This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Crossed-Native Build Weirdness with Current 3.3 Branch
- From: Mohan Embar <gnustuff at thisiscool dot com>
- To: "Ranjit Mathew" <rmathew at hotmail dot com>, gcc at gcc dot gnu dot org, mark at codesourcery dot com
- Cc: GCJ Java <java at gcc dot gnu dot org>
- Date: Sun, 20 Apr 2003 22:36:39 -0500
- Subject: Re: Crossed-Native Build Weirdness with Current 3.3 Branch
- Reply-to: gnustuff at thisiscool dot com
Ranjit,
>I'm getting some weirdness when I do the whole crossed, then crossed-native
>thing with the current 3.3 branch on MingW. Have you tried building with the latest
>3.3 code? For example, the crossed-native jar.exe crashes with a GPF (or whatever
>they call it nowadays).
I've traced this to this fix:
PR c++/7910
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/i386/winnt.c.diff?cvsroot=gcc&r1=1.42&r2=1.43
This can be verified by building a i686-pc-linux-gnu -> i386-pc-mingw32 cross and
the using this to build fastjar. When you try to execute the built jar.exe on Windows,
it GPFs. If you replace winnt.c with an older one and rebuild the cross compiler, it
compiles a working jar.exe. (Mark: Ranjit knows what I mean by all of this.) Unlike
the example in PR c++/7910, we build fastjar with --disable-shared.
Mark: My original message was posted to the java mailing list:
http://gcc.gnu.org/ml/java/2003-04/msg00313.html
I realize the above instructions to reproduce the bug are probably cryptic for you.
Ranjit and Mark: if you want me to help out with this any further, let me
know how. I realize that I probably haven't followed proper protocol,
haven't provided clear enough instructions to reproduce this,
have posted things to the wrong place, etc.
-- Mohan
http://www.thisiscool.com/
http://www.animalsong.org/