[igt-dev] [PATCH i-g-t] tests/perf: Fix B0 counter check in gen12-mi-rpc

[Lionel Landwerlin]

On 22/04/2020 18:47, Umesh Nerlige Ramappa wrote:
> On Tue, Apr 21, 2020 at 11:06:27PM +0300, Lionel Landwerlin wrote:
>> On 21/04/2020 21:22, Umesh Nerlige Ramappa wrote:
>>> The gen12-mi-rpc test checks for B0 counter to be zero, however,
>>> occassionally the counter reads a non-zero value. To ensure the test
>>> works as expected, we check B0 with the preset value.
>>
>>
>> What the value being reported?
>>
>>
>> Sounds like we should ask around see if there isn't a deeper problem 
>> with the HW configuration.
>>
>
> I have seen values 0x121 and 0x1ec. The issue occurs sometimes on 
> reboots. Subsequent runs I see that the counter is 0.
>
> Thanks,
> Umesh


Oh... I thought this test was using sanity_check_reports() on deltas 
between 2 reports.

Yeah I guess your change is fine then.

The test could be improved with sanity_check_reports.


-Lionel


>
>>
>> -Lionel
>>
>>
>>>
>>> Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa at intel.com>
>>> Fixes: https://gitlab.freedesktop.org/drm/intel/-/issues/1085
>>> ---
>>>  tests/perf.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/tests/perf.c b/tests/perf.c
>>> index b7d5e50e..9d0ed45c 100644
>>> --- a/tests/perf.c
>>> +++ b/tests/perf.c
>>> @@ -3002,7 +3002,7 @@ gen12_test_mi_rpc(void)
>>>       */
>>>      igt_assert_eq(report32[0], REPORT_ID);
>>>      igt_assert_neq(report32[1], 0);
>>> -    igt_assert_eq(report32[format.b_off >> 2], 0);
>>> +    igt_assert_neq(report32[format.b_off >> 2], 0x80808080);
>>>      igt_assert_eq(report32[format_size_32], 0x80808080);
>>>      drm_intel_bo_unmap(bo);
>>
>>