Re: [PATCH] VT-d: improve RMRR validity checking

[Weidong Han <weidong.han@intel.com>]

Alex Williamson wrote:
> On Thu, 2010-03-11 at 10:11 +0800, Weidong Han wrote:
>    
>   
>> Alex, you are right. IOAPICs can be included in any DRHDs. Pls try 
>> following patch, if no problem, I will submit it.
>>     
>
> Thanks Weidong.  This of course works, but I still get this output:
>
> (XEN) IOAPIC[0]: apic_id 8, version 32, address 0xfec00000, GSI 0-23
> (XEN) IOAPIC[1]: apic_id 0, version 32, address 0xfec08000, GSI 24-47
> (XEN) IOAPIC[2]: apic_id 10, version 32, address 0xfec10000, GSI 48-71
> (XEN) Enabling APIC mode:  Phys.  Using 3 I/O APICs
> (XEN) [VT-D]dmar.c:421:   Non-existent device (84:13.0) is reported in this DRHD's scope!
> (XEN) [VT-D]dmar.c:442:   There are devices under device scope are not PCI discoverable! if xen fails at VT-d enabling, pls try option iommu=workaround_bios_bug.
>
> So now we've effectively relegated this code to printing things that
> look like errors for both actual bad DMAR tables and 100% spec compliant
> tables.  At a minimum, I think these dprintks need to be reduce to info
> or debug level since they're effectively just spewing out noise.  Can't
> we put a tag for the device type in the list of scope devices and skip
> checking discoverable PCI devices for IOAPICs?  Do we need to do the
> same for HPETs?  Thanks,
>   
good suggestion to check the device type. I cooked a new patch. It skips 
checking IOAPIC and HPET. pls have a try.

diff -r cadf1bae9ee2 xen/drivers/passthrough/vtd/dmar.c
--- a/xen/drivers/passthrough/vtd/dmar.c    Thu Feb 25 18:26:45 2010 +0800
+++ b/xen/drivers/passthrough/vtd/dmar.c    Thu Mar 11 19:39:45 2010 +0800
@@ -407,9 +407,15 @@ acpi_parse_one_drhd(struct acpi_dmar_ent
     {
         u8 b, d, f;
         int i, invalid_cnt = 0;
+        struct acpi_dev_scope *acpi_scope = dev_scope_start;

         for ( i = 0; i < dmaru->scope.devices_cnt; i++ )
         {
+            acpi_scope += (i == 0) ? 0 : acpi_scope->length;
+            if ( acpi_scope->dev_type == ACPI_DEV_IOAPIC ||
+                 acpi_scope->dev_type == ACPI_DEV_MSI_HPET )
+                continue;
+
             b = PCI_BUS(dmaru->scope.devices[i]);
             d = PCI_SLOT(dmaru->scope.devices[i]);
             f = PCI_FUNC(dmaru->scope.devices[i]);


> Alex
>
>
>   
>> diff -r cadf1bae9ee2 xen/drivers/passthrough/vtd/dmar.c
>> --- a/xen/drivers/passthrough/vtd/dmar.c    Thu Feb 25 18:26:45 2010 +0800
>> +++ b/xen/drivers/passthrough/vtd/dmar.c    Thu Mar 11 17:49:40 2010 +0800
>> @@ -437,11 +437,9 @@ acpi_parse_one_drhd(struct acpi_dmar_ent
>>              else
>>              {
>>                  dprintk(XENLOG_WARNING VTDPREFIX,
>> -                    "  The DRHD is invalid due to there are devices under "
>> -                    "its scope are not PCI discoverable! Pls try option "
>> -                    "iommu=force or iommu=workaround_bios_bug if you "
>> -                    "really want VT-d\n");
>> -                ret = -EINVAL;
>> +                    "  There are devices under device scope are not PCI "
>> +                    "discoverable! if xen fails at VT-d enabling, pls try "
>> +                    "option iommu=workaround_bios_bug.\n");
>>              }
>>          }
>>          else
>>
>>     
>
>
>
>