[PATCH i-g-t] tests/intel/xe_eu_stall: Skip test on SR-IOV VF

[Bernatowicz, Marcin]

Hi Jakub,

On 5/23/2025 2:23 PM, Kolakowski, Jakub1 wrote:
> Hi Marcin,
> I was trying to align the code of the IGT test with kernel code, where we have an explicit check for SR-IOV VF.
> See https://gitlab.freedesktop.org/drm/tip/-/commit/1d622a4fe2b9a30cd4af2e858d793d05f8a82774
>
> Jakub

Still, I wonder if relying on VF checks is the best long-term approach -
we might enable the feature on a VF at some point, and the test would 
keep skipping.
What about other apps using this feature - how should they detect 
availability?
If they just try to use it on a VF, do they get ENODEV or something else?
Wouldn’t it be more robust to base this on a real feature check from 
userspace if possible?
--
marcin

> 
> -----Original Message-----
> From: Bernatowicz, Marcin <marcin.bernatowicz at linux.intel.com>
> Sent: Friday, May 23, 2025 12:40 PM
> To: Kolakowski, Jakub1 <jakub1.kolakowski at intel.com>; igt-dev at lists.freedesktop.org
> Cc: Chegondi, Harish <harish.chegondi at intel.com>; Dixit, Ashutosh <ashutosh.dixit at intel.com>; Adam Miszczak <adam.miszczak at linux.intel.com>; Laguna, Lukasz <lukasz.laguna at intel.com>
> Subject: Re: [PATCH i-g-t] tests/intel/xe_eu_stall: Skip test on SR-IOV VF
> 
> 
> 
> On 5/22/2025 7:24 PM, Jakub Kolakowski wrote:
>> EU stall sampling is not supported on SR-IOV VF. Skip the test on
>> SR-IOV VF.
>>
>> Cc: Harish Chegondi <harish.chegondi at intel.com>
>> Cc: Ashutosh Dixit <ashutosh.dixit at intel.com>
>> Cc: Adam Miszczak <adam.miszczak at linux.intel.com>
>> Cc: Lukasz Laguna <lukasz.laguna at intel.com>
>> Cc: Marcin Bernatowicz <marcin.bernatowicz at linux.intel.com>
>> Signed-off-by: Jakub Kolakowski <jakub1.kolakowski at intel.com>
>> ---
>>    tests/intel/xe_eu_stall.c | 3 ++-
>>    1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/tests/intel/xe_eu_stall.c b/tests/intel/xe_eu_stall.c
>> index 411c30871..aa94d9565 100644
>> --- a/tests/intel/xe_eu_stall.c
>> +++ b/tests/intel/xe_eu_stall.c
>> @@ -44,6 +44,7 @@
>>    
>>    #include "igt.h"
>>    #include "igt_core.h"
>> +#include "igt_sriov_device.h"
>>    #include "xe_drm.h"
>>    #include "xe/xe_ioctl.h"
>>    
>> @@ -653,7 +654,7 @@ igt_main_args("e:g:o:r:u:w:", long_options, help_str, opt_handler, NULL)
>>    		drm_fd = drm_open_driver(DRIVER_XE);
>>    		igt_require_fd(drm_fd);
>>    		devid = intel_get_drm_devid(drm_fd);
>> -		igt_require(IS_PONTEVECCHIO(devid) || intel_graphics_ver(devid) >= IP_VER(20, 0));
>> +		igt_require(!intel_is_vf_device(drm_fd) && (IS_PONTEVECCHIO(devid)
>> +|| intel_graphics_ver(devid) >= IP_VER(20, 0)));
>>    		igt_require_f(igt_get_gpgpu_fillfunc(devid), "no gpgpu-fill function\n");
>>    		igt_require_f(!stat(OBSERVATION_PARANOID, &sb), "no observation_paranoid file\n");
>>    		if (output_file) {
> 
> Don't we have another option to check the feature is unavailable (ex. an ioctl returning ENODEV) ?
>