Re: [PATCH 2/5] drm/i915: Correctly enable blacklight adjustment via DPCD

From: Jani Nikula <jani.nikula@linux.intel.com>
To: intel-gfx@lists.freedesktop.org
Cc: Puthikorn Voravootivat <puthik@chromium.org>
Subject: Re: [PATCH 2/5] drm/i915: Correctly enable blacklight	adjustment via DPCD
Date: Thu, 09 Mar 2017 12:03:15 +0200	[thread overview]
Message-ID: <87o9xahin0.fsf@intel.com> (raw)
In-Reply-To: <20170308213053.194062-3-puthik@chromium.org>

On Wed, 08 Mar 2017, Puthikorn Voravootivat <puthik@chromium.org> wrote:
> intel_dp_aux_enable_backlight() assumed that the register
> BACKLIGHT_BRIGHTNESS_CONTROL_MODE can only has value 01
> (DP_EDP_BACKLIGHT_CONTROL_MODE_PRESET) when initialize.
>
> This patch fixed that by handling all cases of that register.
>
> Signed-off-by: Puthikorn Voravootivat <puthik@chromium.org>
> ---
>  drivers/gpu/drm/i915/intel_dp_aux_backlight.c | 29 +++++++++++++++++++++------
>  1 file changed, 23 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/intel_dp_aux_backlight.c b/drivers/gpu/drm/i915/intel_dp_aux_backlight.c
> index abfb09c1269f..16cd1265d264 100644
> --- a/drivers/gpu/drm/i915/intel_dp_aux_backlight.c
> +++ b/drivers/gpu/drm/i915/intel_dp_aux_backlight.c
> @@ -101,15 +101,32 @@ static void intel_dp_aux_enable_backlight(struct intel_connector *connector)
>  {
>  	struct intel_dp *intel_dp = enc_to_intel_dp(&connector->encoder->base);
>  	uint8_t dpcd_buf = 0;
> +	uint8_t edp_backlight_mode = 0;
>  
>  	set_aux_backlight_enable(intel_dp, true);
>  
> -	if ((drm_dp_dpcd_readb(&intel_dp->aux,
> -			       DP_EDP_BACKLIGHT_MODE_SET_REGISTER, &dpcd_buf) == 1) &&
> -	    ((dpcd_buf & DP_EDP_BACKLIGHT_CONTROL_MODE_MASK) ==
> -	     DP_EDP_BACKLIGHT_CONTROL_MODE_PRESET))
> -		drm_dp_dpcd_writeb(&intel_dp->aux, DP_EDP_BACKLIGHT_MODE_SET_REGISTER,
> -				   (dpcd_buf | DP_EDP_BACKLIGHT_CONTROL_MODE_DPCD));

That's clearly bonkers, it ends up selecting
DP_EDP_BACKLIGHT_CONTROL_MODE_PRODUCT by accident, not
DP_EDP_BACKLIGHT_CONTROL_MODE_DPCD, because it doesn't mask the bits.

> +	if (!drm_dp_dpcd_readb(&intel_dp->aux,
> +			       DP_EDP_BACKLIGHT_MODE_SET_REGISTER, &dpcd_buf)) {
> +		return;
> +	}
> +
> +	edp_backlight_mode = dpcd_buf & DP_EDP_BACKLIGHT_CONTROL_MODE_MASK;
> +
> +	switch (edp_backlight_mode) {
> +	case DP_EDP_BACKLIGHT_CONTROL_MODE_PWM:
> +	case DP_EDP_BACKLIGHT_CONTROL_MODE_PRESET:
> +		dpcd_buf &= ~DP_EDP_BACKLIGHT_CONTROL_MODE_MASK;
> +		dpcd_buf |= DP_EDP_BACKLIGHT_CONTROL_MODE_DPCD;
> +		drm_dp_dpcd_writeb(&intel_dp->aux,
> +			DP_EDP_BACKLIGHT_MODE_SET_REGISTER, dpcd_buf);
> +		break;
> +
> +	/* Do nothing in these cases */
> +	case DP_EDP_BACKLIGHT_CONTROL_MODE_DPCD:
> +	case DP_EDP_BACKLIGHT_CONTROL_MODE_PRODUCT:

Based on the current requirements for using DPCD backlight control, the
only power-on default we should end up here with is
DP_EDP_BACKLIGHT_CONTROL_MODE_PRESET.

In particular, I don't think we should accept the PRODUCT mode without
ensuring we output PWM to the eDP pins as well. And currently DPCD and
PWM controls are mutually exclusive.

BR,
Jani.

> +	default:
> +		break;
> +	}
>  }
>  
>  static void intel_dp_aux_disable_backlight(struct intel_connector *connector)

-- 
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx