This is the mail archive of the
mailing list for the Java project.
Re: CVS MinGW (Crossed-)Native Compiler is Toast
- From: Mohan Embar <gnustuff at thisiscool dot com>
- To: Ranjit Mathew <rmathew at hotmail dot com>
- Cc: GCJ Java <java at gcc dot gnu dot org>
- Date: Tue, 23 Sep 2003 22:49:47 -0500
- Subject: Re: CVS MinGW (Crossed-)Native Compiler is Toast
- Reply-to: gnustuff at thisiscool dot com
First of all, thanks to you and everyone else who exhorted me to
perservere with GDB.
I think I have a faint idea of one of the things that seems to be
Line 10300 of parse.y has this call:
list = lookup_method_invoke (0, identifier_wfl, type, identifier, args);
Upon entry into this function, some prologue code seems to be
executed in gcc/gcc/config/i386/cygwin.asm which trashes
the first parameter, changing it from a 0 to garbage. It's a downward
spiral from there.
Does this ring a bell to you?
I am using the pristine sources from CVS which don't have any
mingw-local patches. Is this a recipe for disaster? cygwin.asm isn't
present in the 3.3 patchset. I do, however, see -Wl,--stack=0x00800000
in the LDFLAGS. I wonder if I need this.
I'm just sort of talking out loud here.
I haven't examined in detail what's going awry in cygwin.asm because
I wanted to see if this rang a bell with you. I'll try rebuilding with
the --stack switch and going to bed.
(Hmm. Not really a gcj-related post.)