2.91.5 failed compilation.

[Jean-Pierre Radley <jpr@jpr.com>]

SCO OSR 5.0.4

2.91.4 compiled readily, but with 2.91.5, I end up with:

make[103]: Circular libobjc.a <- libgcc2.ready dependency dropped.
for i in `/s/tools/egcs/obj/gcc/xgcc -B/s/tools/egcs/obj/gcc/ --print-multi-lib`
; do \
  dir=`echo $i | sed -e 's/;.*$//'`; \
  flags=`echo $i | sed -e 's/^[^;]*;//' -e 's/@/ -/g'`; \
  make GCC_FOR_TARGET="/s/tools/egcs/obj/gcc/xgcc -B/s/tools/egcs/obj/gcc/" \
    AR="ar" AR_FLAGS="rc" CC="gcc" CFLAGS="-O3" \
    RANLIB=":" RANLIB_TEST="false" \
    LANGUAGES="c proto gcov  c++ f77 objc" \
    HOST_PREFIX="" HOST_PREFIX_1="loser-" \
    LIBGCC2_CFLAGS="-O2   -DIN_GCC    -O3 -I./include   -g1  -DIN_LIBGCC2 -D__GC
C_FLOAT_NOT_NEEDED -fexceptions
     ${flags}" \
    MULTILIB_CFLAGS="${flags}" \
    LIBGCC1="libgcc1.a" LIBGCC2="libgcc2.a" \
    dir="${dir}" stmp-multilib-sub; \
  if [ $? -eq 0 ] ; then true; else exit 1; fi; \
done
make[104]: Entering directory `/s/tools/egcs/obj/gcc'
rm -f libgcc2.a
if [ -d . ]; then \
  cd .; \
  rm -f libgcc.a ; \
else true; \
fi
make GCC_FOR_TARGET="/s/tools/egcs/obj/gcc/xgcc -B/s/tools/egcs/obj/gcc/" \
  AR="ar" AR_FLAGS="rc" CC="gcc" CFLAGS="-O3" \
  HOST_PREFIX="" HOST_PREFIX_1="loser-" \
  LANGUAGES="c proto gcov  c++ f77 objc" \
  LIBGCC2_CFLAGS="-O2   -DIN_GCC    -O3 -I./include   -g1  -DIN_LIBGCC2 -D__GCC_
FLOAT_NOT_NEEDED -fexceptions
   " libgcc2.a
make[105]: Entering directory `/s/tools/egcs/obj/gcc'
/bin//true: cannot open any more files
make[105]: *** [multilib.h] Error 1
make[105]: Leaving directory `/s/tools/egcs/obj/gcc'
make[104]: *** [stmp-multilib-sub] Error 2
make[104]: Leaving directory `/s/tools/egcs/obj/gcc'
make[103]: *** [stmp-multilib] Error 1
make[103]: Leaving directory `/s/tools/egcs/obj/gcc'
make[102]: *** [stmp-multilib-sub] Error 2
...etc...


I do note that the gcc/Makefile differs between the two versions at
line 2072 by the addition of $(STMP_FIXPROTO).

If I leave that out, then I get different errors, about FPIC
mixed with mcoff.

If I run make several times in a row, alternating with having and then
not having $(STMP_FIXPROTO) on that line, I eventually
succeed.

-- 
Jean-Pierre Radley <jpr@jpr.com>  XC/XT Custodian   Sysop, CompuServe SCOForum