[PlanetCCRMA] help debug realtime priority-induced bug in Kmid 2.3.0

Niels Mayer nielsmayer at gmail.com
Thu Jun 3 21:28:57 PDT 2010

FYI, this issue was resolved and didn't have anything to do w/ realtime prio
SVN commit 1134129 by pedrol:

ALSA backend. Removed a call to ALSAMIDIOutput::reloadDeviceList() from the
MIDI event handling thread, because this method is not thread safe and it is
not needed since 2.3.0 anyway. This fixes an ugly crash.
BUG: 240394

 M  +1 -1      CMakeLists.txt
 M  +2 -4      alsa/alsamidiobject.cpp
 M  +0 -3      alsa/alsamidioutput.cpp

WebSVN link: http://websvn.kde.org/?view=rev&revision=1134129

https://bugs.kde.org/show_bug.cgi?id=240391 remains open and
unconfirmed; similar to the other bug, I hypothesize that it's
multiple threads freeing the same piece of memory, or some other
thread-related malloc-clobbering. I can routinely crash on a 4 core
machine, and more infrequently on a 2 core machine. Haven't tried
locking to one core to see if that resolves the problem, or whether
the threading/corruption issue from the other bug is also the cause of
this problem.

-- Niels

More information about the PlanetCCRMA mailing list