Hi Hans,

On Friday 04 February 2011 14:19:19 Hans Verkuil wrote:

On Friday, February 04, 2011 13:34:02 Sakari Ailus wrote:

...

Hi,

And many thanks for the comments!

Hans Verkuil wrote: ...

...

...

diff --git a/include/media/media-entity.h b/include/media/media-entity.h index b82f824..114541a 100644 --- a/include/media/media-entity.h +++ b/include/media/media-entity.h @@ -81,6 +81,8 @@ struct media_entity {

struct media_pad *pads; /* Pads array (num_pads elements) */ struct media_link *links; /* Links array (max_links elements)*/

• int use_count; /* Use count for the entity. */

Isn't unsigned better?

Could be. The result, though, will be slightly more difficult checking for bad use count --- which always is a driver bug.

me->use_count += change; WARN_ON(me->use_count < 0);

we must do something like this:

if (change < 0)

WARN_ON(me->use_count < (unsigned)-change);

me->use_count += change;

I'd perhaps also go with unsigned int; the choice for signed was made mainly since the above check and with signed int the check was more trivial.

I saw this WARN_ON as well. I think there is a good reason for that WARN_ON, but I think a comment in the header explaining why it is an int will be useful. If I trip over it, then others will as well :-)

I'll add a comment. Thanks.