GeForce FX 5200 Repeatable Lockups w/4363

[wrex]

I'm also having regular lockups with the 4363 driver (screen freezes, X consumes 100% of the cpu, framebuffer virtual consoles garbled).

I seem to get the lockup eventually no matter what I'm running, but the "pyro" eye candy in xscreensaver causes a lockup within 5 or 10 minutes every time without fail.

I've tried everything others have reported in this forum. I've totally disabled agpart in my kernel, tried disabling almost everything in my XF86Config, and tried all the various kernel boot options others have suggested.

No joy.

Hardware:
Soyo Dragon Ultra KT333 mobo
Athlon-XP 1900+
512 MB CL2 memory (tried CL2.5 still no joy)
PNY Verto GeForce FX5200 PCI
(twinview crt)

I'm running gentoo linux with the "gentoo-sources-2.4.20-r5" kernel (kernel config attached). Most everything was compiled with "-march=athlon-xp -O3 -pipe".

rex@megrims rex $ cat /proc/driver/nvidia/cards/0
Model: GeForce FX 5200
IRQ: 11
Video BIOS: 04.34.20.15.00
Card Type: PCI

rex@megrims rex $ cat /proc/cmdline
root=/dev/hda5 noapic pci=noacpi pci=biosirq mem=nopentium


My latest XF86Config and XFree86 log are attached.

Any help or suggestions will be very much appreciated.

[wrex]

Just tried the 4349 driver and still got a lockup within just a few minutes of starting pyro.

[LoOoD]

I also got a dual amd box w/ a AGP version of the same video card. I also had lock ups.
Things I did:

In bios set to 1.1 MP specs
use noapic option at boot. (not sure if this disables smp or not, /proc/interrupts looks funny)

That helped alot for me but I do still get lockups once in a while.. just less of them..

[wrex]

Originally posted by LoOoD
In bios set to 1.1 MP specs
use noapic option at boot. (not sure if this disables smp or not, /proc/interrupts looks funny)


I'm on a single processor system, and I've already got "noapic" in my kernel options (see "cat /proc/cmdline" above).

Last night I also tried compiling and running a 2.4.21 kernel, but no change.


That helped alot for me but I do still get lockups once in a while.. just less of them..

That sounds like my current behavior. I haven't had a lockup yet when I'm just typing in xterms. It always happens when I'm dragging windows around or something similar.

Unfortunately even occasional lockups are totally unacceptable to me -- I'm turning this into my primary workstation. I'll have to get another display adapter if I can't even come up with a workaround.

The one thing that has very repeatably demonstrated the problem for me is running "pyro" (I suspect other screen hacks might exhibit the bug as well, but I haven't verified).

LoOod, if you have xscreensaver installed, could you please try running "xscreensaver-demo" and previewing the "pyro" screen hack for a few minutes. This invariably causes a lockup on my machine within a few minutes. Reproducing a problem is often half the battle -- if this causes the problem on your system as well we'll at least have a test mechanism for potential fixes.

Thanks.

[wrex]

Well, I'm embarassed to admit that my problem appears to have been due to a bad bios setting all along.

I'm still not sure exactly *which* bios setting was the problem though. . . .

I selected the "failsafe defaults" option in the bios settings screen, got rid of all of my kernel arguments ("noapic", pci="noacpi", etc.) and set my XF86Config back to its defaults (NvAGP=3, etc.). No more lockups -- hooray!

The odd thing is that I then selected the "performance defaults" and then one-by-one brought all of the settings back to where I thought they were before, testing with pyro each step of the way, but I haven't been able to reproduce the lockups any more.

I'm forced to conclude that I must have somehow been unintentionally overclocking something. (I note that glxgears has now dropped by about 50 to 1880 FPS with my current bios settings -- something must indeed be different).

I wonder how many other folks suffering from freezes/lockups with the 4363 driver are also affected by a bad bios setting? I'm afraid all the apic/acpi/NvAGP/mem=nopentium/modules.conf suggestions in this forum were all a bunch of red herrings in my case.

I wish I had tried the "failsafe defaults" first! I did try slowing down some of the memory settings, etc., but going to failsafe defaults and then creeping back up to my desired (NOT OVERCLOCKED!) settings definitely seems to have done the trick for me.

Hope this saves someone else from the same frustration. I'm now quite happy with twinview on my FX 5200 card with the 4363 driver.

[abject]

Hmph. Well, I still get easily repeatable X lockups with 4349 and my one-week old FX5200 (not ultra). Can't install 4363 as there is no Debian Way ;) yet. All 4 NvAGP configs tried, no joy.

PNY Verto GeForce FX 5200 128M AGP (Dual VGA)
ECS K7s5a-Pro
512Mb
40 Gb 7200 Seagate ATA IV
19" Hitachi CM751

Mixed Debian Sarge/Sid (Sid for KDE and Gnome, mostly). X-Windows-System from Unstable yesterday (5/26) 4.2.1-6, IIRC.

Runs OK in Vesa 10x7@75Hz :mad:

Lost the XFree86.0.log on the last reboot.

Did learn about the -s switch on LILO, tho'

[wrex]

Originally posted by abject
Hmph. Well, I still get easily repeatable X lockups with 4349 and my one-week old FX5200 (not ultra). Can't install 4363 as there is no Debian Way ;) yet. All 4 NvAGP configs tried, no joy.

