Re: [Intel-gfx] [PATCH 3/3] drm/i915: Pick the backlight controller based on VBT on ICP+

[Rodrigo Vivi <rodrigo.vivi@intel.com>]

On Tue, Feb 07, 2023 at 08:43:37AM +0200, Ville Syrjala wrote:
> From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> 
> Use the second backlight controller on ICP+ if the VBT asks
> us to do so.
> 
> On pre-MTP we also check the chicken bit to make sure the
> pins have been correctly muxed by the firmware.
> 

It looks like CC: stable was added while merging this patch.
But it doesn't go clean. build fails due to s/dev_priv/i915
and also due to the lack of ICP_SECOND_PPS_IO_SELECT that
was added by another patch.

So we need a backported version of this patch to be included
in the stable trees... how far we will go in the stable tree?

At this point of -rc7 I even wonder it will be better to stay
for 6.2 to get released and then send the backported version
to the stable ml directly...

> Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/8016
> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> ---
>  .../gpu/drm/i915/display/intel_backlight.c    | 34 +++++++++++++++++--
>  1 file changed, 31 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_backlight.c b/drivers/gpu/drm/i915/display/intel_backlight.c
> index 5b7da72c95b8..a4e4b7f79e4d 100644
> --- a/drivers/gpu/drm/i915/display/intel_backlight.c
> +++ b/drivers/gpu/drm/i915/display/intel_backlight.c
> @@ -1431,6 +1431,30 @@ bxt_setup_backlight(struct intel_connector *connector, enum pipe unused)
>  	return 0;
>  }
>  
> +static int cnp_num_backlight_controllers(struct drm_i915_private *i915)
> +{
> +	if (INTEL_PCH_TYPE(i915) >= PCH_DG1)
> +		return 1;
> +
> +	if (INTEL_PCH_TYPE(i915) >= PCH_ICP)
> +		return 2;
> +
> +	return 1;
> +}
> +
> +static bool cnp_backlight_controller_is_valid(struct drm_i915_private *i915, int controller)
> +{
> +	if (controller < 0 || controller >= cnp_num_backlight_controllers(i915))
> +		return false;
> +
> +	if (controller == 1 &&
> +	    INTEL_PCH_TYPE(i915) >= PCH_ICP &&
> +	    INTEL_PCH_TYPE(i915) < PCH_MTP)
> +		return intel_de_read(i915, SOUTH_CHICKEN1) & ICP_SECOND_PPS_IO_SELECT;
> +
> +	return true;
> +}
> +
>  static int
>  cnp_setup_backlight(struct intel_connector *connector, enum pipe unused)
>  {
> @@ -1440,10 +1464,14 @@ cnp_setup_backlight(struct intel_connector *connector, enum pipe unused)
>  
>  	/*
>  	 * CNP has the BXT implementation of backlight, but with only one
> -	 * controller. TODO: ICP has multiple controllers but we only use
> -	 * controller 0 for now.
> +	 * controller. ICP+ can have two controllers, depending on pin muxing.
>  	 */
> -	panel->backlight.controller = 0;
> +	panel->backlight.controller = connector->panel.vbt.backlight.controller;
> +	if (!cnp_backlight_controller_is_valid(i915, panel->backlight.controller)) {
> +		drm_dbg_kms(&i915->drm, "Invalid backlight controller %d, assuming 0\n",
> +			    panel->backlight.controller);
> +		panel->backlight.controller = 0;
> +	}
>  
>  	pwm_ctl = intel_de_read(i915,
>  				BXT_BLC_PWM_CTL(panel->backlight.controller));
> -- 
> 2.39.1
>