On Monday, June 20, 2011 16:09:32 Laurent Pinchart wrote:
> Hi Hans,
> 
> Thanks for the patch.
> 
> On Tuesday 14 June 2011 17:22:29 Hans Verkuil wrote:
> > From: Hans Verkuil <hans.verk...@cisco.com>
> > 
> > When the event queue for a subscribed event is full, then the oldest
> > event is dropped. It would be nice if the contents of that oldest
> > event could be merged with the next-oldest. That way no information is
> > lost, only intermediate steps are lost.
> > 
> > This patch adds an optional merge function that will be called to do
> > this job and implements it for the control event.
> > 
> > Signed-off-by: Hans Verkuil <hans.verk...@cisco.com>
> > ---
> >  drivers/media/video/v4l2-event.c |   27 ++++++++++++++++++++++++++-
> >  include/media/v4l2-event.h       |    5 +++++
> >  2 files changed, 31 insertions(+), 1 deletions(-)
> > 
> > diff --git a/drivers/media/video/v4l2-event.c
> > b/drivers/media/video/v4l2-event.c index 9e325dd..aeec2d5 100644
> > --- a/drivers/media/video/v4l2-event.c
> > +++ b/drivers/media/video/v4l2-event.c
> > @@ -113,6 +113,7 @@ static void __v4l2_event_queue_fh(struct v4l2_fh *fh,
> > const struct v4l2_event *e {
> >     struct v4l2_subscribed_event *sev;
> >     struct v4l2_kevent *kev;
> > +   bool copy_payload = true;
> > 
> >     /* Are we subscribed? */
> >     sev = v4l2_event_subscribed(fh, ev->type, ev->id);
> > @@ -130,12 +131,23 @@ static void __v4l2_event_queue_fh(struct v4l2_fh *fh,
> > const struct v4l2_event *e sev->in_use--;
> >             sev->first = sev_pos(sev, 1);
> >             fh->navailable--;
> > +           if (sev->merge) {
> > +                   if (sev->elems == 1) {
> > +                           sev->merge(&kev->event, ev, &kev->event);
> > +                           copy_payload = false;
> > +                   } else {
> > +                           struct v4l2_kevent *second_oldest =
> > +                                   sev->events + sev_pos(sev, 0);
> > +                           sev->merge(&second_oldest->event, 
> > &second_oldest->event, &kev-
> >event);
> > +                   }
> > +           }
> >     }
> > 
> >     /* Take one and fill it. */
> >     kev = sev->events + sev_pos(sev, sev->in_use);
> >     kev->event.type = ev->type;
> > -   kev->event.u = ev->u;
> > +   if (copy_payload)
> > +           kev->event.u = ev->u;
> >     kev->event.id = ev->id;
> >     kev->event.timestamp = *ts;
> >     kev->event.sequence = fh->sequence;
> > @@ -184,6 +196,17 @@ int v4l2_event_pending(struct v4l2_fh *fh)
> >  }
> >  EXPORT_SYMBOL_GPL(v4l2_event_pending);
> > 
> > +static void ctrls_merge(struct v4l2_event *dst,
> > +                   const struct v4l2_event *new,
> > +                   const struct v4l2_event *old)
> > +{
> > +   u32 changes = new->u.ctrl.changes | old->u.ctrl.changes;
> > +
> > +   if (dst == old)
> > +           dst->u.ctrl = new->u.ctrl;
> > +   dst->u.ctrl.changes = changes;
> > +}
> > +
> >  int v4l2_event_subscribe(struct v4l2_fh *fh,
> >                      struct v4l2_event_subscription *sub, unsigned elems)
> >  {
> > @@ -210,6 +233,8 @@ int v4l2_event_subscribe(struct v4l2_fh *fh,
> >     sev->flags = sub->flags;
> >     sev->fh = fh;
> >     sev->elems = elems;
> > +   if (ctrl)
> > +           sev->merge = ctrls_merge;
> > 
> >     spin_lock_irqsave(&fh->vdev->fh_lock, flags);
> >     found_ev = v4l2_event_subscribed(fh, sub->type, sub->id);
> > diff --git a/include/media/v4l2-event.h b/include/media/v4l2-event.h
> > index 8d681e5..111b2bc 100644
> > --- a/include/media/v4l2-event.h
> > +++ b/include/media/v4l2-event.h
> > @@ -55,6 +55,11 @@ struct v4l2_subscribed_event {
> >     struct v4l2_fh          *fh;
> >     /* list node that hooks into the object's event list (if there is one) 
> > */
> >     struct list_head        node;
> > +   /* Optional callback that can merge two events.
> > +      Note that 'dst' can be the same as either 'new' or 'old'. */
> 
> This can lead to various problems in drivers, if the code forgets that 
> changing dst will change new or old. Would it be possible to make it a two 
> arguments (dst, src) function ?

This has been split into a replace and a merge function in the final patch in
my pull request. It's the only change I made (besides documentation) compared
to the RFCv1. This merge function was a bit too obscure and splitting it up
into two callbacks worked much better.

Regards,

        Hans
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to