Robert Watson wrote: > Just to make sure: you're not running with WITNESS in your kernel, right? > That adds substantial overhead to every lock operation, and can really > impact interrupt latency. No, I'm not using WITNESS - doing so has no effect anyway (I tested). > Also, you might want to see if your sound device is sharing an interrupt > with another device that isn't marked MPSAFE, or sharing an interrupt at > all, which also substantially impacts interrupt latency. We're working on > pushing Giant out of various device drivers, but are only partway down > that road, which might well be the problem. What's a good way to check that, apart from going through dmesg? I'm gonna have to do that tomorrow, so tired I can't focus my eyes properly anymore ;) Thanks, /Eirik > > Robert N M Watson FreeBSD Core Team, TrustedBSD Projects > robert_at_fledge.watson.org Senior Research Scientist, McAfee Research > > On Wed, 17 Dec 2003, Eirik Oeverby wrote: > > >>Hi all, >> >>For a long time I have had problems with sound on my ThinkPad, which has >>a rather sucky Crystal chipset. The problems consist of skips and pops >>and crackles when playing back audio, and makes it hopeless to listen to >>any kind of music while working (or playing ;). >> >>There have been attempts at fixing this earlier, which has helped >>somewhat, but I always thought this was a problem on my chipset only - >>but this does not seem to be the case. >> >>I have now tested in a machine that has a SB Live! card (snd_emu10k1 >>module), and I'm experience the same problems, just even more pronounced >>than with the Crystal chipset. >> >>The interesting thing is - this is a non-issue in all 4.x versions of >>FreeBSD, and it's a non-issue in all other OSes I have tried on the >>respective computers. I am therefore led to believe there has been some >>kind of screwup with the PCM or chipset drivers in -CURRENT, or perhaps >>in some underlying code. This has been a problem atleast since >>5.1-RELEASE, and I do believe (but cannot confirm) it was a problem with >>5.0-RELEASE aswell. Choice of scheduler (SCHED_4BSD or SCHED_ULE) has no >>influence, neither has there been any difference before or after the >>i386 interrupt code switch. >> >>I home someone can have a look at this, it is of minor importance but >>major annoyance - and personally I think it should be (attempted) fixed >>before 5.2-RELEASE, if possible. It does make a bad impression on >>people, even in a FreeBSD-house such as the one I am working for right >>now. >> >>Best regards, >>Eirik Oeverby >> >>_______________________________________________ >>freebsd-current_at_freebsd.org mailing list >>http://lists.freebsd.org/mailman/listinfo/freebsd-current >>To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org" >> > >Received on Wed Dec 17 2003 - 13:04:51 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:34 UTC