Re: [xen-unstable bisection] complete test-amd64-i386-rhel6hvm-intel

[Keir Fraser <keir.xen@gmail.com>]

On 21/11/2011 19:43, "Jean Guyader" <jean.guyader@gmail.com> wrote:

> On 21 November 2011 18:47, Keir Fraser <keir.xen@gmail.com> wrote:
>> On 21/11/2011 11:55, "Keir Fraser" <keir.xen@gmail.com> wrote:
>> 
>>> On 21/11/2011 11:37, "Ian Jackson" <Ian.Jackson@eu.citrix.com> wrote:
>>> 
>>>> xen.org writes ("[xen-unstable bisection] complete
>>>> test-amd64-i386-rhel6hvm-intel"):
>>>>> branch xen-unstable
>>>>> xen branch xen-unstable
>>>>> job test-amd64-i386-rhel6hvm-intel
>>>>> test redhat-install
>>>>> 
>>>>> Tree: linux git://github.com/jsgf/linux-xen.git
>>>>> Tree: qemu git://hg.uk.xensource.com/HG/qemu-xen-unstable.git
>>>>> Tree: xen http://xenbits.xen.org/staging/xen-unstable.hg
>>>>> 
>>>>> *** Found and reproduced problem changeset ***
>>>>> 
>>>>>   Bug is in tree:  xen http://xenbits.xen.org/staging/xen-unstable.hg
>>>>>   Bug introduced:  7a9a1261a6b0
>>>>>   Bug not present: 9a1a71f7bef2
>>>> 
>>>> This seems to have completely broken HVM ...
>>> 
>>> I'll revert if there's no fix forthcoming.
>> 
>> I hear silence so I will revert the series tomorrow morning.
>> 
> 
> Ok. I didn't managed to replicate the issue yet.

Actually, it wasn't too hard to work out. This bisection is misleading
though, as it's zeroed in on the RCU locking bug, which is already fixed.
The bug is actually in a later changeset which modifies hvmloader.

Looking at the hvmloader/pci.c changes, the unconditional assignment to
low_mem_pgend after the loop is obviously wrong. As is removing the handling
for high_mem_pgend==0. I checked in a reworked version that is closer to the
original code. 

Hopefully our tests will work again now.

 -- Keir

> Jean
> 
>>>  -- Keir
>>> 
>>>>>   changeset:   24163:7a9a1261a6b0
>>>>>   user:        Jean Guyader <jean.guyader@eu.citrix.com>
>>>>>   date:        Fri Nov 18 13:41:33 2011 +0000
>>>>> 
>>>>>       add_to_physmap: Move the code for XENMEM_add_to_physmap
>>>>> 
>>>>>       Move the code for the XENMEM_add_to_physmap case into it's own
>>>>>       function (xenmem_add_to_physmap).
>>>>> 
>>>>>       Signed-off-by: Jean Guyader <jean.guyader@eu.citrix.com>
>>>>>       Committed-by: Keir Fraser <keir@xen.org>
>>>> 
>>>> Ian.
>>> 
>>> 
>> 
>> 
>> 
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xensource.com
>> http://lists.xensource.com/xen-devel
>>