[LAU] KVM virtual machines and jack

[torbenh]

On Mon, Dec 07, 2009 at 01:48:04AM -0600, Reuben Martin wrote:
> On Sun, Dec 6, 2009 at 4:14 AM, Patrick Shirkey
> <pshirkey at boosthardware.com> wrote:
> >
> > On 12/06/2009 03:20 PM, Reuben Martin wrote:
> >> On Sat, Dec 5, 2009 at 8:39 PM, Paul Davis<paul at linuxaudiosystems.com>  wrote:
> >>
> >>> On Sat, Dec 5, 2009 at 2:51 PM, Reuben Martin<reuben.m at gmail.com>  wrote:
> >>>
> >>>> Does anybody know if it is possible to run a guest OS under KVM and
> >>>> have the ins and outs of the virtual audio device presented to that
> >>>> OS, interface with JACK on the host linux system?
> >>>>
> >>>> I've done a little googling on this but couldn't find anything.
> >>>> (mostly because "jack" is ambiguous, and "KVM" can mean a lot of
> >>>> things)
> >>>>
> >>> unless the VM software that provides fake audio devices to the guest
> >>> OS knows about JACK internally or can be configured to use it, this is
> >>> not happening.
> >>> on the other hand, if the VM software can simply use the ALSA JACK
> >>> plugin, that could potentially work (lots and lots of latency though).
> >>>
> >>>
> >> In that case I guess my best bet would probably be to petition KVM
> >> development to add support for JACK. Currently I believe it supports
> >> ALSA, OSS, SDL and PulseAudio.
> >>
> >>
> >
> > If KVM already supports those api's then it can  be used with jack via
> > several different methods. What is the problem that you are having exactly?
> >
> 
> No problems yet. I'm putting together a new system and had thought of
> using Windows and Mac as guest VM systems under KVM for running things
> that don't exist for Linux and/or don't work well with WINE. And I
> wanted to see if anybody had tried routing the VM audio ins / outs to
> JACK rather than directly to an audio sink.
> 
> Wanted to see if anybody had seriously tried this before I gave it a go.

i fixed an issue with windows in netone.
seems like i can get down to around 26ms roundtrip latency with not many
netxruns.
50ms seems to be stable. 

note that i am not running the VM SCHED_FIFO. 
these tests are done with win7 in a VirtualBox

the host period size is 256. i guess using 1024 would be better.

these big latencies are not a real problem, when you use sequencers.
netjack1 provides a latency compensated jack Transport on the slave
side. 

Note that there is an asio adapter for jack on windows.
so the missing part is getting the jack transport into a windows
app. i have written jack_trans2midi which generates midi clock.

so you just need some virtual midi device.

http://www.youtube.com/watch?v=J0WnR0O-4g8

maybe i should redo this video with bigger console fonts :)
i needed to use a quite big latency due to some bugs and the socket
buffer size of windows being much too small.

these bugs should be fixed now.

i just need to update the celt and then i will roll a jacknone release.



> 
> -Reuben
> _______________________________________________
> Linux-audio-user mailing list
> Linux-audio-user at lists.linuxaudio.org
> http://lists.linuxaudio.org/mailman/listinfo/linux-audio-user

-- 
torben Hohn