system freezes after upgrading kernel
 

I had to upgrade to 2.6.10 (from /testing) so that I could get my new (to me) graphics card working.

It is now working, and most of the things that the upgrade broke are fixed.

However this is the third time that I have started this thread, because each of the other two time the system has frozen. A complete lock up which required a reset.

These freezes are apparently random, and on looking in the logs and dmesg appear to have no cause.

Am I just looking in the wrong place or is there a problem with 2.6.10

Samac

well....can you give your hardware-setup. ?
if i understand correct, this occured after installing a " new " graph. card. ?
if it's pci, you could try another slot.
otherwise reseat the card. ( might be just a bad contact. ).

egag

Post your video card specs, if it is agp, maybe the lock up is due to a bad setting in bios or something wrong with agp module/video card driver, it may be a RAM problem too...

What driver do you use for your card in /etc/X11/xorg.conf ?

I am replying from my backup system Slackware 10.1 (2.4.29) because the 2.6.10 was freezing too often to even get this reply typed.

2.4.29 seems stable therefore I would suggest that the problem is not hardware related, but rather configuration related.

I created a script that gives tons of information and writes it all to a file called si.txt (1300+ lines), but then realised that you could not post attachments.

So I have posted it to

www.samac.pwp.blueyonder.co.uk/si.html

www.samac.pwp.blueyonder.co.uk/nvidia-bug-report.html

A thought that I had, was to run an upgrade using:

upgradepkg --reinstall --install-new /root/packages/*/*.tgz

just to see if that would help (but with my luck it wouldn't)

Or perhaps rebuilding the kernel modules (see previous line)

Samac

ah, if you're sure it's not a hw-prob. then the graph-driver could be it.
still don't know what hw. you use....? ( and what driver.. )

egag

See edit, I have added a massive nvidia bug report file.

UPDATE

Kernel version 2.6.10 freezes with both "nvidia" driver and "nv" driver, whereas 2.4.29 is stable with "nv" driver.

Which suggests to me that the graphics card/driver combination are relatively stable and it is perhaps an interaction between them and the kernel, or in fact some other drivers and the kernel that is causing the problem.

I had a ssh conection between the Zaurus and the main machine during a crash to see if it was just the keyboard and mouse that were dead, and also to poke around to see if there were any clues, but the ssh terminal was frozen too, so it seems to be a system wide failure, not just X or the input devices.

Samac

For the nvidia driver :
Did you try to comment your "VideoRam 262144" line in xorg.conf ?

Maybe an agp problem, try put that line, just below ' Driver "nvidia" ' :
Option "NvAGP" "1"

Then try :
/sbin/rmmod sis_agp
/sbin/rmmod agpgart
startx

I have modified xorg.conf, but am having big troubles removing the modules.

rmmod will not remove them as they are in use, and modifying /etc/rc.modules had no effect.

Where are they being loaded from?

Samac

First exit X session, then rmmod nvidia (if present), rmmod agpgart and sis_agp (maybe try rmmod sis_agp before), then modprobe nvidia and startx

Hi,

Did that but the modules still would not remove. Eventually tried "rmmod -f" this temporarily disabled the modules.

insmod nvidia didn't work, yet modprobe nvidia did, but that appears just to have put sis_agp and agpgart back in.

How can you remove modules that are being used?

Is there a file that loads these modules at start, then I can modify it and reboot.

Samac

If you're using hotplug (more than likelly with 2.6.10 kernel), tou could add sis_agp and agpgart in /etc/hotplug/blacklist to prevent them to load at boot

OK that worked, now we will have to wait and see if the system is stable, I will update you.

Many thanks

Samac

Three minutes later ...........


So now I'm back in 2.4.29, could I be better going back to my old graphics card and forgetting about "The joy of Doom3".

No I will not give up.

Any other suggestions keefaz

At least you know it is not an agp problem. You can revert to default, comment the Agp line in xorg.conf and delete sis_agp and agpgart in /etc/hotplug/blacklist

I see you enabled Linux Plug and Play Support from your log, maybe try a kernel compilation without this option.

Make sure you did not enabled APM in config as you use ACPI

I see some "pciehp" error in your log, try put pciehp in /etc/hotplug/blacklist

OK pciehp is black listed, and all the error messages have disappeared.

Good to know the graphics system seems OK.

APM is compiled into the kernel, but it doesn't seem to be getting used anywhere. If the problems continue I will remove APM and Plug and Play and attempt a recompile (even though my track record at this is appalling, thats why I'm using the standard kernel from /testing).

I will keep you informed, but things may be improving as I have almost finished this reply without a freeze.

In fact I have finished

samac

Still freezing.

I will try a kernel re-compile at some point, probably on Monday afternoon.

Samac

I have recompiled a new kernel from source 2.6.10 and all the problems seem to have gone. Many thanks for all your help, especially KEEFAZ.

Samac