Re: [RFCv4 06/30] Bluetooth: Timers fixes

[Andrei Emeltchenko <Andrei.Emeltchenko.news@gmail.com>]

Hi Ulisses,

On Tue, Mar 20, 2012 at 09:21:56AM -0300, Ulisses Furquim wrote:
> > -static inline void l2cap_set_timer(struct l2cap_chan *chan,
> > -                                       struct delayed_work *work, long
> > timeout)
> > -{
> > -       BT_DBG("chan %p state %s timeout %ld", chan,
> > -                                       state_to_string(chan->state),
> > timeout);
> > -
> > -       if (!cancel_delayed_work(work))
> > -               l2cap_chan_hold(chan);
> > -       schedule_delayed_work(work, timeout);
> > -}
> > -
> >  static inline bool l2cap_clear_timer(struct l2cap_chan *chan,
> >                                        struct delayed_work *work)
> >  {
> >        bool ret;
> >
> > -       ret = cancel_delayed_work(work);
> > +       ret = (delayed_work_pending(work) && cancel_delayed_work(work));
> >        if (ret)
> >                l2cap_chan_put(chan);
> >
> >        return ret;
> >  }
> >
> > +static inline void l2cap_set_timer(struct l2cap_chan *chan,
> > +                                       struct delayed_work *work, long
> > timeout)
> > +{
> > +       BT_DBG("chan %p state %s timeout %ld", chan,
> > +                                       state_to_string(chan->state),
> > timeout);
> > +
> > +       l2cap_clear_timer(chan, work);
> > +
> > +       l2cap_chan_hold(chan);
> > +       schedule_delayed_work(work, timeout);
> > +}
> > +
> What are you fixing here? Care to explain and send an example? This
> needs to be clear in the commit message so we don't have more
> confusion with timer handling as we had this code for quite some time.

I have just sent the patch with proper commit message. I think you assume
that cancel_delayed_work returns not zero if it cancels delayed work but
it returns not zero if there is no work running (even if there was no timer
-- as far as I understood and comment apply).

Best regards 
Andrei Emeltchenko