On Thu, 24 Jan 2002, Kai Vehmanen wrote: > After applying this, I stopped getting "Unexpected hw_pointer" and "Lost > interrupt: hw_ptr = %d expected %d" errors from pcm_lib.c. Without the > patch, I get these errors with all parameters I've tried (even 48kHz!) if > both stream directions are active.
This was bogus. The errors were caused by invalid period and buffersizes set by ecasound. I'm not yet sure whether there's a bug in snd_pcm_hw_params_set_buffer_time_near() and snd_pcm_hw_params_set_period_size_near() functions, or was ecasound just using the API incorrectly. Anyway, in certain situations the above functions resulted in quite strange buffering params. Changing ecasound to use snd_pcm_hw_params_set_buffer_size() and snd_pcm_hw_params_set_period_size_near() instead fixed the problem. Anyway, the patch is still valid. I'm able to run "./latency -r 44100 -m 512 -M 512 -c 2 -s 2000 -p" for long periods of time without problems (like Maarten's original test, but with 2 channels). With '-c 1' there is still something funny going on. It works for a while, around 85secs, but then the audio goes bad. With the patch applied, the resulting bad audio sounds different. It's more like a really short delay than distortion. With '-r 44100', I can't reproduce the problem. With '-r 44099' it happens every time, and usually close to 85secs. Weird... -- http://www.eca.cx Audio software for Linux! _______________________________________________ Alsa-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/alsa-devel
