Kernel hang with latest snapshot

[Li Guanglei]

David Boreham wrote:
> Li Guanglei wrote:
> 
>> The 2.6.17-1.2187_FC5-i686 debuginfo package is too big to download 
>> for me :(. But I tried 2.6.18/i686, it works really well. Probing for 
>> almost half an hour without crash the kernel.
>>
> Hi, perhaps there is an incompatibility between the current stap and the 
> RH kernel.
> I will see if I can make a stock kernel work (I'm assuming that's what 
> you mean
> by 2.6.18 above?).
> 

Yes. I am using 2.6.18

>>
>> I also checked in some bugfix for LKET on i386. I tried to turn on all 
>> probes(addevent.*) and it works ok. You can try the latest CVS and let 
>> me know if you have any problem.
> 
> Hi, just tried the current cvs code. The compile error is gone. 
> Unfortunately the kernel hang is still present :(
> 
> I may also try a 'real' machine vs a vmware instance.
> 

I tried addevent.iosyscall.*, addevent.scsi.* , addevent.syscall.*, 
and they all seem stable.

I tested addevent.* twice, which will turn on a lot of probes. The 
first time it crash after 60 seconds. The second time it keep running 
for almost 30 minutes without crash while I am running other 
application at the same time. So I can't say addevent.* won't crash 
your system, but you can try it or try with a smaller set of trace 
hooks :)

btw, I updated elfutils to 0.124 in the second test of addevent.*

- Guanglei