Xorg hang with 100% cpu, any driver version, GTX275

[OliW]

Originally Posted by Licaon

@bones_was_here: he had a different (solved) problem: http://www.nvnews.net/vbulletin/showthread.php?t=159771

@Licaon: Check the dates. And that wasn't even a problem that was just my idiocy in thinking I could get away with a low powered PSU.

[Licaon]

Originally Posted by OliW

@Licaon: Check the dates. And that wasn't even a problem that was just my idiocy in thinking I could get away with a low powered PSU.

oh, ok, so can you read http://www.nvnews.net/vbulletin/showthread.php?t=46678 and generate a good report with the debug ( -logverbose 6 ) info as instructed after the issue appears ?

[bones_was_here]

Tried upgrading this card from VBIOS 62.00.60.00.01 to 62.00.60.00.70 with one of these http://www.techpowerup.com/vgabios/i...ace=&memSize=0

Needed to change the PCI Device ID and PCI Vendor ID with NiBiTor to get a clean flash with no warnings. The flash was successful and the card runs as normal, displaying the new VBIOS version in nvidia-settings.

However, I can still reproduce the X hang.

Does anyone know the whereabouts of a changelog for NVIDIA VBIOS?

[OliW]

bones_was_here Are you running some variant of "the magic 200 line kernel responsiveness patch", be that the kernel patch itself or its bash equivalent.

Like this: http://www.webupd8.org/2010/11/alter...nel-patch.html

I have been and I've just stopped and I've observed that a lot of my freezes have gone. They still happen occasionally, just a lot less often.

[bones_was_here]

Nope, vanilla 2.6.37 here (EDIT but it happens also with 2.6.31 which was compiled way before that patch was written)

[bones_was_here]

Surely this must be solvable, or at least a logical explanation found.
At risk of repeating myself: I can reproduce it using freely downloadable software and am happy to provide any information or do any testing that might help, but I am not a programmer and don't really know how to proceed further with debugging this.

I've been messing with the card's clock speeds, certain settings increase the chance of it happening, but none so far completely prevent it.

[377684.447] [mi] EQ overflowing. The server is probably stuck in an infinite loop.
[377684.447] 
Backtrace:
[377684.467] 0: /usr/bin/X (xorg_backtrace+0x28) [0x45ceb8]
[377684.467] 1: /usr/bin/X (mieqEnqueue+0x1f4) [0x45cd34]
[377684.467] 2: /usr/bin/X (xf86PostMotionEventP+0xc4) [0x46ddb4]
[377684.467] 3: /usr/lib/xorg/modules/input/evdev_drv.so (0x7f1a5fc4e000+0x52fc) [0x7f1a5fc532fc]
[377684.467] 4: /usr/bin/X (0x400000+0x74277) [0x474277]
[377684.467] 5: /usr/bin/X (0x400000+0x1046f3) [0x5046f3]
[377684.467] 6: /lib/libpthread.so.0 (0x7f1a663db000+0xef60) [0x7f1a663e9f60]
[377684.467] 7: /lib/libpthread.so.0 (0x7f1a663db000+0xffff80e599225177) [0xffffffffff600177]
[377685.470] (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000d944, 0x0000d944)
[377686.128] [mi] EQ overflowing. The server is probably stuck in an infinite loop.
[377686.128] 
Backtrace:
[377686.128] 0: /usr/bin/X (xorg_backtrace+0x28) [0x45ceb8]
[377686.165] 1: /usr/bin/X (mieqEnqueue+0x1f4) [0x45cd34]
[377686.165] 2: /usr/bin/X (xf86PostMotionEventP+0xc4) [0x46ddb4]
[377686.165] 3: /usr/lib/xorg/modules/input/evdev_drv.so (0x7f1a5fc4e000+0x52fc) [0x7f1a5fc532fc]
[377686.165] 4: /usr/bin/X (0x400000+0x74277) [0x474277]
[377686.165] 5: /usr/bin/X (0x400000+0x1046f3) [0x5046f3]
[377686.165] 6: /lib/libpthread.so.0 (0x7f1a663db000+0xef60) [0x7f1a663e9f60]
[377686.165] 7: /lib/libpthread.so.0 (0x7f1a663db000+0xffff80e599225177) [0xffffffffff600177]
[377688.472] (WW) NVIDIA(0): WAIT (2, 6, 0x8000, 0x0000e1a4, 0x0000e360)
[377695.472] (WW) NVIDIA(0): WAIT (1, 6, 0x8000, 0x0000e1a4, 0x0000e360)
[377703.534] (WW) NVIDIA(0): WAIT (2, 6, 0x8000, 0x0000e1a4, 0x0000e370)
[377710.534] (WW) NVIDIA(0): WAIT (1, 6, 0x8000, 0x0000e1a4, 0x0000e370)

[Licaon]

Originally Posted by bones_was_here

Surely this must be solvable, or at least a logical explanation found.

