Screen/system is dead on resume (unable to resume with all current drivers)

With the latest drivers (NVIDIA 319.23) and kernel 3.9.4, my system is unable to resume from suspend.

X.org log on resume contains this:

(II) NVIDIA(0): Setting mode "DFP-0:nvidia-auto-select{}"
(WW) NVIDIA(0): WAIT (2, 8, 0x8000, 0x0000fa10, 0x0000202c)
(WW) NVIDIA(0): WAIT (1, 8, 0x8000, 0x0000fa10, 0x0000202c)

Ctrl+Alt+Backspace doesn’t help because the keyboard is seemingly dead - I’m only able to SysRq+S/U/B.

This is from dmesg:

Jun 25 00:06:12 localhost kernel: [12687.804267] Restarting tasks ... done.
Jun 25 00:06:12 localhost kernel: [12687.853262] NVRM: GPU at 0000:01:00: GPU-136382c0-06fa-2c0f-977a-4f04b1755070
Jun 25 00:06:12 localhost kernel: [12687.853266] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000088 0100cb4f 00000007 00000000
Jun 25 00:06:12 localhost kernel: [12687.853299] NVRM: Xid (0000:01:00): 56, CMDre 00000000 0000008c 00000000 00000005 0000102b
Jun 25 00:06:13 localhost kernel: [12688.852064] NVRM: Xid (0000:01:00): 31, Ch 00000000, engmask 00000101, intr 10000000
Jun 25 00:06:15 localhost kernel: [12690.852480] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00001005
Jun 25 00:06:52 localhost kernel: [12727.355961] SysRq : Emergency Sync
Jun 25 00:06:52 localhost kernel: [12727.359121] Emergency Sync complete
Jun 25 00:06:53 localhost kernel: [12728.733663] SysRq : Emergency Remount R/O

My GPU is not overclocked, it’s an NVIDIA Corporation GK106 [GeForce GTX 660] (rev a1).

This only seems to happen if the computer has been suspended for a reasonably long time; anything less than a quarter of an hour will usually resume successfully.
nvidia-bug-report.7z (51.1 KB)

No problems on 9500GT and 9800GT.
I guess i’ll stay with them and my next GPU will be an ATI with opensource drivers as soon as they implement Power managment, seriously.

I’d test this problem with the 8800GT that I have but I’m too lazy to swap GPUs ;-) - but something tells me this problem concerns only GTX 660.

This bug is probably resolved in 319.32 (edit: it’s not) - the first resume has been successful, I’ll recheck it in the next few days.

No, this bug is not fixed.

[  3053.800] (WW) NVIDIA(0): WAIT (2, 8, 0x8000, 0x00001ce8, 0x00003dd4)
[  3060.800] (WW) NVIDIA(0): WAIT (1, 8, 0x8000, 0x00001ce8, 0x00003dd4)

Jun 28 00:02:27 localhost kernel: [ 3041.592980] NVRM: GPU at 0000:01:00: GPU-136382c0-06fa-2c0f-977a-4f04b1755070
Jun 28 00:02:27 localhost kernel: [ 3041.592984] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000088 0100cb4f 00000007 00000000
Jun 28 00:02:27 localhost kernel: [ 3041.593018] NVRM: Xid (0000:01:00): 56, CMDre 00000000 0000008c 00000000 00000005 0000102b
Jun 28 00:02:28 localhost kernel: [ 3042.591107] NVRM: Xid (0000:01:00): 31, Ch 00000000, engmask 00000101, intr 10000000
Jun 28 00:02:30 localhost kernel: [ 3044.591920] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00001005

Beta 325.08 is also faulty.

NVIDIA drivers 313.30 are also faulty. Damn, what’s going on?

[  5045.039] (II) NVIDIA(0): Setting mode "DFP-0:nvidia-auto-select"
[  5051.042] (WW) NVIDIA(0): WAIT (2, 8, 0x8000, 0x000051e8, 0x000072d4)
[  5058.042] (WW) NVIDIA(0): WAIT (1, 8, 0x8000, 0x000051e8, 0x000072d4)

Jul  5 13:46:08 localhost kernel: [ 5037.337229] Restarting tasks ... done.
Jul  5 13:46:08 localhost kernel: [ 5037.393659] NVRM: GPU at 0000:01:00: GPU-136382c0-06fa-2c0f-977a-4f04b1755070
Jul  5 13:46:08 localhost kernel: [ 5037.393663] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000088 0100cc5f 00000004 00000008
Jul  5 13:46:08 localhost kernel: [ 5037.393696] NVRM: Xid (0000:01:00): 56, CMDre 00000000 0000008c 00000000 00000005 0000102b
Jul  5 13:46:09 localhost kernel: [ 5038.391555] NVRM: Xid (0000:01:00): 31, Ch 00000000, engmask 00000101, intr 10000000
Jul  5 13:46:11 localhost kernel: [ 5040.393052] NVRM: Xid (0000:01:00): 56, CMDre 00000000 00000080 00000000 00000005 00001005
Jul  5 13:46:29 localhost kernel: [ 5057.439806] SysRq : Emergency Sync
Jul  5 13:46:29 localhost kernel: [ 5057.442617] Emergency Sync complete
Jul  5 13:46:33 localhost kernel: [ 5061.922726] SysRq : Emergency Remount R/O

That’s is with VGA console - so no framebuffer at all.

Bump.

I have the same problem.

3.9.9-301.fc19.x86_64
NVIDIA Corporation GK107M [GeForce GT 650M Mac Edition]

After a kernel upgrade I am not able to resume from suspend.

Can you post your X.org/dmesg messages?

or just append

sudo nvidia-bug-report.sh

Wow, I’ve found a person with the exact same problem.

His report is here.

I would love to, but I have to find out how get into my machine after failed resume.
I have no idea where to find SysRq+S/U/B key on my Mac keyboard :).

slune,

Take a look at this:

http://bobcopeland.com/blog/2008/10/sysrq-on-macbook/
http://ubuntuforums.org/showthread.php?t=762665
http://0pointer.de/lennart/projects/keyfuzz/

How did you get those logs. I am just not able to get into the machine and after hard reset no signs of an error in X logs or dmsg or … .

Btw I am always not able to resume.

Have you set up SysRq? It’s one way of dumping logs - you could also try serial console, but it’s too much hassle.

I use nVidia driver version 319.32 and kernel 3.7.10-1.16-desktop on openSUSE 12.3. No problems here, I can resume from suspend to RAM and from suspend to DISK. Though, openSUSE uses pm-utils / suspend packages and not the kernel ACPI code for suspending.

Bump.

I got sysreq working and still have the same issue - nothing in logs. I am able to reboot with alt+sysreq+b. Not sure if it is somehow connected with a full disk encryption.

I will try an older kernel (3.8.X) and we will see.

BTW with Nouveau suspend works, but battery life is terrible and ntb is overheating.

nvidia-bug-report.log.gz (53 KB)

You first need to SysRq + S (Sync) and SysRq + U (remount read only) - this way your system will be able to physically put log messages to the disk.

Yes I tried that - No luck so far. I tried this sequence: REISUB. The same.
Have to investigate it more.

Thanks for help!

It must be Sync then Umount ;-)

An even better usage is https://en.wikipedia.org/wiki/Magic_SysRq_key#Uses

“Raising Elephants Is So Utterly Boring”, “Reboot Even If System Utterly Broken” ;-)