sorry Georg, it`s ok now, my bad, I made a double lock in my calling
chain ....
å?¨ 2015å¹´05æ??20æ?¥ 15:04, golden å??é??:
> Hi Georg
>
> static void pa_stream_timing_callback(pa_stream*s, int
> success, void *userdata)
> {
> NCLOGD("pa_stream_timing_callback -");
> pa_threaded_mainloop_signal((pa_threaded_mainloop*)userdata, 0);
> }
>
> int64_t audiostream::getBufferLatencySize()
> {
>
> pa_threaded_mainloop_lock(m_mainloop);
> pa_operation* o = pa_stream_update_timing_info(s,
> pa_stream_timing_callback, m_mainloop);
> LOGD("enter PA_WAIT_OPEARION2DONE loop");
> if (o != NULL) {
> while (pa_operation_get_state(o) == PA_OPERATION_RUNNING)
> {
> pa_threaded_mainloop_wait(m_mainloop);
> NCLOGD("pa_operation_get_state(o) = %d",
> pa_operation_get_state(o));
> }
> pa_operation_unref(o);
> o = NULL;
> } else {
> LOGD("PA_WAIT_OPEARION2DONE operation is NULL");
> }
>
> NCLOGD("exit PA_WAIT_OPEARION2DONE loop");
> pa_threaded_mainloop_unlock(m_mainloop);
>
>
> const pa_timing_info* t = pa_stream_get_timing_info(s);
> if (!t) {
> return -1;
> }
> return t->write_index - t->read_index;
>
> }
>
> this is the modified code, still the same problem.
> It feels like if auto timing update is enabled,
> pa_stream_update_timing_info called in
> another thread will be missed, and callback is not notified any more.
>
> That`s my test result.
>
> BR,
> Lixin
>
>
> å?¨ 2015å¹´05æ??20æ?¥ 13:49, Georg Chini å??é??:
>> Dear Lixin,
>>
>> On 20.05.2015 04:10, golden wrote:
>>> Dear PulseAudio forks,
>>>
>>>
>>> I`m making a function to get the buffer latency of
>>> pa_stream(with PA_STREAM_AUTO_TIMING_UPDATE flag), here is what it
>>> looks like
>>
>> What type of mainloop are you using?
>>
>>>
>>>
>>> void operation_done_callback(pa_stream* , bool, void*)
>>> {
>>> LOGD("operation_done");
>>> }
>>>
>>> int64_t audiostream::getBufferLatencySize()
>>> {
>>> // m_stream is a pointer of pa_stream
>>> pa_operation* o = pa_stream_update_timing_info(m_stream,
>>> operation_done_callback, this);
>>> LOGD("enter PA_WAIT_OPEARION2DONE loop");
>>> if (o != NULL) {
>>> while (sync && pa_operation_get_state(o) !=
>>> PA_OPERATION_DONE)
>>> ;
>>
>> Where does sync come from?
>> Maybe you try
>> while (sync && pa_operation_get_state(o) == PA_OPERATION_RUNNING)
>> instead, then at least the function should always return. The way you
>> implemented it, the function will never return if
>> pa_operation_get_state()
>> returns PA_OPERATION_CANCELLED.
>>
>>> pa_operation_unref(o);
>>> o = NULL;
>>> } else {
>>> LOGD("PA_WAIT_OPEARION2DONE operation is NULL");
>>> }
>>>
>>> LOGD("exit PA_WAIT_OPEARION2DONE loop");
>>>
>>> const pa_timing_info* t = pa_stream_get_timing_info(s);
>>> if (!t) {
>>> return -1;
>>> }
>>> return t->write_index - t->read_index;
>>> }
>>>
>>> I want that the user can get the accurate result when he calls
>>> this function,
>>> so every time if user called getBufferLatencySize, I call
>>> pa_stream_update_timing_info and wait this operation done, and this
>>> function is called in user thread.
>>>
>>> Sometimes this operation is never finished, and the operation
>>> done callback is never called.
>>>
>>> Don't know what I'm doing wrong, any advice will be
>>> appreciated, Thanks!
>>>
>>>
>>>
>>>
>>> BR,
>>> Lixin
>>> _______________________________________________
>>> pulseaudio-discuss mailing list
>>> pulseaudio-discuss at lists.freedesktop.org
>>> http://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
>>
>> _______________________________________________
>> pulseaudio-discuss mailing list
>> pulseaudio-discuss at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
>
