Ok, perhaps the problem lies with real-time mode. ALSA has some flags that let you open a stream in "real-time" mode, and we tell explicitly tell PortAudio to do this if you're on Linux. (I think I made this change between 1.6.0 and 1.6.1.) It's possible that the kernel (maybe ALSA) has some sort of runaway process shutdown mechanism (like JACK's watchdog) that's giving Mixxx the boot because we're using up too much CPU time. That mechanism is probably only enabled when ALSA's in real-time mode.
The attached patch should disable the use of ALSA's realtime mode in Mixxx. If Owen or Jim could please test this and report back, it would be greatly appreciated. If this fixes the problem, then I think some of the nasty things we've done in our real-time callback code are finally catching up with us, and we end up waiting/blocking on things for too long. (Otherwise I don't understand why my C2D E8500 has this problem. It's a fast CPU.) On the other hand, I don't understand why neither Windows nor Mac OS X have this problem. I can push much lower latencies on my 2.2 GHz Macbook with the same crappy onboard Intel HDA chip than on my 3.0 GHz Linux box. Thanks, Albert On Sat, Jun 27, 2009 at 6:33 AM, James Evans<[email protected]> wrote: > Albert Santoni wrote: >> >> ... which is probably telling us that the bug isn't in Mixxx directly, >> but that's moot since this bug will affect Mixxx (ALSA) users on >> Ubuntu 8.10/9.04/9.10. Is there anyone who _hasn't_ seen Mixxx the >> audio stop (ever) on Ubuntu 8.10+? >> > > I was having no trouble with Intrepid (8.10) so the problem appears in > Ububntu versions 9.04+. > > Jim. > >
alsa-nort.patch
Description: Binary data
------------------------------------------------------------------------------
_______________________________________________ Mixxx-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mixxx-devel
