Re: [patch, fortran] PR18428: use libcpp for preprocessing (review of C/C++ and fortran maintainers needed)

[Daniel Franke <franke dot daniel at gmail dot com>]

Follow-up to:
	http://gcc.gnu.org/ml/fortran/2008-05/msg00039.htm

On Sunday 04 May 2008 17:29:52 FX wrote:
> The Fortran parts are OK provided you answer the questions below and:
>    -- provide a patch for the news page (http://gcc.gnu.org/gcc-4.4/
> changes.html),
>    -- rewrite the doc for preprocessing (http://gcc.gnu.org/
> onlinedocs/gfortran/Preprocessing-and-conditional-compilation.html);

See additional mail for the doc/wwwdocs changes.
Questions are answered below.


> > @@ -354,6 +357,9 @@ gfc_post_options (const char **pfilename
> > +/* FIXME: return gfc_cpp_preprocess_only (); */
> I don't understand this FIXME: why should we do that? And why isn't
> it done here?

Added the explanation given in the mail cited above to the comment.


> > +d
> > +Fortran Joined
> > +-d[DIMN]	Dump details about macro names and definitions during
> > preprocessing
>
> Aren't you missing the 'U' in that list?

Yes. The patch was also lacking handling of 'U' - added.


> > +/* Objects defined in libcpp:  */
> > +static cpp_options *cpp_option = NULL;
> > +static cpp_reader *cpp_in = NULL;
>
> [...] another wording would be clearer.

This now reads: "/* Structures used with libcpp:  */"


> > +  /* Initialize CPP built-ins; '1' corresponds to 'flag_hosted'
> > +     in C, defines __STDC_HOSTED__?!  */
> > +  cpp_init_builtins (cpp_in, 1);
>
> I don't think we want to define __STDC_HOSTED__. On the other hand,
> does your patch still define the CPU, OS and OBJFMT builtins? (macros
> TARGET_CPU_CPP_BUILTINS, TARGET_OS_CPP_BUILTINS and
> TARGET_OBJFMT_CPP_BUILTINS) I think we should. Other macros that we
> used to emit and I think we will now miss include: __VERSION__, _LP64
> and __LP64__, __OPTIMIZE_SIZE__, __OPTIMIZE__, __FAST_MATH__,
> __SUPPORT_SNAN__, __FINITE_MATH_ONLY__, __pic__, __PIC__, __pie__,
> __PIE__, and the __SIZEOF_*__ macros (which are I think potentially
> useful).

I don't know what __STDC_HOSTED__ is used for, but cpp_init_builtins() defines 
it by default. Added as much as of the other macros as I could. See 
cpp_define_builtins() for more specific information. 

Please note that due to __SIZEOF_*, the initializaiton sequence had to 
reorganized slightly. The way it was, *_type_node were not initialized 
properly before using them in cpp_init_builtins().


> > +  if (gfc_cpp_option.working_directory
> > +      && gfc_cpp_option.preprocess_only && !
> > gfc_cpp_option.no_line_commands)
> > +    pp_dir_change (cpp_in, get_src_pwd ());
>
> I don't understand why it only happens with
> gfc_cpp_option.preprocess_only: what happens to debugging
> information? The documentation says that "This option is implicitly
> enabled if debugging information is enabled", is this also the case
> here?

No. I missed that completly; c-opts (sanitize_cpp_opts) had the corresponding 
snippet which I added to cpp.c (gfc_cpp_post_options). Thanks for catching 
this!


Attached update of the Fortran side of the previous patch not only fixes FX' 
requests, but also adds yet another option, namely -nocpp. Tobias B. asked 
for this as a possibility to disable preprocessing if it would otherwise be 
enabled by the respective file extensions.

Btw, the C-side of things still needs reviewing. Condensed patch:
	http://gcc.gnu.org/ml/fortran/2008-05/msg00248.html
	
Bootstrapped and regression tested on i686-pc-linux-gnu. Ok for trunk?

