
insyzygy (setthiry) wrote :

I can't suspend with the nv drivers at all.
This may be due to the fact that in order to get suspend to work with the nividia card I had to change some things in
/etc/default/acpi-support.
 When I resume from suspend with nv drivers I just get a blank screen and then the fan goes to max indicating
the CPU is pegged and I have to hard reboot. Do you still want the above info for the nvidia-glx drivers.

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

insyzygy:
That's really quite strange. But yes, please attach the requested information anyway.

Revision history for this message
insyzygy (setthiry) wrote :

I'm having trouble going back to the nvidia-glx drivers to produce the log files for you, (though the nvidia-glx-legacy works fine).

I have an xorg.conf file that works fine with nvidia-glx-legacy.

If I install nvidia-glx-new using apt-get then when I reboot X-windows fails to start

it gives the error no screens found and reports

"the nvidia GeForce 4 4200 Go GPU installed in this system is suported trhough Nvidia 1.0-96xx legacy drivers,
the 1.0-9755 Nvidia driver will ignore this GPU. continuing probe."

I know that I wasn't using nvidia-legacy drivers when i first installed fiesty because the nvidia splash screen is different with the nvidia-glx-legacy drivers (its the same one as in dapper) than the nvidia splash screen that was displayed with the nvidia-glx drivers.

Any suggestions?

                                                                                          Josh

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

insyzygy:
Your card isn't supported by nvidia-glx-new. However when removed nvidia-glx-new leaves a residual configuration file that can cause problems if you go on to install the other nvidia drivers. You will have to purge uninstall (apt-get --purge remove nvidia-glx-new ) before installing nvidia-glx ...

Please post the output of:
ls -al /lib/linux-restricted-modules/

Revision history for this message
insyzygy (setthiry) wrote :
Revision history for this message
insyzygy (setthiry) wrote :
Revision history for this message
insyzygy (setthiry) wrote :
Revision history for this message
insyzygy (setthiry) wrote :

 dpkg -l nvidia-\* | grep ii
ii nvidia-glx 1.0.9631+2.6.20.5-15.20 NVIDIA binary XFree86 4.x/X.Org driver
ii nvidia-kernel-common 20051028+1ubuntu7 NVIDIA binary kernel module common files

lspci | grep -i nv
01:00.0 VGA compatible controller: nVidia Corporation NV28 [GeForce4 Ti 4200 Go AGP 8x] (rev a1)

ls -al /lib/linux-restricted-modules/
total 68
drwxr-xr-x 15 root root 4096 2007-04-30 02:40 .
drwxr-xr-x 19 root root 8192 2007-04-20 19:38 ..
drwxr-xr-x 21 root root 4096 2006-05-25 02:50 2.6.15-23-386
drwxr-xr-x 21 root root 4096 2006-06-15 20:35 2.6.15-25-386
drwxr-xr-x 19 root root 4096 2006-06-21 22:25 2.6.15-25-686
drwxr-xr-x 21 root root 4096 2006-07-22 16:29 2.6.15-26-386
drwxr-xr-x 19 root root 4096 2006-07-22 16:29 2.6.15-26-686
drwxr-xr-x 21 root root 4096 2006-09-27 01:15 2.6.15-27-386
drwxr-xr-x 19 root root 4096 2006-09-27 01:16 2.6.15-27-686
drwxr-xr-x 21 root root 4096 2007-04-10 02:52 2.6.15-28-386
drwxr-xr-x 19 root root 4096 2007-04-10 02:52 2.6.15-28-686
drwxr-xr-x 18 root root 4096 2007-04-20 10:36 2.6.17-11-386
drwxr-xr-x 16 root root 4096 2007-04-20 10:36 2.6.17-11-generic
drwxr-xr-x 19 root root 4096 2007-04-20 19:11 2.6.20-15-386
drwxr-xr-x 17 root root 4096 2007-04-20 19:11 2.6.20-15-generic
-rw-r--r-- 1 root root 58 2007-04-29 13:55 .nvidia_new_installed

Revision history for this message
insyzygy (setthiry) wrote :

One additional oddity. I was able to get the nvidia-glx drivers to work using by removing the other nvidia packages with
the --purge.
However, everytime I reboot X fails to start and i have to do

rmmod nvidia
nvidia-glx-config enable
/etc/init.d/gdm start

To get X to start with the glx modules.

Is there something additional I need to do so that I don't have to do the above each time. I have read in the forums about
other people with this problem. The error when X tries to start is api mismatch.

Also I checked that as before using the nvidia-glx module I can suspend and resume one time, and the next time I suspend it hangs.

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

insyzygy:
I would need to see a dmesg from a boot where it failed. Could you also check that /lib/linux-restricted-modules/.nvidia_new_installed does not exist (if it still does then remove just that file and reboot again)?

I also have an idea about the single suspend and resume but we should address this booting issue first.

