patches/gcc/4.3.1/180-superh-default-multilib.patch
author "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Mon Jan 05 23:02:43 2009 +0000 (2009-01-05)
changeset 1126 1ab3d2e08c8b
parent 746 b150d6f590fc
permissions -rw-r--r--
Split CT_ExtractAndPatch in two: CT_Extract and CT_Patch:
- it is unworkable to have CT_ExtactAndPAtch cope with all those silly glibc addons:
- they can have 'short' (as 'ports') or 'long' (as glibc-ports-2.7) names
- patches are against eithe the short or long name, but non-uniformly use one or the other
- it is the reposibility of the component (glibc in this case) to handle corner cases such as those
- update all components to use the new functions

/trunk/scripts/build/tools/000-template.sh | 3 2 1 0 +-
/trunk/scripts/build/tools/100-libelf.sh | 3 2 1 0 +-
/trunk/scripts/build/tools/200-sstrip.sh | 3 2 1 0 +-
/trunk/scripts/build/kernel/linux.sh | 3 2 1 0 +-
/trunk/scripts/build/binutils.sh | 3 2 1 0 +-
/trunk/scripts/build/cc/gcc.sh | 3 2 1 0 +-
/trunk/scripts/build/debug/000-template.sh | 3 2 1 0 +-
/trunk/scripts/build/debug/100-dmalloc.sh | 3 2 1 0 +-
/trunk/scripts/build/debug/400-ltrace.sh | 3 2 1 0 +-
/trunk/scripts/build/debug/300-gdb.sh | 9 6 3 0 +++--
/trunk/scripts/build/debug/500-strace.sh | 7 3 4 0 ++--
/trunk/scripts/build/debug/200-duma.sh | 19 8 11 0 ++++------
/trunk/scripts/build/libc/glibc.sh | 14 12 2 0 ++++++-
/trunk/scripts/build/libc/uClibc.sh | 13 9 4 0 +++++--
/trunk/scripts/build/libc/eglibc.sh | 14 12 2 0 ++++++-
/trunk/scripts/build/gmp.sh | 3 2 1 0 +-
/trunk/scripts/build/mpfr.sh | 3 2 1 0 +-
/trunk/scripts/functions | 68 36 32 0 +++++++++++++++++++-----------------
18 files changed, 108 insertions(+), 69 deletions(-)
yann@552
     1
Original patch from gentoo: gentoo/src/patchsets/gcc/4.3.1/gentoo/53_all_gcc4-superh-default-multilib.patch
yann@552
     2
The gcc-3.x toolchains would contain all the targets by default.  With gcc-4,
yann@552
     3
you have to actually list out the multilibs you want or you will end up with
yann@552
     4
just one when using targets like 'sh4-linux-gnu'.
yann@552
     5
yann@552
     6
The resulting toolchain can't even build a kernel as the kernel needs to build
yann@552
     7
with the nofpu flag to be sure that no fpu ops are generated.
yann@552
     8
yann@552
     9
Here we restore the gcc-3.x behavior; the additional overhead of building all
yann@552
    10
of these multilibs by default is negligible.
yann@552
    11
yann@552
    12
http://bugs.gentoo.org/140205
yann@552
    13
yann@552
    14
diff -durN gcc-4.3.1.orig/gcc/config.gcc gcc-4.3.1/gcc/config.gcc
yann@552
    15
--- gcc-4.3.1.orig/gcc/config.gcc	2008-05-21 10:54:15.000000000 +0200
yann@552
    16
+++ gcc-4.3.1/gcc/config.gcc	2008-06-10 14:58:06.000000000 +0200
yann@552
    17
@@ -2278,7 +2278,7 @@
yann@552
    18
 	if test x${sh_multilibs} = x ; then
yann@552
    19
 		case ${target} in
yann@552
    20
 		sh64-superh-linux* | \
yann@552
    21
-		sh[1234]*)	sh_multilibs=${sh_cpu_target} ;;
yann@552
    22
+		sh[1234]*)	sh_multilibs=`cd ${srcdir}/config/sh ; echo t-mlib-sh[1-4]* | sed 's:t-mlib-sh:,m:g;s: ::g'` ;;
yann@552
    23
 		sh64* | sh5*)	sh_multilibs=m5-32media,m5-32media-nofpu,m5-compact,m5-compact-nofpu,m5-64media,m5-64media-nofpu ;;
yann@552
    24
 		sh-superh-*)	sh_multilibs=m4,m4-single,m4-single-only,m4-nofpu ;;
yann@552
    25
 		sh*-*-linux*)	sh_multilibs=m1,m3e,m4 ;;