Re: [Jack-Devel] FFADO issues after freewheeling (was: Re: update jack from 1.9.7 to 1.9.8)

On Thu, Mar 08, 2012 at 11:03:57AM +0000, Fons Adriaensen wrote:
> On Thu, Mar 08, 2012 at 09:14:43PM +1030, Jonathan Woithe wrote:
>  
> > In any case, what this tends to prove is that the basic layers of FFADO are
> > just fine with freewheeling.  The question simply comes down to whether the
> > driver in use is able to successfully restart streaming.  If they can't it's
> > definitely a bug which needs dealing with, but it's bug in the driver rather
> > than the FFADO infrastructure.  Obviously I'll look into the Fireface issue
> > since I now know about it; if freewheeling causes trouble with other
> > interfaces please let us know so they too can be investigated.
> 
> Why should any part of FFADO be aware of freewheeling ? 
> Basic I/O just continues, silence is output, input is
> discarded. The only difference is that the rest of the
> cycle becomes independent of this.

FFADO isn't aware of freewheeling as such.  However, when freewheeling is
started the backend streaming is stopped for the reasons Paul mentioned. 
When jack is taken out of freewheeling the backend is restarted again. 
FFADO sees the stop/start cycle but doesn't know (or care) that this is
because of freewheeling.

I would expect that if FFADO drivers have difficulties with freewheeling it
will generally be when jack is taken out of freewheeling mode.  Stopping the
streaming (moving into freewheeling) is well debugged since it has to happen
when FFADO is closed down.  It's the path involving the restart of streaming
with previously used objects that has seen much less testing (and it's
something I never even thought of testing until this subject came up).

Regards
  jonathan

1331213775.25681_0.ltw:2,a <20120308133541.GA2733 at marvin dot atrad dot com dot au>