X is still locked even when the new 1.0.4496 driver is used

[jlin]

Hi everybody,


I have been expecting this new driver for many weeks since some people reported that the older driver 1.0.4369 will cause X being locked for GeForce FX 5200 and I unfortunately have the same graphics card. The sympton is similar to what have been reported: the X terminal simply gets locked after some time, probably a few minutes. I am not sure what is the real cause. If I use "top" to check the CPU loading, I simply see the process X is taking 99.9% of the CPU load (I could still remotely login from other machines and execute some commands).

Could anyone be so kind to tell me what I can do to fix the bug? Are there some procedures that I can use to check what has been happening? Sorry if my description is not clear enough, but I will supply more if you ask me. I will greatly appreciate any hint and help.

[jaymz]

first guess would be you are using the agp support in your XF86Config-4 file. can you post it to the forum?

[pe1chl]

I have a 5200 chipset card and it works fine with that driver...

[jlin]

Hi


Thanks for your replies. My XF86Config, which cause the lock, is as attached. I therefore modify the driver name to "nv", instead of "nvidia" to keep it alive. However, the glx module will not be found:

(EE) [GLX]: Failed to add GLX extension (NVIDIA XFree86 driver not found)


(from /var/log/XFree86.0.log),

which is expected.

Could you please tell me how you make it work with nvidia driver? I need GLX because I have many OpenGL applications.

Thanks!

[jaymz]

Originally posted by jlin
My XF86Config, which cause the lock, is as attached.

I don't see the attachment, could you try to re-post? thanks.

[jlin]

My XF86Config is reattached. I didn't notice that this board only accept files with special extensions.


Thanks!

[jaymz]

First thing i would do is change the "device" section:


Section "Device"
Identifier "Videocard0"
Driver "nvidia"
VendorName "Videocard vendor"
BoardName "NVIDIA GeForce FX (generic)"
Option "NvAGP" "1"
EndSection

[jaymz]

If that doesn't help, please attach a copy of your last XFree86.0.log

[jlin]

Many thanks to jaymz for the suggestion. I did the modification to XF86Config but the sympton appears to be very similar. It seems that I stepped on some sort of "land mine" when I took a certain actions that I cannot precisely identify.
For example, I found that the X got locked immediately after I lunch an molecular display program, molmol, which requires OpenGL. Also, I found the X became locked when I opened the web site of NVidia using mozilla, but it can survive longer when I browsed other pages.

Attached is the XFree86.0.log that I saved after the crash due to launching of the OpenGL molecular display program.

Thanks!

[flea]

Hello,

I get X blocked (99% of proc time) ANY time I scroll an OpenOffice document page with my wheel mouse. Screen is freezed, mouse pointer moves right as should do with X, keyboard seems freezed also, as keyboard lights don't toggle when i press caps lock.

kill -HUP won't do, I need to kill X (logged in from another computer) then gdm greeter restarts things. XFree log doesn't say nothing interesting, I'm using rh 9 (not customized) with GeForce2 MX and plain regular configuration. Problem encountered with 4496 and 4363, not with previous driver versions (but with less recent drivers I had rh 8.0, glibc 2.3 issues?)

Any suggestions?

[cbk]

I have the same problem with X being unstable. I have tryed instlling the drivers but I still get the same error in my XFree log.
Thank you jaymz for your help. after making the changes as per your post, X seams to last longer before it locks up and restarts. I have noticed that I don't have the /dev/nvidia* file on my system after the install. Anyone got a idea on how to fix this and why it went wrong on my system. I'm running 2.4.21-0.13mdk, Mandrake 9.1, On a Shuttle XPC SN41G2, AMD XPC 2500+, nfoce2 chipset.

[johannesprix]

Same with my nforce2 on my SuSE 8.0 based system, (see the other thread I created for detailes machine specs, config and log files, bios version, hardware specs).

I'd be most grateful if someone could finally eliminate this bug. Please tell me if you need any other information on versions and the like.

Sincerly, Johannes.

UPDATE: After enabling option "NvAGP" "0" the problem seems to have gone away. That's good, cause I want to write an OpenGL based ap, and without OpenGL, that's kind'a hard

See ya, jp.