[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Swarm-Support] possible memory problem on mac g5
From: |
Bill Northcott |
Subject: |
Re: [Swarm-Support] possible memory problem on mac g5 |
Date: |
Tue, 6 Apr 2004 13:20:10 +1000 |
On 06/04/2004, at 12:12 PM, Paul Box wrote:
Have you tried the code on a G4 Mac?
I have tried it on a titanium notebook using osX 10.3, and I have the
same results (crash).
Which compiler are you using?
Paul-Boxs-Computer:~/swarm paulbox$ gcc -v
Reading specs from /usr/lib/gcc/powerpc-apple-darwin/3.3.1/specs
Configured with:
/Groups/billn/Public/Darwin/gcc3/build/obj/src/configure
--disable-checking --prefix=/usr --mandir=${prefix}/share/man
--enable-languages=c,objc,c++,objc++
--program-transform-name=/^[cg][^.-]*$/s/$/-3.3/
--with-gxx-include-dir=${prefix}/include/gcc/darwin/3.3/c++
--build=powerpc-apple-darwin --host=powerpc-apple-darwin
--target=powerpc-apple-darwin
Thread model: posix
gcc version 3.3.1 (Apple Computer, Inc. build 1555)
I have recently revived the Swarm test routines the build of which had
been broken by one of my fixes.
This revealed some problems on MacOS X and I have traced some of them
to broken compiler builtin functions used in sendmsg.c and one place in
defobj. Some of these problems had already been found and fixed in
late 2003 by the gcc maintainers. The patches will be in gcc3.4 but
have not yet made it into Apple sources (at least not the public ones).
I have built a patched version of the Apple compiler and this is now
available on my web site. It would be a good idea to use this.
However, you will probably also need to rebuild the libraries with the
patched compiler. I have not done this yet because I have not squashed
the last of the bugs. (I had to stop because of teaching commitments).
If you like, and I can find time, I will put up a debug build of the
partially patched Swarm libraries.
Other wise I will try your code on my setup and see if I can pin down
the problem.
Bill Northcott
If anyone is interested in trying this out for themselves, I have
placed a copy of the offending code at
http://geohive.nr.usu.edu/~sanduku/swarm/tyler-17-10-03.tar.gz
If you enter the string 't50' in the initialInfestation value in the
DataFileSwarm window, you should find the offending behavior
happening after 26 time steps.
The file is largish (5.3M)-- sorry about the size.
The specs on the new compiler are:
[PBG4-BN:~] billn% cc -v
Reading specs from /usr/lib/gcc/powerpc-apple-darwin/3.3.1/specs
Configured with:
/Users/billn/Public/Darwin/gcc_os-1555/build/obj/src/configure
--disable-checking --prefix=/usr --mandir=${prefix}/share/man
--enable-languages=c,objc,c++,objc++
--program-transform-name=/^[cg][^.-]*$/s/$/-3.3/
--with-gxx-include-dir=${prefix}/include/gcc/darwin/3.3/c++
--build=powerpc-apple-darwin --host=powerpc-apple-darwin
--target=powerpc-apple-darwin
Thread model: posix
gcc version 3.3.1 (Apple Computer, Inc. build 1555)