Re: Testing status of HVM (Intel VT) on 64bit XEN unstable c/s 11686

[Ed Smith <esmith@virtualiron.com>]

OK it took a lot longer to reproduce but I did hit the
Error: shadow lock held by shadow_update_cr3 crash running
mtest01 on a RHEL4U2-32bit SMP HVM guest.  I was unable to
reproduce any of the other shadow crashes.

Details in my next test report post for c/s 11731.

Ed


Ed Smith wrote:

> Thanks Tim,
> 
> I'm testing with c/s 11731 now and it looks good.  I'll be posting my
> test report shortly.
> 
> Ed
> 
> -----Original Message-----
> From: Tim Deegan [mailto:Tim.Deegan@xensource.com] 
> Sent: Monday, October 09, 2006 12:05 PM
> To: Ed Smith
> Cc: Xen Devel
> Subject: Re: [Xen-devel] Testing status of HVM (Intel VT) on 64bit XEN
> unstable c/s 11686
> 
> Hi Ed,
> 
> Changeset 11729:6e932f32662c of xen-unstable should fix these crashes.
> 
> Cheers,
> 
> Tim.
> 
> At 09:22 -0400 on 04 Oct (1159953769), Ed Smith wrote:
>> These bugs happen every time for me.  To reproduce the bug I do the
>> following:
>>
>> - Boot a release build of XEN
>> - Launch a RHEL4U2-32bit HVM guest, 256MB, 2CPU SMP kernel, 4GB swap 
>> file, local SATA disk
>> - ssh into the guest and run the following test:
>>   # while [ 1 ]; do mtest01 -v -p80; done
>> - Within 1 or 2 passes of the test I hit one of the several crashes 
>> reported.
>> - Nothing else is running on the guest and this is the only guest 
>> running on XEN.
>>
>> Hope this helps,
>> Ed
>>
>>
>> -----Original Message-----
>> From: Tim Deegan [mailto:Tim.Deegan@xensource.com]
>> Sent: Wednesday, October 04, 2006 9:13 AM
>> To: Ed Smith
>> Cc: Xen Devel
>> Subject: Re: [Xen-devel] Testing status of HVM (Intel VT) on 64bit XEN
> 
>> unstable c/s 11686
>>
>> Hi Ed,
>>
>> At 16:05 -0400 on 03 Oct (1159891531), Ed Smith wrote:
>>> - NEW: Domain crash running ltp "mtest01 -p80" on 32bit SMP
>>>   HVM guest and then XEN locked up requiring a hard reset:
>>>   domain_crash_sync called from private.h:495 (failure.8)
>>> - XEN crash running ltp "mtest01 -p80" on 32bit SMP HVM guest:
>>>   BUG at multi.c:2967 from sh_page_fault__shadow_3_guest_3 
>>> (failure.3)
>> We're having some difficulty reproducing these failures.  Are you 
>> running full ltp tests or just the mtest01 binary?  Are you running 
>> anything else in the VM at the time?  How often do the failures occur?
>> Was there a difference between the tests which caused the two failures
> 
>> or did they happen on different runs of the same test?
>>
>> Cheers,
>>
>> Tim.