[pulseaudio-discuss] paplay stuck while playing over Bluetooth without any error messages

Tanu Kaskinen tanu.kaskinen at linux.intel.com
Thu Jun 5 06:34:30 PDT 2014


On Mon, 2014-06-02 at 16:01 +0100, tony wrote:
> Hello,
> 	I am trying to play audio over Bluetooth(BlueZ, 4.101) Headset profile 
> using my tegra board as audio Source running ubuntu 13.04. The pulse 
> audio version used is v5.0.
> 	From the bluez logs, I can see the headset is connected and the state 
> is converted to "HEADSET_STATE_PLAYING" as soon as I start playing a 
> *.wav file using pacat
> 
> "./src/pacat -vvvv -p -d bluez_sink.90_03_B7_30_E5_84 
> ../test3_PCM_Mono_8SS_8000Hz.wav"
> 
> But pacat seems to be stuck at Time: 0.0 until I interrupt the binary 
> (Ctrl+C). The corresponding (terminal, blueZ, pulseAudio )logs are 
> attached.
> 
> 	As far as I could debug, I see the "pacat", sends an 
> "PA_COMMAND_DRAIN_PLAYBACK_STREAM" () which is received and acted upon 
> by "command_drain_playback_stream". I traced the data flow till 
> "sink_input_process_msg" which returns 0 (~ 1540. The 
> PLAYBACK_STREAM_MESSAGE_DRAIN_ACK is not returned in the if loop). But 
> my knowledge in pulse audio is limited and am not sure whether it is 
> expected. Can somebody please give me any pointers what might be going 
> wrong?

Probably the headset is just not consuming any audio, and since there
are no errors in the pulseaudio log, it's hard to know what might be
wrong. Maybe bluez developers can figure something out from the bluez
log; I can't.

Is the test file very short? Draining should happen only when pacat
reaches the end of the file. To rule out any pulseaudio bug related to
handling short streams, you could test with a longer file that pacat
can't consume in one go.

The log level in the pulseaudio log seems to be "info", not "debug", so
you could increase the log verbosity and see if there are any
interesting debug messages.

-- 
Tanu



More information about the pulseaudio-discuss mailing list