Re: [media-ctl PATCH 1/1] libmediactl: Implement MEDIA_ENT_ID_FLAG_NEXT in media_get_entity_by_id()

[Sakari Ailus <sakari.ailus@iki.fi>]

Hi Laurent,

Laurent Pinchart wrote:
> On Saturday 14 January 2012 20:33:36 Sakari Ailus wrote:
>> Signed-off-by: Sakari Ailus<sakari.ailus@iki.fi>
>> ---
>>   src/mediactl.c |    9 +++++++--
>>   src/mediactl.h |    4 +++-
>>   2 files changed, 10 insertions(+), 3 deletions(-)
>>
>> diff --git a/src/mediactl.c b/src/mediactl.c
>> index 5b8c587..f62fcdf 100644
>> --- a/src/mediactl.c
>> +++ b/src/mediactl.c
>> @@ -81,8 +81,13 @@ struct media_entity *media_get_entity_by_id(struct
>> media_device *media, for (i = 0; i<  media->entities_count; ++i) {
>>   		struct media_entity *entity =&media->entities[i];
>>
>> -		if (entity->info.id == id)
>> -			return entity;
>> +		if (!(id&  MEDIA_ENT_ID_FLAG_NEXT)) {
>> +			if (entity->info.id == id)
>> +				return entity;
>> +		} else {
>> +			if (entity->info.id>= (id&  ~MEDIA_ENT_ID_FLAG_NEXT)
>> +				return entity;
>> +		}
>
> Just one question that hasn't crossed my mind before, why do you need this ?
> If you want to enumerate entities in an application you can just iterate over
> media_device::entities.

We do have the MEDIA_ENT_ID_FLAG_NEXT flag which is intended to help in 
entity enumeration. Currently the range of entity ids is contiguous in 
all practical implementation but will that always be the case, also in 
the future? A few things might break in the kernel if the range is 
non-contiguous as well, but that's still internal to the kernel.

However, this is a user space library and if this interface change is 
not made, we essentially are making a promise that the entity ranges 
will always be contiguous.

I wouldn't as there's no need to do so.

I you think about programmable hardware, entities there are logical 
rather than physical and their existence may be dependent on multiple 
factors.

-- 
Sakari Ailus
sakari.ailus@iki.fi