[Intel-gfx] [PATCH 39/49] drm/i915/chv: use vlv_PLL_is_optimal in chv_find_best_dpll

[Jesse Barnes]

On 03/19/2015 01:55 PM, Imre Deak wrote:
> On Thu, 2015-03-19 at 13:34 -0700, Jesse Barnes wrote:
>> On 03/17/2015 02:40 AM, Imre Deak wrote:
>>> Prepare chv_find_best_dpll to be used for BXT too, where we want to
>>> consider the error between target and calculated frequency too when
>>> choosing a better PLL configuration.
>>>
>>> No functional change.
>>>
>>> Signed-off-by: Imre Deak <imre.deak at intel.com>
>>> ---
>>>  drivers/gpu/drm/i915/intel_display.c | 26 ++++++++++++++++++++------
>>>  1 file changed, 20 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
>>> index 5874512..9ca84a2 100644
>>> --- a/drivers/gpu/drm/i915/intel_display.c
>>> +++ b/drivers/gpu/drm/i915/intel_display.c
>>> @@ -786,6 +786,16 @@ static bool vlv_PLL_is_optimal(struct drm_device *dev, int target_freq,
>>>  			       unsigned int best_error_ppm,
>>>  			       unsigned int *error_ppm)
>>>  {
>>> +	/*
>>> +	 * For CHV ignore the error and consider only the P value.
>>> +	 * Prefer a bigger P value based on HW requirements.
>>> +	 */
>>> +	if (IS_CHERRYVIEW(dev)) {
>>> +		*error_ppm = 0;
>>> +
>>> +		return calculated_clock->p > best_clock->p;
>>> +	}
>>> +
>>>  	if (WARN_ON_ONCE(!target_freq))
>>>  		return false;
>>>  
>>> @@ -864,11 +874,13 @@ chv_find_best_dpll(const intel_limit_t *limit, struct intel_crtc *crtc,
>>>  		   intel_clock_t *best_clock)
>>>  {
>>>  	struct drm_device *dev = crtc->base.dev;
>>> +	unsigned int best_error_ppm;
>>>  	intel_clock_t clock;
>>>  	uint64_t m2;
>>>  	int found = false;
>>>  
>>>  	memset(best_clock, 0, sizeof(*best_clock));
>>> +	best_error_ppm = 1000000;
>>>  
>>>  	/*
>>>  	 * Based on hardware doc, the n always set to 1, and m1 always
>>> @@ -882,6 +894,7 @@ chv_find_best_dpll(const intel_limit_t *limit, struct intel_crtc *crtc,
>>>  		for (clock.p2 = limit->p2.p2_fast;
>>>  				clock.p2 >= limit->p2.p2_slow;
>>>  				clock.p2 -= clock.p2 > 10 ? 2 : 1) {
>>> +			unsigned int error_ppm;
>>>  
>>>  			clock.p = clock.p1 * clock.p2;
>>>  
>>> @@ -898,12 +911,13 @@ chv_find_best_dpll(const intel_limit_t *limit, struct intel_crtc *crtc,
>>>  			if (!intel_PLL_is_valid(dev, limit, &clock))
>>>  				continue;
>>>  
>>> -			/* based on hardware requirement, prefer bigger p
>>> -			 */
>>> -			if (clock.p > best_clock->p) {
>>> -				*best_clock = clock;
>>> -				found = true;
>>> -			}
>>> +			if (!vlv_PLL_is_optimal(dev, target, &clock, best_clock,
>>> +						best_error_ppm, &error_ppm))
>>> +				continue;
>>> +
>>> +			*best_clock = clock;
>>> +			best_error_ppm = error_ppm;
>>> +			found = true;
>>>  		}
>>>  	}
>>>  
>>>
>>
>> Looking at it again, maybe vlv_PLL_is_better() might be a better name.
> 
> Ok, will change it.
> 
>> Also, could you just make the ppm variable a scratch one and ignore it?
>>  It just gets set to 0 no matter what, right?
> 
> For CHV yes. But the next patch takes the same function into use in BXT
> too, where error_ppm will be set to the actual error, the same way as on
> VLV.

Oh right it's a patch *series*.  Ignore the noise. :)

The rename is optional too; I just had to think about it when I saw
>>> +		return calculated_clock->p > best_clock->p;
since in that case we're really checking whether the new calculated p
value is better (higher) than the last one.  The _is_optimal() name made
me think it should be >= or something.  But it's not a big deal.

Thanks,
Jesse