On 7/24/2025 2:22 PM, Philippe Mathieu-Daudé wrote:
> On 24/7/25 08:47, Sairaj Kodilkar wrote:
>> Support the writes to the status register so that guest can reset the
>> EventOverflow, EventLogInt, ComWaitIntr, etc bits after servicing the
>> respective interrupt.
>>
>> Signed-off-by: Sairaj Kodilkar <sarunkod@amd.com>
>> Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
>> ---
>> hw/i386/amd_iommu.c | 3 +++
>> 1 file changed, 3 insertions(+)
>>
>> diff --git a/hw/i386/amd_iommu.c b/hw/i386/amd_iommu.c
>> index bbffd07b4e48..7c2fa80d14ff 100644
>> --- a/hw/i386/amd_iommu.c
>> +++ b/hw/i386/amd_iommu.c
>> @@ -850,6 +850,9 @@ static void amdvi_mmio_write(void *opaque, hwaddr
>> addr, uint64_t val,
>> amdvi_mmio_reg_write(s, size, val, addr);
>> amdvi_handle_pprtail_write(s);
>> break;
>> + case AMDVI_MMIO_STATUS:
>> + amdvi_mmio_reg_write(s, size, val, addr);
>> + break;
>
> Should we add:
>
> default:
> g_assert_not_reached();
I am against adding a failure here. Because when xtsup=on, linux tries
to write all the XT registers ( which we dont support yet, because of
which event logging does not work in this mode). Although future cleanup
patches will add this support, I dont want to break things now.
I think amdvi_mmio_trace_write() is sufficient for debugging purpose,
which prints "UNHANDLED" for unsupported registers.
Thanks
Sairaj
>
> to catch unimplemented cases?
>
>> }
>> }
>