Regards

	Daniel

Index: cpp.c
===================================================================
--- cpp.c	(revision 0)
+++ cpp.c	(revision 0)
@@ -0,0 +1,991 @@
+#include "config.h"
+#include "system.h"
+#include "coretypes.h"
+#include "tm.h"
+#include "tree.h"
+#include "version.h"
+#include "flags.h"
+
+
+#include "options.h"
+#include "gfortran.h"
+#include "target.h"
+#include "toplev.h"
+#include "diagnostic.h"
+
+#include "../../libcpp/internal.h"
+#include "cpp.h"
+#include "incpath.h"
+
+
+/* Holds switches parsed by gfc_cpp_handle_option (), but whose
+   handling is deferred to gfc_cpp_init ().  */
+typedef struct
+{
+    enum opt_code code;
+    const char *arg;
+}
+gfc_cpp_deferred_opt_t;
+
+
+/* Defined and undefined macros being queued for output with -dU at
+   the next newline.  */
+typedef struct gfc_cpp_macro_queue
+{
+  struct gfc_cpp_macro_queue *next;	/* Next macro in the list.  */
+  char *macro;				/* The name of the macro if not
+					   defined, the full definition if
+					   defined.  */
+} gfc_cpp_macro_queue;
+static gfc_cpp_macro_queue *cpp_define_queue, *cpp_undefine_queue;
+
+struct
+{
+  /* Argument of -cpp, implied by SPEC;
+     if NULL, preprocessing disabled.  */
+  const char *temporary_filename;
+
+  const char *output_filename;          /* -o <arg>  */
+  int preprocess_only;                  /* -E  */
+  int discard_comments;                 /* -C  */
+  int discard_comments_in_macro_exp;    /* -CC  */
+  int print_include_names;              /* -H  */
+  int no_line_commands;                 /* -P  */
+  char dump_macros;                     /* -d[DMNU]  */
+  int dump_includes;                    /* -dI  */
+  int working_directory;                /* -fworking-directory  */
+  int no_predefined;                    /* -undef */
+  int standard_include_paths;           /* -nostdinc */
+  int verbose;                          /* -v */
+
+  const char *multilib;                 /* -imultilib <dir>  */
+  const char *prefix;                   /* -iprefix <dir>  */
+  const char *sysroot;                  /* -isysroot <dir>  */
+
+  /* Options whose handling needs to be deferred until the
+     appropriate cpp-objects are created:
+      -A predicate=answer
+      -D <macro>[=<val>]
+      -U <macro>  */
+  gfc_cpp_deferred_opt_t *deferred_opt;
+  int deferred_opt_count;
+}
+gfc_cpp_option;
+
+/* Structures used with libcpp:  */
+static cpp_options *cpp_option = NULL;
+static cpp_reader *cpp_in = NULL;
+
+/* Defined in toplev.c.  */
+extern const char *asm_file_name;
+
+
+
+
+/* Encapsulates state used to convert a stream of cpp-tokens into
+   a text file.  */
+static struct
+{
+  FILE *outf;			/* Stream to write to.  */
+  const cpp_token *prev;	/* Previous token.  */
+  const cpp_token *source;	/* Source token for spacing.  */
+  int src_line;			/* Line number currently being written.  */
+  unsigned char printed;	/* Nonzero if something output at line.  */
+  bool first_time;		/* cb_file_change hasn't been called yet.  */
+} print;
+
+/* General output routines.  */
+static void scan_translation_unit (cpp_reader *);
+static void scan_translation_unit_trad (cpp_reader *);
+
+/* Callback routines for the parser. Most of these are active only
+   in specific modes.  */
+static void cb_file_change (cpp_reader *, const struct line_map *);
+static void cb_line_change (cpp_reader *, const cpp_token *, int);
+static void cb_define (cpp_reader *, source_location, cpp_hashnode *);
+static void cb_undef (cpp_reader *, source_location, cpp_hashnode *);
+static void cb_def_pragma (cpp_reader *, source_location);
+static void cb_include (cpp_reader *, source_location, const unsigned char *,
+			const char *, int, const cpp_token **);
+static void cb_ident (cpp_reader *, source_location, const cpp_string *);
+static void cb_used_define (cpp_reader *, source_location, cpp_hashnode *);
+static void cb_used_undef (cpp_reader *, source_location, cpp_hashnode *);
+void pp_dir_change (cpp_reader *, const char *);
+
+static int dump_macro (cpp_reader *, cpp_hashnode *, void *);
+static void dump_queued_macros (cpp_reader *);
+
+
+static void
+cpp_define_builtins (cpp_reader *pfile)
+{
+  int major, minor, patchlevel;
+
+  /* Initialize CPP built-ins; '1' corresponds to 'flag_hosted'
+     in C, defines __STDC_HOSTED__?!  */
+  cpp_init_builtins (pfile, 0);
+
+  /* Initialize GFORTRAN specific builtins.
+     These are documented.  */
+  if (sscanf (BASEVER, "%d.%d.%d", &major, &minor, &patchlevel) != 3)
+    {
+      sscanf (BASEVER, "%d.%d", &major, &minor);
+      patchlevel = 0;
+    }
+  cpp_define_formatted (pfile, "__GNUC__=%d", major);
+  cpp_define_formatted (pfile, "__GNUC_MINOR__=%d", minor);
+  cpp_define_formatted (pfile, "__GNUC_PATCHLEVEL__=%d", patchlevel);
+
+  cpp_define (pfile, "__GFORTRAN__=1");
+  cpp_define (pfile, "_LANGUAGE_FORTRAN=1");
+
+  if (gfc_option.flag_openmp)
+    cpp_define (pfile, "_OPENMP=200505");
+
+
+  /* More builtins that might be useful, but are not documented
+     (in no particular order).  */
+  cpp_define_formatted (pfile, "__VERSION__=\"%s\"", version_string);
+
+  if (flag_pic)
+    {
+      cpp_define_formatted (pfile, "__pic__=%d", flag_pic);
+      cpp_define_formatted (pfile, "__PIC__=%d", flag_pic);
+    }
+  if (flag_pie)
+    {
+      cpp_define_formatted (pfile, "__pie__=%d", flag_pie);
+      cpp_define_formatted (pfile, "__PIE__=%d", flag_pie);
+    }
+
+  if (optimize_size)
+    cpp_define (pfile, "__OPTIMIZE_SIZE__");
+  if (optimize)
+    cpp_define (pfile, "__OPTIMIZE__");
+
+  if (fast_math_flags_set_p ())
+    cpp_define (pfile, "__FAST_MATH__");
+  if (flag_signaling_nans)
+    cpp_define (pfile, "__SUPPORT_SNAN__");
