RE: [Xen-ia64-devel] RFC: patch to improve Xen stability.

["Magenheimer, Dan (HP Labs Fort Collins)" <dan.magenheimer@xxxxxx>]

Excellent progress!

The crash is almost certainly some rare combination of
two events.  For example, perhaps a clock tick is occurring
while the register stack is "in an unstable state" during
a fork?

This suggests a theory:  Perhaps the frequency decreased
as a result of reducing HZ?  It might be worth an experiment
to increase HZ (in Xenlinux)  back to 1000 and see if
it increases the frequency that the problem occurs.  Making
it occur more often should make it easier to find.

Dan

> -----Original Message-----
> From: xen-ia64-devel-bounces@xxxxxxxxxxxxxxxxxxx 
> [mailto:xen-ia64-devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf 
> Of Tristan Gingold
> Sent: Thursday, April 27, 2006 7:09 AM
> To: xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [Xen-ia64-devel] RFC: patch to improve Xen stability.
> 
> Le Jeudi 27 Avril 2006 13:06, Tristan Gingold a écrit :
> > Hi,
> >
> > here is a patch which seems to improve stability.
> > Until now, it was able to compile 45 times the linux kernel 
> in a raw,
> > without faults.
> > I will do more checks during the week-end.
> >
> > The patch has 3 fixes:
> > * 2 obvious fix related to SMP-guest
> > * 1 fix: make itc and itr consistent, thus removing dtlb_pte.
> As expected, this patch does not fix all the instability.  
> The crashes are 
> less frequent but more localized. Surprise: most of the 
> crashes occur in 
> libc:fork.o, after the system call.  They are due to bad 
> values in stacked 
> register.
> 
> Tristan.
> 
> 
> 
> _______________________________________________
> Xen-ia64-devel mailing list
> Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-ia64-devel
> 

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel