Re: [PATCH v3 1/2] mac80211_hwsim: fix duplicate beacons on TSF adjust

[Johannes Berg <johannes@sipsolutions.net>]

On Wed, 2014-01-15 at 09:55 -0800, Thomas Pedersen wrote:

> >> +     if (!hrtimer_callback_running(&data->beacon_timer.timer) &&
> >> +         !test_bit(TASKLET_STATE_RUN, &data->beacon_timer.tasklet.state))
> >> +             tasklet_hrtimer_cancel(&data->beacon_timer);
> >
> > That test_bit() really seems suspicious - there are no other users in
> > the tree except for the internal tasklet code... What are you trying to
> > do?
> 
> I'm trying to avoid calling tasklet_hrtimer_cancel() recursively, that
> is, when mac80211_hwsim_beacon_sched() is called from within the
> hrtimer tasklet itself. Looking at it again this does seem ugly. Would
> it be acceptable to pass a flag to _beacon_sched() indicating whether
> a reschedule is taking place from within the tasklet? 

That seems fine to me.

> We could also
> have a mac80211_hwsim_beacon_sched() (which would cance the hrtimer
> tasklet) and __mac80211_hwsim_beacon_sched() pair, but I'm not sure if
> that nomenclature applies here.

Whichever causes less code duplication? Could one call the other maybe?
Dunno, up to you. I was really just thinking that the test_bit() was odd
since nobody else uses it.

johannes