Re: [PATCH v3] drm/i915/hdmi: preserve cached EDID on DDC failure when connector is forced

[Jani Nikula <jani.nikula@linux.intel.com>]

On Sat, 24 Oct 2026, Aman Sharma <amansharma6122002@gmail.com> wrote:
> When an HDMI connector is forced enabled (e.g. via video=HDMI-A-1:e for
> laptop clamshell operation) and the external monitor enters DPMS standby,
> some monitors (such as the BenQ GW2790Q) power down their internal DDC/I2C
> controller, causing SCDC/EDID reads to fail with -ENXIO (-6).
>
> Currently, intel_hdmi_detect() and intel_hdmi_force() unconditionally call
> intel_hdmi_unset_edid() before calling intel_hdmi_set_edid(). When
> drm_edid_read_ddc() returns NULL during monitor standby, the connector's
> cached detect_edid is wiped out and
> drm_helper_probe_single_connector_modes() falls back to the 1024x768 VESA
> no-EDID mode. When the monitor wakes back up, because the connector status
> remained forced to connector_status_connected, no hotplug state transition
> occurs and the display remains stuck at 1024x768 until a manual mode probe
> (such as toggling the laptop lid) is triggered.
>
> Fix this by moving intel_hdmi_unset_edid() into intel_hdmi_set_edid() after
> checking whether a forced-on connector (DRM_FORCE_ON or
> DRM_FORCE_ON_DIGITAL) already has a valid cached EDID (detect_edid and
> edid_blob_ptr) when DDC fails, and preserving connector_status_connected
> on Gen11+ digital port disconnect when a forced-on connector has a cached
> EDID. When the monitor is awake or a different monitor is plugged in,
> drm_edid_read_ddc() succeeds, resets the previous EDID and dp_dual_mode
> state via intel_hdmi_unset_edid(), and updates the connector with the new
> EDID.

Connector forcing and EDID reading/caching need to remain
orthogonal. The patch at hand makes a complex thing more complex.

Please file a bug as described at [1], reproducing the problem.

[1] https://drm.pages.freedesktop.org/intel-docs/how-to-file-i915-bugs.html

> v3:
> - Explicitly check for DRM_FORCE_ON / DRM_FORCE_ON_DIGITAL so DRM_FORCE_OFF
>   still clears the cached EDID as expected.
> - Report connector_status_connected on Gen11+ disconnect when a forced-on
>   connector retains its cached EDID, and jump to out in
>   intel_hdmi_set_edid() so CEC physical address remains valid and
>   drm_helper_probe_detect_ctx() does not increment epoch_counter on
>   standby HPD pulses.
>
> v2:
> - Move intel_hdmi_unset_edid() into intel_hdmi_set_edid() so hotplug
>   detection via intel_hdmi_detect() also preserves the cached EDID on
>   forced connectors and resets dp_dual_mode when updating EDID.
> - Clear detect_edid on Gen11+ disconnect when the connector is unforced,
>   and verify edid_blob_ptr is non-NULL before retaining detect_edid.
>
> Signed-off-by: Aman Sharma <amansharma6122002@gmail.com>
> ---
>  drivers/gpu/drm/i915/display/intel_hdmi.c | 31 ++++++++++++++++++-----
>  1 file changed, 25 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_hdmi.c b/drivers/gpu/drm/i915/display/intel_hdmi.c
> index b6fa1a1558b8..a14a857dfe94 100644
> --- a/drivers/gpu/drm/i915/display/intel_hdmi.c
> +++ b/drivers/gpu/drm/i915/display/intel_hdmi.c
> @@ -2570,6 +2570,21 @@ intel_hdmi_set_edid(struct drm_connector *_connector)
>  		intel_gmbus_force_bit(ddc, false);
>  	}
>  
> +	/*
> +	 * If the connector is forced on (e.g. video=HDMI-A-1:e) and the sink's
> +	 * DDC/I2C bus is powered down during DPMS standby, keep the previously
> +	 * cached EDID rather than wiping it and falling back to 1024x768.
> +	 */
> +	if (!drm_edid &&
> +	    (connector->base.force == DRM_FORCE_ON ||
> +	     connector->base.force == DRM_FORCE_ON_DIGITAL) &&
> +	    connector->detect_edid && connector->base.edid_blob_ptr) {

Drivers also have no business looking at edid_blob_ptr.

BR,
Jani.

> +		connected = true;
> +		goto out;
> +	}
> +
> +	intel_hdmi_unset_edid(&connector->base);
> +
>  	/* Below we depend on display info having been updated */
>  	drm_edid_connector_update(&connector->base, drm_edid);
>  
> @@ -2580,6 +2595,7 @@ intel_hdmi_set_edid(struct drm_connector *_connector)
>  		connected = true;
>  	}
>  
> +out:
>  	intel_display_power_put(display, POWER_DOMAIN_GMBUS, wakeref);
>  
>  	cec_notifier_set_phys_addr(intel_hdmi->cec_notifier,
> @@ -2611,10 +2627,15 @@ intel_hdmi_detect(struct drm_connector *_connector, bool force)
>  	wakeref = intel_display_power_get(display, POWER_DOMAIN_GMBUS);
>  
>  	if (DISPLAY_VER(display) >= 11 &&
> -	    !intel_digital_port_connected(encoder))
> +	    !intel_digital_port_connected(encoder)) {
> +		if ((connector->base.force == DRM_FORCE_ON ||
> +		     connector->base.force == DRM_FORCE_ON_DIGITAL) &&
> +		    connector->detect_edid && connector->base.edid_blob_ptr)
> +			status = connector_status_connected;
> +		else
> +			intel_hdmi_unset_edid(&connector->base);
>  		goto out;
> -
> -	intel_hdmi_unset_edid(&connector->base);
> +	}
>  
>  	if (intel_hdmi_set_edid(&connector->base))
>  		status = connector_status_connected;
> @@ -2638,10 +2659,10 @@ intel_hdmi_force(struct drm_connector *_connector)
>  	if (!intel_display_driver_check_access(display))
>  		return;
>  
> -	intel_hdmi_unset_edid(&connector->base);
> -
> -	if (connector->base.status != connector_status_connected)
> +	if (connector->base.status != connector_status_connected) {
> +		intel_hdmi_unset_edid(&connector->base);
>  		return;
> +	}
>  
>  	intel_hdmi_set_edid(&connector->base);
>  }

-- 
Jani Nikula, Intel