PNY Verto GeForce FX 5200 128M AGP (Dual VGA)
...

FWIW, that is almost exactly the same card I'm using (purchased at Fry's a few weeks ago) except that mine is PCI and not AGP.

I haven't had a freeze-up since I fiddled with the bios settings.

I'm mostly Debian-ignorant, but I'm not sure I understand why you can't install 4363? There are gentoo ebuilds available for 4363, is something preventing a .deb from being created similarly?

Though I prefer having my OS's package manager control things too, one of the first things I tried when I was having problems was running the verbatim nvidia installer. I went back to using the gentoo portage packages, but for just a handful of object files and libraries (nvidia.o and the glx stuff) I would think it would be preferable to run the latest driver rather than hold out for "package manager purity" (especially if you're having problems).

Good luck!

[Medic6666]

No AGP on my board.


I get the same lock ups when moving windows. And I can guarantee it will lock when using a Web Browser.

I have done all the readme said and my Bios is set correctly.

I have the new drivers installed and still it locks.

any ideas?

Cheers


Medic6666

[LoOoD]

time to try 4699 ..

[Medic6666]

I have these installed now and I can still make the problem happen.

:(

[LoOoD]

cat /proc/interrupts?
lspci -v?

[Medic6666]

VGA Compatible Controller: nVidia Corporation: Unknown Device 0322 (rev a1)

:(


Cat dont work :( :(



Medic6666

[LoOoD]

post the entire output.. and try doing it as root.

/sbin/lspci -v ; cat /proc/interrupts

[Medic6666]

Interrupts says: 9 6304276 XT-PIC nvidia



lspci -v says:
00:00.0 Host bridge: Intel Corp. 82810 GMCH [Graphics Memory Controller Hub] (rev 03)
Subsystem: Intel Corp. 82810 GMCH [Graphics Memory Controller Hub]
Flags: bus master, fast devsel, latency 0

00:1e.0 PCI bridge: Intel Corp. 82801AA PCI Bridge (rev 02) (prog-if 00 [Normal decode])
Flags: bus master, fast devsel, latency 0
Bus: primary=00, secondary=01, subordinate=01, sec-latency=32
I/O behind bridge: 0000b000-0000bfff
Memory behind bridge: ede00000-efefffff
Prefetchable memory behind bridge: ddc00000-edcfffff

00:1f.0 ISA bridge: Intel Corp. 82801AA ISA Bridge (LPC) (rev 02)
Flags: bus master, medium devsel, latency 0

00:1f.1 IDE interface: Intel Corp. 82801AA IDE (rev 02) (prog-if 80 [Master])
Subsystem: Intel Corp. 82801AA IDE
Flags: bus master, medium devsel, latency 0
I/O ports at ffa0 [size=16]

00:1f.5 Multimedia audio controller: Intel Corp. 82801AA AC'97 Audio (rev 02)
Subsystem: Intel Corp.: Unknown device 4001
Flags: bus master, medium devsel, latency 0, IRQ 11
I/O ports at dc00 [size=256]
I/O ports at d800 [size=64]

01:00.0 VGA compatible controller: nVidia Corporation: Unknown device 0322 (rev a1) (prog-if 00 [VGA])
Flags: bus master, 66Mhz, medium devsel, latency 248, IRQ 9
Memory at ee000000 (32-bit, non-prefetchable) [size=16M]
Memory at e0000000 (32-bit, prefetchable) [size=128M]
Expansion ROM at efee0000 [disabled] [size=128K]
Capabilities: [60] Power Management version 2

01:02.0 Ethernet controller: VIA Technologies, Inc. VT86C100A [Rhine] (rev 06)
Subsystem: D-Link System Inc DFE-530TX rev A
Flags: bus master, medium devsel, latency 128, IRQ 10
I/O ports at bc00 [size=128]
Memory at efedff80 (32-bit, non-prefetchable) [size=128]
Expansion ROM at efec0000 [disabled] [size=64K]


Cheers

Medic6666

[Medic6666]

I am still having this trouble.

I have installed the latest drivers.

I have posted some logs here
http://www.mysite123.co.uk/temp/logs.html

and also acouple of screen shots here
http://www.mysite123.co.uk/temp/snaps.html


It only seems to lock when in a web browser. I have ran KOffice, Xchat, gIMP, gaim all at the same time and it all works fine.

I have Galeon, Mozilla, konq and it does the same in all of them?????


help!


:(

Medic6666

[LoOoD]

two video cards on the same IRQ??..

[Medic6666]

The problem of the lockups started happening when I installed my new GFX card. I managed to find a updated bios (which was not on my mobo website) and that let me disable my onboard GFX chip.

This made no difference. I also have the latest drivers from Nvidia and have the XF86Config configured properly.

Then I managed to get a cheap 100meg network card and decided to install this so my whole LAN would be 100meg. Since uninstalling my old network card and installing the new one the problem has stopped.

I am unsure why installing a Nvidia GFX Card and Drivers would affect the network card but this seems the case for me.

I have double checked my logs and there is no conflicts or error for the network card, but as it is now working I am not going to bother trying to work it out :)

Thank for all the help you have given ;)

Medic6666