Re: [PATCH] drm/amd/display: Remove unwanted drm edid references

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

On Fri, 15 Sep 2023, Harry Wentland <harry.wentland@amd.com> wrote:
> On 2023-09-05 13:13, Alex Hung wrote:
>> [WHY]
>> edid_override and drm_edid_override_connector_update, according to drm
>> documentation, should not be referred outside drm_edid.
>> 
>> [HOW]
>> Remove and replace them accordingly.
>> 
>> Signed-off-by: Alex Hung <alex.hung@amd.com>
>> ---
>>  .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 23 ++-----------------
>>  1 file changed, 2 insertions(+), 21 deletions(-)
>> 
>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> index 1bb1a394f55f..f6a255773242 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> @@ -6372,15 +6372,12 @@ amdgpu_dm_connector_late_register(struct drm_connector *connector)
>>  static void amdgpu_dm_connector_funcs_force(struct drm_connector *connector)
>>  {
>>  	struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector);
>> +	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
>>  	struct dc_link *dc_link = aconnector->dc_link;
>>  	struct dc_sink *dc_em_sink = aconnector->dc_em_sink;
>>  	struct edid *edid;
>>  
>> -	if (!connector->edid_override)
>> -		return;
>> -
>> -	drm_edid_override_connector_update(&aconnector->base);
>> -	edid = aconnector->base.edid_blob_ptr->data;
>> +	edid = drm_get_edid(connector, &amdgpu_connector->ddc_bus->aux.ddc);
>
> Looks like we only call this in the case where a connector is forced, so
> drm_get_edid will never try to read the edid from the ddc but always gives
> us the override_edid. Please spell that out in the commit description so
> anyone else looking at the patch doesn't have to trace it themselves.

Connector forcing is only about forcing the connector status. The probe
helper will call ->force instead of ->detect.

But this has nothing to do with override_edid. That's completely
orthogonal.

Here, you can call drm_get_edid() if you like. Connector forcing just
bypasses the DDC probe for determining connector status. If connector is
forced off, you won't get the EDID, regardless of override/firmware
EDID, but if it's forced on, the EDID read proceeds.

And the EDID read has the priority order 1) override EDID if set via
edid_override debugfs, 2) firmware EDID if set via edid_firmware module
parameter, and 3) regular DDC read.

BR,
Jani.

>
>>  	aconnector->edid = edid;
>>  
>>  	/* Update emulated (virtual) sink's EDID */
>> @@ -6421,22 +6418,6 @@ static void create_eml_sink(struct amdgpu_dm_connector *aconnector)
>>  	};
>>  	struct edid *edid;
>>  
>> -	if (!aconnector->base.edid_blob_ptr) {
>
> Can edid_blob_ptr never be NULL?
>
> Harry
>
>> -		/* if connector->edid_override valid, pass
>> -		 * it to edid_override to edid_blob_ptr
>> -		 */
>> -
>> -		drm_edid_override_connector_update(&aconnector->base);
>> -
>> -		if (!aconnector->base.edid_blob_ptr) {
>> -			DRM_ERROR("No EDID firmware found on connector: %s ,forcing to OFF!\n",
>> -					aconnector->base.name);
>> -
>> -			aconnector->base.force = DRM_FORCE_OFF;
>> -			return;
>> -		}
>> -	}
>> -
>>  	edid = (struct edid *) aconnector->base.edid_blob_ptr->data;
>>  
>>  	aconnector->edid = edid;
>

-- 
Jani Nikula, Intel