Re: [Qemu-devel] [Xen-devel] [PATCH V6 10/11] Introduce Xen PCI Passthrough, MSI (3/3)

[Jan Beulich]

>>> On 14.02.12 at 14:46, Anthony PERARD <address@hidden> wrote:
> On Tue, Feb 14, 2012 at 13:13, Jan Beulich <address@hidden> wrote:
>>>>> On 13.02.12 at 13:20, Anthony PERARD <address@hidden> wrote:
>>> From: Jiang Yunhong <address@hidden>
>>>
>>> A more complete history can be found here:
>>> git://xenbits.xensource.com/qemu-xen-unstable.git
>>
>> This needs to be updated to include the changes in
>> 
> http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commitdiff;h=bb36d63 
> 2e4cabf47882adff07a45c6702c4a5b30
>> (and hopefully the broken function that got fixed in
>> 
> http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commitdiff;h=8cc8a36 
> 51c9c5bc2d0086d12f4b870fc525b9387
>> didn't even make it into this patch set).
> 
> For the first patch/link, the 'volatile' variable should already be in
> the patch (without a separate patch :( ).

Good that you said "volatile" - that was the wrong link. The correct one
is
http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commitdiff;h=56d7747a3cf811910c4cf865e1ebcb8b82502005

I'm sorry for that,
Jan

> For the second, yes,
> unregister_iomem is done by other functions in QEMU.
> 
>> In particular it must be avoided
>> to map the MSI-X table with PROT_WRITE, and the respective MMIO
>> range should not get assigned to the guest at all.
> 
> I will check that.
> 
> Regards,
> 
> -- 
> Anthony PERARD