Re: [FFADO-user] motu ultralite mk3 + ubuntu13 - jack doesn't start

[Jonathan W. <jw...@ju...>]

Hi Miguel

Fri, 2013-03-29 às 20:11 +0000, Miguel Negrao escreveu:
> I've been looking into switching from osx to linux some time, 
> unfortunately main main activity is computer music and I own a motu 
> ultralite mk3 (not hybrid, firewire only), so I really need that card 
> working in whatever os I use. I've seen in the release notes of ffado 
> 2.1 that there is experimental support for the motu, so I was hopping to 
> get everything working in linux so I could finally switch. ...

While there's no guarantee of getting the ultralite mk3 working due to
MOTU's well publicised hostility to Linux, we can make every attempt to do
so.  Based on what you've provided so far I think it's fairly likely that we
can at least get audio streaming working properly (access to the DSP
components of the interface may be a different matter - again, we'll deal
with that later).  But first things first; if you are able to run tests and
provide feedback I think we can make a pretty good stab at this.

As Matt Collier pointed out, your machine currently has a generic kernel and
this can cause issues for audio work in general (not just FFADO).  However,
in our recent experience it should not be necessary to go all the way to a
kernel running the realtime patch in most cases.  Shifting to a "normal"
kernel compiled with the "low latency desktop" setting (also known as a
PREEMPT kernel) should be all that's needed.  Most distributions have such a
kernel in their repositories for users to install if they wish to.

Having said that, if you want to go down the RT patch route then that's fine
too; it's just that it's possibly more work. :-)

Setting IRQ priorities is also only necessary in very specific
circumstances, and certainly for the initial tests I wouldn't worry too much
about doing that.  Only later, if you require very low latencies, would it
be worth looking into this (very low latency operation may also require the
RT kernel, but let's get the basics working first).

> Also, you can have a look at the output of :
> 
> >  jackd -d firewire -r 44100 -v5 > /tmp/broken.log 2>&1
> 
> here: https://gist.github.com/anonymous/5273530

This is very instructive because it proves that the device does in fact
start streaming and delivers a packet which looks half-way reasonable.  This
gives us a good start.  The primary problem is associated with the rate
calculation: ffado is unable to obtain sync with the device properly and
this would be the reason why it eventually exits with an error.  While this
could be a result of the non-preempt kernel, I don't think it is.  It is
more likely due to FFADO having the wrong idea about the packet contents,
and this in turn would be due to the incomplete testing we've been able to
do with the ultralite mk3.

To progress this, I think the first thing to do would be to install the
preempt kernel from your distribution's repository if such a kernel is
available from there.  It shouldn't take a lot of time and for all practical
purposes removes the kernel from the equation.  Once that's done I'd like
you to re-run jackd using the following command line:

  jackd -d firewire -r 44100 -v6 -n4 -p1024 > /tmp/ffado.log 2>&1

(use whatever name you like for the log).  Then post the resulting log. 
This will give me a better idea as to what FFADO thinks the device is doing. 
If you happen to know the configuration state of optional I/Os (ADAT for
example, but then again I can't remember if the ultralite mk3 has this) then
that would be useful to know.

Regards
  jonathan