[Intel-gfx] [PATCH] drm/i915: Move the min_pixclk[] handling to the end of readout

[Ville Syrjälä]

On Thu, Dec 22, 2016 at 02:10:25PM +0100, Maarten Lankhorst wrote:
> Op 20-12-16 om 16:39 schreef ville.syrjala at linux.intel.com:
> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> >
> > Trying to determine the pixel rate of the pipe can't be done until we
> > know the clock, which means it can't be done until the encoder
> > .get_config() hooks have been called. So let's move the min_pixclk[]
> > stuff to the end of intel_modeset_readout_hw_state() when we actually
> > have gathered all the required infromation.
> >
> > Cc: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>
> > Cc: Mika Kahola <mika.kahola at intel.com>
> > Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira at intel.com>
> > Fixes: 565602d7501a ("drm/i915: Do not acquire crtc state to check clock during modeset, v4.")
> > Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > ---
> >  drivers/gpu/drm/i915/intel_display.c | 32 ++++++++++++++++----------------
> >  1 file changed, 16 insertions(+), 16 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> > index ef5dde5ab1cf..d8effd4da034 100644
> > --- a/drivers/gpu/drm/i915/intel_display.c
> > +++ b/drivers/gpu/drm/i915/intel_display.c
> > @@ -17033,7 +17033,6 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)
> >  
> >  	for_each_intel_crtc(dev, crtc) {
> >  		struct intel_crtc_state *crtc_state = crtc->config;
> > -		int pixclk = 0;
> >  
> >  		__drm_atomic_helper_crtc_destroy_state(&crtc_state->base);
> >  		memset(crtc_state, 0, sizeof(*crtc_state));
> > @@ -17045,23 +17044,9 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)
> >  		crtc->base.enabled = crtc_state->base.enable;
> >  		crtc->active = crtc_state->base.active;
> >  
> > -		if (crtc_state->base.active) {
> > +		if (crtc_state->base.active)
> >  			dev_priv->active_crtcs |= 1 << crtc->pipe;
> >  
> > -			if (INTEL_GEN(dev_priv) >= 9 || IS_BROADWELL(dev_priv))
> > -				pixclk = ilk_pipe_pixel_rate(crtc_state);
> > -			else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
> > -				pixclk = crtc_state->base.adjusted_mode.crtc_clock;
> > -			else
> > -				WARN_ON(dev_priv->display.modeset_calc_cdclk);
> > -
> > -			/* pixel rate mustn't exceed 95% of cdclk with IPS on BDW */
> > -			if (IS_BROADWELL(dev_priv) && crtc_state->ips_enabled)
> > -				pixclk = DIV_ROUND_UP(pixclk * 100, 95);
> > -		}
> > -
> > -		dev_priv->min_pixclk[crtc->pipe] = pixclk;
> > -
> >  		readout_plane_state(crtc);
> >  
> >  		DRM_DEBUG_KMS("[CRTC:%d:%s] hw state readout: %s\n",
> > @@ -17134,6 +17119,8 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)
> >  	}
> >  
> >  	for_each_intel_crtc(dev, crtc) {
> > +		int pixclk = 0;
> > +
> >  		crtc->base.hwmode = crtc->config->base.adjusted_mode;
> >  
> >  		memset(&crtc->base.mode, 0, sizeof(crtc->base.mode));
> > @@ -17161,10 +17148,23 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)
> >  			 */
> >  			crtc->base.state->mode.private_flags = I915_MODE_FLAG_INHERITED;
> >  
> > +			if (INTEL_GEN(dev_priv) >= 9 || IS_BROADWELL(dev_priv))
> > +				pixclk = ilk_pipe_pixel_rate(crtc->config);
> > +			else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
> > +				pixclk = crtc->config->base.adjusted_mode.crtc_clock;
> > +			else
> > +				WARN_ON(dev_priv->display.modeset_calc_cdclk);
> > +
> > +			/* pixel rate mustn't exceed 95% of cdclk with IPS on BDW */
> > +			if (IS_BROADWELL(dev_priv) && crtc->config->ips_enabled)
> > +				pixclk = DIV_ROUND_UP(pixclk * 100, 95);
> > +
> >  			drm_calc_timestamping_constants(&crtc->base, &crtc->base.hwmode);
> >  			update_scanline_offset(crtc);
> >  		}
> >  
> > +		dev_priv->min_pixclk[crtc->pipe] = pixclk;
> > +
> >  		intel_pipe_config_sanity_check(dev_priv, crtc->config);
> >  	}
> >  }
> 
> Looks good. Could you do intel_crtc_state = to_intel_crtc_state(crtc->base.state) at the start
> and fix all ->state and ->config dereferences in this loop too?
> This is a cleanup I'm trying to apply to the whole driver. Atomic state should be
> passed in where possible, and crtc->config/state dereferences should be avoided as much as we can.
> Eventually I want to get rid of crtc->config altogether. :)

I posted that as a followup.

> 
> Reviewed-by: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>

Patch pushed to dinq. Thanks for the reviews.

-- 
Ville Syrjälä
Intel OTC