Re: [Xen-devel] Xen security advisory CVE-2011-1898 - VT-d (PCI passthrough) MSI

[Keir Fraser <keir@xxxxxxx>]

On 24/05/2011 16:15, "Ian Jackson" <Ian.Jackson@xxxxxxxxxxxxx> wrote:

> Ian Campbell writes ("Re: [Xen-devel] Xen security advisory CVE-2011-1898 -
> VT-d (PCI passthrough) MSI"):
>> IOMMU: Fail if intremap is not available and iommu=required/force.
>> 
>> Rather than sprinkling panic()s throughout the setup code hoist the check up
>> into common code.
>> 
>> Signed-off-by: Ian Campbell <ian.campbell@xxxxxxxxxx>
> 
> Acked-by: Ian Jackson <ian.jackson@xxxxxxxxxxxxx>
> 
> Keir, do you think we should apply this then ?

<sigh> take your pick really. Majority opinion is on the side of this
revised patch, however Intel are the primary maintainers of this code and
they clearly do not like it. If I have a casting vote here, I would be
inclined to plump in favour of the revised patch -- we already have iommu=on
as a best-effort option, and I believe iommu=force could be stronger than it
is. However Joseph's claim that the non-DoS vulns may all now be handled is
not as unconvincing as some seem to believe (and I was in that camp for a
while) -- I can't really see how the attack vector can be successfully
exploited now my mitigation patch is in the tree. So I'm not strongly
inclined one way or the other really.

 -- Keir

> Ian.
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-devel



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel