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

Bug#410807: Patch for "kernel BUG at drivers/xen/core/evtchn.c:481"



On Sun, Jun 15, 2008 at 09:28:47PM +0100, Ian Campbell wrote:
> Hi,
> 
> I believe this issue is fixed by
> http://xenbits.xensource.com/xen-unstable.hg?rev/914304b3a3da
> 
> Without the patch (2.6.18.dfsg.1-18etch1) I was able to reproduce the
> problem by repeatedly starting 4 guest domains in parallel[0] in a
> around 5-6 hours or 288 iterations (first time I repro'd it was a couple
> of hours, I wasn't counting time or iterations that time).
> 
> With the patch applied on top of the svn kernel
> (2.6.18.dfsg.1-21.0.hellion0) the test survived overnight, around 16
> hours and 850 iterations.
> 
> ???I've attached a version of this backported to the Etch kernel (some of
> the function names differ). ???I also attached the svn diff-of-a-diff
> against revision in r11588, 2.6.18.dfsg.1-22.

Thanks Ian! Please go ahead and commit. Note that I think you need a
'#' in front of the bug number for the closer syntax to work.

> This issue is already fixed in the Lenny/Sid kernels which use the
> paravirt_ops kernel port which has completely reimplemented event
> channel code.
> 
> Cheers,
> Ian.
> 
> [0] n=0 ; while : ; do date ; echo iteration $n ; xm create
> debian-x86_32p-1 & xm create debian-x86_32p-2 & xm create
> rhel44-x86_32p-1 kernel=/boot/vmlinuz-2.6.18.8-x86_32p-xenU ramdisk= &
> xm create rhel41-x86_32p-1 kernel=/boot/vmlinuz-2.6.18.8-x86_32p-xenU
> ramdisk= & sleep 1m ; xm dest debian-1 ; xm dest debian-2 ; xm dest
> rhel41-1 ; xm dest rhel44-1 ; sleep 5s ; n=$(($n + 1)) ; done
> 






-- 
dann frazier




Reply to: