[Alsa-user] Odd timing behavior observed running OSS emulation on Alsa

[Roger D. <rb...@cs...>]

When using the OSS API under Alsa 0.5.9, I recorded the time at which each
audio block was written. You would expect the time to be approximately
64/41000 seconds/block for 64 frame blocks and a sample rate of 44100. This
is usually true, but a graph of the times vs. block number shows some jumps
as if either more buffer space opened up, some samples were dropped, or
there are jumps in the time reference. I verified that samples are not being
dropped, and this problem only occurs with OSS emulation. (Direct ALSA works
as expected, and the OSS code works as expected on an non-alsa machine).
This seems to indicate something strange about OSS emulation, buffers, and
timing. I put some graphs on http://www.cs.cmu.edu/~rbd/oss_alsa/. If anyone
knows what's going on here, I'd sure like to know. (Please reply to or cc to
rb...@cs...).

	Roger Dannenberg