Re: [patch] voluntary-preempt-2.6.9-rc1-bk4-R0

["K.R. Foley" <kr@cybsft.com>]

Ingo Molnar wrote:
> * K.R. Foley <kr@cybsft.com> wrote:
> 
> 
>>Managed to hang the system again under heavy load. This time with the
>>above patch:
>>
>>http://www.cybsft.com/testresults/crashes/2.6.9-rc1-bk4-R0.txt
>>
>>Last time was with Q7:
>>
>>http://www.cybsft.com/testresults/crashes/2.6.9-rc1-bk4-Q7.txt
> 
> 
> seems to be the same thing - an unbalanced preemption count, possibly
> due to some locking error. Unfortunately the assert catches the
> imbalance only at exit time. (it's unlikely that the do_exit() code is
> buggy.)
> 
> i'll add a new feature to debug this: when crashing on an assert and
> tracing is enabled the trace leading up to the crash will be printed to
> the console. How did you capture the crash - was it in the syslog or do
> you have serial logging? Maybe this is not the real crash but only a
> followup crash?
> 
> 	Ingo
> 

This actually came from syslog. When it happened it was completely 
locked, so I have no idea if there was more that didn't make it to disk. 
Of course the monitor was blanked because most of the access is via the 
network. I could probably enable serial logging if necessary because 
it's sitting next to a couple of other machines.

kr