X-Git-Url: http://git.sourceforge.jp/view?a=blobdiff_plain;f=gcc%2Fc-lex.c;h=23abaad6c10816d987be47e0bab114ebeb037936;hb=3e04ac919d4b400683517fc21277e3a7e1084af0;hp=ec16c15397f2c7a7ed15a656741c96fcb89bf5e3;hpb=2cbf1359092416d271b02844198e7ad6e5e56284;p=pf3gnuchains%2Fgcc-fork.git diff --git a/gcc/c-lex.c b/gcc/c-lex.c index ec16c15397f..23abaad6c10 100644 --- a/gcc/c-lex.c +++ b/gcc/c-lex.c @@ -1,12 +1,13 @@ /* Mainly the interface between cpplib and the C front ends. Copyright (C) 1987, 1988, 1989, 1992, 1994, 1995, 1996, 1997 - 1998, 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc. + 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2007 + Free Software Foundation, Inc. This file is part of GCC. GCC is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free -Software Foundation; either version 2, or (at your option) any later +Software Foundation; either version 3, or (at your option) any later version. GCC is distributed in the hope that it will be useful, but WITHOUT ANY @@ -15,9 +16,8 @@ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License -along with GCC; see the file COPYING. If not, write to the Free -Software Foundation, 59 Temple Place - Suite 330, Boston, MA -02111-1307, USA. */ +along with GCC; see the file COPYING3. If not see +. */ #include "config.h" #include "system.h" @@ -27,7 +27,6 @@ Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include "real.h" #include "rtl.h" #include "tree.h" -#include "expr.h" #include "input.h" #include "output.h" #include "c-tree.h" @@ -41,67 +40,50 @@ Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include "tm_p.h" #include "splay-tree.h" #include "debug.h" - -#ifdef MULTIBYTE_CHARS -#include "mbchar.h" -#include -#endif /* MULTIBYTE_CHARS */ - -/* The current line map. */ -static const struct line_map *map; - -/* The line used to refresh the lineno global variable after each token. */ -static unsigned int src_lineno; +#include "target.h" /* We may keep statistics about how long which files took to compile. */ static int header_time, body_time; static splay_tree file_info_tree; -/* File used for outputting assembler code. */ -extern FILE *asm_out_file; - -#undef WCHAR_TYPE_SIZE -#define WCHAR_TYPE_SIZE TYPE_PRECISION (wchar_type_node) - -/* Number of bytes in a wide character. */ -#define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT) - int pending_lang_change; /* If we need to switch languages - C++ only */ int c_header_level; /* depth in C headers - C++ only */ -/* Nonzero tells yylex to ignore \ in string constants. */ -static int ignore_escape_flag; - -static tree interpret_integer PARAMS ((const cpp_token *, unsigned int)); -static tree interpret_float PARAMS ((const cpp_token *, unsigned int)); -static enum integer_type_kind - narrowest_unsigned_type PARAMS ((tree, unsigned int)); -static enum integer_type_kind - narrowest_signed_type PARAMS ((tree, unsigned int)); -static tree lex_string PARAMS ((const unsigned char *, unsigned int, - int)); -static tree lex_charconst PARAMS ((const cpp_token *)); -static void update_header_times PARAMS ((const char *)); -static int dump_one_header PARAMS ((splay_tree_node, void *)); -static void cb_line_change PARAMS ((cpp_reader *, const cpp_token *, int)); -static void cb_ident PARAMS ((cpp_reader *, unsigned int, - const cpp_string *)); -static void cb_def_pragma PARAMS ((cpp_reader *, unsigned int)); -static void cb_define PARAMS ((cpp_reader *, unsigned int, - cpp_hashnode *)); -static void cb_undef PARAMS ((cpp_reader *, unsigned int, - cpp_hashnode *)); +/* If we need to translate characters received. This is tri-state: + 0 means use only the untranslated string; 1 means use only + the translated string; -1 means chain the translated string + to the untranslated one. */ +int c_lex_string_translate = 1; + +/* True if strings should be passed to the caller of c_lex completely + unmolested (no concatenation, no translation). */ +bool c_lex_return_raw_strings = false; + +static tree interpret_integer (const cpp_token *, unsigned int); +static tree interpret_float (const cpp_token *, unsigned int); +static tree interpret_fixed (const cpp_token *, unsigned int); +static enum integer_type_kind narrowest_unsigned_type + (unsigned HOST_WIDE_INT, unsigned HOST_WIDE_INT, unsigned int); +static enum integer_type_kind