Re: [Linux-kernel] RFC: supporting adv7604.c under soc_camera/rcar_vin

[William Towle <william.towle@codethink.co.uk>]

On Wed, 4 Mar 2015, Hans Verkuil wrote:
> On 03/04/15 10:51, William Towle wrote:
>>     if (timings->pad >= state->source_pad)
>>             return -EINVAL;
>>   It suffices to comment out this line, but clearly this is not ideal.
>> Depending on the intended semantics, should it be filtering out all pad
>> IDs not matching the active one, or all pad IDs that are not valid
>> input sources? Unfortunately the lager board's adv7180 chip is too
>> simple to make a sensible comparison case (that we can also run tests
>> on) here.
>
> The adv7604 code is not ideal, although the pad test is valid (you shouldn't
> be able to ask timings for pads that do not exist).

   Right, thanks. It seems I have initialisation code that is making
inappropriate assumptions earlier on. I'll investigate this.


>>   Please advise. Comments would also be welcome regarding whether the
>> shims describe changes that should live in the driver or elsewhere in
>> soc_camera/rcar_vin in an acceptable solution.
>
> I'm not entirely sure what it is you are referring to.

   Amongst our various modifications to soc_camera/rcar_vin we have a
'struct media_pad' object in order to communicate with the adv7604
driver, and latterly an array of 'struct v4l2_subdev_pad_config's to
handle format information ... but there is more to be done, obviously,
and you have pointed us in the right direction above :)

Cheers,
   Wills.