This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: egcs-2.91.16 980328: gcc -O2 gets fatal signal 6 on regex.c
- To: Jim Wilson <wilson at cygnus dot com>
- Subject: Re: egcs-2.91.16 980328: gcc -O2 gets fatal signal 6 on regex.c
- From: Ulrich Drepper <drepper at cygnus dot com>
- Date: 06 Apr 1998 20:09:35 -0700
- Cc: meyering at eng dot ascend dot com, egcs-bugs at cygnus dot com, scox at cygnus dot com
- References: <199804070238.TAA11947@rtl.cygnus.com>
- Reply-To: drepper at cygnus dot com (Ulrich Drepper)
Jim Wilson <wilson@cygnus.com> writes:
> Richard has already installed his own patch. Please resubmit your own patch
> separately, instead of mixed up with other people's patches. It looks like
> you have some pic label stuff that we need.
In fact almost all the changes are done by Richard so I wanted to
provide him the possiblity to see which patch worked. But I certainly
can update my patch based on what is checked in.
> I don't see the point of any of this. My egcs build does define __i686__.
> Yours doesn't? Why?
The problem is a different. If you configure your gcc for
i686-linux-gnu and than later want to use the new gcc like this:
gcc -mpentium
the problem is that __i686__ still is defined. This is wrong since
now __i586__ (and __i386__) must be defined. In headers I want to use
#ifdef __i686__
some asm using i686 opcodes
#else
boring old i386 code
#endif
which is impossible with the current specs files (unless you
explicitly use -u__i686__).
-- Uli
---------------. drepper at gnu.org ,-. 1325 Chesapeake Terrace
Ulrich Drepper \ ,-------------------' \ Sunnyvale, CA 94089 USA
Cygnus Solutions `--' drepper at cygnus.com `------------------------