Re: [igt-dev] [PATCH i-g-t 2/2] test/i915/gem_eio: Skip suspend test on smallbar machine

["Das, Nirmoy" <nirmoy.das@linux.intel.com>]

On 1/20/2023 4:45 PM, Matthew Auld wrote:
> On 20/01/2023 10:07, Nirmoy Das wrote:
>> Suspending machine requires backing up gem objects to
>> system memory. CPU non-visible lmem objects can only be
>> backed up with the help of GPU on a smallbar device and if
>> the gpu is wedged then such backup action will fail with -EIO.
>>
>> This test sets the gpu to wedge state before suspending the
>> machine which will always fail on smallbar machine.
>>
>> References: https://gitlab.freedesktop.org/drm/intel/-/issues/7896
>> Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
>
> For the series,
> Reviewed-by: Matthew Auld <matthew.auld@intel.com>


Thanks, Matt! Also please help merging these.

>
> I guess this would also fail on full-bar, if there were some userspace 
> lmem objects allocated during the test (due to ccs).


True that can happen. Not sure if we can do anything in that case.

Regards,

Nirmoy

>
>
>> ---
>>   tests/i915/gem_eio.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/tests/i915/gem_eio.c b/tests/i915/gem_eio.c
>> index 70e82b811..8dbaa7a75 100644
>> --- a/tests/i915/gem_eio.c
>> +++ b/tests/i915/gem_eio.c
>> @@ -500,6 +500,7 @@ static void test_wait(int fd, unsigned int flags, 
>> unsigned int wait)
>>     static void test_suspend(int fd, int state)
>>   {
>> +    igt_require(!gem_has_smallbar(fd));
>>       /* Do a suspend first so that we don't skip inside the test */
>>       igt_system_suspend_autoresume(state, SUSPEND_TEST_DEVICES);