[nvidia-glx][nvidia-glx-new] black/blank screen on resume after suspend/hibernate/vt switch when using compiz/beryl/desktop effects unless sync to vblank is disabled

Actually, it's the driver that's supposed to bring back the textures. This problem has existed with the nvidia drivers longer than compiz has existed so I'm not sure we can expect a fix anytime soon. Reassigning to nvidia-glx.

Oh right. I wonder if it's possible to force compiz to put them back when the machine resumes from suspend. It's a bit ugly, but may be an OK workaround.

Do you also get a blank screen if you switch virtual consoles?

If I'm using compiz and switch to any other console (ctrl-alt-f1) and then switch back (ctrl-alt-f7) I get a blank screen with just a mouse pointer.

I ALSO get a blank screen when I come back from suspend whether or not I am using compiz.

I wonder if we are talking about two seperate bugs here.

Eythian:
Could you add the output of
dpkg -l nvidia-glx\* | grep ii
to this bug?

Setting to needsinfo pending reply from Eythian.

I just tested it, I do get a blank screen after switching to a console and back.

The command:
$ dpkg -l nvidia-glx\* | grep ii
ii nvidia-glx 1.0.9631+2.6.20.5-15.20 NVIDIA binary XFree86 4.x/X.Org driver

Setting back to confirmed.

You _might_ be able to work around this by blacklisting your default AGP driver and using NvAGP instead...

(this bug also applies to nvidia-glx-new)

Just a note, this bug also happens on my desktop which doesn't have AGP so no AGP driver. I haven't tried the different module on the laptop yet, I haven't had occasion to reboot.

Eythian:
You will have a pcigart which is often driven in a similar fashion. It is unclear whether the NvAGP option still affects pcigart users...

I have the very same issue here. Is there anything I can help to speed up a fix of that?

There is a fix here (https://help.ubuntu.com/community/NvidiaLaptopBinaryDriverSuspend) but it did not work for me. I have the same bug too with a GeForce Go 7300.

does not work for me either - same issue

Just to be clear, this bug is only in relation to people who have working suspend and resume with the nvidia binary drivers _until_ they use compiz and beryl. If suspend and resume does not work and you are not using beryl/compiz then you probably should search for a more appropriate bug.

yes that's clear. hibernate and resume works with compiz disalbed but not with it enabled. nvidia card with nvidia binary driver.

I have found a workaround for making suspend work, finally, using the nvidia drivers. I had the same resume problem, whether or not compiz/beryl was enabled, but now suspend works very fine :-)
https://bugs.launchpad.net/ubuntu/+source/linux-restricted-modules-2.6.20/+bug/111380

hi i have the same problem and if i switch to metacity before suspending then everything goes well I can resume my computer back.
But if I keep beryl on I just get a blank screen. I tried all the solution that are mentioned on this page non worked for me.

@Sitsofe: I may need some pointers in using the nvidia AGP, I added 'agpgart' to /etc/modprobe.d/blacklist, added:
Option "NvAGP" "1"
to the video card section of x.org, and then rebooted, however:
$ lsmod|grep agp
intel_agp 25116 1
agpgart 35400 2 nvidia,intel_agp

it appears to be loading anyway.

Eythian:
You want to blacklist intel_agp not agpgart...

does pcigart exist?and if so as what? there is nvidia-agp module

$ lsmod|grep agp
nvidia_agp 9500 0
agpgart 35788 3 drm,nvidia,nvidia_agp

is what I get.
on resume from compiz session I see a lit black screen then unlit black screen the lit black screen with small colored box behind where the pointer happens to be.
can move the pointer ... small colored box stays where it is, but the keyboard doesn't respond, however the power button does so ...

HP Compaq presario v6001au nvidia 6150

get this too

May 2 16:55:30 lapy2 kernel: [458029.148000] RPC: failed to contact portmap (errno -5).

