[Intel-gfx] [PATCH 6/6] drm/i915: Ignore LFP2 for now

[Ville Syrjälä]

On Wed, Nov 09, 2022 at 01:16:49PM +0200, Ville Syrjala wrote:
> From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> 
> There are tons of ADL machines in the wild where the VBT has
> two child devices (eDP and HDMI) for port B. Our code can't
> handle that as the whole parse_ddi_port() stuff assumes a
> single child device per port. So even if we fix the PPS issues
> around the (typically) non-present second eDP port we still
> won't be able to use the HDMI port after the eDP init fails.
> 
> The easiest way to get the HDMI port operational is to just
> ignore the second eDP child device entirely. Dual eDP systems
> should be exceedingly rare anyway.

One thing I just realized when staring at the logs is that
the BIOS seems to leave the VDD on for the second PPS. So
with this patch we'll never actually turn it off :/

> 
> Cc: Animesh Manna <animesh.manna at intel.com>
> Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_bios.c | 31 ++++++++++++++++++++---
>  1 file changed, 27 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_bios.c b/drivers/gpu/drm/i915/display/intel_bios.c
> index 64f927f6479d..1011ccc2836b 100644
> --- a/drivers/gpu/drm/i915/display/intel_bios.c
> +++ b/drivers/gpu/drm/i915/display/intel_bios.c
> @@ -2705,10 +2705,33 @@ static void parse_ddi_port(struct intel_bios_encoder_data *devdata)
>  	}
>  
>  	if (i915->display.vbt.ports[port]) {
> -		drm_dbg_kms(&i915->drm,
> -			    "More than one child device for port %c in VBT, using the first.\n",
> -			    port_name(port));
> -		return;
> +		const struct child_device_config *other_child =
> +			&i915->display.vbt.ports[port]->child;
> +
> +		/*
> +		 * FIXME: Temporary hack for many ADL machines where
> +		 * the VBT declares two eDPs, the second of which
> +		 * conflicts with an external HDMI port. Ideally
> +		 * we should fall back to consulting the HDMI port
> +		 * child device after the eDP init fails, but that
> +		 * does not work with the current port based
> +		 * i915->display.vbt.ports[] stuff.
> +		 *
> +		 * Hopefully this does not conflict with any other
> +		 * bogus VBT case...
> +		 */
> +		if (other_child->handle == DEVICE_HANDLE_LFP2 &&
> +		    child->handle != DEVICE_HANDLE_LFP2) {
> +			drm_dbg_kms(&i915->drm,
> +				    "More than one child device for port %c in VBT, using the last.\n",
> +				    port_name(port));
> +			i915->display.vbt.ports[port] = NULL;
> +		} else {
> +			drm_dbg_kms(&i915->drm,
> +				    "More than one child device for port %c in VBT, using the first.\n",
> +				    port_name(port));
> +			return;
> +		}
>  	}
>  
>  	sanitize_device_type(devdata, port);
> -- 
> 2.37.4

-- 
Ville Syrjälä
Intel