LE Audio: all members of "sink" input argument of pa_sync_lost are zeros

[ladivin]

Hello All,

I'm playing with LE Audio broadcast sinks and sources and I stumbled upon following issue. pa_sync_lost callback from bt_audio_broadcast_sink_cb gets one input argument - pointer to bt_audio_broadcast_sink structure called sink. When PA sync is lost and pa_sync_lost is called, all members of sink are zeros inside of this callback. One of the consequences is, that it's not possible to use sink->broadcast_id to determine to which source the lost PA belonged.

The reason all members of sink are zeroed lies in subsys/bluetooth/audio/broadcast_sink.c, in fuction:

static void pa_term(struct bt_le_per_adv_sync *sync,
		    const struct bt_le_per_adv_sync_term_info *info)
{
        ...

	broadcast_sink_cleanup(sink);
	SYS_SLIST_FOR_EACH_CONTAINER(&sink_cbs, listener, _node) {
		if (listener->pa_sync_lost != NULL) {
			listener->pa_sync_lost(sink);
		}
	}
}

As we can see, broadcast_sink_cleanup(sink), that memsets all sink members to zero, is called before the pa_sync_lost callbacks.

What is the reason to call broadcast_sink_cleanup(sink) first? Is it so that the pa_sync_lost callbacks don't get pointers to elements that are no longer valid because the PA was lost already? Or would it be safe to call the cleanup after the callbacks? In my opinion, there should be either an update to pa_sync_lost documentation notifying about this behaviour, or change of the code that would move the callback further down. Either way, I'm more then willing to prepare a PR. But before I do that, I just wanted to make sure I'm not missing anything.

Thanks,

Ladivin