[Intel-gfx] [PATCH] drm/i915: enhance legacy HPD disconnection flow for 4K pipe compute in GLK

Ville Syrjälä ville.syrjala at linux.intel.com
Fri Feb 19 15:36:05 UTC 2021


On Fri, Feb 19, 2021 at 02:25:39PM +0800, Gary C Wang wrote:
> HDMI PHY is not available to use when its HDMI disaply plug-in, and power-off
> then power-on as soon as getting a hotplug. In above cases where there's a HDMI
> connector physically connected but it can't be used by GLK with 4K pipe then blank
> screen (lacking of edid-update & mode-probing) then need return false, since the
> rest of the driver should pretty much treat the port as disconnected.

HPD live status has been notoriously unreliable on HDMI,
hence we don't use it. Dunno if we could start to trust it on
GLK.

> 
> As previous result, handshaking through is required around connect and disconnect.
> Otherwise it would be in a inconsistent state as port is disconnected but with a
> valid HDMI type.

Not sure what handshaking you're talking about. HDMI detection
itself only uses GMBUS so it has no relationship to any PHY stuff.

Anyway this looks a bit like duct tape that happens to avoid some
unknown sequence of events that cause a black screen. I suspect
it may have something to do with the scrambling/bit clock ratio
stuff. I'll comment on the bug a bit...

> 
> Also setting it to return HDMI disconnect for any future calls to
> intel_digital_port_connected(), this way we don't need to check if port is marked
> as safe everytime.
> 
> References: https://gitlab.freedesktop.org/drm/intel/-/issues/3092
> Test-steps: setup HDMI 4K at 60Hz in GLK then to power monitor off then on to get display
> recovery correctly
> 
> Cc: Gordon Sylin <gordon.sylin at intel.com>
> Tested-by: Gary C Wang <gary.c.wang at intel.com>
> Reviewed-by: Gordon Sylin <gordon.sylin at intel.com>
> Signed-off-by: Gary C Wang <gary.c.wang at intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_hdmi.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_hdmi.c b/drivers/gpu/drm/i915/display/intel_hdmi.c
> index 7f384f259fc8..039cdbfe71a0 100644
> --- a/drivers/gpu/drm/i915/display/intel_hdmi.c
> +++ b/drivers/gpu/drm/i915/display/intel_hdmi.c
> @@ -2705,7 +2705,7 @@ intel_hdmi_detect(struct drm_connector *connector, bool force)
>  
>  	wakeref = intel_display_power_get(dev_priv, POWER_DOMAIN_GMBUS);
>  
> -	if (INTEL_GEN(dev_priv) >= 11 &&
> +	if ((INTEL_GEN(dev_priv) >= 11 || IS_GEMINILAKE(dev_priv)) &&
>  	    !intel_digital_port_connected(encoder))
>  		goto out;
>  
> -- 
> 2.17.1

-- 
Ville Syrjälä
Intel


More information about the Intel-gfx mailing list