27 Mar
2011
27 Mar
'11
9:14 p.m.
------------------------------
Message: 2
Date: Sat, 26 Mar 2011 16:04:51 +0100 (CET)
From: Julien Claassen <julien(a)c-lab.de>
Subject: [LAU] persisting jack_connect problem (connection process
doesn't finish)
To: linux-audio-user <linux-audio-user(a)lists.linuxaudio.org>
Cc: Linux-Audio-Dev <linux-audio-dev(a)lists.linuxaudio.org>
Message-ID: <alpine.DEB.2.00.1103261601520.901(a)britney.spears.net>
Content-Type: TEXT/PLAIN; format=flowed; charset=US-ASCII
Hello Everyone1
Sorry for cross posting. I just repulled Jackmp from svn, but the problem
continues.
Looking at the jack_connect source I can't see anything wrong, so it seems,
that the server doesn't properly notify jack_connect/jack_disconnect of the
finishing of the connection/disconnection process.
Does anyone else have that problem or does anyone have a possible solution?
many thanks and warm regards
Julien
What is the problem in the first place? What applications? what context?
Thanks.
Stéphane
27 Mar
27 Mar
9:23 p.m.
New subject: persisting jack_connect problem
Hello Stephane!
The problem in the first place is jack_connect - or maybe the jackd part
behind it). Jack_connect is the commandline based connection/disconnection
client, that comes with jack (and jackmp).
I use it on the commandline to connect two ports, did you never come across
it?
jack_connect system:capture_3 system:playback_1
Something like this. The normal behaviour is:
connect inut to output
Print a short message about engine samplerate or something
exit.
That way you can use it for all kinds of text based purposes or for
scripting.
But if the connection is made, yet the aplication doesn't return somehow
(from what I could gather, stuck somewhere in the connection proces, perhaps
not getting the clear signal), then you can't script with it and even normal
manual usage on the commandline is VERY awkward.
There is a callback in jack_connect, which is given to jack. It just sets a
simple integer from 0 to 1. Somewhere in the program (after the connection
call), there's a while-loop waitng for this variable to become 1. I suspect,
there it sticks. Couldn't see any other problem there. Well maybe the
original call to the connection function from the jack API itself doesn't
return.
Any ideas, suggestions? I could put some debugging prints in there to test
further, but I thought, that someone else might have noticed this and could
give me a hint or just a note, if/when it's fixed. As I said, I repulled three
days after the first build and there were changes, but nothing changed for
jack_connect.
Warm regards
Julien
--------
Music was my first love and it will be my last (John Miles)
======== FIND MY WEB-PROJECT AT: ========
http://ltsb.sourceforge.net
the Linux TextBased Studio guide
======= AND MY PERSONAL PAGES AT: =======
http://www.juliencoder.de
28 Mar
28 Mar
9:56 a.m.
New subject: persisting jack_connect problem
----- "Julien Claassen" <julien(a)c-lab.de> wrote:
There is a callback in jack_connect, which is given
to jack. It
just sets a
simple integer from 0 to 1. Somewhere in the program (after the
connection
call), there's a while-loop waitng for this variable to become 1. I
suspect,
there it sticks. Couldn't see any other problem there. Well maybe the
try to change:
int done = 0;
to:
int volatile done = 0;
in example-clients/connect.c
(if that's the place you're talking about)
10:22 a.m.
New subject: persisting jack_connect problem
Le 28 mars 2011 à 09:56, Cedric Roux a écrit :
----- "Julien Claassen"
<julien(a)c-lab.de> wrote:
Hi Julien,
I'll try to do some deeper tests ASAP (hope this week).
Stéphane
There is a callback in jack_connect, which is
given to jack. It
just sets a
simple integer from 0 to 1. Somewhere in the program (after the
connection
call), there's a while-loop waitng for this variable to become 1. I
suspect,
there it sticks. Couldn't see any other problem there. Well maybe the
try to change:
int done = 0;
to:
int volatile done = 0;
in example-clients/connect.c
(if that's the place you're talking about)
10:23 a.m.
New subject: persisting jack_connect problem
Hello Cedric!
I just changed it to int volatile done = 0 and added some debugging output.
It jumps into the connecting branch (if(connect) ), then there's the
expression:
if (jack_connect(client, jack_port_name(src_port),...
There it fails (i.e., it doesn't execute the following goto exit. Instead,
it executed my prepare else and then hung around the while-loop forever. Well,
I didn't wait that long. :-)
So any thoughts on that?