narrowest_signed_type + (unsigned HOST_WIDE_INT, unsigned HOST_WIDE_INT, unsigned int); +static enum cpp_ttype lex_string (const cpp_token *, tree *, bool); +static tree lex_charconst (const cpp_token *); +static void update_header_times (const char *); +static int dump_one_header (splay_tree_node, void *); +static void cb_line_change (cpp_reader *, const cpp_token *, int); +static void cb_ident (cpp_reader *, unsigned int, const cpp_string *); +static void cb_def_pragma (cpp_reader *, unsigned int); +static void cb_define (cpp_reader *, unsigned int, cpp_hashnode *); +static void cb_undef (cpp_reader *, unsigned int, cpp_hashnode *); void -init_c_lex () +init_c_lex (void) { struct cpp_callbacks *cb; struct c_fileinfo *toplevel; - /* Set up filename timing. Must happen before cpp_read_main_file. */ - file_info_tree = splay_tree_new ((splay_tree_compare_fn)strcmp, - 0, - (splay_tree_delete_value_fn)free); + /* The get_fileinfo data structure must be initialized before + cpp_read_main_file is called. */ toplevel = get_fileinfo (""); if (flag_detailed_statistics) { @@ -109,7 +91,7 @@ init_c_lex () body_time = get_run_time (); toplevel->time = body_time; } - + cb = cpp_get_callbacks (parse_in); cb->line_change = cb_line_change; @@ -120,8 +102,8 @@ init_c_lex () /* Set the debug callbacks if we can use them. */ if (debug_info_level == DINFO_LEVEL_VERBOSE - && (write_symbols == DWARF_DEBUG || write_symbols == DWARF2_DEBUG - || write_symbols == VMS_AND_DWARF2_DEBUG)) + && (write_symbols == DWARF2_DEBUG + || write_symbols == VMS_AND_DWARF2_DEBUG)) { cb->define = cb_define; cb->undef = cb_undef; @@ -129,17 +111,21 @@ init_c_lex () } struct c_fileinfo * -get_fileinfo (name) - const char *name; +get_fileinfo (const char *name) { splay_tree_node n; struct c_fileinfo *fi; + if (!file_info_tree) + file_info_tree = splay_tree_new ((splay_tree_compare_fn) strcmp, + 0, + (splay_tree_delete_value_fn) free); + n = splay_tree_lookup (file_info_tree, (splay_tree_key) name); if (n) return (struct c_fileinfo *) n->value; - fi = (struct c_fileinfo *) xmalloc (sizeof (struct c_fileinfo)); + fi = XNEW (struct c_fileinfo); fi->time = 0; fi->interface_only = 0; fi->interface_unknown = 1; @@ -149,8 +135,7 @@ get_fileinfo (name) } static void -update_header_times (name) - const char *name; +update_header_times (const char *name) { /* Changing files again. This means currently collected time is charged against header time, and body time starts back at 0. */ @@ -165,9 +150,7 @@ update_header_times (name) } static int -dump_one_header (n, dummy) - splay_tree_node n; - void *dummy ATTRIBUTE_UNUSED; +dump_one_header (splay_tree_node n, void * ARG_UNUSED (dummy)) { print_time ((const char *) n->key, ((struct c_fileinfo *) n->value)->time); @@ -175,7 +158,7 @@ dump_one_header (n, dummy) } void -dump_time_statistics () +dump_time_statistics (void) { struct c_fileinfo *file = get_fileinfo (input_filename); int this_time = get_run_time (); @@ -185,24 +168,27 @@ dump_time_statistics () print_time ("header files (total)", header_time); print_time ("main file (total)", this_time - body_time); fprintf (stderr, "ratio = %g : 1\n", - (double)header_time / (double)(this_time - body_time)); + (double) header_time / (double) (this_time - body_time)); fprintf (stderr, "\n******\n"); splay_tree_foreach (file_info_tree, dump_one_header, 0); } static void -cb_ident (pfile, line, str) - cpp_reader *pfile ATTRIBUTE_UNUSED; - unsigned int line ATTRIBUTE_UNUSED; - const cpp_string *str ATTRIBUTE_UNUSED; +cb_ident (cpp_reader * ARG_UNUSED (pfile), + unsigned int ARG_UNUSED (line), + const cpp_string * ARG_UNUSED (str)) { #ifdef ASM_OUTPUT_IDENT - if (! flag_no_ident) + if (!flag_no_ident) { /* Convert escapes in the string. */ - tree value ATTRIBUTE_UNUSED = lex_string (str->text, str->len, 0); - ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (value)); + cpp_string cstr = { 0, 0 }; + if (cpp_interpret_string (pfile, str, 1, &cstr, false)) + { + ASM_OUTPUT_IDENT (asm_out_file, (const char *) cstr.text); + free (CONST_CAST (unsigned char *, cstr.text)); + } } #endif } @@ -210,32 +196,44 @@ cb_ident (pfile, line, str) /* Called at the start of every non-empty line. TOKEN is the first lexed token on the line. Used for diagnostic line numbers. */ static void -cb_line_change (pfile, token, parsing_args) - cpp_reader *pfile ATTRIBUTE_UNUSED; - const cpp_token *token; - int parsing_args ATTRIBUTE_UNUSED; +cb_line_change (cpp_reader * ARG_UNUSED (pfile), const cpp_token *token, + int parsing_args) { - src_lineno = SOURCE_LINE (map, token->line); + if (token->type != CPP_EOF && !parsing_args) +#ifdef USE_MAPPED_LOCATION + input_location = token->src_loc; +#else + { + source_location loc = token->src_loc; + const struct line_map *map = linemap_lookup (line_table, loc); + input_line = SOURCE_LINE (map, loc); + } +#endif } void -fe_file_change (new_map) - const struct line_map *new_map; +fe_file_change (const struct line_map *new_map) { - unsigned int to_line = SOURCE_LINE (new_map, new_map->to_line); + if (new_map == NULL) + return; if (new_map->reason == LC_ENTER) { /* Don't stack the main buffer on the input stack; we already did in compile_file. */ - if (map == NULL) - main_input_filename = new_map->to_file; - else + if (!MAIN_FILE_P (new_map)) { - int included_at = SOURCE_LINE (new_map - 1, new_map->from_line - 1); +#ifdef USE_MAPPED_LOCATION + int included_at = LAST_SOURCE_LINE_LOCATION (new_map - 1); - lineno = included_at; + input_location = included_at; + push_srcloc (new_map->start_location); +#else + int included_at = LAST_SOURCE_LINE (new_map - 1); + + input_line = included_at; push_srcloc (new_map->to_file, 1); +#endif (*debug_hooks->start_source_file) (included_at, new_map->to_file); #ifndef NO_IMPLICIT_EXTERN_C if (c_header_level) @@ -254,29 +252,27 @@ fe_file_change (new_map) if (c_header_level && --c_header_level == 0) { if (new_map->sysp == 2) - warning ("badly nested C headers from preprocessor"); + warning (0, "badly nested C headers from preprocessor"); --pending_lang_change; } #endif pop_srcloc (); - - (*debug_hooks->end_source_file) (to_line); + + (*debug_hooks->end_source_file) (new_map->to_line); } update_header_times (new_map->to_file); in_system_header = new_map->sysp != 0; +#ifdef USE_MAPPED_LOCATION + input_location = new_map->start_location; +#else input_filename = new_map->to_file; - lineno = to_line; - map = new_map; - - /* Hook for C++. */ - extract_interface_info (); + input_line = new_map->to_line; +#endif } static void -cb_def_pragma (pfile, line) - cpp_reader *pfile; - unsigned int line; +cb_def_pragma (cpp_reader *pfile, source_location loc) { /* Issue a warning message if we have been asked to do so. Ignore unknown pragmas in system headers unless an explicit @@ -285,6 +281,14 @@ cb_def_pragma (pfile, line) { const unsigned char *space, *name; const cpp_token *s; +#ifndef USE_MAPPED_LOCATION + location_t fe_loc; + const struct line_map *map = linemap_lookup (line_table, loc); + fe_loc.file = map->to_file; + fe_loc.line = SOURCE_LINE (map, loc); +#else + location_t fe_loc = loc; +#endif space = name = (const unsigned char *) ""; s = cpp_get_token (pfile); @@ -296,62 +300,58 @@ cb_def_pragma (pfile, line) name = cpp_token_as_text (pfile, s); } - lineno = SOURCE_LINE (map, line); - warning ("ignoring #pragma %s %s", space, name); + warning (OPT_Wunknown_pragmas, "%Hignoring #pragma %s %s", + &fe_loc, space, name); } } /* #define callback for DWARF and DWARF2 debug info. */ static void -cb_define (pfile, line, node) - cpp_reader *pfile; - unsigned int line; - cpp_hashnode *node; +cb_define (cpp_reader *pfile, source_location loc, cpp_hashnode *node) { - (*debug_hooks->define) (SOURCE_LINE (map, line), + const struct line_map *map = linemap_lookup (line_table, loc); + (*debug_hooks->define) (SOURCE_LINE (map, loc), (const char *) cpp_macro_definition (pfile, node)); } /* #undef callback for DWARF and DWARF2 debug info. */ static void -cb_undef (pfile, line, node) - cpp_reader *pfile ATTRIBUTE_UNUSED; - unsigned int line; - cpp_hashnode *node; +cb_undef (cpp_reader * ARG_UNUSED (pfile), source_location loc, + cpp_hashnode *node) { - (*debug_hooks->undef) (SOURCE_LINE (map, line), + const struct line_map *map = linemap_lookup (line_table, loc); + (*debug_hooks->undef) (SOURCE_LINE (map, loc), (const char *) NODE_NAME (node)); } -int -c_lex (value) - tree *value; +/* Read a token and return its type. Fill *VALUE with its value, if + applicable. Fill *CPP_FLAGS with the token's flags, if it is + non-NULL. */ + +enum cpp_ttype +c_lex_with_flags (tree *value, location_t *loc, unsigned char *cpp_flags) { + static bool