Re: [FFADO-devel] MOTU Ultralite Hybrid

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

Hi Jeff

> I've got a MOTU Ultralite Hybrid.

Ah, cool.

> I downloaded the latest trunk, and compiled it with debug. I tried testing
> a simple streaming example, but it's not working.

Could you tell me exactly what you did here (as in what commands did you run
to run this "simple streaming example"?

> The problem appears to be in StreamProcessorManager::startDryRunning(). Up
> to that point, things seem to be going ok.

Could you please send me off-list the full output from jackd?  That will
confirm whether things are indeed running ok or if there's something else
going on behind the scenes.

> startDryRunning times out while waiting for the receive SP to 
> transition to dry running state. (The transmit SP does successfully 
> transition to dry-running state).

Hmm, that would tend to indicate that ffado didn't see any data coming from
the MOTU.  That is most strange.

> This appears like it's going to be difficult to debug ...

Not really.  I think it's most likely that the timeout is simply the last
thing in a chain of failures.  The earlier problems might be rather subtle
though so it may appear at first glance that the timeout is the only error.

The timeout on the rx stream is a little strange.  With the MOTUs once
they've been instructed to start streaming they should do so without any
further ado.  The fact that yours appears not to is therefore somewhat
unexpected.

> So when I tried to set some debug breakpoints, I found 
> that I triggered the time-out error ...

That would be true.  Breakpoint debugging of this particular fault is
unlikely to result in any insights though because the fundamental problem is
a lack of data from the MOTU.

> So is there someone familiar enough with the MOTU support I can talk to
> about helping analyze the debug output, or discuss about working around
> the timeout so that I can get to the real problem?

I wrote the MOTU driver so hopefully I can help. :-)

First a word about the hybrid.  I was working with someone earlier this year
with this device and we managed to get to the point where streaming did
appear to work, at least for some sample rates.  Furthermore, the streaming
system starts successfully for the 828Mk3 device which shares the same
streaming engine as the hybrid in firewire mode.  Therefore there's a
reasonable chance that the streaming system for these devices is fine; that
would indicate a lower level problem in your case.

As I mentioned above, the first step is for you to get me the full output
log from ffado so I can cast an eye over it to see if anything jumps out as
wrong.

To do this I recommend you run the following command:

  jackd -P70 -R -d firewire -r 48000 -p 1024 -n 4 -v 6 > ffadojack.log 2>&1

if your shell is bash, or

  jackd -P70 -R -d firewire -r 48000 -p 1024 -n 4 -v 6 >& ffadojack.log

if it's tcsh.  This will result in a rather large ffadojack.log file which
you can email to me off-list.  The contents of this will give me a place to
start looking for issues, and we can take things from there.

There are a few other things to ask at this point too.  Firstly, what
firewire chipset does your firewire port utilise.  You should be able to
identify it using the output of

  lspci | grep 1394

Also, do you know if your system uses a binary video card driver?

Finally, does your firewire card share its interrupt with anything?  This
can be determined by looking at the output from

  cat /proc/interrupts | grep ohci

Regards
  jonathan