Re: [RFC v2 2/5] OMAPDSS: DT: Get source endpoint by matching reg-id

[Archit Taneja <archit@ti.com>]

On Tuesday 27 May 2014 03:54 PM, Tomi Valkeinen wrote:
> On 27/05/14 12:49, Archit Taneja wrote:
>> On Tuesday 27 May 2014 02:04 PM, Tomi Valkeinen wrote:
>
>>> Hmm, maybe the above function, and the helper functions, should be
>>> changed a bit more, so that this function would do:
>>>
>>> ep = omapdss_of_get_first_endpoint(node);
>>> src_port = omapdss_of_get_remote_port(ep);
>>> src = omap_dss_find_output_by_port_node(src_port);
>>
>> src_port will be just on level up(the parent node), the source node will
>> be still a couple of hops up.
>>
>> Getting the source node would require us to do some DT related hopping
>> in omap_dss_find_output_by_port_node(). And it'll also require parsing
>> of the dss output_list in output.c. I'm worried it'll be a bit messy,
>> and we might end up adding some DT parsing in output.c
>
> But now the function above is messy =).
>
> Well, I haven't thought what the code would actually be. I just thought
> the functions I wrote would be logical in this context. As the output
> omap_dss_device more or less is a "port", it would feel logical to ask
> for the omap_dss_device by giving the port node.
>
> But you're right, the function in output.c would need to do a bit more
> than now. Then again, we could add helper functions to dss-of.c, so that
> output.c wouldn't need to do them manually.
>
> I guess the helper funcs in this case would be:
>
> struct device_node *dss_of_port_get_parent_device(struct device_node *port);
>
> int dss_of_port_get_port_number(struct device_node *port);

These look fine. I'll update the patch.

Archit