Try another USB keyboard and/or mouse ( but not one from Logitech ) see this post: http://www.nvnews.net/vbulletin/show...04#post2398504 as you have a Logitech USB device and you Xorg.0.log shows [mi] overflow errors

[bones_was_here]

Thanks for the link, but I've had this problem longer than I've had the keyboard. Previously I had a saitek keyboard and a m$ mouse, but alt+tab from Nexuiz caused a hang at least as often then as now. I will probably try the QUEUE_SIZE hack at some point when I have more time.

[Unixnut]

Has anyone ever found a solution to this?

Up until recently my Linux system was working fine, CUDA processing and dual screen displays with a nvidia GTX 280. While it all worked fine the OS was rather old (first installed 2006) and was running old kernel/libraries/nvidia-driver/etc... so I decided it was time to reinstall the OS.

Turns out this was a big mistake. Since reinstalling the OS I've been unable to get the card to work with the nvidia driver.

The card works with the open source nouveau driver, but with no GL support, no acceleration and no CUDA (logically). This coupled with the fact it was working fine leads me to believe that it isn't a hardware fault (what kind of hardware fault would occur due to a new OS/driver? Yet work with a basic driver?).

As mentioned in this thread. When I start X the screen goes blank and X hangs at 100% CPU forever. As I'm not sure if this is due to the same issue I decided to post here (I can post a separate thread in future if this is unrelated).

When I run, X:


dmesg shows the following:

NVRM: loading NVIDIA UNIX x86_64 Kernel Module 295.49 Mon Apr 30 23:46:33 PDT 2012
NVRM: Xid (0000:07:00): 56, CMDre 00000000 00000088 0100cb05 00000004 00000084
NVRM: Xid (0000:07:00): 56, CMDre 00000000 0000008c 00000000 00000005 00000008
NVRM: Xid (0000:07:00): 56, CMDre 00000000 00000088 0100cb0b 00000004 00000084
NVRM: Xid (0000:07:00): 56, CMDre 00000000 0000008c 00000000 00000005 00000008
NVRM: Xid (0000:07:00): 56, CMDre 00000000 00000088 0100cb11 00000004 00000084
NVRM: Xid (0000:07:00): 56, CMDre 00000000 0000008c 00000000 00000005 00000008
NVRM: Xid (0000:07:00): 56, CMDre 00000000 00000080 00000000 00000005 00000008
NVRM: Xid (0000:07:00): 6, PE0001
NVRM: Xid (0000:07:00): 6, PE0001

After killing X:
NVRM: Xid (0000:07:00): 6, PE007e
NVRM: Xid (0000:07:00): 6, PE007e

I have attached the Xorg log file, and the results of nvidia-bug-report.

I've run out of ideas as to what it can be . I've updated the BIOS, tried both the latest 3.X kernels, downgraded to 2.6.XX kernels, tried multiple different nvidia drivers, and still the same problems. I have gone back to wishing I never reinstalled the OS to begin with :/

Despite this the machine is stable, I can remotely SSH in and use it. CTRL-C'ing X org doesn't work, only a kill -9 will do it. Occasionally the entire machine can hang, needing a reset, but that is uncommon.

I will include the last few lines of strace running of X, not sure if it is useful, but can include the whole thing if needed.

strace output:

mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fa6931aa000
read(17, "EnableVia4x: 0\nEnableALiAGP: 0\nN"..., 1024) = 456
close(17) = 0
munmap(0x7fa6931aa000, 4096) = 0
stat("/dev/nvidia0", {st_mode=S_IFCHR|0666, st_rdev=makedev(195, 0), ...}) = 0
open("/dev/nvidia0", O_RDWR) = 17
ioctl(10, 0xc048464d, 0x7fff8355e210) = 0
open("/proc/interrupts", O_RDONLY) = 20
fstat(20, {st_mode=S_IFREG|0444, st_size=0, ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fa6931aa000
read(20, " CPU0 CPU1 "..., 1024) = 1024
read(20, " 0 Performance pending work"..., 1024) = 437
read(20, "", 1024) = 0
read(20, "", 1024) = 0
close(20) = 0
munmap(0x7fa6931aa000, 4096) = 0
ioctl(17, 0xc01446ce, 0x7fff8355e520) = 0
ioctl(10, 0xc0144628, 0x7fff8355e560) = 0
ioctl(16, 0xc020462b, 0x7fff8355e4e0) = 0
ioctl(10, 0xc020462b, 0x7fff8355e550) = 0
ioctl(10, 0xc030464e, 0x7fff8355e530) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, 13, 0xfd647000) = 0x7fa6931aa000




Any help sorely appreciated! I've been trying to get this work for weeks!! The card has never been OC'ed or otherwise run beyond its normal parameters. Temps were avg 30-40 degrees throughout.