Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Random lockups



 Don Levey wrote: 
> Clearly I'm tickling something, but I have no idea what.  Short of 
> ripping the  thing apart and replacing each part individually (way too 
> expensive for me right about now) does anyone have any suggestions? 

I've had problems like you describe in the past on my home computers, and I 
too blamed the video card / proprietary drivers. 

But now for a project at work we're using PC104 boards and heat dissipation is 
a bit of an issue (it's in a sealed box).  We're seeing hard OS lockups 
periodically (only when it's in the sealed box, it'll run for weeks on end 
when it's sitting on my desk).  No syslog messages, no crash info on the 
console, nothing. I'm fairly certain its head-related, but I'm having trouble 
explaining why that would cause a hard OS lockup. 

My old theory with my home machine and the video card was that the video card 
would overheat and freeze, and in the process somehow lock up the PCI bus 
and/or a DMA channel, which totally locks up the machine. 

I suppose the same thing could be happening here, in that one of chips 
(perhaps not the CPU) is overheating and disrupting one of the critical buses. 

Back to your problem, and suggestions for you: 
  - add a heatsink to your Northbridge and/or Southbridge chips (southbridge 
is the PCI-bus controller) 
  - if your CPU has frequency scaling, force it to run at the slowest speed 
(ie using the least amount of power).  Its probably defaulting to the 
"on-demand" governor, which only pumps up the processor speed when you do 
something, like move/select files in Thunar. 
  - see if you can predictably trigger a lockup by doing a subsystem-specific 
test, like bonnie++ for disk I/O, glxgears for video+cpu, etc. 

HTH, 
Matt 

-- 
This message has been scanned for viruses and 
dangerous content by MailScanner, and is 
believed to be clean. 

_______________________________________________ 
Discuss mailing list 
[hidden email] 
http://lists.blu.org/mailman/listinfo/discuss
 


BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!



Boston Linux & Unix / webmaster@blu.org