May 1 18:55:54 lapy2 kernel: [378852.974000] Disabling non-boot CPUs ...
May 1 18:55:54 lapy2 kernel: [378852.974000] kvm: disabling virtualization on CPU1
May 1 18:55:54 lapy2 kernel: [378852.974000] BUG: at arch/i386/kernel/smp.c:692 smp_call_function_single()
May 1 18:55:54 lapy2 kernel: [378852.974000] [smp_call_function_single+244/352] smp_call_function_single+0xf4/0x160
May 1 18:55:54 lapy2 kernel: [378852.974000] [vprintk+924/944] vprintk+0x39c/0x3b0
May 1 18:55:54 lapy2 kernel: [378852.974000] [_spin_lock+13/112] _spin_lock+0xd/0x70
May 1 18:55:54 lapy2 kernel: [378852.974000] [<f8e8fc20>] decache_vcpus_on_cpu+0x10/0xa0 [kvm]
May 1 18:55:54 lapy2 kernel: [378852.974000] [<f8e8fd53>] kvm_cpu_hotplug+0x73/0x80 [kvm]
May 1 18:55:54 lapy2 kernel: [378852.974000] [notifier_call_chain+37/48] notifier_call_chain+0x25/0x30
May 1 18:55:54 lapy2 kernel: [378852.974000] [_cpu_down+73/528] _cpu_down+0x49/0x210
May 1 18:55:54 lapy2 kernel: [378852.975000] [disable_nonboot_cpus+108/272] disable_nonboot_cpus+0x6c/0x110
May 1 18:55:54 lapy2 kernel: [378852.975000] [printk+27/32] printk+0x1b/0x20
May 1 18:55:54 lapy2 kernel: [378852.975000] [enter_state+124/496] enter_state+0x7c/0x1f0
May 1 18:55:54 lapy2 kernel: [378852.975000] [state_store+177/192] state_store+0xb1/0xc0
May 1 18:55:54 lapy2 kernel: [378852.975000] [state_store+0/192] state_store+0x0/0xc0
May 1 18:55:54 lapy2 kernel: [378852.975000] [subsys_attr_store+41/64] subsys_attr_store+0x29/0x40
May 1 18:55:54 lapy2 kernel: [378852.975000] [sysfs_write_file+154/256] sysfs_write_file+0x9a/0x100
May 1 18:55:54 lapy2 kernel: [378852.975000] [vfs_write+190/400] vfs_write+0xbe/0x190
May 1 18:55:54 lapy2 kernel: [378852.975000] [sysfs_write_file+0/256] sysfs_write_file+0x0/0x100
May 1 18:55:54 lapy2 kernel: [378852.975000] [sys_write+65/112] sys_write+0x41/0x70
May 1 18:55:54 lapy2 kernel: [378852.975000] [sysenter_past_esp+105/169] sysenter_past_esp+0x69/0xa9
May 1 18:55:54 lapy2 kernel: [378852.975000] =======================
May 1 18:55:54 lapy2 kernel: [378853.085000] CPU 1 is now offline
May 1 18:55:54 lapy2 kernel: [378853.085000] SMP alternatives: switching to UP code
May 1 18:55:54 lapy2 kernel: [378853.088000] CPU1 is down

vivichrist:
portmap is to do with network services (like NFS or NIS) so is almost certainly nothing to do with this bug. The BUG stuff looks similarly unrelated to this particular report and you are probably best off seeing whether you can reproduce that issue when using the open source nv drivers and filing a new bug report. This bug is solely about the screen being black after resuming or switching VTs while using compiz/beryl...

dame ... was hoping ... praying that there was something in there that could be useful.

I can't see the point in testing the nv driver since opengl graphics hardware acceleration is the culprit.

I get black screen on switching VTs regardless of compiz/beryl.

vivichrist:
I would still test with the nv drivers. If it happens when you are using those too then you have a completely different issue. However, whatever you do, I would recommend filing a new bug and posting the link to your new bug from here.

Exactly the same results: when I wake from suspend it juts sits at a black screen. Mouse moved ,and area under where the mouse starts has some graphical corruption, but other than that pure black and it stays stuck there.

