Pkg xen devel - Aug 2012 - Bug#684661: Xen BUG at pci_amd_iommu.c:33

Hi Wei,

A Debian user has hit this message and reported it in
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=684661

I think it is the BUG_ON in:
        struct amd_iommu *find_iommu_for_device(int bdf)
        {
            BUG_ON ( bdf >= ivrs_bdf_entries );
            return ivrs_mappings[bdf].iommu;
        }

It looks like ivrs_bdf_entries comes from ACPI. Unfortunately the bug
report is a bit vague about things like stack traces etc so it's hard to
say where BDF came from.

Any ideas? I'll also follow up to the submitter to try and get some more
details.

Ian.

On 08/17/2012 01:46 PM, Ian Campbell wrote:
> Hi Wei,
>
> A Debian user has hit this message and reported it in
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=684661
>
> I think it is the BUG_ON in:
>          struct amd_iommu *find_iommu_for_device(int bdf)
>          {
>              BUG_ON ( bdf>= ivrs_bdf_entries );
>              return ivrs_mappings[bdf].iommu;
>          }
>
> It looks like ivrs_bdf_entries comes from ACPI. Unfortunately the bug
> report is a bit vague about things like stack traces etc so it's hard
to
> say where BDF came from.
>
> Any ideas? I'll also follow up to the submitter to try and get some
more
> details.
It difficult to identify the issue from the description and I think a 
full serial boot log would be great. Also using iommu=debug as boot 
option is very helpful to get more information.

Thanks,
Wei
> Ian.
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel at lists.xen.org
> http://lists.xen.org/xen-devel
>

On Fri, 2012-08-17 at 13:10 +0100, Wei Wang wrote:
> On 08/17/2012 01:46 PM, Ian Campbell wrote:
> > Hi Wei,
> >
> > A Debian user has hit this message and reported it in
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=684661
> >
> > I think it is the BUG_ON in:
> >          struct amd_iommu *find_iommu_for_device(int bdf)
> >          {
> >              BUG_ON ( bdf>= ivrs_bdf_entries );
> >              return ivrs_mappings[bdf].iommu;
> >          }
> >
> > It looks like ivrs_bdf_entries comes from ACPI. Unfortunately the bug
> > report is a bit vague about things like stack traces etc so
it''s hard to
> > say where BDF came from.
> >
> > Any ideas? I''ll also follow up to the submitter to try and
get some more
> > details.
> 
> It difficult to identify the issue from the description and I think a 
> full serial boot log would be great. Also using iommu=debug as boot 
> option is very helpful to get more information.
Thanks, I''ve added that to my request to the submitter.

Ian.