Re: [PATCH 14/33] tests: acpi: update expected blobs

[Fabiano Rosas <farosas@suse.de>]

Igor Mammedov <imammedo@redhat.com> writes:

> On Wed, 1 Mar 2023 19:27:33 -0500
> "Michael S. Tsirkin" <mst@redhat.com> wrote:
>
>> On Fri, Feb 24, 2023 at 04:37:53PM +0100, Igor Mammedov wrote:
>> > only following context change:
>> >  -  Local1 = Zero
>> >     If ((Arg0 != ToUUID ("e5c937d0-3553-4d7a-9117-ea4d19c3434d") /* Device Labeling Interface */))
>> >     {
>> >         Return (Local0)
>> >  ...
>> >         Return (Local0)
>> >     }
>> > 
>> >  +  Local1 = Zero
>> >     Local2 = AIDX (DerefOf (Arg4 [Zero]), DerefOf (Arg4 [One]
>> > 
>> > Signed-off-by: Igor Mammedov <imammedo@redhat.com>  
>> 
>> Nope:
>> 
>> diff -ru -N -IDisassembly -IChecksum '-I*     Length           ' old/asl/tests/data/acpi/q35/DSDT.viot.dsl new/asl/tests/data/acpi/q35/DSDT.viot.dsl
>> :--- old/asl/tests/data/acpi/q35/DSDT.viot.dsl  2023-03-01 19:22:57.636454958 -0500
>> :+++ new/asl/tests/data/acpi/q35/DSDT.viot.dsl  2023-03-01 19:22:58.451460462 -0500
>> :@@ -148,7 +148,6 @@
>>                      {
>>                           0x00                                             // .
>>                      }
>> -                Local1 = Zero
>>                  If ((Arg0 != ToUUID ("e5c937d0-3553-4d7a-9117-ea4d19c3434d") /* Device Labeling Interface */))
>>                  {
>>                      Return (Local0)
>> :@@ -159,12 +158,14 @@
>>                      Return (Local0)
>>                  }
>>  
>> +                Local1 = Zero
>>                  Local2 = AIDX (DerefOf (Arg4 [Zero]), DerefOf (Arg4 [One]
>>                      ))
>>                  If (!((Local2 == Zero) | (Local2 == 0xFFFFFFFF)))
>>                  {
>>                      Local1 |= One
>>                      Local1 |= (One << 0x07)
>> +                    Local1 |= (One << 0x05)
>>                  }
>>  
>>                  Local0 [Zero] = Local1
>> 
>> 
>> and the funny thing is, the second change is in the expected
>> file but not in the code so this patch causes the test to fail.
>
> above diff is remnants of some old version of the patch before
> patches were reshuffled.
>
> for some reason q35/viot test gets skipped => no table rebuild => stale DSDT.viot table
>
> reverting recently merged commit c471eb4f40, I can get q35/viot test back:
>
> commit c471eb4f40445908c1be7bb11a37ac676a0edae7
> Author: Fabiano Rosas <farosas@suse.de>
> Date:   Wed Feb 8 16:46:57 2023 -0300
>
>     tests/qtest: Check for devices in bios-tables-test
>
> so checks aren't working as expected on my host for some reason
>

Was this a legitimate skip or do we have a bug in the way we check for
devices? I.e. did you have CONFIG_VIRTIO_IOMMU in the build?

If we avoid the checks during rebuild and a device is not present in the
build anyway, then that QEMU invocation would crash.