6 Jun
2015
6 Jun
'15
3:24 a.m.
Hallo,
I'm an active FreeBSD user and I'm trying to nail down a problem of
zombies caused by KDE components which make use of jack-audio-connection-kit-0.121.3
To get this debugged I have inserted a wrapper script as
/usr/local/bin/jackd which does some logging and calls at the end the
real jackd as jackd.bin (see below). I've stumbled over something in the
source and man page of jackd which I do not understand:
The man page explains the flag -l as for changing the LISTEN port, while
the implementation in the C-source use the flag -l to just print the
actual working directory (/tmp) and do exit(0); see below.
Perhaps, the KDE components do not expect the termination of the forked
jackd server and resulting to state zombie.
Do I misunderstand something stupidly wrong? Thanks for a helping hand.
matthias
----- Forwarded message from Matthias Apitz <guru(a)unixarea.de> -----
Date: Fri, 5 Jun 2015 20:04:53 +0200
From: Matthias Apitz <guru(a)unixarea.de>
To: Tobias Berner <tcberner(a)gmail.com>om>, kde-freebsd(a)kde.org, multimedia(a)freebsd.org
Subject: [kde-freebsd] KDE && jackd flag -l (was: kbiff && zombies)
I have modified the shell wrapper script for jackd as:
#!/bin/sh
printf "\nnew jackd call, pid: %d\n" $$ >> /tmp/jackd
echo args: $0 $* >> /tmp/jackd
/usr/local/bin/jackd.bin $* &
echo last pg pid: $! >> /tmp/jackd
ps axl >> /tmp/jackd
and it turns out that all KDE components, already on KDE start the
/usr/local/bin/knotify4, are calling jackd as
/usr/local/bin/jackd -l
which is fine according the man page of jackd:
-l, --listen-port int
The socket port we are listening on for sync
packets (default: 3000)
but the source of jackd implements with -l only the printout of the
current working directory and exit(0):
case 'l':
/* special flag to allow libjack to determine
* jackd's idea of where tmpdir is */
printf ("%s\n", jack_tmpdir);
exit (0);
I have search in the KDE sources, without any luck, where 'jackd -l' is
issued to understand the problem; no luck; and for the moment I only can
say that there is some missimplementation, either in KDE or in jackd.
matthias
_______________________________________________
kde-freebsd mailing list
kde-freebsd(a)kde.org
https://mail.kde.org/mailman/listinfo/kde-freebsd
See also http://freebsd.kde.org/ for latest information
----- End forwarded message -----
--
Matthias Apitz, guru(a)unixarea.de, http://www.unixarea.de/ +49-170-4527211
+49-176-38902045
"Wenn der Mensch von den Umständen gebildet wird, so muß man die Umstände menschlich
bilden."
"Si el hombre es formado por las circunstancias entonces es necesario formar
humanamente
las circunstancias", Karl Marx in Die heilige Familie / La sagrada familia (MEW 2,
138)
6 Jun
6 Jun
6:47 a.m.
On Sat, Jun 6, 2015 at 4:24 AM, Matthias Apitz <guru(a)unixarea.de> wrote:
Hallo,
I'm an active FreeBSD user and I'm trying to nail down a problem of
zombies caused by KDE components which make use of
jack-audio-connection-kit-0.121.3
To get this debugged I have inserted a wrapper script as
/usr/local/bin/jackd which does some logging and calls at the end the
real jackd as jackd.bin (see below). I've stumbled over something in the
source and man page of jackd which I do not understand:
The man page explains the flag -l as for changing the LISTEN port, while
the implementation in the C-source use the flag -l to just print the
actual working directory (/tmp) and do exit(0); see below.
You're confused.
JACK Command line structure:
jackd [ server options ] -d [ backend name ] [ backend options ]
the same option may appear twice, once as a server option and once as a
backend option. The semantics of each appearance are totally different.
-l is both a server option (meaning: print out the location of the
directory where FIFOs exist to communicate with the the server), and an
option for the net1 backend (meaning: define the listen port for sync
packets).
7:11 a.m.
