[PATCH 24/35] drm/i915/vrr: Adjust Vtotal for MSA for fixed timings

[Nautiyal, Ankit K]

On 1/25/2025 2:31 AM, Ville Syrjälä wrote:
> On Fri, Jan 24, 2025 at 08:30:09PM +0530, Ankit Nautiyal wrote:
>> DP sink uses MSA timings for the fixed refresh rate mode.
>> For using VRR timing generator for fixed refresh rate mode, the HW prepares
>> the Vtotal for the MSA from the VMAX register. Since the MSA Vtotal is
>> one-based while Vmax is zero-based we need to take care of this while
>> setting Vtotal value for the fixed refresh rate mode.
>>
>> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal at intel.com>
>> ---
>>   drivers/gpu/drm/i915/display/intel_vrr.c | 14 +++++++++++++-
>>   1 file changed, 13 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/i915/display/intel_vrr.c b/drivers/gpu/drm/i915/display/intel_vrr.c
>> index f485c3e35d68..5a1b0d61c483 100644
>> --- a/drivers/gpu/drm/i915/display/intel_vrr.c
>> +++ b/drivers/gpu/drm/i915/display/intel_vrr.c
>> @@ -264,6 +264,18 @@ void intel_vrr_compute_vrr_timings(struct intel_crtc_state *crtc_state, int vmin
>>   	crtc_state->mode_flags |= I915_MODE_FLAG_VRR;
>>   }
>>   
>> +static
>> +int adjust_vtotal_for_msa(int vtotal)
>> +{
>> +	/*
>> +	 * DP sink uses MSA timings for the fixed refresh rate mode.
>> +	 * The HW prepares the Vtotal for the MSA from the VMAX register.
>> +	 * Since the MSA Vtotal is one-based while Vmax is zero-based we need to
>> +	 * take care of this while setting Vtotal value.
> Hang on. If the hardware is broken enough that it forgets to put the +1
> into MSA then that means we should be increasing vmax if we want the
> proper number to appear there. But that would actually just increase
> the length of the frame and the MSA would still be off by one.

I was observing blanking on the DP panel, when vmin,vmax,flipline were 
set to vtotal and MSA_TIMING_PAR_IGNORE_EN set to 0.


When I had tested with DP analyser, the MSA seemed to be invalid as the 
timing taken from VMAX register was 1 more than the required value.

After setting VMIN VMAX FLIPLINE with vtotal -1 the issue was no more 
seen on DP panel.


But we are already writing VMAX -1 into the register. I'll check again, 
if I missed something else and get back on this.


Regards,

Ankit


>> +	 */
>> +	return vtotal - 1;
>> +}
>> +
>>   static
>>   int intel_vrr_compute_vmin(struct intel_crtc_state *crtc_state)
>>   {
>> @@ -274,7 +286,7 @@ int intel_vrr_compute_vmin(struct intel_crtc_state *crtc_state)
>>   	 * For this set the vmin as crtc_vtotal. With this we never need to
>>   	 * change anything to do with the guardband.
>>   	 */
>> -	return crtc_state->hw.adjusted_mode.crtc_vtotal;
>> +	return adjust_vtotal_for_msa(crtc_state->hw.adjusted_mode.crtc_vtotal);
>>   }
>>   
>>   static
>> -- 
>> 2.45.2