Log is below. This is on a laptop with a 7600gt. I am using the restricted drivers module. Option "AddARGBGLXVisuals" "True" is in xorg.conf but compiz is enabled. Note that vbetool is very unhappy.

May 10 13:47:39 robertj-laptop gnome-power-manager: (robertj) Suspending computer because the DBUS method Suspend() was invoked
May 10 13:47:41 robertj-laptop kernel: [ 4040.231980] ADDRCONF(NETDEV_UP): eth1: link is not ready
May 10 13:47:41 robertj-laptop kernel: [ 4040.487514] ACPI: PCI interrupt for device 0000:04:00.0 disabled
May 10 13:47:45 robertj-laptop kernel: [ 4042.748509] Disabling non-boot CPUs ...
May 10 13:47:45 robertj-laptop kernel: [ 4042.863337] CPU 1 is now offline
May 10 13:47:45 robertj-laptop kernel: [ 4042.863342] SMP alternatives: switching to UP code
May 10 18:19:30 robertj-laptop -- MARK --
May 10 18:19:30 robertj-laptop kernel: [ 4042.870139] CPU1 is down
May 10 18:19:30 robertj-laptop kernel: [ 4042.870142] Stopping tasks ... done.
May 10 18:19:30 robertj-laptop kernel: [ 4043.086935] Suspending console(s)
May 10 18:19:30 robertj-laptop kernel: [ 4046.004631] ACPI: PCI interrupt for device 0000:06:04.2 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4046.017803] ohci1394 does not fully support suspend and resume yet
May 10 18:19:30 robertj-laptop kernel: [ 4046.037763] eth1: Going into suspend...
May 10 18:19:30 robertj-laptop kernel: [ 4046.148953] ACPI: PCI interrupt for device 0000:02:00.0 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4046.161578] NVRM: RmPowerManagement: 3
May 10 18:19:30 robertj-laptop kernel: [ 4047.192118] ACPI: PCI interrupt for device 0000:00:1f.2 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4047.205133] ACPI: PCI interrupt for device 0000:00:1d.7 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4047.221004] ACPI: PCI interrupt for device 0000:00:1d.3 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4047.221041] ACPI: PCI interrupt for device 0000:00:1d.2 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4047.221078] ACPI: PCI interrupt for device 0000:00:1d.1 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4047.221114] ACPI: PCI interrupt for device 0000:00:1d.0 disabled
May 10 18:19:30 robertj-laptop kernel: [ 4047.228952] ACPI: PCI interrupt for device 0000:00:1b.0 disabled
May 10 18:19:30 robertj-laptop kernel: [ 0.717799] ACPI: PCI Interrupt 0000:00:1b.0[A] -> GSI 22 (level, low) -> IRQ 22
May 10 18:19:30 robertj-laptop kernel: [ 0.921854] ACPI: PCI Interrupt 0000:00:1d.0[A] -> GSI 23 (level, low) -> IRQ 23
May 10 18:19:30 robertj-laptop kernel: [ 0.921923] usb usb2: root hub lost power or was reset
May 10 18:19:30 robertj-laptop kernel: [ 0.921946] ACPI: PCI Interrupt 0000:00:1d.1[B] -> GSI 19 (level, low) -> IRQ 19
May 10 18:19:30 robertj-laptop kernel: [ 0.922013] usb usb3: root hub lost power or was reset
May 10 18:19:30 robertj-laptop kernel: [ 0.922036] ACPI: PCI Interrupt 0000:00:1d.2[C] -> GSI 18 (level, low) -> IRQ 18
May 10 18:19:30 robertj-la...

Eythian:
Does disabling sync to vblank by doing
gconftool -s /apps/compiz/general/screen0/options/sync_to_vblank-t boolean false
make any difference?

I was having the same problem with beryl. Turning off sync-to-vblank fixed it for me.

The same as bigfish with beryl : I disabled the vsync through Beryl Settings Manager > General options > Vblank sync (unchecked it).
Work nice from suspend / resume now ;-)

Disabling sync to vblank makes it seem to work now. thanks!

