Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 04-21-04, 12:52 AM   #1
barrettabel
Registered User
 
Join Date: Jan 2004
Posts: 8
Default X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

Hello All,

I've been having some grief with my Nvidia card - a Asus FX 5600..
It runs well until some random interval X will lock with the last image on the screen and become completely unreponsive to input (ctrl bckspace / mouse /etc). I ssh in from another computer and find X pegging a cpu (99.9% of one cpu's time) - the only resolution is a reboot or killing X with a SIGTERM.

This occurs randomly, I can have xterm and vi running or a Sybyl (Molecular modeling package -heavy 3d). - makes no difference. The system runs perfectly fine with the opensource "nv" driver or another video card (Ati FireGL). Also - swapped out the video card for an identical model to no avail.

The system is a dual 1.6ghz Opteron with 2 gigs of ram, runnning Suse 9 AMD64 (64 bit version) and for diagnostic purposes I also tried the 32bit version with the same results.

- Also tried pulling out 1gb of memory / tests fine ( we have a memory tester)

The motherboard is a Tyan S2875 running the most current bios, 1.03.
I'm driving two CRT's (and for testing purposes, tried a single) via TwinView.
Bios is the latest on the video card as well.

I've tried every combination of NvAGP "0-3", tweaked bios settings, and tried passing different combinations of arguments to the kernel (suggestions when I googled around) - i.e. apm=off noapic iommu=noagp pci=noacpi mem=nopentium.

Work in a Molecular modeling lab, and were going to get 15 Quadro FX2000/FX3000, but that may be a mistake ( anyone else work with these?)


More details below -Sorry for being verbose, wanted to get all the info out

I have the driver latest driver installed, here is some info

cat /proc/driver/nvidia/agp/card
Fast Writes: Supported
SBA: Supported
AGP Rates: 8x 4x
Registers: 0x1f000e1b:0x00000000

Host Bridge: Advanced Micro Devices [AMD] AMD-8151 System Controller
Fast Writes: Supported
SBA: Supported
AGP Rates: 8x 4x
Registers: 0x1f000b7b:0x00000200

Model: GeForce FX 5600
IRQ: 16
Video BIOS: ??.??.??.??.??
Card Type: AGP

NVRM version: NVIDIA Linux x86 NVIDIA Kernel Module 1.0-5336 Wed Jan 14 18:29:26 PST 2004
GCC version: gcc version 3.3.1 (SuSE Linux)


** Gcc version
gcc -v
Reading specs from /usr/lib/gcc-lib/i586-suse-linux/3.3.1/specs
Configured with: ../configure --enable-threads=posix --prefix=/usr --with-local-prefix=/usr/local --infodir=/usr/share/info --mandir=/usr/share/man --libdir=/usr/lib --enable-languages=c,c++,f77,objc,java,ada --disable-checking --enable-libgcj --with-gxx-include-dir=/usr/include/g++ --with-slibdir=/lib --with-system-zlib --enable-shared --enable-__cxa_atexit i586-suse-linux
Thread model: posix
gcc version 3.3.1 (SuSE Linux)


** Kernel ( on the 32bit Suse)
uname -a
Linux polymerase 2.4.21-202-smp4G #1 SMP Fri Apr 2 21:32:50 UTC 2004 i686 athlon i386 GNU/Linux

Any ideas?

Thanks
Barrett
Attached Files
File Type: txt Xlog.txt (45.4 KB, 90 views)
File Type: txt XF86Config.txt (7.0 KB, 118 views)
barrettabel is offline   Reply With Quote
Old 04-21-04, 03:24 AM   #2
whig
550Ti
 
Join Date: Jan 2004
Location: New Zealand
Posts: 854
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

Try removing those sundry devices in your XF86Config, eg:

InputDevices "/dev/ttyS8"
whig is offline   Reply With Quote
Old 04-21-04, 11:50 AM   #3
barrettabel
Registered User
 
Join Date: Jan 2004
Posts: 8
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

Thanks,

I made the changes ( and cleaned up the XF86Config file ) - it just locked up as usual- any other ideas?

