[Intel-gfx] [PATCH v3] drm/i915: Restrict qgv points which don't have enough bandwidth.

[Ville Syrjälä]

On Mon, Oct 14, 2019 at 09:33:32AM -0700, James Ausmus wrote:
> On Mon, Oct 14, 2019 at 05:50:18PM +0300, Ville Syrjälä wrote:
> > On Mon, Oct 14, 2019 at 02:13:31PM +0300, Lisovskiy, Stanislav wrote:
> > > On Fri, 2019-10-11 at 16:49 -0700, James Ausmus wrote:
> > > > > +				new_qgv_points_mask |= new_mask_bit;
> > > > > +	}
> > > > > +
> > > > > +	ret = icl_pcode_restrict_qgv_points(dev_priv,
> > > > > new_qgv_points_mask);
> > > > > +	if (ret < 0)
> > > > > +		DRM_DEBUG_KMS("Could not restrict required gqv
> > > > > points(%d)\n", ret);
> > > > 
> > > > s/gqv/qgv/
> > > > 
> > > > 
> > > > Also, if we fail masking off the qgv points that can't support our BW
> > > > req, shouldn't we handle that failure somehow - maybe just disable
> > > > SAGV
> > > > entirely?  Better we lose power than have flickering screens...
> > 
> > Sounds like dead code to me. My approach is: don't deal with hw/firmware
> > failures until they are proven to exist.
> > 
> > The debug msg should be an error so that we get a bug report if this
> > ever happens.
> 
> That works - however, I think we should return the error rather than
> continuing.

No. We're way past the point of no return.

-- 
Ville Syrjälä
Intel