Subject: gcc4 status
To: None <>
From: matthew green <>
List: tech-toolchain
Date: 05/22/2006 14:33:00
hi folks.

here's the current list of remaining issues i know of:

	- sh3 kernels need movmem() and maybe friends from libgcc
	  implemented in libkern

	- make -j was known broken last i heard.

	- tools/Makefile.gmakehost and tools/gcc/Makefile need
	  cleanup.  they currently hard code using "/usr/bin/env -i"
	  to wipe out MAKEFLAGS from the environment (our make
	  and gmake flags aren't very compatible...)  there are
	  also issues with variables on the command line vs. the
	  environment vs. gmake -e.  (what's there, or will be in
	  a short while, works but it's ugly.)

	- sun2 (m68010) mknative fails as libobjc configure use the
	  -fobjc-exceptions option and m68000--netbsd-gcc does not
	  seem to support that option.  no idea why yet.

	- vax is not done at all

	- ns32k is missing and will stay that way.  sorry, simon. :-)

	- testing on many platforms.  i386, amd64 & sparc64 kernels
	  are known to work.  those and sparc/powerpc userlands works
	  but pretty much everything else needs testing.  (i guess
	  alpha kernels work, thanks drochner.)

	- pkgsrc.  dogcow@ has verified that firefox can build from
	  pkgsrc on i386, but a bulkbuild for some faster platforms
	  would be nice to have...

i386, x86_64 and sparc64 could probably switch now but i'm going to
leave that to the port masters / others.  powerpc could when people
verify that all the non-macppc powerpc platforms build..  i will
probably look at evbppc myself.