I appreciate your input ..

Bear
barrettabel is offline   Reply With Quote
Old 04-21-04, 12:34 PM   #4
Corona688
Registered User
 
Join Date: Feb 2004
Posts: 93
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

I've had the exact same problem for the last year or so, with widely different cards(Geforce2MX, GeforceFX), kernels(2.4.x, 2.6.x), systems(single duron, dual opteron), and distributions(mandrake, gentoo). It must be a driver issue of some sort.

Once again, I'm obviously not the only one waiting with bated breath for the next amd64 driver release
Corona688 is offline   Reply With Quote
Old 04-21-04, 12:44 PM   #5
chroma
Registered User
 
Join Date: Apr 2004
Posts: 4
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

I seem to have a very similar problem on similar hardware:

2x2200+ Athlon MPs
Tyan Tiger MPX (S2466N-4M)
Abit Siluro Geforce Ti4200
1.5GB RAM (3x512MB ECC PC2100)
Debian Sid, 2.6.4 kernel

At seemingly random intervals X will lock up crash and I'll get a pattern of red or blue largeish pixels on a white background. The crashes are basically random, although I've never been up for more than 5 or 6 days. If I ssh in to the box after it has crashed, I find X is taking 100% CPU. If I attach gdb to the X process and grab a backtrace, I find that X is in an infinite loop over gettimeofday() in the getTimeInMillis() function. What happens when you do the same thing, barrettabel?

The problem also seems to be unaffected by load: it has even happened while the box is unattended a few times. Heavy load, graphical or otherwise, does not seem to make a difference. The only correlation I can think of is that the crashes seem slightly more frequent when I'm using rdesktop and listening to internet radio. That is certainly not the only time the thing crashes though.

I've swapped the Ti4200 for an old Matrox G400 and I have yet to experience a crash, so it seems to be either the nvidia driver or the card itself. FWIW, I've had this problem since the 2.4.18 kernel.

Please let me know if you (or anyone else) figures this one out.


chroma
chroma is offline   Reply With Quote
Old 04-21-04, 11:29 PM   #6
chroma
Registered User
 
Join Date: Apr 2004
Posts: 4
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

The plot thickens... I had put my Ti4200 in another computer (Soltek SL-NV400-64, 2100+ XP) running a Debian stock 2.6.5 kernel. I set it up to use the nvidia driver, and lo and behold, ten minutes after my significant other logs in, she gets the same crash as above: red dots, 100% CPU usage and X looping over gettimeofday()...

So, it definitely is the card/driver. I switched X over to the nv driver, so we'll see how that goes.


chroma

Last edited by chroma; 04-21-04 at 11:46 PM.
chroma is offline   Reply With Quote
Old 04-22-04, 12:07 AM   #7
barrettabel
Registered User
 
Join Date: Jan 2004
Posts: 8
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

> The crashes are basically random, although I've never been up for more than 5 or 6 days.
I've never been able to make it more than a day and a half.

>If I ssh in to the box after it has crashed, I find X is taking 100% CPU. If I attach gdb to the X process and grab a backtrace, I find that X is in an infinite loop over gettimeofday() in the getTimeInMillis() function. What happens when you do the same thing, barrettabel?

My notes are at work - I'll post details when I get there.

>The problem also seems to be unaffected by load: it has even happened while the box is unattended a few times. Heavy load, graphical or otherwise, does not seem to make a difference. The only correlation I can think of is that the crashes seem slightly more frequent when I'm using rdesktop and listening to internet radio. That is certainly not the only time the thing crashes though.

I'm an Internet radio junkie too, but I haven't seen that make any difference. The system is generally pegged most of the day running simulations, but it will exhibit this behavior at idle too.

I've found the same behavior with a Fx5200/5900/ and a different 5600. I'm making the wild guess it is the drivers. I've tried the 32 bit drivers ( on a 32bit os) and the amd64 (on the 64bit os) with the same results.