Revision history for this message
insyzygy (setthiry) wrote :

That fixed it.

I removed /lib/linux-restricted-modules/.nvidia_new_installed and then I rebooted and it went into X windows with no trouble.

Revision history for this message
insyzygy (setthiry) wrote :

That fixed the boot issue (suspend still doesn't work properly)

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

(The .nvidia_new_installed issue is bug #106365 )

Does setting
POST_VIDEO=false
in /etc/default/acpi-support make a difference?

Revision history for this message
insyzygy (setthiry) wrote :

Attached is my acpi-support file. I already was using POST_VIDEO=False. I had to tweak these a bit
in particular save_vbe_state, save_video_pci_state,and POST_VIDEO. I was following various forums until it worked.

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

I'm not sure if
SAVE_VIDEO_PCI_STATE=true
actually needs to be false. However it sounds like you couldn't suspend at all unless this was set.

I'm curious, if you back up your /etc/default/acpi-support and reset it to the defaults, can you suspend and resume when using the nv driver?

Revision history for this message
insyzygy (setthiry) wrote :

Is there a way to reset to the defaults automatically, or can you tell me which lines to change ( I don't remember which ones I modified.)

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

SAVE_VBE_STATE=false
POST_VIDEO=true
# SAVE_VIDEO_PCI_STATE=true
are probably the lines you need to change.

Revision history for this message
insyzygy (setthiry) wrote :

Using the nv drivers with acpi-support modified as above I can suspend/resume multiple times.

Not sure if its relevant but using the nv drivers I have a different screen resolution than with the nvidia drivers as my display
is best with a 1680x1050 resolution.

                                                                                                                                                                    Josh

Revision history for this message
insyzygy (setthiry) wrote :

An additional observation. If I suspend and resume with the nv drivers. On resume switching to a virtual terminal shuts off the screen and I have to reboot to get it back on.
As long as I stay in X windows everything is fine and I can suspend and resume as many times as I please.

But if I do ctrl-alt-f1 then I get a blank screen and after ctrl-alt-f7 it is still blank. I can tell that everything is still working because ctl-alt-backspace produces a chime indicating that X windows went to the login screen, my monitor is just blank.

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

insyzygy:
OK your testing seems to suggest this is either an AGP or an nvidia binary driver issue. Can you switch back to the nvidia driver and try adding NvAGP as described on https://help.ubuntu.com/community/NvidiaLaptopBinaryDriverSuspend and then reboot and retest? This will disable AGP and allow us to narrow the problem down.

With regards to the black screen when switching to a VT when after resuming with the nv drivers - please file a new bug on that.

Revision history for this message
insyzygy (setthiry) wrote :

I believe this is how my xorg.conf file was originally set up before I started changing drivers around.
Anyway I added the line as described and also set POST_VIDEO=false in acpi-support.

With these setting I get the original behavious of the bug report. I can suspend and resume one time. The second time I suspend it immediately wakes up and hangs with
text on the screen and an unresponsive keyboard. (Oddly I can swtich virtuals terminals but can't log in as actual characters keys have no effect). I have to do a hard power off to recover.

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

insyzygy:
Have you enabled desktop effects (in other words are you using compiz/beryl)?

Revision history for this message
insyzygy (setthiry) wrote :

Desktop effects are not enabled.I did enable them for a bit when I installed but then turned them off.

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote : Re: [nvidia-glx] can suspend/resume once, suspending a second time hangs

Moving out of needsinfo and setting back to unconfirmed. Punting from acpi -> linux-restricted-modules-2.6.20 because it does seem to be entirely dependent on the binary drivers as stated by the reporter in an earlier comment.

Changed in acpi:
status: Needs Info → Unconfirmed
Revision history for this message
cement_head (andorjkiss) wrote :

I can confirm on Dell Inspiron 8100.

This is a RECENT development; over the last two weeks - ACPI update or HAL update

Was working flawlessly before (Suspend 10 or more times in a row).

Also getting the ACPI ATA:timing error.

- CH

Revision history for this message
Lars Andersson (larsand) wrote :

Hi,

I'm experiencing the very same problems as described above, running Feisty on a Dell Latitude D800 with a GeForce4 Ti 4200 Go AGP 8x, nvidia drivers 9639. I noticed that suspend/resume seems to work fine when used from the console (with POST_VIDEO=true), even when the nvidia module is loaded. But in X, it hangs when trying to suspend the second time after boot (POST_VIDEO=false). Has any of you come up with a solution? Let me know if I can provide any more useful information.

Cheers,

Lars

Revision history for this message
cement_head (andorjkiss) wrote : Re: [Bug 109529] Re: [nvidia-glx] can suspend/resume once, suspending a second time hangs, nvidia-glx-legacy works

Hi Lars,

    Using Synaptic, force downgrade of the following 4 packages: "hal;
hal-device-manager; libhal1; libhal-storage1". Once that is done, LOCK
the version. Then only use the GUI of update manager to do updates, or
Synaptic (apt-get upgrade) with update the bad HAL packages.

