[Intel-gfx] [PATCH 1/2] drm/i915: Fix g4x+ sprite dotclock limit for upscaling

[Ville Syrjälä]

On Fri, Sep 11, 2020 at 09:03:36PM +0300, Jani Nikula wrote:
> On Thu, 06 Feb 2020, Ville Syrjala <ville.syrjala at linux.intel.com> wrote:
> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> >
> > Even if we're not doing downscaling we should account for
> > some of the extra dotclock limitations for g4x+ sprites. In
> > particular we must never exceed the 90% rule, and with RGB
> > that limits actually drops to 80%.
> >
> > So instead of bailing out when upscaling let's clamp the
> > scaling factor appropriately and go through the rest of
> > calculation normally. By luck we already did the full
> > calculations for the 1:1 case.
> >
> > Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_sprite.c | 3 +--
> >  1 file changed, 1 insertion(+), 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_sprite.c b/drivers/gpu/drm/i915/display/intel_sprite.c
> > index 7abeefe8dce5..6e2e22d9bbaa 100644
> > --- a/drivers/gpu/drm/i915/display/intel_sprite.c
> > +++ b/drivers/gpu/drm/i915/display/intel_sprite.c
> > @@ -1611,8 +1611,7 @@ static int g4x_sprite_min_cdclk(const struct intel_crtc_state *crtc_state,
> >  	hscale = drm_rect_calc_hscale(&plane_state->uapi.src,
> >  				      &plane_state->uapi.dst,
> >  				      0, INT_MAX);
> > -	if (hscale < 0x10000)
> > -		return pixel_rate;
> > +	hscale = max(hscale, 0x10000u);
> 
> It bugs me that drm_rect seems to be used for both integer and 16.16
> fixed point and whatnot.

In theory it can use any arbitrary precision. There are a few
functions which do assume .0 or .16 though.

> 
> It also gives me an uneasy feeling that hscale is uint while
> drm_rect_calc_hscale() may return -ERANGE... but I guess shouldn't
> happen.

Yeah, should not happen since we've already done this
same calculation (+check for <0) earlier. I've occasionally
pondered about stashing the h/vscale from that first check
into the plane state so we wouldn't have to redo it here.
But I never found sufficient motivation to actually do it.

> 
> All in all,
> 
> Reviewed-by: Jani Nikula <jani.nikula at intel.com>

Ta.

> 
> 
> >  
> >  	/* Decimation steps at 2x,4x,8x,16x */
> >  	decimate = ilog2(hscale >> 16);
> 
> -- 
> Jani Nikula, Intel Open Source Graphics Center

-- 
Ville Syrjälä
Intel