Re: [PATCH v2 09/26] media: v4l: async: refactor v4l2_async_create_ancillary_links

[Laurent Pinchart]

On Wed, Apr 24, 2024 at 06:17:31PM +0000, Sakari Ailus wrote:
> On Wed, Apr 24, 2024 at 12:55:20PM +0200, Hans Verkuil wrote:
> > On 19/04/2024 11:47, Ricardo Ribalda wrote:
> > > Return 0 without checking IS_ERR or PTR_ERR if CONFIG_MEDIA_CONTROLLER
> > > is not enabled.
> > > 
> > > This makes cocci happier:
> > > 
> > > drivers/media/v4l2-core/v4l2-async.c:331:23-30: ERROR: PTR_ERR applied after initialization to constant on line 319
> > > 
> > > Signed-off-by: Ricardo Ribalda <ribalda@xxxxxxxxxxxx>
> > > ---
> > >  drivers/media/v4l2-core/v4l2-async.c | 7 +++----
> > >  1 file changed, 3 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/drivers/media/v4l2-core/v4l2-async.c b/drivers/media/v4l2-core/v4l2-async.c
> > > index 4bb073587817..915a9f3ea93c 100644
> > > --- a/drivers/media/v4l2-core/v4l2-async.c
> > > +++ b/drivers/media/v4l2-core/v4l2-async.c
> > > @@ -316,9 +316,10 @@ v4l2_async_nf_try_all_subdevs(struct v4l2_async_notifier *notifier);
> > >  static int v4l2_async_create_ancillary_links(struct v4l2_async_notifier *n,
> > >  					     struct v4l2_subdev *sd)
> > >  {
> > > -	struct media_link *link = NULL;
> > > +	struct media_link *link;
> > >  
> > > -#if IS_ENABLED(CONFIG_MEDIA_CONTROLLER)
> > > +	if (!IS_ENABLED(CONFIG_MEDIA_CONTROLLER))
> > > +		return 0;
> > >  
> > >  	if (sd->entity.function != MEDIA_ENT_F_LENS &&
> > >  	    sd->entity.function != MEDIA_ENT_F_FLASH)
> > > @@ -326,8 +327,6 @@ static int v4l2_async_create_ancillary_links(struct v4l2_async_notifier *n,
> > >  
> > >  	link = media_create_ancillary_link(&n->sd->entity, &sd->entity);
> > >  
> > > -#endif
> > > -
> > >  	return IS_ERR(link) ? PTR_ERR(link) : 0;
> > >  }
> > 
> > I think I would prefer:
> > 
> > static int v4l2_async_create_ancillary_links(struct v4l2_async_notifier *n,
> > 					     struct v4l2_subdev *sd)
> > {
> > #if IS_ENABLED(CONFIG_MEDIA_CONTROLLER)
> > 	struct media_link *link;
> > 
> > 	...
> > 
> > 	return IS_ERR(link) ? PTR_ERR(link) : 0;
> > #else
> > 	return 0;
> > #endif
> > }
> > 
> 
> Me, too.

I actually prefer Ricardo's proposal :-)

-- 
Regards,

Laurent Pinchart