+
+  cpp_define_formatted (pfile, "__FINITE_MATH_ONLY__=%d", flag_finite_math_only);
+
+  /* Definitions for LP64 model. */
+  if (TYPE_PRECISION (long_integer_type_node) == 64
+      && POINTER_SIZE == 64
+      && TYPE_PRECISION (integer_type_node) == 32)
+    {
+      cpp_define (pfile, "_LP64");
+      cpp_define (pfile, "__LP64__");
+    }
+
+  /* Define NAME with value TYPE size_unit.
+     The C-side also defines __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__
+     __SIZEOF_PTRDIFF_T__, however, fortran seems to lack the
+     appropriate type nodes.  */
+
+#define define_type_sizeof(NAME, TYPE)                             \
+    cpp_define_formatted (pfile, NAME"="HOST_WIDE_INT_PRINT_DEC,   \
+                          tree_low_cst (TYPE_SIZE_UNIT (TYPE), 1))
+
+  define_type_sizeof ("__SIZEOF_INT__", integer_type_node);
+  define_type_sizeof ("__SIZEOF_LONG__", long_integer_type_node);
+  define_type_sizeof ("__SIZEOF_LONG_LONG__", long_long_integer_type_node);
+  define_type_sizeof ("__SIZEOF_SHORT__", short_integer_type_node);
+  define_type_sizeof ("__SIZEOF_FLOAT__", float_type_node);
+  define_type_sizeof ("__SIZEOF_DOUBLE__", double_type_node);
+  define_type_sizeof ("__SIZEOF_LONG_DOUBLE__", long_double_type_node);
+  define_type_sizeof ("__SIZEOF_SIZE_T__", size_type_node);
+
+#undef define_type_sizeof
+
+  /* The defines below are necessary for the TARGET_* macros.
+
+     FIXME:  Note that builtin_define_std() actually is a function
+     in c-cppbuiltin.c which uses flags undefined for Fortran.
+     Let's skip this for now. If needed, one needs to look into it
+     once more.  */
+
+# define builtin_define(TXT) cpp_define (pfile, TXT)
+# define builtin_define_std(TXT)
+# define builtin_assert(TXT) cpp_assert (pfile, TXT)
+
+  TARGET_CPU_CPP_BUILTINS ();
+  TARGET_OS_CPP_BUILTINS ();
+  TARGET_OBJFMT_CPP_BUILTINS ();
+
+#undef builtin_define
+#undef builtin_define_std
+#undef builtin_assert
+}
+
+bool
+gfc_cpp_enabled (void)
+{
+  return gfc_cpp_option.temporary_filename != NULL;
+}
+
+bool
+gfc_cpp_preprocess_only (void)
+{
+  return gfc_cpp_option.preprocess_only;
+}
+
+const char *
+gfc_cpp_temporary_file (void)
+{
+  return gfc_cpp_option.temporary_filename;
+}
+
+void
+gfc_cpp_init_options (unsigned int argc,
+		      const char **argv ATTRIBUTE_UNUSED)
+{
+  /* Do not create any objects from libcpp here. If no
+     preprocessing is requested, this would be wasted
+     time and effort.
+
+     See gfc_cpp_post_options() instead.  */
+
+  gfc_cpp_option.temporary_filename = NULL;
+  gfc_cpp_option.output_filename = NULL;
+  gfc_cpp_option.preprocess_only = 0;
+  gfc_cpp_option.discard_comments = 1;
+  gfc_cpp_option.discard_comments_in_macro_exp = 1;
+  gfc_cpp_option.print_include_names = 0;
+  gfc_cpp_option.no_line_commands = 0;
+  gfc_cpp_option.dump_macros = '\0';
+  gfc_cpp_option.dump_includes = 0;
+  gfc_cpp_option.working_directory = -1;
+  gfc_cpp_option.no_predefined = 0;
+  gfc_cpp_option.standard_include_paths = 1;
+  gfc_cpp_option.verbose = 0;
+
+  gfc_cpp_option.multilib = NULL;
+  gfc_cpp_option.prefix = NULL;
+  gfc_cpp_option.sysroot = NULL;
+
+  gfc_cpp_option.deferred_opt = XNEWVEC (gfc_cpp_deferred_opt_t, argc);
+  gfc_cpp_option.deferred_opt_count = 0;
+}
+
+int
+gfc_cpp_handle_option (size_t scode, const char *arg, int value ATTRIBUTE_UNUSED)
+{
+  int result = 1;
+  enum opt_code code = (enum opt_code) scode;
+
+  switch (code)
+  {
+    default:
+      result = 0;
+      break;
+
+    case OPT_cpp:
+      gfc_cpp_option.temporary_filename = arg;
+      break;
+
+    case OPT_nocpp:
+      gfc_cpp_option.temporary_filename = 0L;
+      break;
+
+    case OPT_d:
+      for ( ; *arg; ++arg)
+        switch (*arg)
+	{
+	  case 'D':
+	  case 'M':
+	  case 'N':
+	  case 'U':
+	    gfc_cpp_option.dump_macros = *arg;
+	    break;
+
+	  case 'I':
+	    gfc_cpp_option.dump_includes = 1;
+	    break;
+	}
+      break;
+
+    case OPT_fworking_directory:
+      gfc_cpp_option.working_directory = value;
+      break;
+
+    case OPT_imultilib:
+      gfc_cpp_option.multilib = arg;
+      break;
+
+    case OPT_iprefix:
+      gfc_cpp_option.prefix = arg;
+      break;
+
+    case OPT_isysroot:
+      gfc_cpp_option.sysroot = arg;
+      break;
+
+    case OPT_iquote:
+    case OPT_isystem:
+      gfc_cpp_add_include_path (xstrdup(arg), true);
+      break;
+
+    case OPT_nostdinc:
+      gfc_cpp_option.standard_include_paths = value;
+      break;
+
+    case OPT_o:
+      if (!gfc_cpp_option.output_filename)
+	gfc_cpp_option.output_filename = arg;
+      else
+	gfc_fatal_error ("output filename specified twice");
+      break;
+
+    case OPT_undef:
+      gfc_cpp_option.no_predefined = value;
+      break;
+
+    case OPT_v:
+      gfc_cpp_option.verbose = value;
+      break;
+
+    case OPT_A:
+    case OPT_D:
+    case OPT_U:
+      gfc_cpp_option.deferred_opt[gfc_cpp_option.deferred_opt_count].code = code;
+      gfc_cpp_option.deferred_opt[gfc_cpp_option.deferred_opt_count].arg = arg;
+      gfc_cpp_option.deferred_opt_count++;
+      break;
+
+    case OPT_C:
+      gfc_cpp_option.discard_comments = 0;
+      break;
+
+    case OPT_CC:
+      gfc_cpp_option.discard_comments = 0;
+      gfc_cpp_option.discard_comments_in_macro_exp = 0;
+      break;
+
+    case OPT_E:
+      gfc_cpp_option.preprocess_only = 1;
+      break;
+
+    case OPT_H:
+      gfc_cpp_option.print_include_names = 1;
+      break;
+
+    case OPT_P:
+      gfc_cpp_option.no_line_commands = 1;
+      break;
+  }
+
+  return result;
+}
+
+
+void
+gfc_cpp_post_options (void)
+{
+  /* Any preprocessing-related option without '-cpp' is considered
