Excerpts from Christopher Brannon's message of 2010-08-25 20:12:38 +0200:
Philipp Überbacher <hollunder@lavabit.com> writes:
As it happens, this article about algoscore says that it does require either -O2 or -O3 as CFLAGS: http://www.linuxjournal.com/content/algoscore-music-numbers
Sorry, but this doesn't seem plausible. I did get some useful output from valgrind. The invalid reads and writes are happening in src/nasal/hash.c, at line 132, in the resize function. The malloc call in that function doesn't seem to be allocating enough memory. The author is aligning his pointers, and the malloc call doesn't account for the padding. I added a fudge factor to malloc's size argument. The patch is here: <http://pastebin.ca/1925235>.
Unfortunately, I can't do much to test this program, but I can say that it no longer crashes with a backtrace.
-- Chris
The strange thing here is that it didn't crash when it was built from ROX-filer, which I believe just executes AppRun, a simple shell script to compile algoscore. One thing that causes it to crash here is 'csound_test.as' in the examples folder. The midi_test.as example works fine though. -- Philipp -- "Wir stehen selbst enttäuscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan