Re: random crash in post_kvm_run()

[Avi Kivity <avi@redhat.com>]

On 07/05/2010 10:42 AM, Avi Kivity wrote:
> Please don't top-post.
>
> On 07/03/2010 05:23 PM, BuraphaLinux Server wrote:
>> Ok, I kept going like you said.   Here is what it said:
>>
>> $git bisect good
>> 44ea2b1758d88ad822e65b1c4c21ca6164494e27 is the first bad commit
>> commit 44ea2b1758d88ad822e65b1c4c21ca6164494e27
>> Author: Avi Kivity<avi@redhat.com>
>> Date:   Sun Sep 6 15:55:37 2009 +0300
>>
>>      KVM: VMX: Move MSR_KERNEL_GS_BASE out of the vmx autoload msr area
>>
>>      Currently MSR_KERNEL_GS_BASE is saved and restored as part of the
>>      guest/host msr reloading.  Since we wish to lazy-restore all the 
>> other
>>      msrs, save and reload MSR_KERNEL_GS_BASE explicitly instead of 
>> using
>>      the common code.
>>
>>      Signed-off-by: Avi Kivity<avi@redhat.com>
>
> That doesn't make any sense.  This commit shouldn't affect anything in 
> user-kernel communications.
>
> Can you describe your environment?  I'll try to reproduce it.
>

I was able to reproduce it, and the commit does make sense.

The faulting instruction is

   0x807182a <post_kvm_run+10>     mov    %gs:0x14,%eax

which is a stack guard fetch.  It shouldn't ever fault - so it looks 
like %gs is corrupted, and indeed the commit plays with %gs.

I'll investigate further.

-- 
error compiling committee.c: too many arguments to function