El d�a Saturday, June 06, 2015 a las 07:47:58AM -0400, Paul Davis escribi�:
On Sat, Jun 6, 2015 at 4:24 AM, Matthias Apitz
<guru(a)unixarea.de> wrote:
Maybe that I'm confused. But, the wrapper script shows that all the KDE
components are only exacuting:
/usr/local/bin/jackd -l
without any further args and without any jackd running after this as
some server. Why the KDE folks are doing this? Are they confused too?
matthias
Hallo,
I'm an active FreeBSD user and I'm trying to nail down a problem of
zombies caused by KDE components which make use of
jack-audio-connection-kit-0.121.3
To get this debugged I have inserted a wrapper script as
/usr/local/bin/jackd which does some logging and calls at the end the
real jackd as jackd.bin (see below). I've stumbled over something in the
source and man page of jackd which I do not understand:
The man page explains the flag -l as for changing the LISTEN port, while
the implementation in the C-source use the flag -l to just print the
actual working directory (/tmp) and do exit(0); see below.
You're confused.
JACK Command line structure:
jackd [ server options ] -d [ backend name ] [ backend options ]
the same option may appear twice, once as a server option and once as a
backend option. The semantics of each appearance are totally different.
-l is both a server option (meaning: print out the location of the
directory where FIFOs exist to communicate with the the server), and an
option for the net1 backend (meaning: define the listen port for sync
packets).
8:21 a.m.
On Sat, Jun 6, 2015 at 8:11 AM, Matthias Apitz <guru(a)unixarea.de> wrote:
Maybe that I'm confused. But, the wrapper script shows that all the KDE
components are only exacuting:
/usr/local/bin/jackd -l
without any further args and without any jackd running after this as
some server. Why the KDE folks are doing this? Are they confused too?
that command just prints out the location of the directory where the FIFOs
to communicate with the server can be found.
8:54 a.m.
El sábado, 6 de junio de 2015 15:21:31 (CEST), Paul Davis escribió:
On Sat, Jun 6, 2015 at 8:11 AM, Matthias Apitz
<guru(a)unixarea.de> wrote:
And what should be issued by the KDE components to start a server? Thanks.
--
Sent from my Ubuntu phone
http://www.unixarea.de/
Maybe that I'm confused. But, the wrapper script shows that all the KDE
components are only exacuting:
/usr/local/bin/jackd -l
without any further args and without any jackd running after this as
some server. Why the KDE folks are doing this? Are they confused too?
that command just prints out the location of the directory where the FIFOs
to communicate with the server can be found.
8:54 a.m.
On Sat, June 6, 2015 7:11 am, Matthias Apitz wrote:
Maybe that I'm confused. But, the wrapper script
shows that all the KDE
components are only executing:
/usr/local/bin/jackd -l
without any further args and without any jackd running after this as
some server. Why the KDE folks are doing this? Are they confused too?
That seems very odd, usually KDE would have no need to use jackd.
Where is that call to jackd -l made? I can check on a Fedora distribution.
Is that really in the KDE sources, and not something added by the
distribution packagers?
--
Chris Caudle
1:49 p.m.
El día Saturday, June 06, 2015 a las 08:53:54AM -0500, Chris Caudle escribió:
On Sat, June 6, 2015 7:11 am, Matthias Apitz wrote:
As I said, I substituted /usr/local/bin/jackd with some shell script to
catch the calls, log its arguments and check with shell commands the PID and the process
chain. The calls of '/usr/local/bin/jackd -l' are coming down from:
/usr/local/bin/knotify4
on start of the desktop via 'startx' and from
/usr/local/bin/kbiff
when mail arrives to the mbox.
matthias
--
Matthias Apitz, guru(a)unixarea.de, http://www.unixarea.de/ +49-170-4527211
+49-176-38902045
"Wenn der Mensch von den Umständen gebildet wird, so muß man die Umstände menschlich
bilden."
"Si el hombre es formado por las circunstancias entonces es necesario formar
humanamente
las circunstancias", Karl Marx in Die heilige Familie / La sagrada familia (MEW 2,
138)
Maybe that I'm confused. But, the wrapper
script shows that all the KDE
components are only executing:
/usr/local/bin/jackd -l
without any further args and without any jackd running after this as
some server. Why the KDE folks are doing this? Are they confused too?
That seems very odd, usually KDE would have no need to use jackd.
Where is that call to jackd -l made? I can check on a Fedora distribution.
Is that really in the KDE sources, and not something added by the
distribution packagers?
3518
days inactive
3518
days old
jackaudio@lists.linuxaudio.org
6 comments
3 participants
participants (3)
-
Chris Caudle -
Matthias Apitz -
Paul Davis