[src/trunk]: src/sys/arch/mips/conf gcc does not pass floating point options ...

[macallan <macallan%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/eaaf06b5f01a
branches:  trunk
changeset: 813789:eaaf06b5f01a
user:      macallan <macallan%NetBSD.org@localhost>
date:      Thu Feb 18 20:50:44 2016 +0000

description:
gcc does not pass floating point options to the assembler
by default, because it is afraid that the stricter tests
will break userland code. The new binutils is pickier about
this. Gcc 5.x fixes the issue so for now, set explicitly
the assembler soft-float flags when we build the kernel.
see: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64569

diffstat:

 sys/arch/mips/conf/Makefile.mips |  14 +++++++++++++-
 1 files changed, 13 insertions(+), 1 deletions(-)

diffs (28 lines):

diff -r ef1f0832db7b -r eaaf06b5f01a sys/arch/mips/conf/Makefile.mips
--- a/sys/arch/mips/conf/Makefile.mips  Thu Feb 18 20:33:40 2016 +0000
+++ b/sys/arch/mips/conf/Makefile.mips  Thu Feb 18 20:50:44 2016 +0000
@@ -1,4 +1,4 @@
-#      $NetBSD: Makefile.mips,v 1.62 2015/08/24 14:04:24 uebayasi Exp $
+#      $NetBSD: Makefile.mips,v 1.63 2016/02/18 20:50:44 macallan Exp $
 
 # Makefile for NetBSD
 #
@@ -46,6 +46,18 @@
 CPPFLAGS+=     -D${MACHINE}
 DEFGP?=                -G 0
 GP?=           ${DEFGP}
+
+# XXX
+# gcc does not pass floating point options to the assembler
+# by default, because it is afraid that the stricter tests
+# will break userland code. The new binutils is pickier about
+# this. Gcc 5.x fixes the issue so for now, set explicitly
+# the assembler soft-float flags when we build the kernel.
+# see: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64569
+
+CFLAGS+=       -Wa,-msoft-float
+COPTS.mips_fpu.c+=     -Wa,-mhard-float
+
 CFLAGS+=       ${GP} -mno-abicalls -msoft-float -ffixed-24
 .if defined(LP64) && ${LP64} == "yes"
 .if ${MACHINE_ARCH} == "mips64eb" || ${MACHINE_ARCH} == "mips64el"