no_more_pch; const cpp_token *tok; + enum cpp_ttype type; + unsigned char add_flags = 0; - retry: timevar_push (TV_CPP); - do - tok = cpp_get_token (parse_in); - while (tok->type == CPP_PADDING); - timevar_pop (TV_CPP); - - /* The C++ front end does horrible things with the current line - number. To ensure an accurate line number, we must reset it - every time we return a token. */ - lineno = src_lineno; + retry: +#ifdef USE_MAPPED_LOCATION + tok = cpp_get_token_with_location (parse_in, loc); +#else + tok = cpp_get_token (parse_in); + *loc = input_location; +#endif + type = tok->type; - *value = NULL_TREE; - switch (tok->type) + retry_after_at: + switch (type) { - /* Issue this error here, where we can get at tok->val.c. */ - case CPP_OTHER: - if (ISGRAPH (tok->val.c)) - error ("stray '%c' in program", tok->val.c); - else - error ("stray '\\%o' in program", tok->val.c); + case CPP_PADDING: goto retry; - + case CPP_NAME: *value = HT_IDENT_TO_GCC_IDENT (HT_NODE (tok->val.node)); break; @@ -365,9 +365,14 @@ c_lex (value) case CPP_N_INVALID: /* cpplib has issued an error. */ *value = error_mark_node; + errorcount++; break; case CPP_N_INTEGER: + /* C++ uses '0' to mark virtual functions as pure. + Set PURE_ZERO to pass this information to the C++ parser. */ + if (tok->val.str.len == 1 && *tok->val.str.text == '0') + add_flags = PURE_ZERO; *value = interpret_integer (tok, flags); break; @@ -376,11 +381,85 @@ c_lex (value) break; default: - abort (); + gcc_unreachable (); } } break; + case CPP_ATSIGN: + /* An @ may give the next token special significance in Objective-C. */ + if (c_dialect_objc ()) + { +#ifdef USE_MAPPED_LOCATION + location_t atloc = *loc; + location_t newloc; +#else + location_t atloc = input_location; +#endif + + retry_at: +#ifdef USE_MAPPED_LOCATION + tok = cpp_get_token_with_location (parse_in, &newloc); +#else + tok = cpp_get_token (parse_in); +#endif + type = tok->type; + switch (type) + { + case CPP_PADDING: + goto retry_at; + + case CPP_STRING: + case CPP_WSTRING: + type = lex_string (tok, value, true); + break; + + case CPP_NAME: + *value = HT_IDENT_TO_GCC_IDENT (HT_NODE (tok->val.node)); + if (objc_is_reserved_word (*value)) + { + type = CPP_AT_NAME; + break; + } + /* FALLTHROUGH */ + + default: + /* ... or not. */ + error ("%Hstray %<@%> in program", &atloc); +#ifdef USE_MAPPED_LOCATION + *loc = newloc; +#endif + goto retry_after_at; + } + break; + } + + /* FALLTHROUGH */ + case CPP_HASH: + case CPP_PASTE: + { + unsigned char name[4]; + + *cpp_spell_token (parse_in, tok, name, true) = 0; + + error ("stray %qs in program", name); + } + + goto retry; + + case CPP_OTHER: + { + cppchar_t c = tok->val.str.text[0]; + + if (c == '"' || c == '\'') + error ("missing terminating %c character", (int) c); + else if (ISGRAPH (c)) + error ("stray %qc in program", (int) c); + else + error ("stray %<\\%o%> in program", (int) c); + } + goto retry; + case CPP_CHAR: case CPP_WCHAR: *value = lex_charconst (tok); @@ -388,29 +467,51 @@ c_lex (value) case CPP_STRING: case CPP_WSTRING: - *value = lex_string (tok->val.str.text, tok->val.str.len, - tok->type == CPP_WSTRING); + if (!c_lex_return_raw_strings) + { + type = lex_string (tok, value, false); + break; + } + *value = build_string (tok->val.str.len, (const char *) tok->val.str.text); + break; + + case CPP_PRAGMA: + *value = build_int_cst (NULL, tok->val.pragma); break; /* These tokens should not be visible outside cpplib. */ case CPP_HEADER_NAME: case CPP_COMMENT: case CPP_MACRO_ARG: - abort (); + gcc_unreachable (); - default: break; + default: + *value = NULL_TREE; + break; } - return tok->type; + if (cpp_flags) + *cpp_flags = tok->flags | add_flags; + + if (!no_more_pch) + { + no_more_pch = true; + c_common_no_more_pch (); + } + + timevar_pop (TV_CPP); + + return type; } /* Returns the narrowest C-visible unsigned type, starting with the - minimum specified by FLAGS, that can fit VALUE, or itk_none if + minimum specified by FLAGS, that can fit HIGH:LOW, or itk_none if there isn't one. */ + static enum integer_type_kind -narrowest_unsigned_type (value, flags) - tree value; - unsigned int flags; +narrowest_unsigned_type (unsigned HOST_WIDE_INT low, + unsigned HOST_WIDE_INT high, + unsigned int flags) { enum integer_type_kind itk; @@ -421,22 +522,23 @@ narrowest_unsigned_type (value, flags) else