On 4/21/05, michael higgins <[EMAIL PROTECTED]> wrote: > On Thu, 21 Apr 2005 09:49:50 -0700 > Mark Knecht <[EMAIL PROTECTED]> wrote: > > > Hi, > > I'm just poking around looking at kernels again this morning. Maybe > > I've missed it but I cannot find one in portage that uses Ingo > > Molnar's realtime-preempt patches. Is there one? I am now using both > > ck-sources as well as the realtime-lsm module from portage. Many > > thanks to who ever helped with getting that in there. > > > > Are there enough people using Gentoo for realtime audio that it > > would warrent some sorrt of specialized kernel project? I expect so. > > Is there interest? > > Yes, there is interest. At least, I'm interested. > > I've been using 2.6.10-gentoo-r6 with > > CONFIG_SECURITY_REALTIME=m > > using the patch from here: http://www.joq.us/realtime/ at the suggestion of > someone on this list... and getting some good low latency in, say, jack rack.
What are you seeing? > > But, It seems there's more than one way to do this. > > I was _trying_ to find a patch I could apply to the 2.6.11 kernel for > realtime pre-emption, but this isn't it, I guess. No, it isn't. This is a patch that allows you to run with realtime privileges as a normal user. This is good but it's no the complete story I think. I know read about Molnar's realtime-preempt patches somewhere along the way, maybe even tried to apply them to the gentoo kernel sources, but this didn't work out. I'm essentially clueless in this dept. OK, so Ingo's patches are here: (at least I think!) http://people.redhat.com/mingo/ He does both low-latency patches and realtime-preempt patches. I think they are both used in the PlanetCCRMA kernels, along with the realtime-lsm modules to allow a user to take full advantage of this stuff. That's the way I understand it anyway. > > I also read somewhere along the way that maybe 2.6.11 isn't as good, yet, as > 2.6.10 for audio. But I don't believe everything I read. > > Anyway, thanks for bringing this up. It'd be great to have a ready-made > Gentoo realtime-preempt kernel. I think so. First we should make sure we know what we want and then possibly try writing an ebuild to make it more or less automatic. Thanks, Mark -- gentoo-user@gentoo.org mailing list
