Re: [PATCH i-g-t v1] kms/vrr: Trigger modeset for non-preferred override mode

["Naladala, Ramanaidu" <Ramanaidu.naladala@intel.com>]

Hi Jevan,

Thanks for the feedback.

On 7/13/2026 9:05 AM, B, Jeevan wrote:
>> -----Original Message-----
>> From: igt-dev <igt-dev-bounces@lists.freedesktop.org> On Behalf Of Naladala
>> Ramanaidu
>> Sent: Saturday, July 4, 2026 10:49 AM
>> To: igt-dev@lists.freedesktop.org
>> Cc: Golani, Mitulkumar Ajitkumar <mitulkumar.ajitkumar.golani@intel.com>;
>> Nautiyal, Ankit K <ankit.k.nautiyal@intel.com>; Naladala, Ramanaidu
>> <ramanaidu.naladala@intel.com>
>> Subject: [PATCH i-g-t v1] kms/vrr: Trigger modeset for non-preferred override
>> mode
> Need to change kms/vrr to tests/kms_vrr
>> Ensure a modeset is triggered when overriding to a mode that is not marked as
>> PREFERRED or DRIVER. This guarantees that the new refresh rate mode is
>> properly programmed into hardware instead of relying on an existing state.
>>
>> Observed fastset failures when the preferred mode has a low clock and
>> switching to a higher clock mode via fastset, which led to test failures. Forcing
>> an atomic commit in such cases avoids these issues.
>>
>> Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/7397
>>
>> Signed-off-by: Naladala Ramanaidu <ramanaidu.naladala@intel.com>
>> ---
>>   tests/kms_vrr.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/tests/kms_vrr.c b/tests/kms_vrr.c index 82eb8554a..d583d3cc3
>> 100644
>> --- a/tests/kms_vrr.c
>> +++ b/tests/kms_vrr.c
>> @@ -948,6 +948,10 @@ static bool output_constraint(data_t *data,
>> igt_output_t *output, uint32_t flags
>>   	data->range.max = data->switch_modes[HIGH_RR_MODE].vrefresh;
>>   	igt_output_override_mode(output, &data-
>>> switch_modes[HIGH_RR_MODE]);
>> +	if (!(data->switch_modes[HIGH_RR_MODE].type &
>> +	    (DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER)))
>> +			igt_display_commit2(&data->display,
>> COMMIT_ATOMIC);
>> +
> Can the modeset enforcement be moved to the actual mode-switch/commit path instead of committing from output_constraint()?
> That would avoid introducing side effects in what currently appears to be a constraint-discovery helper. ??
Yes possible. I will fix the changes in the next revision.
>
>>   	/* Search for a low refresh rate mode. */
>>   	if (!(flags & (TEST_SEAMLESS_VRR | TEST_SEAMLESS_DRRS |
>> TEST_LINK_OFF)))
>>   		return true;
>> --
>> 2.43.0