itk = itk_unsigned_long_long; - /* int_fits_type_p must think the type of its first argument is - wider than its second argument, or it won't do the proper check. */ - TREE_TYPE (value) = widest_unsigned_literal_type_node; - for (; itk < itk_none; itk += 2 /* skip unsigned types */) - if (int_fits_type_p (value, integer_types[itk])) - return itk; + { + tree upper = TYPE_MAX_VALUE (integer_types[itk]); + + if ((unsigned HOST_WIDE_INT) TREE_INT_CST_HIGH (upper) > high + || ((unsigned HOST_WIDE_INT) TREE_INT_CST_HIGH (upper) == high + && TREE_INT_CST_LOW (upper) >= low)) + return itk; + } return itk_none; } /* Ditto, but narrowest signed type. */ static enum integer_type_kind -narrowest_signed_type (value, flags) - tree value; - unsigned int flags; +narrowest_signed_type (unsigned HOST_WIDE_INT low, + unsigned HOST_WIDE_INT high, unsigned int flags) { enum integer_type_kind itk; @@ -447,22 +549,23 @@ narrowest_signed_type (value, flags) else itk = itk_long_long; - /* int_fits_type_p must think the type of its first argument is - wider than its second argument, or it won't do the proper check. */ - TREE_TYPE (value) = widest_unsigned_literal_type_node; for (; itk < itk_none; itk += 2 /* skip signed types */) - if (int_fits_type_p (value, integer_types[itk])) - return itk; + { + tree upper = TYPE_MAX_VALUE (integer_types[itk]); + + if ((unsigned HOST_WIDE_INT) TREE_INT_CST_HIGH (upper) > high + || ((unsigned HOST_WIDE_INT) TREE_INT_CST_HIGH (upper) == high + && TREE_INT_CST_LOW (upper) >= low)) + return itk; + } return itk_none; } /* Interpret TOKEN, an integer with FLAGS as classified by cpplib. */ static tree -interpret_integer (token, flags) - const cpp_token *token; - unsigned int flags; +interpret_integer (const cpp_token *token, unsigned int flags) { tree value, type; enum integer_type_kind itk; @@ -471,18 +574,19 @@ interpret_integer (token, flags) integer = cpp_interpret_integer (parse_in, token, flags); integer = cpp_num_sign_extend (integer, options->precision); - value = build_int_2_wide (integer.low, integer.high); /* The type of a constant with a U suffix is straightforward. */ if (flags & CPP_N_UNSIGNED) - itk = narrowest_unsigned_type (value, flags); + itk = narrowest_unsigned_type (integer.low, integer.high, flags); else { /* The type of a potentially-signed integer constant varies depending on the base it's in, the standard in use, and the length suffixes. */ - enum integer_type_kind itk_u = narrowest_unsigned_type (value, flags); - enum integer_type_kind itk_s = narrowest_signed_type (value, flags); + enum integer_type_kind itk_u + = narrowest_unsigned_type (integer.low, integer.high, flags); + enum integer_type_kind itk_s + = narrowest_signed_type (integer.low, integer.high, flags); /* In both C89 and C99, octal and hex constants may be signed or unsigned, whichever fits tighter. We do not warn about this @@ -506,10 +610,11 @@ interpret_integer (token, flags) if (itk_u < itk_unsigned_long) itk_u = itk_unsigned_long; itk = itk_u; - warning ("this decimal constant is unsigned only in ISO C90"); + warning (0, "this decimal constant is unsigned only in ISO C90"); } - else if (warn_traditional) - warning ("this decimal constant would be unsigned in ISO C90"); + else + warning (OPT_Wtraditional, + "this decimal constant would be unsigned in ISO C90"); } } } @@ -524,15 +629,15 @@ interpret_integer (token, flags) if (itk > itk_unsigned_long && (flags & CPP_N_WIDTH) != CPP_N_LARGE - && ! in_system_header && ! flag_isoc99) - pedwarn ("integer constant is too large for \"%s\" type", + && !in_system_header && !flag_isoc99) + pedwarn ("integer constant is too large for %qs type", (flags & CPP_N_UNSIGNED) ? "unsigned long" : "long"); - TREE_TYPE (value) = type; + value = build_int_cst_wide (type, integer.low, integer.high); /* Convert imaginary to a complex type. */ if (flags & CPP_N_IMAGINARY) - value = build_complex (NULL_TREE, convert (type, integer_zero_node), value); + value = build_complex (NULL_TREE, build_int_cst (type, 0), value); return value; } @@ -540,60 +645,100 @@ interpret_integer (token, flags) /* Interpret TOKEN, a floating point number with FLAGS as classified by cpplib. */ static tree -interpret_float (token, flags) - const cpp_token *token; - unsigned int flags; +interpret_float (const cpp_token *token, unsigned int