From bug-request at octave dot org Wed Jan 19 22:43:28 2005 Subject: Re: Serious memory leak in octave (this may be a duplicate bug report) From: Dan Nessett To: Per Persson CC: Date: Wed, 19 Jan 2005 20:16:22 -0600 I downloaded octave 2.1.64 and built from source. The problem remains. Since I pulled the tarball into a directory separate from /sw, I had to play with some flags to get the compile and load to work. The main problem was that make kept complaining about Readline 4.2 not being installed. I have Readline 4.3.15 installed by Fink. Also, I had to use the fix for f77 errors. In order to get around these problems I used the following configure execute line: LDFLAGS=-L/sw/lib CPPFLAGS=-I/sw/include FLIBS=-lg2c ./configure --enable-shared --enable-dl --disable-static The make complained about multiple definitions, specifically (this sequence occurred many times): ld: warning multiple definitions of symbol _xerbla_ ../libcruft/blas-xtra/xerbla.o definition of _xerbla_ in section (__TEXT,__text)../libcruft/libcruft.dylib(single module) definition of _xerbla_ ld: warning multiple definitions of symbol _BC /sw/lib/libreadline.dylib(terminal.so) definition of _BC /sw/lib/libncurses.dylib(lib_termcap.lo) definition of _BC ld: warning multiple definitions of symbol _UP /sw/lib/libreadline.dylib(terminal.so) definition of _UP /sw/lib/libncurses.dylib(lib_termcap.lo) definition of _UP ld: warning multiple definitions of symbol _PC /sw/lib/libreadline.dylib(terminal.so) definition of _PC /sw/lib/libncurses.dylib(lib_tputs.lo) definition of _PC Also, when I ran make check, it failed with the following message: make -f octMakefile check make -C test check WARNING: Couldn't find the global config file. WARNING: Couldn't find tool init file Illegal Argument "OCTAVE=../src/octave" try "runtest --help" for option list make[2]: *** [check] Error 1 make[1]: *** [check] Error 2 make: *** [check] Error 2 However, I decided to run make install anyway. Octave seemed to come up fine. I then executed the two octave commands that cause the problem and observed the same behavior. In fact it actually got a bit worse. Octave expanded to 2.64G before the second command (p - i.e., print p) completed. Dan On 1/19/05 11:05 AM, "Per Persson" wrote: > > On Jan 19, 2005, at 19:05, Dan Nessett wrote: > >> Per, >> >> Moving to octave 2.1.50 doesn't fix the problem. As you can see from >> the >> following output from top, the number of pageouts greatly exceeds the >> number >> of pageins, which is another indication of a memory leak. >> >> Is there anyway I can help debug this problem? Is it possible this is >> a gcc >> 3.3 problem? I did not have the problem before I moved to it. > > I haven't seen that particular problem on 10.2, so I don't (yet) think > the compiler is to blame. > > Try building octave 2.1.64 from source. You might find some helpful > comments on the wiki and > in the mailing list archives. > > /Per > ------------------------------------------------------------- Octave is freely available under the terms of the GNU GPL. Octave's home on the web: http://www.octave.org How to fund new projects: http://www.octave.org/funding.html Subscription information: http://www.octave.org/archive.html -------------------------------------------------------------