Gcc head has been failing to bootstrap for me for weeks now. I'm astounded to find this going on, and not finding a match in bugzilla; I don't think I have anything special in my set-up. This is a vanilla Mac G$ running OS X 10.3.5 "Panther". Is nobody else seeing this? No options to configure except a --prefix. The bootstrap ends with: ./xgcc -B./ -B/Users/bothner/GNU/install-gcc/powerpc-apple-darwin7.5.0/bin/ -isy stem /Users/bothner/GNU/install-gcc/powerpc-apple-darwin7.5.0/include -isystem / Users/bothner/GNU/install-gcc/powerpc-apple-darwin7.5.0/sys-include -L/Users/bot hner/GNU/build-gcc.head/gcc/../ld -DIN_GCC -W -Wall -Wwrite-strings -Wstrict- prototypes -Wmissing-prototypes -Wold-style-definition -isystem ./include -I. -I. -I/Users/bothner/GNU/gcc.head/gcc -I/Users/bothner/GNU/gcc.head/gcc/. -I/Use rs/bothner/GNU/gcc.head/gcc/../include -I./../intl -I/Users/bothner/GNU/gcc.head /gcc/../libcpp/include \ -c /Users/bothner/GNU/gcc.head/gcc/config/darwin-crt2.c -o crt2.o /var/tmp//ccLVOez4.s:1:Unknown pseudo-op: .machine /var/tmp//ccLVOez4.s:1:Rest of line ignored. 1st junk character valued 112 (p). /var/tmp//ccLVOez4.s:319:Unknown pseudo-op: .subsections_via_symbols make[2]: *** [crt2.o] Error 1
You need a new cctools see: http://gcc.gnu.org/ml/gcc-patches/2004-10/msg01222.html for details.
I think this should remain open as long as we don't have a configure-time check to warn about an invalid assembler. Otherwise the support burden will be intolerable. Kelley's configure patch to check the assembler was reverted on Oct 20 (I missed why); we clearly need *something*.
It was reverted because it made a cross compiler to powerpc-darwin imposiable if you only wantted a cc1 and not a full fledge compiler (which is useful when debuging bugs).
Confirmed, this is minor as the problem is that you need a new binutils as reported before.
It's minor if we accept that people who run Darwin without the updated binutils and try to build from source will get an error message without any clue about what to do. Perhaps we can hope there aren't very many such people, or all/most of them will have read the release notes. But it seems to me obvious that re-applying Kelley's patch (imperfect though it might be) is an improvement over the current situation.
*** Bug 18627 has been marked as a duplicate of this bug. ***
Subject: Bug 18142 CVSROOT: /cvs/gcc Module name: gcc Changes by: bothner@gcc.gnu.org 2004-11-28 06:06:29 Modified files: gcc : configure.ac configure ChangeLog Log message: PR bootstrap/18142 Re-applied following, with two modifications: add a URL for the fixed cctools; only exit if $build = $target. 2004-10-18 Kelley Cook <kcook@gcc.gnu.org> * configure.ac (powerpc-*-darwin*): Require assembler to support .machine directive. * configure: Regenerate. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/configure.ac.diff?cvsroot=gcc&r1=2.85&r2=2.86 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/configure.diff?cvsroot=gcc&r1=1.876&r2=1.877 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.6594&r2=2.6595
I consider the bug fixed now that configure gives a useful error message telling people to get a fixed cctools. Though with one reservation: 18627 comment #2 reports problems with installing the new cctools.