flags) { tree type; tree value; REAL_VALUE_TYPE real; char *copy; size_t copylen; - const char *typename; - /* FIXME: make %T work in error/warning, then we don't need typename. */ - if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) - { + /* Decode _Fract and _Accum. */ + if (flags & CPP_N_FRACT || flags & CPP_N_ACCUM) + return interpret_fixed (token, flags); + + /* Decode type based on width and properties. */ + if (flags & CPP_N_DFLOAT) + if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) + type = dfloat128_type_node; + else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL) + type = dfloat32_type_node; + else + type = dfloat64_type_node; + else + if (flags & CPP_N_WIDTH_MD) + { + char suffix; + enum machine_mode mode; + + if ((flags & CPP_N_WIDTH_MD) == CPP_N_MD_W) + suffix = 'w'; + else + suffix = 'q'; + + mode = targetm.c.mode_for_suffix (suffix); + if (mode == VOIDmode) + { + error ("unsupported non-standard suffix on floating constant"); + errorcount++; + + return error_mark_node; + } + else if (pedantic) + pedwarn ("non-standard suffix on floating constant"); + + type = c_common_type_for_mode (mode, 0); + gcc_assert (type); + } + else if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) type = long_double_type_node; - typename = "long double"; - } - else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL - || flag_single_precision_constant) - { + else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL + || flag_single_precision_constant) type = float_type_node; - typename = "float"; - } - else - { + else type = double_type_node; - typename = "double"; - } /* Copy the constant to a nul-terminated buffer. If the constant has any suffixes, cut them off; REAL_VALUE_ATOF/ REAL_VALUE_HTOF can't handle them. */ copylen = token->val.str.len; - if ((flags & CPP_N_WIDTH) != CPP_N_MEDIUM) - /* Must be an F or L suffix. */ - copylen--; - if (flags & CPP_N_IMAGINARY) - /* I or J suffix. */ - copylen--; + if (flags & CPP_N_DFLOAT) + copylen -= 2; + else + { + if ((flags & CPP_N_WIDTH) != CPP_N_MEDIUM) + /* Must be an F or L or machine defined suffix. */ + copylen--; + if (flags & CPP_N_IMAGINARY) + /* I or J suffix. */ + copylen--; + } - copy = alloca (copylen + 1); + copy = (char *) alloca (copylen + 1); memcpy (copy, token->val.str.text, copylen); copy[copylen] = '\0'; - real_from_string (&real, copy); - real_convert (&real, TYPE_MODE (type), &real); + real_from_string3 (&real, copy, TYPE_MODE (type)); - /* A diagnostic is required for "soft" overflow by some ISO C - testsuites. This is not pedwarn, because some people don't want - an error for this. - ??? That's a dubious reason... is this a mandatory diagnostic or - isn't it? -- zw, 2001-08-21. */ - if (REAL_VALUE_ISINF (real) && pedantic) - warning ("floating constant exceeds range of \"%s\"", typename); + /* Both C and C++ require a diagnostic for a floating constant + outside the range of representable values of its type. Since we + have __builtin_inf* to produce an infinity, this is now a + mandatory pedwarn if the target does not support infinities. */ + if (REAL_VALUE_ISINF (real)) + { + if (!MODE_HAS_INFINITIES (TYPE_MODE (type))) + pedwarn ("floating constant exceeds range of %qT", type); + else + warning (OPT_Woverflow, "floating constant exceeds range of %qT", type); + } + /* We also give a warning if the value underflows. */ + else if (REAL_VALUES_EQUAL (real, dconst0)) + { + REAL_VALUE_TYPE realvoidmode; + int overflow = real_from_string (&realvoidmode, copy); + if (overflow < 0 || !REAL_VALUES_EQUAL (realvoidmode, dconst0)) + warning (OPT_Woverflow, "floating constant truncated to zero"); + } /* Create a node with determined type and value. */ value = build_real (type, real); @@ -603,112 +748,261 @@ interpret_float (token, flags) return value; } +/* Interpret TOKEN, a fixed-point number with FLAGS as classified + by cpplib. */ + static tree -lex_string (str, len, wide) - const unsigned char *str; - unsigned int len; - int wide; +interpret_fixed (const cpp_token *token, unsigned int flags) { + tree type; tree value; - char *buf = alloca ((len + 1) * (wide ? WCHAR_BYTES : 1)); - char *q = buf; - const unsigned char *p = str, *limit = str + len; - cppchar_t c; - -#ifdef MULTIBYTE_CHARS - /* Reset multibyte conversion state. */ - (void) local_mbtowc (NULL, NULL, 