------- Comment #8 from howarth at nitro dot med dot uc dot edu 2006-09-09 19:44 ------- Eric, I've been running autoheader and autoconf in the gcc source subdirectory. It does cause configure to run the test which results in...
configure:13732: checking assembler for .literal16 configure:13745: /sw/lib/odcctools/bin/as -o conftest.o conftest.s >&5 conftest.s:1:Unknown pseudo-op: .literal16 configure:13748: $? = 1 configure: failed program was .literal16 configure:13759: result: no in config.log. However the compilation of darwin.c still throws the errors I mentioned before about HAVE_GAS_LITERAL16 being undeclared. If I understand the proposed patch correctly, it will only create HAVE_GAS_LITERAL16 being set to 1 if the literal16 configure test passes but will leave it undefined otherwise. Don't we have to have it set to 0 if the configure test fails instead? Since HAVE_GAS_LITERAL16 isn't part of a preprocessor statement but is actual c code, doesn't it always have to be defined to some value? -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28995