The field taking the value 7 (resulting in 18-bit DIDs when using the
calculation in cap_ndoms(), when the DID fields are only 16 bits wide)
is reserved. Instead of misbehaving in case we would encounter such an
IOMMU, refuse to use it.

Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/xen/drivers/passthrough/vtd/iommu.c
+++ b/xen/drivers/passthrough/vtd/iommu.c
@@ -1267,8 +1267,11 @@ int __init iommu_alloc(struct acpi_drhd_
 
     quirk_iommu_caps(iommu);
 
+    nr_dom = cap_ndoms(iommu->cap);
+
     if ( cap_fault_reg_offset(iommu->cap) +
          cap_num_fault_regs(iommu->cap) * PRIMARY_FAULT_REG_LEN > PAGE_SIZE ||
+         ((nr_dom - 1) >> 16) /* I.e. cap.nd > 6 */ ||
          (has_register_based_invalidation(iommu) &&
           ecap_iotlb_offset(iommu->ecap) >= PAGE_SIZE) )
     {
@@ -1294,8 +1297,6 @@ int __init iommu_alloc(struct acpi_drhd_
     if ( !ecap_coherent(iommu->ecap) )
         iommu_non_coherent = true;
 
-    nr_dom = cap_ndoms(iommu->cap);
-
     if ( nr_dom <= DOMID_MASK * 2 + cap_caching_mode(iommu->cap) )
     {
         /* Allocate domain id (bit) maps. */

On Mon, Apr 11, 2022 at 11:37:28AM +0200, Jan Beulich wrote:
> The field taking the value 7 (resulting in 18-bit DIDs when using the
> calculation in cap_ndoms(), when the DID fields are only 16 bits wide)
> is reserved. Instead of misbehaving in case we would encounter such an
> IOMMU, refuse to use it.
> 
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>

I would maybe prefer to get more specific error message rather than
"IOMMU: unsupported" and a dump of the iommu registers.

Thanks, Roger.

On 12.04.2022 11:22, Roger Pau Monné wrote:
> On Mon, Apr 11, 2022 at 11:37:28AM +0200, Jan Beulich wrote:
>> The field taking the value 7 (resulting in 18-bit DIDs when using the
>> calculation in cap_ndoms(), when the DID fields are only 16 bits wide)
>> is reserved. Instead of misbehaving in case we would encounter such an
>> IOMMU, refuse to use it.
>>
>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> 
> Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>

Thanks.

> I would maybe prefer to get more specific error message rather than
> "IOMMU: unsupported" and a dump of the iommu registers.

Perhaps, but this extends to other properties being checked as well then,
I would say, and hence may want to be the subject of yet another patch.

Jan