0); -#endif + FIXED_VALUE_TYPE fixed; + char *copy; + size_t copylen; - while (p < limit) - { -#ifdef MULTIBYTE_CHARS - wchar_t wc; - int char_len; + copylen = token->val.str.len; - char_len = local_mbtowc (&wc, (const char *) p, limit - p); - if (char_len == -1) + if (flags & CPP_N_FRACT) /* _Fract. */ + { + if (flags & CPP_N_UNSIGNED) /* Unsigned _Fract. */ { - warning ("ignoring invalid multibyte character"); - char_len = 1; - c = *p++; + if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) + { + type = unsigned_long_long_fract_type_node; + copylen -= 4; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_MEDIUM) + { + type = unsigned_long_fract_type_node; + copylen -= 3; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL) + { + type = unsigned_short_fract_type_node; + copylen -= 3; + } + else + { + type = unsigned_fract_type_node; + copylen -= 2; + } } - else + else /* Signed _Fract. */ { - p += char_len; - c = wc; - } -#else - c = *p++; -#endif - - if (c == '\\' && !ignore_escape_flag) - c = cpp_parse_escape (parse_in, &p, limit, wide); - - /* Add this single character into the buffer either as a wchar_t, - a multibyte sequence, or as a single byte. */ - if (wide) + if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) + { + type = long_long_fract_type_node; + copylen -= 3; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_MEDIUM) + { + type = long_fract_type_node; + copylen -= 2; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL) + { + type = short_fract_type_node; + copylen -= 2; + } + else + { + type = fract_type_node; + copylen --; + } + } + } + else /* _Accum. */ + { + if (flags & CPP_N_UNSIGNED) /* Unsigned _Accum. */ { - unsigned charwidth = TYPE_PRECISION (char_type_node); - unsigned bytemask = (1 << charwidth) - 1; - int byte; - - for (byte = 0; byte < WCHAR_BYTES; ++byte) + if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) { - int n; - if (byte >= (int) sizeof (c)) - n = 0; - else - n = (c >> (byte * charwidth)) & bytemask; - if (BYTES_BIG_ENDIAN) - q[WCHAR_BYTES - byte - 1] = n; - else - q[byte] = n; + type = unsigned_long_long_accum_type_node; + copylen -= 4; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_MEDIUM) + { + type = unsigned_long_accum_type_node; + copylen -= 3; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL) + { + type = unsigned_short_accum_type_node; + copylen -= 3; + } + else + { + type = unsigned_accum_type_node; + copylen -= 2; } - q += WCHAR_BYTES; } -#ifdef MULTIBYTE_CHARS - else if (char_len > 1) - { - /* We're dealing with a multibyte character. */ - for ( ; char_len >0; --char_len) + else /* Signed _Accum. */ + { + if ((flags & CPP_N_WIDTH) == CPP_N_LARGE) + { + type = long_long_accum_type_node; + copylen -= 3; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_MEDIUM) + { + type = long_accum_type_node; + copylen -= 2; + } + else if ((flags & CPP_N_WIDTH) == CPP_N_SMALL) { - *q++ = *(p - char_len); + type = short_accum_type_node; + copylen -= 2; + } + else + { + type = accum_type_node; + copylen --; } } -#endif - else + } + + copy = (char *) alloca (copylen + 1); + memcpy (copy, token->val.str.text, copylen); + copy[copylen] = '\0'; + + fixed_from_string (&fixed, copy, TYPE_MODE (type)); + + /* Create a node with determined type and value. */ + value = build_fixed (type, fixed); + + return value; +} + +/* Convert a series of STRING and/or WSTRING tokens into a tree, + performing string constant concatenation. TOK is the first of + these. VALP is the location to write the string into. OBJC_STRING + indicates whether an '@' token preceded the incoming token. + Returns the CPP token type of the result (CPP_STRING, CPP_WSTRING, + or CPP_OBJC_STRING). + + This is unfortunately more work than it should be. If any of the + strings in the series has an L prefix, the result is a wide string + (6.4.5p4). Whether or not the result is a wide string affects the + meaning of octal and hexadecimal escapes (6.4.4.4p6,9). But escape + sequences do not continue across the boundary between two strings in + a series (6.4.5p7), so we must not lose the boundaries. Therefore + cpp_interpret_string takes a vector of cpp_string structures, which + we must arrange to provide. */ + +static enum cpp_ttype +lex_string (const cpp_token *tok, tree *valp, bool objc_string) +{ + tree value; + bool