Hannu Savolainen wrote: > Stéphane Letz wrote: >> Ok. Now testing with O_EXCL and SNDCTL_DSP_COOKEDMODE works, bit I see >> some strange of the device. The overruns/underrun detection code based >> on SNDCTL_DSP_GETERROR does not see underrun anymore although the >> sound get weird with high CPU load (Writing another buffer in the out >> device solve the "weird sound" problem...). The same overruns/underrun >> detection code was working when opening the /dev/dsp0 device (that >> uses vmix AFAIK) >> >> This occurs on OpenSolaris with internal HDAudio using published OSS >> 4.0. Are tey know problems with this card with OSS 4.0 ? >> > I don't think this is caused by problems with any particular card. > Underruns/overruns will happen if the CPU load gets too high (more than > 80-90%). In this situation the application is not able to process audio > data as fast as it should. It's often possible to decrease the problem > by using higher priority (see man sched_setscheduler). > > It is just necessary to write enough silent data to the output device > before starting the read/write loop. It is normal that the other > processes running in the computer occasionally delay the audio process > by few milliseconds. If there is not enough data in the playback buffer > then this delay will cause an underrun. In addition recorded data will > get accumulated in the recording device during the delay. This will > sooner or later cause recording overruns if the application doesn't > handle this properly. Simple read/write loop (without any > usleep/whatever calls) will recover from this situation automatically. > > Best regards, > > Hannu > _______________________________________________ > oss-devel mailing list > oss-devel@mailman.opensound.com > http://mailman.opensound.com/mailman/listinfo/oss-devel >
Hi, Have you tried to use the fulldup.c test program to see if this works http://mercurial.opensound.com/?file/2fed97c765ec/tutorials/sndkit/samples/fulldup.c If fulldup.c shows problems then there are issues with vmix. Have you tried disabling vmix to see if the native device behaves differently than the vmix device? regards Dev Mazumdar -- -------------------------------------------------------------------------- 4Front Technologies 4035 Lafayette Place, Unit F, Culver City, CA 90232, USA Tel: 310 202 8530 Fax: 310 202 0496 URL: http://www.opensound.com -------------------------------------------------------------------------- _______________________________________________ oss-devel mailing list oss-devel@mailman.opensound.com http://mailman.opensound.com/mailman/listinfo/oss-devel
