Tracking hard lockups

BradB · Posted: Thu Jan 30, 2003 9:27 am Post subject: Tracking hard lockups

Hi there, I'm writting an openGL app, using glut as the wrapper. This is something I've done many times before. For some reason every now & then my app locks up & freezes gentoo hard. I can't kill X or switch to a console. I have an nVidia Geforce2 GTS running nVidia binary drivers.
I could change to the nv driver, but I'd like to keep my acceleration. What my question is : are there any logging systems I can enable or trace through to find out what's going on?

Also - uname -a gives

Linux workstation 2.4.19-gentoo-r9 #17 SMP Sat Dec 14 01:57:10 NZDT 2002 i686 AMD Athlon(tm) Processor AuthenticAMD GNU/Linux

This is the first problem of this sort I have seen (& I haven't upgraded any system stuff lately)

Cheers
Brad

rac · Posted: Thu Jan 30, 2003 6:49 pm Post subject:

Maybe compiling magic SysRq support into your kernel could help.
_________________
For every higher wall, there is a taller ladder

jivera · n00b Joined: 26 Jan 2003 Posts: 14 Location: Plano, Tx

I've been having the same problems actually... my problems seem mostly when I'm working with gedit (but it's possible it's just a coincidence).

I have a GeForce2 MX400 (PCI) and Athlon XP (wasn't at one time there some fix people were recommending for GeForce/Athlon users? I never had any problems at the time so I don't remember what the symptoms were or the cure and I wouldn't be surprised if it's been fixed since then but still...).

ebrostig · Posted: Fri Jan 31, 2003 1:37 am Post subject: Re: Tracking hard lockups

BradB · Posted: Sun Feb 02, 2003 10:02 am Post subject:

Hi all. To ebrostig: this could be mem management, but I'm not doing anything fancy - I don't even alloc off the heap - and in the past all mem errors usually exhibit as segfaults.
I compiled magic SysRq into my kernel & tested that it worked. The problem happened again & the SysRq commands did not work, they didn't kill the process or sync or reboot. Strangely enough while I was playing around the app started again for a second. As I waited the app continued to "burst" - maybe 0.5 of a second activity every 10 secs or so. I was still unable to switch consoles, magic SysRq, etc.
I've now setup sshd so I can open a remote console from my laptop - is there anyway I can pipe kernel messages to this console so I can further trace what is going on?
I'm also planning on
1) using nv driver rather than the nVidia binaries
2) vanilla kernel
but I'm away for a while, so I won't report back for a week or two.

Cheers
Brad