+     an error.  */
+  if (!gfc_cpp_enabled ()
+      && (gfc_cpp_preprocess_only ()
+          || !gfc_cpp_option.discard_comments
+          || !gfc_cpp_option.discard_comments_in_macro_exp
+          || gfc_cpp_option.print_include_names
+          || gfc_cpp_option.no_line_commands
+          || gfc_cpp_option.dump_macros
+          || gfc_cpp_option.dump_includes))
+    gfc_fatal_error("To enable preprocessing, use -cpp");
+
+  cpp_in = cpp_create_reader (CLK_GNUC89, NULL, line_table);
+  if (!gfc_cpp_enabled())
+    return;
+
+  gcc_assert (cpp_in);
+
+  /* The cpp_options-structure defines far more flags than those set here.
+     If any other is implemented, see c-opt.c (sanitize_cpp_opts) for
+     inter-option dependencies that may need to be enforced.  */
+  cpp_option = cpp_get_options (cpp_in);
+  gcc_assert (cpp_option);
+
+  /* TODO: allow non-traditional modes, e.g. by -cpp-std=...?  */
+  cpp_option->traditional = 1;
+  cpp_option->cplusplus_comments = 0;
+
+  cpp_option->pedantic = pedantic;
+  cpp_option->inhibit_warnings = inhibit_warnings;
+
+  cpp_option->dollars_in_ident = gfc_option.flag_dollar_ok;
+  cpp_option->discard_comments = gfc_cpp_option.discard_comments;
+  cpp_option->discard_comments_in_macro_exp = gfc_cpp_option.discard_comments_in_macro_exp;
+  cpp_option->print_include_names = gfc_cpp_option.print_include_names;
+  cpp_option->preprocessed = gfc_option.flag_preprocessed;
+
+  if (gfc_cpp_option.working_directory == -1)
+    gfc_cpp_option.working_directory = (debug_info_level != DINFO_LEVEL_NONE);
+
+  cpp_post_options (cpp_in);
+
+  /* If an error has occurred in cpplib, note it so we fail immediately.  */
+  errorcount += cpp_errors (cpp_in);
+
+  gfc_cpp_register_include_paths ();
+}
+
+
+void
+gfc_cpp_init_0 (void)
+{
+  struct cpp_callbacks *cb;
+
+  cb = cpp_get_callbacks (cpp_in);
+  cb->file_change = cb_file_change;
+  cb->line_change = cb_line_change;
+  cb->ident = cb_ident;
+  cb->def_pragma = cb_def_pragma;
+
+  if (gfc_cpp_option.dump_includes)
+    cb->include = cb_include;
+
+  if ((gfc_cpp_option.dump_macros == 'D')
+      || (gfc_cpp_option.dump_macros == 'N'))
+    {
+      cb->define = cb_define;
+      cb->undef  = cb_undef;
+    }
+
+  if (gfc_cpp_option.dump_macros == 'U')
+    {
+      cb->before_define = dump_queued_macros;
+      cb->used_define = cb_used_define;
+      cb->used_undef = cb_used_undef;
+    }
+
+  /* Initialize the print structure.  Setting print.src_line to -1 here is
+     a trick to guarantee that the first token of the file will cause
+     a linemarker to be output by maybe_print_line.  */
+  print.src_line = -1;
+  print.printed = 0;
+  print.prev = 0;
+  print.first_time = 1;
+
+  if (gfc_cpp_preprocess_only ())
+    {
+      if (gfc_cpp_option.output_filename)
+	{
+	  /* This needs cheating: with "-E -o <file>", the user wants the
+	     preprocessed output in <file>. However, if nothing is done
+	     about it <file> is also used for assembler output. Hence, it
+	     is necessary to redirect assembler output (actually nothing
+	     as -E implies -fsyntax-only) to another file, otherwise the
+	     output from preprocessing is lost.  */
+	  asm_file_name = gfc_cpp_option.temporary_filename;
+
+	  print.outf = fopen (gfc_cpp_option.output_filename, "w");
+	  if (print.outf == NULL)
+	    gfc_fatal_error ("opening output file %s: %s",
+			     gfc_cpp_option.output_filename, strerror(errno));
+	}
+      else
+	print.outf = stdout;
+    }
+  else
+    {
+      print.outf = fopen (gfc_cpp_option.temporary_filename, "w");
+      if (print.outf == NULL)
+	gfc_fatal_error ("opening output file %s: %s",
+			 gfc_cpp_option.temporary_filename, strerror(errno));
+    }
+
+  gcc_assert(cpp_in);
+  if (!cpp_read_main_file (cpp_in, gfc_source_file))
+    errorcount++;
+}
+
+void
+gfc_cpp_init (void)
+{
+  int i;
+
+  cpp_change_file (cpp_in, LC_RENAME, _("<built-in>"));
+  if (!gfc_cpp_option.no_predefined)
+    cpp_define_builtins (cpp_in);
+
+  /* Handle deferred options from command-line.  */
+  cpp_change_file (cpp_in, LC_RENAME, _("<command-line>"));
+
+  for (i = 0; i < gfc_cpp_option.deferred_opt_count; i++)
+    {
+      gfc_cpp_deferred_opt_t *opt = &gfc_cpp_option.deferred_opt[i];
+
+      if (opt->code == OPT_D)
+	cpp_define (cpp_in, opt->arg);
+      else if (opt->code == OPT_U)
+	cpp_undef (cpp_in, opt->arg);
+      else if (opt->code == OPT_A)
+	{
+	  if (opt->arg[0] == '-')
+	    cpp_unassert (cpp_in, opt->arg + 1);
+	  else
+	    cpp_assert (cpp_in, opt->arg);
+	}
+    }
+
+  if (gfc_cpp_option.working_directory
+      && gfc_cpp_option.preprocess_only && !gfc_cpp_option.no_line_commands)
+    pp_dir_change (cpp_in, get_src_pwd ());
+}
+
+try
+gfc_cpp_preprocess (const char *source_file)
+{
+  if (!gfc_cpp_enabled ())
+    return FAILURE;
+
+  cpp_change_file (cpp_in, LC_RENAME, source_file);
+
+  if (cpp_option->traditional)
+    scan_translation_unit_trad (cpp_in);
+  else
+    scan_translation_unit (cpp_in);
+
+  /* -dM command line option.  */
+  if (gfc_cpp_option.dump_macros == 'M')
+    {
+      putc ('\n', print.outf);
+      cpp_forall_identifiers (cpp_in, dump_macro, NULL);
+    }
+
+  if (!gfc_cpp_preprocess_only ()
+      || (gfc_cpp_preprocess_only () && gfc_cpp_option.output_filename))
+    fclose (print.outf);
+
+  return SUCCESS;
+}
+
+void
+gfc_cpp_done (void)
+{
+  if (!gfc_cpp_enabled ())
+    return;
+
+  /* TODO: if dependency tracking was enabled, call
+     cpp_finish() here to write dependencies.
+
+     Use cpp_get_deps() to access the current source's
+     dependencies during parsing. Add dependencies using
+     the mkdeps-interface (defined in libcpp).  */
+