Setting back to confirmed after reply.

this worked for me to ... yay now I can have compiz all the time.

no improvement here

asus z71vp w/ go6600 v/ 9755

ignore above, didn't execute correctly, will try again, and report.

Sitsofe rocks! I've finally got compiz _and_ suspend on my Dell D820 with nvidia/Feisty. Ubuntu is great, but add GL Effects and it's perfect. What is sync to vblank anyway?

Hey Bingo YOU rock! How about that? I'll just subscribe you so you see this reply though (you may want to unsubscribe yourself from this bug to avoid seeing replies).

As to your query, sync to vblank is there to stop tearing (at the cost of speed). Your monitor has a refresh rate and if things are not timed to be in sync with when a vertical blank takes place you can get tearing (imagine you have a black screen and you go to draw a white cube. If for whatever reason you can't draw the screen with the cube in less than the time it takes to refresh or you are unlucky then you will end up with half a cube at the bottom and a black screen at the top). See http://www.mythtv.org/wiki/index.php/Frame_display_timing for a better description.

(PS: if you want to see replies it is best to subscribe yourself to the bug you pose the question in)

I confirmed this.On my Thinkpad R31 it also happen."sync to vblank" disabling make hiberate work.

anyone else experiencing this bug in gutsy? in feisty i had this problem, used the 'sync to vblank' fix with the beryl settings manager. now in gutsy, i'm using compiz, but the sync-to-vblank setting no longer fixes things. this is on a dell latitude d820 with the nvidia driver. switching from compiz to metacity fixes the issue.

unsubscribe

I am having the same problem in gusty. All works well when compiz is disabled. But when enabled, PC does not resume from hibernation unless sync-to-vblank setting is disabled. The used to be fix, disabling sync-to-vblank, no longer completely works. I get my desktop, but things are major screwy. I can make out the desktop, but there are artifacts all over the place that change with cursor movement, lines dancing all over, etc. Hard to explain because its so screwy.

I've also been having this problem in Gutsy. Disabling "Sync to VBlank" doesn't seem to have any effect on the problem. It suspends/resumes perfectly when im not running compiz. However, when i am running compiz i get a black screen with a cursor on resume. If i restart x (ctrl+alt+backspace) Then i can log in, however i can no longer connect to wireless networks, or open any applications. When i try to shut down i get an "unregister_device" error. It seems like its trying to shut down the wireless service but cant for some reason. This error just repeats itself until i restart.

This is marked as a duplicate of #88152, but their descriptions look fairly disparate. I do not have any problems switching between virtual terminals, nor do I have the nvidia driver.

I wrote a comment in #136774:11 which upon closer examination might belong in this bug. I think I am seeing the same on a fresh install of Gutsy on a Toshiba Libretto U105 with Intel driver. Anybody care to comment on whether this is the right place to look for a solution?

So with Gutsy, if I understand correctly, I should even be able to get Compiz if I install some extra compiz management stuff (what exactly?) and/or fiddle with the gconftool command line in comment #28?

... And I guess I should mention for those who don't follow the link to my other comment that disabling Compiz solved the problem for me as a workaround, but it would of course be nice to have the option of having a bit of pretty Compiz bling on this puppy. But having working suspend AND hibernate is already an improvement over Feisty for me.

and then its back with vengeance (but not 64bit gutsy ... have to use noapic all the time and works flawlessly).
but vt switching works fine.
also, it seems to me to be time dependant i.e the longer my laptop is up and running (i will confirm this)....

Nov 16 12:17:35 comkrapper kernel: [ 1.964000] Magic number: 15:224:279
Nov 16 12:17:35 comkrapper kernel: [ 1.964000] hash matches device ttyy7
Nov 16 12:17:35 comkrapper kernel: [ 1.964000] hash matches device ttyva
Nov 16 12:17:35 comkrapper kernel: [ 1.964000] hash matches device tty16

sorry also this seems to me to coincide with video stuter (avi playback video will freeze and resume but audio continues)