[Intel-gfx] [PATCH 10/10] drm/i915/cnl: Fix DP max rate for Cannonlake with port F.

[Rodrigo Vivi]

On Tue, Jan 30, 2018 at 08:46:49PM +0000, Ville Syrjälä wrote:
> On Tue, Jan 30, 2018 at 12:42:12PM -0800, Rodrigo Vivi wrote:
> > On Tue, Jan 30, 2018 at 07:45:03AM +0000, Jani Nikula wrote:
> > > On Mon, 29 Jan 2018, Rodrigo Vivi <rodrigo.vivi at intel.com> wrote:
> > > > On CNL SKUs that uses port F,  max DP rate is 8.1G for all
> > > > ports when we have the elevated voltage (higher than 0.85V).
> > > >
> > > > v2: Make commit message more generic.
> > > > v3: Move conditions to a helper to get easier to read. (Ville).
> > > > v4: Add a mention to the numerical voltage on commit
> > > >     message per Manasi request.
> > > > v5: Thanks CI! "error: control reaches end of non-void function"
> > > >
> > > > Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > > > Cc: Lucas De Marchi <lucas.demarchi at intel.com>
> > > > Cc: Manasi Navare <manasi.d.navare at intel.com>
> > > > Signed-off-by: Rodrigo Vivi <rodrigo.vivi at intel.com>
> > > > Reviewed-by: Manasi Navare <manasi.d.navare at intel.com>
> > > > ---
> > > >  drivers/gpu/drm/i915/intel_dp.c | 31 ++++++++++++++++++++++++-------
> > > >  1 file changed, 24 insertions(+), 7 deletions(-)
> > > >
> > > > diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
> > > > index 86a5e8bfe2a6..1f10bdb855e7 100644
> > > > --- a/drivers/gpu/drm/i915/intel_dp.c
> > > > +++ b/drivers/gpu/drm/i915/intel_dp.c
> > > > @@ -220,15 +220,36 @@ intel_dp_downstream_max_dotclock(struct intel_dp *intel_dp)
> > > >  	return max_dotclk;
> > > >  }
> > > >  
> > > > +static int cnl_adjusted_max_rate(struct intel_dp *intel_dp, int size)
> > > > +{
> > > > +	struct intel_digital_port *dig_port = dp_to_dig_port(intel_dp);
> > > > +	struct drm_i915_private *dev_priv = to_i915(dig_port->base.base.dev);
> > > > +	enum port port = dig_port->base.port;
> > > > +
> > > > +	u32 voltage = I915_READ(CNL_PORT_COMP_DW3) & VOLTAGE_INFO_MASK;
> > > > +
> > > > +	/* Low voltage SKUs are limited to max of 5.4G */
> > > > +	if (voltage == VOLTAGE_INFO_0_85V)
> > > > +		return size - 2;
> > > > +
> > > > +	/* For this SKU 8.1G is supported in all ports */
> > > > +	if(IS_CNL_WITH_PORT_F(dev_priv))
> > > > +		return size;
> > > > +
> > > > +	/* For other SKUs, max rate on ports A and B is 5.4G */
> > > > +	if (port == PORT_A || port == PORT_D)
> > > > +		return size - 2;
> > > > +
> > > > +	return size;
> > > 
> > 
> > ops, I had missed this email. Since I had resent the series, the old one
> > was on top of my inbox.
> > 
> > > IMO this splits the ARRAY_SIZE() and the (size - 2) adjustments too
> > > much. They were tolerable within one function, but looking at this
> > > function alone, the (size - 2) is a big WTF.
> > > 
> > > I'd just put this all in the same function.
> > 
> > I just split per Ville request to make conditions more readable.
> > I now agree that size outside of the context get uglier.
> > 
> > What about changing:
> > 
> > int num_source_rates
> > const int *source_rates
> > 
> > into:
> > struct {
> > int num;
> > const int *list;
> > int max_available;
> > } source_rates;
> > 
> > So that function or whenever we need like reading from new VBT field
> > we set a max_available, and when going through the list for finding
> > the common rate instead of relying only on num we also check max_available?
> > 
> > Agree?
> > Thoughts?
> 
> I think the obvious solution is to just make this function
> return both the array and its size. Not sure there's much point
> in complicating it more than that.

fair enough. Just to confirm this simple solution you have in mind is this right:
https://pastebin.com/NcsZcfR1
?

But the point in the complicated one is to also address at the same time
the new vbt field:

u8 dp_max_link_rate:2;                                  /* 216 CNL+ */

so we would keep all the adjustments, including the VBT in a single place.

> 
> > 
> > > 
> > > BR,
> > > Jani.
> > > 
> > > 
> > > > +}
> > > > +
> > > >  static void
> > > >  intel_dp_set_source_rates(struct intel_dp *intel_dp)
> > > >  {
> > > >  	struct intel_digital_port *dig_port = dp_to_dig_port(intel_dp);
> > > >  	struct drm_i915_private *dev_priv = to_i915(dig_port->base.base.dev);
> > > > -	enum port port = dig_port->base.port;
> > > >  	const int *source_rates;
> > > >  	int size;
> > > > -	u32 voltage;
> > > >  
> > > >  	/* This should only be done once */
> > > >  	WARN_ON(intel_dp->source_rates || intel_dp->num_source_rates);
> > > > @@ -238,11 +259,7 @@ intel_dp_set_source_rates(struct intel_dp *intel_dp)
> > > >  		size = ARRAY_SIZE(bxt_rates);
> > > >  	} else if (IS_CANNONLAKE(dev_priv)) {
> > > >  		source_rates = cnl_rates;
> > > > -		size = ARRAY_SIZE(cnl_rates);
> > > > -		voltage = I915_READ(CNL_PORT_COMP_DW3) & VOLTAGE_INFO_MASK;
> > > > -		if (port == PORT_A || port == PORT_D ||
> > > > -		    voltage == VOLTAGE_INFO_0_85V)
> > > > -			size -= 2;
> > > > +		size = cnl_adjusted_max_rate(intel_dp, ARRAY_SIZE(cnl_rates));
> > > >  	} else if (IS_GEN9_BC(dev_priv)) {
> > > >  		source_rates = skl_rates;
> > > >  		size = ARRAY_SIZE(skl_rates);
> > > 
> > > -- 
> > > Jani Nikula, Intel Open Source Technology Center
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> 
> -- 
> Ville Syrjälä
> Intel OTC