On Thu, 20 May 2010 12:04:06 +0800 Raymond Yau superquad.vortex2@gmail.com wrote:
2010/5/20 Tyler Conant tconant@taser.com
I'm trying to use dmix on an OMAP processor. I'm experiencing the first 50 ms of the sample getting clipped. When I aplay directly to the hw:0 I don't get the clipping. The difference I see between the alsa configuration between dmix and hw:0 is the start_threshold, stop_threshold and silence_size. When it's direct to hw:0 the start and stop threshold are equal to the buffer size. For dmix the start threshold is 1 and the stop threshold is 1445199872. Is this dmix which is setting the start and stop or is this aplay. I'm only guessing this is the reason why the first part is clipping. The only bug which seem similar is http://bugs.gentoo.org/show_bug.cgi?id=218641 and it's already been closed out. I'm using alsa lib 1.0.23
...
The major difference is access : MMAP_INTERLEAVED
I think I hear the same problem with OMAP3 BeagleBoard. These play fine:
aplay -D hw:0 48k_1kHz0dB100ms.wav aplay -D hw:0 -M 48k_1kHz0dB100ms.wav
But with this there is some crack etc. if played after playing directly to HW device:
aplay -D dmix 48k_1kHz0dB100ms.wav
But I don't hear this crack sound if repeating the command above. Problem appears only when playing to dmix after playing to HW device.
And no, ASoC DAPM is not explaining this. Time between the commands doesn't have any other effect than power up/-down pops.