Re: [Xen-devel] [PATCH 1/2] x86/iommu: avoid mapping the interrupt address range for hwdom

[Jan Beulich <JBeulich@xxxxxxxx>]

On 25.07.2019 12:54, Roger Pau Monné  wrote:
> On Thu, Jul 25, 2019 at 10:22:17AM +0000, Jan Beulich wrote:
>> On 23.07.2019 17:48, Roger Pau Monne wrote:
>>> Current code only prevent mapping the lapic page into the guest
>>> physical memory map. Expand the range to be 0xFEEx_xxxx as described
>>> in the Intel VTd specification section 3.13 "Handling Requests to
>>> Interrupt Address Range".
>>>
>>> AMD also lists this address range in the AMD SR5690 Databook, section
>>> 2.4.4 "MSI Interrupt Handling and MSI to HT Interrupt Conversion".
>>>
>>> Requested-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
>>> Signed-off-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
>>
>> I've committed this on the basis that it shouldn't hurt, but having
>> thought about this some more I'm not really sure I see the point:
>> The IOMMU special cases accesses into this range anyway, to redirect
>> lookup to the interrupt remapping table instead of the DMA remapping
>> one. Hence any mappings inserted into this range are simply useless,
>> but shouldn't otherwise hurt.
> 
> Intel SDM contains:
> 
> "Software must ensure the second-level paging-structure entries are
> programmed not to remap input addresses to the interrupt address
> range. Hardware behavior is undefined for memory requests remapped to
> the interrupt address range."
> 
> In section 3.13 (Handling Requests to Interrupt Address Range).
> 
> Since arch_iommu_hwdom_init/hwdom_iommu_map adds entries to both the
> hap and the iommu page tables (or to hap only if shared) Xen should be
> careful to not map this range because the iommu special cases this
> range, but I'm not sure what hap does.

Hmm, in the shared-pt case this is indeed desirable in any event. I
have to admit that I didn't recall that even in the non-shared case
arch_iommu_hwdom_init() would fiddle with the HAP tables as well,
rather than just the IOMMU ones. I don't think this should remain
this way long term. Or was there a reason for this to be needed?

Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel