Mandrake Linux 9.1 lockups

[omas]

Hi!

I have some problems with essentially all NVIDIA Linux drivers and it shows itself when i primarly run Gnome.

Frequent lockups occur when for example i try to run Mozilla browser when running Gnome.

I thought i had gotten rid of the problems i had when i switched to KDE, but yesterday i had the same kind of lockup when trying to run Mozilla browser within KDE.

It started ok, but the minute i tried to rite something in the addressfield, my machine just froze dead.

This seemes to indicate some problem with my hardware, and the only thing i ever have had a problem with is NVIDIA, so... well you be the judge of the relevanse of this post, but my suspicions leans toward my Creative 3DBlaster Geforce 256 DDR.

I dont know exactly how much info is needed but i have attached /var/log/XFree86.0.log, /var/log/messages and a dmesg output from the machine in question.

If anyone could give me a good pointer in wich direction i should look to solve this problem i would be most thankful.


Regards

/Omas Jakobsson

[omas]

I have read Andy Mechams advise regarding wich information i should post for the ease of troubleshooting a problem, so here is some more information.

OS: Mandrake Linux 9.1
Driver version: 1.0-4349
Kernel: 2.4.21-0.13mdk
Processor: AMD Athlon 700mhz

#less /proc/driver/nvidia/cards/0
Model: GeForce DDR
IRQ: 11
Video BIOS: 02.10.02.04.00
Card Type: AGP

Attached is also my copy of /etc/X11/XF86Config-4

I havent had any problems with resolution or color, so i do not attach any verbose XFree86 logs.

My previos post should give the rest of any interesting info that could solve this...

Regards

/Omas Jakobsson

[Nuitari]

I have a similar problem with Slackware 9 but gnome was stable and kde kept crashing until I removed all of the icons on the desktop.

Though konqueror will have it crash very quickly.

This happenned both in twinview and not.
OS: Slackware 9
Driver version: 1.0-4349
Kernel: 2.4.20
Processor: AMD Duron 700
MotherboarD: Abit Kt7a-RAID

nuitari@gandalf:~$ cat /proc/driver/nvidia/cards/0
Model: GeForce4 Ti 4200 with AGP8X
IRQ: 5
Video BIOS: 04.28.20.21.00
Card Type: AGP

[omas]

Any luck with different configurations?

I know there are som Graphical frontends for configuring nvidias drivers, but i havent found anyone i really find trustworthy.

Someone got any tips?

Regards

/Omas Jakobsson

[Andy Mecham]

Do hangs occur with AGP disabled?

--andy

[omas]

To put it simple. No, with AGP disabled it doesnt lock up.

But my fps with glxgears goes from an average of 2238 to 1479fps.

Thats not good.

This is as i said before not a lockup that is 100% reproducable, it seemes to occur after working the machine in question for a while with heavy 3D-stuff and such.

I have had an Windows XP installation on the same machine in paralell to my Mandrake installation for about a year and never experienced any lockups, and i know AGP isnt disabled in that driver, else it would be rather impossible to play any games i currently use Windows XP for.

My motherboard has a Via chipset and i looked around a bit, but couldnt find any leeds on the issue there, however, i managed a year ago to find an bios update and have made certain i run the latest possible bios, so any bios updates isnt likely to fix it.

What is puzzling is that it doesnt say anything in any logs i can find about what can be wrong.

[kristanb]

Omas - I'm having a similar problem, now with both Mandrake 9.0 and 9.1, and XF86 4.2.x and 4.3.0. I've found, though, that I can ssh in to the box, and when I run top it turns out that X is consuming 95+% of the cpu. I can kill X and start up again and everything will be fine for awhile, then it will happen again. I'm not noticing any consistent pattern - sometimes it's during heavy 3d rendering (my kids like Tux Racer), sometimes during more mundane desktop stuff. I went out and got another card today to rule out hardware failure, and got similar symptoms with the new card. Oh, and I get similar problems with drivers back to the 2960 series. The NV driver that comes with XFree86 doesn't cause the problem (yet), but then it also isn't 3d accelerated.

Don't know if that helped, but the more of us Mandrake users that speak up about this the better.

[omas]

Well, im pleased to hear that this problem isnt in my mind... :)

But i think i got worse symptoms than you, my machine simple freezes and doesnt respond to anything at all.

[afrosheen]

I'm having the same problem. I thought it was fixed when I wasn't using Konqueror because I can game and it's fine (with the newest drivers). It locked again today and sure enough, X was consuming all of the cpu power. I ssh'ed in, killed X and restarted kde and it was business as usual.

I've rolled back to the 4191's again and will see how it goes. I don't remember having these lockups before because my system usually goes a few weeks between reboots (it's a dual boot system).

There's definitely some issue here...

[kristanb]

I still haven't ruled out hardware problems. Every once in awhile X will just die, and I finally had an error message I could do something with (most of the time there is no indication of what's wrong...nothing in the logs, messages, etc). The last time X crashed it reported a signal 11, which apparently can be caused by just about anything, but a bad sector on the swap partition and bad system RAM are common causes. Bad swap space could also explain why I only get these problems in Linux/X11 (and not Win98), and why they are so erratic. When I get a chance I'm going to mess around with the swap to see if I can make the problem go away - I'll let you know.

