Hi Tanu, On Tue, Dec 11, 2012 at 10:20 AM, Tanu Kaskinen <tanuk at iki.fi> wrote: > On Tue, 2012-12-11 at 09:08 +0100, Mikel Astiz wrote: >> Hi Tanu, >> >> On Tue, Dec 11, 2012 at 6:10 AM, Tanu Kaskinen <tanuk at iki.fi> wrote: >> > On Mon, 2012-12-10 at 08:30 +0100, Mikel Astiz wrote: >> >> From: Mikel Astiz <mikel.astiz at bmw-carit.de> >> >> >> >> Move the connection of sink/source-related hooks to module >> >> initialization and shutdown, to group all of them together. There is >> >> no need to connect them every time the card profile is changed. >> > >> > sink_state_changed_cb and source_state_changed_cb have to be modified a >> > bit, basically just return without doing anything if USE_SCO_OVER_PCM() >> > returns false. With the current code this patch can cause an assertion >> > crash because of this in the beginning of sco_over_pcm_state_update(): >> > >> > pa_assert(USE_SCO_OVER_PCM(u)); >> >> I don' think this scenario is possible since sco_source and sco_sink >> should be NULL and thus sco_over_pcm_state_update() would never be >> called, due to the comparison in both xxx_state_changed_cb(). > > Why do you say sco_source and sco_sink should be NULL? They're set > during module initialization and never changed, so they definitely can > be non-NULL. USE_SCO_OVER_PCM() returns false if the current profile is > not HSP, but with your patch xxx_state_changed_cb() can be called also > when some other profile is active, and thus crashing may occur. You're right, I didn't actually realize this. As a general comment, it'd be good if someone could test the SCO_OVER_PCM part since I'm not able to do it and there's been a lot changes. It's likely that we have some regressions, I'm afraid. >> However, I agree that we should make this more explicit. We could do >> what you propose or alternatively (as the previous code does) just >> connect the hook only if USE_SCO_OVER_PCM. Any preference? > > I prefer having the hook always connected, for simplicity. I will send v2 with the change you suggest. Cheers, Mikel
