Possible fix for 18452

[Asher Langton]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18452

This bug is involves extra -f* options being passed on to cc1 when a 
file is preprocessed.  In lang-spec.h, we have:

{"@f95-cpp-input",
   "cc1 -E -traditional-cpp -D_LANGUAGE_FORTRAN %(cpp_options) \
       %{E|M|MM:%(cpp_debug_options)}\
       %{!M:%{!MM:%{!E: -o %|.f95 |\n\
     f951 %|.f95 %(cc1_options) %{J*} %{I*}\
       %{!fsyntax-only:%(invoke_as)}}}}", 0, 0, 0},

The problem seems to be the spec string cpp_options  (output from -dumpspecs):

*cpp_options:
%(cpp_unique_options) %1 %{m*} %{std*&ansi&trigraphs} %{W*&pedantic*} 
%{w} %{f*} %{g*:%{!g0:%{!fno-working-directory:-fworking-directory}}} 
%{O*} %{undef} %{save-temps:-fpch-preprocess}

All of the -f* options are being passed along to cc1.  As a fix 
(though not very elegant), we can define a new spec string in place 
of cpp_options, replacing

"%{f*}"

with something like

"%{funderscoring|fsecond_underscore|...|fimplicit-none: ; f*:f%*}"

Unfortunately, this requires specifiying all of the flags we don't 
want to pass.  It might be easier to just replace %{f*} with an 
explicit list of the flags that should be passed, but changes to cc1 
might require updating this list, and so this approach would be 
harder to maintain

Any comments or suggestions?


-Asher

PS: The Cray Pointer patch should be released very soon.  I'm just 
waiting for the paperwork at this point.

-- 
Asher Langton
Institute for Scientific Computing Research
Lawrence Livermore National Laboratory
P.O. Box 808, L-312
Livermore, CA 94551
langton2@llnl.gov