Re: [PATCH i-g-t] tests/kms_getfb: Skip getfb2-accept-ccs subtest on Xe2 platforms

[Karthik B S <karthik.b.s@intel.com>]

On 6/26/2024 2:13 PM, Juha-Pekka Heikkila wrote:
> On 26.6.2024 8.43, Karthik B S wrote:
>>
>> On 6/21/2024 10:02 AM, Pranay Samala wrote:
>>> The getfb2-accept-ccs subtest tests on ccs modifiers.
>>> With Xe2 platforms we no more have the ccs modifiers.
>>>
>>> Adding a check to verify the platform. If it is
>>> Xe2 platform then test will skip, otherwise it
>>> will execute further.
>>>
>>> Signed-off-by: Pranay Samala<pranay.samala@intel.com>
>>> ---
>>>   tests/kms_getfb.c | 3 +++
>>>   1 file changed, 3 insertions(+)
>>>
>>> diff --git a/tests/kms_getfb.c b/tests/kms_getfb.c
>>> index bb9679bc9..81829ca3b 100644
>>> --- a/tests/kms_getfb.c
>>> +++ b/tests/kms_getfb.c
>>> @@ -161,6 +161,9 @@ static void get_ccs_fb(int fd, struct 
>>> drm_mode_fb_cmd2 *ret)
>>>       igt_require(has_addfb2_iface(fd));
>>>       devid = intel_get_drm_devid(fd);
>>>   +    igt_require_f(intel_graphics_ver(devid) < 20,
>>> +              "No ccs modifiers on Xe2\n");
>>> +
>>
>> Hi Pranay,
>>
>> I see that getfb tests are skipping on legacy platforms with this patch.
>> https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_11292/index.html?testfilter=kms_getfb 
>>
>>
>> This should be resolved by replacing 'intel_graphics_ver' with 
>> 'intel_display_ver'. Could you please update this.
>
> intel_display_ver will not help here, that would lead to detecting 
> each machine individually because xe2 vs older platforms have 
> overlapping display versions.
Ah right, this won't work on xe2 platforms with overlapping display 
versions.
>
> Seems Pranay's patch is almost right. That comparison probably should 
> be something like intel_graphics_ver(devid) < IP_VER(20, 0)

Agreed, this should fix to avoid the skips on older platforms. Thanks JP 
for the input.

Thanks,
Karthik.B.S
>
> /Juha-Pekka