wide = false; + size_t concats = 0; + struct obstack str_ob; + cpp_string istr; + + /* Try to avoid the overhead of creating and destroying an obstack + for the common case of just one string. */ + cpp_string str = tok->val.str; + cpp_string *strs = &str; + + if (tok->type == CPP_WSTRING) + wide = true; + + retry: + tok = cpp_get_token (parse_in); + switch (tok->type) + { + case CPP_PADDING: + goto retry; + case CPP_ATSIGN: + if (c_dialect_objc ()) + { + objc_string = true; + goto retry; + } + /* FALLTHROUGH */ + + default: + break; + + case CPP_WSTRING: + wide = true; + /* FALLTHROUGH */ + + case CPP_STRING: + if (!concats) { - *q++ = c; + gcc_obstack_init (&str_ob); + obstack_grow (&str_ob, &str, sizeof (cpp_string)); } + + concats++; + obstack_grow (&str_ob, &tok->val.str, sizeof (cpp_string)); + goto retry; } - /* Terminate the string value, either with a single byte zero - or with a wide zero. */ + /* We have read one more token than we want. */ + _cpp_backup_tokens (parse_in, 1); + if (concats) + strs = XOBFINISH (&str_ob, cpp_string *); - if (wide) + if (concats && !objc_string && !in_system_header) + warning (OPT_Wtraditional, + "traditional C rejects string constant concatenation"); + + if ((c_lex_string_translate + ? cpp_interpret_string : cpp_interpret_string_notranslate) + (parse_in, strs, concats + 1, &istr, wide)) { - memset (q, 0, WCHAR_BYTES); - q += WCHAR_BYTES; + value = build_string (istr.len, (const char *) istr.text); + free (CONST_CAST (unsigned char *, istr.text)); + + if (c_lex_string_translate == -1) + { + int xlated = cpp_interpret_string_notranslate (parse_in, strs, + concats + 1, + &istr, wide); + /* Assume that, if we managed to translate the string above, + then the untranslated parsing will always succeed. */ + gcc_assert (xlated); + + if (TREE_STRING_LENGTH (value) != (int) istr.len + || 0 != strncmp (TREE_STRING_POINTER (value), + (const char *) istr.text, istr.len)) + { + /* Arrange for us to return the untranslated string in + *valp, but to set up the C type of the translated + one. */ + *valp = build_string (istr.len, (const char *) istr.text); + valp = &TREE_CHAIN (*valp); + } + free (CONST_CAST (unsigned char *, istr.text)); + } } else { - *q++ = '\0'; + /* Callers cannot generally handle error_mark_node in this context, + so return the empty string instead. cpp_interpret_string has + issued an error. */ + if (wide) + value = build_string (TYPE_PRECISION (wchar_type_node) + / TYPE_PRECISION (char_type_node), + "\0\0\0"); /* widest supported wchar_t + is 32 bits */ + else + value = build_string (1, ""); } - value = build_string (q - buf, buf); + TREE_TYPE (value) = wide ? wchar_array_type_node : char_array_type_node; + *valp = fix_string_type (value); - if (wide) - TREE_TYPE (value) = wchar_array_type_node; - else - TREE_TYPE (value) = char_array_type_node; - return value; + if (concats) + obstack_free (&str_ob, 0); + + return objc_string ? CPP_OBJC_STRING : wide ? CPP_WSTRING : CPP_STRING; } /* Converts a (possibly wide) character constant token into a tree. */ static tree -lex_charconst (token) - const cpp_token *token; +lex_charconst (const cpp_token *token) { cppchar_t result; tree type, value; @@ -716,24 +1010,23 @@ lex_charconst (token) int unsignedp; result = cpp_interpret_charconst (parse_in, token, - &chars_seen, &unsignedp); - - /* Cast to cppchar_signed_t to get correct sign-extension of RESULT - before possibly widening to HOST_WIDE_INT for build_int_2. */ - if (unsignedp || (cppchar_signed_t) result >= 0) - value = build_int_2 (result, 0); - else - value = build_int_2 ((cppchar_signed_t) result, -1); + &chars_seen, &unsignedp); if (token->type == CPP_WCHAR) type = wchar_type_node; /* In C, a character constant has type 'int'. In C++ 'char', but multi-char charconsts have type 'int'. */ - else if ((c_language == clk_c) || chars_seen > 1) + else if (!c_dialect_cxx () || chars_seen > 1) type = integer_type_node; else type = char_type_node; - TREE_TYPE (value) = type; + /* Cast to cppchar_signed_t to get correct sign-extension of RESULT + before possibly widening to HOST_WIDE_INT for build_int_cst. */ + if (unsignedp || (cppchar_signed_t) result >= 0) + value = build_int_cst_wide (type, result, 0); + else + value = build_int_cst_wide (type, (cppchar_signed_t) result, -1); + return value; }