Has anyone else posed this question to linux-bugs@nvidia.com? I'd appreciate anyone else chiming in if they have had a similar problem.


chroma, my guess is the nv drivers will work great ( at the expense of 3d accel) as long as you don't have hardware problems.

Thanks all,
barrettabel is offline   Reply With Quote
Old 04-22-04, 12:44 AM   #8
chroma
Registered User
 
Join Date: Apr 2004
Posts: 4
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

AFAIK the nv drivers don't do twinview either, right? On the other box it's not a big deal if the secondary 15" monitor doesn't work, but the G400 I've got in my main box won't do 1600x1200 on the second head (a 19" P95f+). I've been spoiled by dual head so I can't live without it now . If these problems persist, it looks like I'm gonna have to go shopping for a non-nvidia card with an OSS driver that supports 2x1600x1200 dual head . It's a shame because my machine flies with the nvidia drivers...


chroma
chroma is offline   Reply With Quote

Old 04-22-04, 01:14 AM   #9
barrettabel
Registered User
 
Join Date: Jan 2004
Posts: 8
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

AFAIK the nv drivers don't do twinview either, right?

No they don't (somebody correct me if i'm wrong)

I've been spoiled by dual head so I can't live without it now .


I agree, one you go dual, you never go back.

If these problems persist, it looks like I'm gonna have to go shopping for a non-nvidia card with an OSS driver that supports 2x1600x1200 dual head . It's a shame because my machine flies with the nvidia drivers...


Matrox's P650 and P750 do a Dual 1600x1200 and I know they are fairly well supported under linux - My g450 (similar to your g400 I assume) has oss drivers that do dualhead. Rock solid, just a little slow on the 3d.

The only other card I know is an Ati 9200 - it has opensouce 3d accell. drivers (not sure about the dualhead) Otherwise Ati's idea of driver support in the amd64 is the middle finger. Nvidia is the only true 3d player in the Amd64 realm now

If worse comes to worse you can always put in another (shudder) pci video card - works well.
barrettabel is offline   Reply With Quote
Old 04-29-04, 07:21 AM   #10
rednuht
GeForce FX5600 (1.0-5336)
 
rednuht's Avatar
 
Join Date: Sep 2003
Location: uk
Posts: 164
Question Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

>>If I ssh in to the box after it has crashed, I find X is taking 100% CPU. If I >>attach gdb to the X process and grab a backtrace, I find that X is in an >>infinite loop over gettimeofday() in the getTimeInMillis() function.

what steps are need to get these results ?
what are prerequesits for doing a backtrace on X ?
Is GDB interactive allowing you to skip out of the infinite loop ?
rednuht is offline   Reply With Quote
Old 04-30-04, 02:21 AM   #11
chroma
Registered User
 
Join Date: Apr 2004
Posts: 4
Default Re: X lockup on Dual Opteron / Suse 9 AMD 64 / FX 5600

Quote:
Originally Posted by rednuht
what steps are need to get these results ?
what are prerequesits for doing a backtrace on X ?
Is GDB interactive allowing you to skip out of the infinite loop ?
$ ssh [ b0rked machine ]
$ su
Password: [ type root password ]
# ps -A # note XFree86's pid
# gdb --pid=[XFree86 pid]
(gdb) bt # print backtrace

....

If you're lucky X won't totally lock up and you should be able to type 'exit' and detach from the X process. You might even be able to kill the X process (kill -9 [pid]) and then shut the machine down.


chroma
chroma is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
AMD Gooses the Clocks on 'Bulldozer' Opterons News Archived News Items 0 06-04-12 02:24 PM
AMD vs Intel druga runda CPUs, Motherboards And Memory 7 08-20-02 01:09 PM
Suse 8.0 Dual display on Quadro 4 700XGL TheCowStir NVIDIA Linux 3 08-10-02 12:16 PM

All times are GMT -5. The time now is 08:01 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright ©1998 - 2014, nV News.