+  gcc_assert (cpp_in);
+  cpp_undef_all (cpp_in);
+  cpp_clear_file_cache (cpp_in);
+}
+
+/* PATH must be malloc-ed and NULL-terminated.  */
+void
+gfc_cpp_add_include_path (char *path, bool user_supplied)
+{
+  /* CHAIN sets cpp_dir->sysp which differs from 0 if PATH is a system
+     include path. Fortran does not define any system include paths.  */
+  int chain = 0;
+  int cxx_aware = 0;
+
+  add_path (path, chain, cxx_aware, user_supplied);
+}
+
+void
+gfc_cpp_register_include_paths (void)
+{
+  int cxx_stdinc = 0;
+  register_include_chains (cpp_in, gfc_cpp_option.sysroot,
+			   gfc_cpp_option.prefix, gfc_cpp_option.multilib,
+			   gfc_cpp_option.standard_include_paths, cxx_stdinc,
+			   gfc_cpp_option.verbose);
+}
+
+
+
+static void scan_translation_unit_trad (cpp_reader *);
+static void account_for_newlines (const unsigned char *, size_t);
+static int dump_macro (cpp_reader *, cpp_hashnode *, void *);
+
+static void print_line (source_location, const char *);
+static void maybe_print_line (source_location);
+
+
+/* Writes out the preprocessed file, handling spacing and paste
+   avoidance issues.  */
+static void
+scan_translation_unit (cpp_reader *pfile)
+{
+  bool avoid_paste = false;
+
+  print.source = NULL;
+  for (;;)
+    {
+      const cpp_token *token = cpp_get_token (pfile);
+
+      if (token->type == CPP_PADDING)
+	{
+	  avoid_paste = true;
+	  if (print.source == NULL
+	      || (!(print.source->flags & PREV_WHITE)
+		  && token->val.source == NULL))
+	    print.source = token->val.source;
+	  continue;
+	}
+
+      if (token->type == CPP_EOF)
+	break;
+
+      /* Subtle logic to output a space if and only if necessary.  */
+      if (avoid_paste)
+	{
+	  if (print.source == NULL)
+	    print.source = token;
+	  if (print.source->flags & PREV_WHITE
+	      || (print.prev
+		  && cpp_avoid_paste (pfile, print.prev, token))
+	      || (print.prev == NULL && token->type == CPP_HASH))
+	    putc (' ', print.outf);
+	}
+      else if (token->flags & PREV_WHITE)
+	putc (' ', print.outf);
+
+      avoid_paste = false;
+      print.source = NULL;
+      print.prev = token;
+      cpp_output_token (token, print.outf);
+
+      if (token->type == CPP_COMMENT)
+	account_for_newlines (token->val.str.text, token->val.str.len);
+    }
+}
+
+/* Adjust print.src_line for newlines embedded in output.  */
+static void
+account_for_newlines (const unsigned char *str, size_t len)
+{
+  while (len--)
+    if (*str++ == '\n')
+      print.src_line++;
+}
+
+/* Writes out a traditionally preprocessed file.  */
+static void
+scan_translation_unit_trad (cpp_reader *pfile)
+{
+  while (_cpp_read_logical_line_trad (pfile))
+    {
+      size_t len = pfile->out.cur - pfile->out.base;
+      maybe_print_line (pfile->out.first_line);
+      fwrite (pfile->out.base, 1, len, print.outf);
+      print.printed = 1;
+      if (!CPP_OPTION (pfile, discard_comments))
+	account_for_newlines (pfile->out.base, len);
+    }
+}
+
+/* If the token read on logical line LINE needs to be output on a
+   different line to the current one, output the required newlines or
+   a line marker.  */
+static void
+maybe_print_line (source_location src_loc)
+{
+  const struct line_map *map = linemap_lookup (line_table, src_loc);
+  int src_line = SOURCE_LINE (map, src_loc);
+
+  /* End the previous line of text.  */
+  if (print.printed)
+    {
+      putc ('\n', print.outf);
+      print.src_line++;
+      print.printed = 0;
+    }
+
+  if (src_line >= print.src_line && src_line < print.src_line + 8)
+    {
+      while (src_line > print.src_line)
+	{
+	  putc ('\n', print.outf);
+	  print.src_line++;
+	}
+    }
+  else
+    print_line (src_loc, "");
+}
+
+/* Output a line marker for logical line LINE.  Special flags are "1"
+   or "2" indicating entering or leaving a file.  */
+static void
+print_line (source_location src_loc, const char *special_flags)
+{
+  /* End any previous line of text.  */
+  if (print.printed)
+    putc ('\n', print.outf);
+  print.printed = 0;
+
+  if (!gfc_cpp_option.no_line_commands)
+    {
+      const struct line_map *map = linemap_lookup (line_table, src_loc);
+
+      size_t to_file_len = strlen (map->to_file);
+      unsigned char *to_file_quoted =
+         (unsigned char *) alloca (to_file_len * 4 + 1);
+      unsigned char *p;
+
+      print.src_line = SOURCE_LINE (map, src_loc);
+
+      /* cpp_quote_string does not nul-terminate, so we have to do it
+	 ourselves.  */
+      p = cpp_quote_string (to_file_quoted,
+			    (const unsigned char *) map->to_file, to_file_len);
+      *p = '\0';
+      fprintf (print.outf, "# %u \"%s\"%s",
+	       print.src_line == 0 ? 1 : print.src_line,
+	       to_file_quoted, special_flags);
+
+      if (map->sysp == 2)
+	fputs (" 3 4", print.outf);
+      else if (map->sysp == 1)
+	fputs (" 3", print.outf);
+
+      putc ('\n', print.outf);
+    }
+}
+
+static void
+cb_file_change (cpp_reader * ARG_UNUSED (pfile), const struct line_map *map)
+{
+  const char *flags = "";
+
+  if (gfc_cpp_option.no_line_commands)
+    return;
+
+  if (!map)
+    return;
+
+      if (print.first_time)
+	{
+	  /* Avoid printing foo.i when the main file is foo.c.  */
+	  if (!cpp_get_options (cpp_in)->preprocessed)
+	    print_line (map->start_location, flags);
+	  print.first_time = 0;
+	}
+      else
+	{
+	  /* Bring current file to correct line when entering a new file.  */
+	  if (map->reason == LC_ENTER)
+	    {
+	      const struct line_map *from = INCLUDED_FROM (line_table, map);
+	      maybe_print_line (LAST_SOURCE_LINE_LOCATION (from));
+	    }
+	  if (map->reason == LC_ENTER)
+	    flags = " 1";
+	  else if (map->reason == LC_LEAVE)
+	    flags = " 2";