[hennikul]

I had a similar problem. My problem was that my machine crashed when there was a lot of IO activity. I was able to reproduce the crashes even without the X-server running. But allways with the nvidia.o kernel driver installed. Never without that module. I also experienced a couple of X-server hangs which forced me to kill X with Sysrq-ctrl-alt-k.

My setupt is a 1.1GHz Athlon on an ABIT KT7-RAID motherboard and a Geforce4 Ti4200 graphics board. I'm running Mandrake 9.1.

Two days ago I upgraded my Motherboard BIOS. I hadn't done so in quite a long time. After this the problem seems to be gone. I have not experienced any crashes since. You seem to have a similar board, so you might want to try that yourself...

Hope this helps!

Regards,
Henning

[Guardian Bob]

I'm having similar behavior on RH 9. Possibly this could help:
This is probably unrelated, but I experience the same thing on a distroless machine (XF 4.3) when using:

Option "RenderAccel" "1"

In my XF86Config. NVIDIA mark this off by default and say it's experimental, but it used to work prior to the 4xxx series on this GF2Ti, now it's broken.

If you have RenderAccel set, try removing that line.

Cheers,
Alistair Strachan.

--GB

[afrosheen]

It was unset to begin with when it was locking, so it doesn't make any difference set or unset on my machine.

I've rolled back to Mandrake 9 for the time being until everyone figures out what's going on. With mdk9 and the new kernel drivers, everything is rock solid like it used to be.

[3777]

Wierd, once I got everything installed in Slackware 9, I haven't had any problems with KDE whatsoever. My first Distro was Mandrake and everything used to crash, but I think that's a reflection of the Distro rather then the Nvidia drivers. Don't know what problem the Slackware guy is having, maybe it's hardware? I'm runnint a TI 4600 card and the KT400 Chipset:confused:

[afrosheen]

Spare us the distro bigotry. If you haven't noticed, this is a help forum. :rolleyes:

[3777]

Originally posted by afrosheen
Spare us the distro bigotry. If you haven't noticed, this is a help forum. :rolleyes:

It's not distro bigotry, almost every program I ever used in Mandrake would lock up, crash, and freeze, and that was just with the Mandrake drivers. I almost quit Linux because of Mandrake. Really, it's not bigotry, just advice from someone with experience...Mandrake is also the only distro that wouldn't install on my current system. The only other distros I've used are Red Hat and now Slackware, and I've had absolutely none of the problems with either that I always used to have with Mandrake.:rolleyes:

[kristanb]

I did a little more messing around on my Mandrake 9.1 box, using the latest 4349 driver. The problem is so ephemeral that I'm still not completely sure it's fixed, but I haven't had a lock up in about 24 hours, so....

I re-formatted my swap partition. There were no bad blocks reported, but it seemed to fix the problem for awhile, then it happened again.

I had tried setting "NvAgp" "1" before, in response to somebody else's post, but if you're brighter than I you'll notice that it should be "NvAGP" "1" (gotta love case sensitivity). After I fixed the typo, the box has been stable.

I'm still concerned that I could be having a hardware problem, since I don't really know what this option setting does. Is there reason to believe that having the wrong AGP settings could cause flaky performance, rather than preventing X from starting, or locking consistently with a particular app?

[omas]

Well, i also have changed my XF86Config-4.

Instead of your

Option "NvAGP" "1"

or the defaultsetting, i used

Option "NvAGP" "3"

And since then i havent had any lockups... (strange)

I also have played around with my BIOS settings, but havent recorded wich settings i changed there.

[ka-klick]

Well, i also have changed my XF86Config-4.
Instead of your
Option "NvAGP" "1"
or the defaultsetting, i used
Option "NvAGP" "3"


I just tried that for the heck of it, since I hadn't heard anything out of Andy on my own thread since the day after I started it, and it worked (so far). Thanks! :)

[bwkaz]

That's, umm, very strange. For that setting, 3 is the default. If you don't have the option specified, it's as if you have it set like you guys do, at "3".

[ka-klick]

That's, umm, very strange. For that setting, 3 is the default. If you don't have the option specified, it's as if you have it set like you guys do, at "3".


That's what I thought, but for the first time in a couple weeks I have a working X setup that has 3d accelleration. I tried it as a lark, but it did work. I'm glad, because now my wife'll stop being grumpy about not being able to frag me. ;)

[bwkaz]

Hmm.... *shrug*

Well, I guess if it works...

[Kenneth]

On a friends Mandrake 9.1 system (a Gforce 2 card) I have also seen these lockups.

I tried reverting to an older driver, but it didn't help.

I guess the next thing is to try the

Option "NvAGP" "3"

/Kenneth

[Andy Mecham]

I guess the next thing is to try the

Option "NvAGP" "3"That's the default behavior - unless you're specifying a different number to NvAgp, that's what you get.

--andy

[afrosheen]

I bet this stuff is the kernel's fault, not the module. I seem to remember someone mentioning Mandrake using a -pre kernel right before 9.1's release. Then again, it could be X's fault also, since myself and others have noticed that X takes up 99% of the processor's time before it locks. Who knows, all I know is Mandrake 9 sucks compared to 9.1 but at least the drivers work. :p