Re: [FFADO-devel] Saffire Pro 24

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

Hi Bernardo

> > Ok, so what I take from this is that by manually starting jackd like this
> > with jack1 and libffado-svn, things seem to be working as expected.  Is that
> > right?
> 
> ``working'' means that jackd starts, but I get no sound.

Right, ok, thanks for clarifying.

> >> - if I start qjackctl I can see the CPU load and eventual xruns, even
> >> without sound
> >
> > Ok, that's interesting.  Again for my benefit: I understand from this that
> > you don't see xruns when running jackd manually, but if you start it via
> > qjackctl you start to get xruns.  Hmm.
> 
> I can only see the red xruns numbers in qjackctl GUI, but I don't hear
> them as I don't hear anything from the device. =)

That makes sense.  We might leave the qjackctl thing to one side for the
moment and solve the "no audio" problem first since that's obviously the
most pressing issue.

> >> - the audio levels in the MOTU do not change, as if they were "muted"
> >
> > What do you mean by this?  Are you saying that even with vlc playing you get
> > no audio from any of the MOTU outputs?  Or is this referring to inputs as
> > indicated on meterbridge (or some other application)?
> 
> Yes, no audio and no activity in the MOTU physical panel, where
> normally I would see meters going up and down according to the sound.

Right.  Hmm.  Is it possible for you to plug some sort of audio source into
one of the device's inputs and run jack meterbridge (or something) to see if
the audio from that input makes it into the computer?  You may have to use
ffado-mixer and/or the front panel mixer controls to set an appropriate
gain.

> > The good news here is that according to the debug log most ffado components
> > seem to be working.  I've just got to get it clear in my head what *wasn't*
> > working and then we can address that.
> 
> Yes, that's strange, isn't it? And maybe it also occurs with Saffire,
> as it seems it is the same kind of problem.

I am thinking along those lines, but it's still rather odd.  I can't quite
understand what could cause two totally different devices to behave like
this.

Ordinarily I would at this point ask you to double-check the jack port
connections to make sure that things are connected as you expect.  However,
even if these were wrong, with a non-zero "-v" setting as you had here you
still should have received a full-volume 1 kHz tone out of one of the analog
outputs (analog 1 I think on the Ultralite).  This tone is produced within
ffado itself and is completely independent of jack - it should start as soon
as jackd/ffado sync up with the device.

The thing which makes this all the more curious is that as far as I know
there is no "output mute" control on the Ultralite (cuemix has mutes for its
mix sends and mix masters, but that doesn't affect signals sent directly to
an output by the PC, as is happening with the 1 kHz test tone).

Hmm. You said you were working in a 64-bit environment, right?  The
generation of the test tone uses a pretty ugly pointer hack for speed, but
maybe that's broken on 64-bit Linux.  Having said that, "signed int" is
still 32 bits on a 64-bit Linux system AFAIK and all calls under
CondSwapToBus32() are explicit 32-bit types, so maybe it is ok.

You're certainly right about it being strainge since by all accounts jackd
and FFADO are correctly syncing with the device and controlling it.  I'm
going to keep thinking to see if anything springs to mind.  In the meantime
it would be great to know whether audio can be received from the device.

Regards
  jonathan