Re: [Qtractor-devel] Frequent crashes when moving audio parts in loop mode

[Rui N. C. <rn...@rn...>]

On Wed, 21 Oct 2009 00:47:37 +0200, Frank Neumann wrote:
> I manage to crash qtractor (build 1408) rather quickly by:
> 
> - creating a few audio tracks
> - loading a medium-size sample (10-20 seconds) into each track
> - setting a range for loop start and end point (just a few seconds from
>   somewhere in the middle)
> - activating loop mode and starting playback
> - and then moving the audio clips around during activated playback.
>

now that's pushing it way too hard :) although i've been practicing the
same dangerous juggling, i've did not come across any crashes of the sort.
but i know that things gcan get uglier depending on too many circumstances.
eg. sample file format, sample-rate, number of cpu cores, shared libraries
versions, whatever.

please try to find a pattern. do your sample files have anything special,
either common or distinguishable? does the number of tracks make any
difference? is there sample-rate conversion behind the scenes, inferring
that jackd sample rate is not the same of those files? are the number of
channels (mono, stereo, etc.) consistent with respective audio buses?
hardware? kernel? qt4 version? libsndfile? plugins?

yep, too many questions, but its a too many things that can go wrong in
there :)


> Normally, it then only takes a few seconds to make qtractor die,
sometimes
> unfortunately without it leaving a core image behind (despite ulimit -c
> unlimited - it even doesn't say "Segmentation fault" or similar in the
> xterm
> I start it from).
> 
> However, in the cases where I DO get a coredump, all that I find is:
> 
> franky@arwen:/tmp> gdb /usr/local/bin/qtractor ./core 
> [..]
> warning: core file may not match specified executable file.
> Core was generated by `gdb -q --batch --pid=19098 --eval-command=bt'.
> Program terminated with signal 6, Aborted.
> [New process 19233]
> #0  0x00007fa288a6afb5 in ?? ()
> (gdb) bt
> #0  0x00007fa288a6afb5 in ?? ()
> #1  0x00007fa288a6cbc3 in ?? ()
> #2  0x0000000000000001 in ?? ()
> #3  0x00000000011ccad0 in ?? ()
> #4  0x0000000000960043 in ?? ()
> #5  0x0000000000000001 in ?? ()
> #6  0x00000000011cc990 in ?? ()
> #7  0x00000000000003dc in ?? ()
> #8  0x0000000000963020 in ?? ()
> #9  0x0000000000648210 in Ui_qtractorTimeScaleForm::setupUi (this=0x0,
> qtractorTimeScaleForm=0x0) at .ui/ui_qtractorTimeScaleForm.h:105
> Backtrace stopped: previous frame inner to this frame (corrupt stack?)
> 
> Not sure if that helps in debugging..
> 

not quite :/

as i think telling you once or twice, try to reproduce the crash(es) with
a debug executable version (built from ./configure --enable-debug) then run
./qtractor from the console (xterm in your case). upon a crash, a gdb
stack-trace will be automagically spit out. 

try several times, and collect the back-traces on each occasion. try to
look and find a pattern on those too. meanwhile i'll try to do the same.

ah, and thanks for your patience ;)
cheers


> 
> PS: Did my previous mail regarding "sending bank/program changes at song
> start" make it to the list (Sunday 18th, 00:52)? It was a follow-up to
an
> older mail thread, so those using a threaded mail reader might have
missed
> that one..
> 

yes, afaicr it did. maybe i'm missing something.in the archives:
http://sourceforge.net/mailarchive/message.php?msg_name=E1MoPOb-0005TC-00%40smtp06.web.de
-- 
rncbc aka Rui Nuno Capela
rn...@rn...