Warmly yours
Julien
--------
Music was my first love and it will be my last (John Miles)
======== FIND MY WEB-PROJECT AT: ========
http://ltsb.sourceforge.net
the Linux TextBased Studio guide
======= AND MY PERSONAL PAGES AT: =======
http://www.juliencoder.de
10:27 a.m.
New subject: persisting jack_connect problem
Le 28 mars 2011 à 10:23, Julien Claassen a écrit :
Hello Cedric!
I just changed it to int volatile done = 0 and added some debugging output.
It jumps into the connecting branch (if(connect) ), then there's the expression:
if (jack_connect(client, jack_port_name(src_port),...
There it fails (i.e., it doesn't execute the following goto exit. Instead, it
executed my prepare else and then hung around the while-loop forever. Well, I didn't
wait that long. :-)
So any thoughts on that?
Warmly yours
Julien
The only reason I see is that for some reason the connect callback is never called... so
*this* has to be debugged in the first place.
Stéphane
10:53 a.m.
New subject: persisting jack_connect problem
----- "Julien Claassen" <julien(a)c-lab.de> wrote:
Hello Cedric!
I just changed it to int volatile done = 0 and added some debugging
output.
It jumps into the connecting branch (if(connect) ), then there's
the
expression:
if (jack_connect(client, jack_port_name(src_port),...
There it fails (i.e., it doesn't execute the following goto exit.
Instead,
it executed my prepare else and then hung around the while-loop
forever. Well,
I didn't wait that long. :-)
So any thoughts on that?
Warmly yours
Julien
Julien,
add the lines:
/* tell the JACK server that we are ready to roll */
if (jack_activate (client)) {
fprintf (stderr, "cannot activate client");
goto exit;
}
just before the lines:
/* connect the ports. Note: you can't do this before
the client is activated (this may change in the future).
*/
if (connecting) {
and add:
jack_deactivate (client);
just after the lines:
// Wait for connection/disconnection to be effective
while(!done) {
#ifdef WIN32
Sleep(10);
#else
usleep(10000);
#endif
}
that does the trick for me.
You should keep the "volatile" stuff. Without it the program
may loop forever if compiled with gcc and some optimization.
Note: it is normal that it doesn't jump to exit. Go read the
sources of jack-1.9.6, it's clearer. It jumps to exit in case
of error. It should do the while(!done) if things are alright.
11:28 a.m.
New subject: persisting jack_connect problem
Hello Cedirc!
Thanks a lot! It does work now! Wonderful!
Warmly yours
Julien
--------
Music was my first love and it will be my last (John Miles)
======== FIND MY WEB-PROJECT AT: ========
http://ltsb.sourceforge.net
the Linux TextBased Studio guide
======= AND MY PERSONAL PAGES AT: =======
http://www.juliencoder.de
11:33 a.m.
New subject: persisting jack_connect problem
Le 28 mars 2011 à 10:53, Cedric Roux a écrit :
----- "Julien Claassen" <julien(a)c-lab.de> wrote:
OK commited those changes in JACK2 SVN 4223
Stéphane
Hello Cedric!
I just changed it to int volatile done = 0 and added some debugging
output.
It jumps into the connecting branch (if(connect) ), then there's
the
expression:
if (jack_connect(client, jack_port_name(src_port),...
There it fails (i.e., it doesn't execute the following goto exit.
Instead,
it executed my prepare else and then hung around the while-loop
forever. Well,
I didn't wait that long. :-)
So any thoughts on that?
Warmly yours
Julien
Julien,
add the lines:
/* tell the JACK server that we are ready to roll */
if (jack_activate (client)) {
fprintf (stderr, "cannot activate client");
goto exit;
}
just before the lines:
/* connect the ports. Note: you can't do this before
the client is activated (this may change in the future).
*/
if (connecting) {
and add:
jack_deactivate (client);
just after the lines:
// Wait for connection/disconnection to be effective
while(!done) {
#ifdef WIN32
Sleep(10);
#else
usleep(10000);
#endif
}
that does the trick for me.
You should keep the "volatile" stuff. Without it the program
may loop forever if compiled with gcc and some optimization.
Note: it is normal that it doesn't jump to exit. Go read the
sources of jack-1.9.6, it's clearer. It jumps to exit in case
of error. It should do the while(!done) if things are alright.
5057
days inactive
5058
days old
linux-audio-user@lists.linuxaudio.org
8 comments
3 participants
participants (3)
-
Cedric Roux -
Julien Claassen -
Stéphane Letz