+	  print_line (map->start_location, flags);
+	}
+
+}
+
+/* Called when a line of output is started.  TOKEN is the first token
+   of the line, and at end of file will be CPP_EOF.  */
+static void
+cb_line_change (cpp_reader *pfile, const cpp_token *token,
+		int parsing_args)
+{
+  source_location src_loc = token->src_loc;
+
+  if (token->type == CPP_EOF || parsing_args)
+    return;
+
+  maybe_print_line (src_loc);
+  print.prev = 0;
+  print.source = 0;
+
+  /* Supply enough spaces to put this token in its original column,
+     one space per column greater than 2, since scan_translation_unit
+     will provide a space if PREV_WHITE.  Don't bother trying to
+     reconstruct tabs; we can't get it right in general, and nothing
+     ought to care.  Some things do care; the fault lies with them.  */
+  if (!CPP_OPTION (pfile, traditional))
+    {
+      const struct line_map *map = linemap_lookup (line_table, src_loc);
+      int spaces = SOURCE_COLUMN (map, src_loc) - 2;
+      print.printed = 1;
+
+      while (-- spaces >= 0)
+	putc (' ', print.outf);
+    }
+}
+
+static void
+cb_ident (cpp_reader *pfile ATTRIBUTE_UNUSED, source_location line,
+	  const cpp_string *str)
+{
+  maybe_print_line (line);
+  fprintf (print.outf, "#ident %s\n", str->text);
+  print.src_line++;
+}
+
+static void
+cb_define (cpp_reader *pfile ATTRIBUTE_UNUSED, source_location line,
+           cpp_hashnode *node ATTRIBUTE_UNUSED)
+{
+  maybe_print_line (line);
+  fputs ("#define ", print.outf);
+
+  /* 'D' is whole definition; 'N' is name only.  */
+  if (gfc_cpp_option.dump_macros == 'D')
+    fputs ((const char *) cpp_macro_definition (pfile, node),
+	   print.outf);
+  else
+    fputs ((const char *) NODE_NAME (node), print.outf);
+
+  putc ('\n', print.outf);
+  if (linemap_lookup (line_table, line)->to_line != 0)
+    print.src_line++;
+}
+
+static void
+cb_undef (cpp_reader *pfile ATTRIBUTE_UNUSED, source_location line,
+	  cpp_hashnode *node)
+{
+  maybe_print_line (line);
+  fprintf (print.outf, "#undef %s\n", NODE_NAME (node));
+  print.src_line++;
+}
+
+static void
+cb_include (cpp_reader *pfile ATTRIBUTE_UNUSED, source_location line,
+	    const unsigned char *dir, const char *header, int angle_brackets,
+	    const cpp_token **comments)
+{
+  maybe_print_line (line);
+  if (angle_brackets)
+    fprintf (print.outf, "#%s <%s>", dir, header);
+  else
+    fprintf (print.outf, "#%s \"%s\"", dir, header);
+
+  if (comments != NULL)
+    {
+      while (*comments != NULL)
+	{
+	  if ((*comments)->flags & PREV_WHITE)
+	    putc (' ', print.outf);
+	  cpp_output_token (*comments, print.outf);
+	  ++comments;
+	}
+    }
+
+  putc ('\n', print.outf);
+  print.src_line++;
+}
+
+/* Dump out the hash table.  */
+static int
+dump_macro (cpp_reader *pfile, cpp_hashnode *node, void *v ATTRIBUTE_UNUSED)
+{
+  if (node->type == NT_MACRO && !(node->flags & NODE_BUILTIN))
+    {
+      fputs ("#define ", print.outf);
+      fputs ((const char *) cpp_macro_definition (pfile, node),
+	     print.outf);
+      putc ('\n', print.outf);
+      print.src_line++;
+    }
+
+  return 1;
+}
+
+static void
+cb_used_define (cpp_reader *pfile, source_location line ATTRIBUTE_UNUSED,
+		cpp_hashnode *node)
+{
+  gfc_cpp_macro_queue *q;
+  q = XNEW (gfc_cpp_macro_queue);
+  q->macro = xstrdup ((const char *) cpp_macro_definition (pfile, node));
+  q->next = cpp_define_queue;
+  cpp_define_queue = q;
+}
+
+
+/* Callback called when -fworking-director and -E to emit working
+   directory in cpp output file.  */
+
+void
+pp_dir_change (cpp_reader *pfile ATTRIBUTE_UNUSED, const char *dir)
+{
+  size_t to_file_len = strlen (dir);
+  unsigned char *to_file_quoted =
+     (unsigned char *) alloca (to_file_len * 4 + 1);
+  unsigned char *p;
+
+  /* cpp_quote_string does not nul-terminate, so we have to do it ourselves.  */
+  p = cpp_quote_string (to_file_quoted, (const unsigned char *) dir, to_file_len);
+  *p = '\0';
+  fprintf (print.outf, "# 1 \"%s//\"\n", to_file_quoted);
+}
+
+/* Copy a #pragma directive to the preprocessed output.  */
+static void
+cb_def_pragma (cpp_reader *pfile, source_location line)
+{
+  maybe_print_line (line);
+  fputs ("#pragma ", print.outf);
+  cpp_output_line (pfile, print.outf);
+  print.src_line++;
+}
+
+static void
+cb_used_undef (cpp_reader *pfile ATTRIBUTE_UNUSED,
+	       source_location line ATTRIBUTE_UNUSED,
+	       cpp_hashnode *node)
+{
+  gfc_cpp_macro_queue *q;
+  q = XNEW (gfc_cpp_macro_queue);
+  q->macro = xstrdup ((const char *) NODE_NAME (node));
+  q->next = cpp_undefine_queue;
+  cpp_undefine_queue = q;
+}
+
+static void
+dump_queued_macros (cpp_reader *pfile ATTRIBUTE_UNUSED)
+{
+  gfc_cpp_macro_queue *q;
+
+  /* End the previous line of text.  */
+  if (print.printed)
+    {
+      putc ('\n', print.outf);
+      print.src_line++;
+      print.printed = 0;
+    }
+
+  for (q = cpp_define_queue; q;)
+    {
+      gfc_cpp_macro_queue *oq;
+      fputs ("#define ", print.outf);
+      fputs (q->macro, print.outf);
+      putc ('\n', print.outf);
+      print.src_line++;
+      oq = q;
+      q = q->next;
+      gfc_free (oq->macro);
+      gfc_free (oq);
+    }
+  cpp_define_queue = NULL;
+  for (q = cpp_undefine_queue; q;)
+    {
+      gfc_cpp_macro_queue *oq;
+      fprintf (print.outf, "#undef %s\n", q->macro);
+      print.src_line++;
+      oq = q;
+      q = q->next;
+      gfc_free (oq->macro);
+      gfc_free (oq);
+    }
+  cpp_undefine_queue = NULL;
+}
+
+
Index: f95-lang.c
===================================================================
--- f95-lang.c	(revision 135676)
+++ f95-lang.c	(working copy)
@@ -45,6 +45,7 @@ along with GCC; see the file COPYING3.  
 #include "cgraph.h"
 
