On Wed, Feb 06, 2002 at 11:51:54 +0100, Takashi Iwai wrote: > > If I try to start jack with a shortish period time (eg. 512 frames) I get: > > > > Feb 5 14:12:16 dumuzi kernel: ALSA ../alsa-kernel/core/pcm_lib.c:118: Unexpected >hw_pointer value (stream = 1, delta: -256, max jitter = 1024): wrong interrupt >acknowledge? > > Feb 5 14:12:16 dumuzi kernel: ALSA ../alsa-kernel/core/pcm_lib.c:118: Unexpected >hw_pointer value (stream = 1, delta: -256, max jitter = 1024): wrong interrupt >acknowledge? > > Perhaps the minimum period size of m3 is larger than expected in the > code. Basically maestro3 uses timer interrupts to update pcm pointer > as well as on other ess's pci chips, and the messages above can come > when timer interrupts are not enough fine.
I'm getting these at 1024 too, but only occasionally, and when apps start and stop when you would expect overruns. > > Feb 5 14:12:16 dumuzi kernel: ALSA ../alsa-kernel/pci/maestro3.c:1486: bad MAGIC >(0xa15a0302) > > Hmm, it must be fixed. Do you still have this message with the latest > driver? Looks like the source code was changed from the version > you're using (the line doesn't match). No this is gone now. For some reason /etc/init.d/asound restart didn't load the new driver. After rebooting this has gone. > > The main problem though is that when running jack the CPU use often jumps > > to just under 50% or just under 100% for no good reason. It seems like the > > CPU is being used up inside alsa-lib or jacks alsa code, but its not > > obvious which. I don't get this problem on my hammerfall or ens1371. > > I guess this is because of the wrong period size setting. > m3 likely doesn't support the small period size, or, of course this > can be a bug of driver as usual ;) This happens with all period sizes, though the effects are slightly different with large period sizes (eg. 4096), it isn't allways 50 or 100%. I'm afraid I know almost nothing about sondcard hardware, so guessing what's going on in the driver/hardware is beyond me. - Steve _______________________________________________ Alsa-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/alsa-devel
