Re: [PATCH 2 of 2] x86/VT-d: Sandy-Bridge BT98 Erratum

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 29/04/13 16:13, Jan Beulich wrote:
>>>> On 26.04.13 at 20:11, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
>> @@ -288,9 +309,12 @@ void __init platform_quirks_init(void)
>>      /* ioremap IGD MMIO+0x2000 page */
>>      map_igd_reg();
>>  
>> -    /* Tylersburg interrupt remap quirk */
>> +    /* Interrupt remapping quirks */
>>      if ( iommu_intremap )
>> +    {
>>          tylersburg_intremap_quirk();
>> +        snb_bt98_erratum();
>> +    }
> Considering the nature of the erratum, keying this off
> iommu_intremap seems wrong - iommu_qinval ought to be used
> instead - intel_vtd_setup() takes care to disable interrupt
> remapping when queued invalidation is not available, yet when
> you make the above conditional upon iommu_intremap, queued
> invalidation could still get enabled (and used for whatever else
> purposes, now or in the future).
>
> Jan
>

Yes - I mentally had the erratum the wrong way around, as intremap is
the visible impact of it.

I shall respin once hearing back from Xiantao, especially regarding the
first patch.

~Andrew