 #include "gfortran.h"
+#include "cpp.h"
 #include "trans.h"
 #include "trans-types.h"
 #include "trans-const.h"
@@ -275,18 +276,25 @@ gfc_be_parse_file (int set_yydebug ATTRI
 static bool
 gfc_init (void)
 {
-  linemap_add (line_table, LC_ENTER, false, gfc_source_file, 1);
-  linemap_add (line_table, LC_RENAME, false, "<built-in>", 0);
+  if (!gfc_cpp_enabled ())
+    {
+      linemap_add (line_table, LC_ENTER, false, gfc_source_file, 1);
+      linemap_add (line_table, LC_RENAME, false, "<built-in>", 0);
+    }
+  else
+    gfc_cpp_init_0 ();
 
-  /* First initialize the backend.  */
   gfc_init_decl_processing ();
   gfc_static_ctors = NULL_TREE;
 
-  /* Then the frontend.  */
+  if (gfc_cpp_enabled ())
+    gfc_cpp_init ();
+
   gfc_init_1 ();
 
   if (gfc_new_file () != SUCCESS)
     fatal_error ("can't open input file: %s", gfc_source_file);
+
   return true;
 }
 
@@ -294,6 +302,7 @@ gfc_init (void)
 static void
 gfc_finish (void)
 {
+  gfc_cpp_done ();
   gfc_done_1 ();
   gfc_release_include_path ();
   return;
Index: options.c
===================================================================
--- options.c	(revision 135676)
+++ options.c	(working copy)
@@ -31,6 +31,7 @@ along with GCC; see the file COPYING3.  
 #include "tree-inline.h"
 #include "gfortran.h"
 #include "target.h"
+#include "cpp.h"
 
 gfc_option_t gfc_option;
 
@@ -50,8 +51,7 @@ set_default_std_flags (void)
 /* Get ready for options handling.  */
 
 unsigned int
-gfc_init_options (unsigned int argc ATTRIBUTE_UNUSED,
-		  const char **argv ATTRIBUTE_UNUSED)
+gfc_init_options (unsigned int argc, const char **argv)
 {
   gfc_source_file = NULL;
   gfc_option.module_dir = NULL;
@@ -128,6 +128,9 @@ gfc_init_options (unsigned int argc ATTR
   /* -fshort-enums can be default on some targets.  */
   gfc_option.fshort_enums = targetm.default_short_enums ();
 
+  /* Initialize cpp-related options.  */
+  gfc_cpp_init_options(argc, argv);
+
   return CL_Fortran;
 }
 
@@ -354,6 +357,15 @@ gfc_post_options (const char **pfilename
   if (gfc_option.flag_all_intrinsics)
     gfc_option.warn_nonstd_intrinsics = 0;
 
+  gfc_cpp_post_options ();
+
+/* FIXME: return gfc_cpp_preprocess_only ();
+
+   The return value of this function indicates whether the
+   backend needs to be initialized. On -E, we don't need
+   the backend. However, if we return 'true' here, an
+   ICE occurs. Initializing the backend doesn't hurt much,
+   hence, for now we can live with it as is.  */
   return false;
 }
 
@@ -452,6 +464,9 @@ gfc_handle_option (size_t scode, const c
   if (code == N_OPTS)
     return 1;
 
+  if (gfc_cpp_handle_option (scode, arg, value) == 1)
+    return 1;
+
   switch (code)
     {
     default:
@@ -692,7 +707,7 @@ gfc_handle_option (size_t scode, const c
       else
 	gfc_fatal_error ("Unrecognized option to -finit-real: %s",
 			 arg);
-      break;      
+      break;
 
     case OPT_finit_integer_:
       gfc_option.flag_init_integer = GFC_INIT_INTEGER_ON;
Index: scanner.c
===================================================================
--- scanner.c	(revision 135676)
+++ scanner.c	(working copy)
@@ -47,6 +47,7 @@ along with GCC; see the file COPYING3.  
 #include "toplev.h"
 #include "debug.h"
 #include "flags.h"
+#include "cpp.h"
 
 /* Structure for holding module and include file search path.  */
 typedef struct gfc_directorylist
@@ -340,6 +341,7 @@ void
 gfc_add_include_path (const char *path, bool use_for_modules)
 {
   add_path_to_list (&include_dirs, path, use_for_modules);
+  gfc_cpp_add_include_path (xstrdup(path), true);
 }
 
 
@@ -1909,7 +1911,14 @@ gfc_new_file (void)
 {
   try result;
 
