[Intel-gfx] [PATCH 5/5] drm/i915: use the SW-based pw->hw_enabled check instead of reading registers

[Imre Deak]

On Mon, Aug 20, 2018 at 04:31:39PM -0700, Paulo Zanoni wrote:
> I can't find a reason why we would want to call is_enabled(), which
> does a register read, instead of just relying on our tracking with
> hw_enabled. Let's try to trust our hardware sync.

The software state is indirect, so not sure why we should check that
instead. The register read doesn't have considerable overhead and it's
not on a speed critical path anyway.

> 
> Cc: Imre Deak <imre.deak at intel.com>
> Requested-by: José Roberto de Souza <jose.souza at intel.com>
> Signed-off-by: Paulo Zanoni <paulo.r.zanoni at intel.com>
> ---
>  drivers/gpu/drm/i915/intel_runtime_pm.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/intel_runtime_pm.c b/drivers/gpu/drm/i915/intel_runtime_pm.c
> index f38a049861e6..76bb2e06fef1 100644
> --- a/drivers/gpu/drm/i915/intel_runtime_pm.c
> +++ b/drivers/gpu/drm/i915/intel_runtime_pm.c
> @@ -670,9 +670,8 @@ static void assert_can_enable_dc5(struct drm_i915_private *dev_priv)
>  {
>  	struct i915_power_well *pg2 = lookup_power_well(dev_priv,
>  							SKL_DISP_PW_2);
> -	bool pg2_enabled = pg2->desc->ops->is_enabled(dev_priv, pg2);
>  
> -	WARN_ONCE(pg2_enabled, "PG2 not disabled to enable DC5.\n");
> +	WARN_ONCE(pg2->hw_enabled, "PG2 not disabled to enable DC5.\n");
>  
>  	WARN_ONCE((I915_READ(DC_STATE_EN) & DC_STATE_EN_UPTO_DC5),
>  		  "DC5 already programmed to be enabled.\n");
> -- 
> 2.14.4
>