problem in line 2500 of ligcc-3.4.0/libffi/configure: === + echo (cached) \c (cached) + echo yes yes + test set = set CFLAGS=-O2 -g -O2 + test != yes /tmp/gcc-3.4.0/libffi/configure: test: argument expected make[1]: *** [configure-target-libffi] Error 1 make[1]: Leaving directory `/tmp/gcc_build' make: *** [bootstrap] Error 2 nwaibel@blade1000-2 /tmp/gcc_build $ === it comes actually from a macro in gcc-3.4.0/libffi/acinclude.m4 line 13: === AC_DEFUN([AC_FUNC_MMAP_BLACKLIST], [if test $ac_cv_header_sys_mman_h != yes \ || test $ac_cv_func_mmap != yes; then ac_cv_func_mmap_file=no ac_cv_func_mmap_dev_zero=no ac_cv_func_mmap_anon=no else === dblquotes around the variables ($ac_cv_header_sys_mman_h and $ac_cv_func_mmap) help, but i am not 100% sure if the variables should not contain sthg anyway (either yes or no)... my configure line: === ../gcc-3.4.0/configure --with-gnu-as --with-as=/misc/sparc-sun-solaris2. 8/bin/as --with-gnu-ld --with-ld=/misc/sparc-sun-solaris2.8/bin/ld --prefix=/misc/sparc-sun-solaris2.8/opt/gcc/3.4.0 --disable-shared --enable-languages=all --enable-threads --with-system-zlib --disable-nls --enable-multilib --with-cpu=ultrasparc ===
This is partly a dup of bug 15093 but note using solaris's sh will not work even after the fix for that bug is applied (which it is fixed for 3.4.1 already). Read <http://gcc.gnu.org/install/specific.html#*-*- solaris2*> for what you should do so you will not use solaris's sh but instead ksh. *** This bug has been marked as a duplicate of 15093 ***
Info: I ran into the same on Tru64 *with* ksh, but still trying with 3.4.0. checking whether /usr/inst/gcc-3.4.0/gcc/xgcc -B/usr/inst/gcc-3.4.0/gcc/ - B/usr/local/alphaev56-dec-osf5.1/bin/ -B/usr/local/alphaev56-dec-osf5.1/lib/ - isystem /usr/local/alphaev56-dec-osf5.1/include -isystem /usr/local/alphaev56- dec-osf5.1/sys-include accepts -g... (cached) yes /usr/inst/gcc-3.4.0/libffi/configure: test: argument expected make: *** [configure-target-libffi] Error 1