Hi!
On Wed, Jan 19, 2011 at 08:06:38AM -0800, Niels Mayer wrote:
On Sun, Jan 16, 2011 at 6:13 AM, Stefan Westerfeld
<stefan(a)space.twc.de> wrote:
Stefan --
Thanks for making this new version of gst123 available...
I tried my usual torture test on gst123 0.1.4 and it's still having
the same problem on the KCRW live stream, which IMHO stems from some
kind of memory leak related to changing of program text (since other
streams that don't change the program text don't seem to crash). The
crashes happen after several hours of playing, usually after a new
program text label is output on the stream. (note parsing oddities
seen before a crash e.g.
"Title : <HTML> Artist :
"
).
I ran the test by outputting to each of the stereo channels on an
ice1712 soundcard, which allows for five simultaneous gst123 instances
to run (see the end of
http://nielsmayer.com/npm/dot-asoundrc.txt for
details on the setup).
................................
gnulem-411-~/gst123-0.1.4> gst123 -a alsa=66ch12
http://64.12.61.1:80/stream/1046
...
Can you please send me a backtrace made with gdb instead of the one you sent
me? You should be able to get one by running
$ gdb --args gst123 -a alsa=66ch12 http://...
run
...(wait until crash)...
bt full
I've also added the command line option
$ gst123 --full-version
in the git version of gst123, having the output of this running on your machine
would also be helpful (in case your crash can only be reproduced with certain
library versions).
If you compile from git, you need to use
$ ./autogen.sh --enable-debug
to get debugging symbols (without them, the backtrace will not be useful).
Finally, you can try if your crash can be reproduced with the
$ gst123 -a none
option that I added recently, this should be easier (for instance if I want to
reproduce the crash here) than getting your ALSA setup working on another
machine.
Cu... Stefan
--
Stefan Westerfeld, Hamburg/Germany,
http://space.twc.de/~stefan