- Andor

On Sun, 2007-06-10 at 11:56 +0000, Lars Andersson wrote:

> Hi,
>
> I'm experiencing the very same problems as described above, running
> Feisty on a Dell Latitude D800 with a GeForce4 Ti 4200 Go AGP 8x, nvidia
> drivers 9639. I noticed that suspend/resume seems to work fine when used
> from the console (with POST_VIDEO=true), even when the nvidia module is
> loaded. But in X, it hangs when trying to suspend the second time after
> boot (POST_VIDEO=false). Has any of you come up with a solution? Let me
> know if I can provide any more useful information.
>
> Cheers,
>
> Lars
>

_____________________________________________________________
Andor J Kiss
eMAIL: <email address hidden>

Revision history for this message
Lars Andersson (larsand) wrote : Re: [Bug 109529] Re: [nvidia-glx] can suspend/resume once, suspending a second time hangs, nvidia-glx-legacy works

Hi Andor,

Thanks for the suggestions... However, I had this problem already
before trying to upgrade HAL. I tried the new HAL just to see how it
worked, but nothing changed. I have now reverted to the standard
version, and my laptop still hangs when trying to suspend a second
time. As I mentioned, it's probably related to X. I can suspend/resume
several times in a row from the console prompt, but as soon as I have
a gdm login or X session running, it hangs at the second attempt to
suspend.

/Lars

On 6/12/07, cement_head <email address hidden> wrote:
> Hi Lars,
>
> Using Synaptic, force downgrade of the following 4 packages: "hal;
> hal-device-manager; libhal1; libhal-storage1". Once that is done, LOCK
> the version. Then only use the GUI of update manager to do updates, or
> Synaptic (apt-get upgrade) with update the bad HAL packages.
>
> - Andor
>
>
> On Sun, 2007-06-10 at 11:56 +0000, Lars Andersson wrote:
>
> > Hi,
> >
> > I'm experiencing the very same problems as described above, running
> > Feisty on a Dell Latitude D800 with a GeForce4 Ti 4200 Go AGP 8x, nvidia
> > drivers 9639. I noticed that suspend/resume seems to work fine when used
> > from the console (with POST_VIDEO=true), even when the nvidia module is
> > loaded. But in X, it hangs when trying to suspend the second time after
> > boot (POST_VIDEO=false). Has any of you come up with a solution? Let me
> > know if I can provide any more useful information.
> >
> > Cheers,
> >
> > Lars
> >
>
>
> _____________________________________________________________
> Andor J Kiss
> eMAIL: <email address hidden>
>
> --
> [nvidia-glx] can suspend/resume once, suspending a second time hangs, nvidia-glx-legacy works
> https://bugs.launchpad.net/bugs/109529
> You received this bug notification because you are a direct subscriber
> of the bug.
>

Revision history for this message
Mark Rabideau (manyroads) wrote :

Hi all,

I have noticed that I get the following error message on entering and exiting hibernation:

ACPI set timing mode failed (on ata1 & 2)

I am running a Dell e1505n w/
GeForce Go 7300, PCI-E 16x, 256 MB, 200 MHz,
NVIDIA Linux x86 Kernel Module 1.0-9631 Thu Nov 9 17:38:10 PST 2006

I am also using Beryl/Compiz-Fusion
Feisty 7.04 (Ubuntu)

Hibernation works but not correctly... I has frequent screen garbage and ends up with an odd little box in the upper right hand quadrant of the screen just before the Login box appears.

I do not know if this is the same problem as reported by others in this thread (if it is I hope this note helps with the problem resolution). If the posting belongs elsewhere please feel free to move it or "tell me where to go" ;^)

...mark

Revision history for this message
DanielPhillips (dwp-rushpost) wrote :

Hi,

Also having problems with a Dell 8500 which will only suspend / resume one time using nvidia driver. I am actually running Debian Unstable (Sid), currently kernel 2.6.23.1

The second time the suspend / resume cycle will just have a blank screen on resume and no way out except to power off.

I have had this problem for a long time and have only really had suspend working well using the Open Nv driver.

I hope this helps and maybe this bug will get looked at. It would be nice to get this problem solved.

Dan

Revision history for this message
Id2ndR (id2ndr) wrote :

I have an intel G33 based motherboad, and an nvidia 6600 GPU. I have the same trouble with linux 2.6.22 : I can resume from suspend only once (the screen remain blank the second time, and I have to reset the computer).

Revision history for this message
Bryce Harrington (bryce) wrote : linux-restricted-modules-2.6.20 is obsolete

This package has become obsolete so we're closing out the bug report as WONTFIX.
Thanks for reporting it though!

Changed in linux-restricted-modules-2.6.20:
status: New → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

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