solaris2 libjava 500 failures
Jeff Sturm
jsturm@one-point.com
Thu May 3 15:11:00 GMT 2001
On 3 May 2001, Alexandre Oliva wrote:
> On May 3, 2001, Jeff Sturm <jsturm@one-point.com> wrote:
> > LD_FOR_TARGET is `ld' for native builds. Hmm... that's not what gcc uses.
>
> Hmm... This gave me an idea. Does this help?
Thanks. This worked as expected once I added a few missing semicolons.
Revised patch follows.
(I was going to try something a little more drastic, removing the LD/AS/NM
definitions altogether since AFAICT they do no good. They aren't needed
for cross builds or unified trees, and overriding them doesn't do the
right thing anyway. I guess they are left over from some pre-autoconf
days?)
Index: Makefile.in
===================================================================
RCS file: /cvs/gcc/gcc/Makefile.in,v
retrieving revision 1.68.4.5
diff -u -p -r1.68.4.5 Makefile.in
--- Makefile.in 2001/04/13 10:24:35 1.68.4.5
+++ Makefile.in 2001/05/03 22:03:38
@@ -232,6 +232,8 @@ GCC_FOR_TARGET = $$r/gcc/xgcc -B$$r/gcc/
AS_FOR_TARGET = ` \
if [ -f $$r/gas/as-new ] ; then \
echo $$r/gas/as-new ; \
+ elif [ -f $$r/gcc/xgcc ]; then \
+ $(CC_FOR_TARGET) -print-prog-name=as ; \
else \
if [ "$(host_canonical)" = "$(target_canonical)" ] ; then \
echo $(AS); \
@@ -243,6 +245,8 @@ AS_FOR_TARGET = ` \
LD_FOR_TARGET = ` \
if [ -f $$r/ld/ld-new ] ; then \
echo $$r/ld/ld-new ; \
+ elif [ -f $$r/gcc/xgcc ]; then \
+ $(CC_FOR_TARGET) -print-prog-name=ld ; \
else \
if [ "$(host_canonical)" = "$(target_canonical)" ] ; then \
echo $(LD); \
@@ -298,6 +302,8 @@ RANLIB_FOR_TARGET = ` \
NM_FOR_TARGET = ` \
if [ -f $$r/binutils/nm-new ] ; then \
echo $$r/binutils/nm-new ; \
+ elif [ -f $$r/gcc/xgcc ]; then \
+ $(CC_FOR_TARGET) -print-prog-name=nm ; \
else \
if [ "$(host_canonical)" = "$(target_canonical)" ] ; then \
echo $(NM); \
More information about the Gcc-bugs
mailing list