-  result = load_file (gfc_source_file, true);
+  if (gfc_cpp_enabled ())
+    {
+      result = gfc_cpp_preprocess (gfc_source_file);
+      if (!gfc_cpp_preprocess_only ())
+        result = load_file (gfc_cpp_temporary_file (), true);
+    }
+  else
+    result = load_file (gfc_source_file, true);
 
   gfc_current_locus.lb = line_head;
   gfc_current_locus.nextc = (line_head == NULL) ? NULL : line_head->line;
Index: cpp.h
===================================================================
--- cpp.h	(revision 0)
+++ cpp.h	(revision 0)
@@ -0,0 +1,29 @@
+#ifndef GFC_CPP_H
+#define GFC_CPP_H
+
+/* Returns true if preprocessing is enabled, false otherwise.  */
+bool gfc_cpp_enabled (void);
+
+bool gfc_cpp_preprocess_only (void);
+
+const char *gfc_cpp_temporary_file (void);
+
+
+void gfc_cpp_init_0 (void);
+void gfc_cpp_init (void);
+
+void gfc_cpp_init_options (unsigned int argc, const char **argv);
+
+int gfc_cpp_handle_option(size_t scode, const char *arg, int value);
+
+void gfc_cpp_post_options (void);
+
+try gfc_cpp_preprocess (const char *source_file);
+
+void gfc_cpp_done (void);
+
+void gfc_cpp_add_include_path (char *path, bool user_supplied);
+
+void gfc_cpp_register_include_paths (void);
+
+#endif /* GFC_CPP_H */
Index: lang-specs.h
===================================================================
--- lang-specs.h	(revision 135676)
+++ lang-specs.h	(working copy)
@@ -20,35 +20,45 @@
 /* This is the contribution to the `default_compilers' array in gcc.c
    for the f95 language.  */
 
+
+#define F951_CPP_OPTIONS "%{!nocpp: -cpp %g.f90 %(cpp_options)\
+                          %{E|M|MM:%(cpp_debug_options) -fsyntax-only} %{E}}"
+#define F951_OPTIONS     "%(cc1_options) %{J*} %{I*}\
+                          %{!nostdinc:-fintrinsic-modules-path finclude%s}\
+                          %{!fsyntax-only:%(invoke_as)}"
+#define F951_SOURCE_FORM  "%{!ffree-form:-ffixed-form}"
+
+
 {".F",   "@f77-cpp-input", 0, 0, 0},
 {".FOR", "@f77-cpp-input", 0, 0, 0},
 {".FTN", "@f77-cpp-input", 0, 0, 0},
 {".fpp", "@f77-cpp-input", 0, 0, 0},
 {".FPP", "@f77-cpp-input", 0, 0, 0},
 {"@f77-cpp-input",
-  "cc1 -E -lang-fortran -traditional-cpp -D_LANGUAGE_FORTRAN %(cpp_options) \
-      %{E|M|MM:%(cpp_debug_options)}\
-      %{!M:%{!MM:%{!E: -o %|.f |\n\
-    f951 %|.f %{!ffree-form:-ffixed-form} %(cc1_options) %{J*} %{I*}\
-      -fpreprocessed %{!nostdinc:-fintrinsic-modules-path finclude%s} %{!fsyntax-only:%(invoke_as)}}}}", 0, 0, 0},
+    "f951 %i " F951_SOURCE_FORM " " \
+               F951_CPP_OPTIONS " %{!E:" F951_OPTIONS "}", 0, 0, 0},
+{".f",   "@f77", 0, 0, 0},
+{".for", "@f77", 0, 0, 0},
+{".ftn", "@f77", 0, 0, 0},
+{"@f77",
+    "f951 %i " F951_SOURCE_FORM " \
+          %{E:%{!cpp:%egfortran does not support -E without -cpp}} \
+          %{cpp:" F951_CPP_OPTIONS "} %{!E:" F951_OPTIONS "}", 0, 0, 0},
 {".F90", "@f95-cpp-input", 0, 0, 0},
 {".F95", "@f95-cpp-input", 0, 0, 0},
 {".F03", "@f95-cpp-input", 0, 0, 0},
 {".F08", "@f95-cpp-input", 0, 0, 0},
 {"@f95-cpp-input",
-  "cc1 -E -lang-fortran -traditional-cpp -D_LANGUAGE_FORTRAN %(cpp_options) \
-      %{E|M|MM:%(cpp_debug_options)}\
-      %{!M:%{!MM:%{!E: -o %|.f95 |\n\
-    f951 %|.f95 %{!ffixed-form:-ffree-form} %(cc1_options) %{J*} %{I*}\
-      -fpreprocessed %{!nostdinc:-fintrinsic-modules-path finclude%s} %{!fsyntax-only:%(invoke_as)}}}}", 0, 0, 0},
+    "f951 %i " F951_CPP_OPTIONS " %{!E:" F951_OPTIONS "}", 0, 0, 0},
 {".f90", "@f95", 0, 0, 0},
 {".f95", "@f95", 0, 0, 0},
 {".f03", "@f95", 0, 0, 0},
 {".f08", "@f95", 0, 0, 0},
-{"@f95", "%{!E:f951 %i %(cc1_options) %{J*} %{I*}\
-         %{!nostdinc:-fintrinsic-modules-path finclude%s} %{!fsyntax-only:%(invoke_as)}}", 0, 0, 0},
-{".f",   "@f77", 0, 0, 0},
-{".for", "@f77", 0, 0, 0},
-{".ftn", "@f77", 0, 0, 0},
-{"@f77", "%{!E:f951 %i %{!ffree-form:-ffixed-form} %(cc1_options) %{J*} %{I*}\
-         %{!nostdinc:-fintrinsic-modules-path finclude%s} %{!fsyntax-only:%(invoke_as)}}", 0, 0, 0},
+{"@f95", 
+    "f951 %i %{E:%{!cpp:%egfortran does not support -E without -cpp}}\
+          %{cpp:" F951_CPP_OPTIONS "} %{!E:" F951_OPTIONS "}", 0, 0, 0},
+
+
+#undef F951_SOURCE_FORM
+#undef F951_CPP_OPTIONS
+#undef F951_OPTIONS