[PlanetCCRMA] jack memory leak?

Mark Knecht mknecht@controlnet.com
Wed Nov 13 15:28:01 2002


Kai,
   Actually I sent this message to the Planet first, instead of the jack
list, since I was actually running jackstart, a program provided by
Fernando, and not the standard jack. I didn't want to involve the real jack
folks until I'd given Fernando a chance to comment. Fernando compiled this
0.40.0 version a few days ago. I did not notice the problem on the previous
jackstart, but it could have been there.

   I am certainly NOT saying there is a problem with jack at this point, but
it was the only program I started by hand, so I'm curious. It could just as
easily be caused by some other daemon or some other thing I know nothing
about at this point. I'm just asking around. Maybe it's known about to
people more involved than I.

   Jack ran idle the whole time, including the last couple of hours. I'm now
at about 87448K, so it's up a bit, but not too much.

Mark

-----Original Message-----
From: Kai Vehmanen [mailto:kai.vehmanen@wakkanet.fi]
Sent: Wednesday, November 13, 2002 3:04 PM
To: Mark Knecht
Cc: PlanetCCRMA (E-mail); jackit-devel@lists.sourceforge.net
Subject: Re: [PlanetCCRMA] jack memory leak?


Did you use jack during the 4 hours (and if yes, how) or was it running
idle? I've run the current CVS-version of jackd under valgrind (memory
debugger) and it didn't detect any notable memory leaks.

OTOH, although valgrind is a superb tool, it's not 100% perfect
(especially with complex apps like jackd), so leaks are still a
possibility.

On Wed, 13 Nov 2002, Mark Knecht wrote:

>    I'm early in looking for this, but wondered if anyone else has seen any
> evidence of jack having a memory leak? I've noticed over the course of a
day
> or two that memory is getting used up and not released, at least according
> to 'top'. In most of those cases I've run lots of applications and
couldn't
> say who the culprit was, and I still really can't.
>
>    However, this morning I rebooted my machine and tried an experiment.
>
> 1) Reboot, open 1 Eterm, and run top - 50128K used
>
> 2) Start Jack (using PlanetCCRMA jackstart, if important) - 84776K used
>
> 3) 4 hours later - 87380 used
>
>    It's not a perfect experiment as I used the Eterm that was open to run
> nslookup and dig a few times, but it is interesting.
>
>    Can anyone else confirm or deny this?
>
> Thanks,
> Mark
>
> _______________________________________________
> PlanetCCRMA mailing list
> PlanetCCRMA@ccrma.stanford.edu
> http://ccrma-mail.stanford.edu/mailman/listinfo/planetccrma
>

--
 http://www.eca.cx
 Audio software for Linux!