/* Define builtin-in macros for the C family front ends. Copyright (C) 2002, 2003, 2004, 2005 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 version. GCC is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 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, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ #include "config.h" #include "system.h" #include "coretypes.h" #include "tm.h" #include "tree.h" #include "version.h" #include "flags.h" #include "real.h" #include "c-common.h" #include "c-pragma.h" #include "output.h" #include "except.h" /* For USING_SJLJ_EXCEPTIONS. */ #include "toplev.h" #include "tm_p.h" /* Target prototypes. */ #include "target.h" #ifndef TARGET_OS_CPP_BUILTINS # define TARGET_OS_CPP_BUILTINS() #endif #ifndef TARGET_OBJFMT_CPP_BUILTINS # define TARGET_OBJFMT_CPP_BUILTINS() #endif #ifndef REGISTER_PREFIX #define REGISTER_PREFIX "" #endif /* Non-static as some targets don't use it. */ void builtin_define_std (const char *) ATTRIBUTE_UNUSED; static void builtin_define_with_value_n (const char *, const char *, size_t); static void builtin_define_with_int_value (const char *, HOST_WIDE_INT); static void builtin_define_with_hex_fp_value (const char *, tree, int, const char *, const char *, const char *); static void builtin_define_stdint_macros (void); static void builtin_define_type_max (const char *, tree, int); static void builtin_define_type_precision (const char *, tree); static void builtin_define_float_constants (const char *, const char *, const char *, tree); static void define__GNUC__ (void); /* Define NAME with value TYPE precision. */ static void builtin_define_type_precision (const char *name, tree type) { builtin_define_with_int_value (name, TYPE_PRECISION (type)); } /* Define the float.h constants for TYPE using NAME_PREFIX, FP_SUFFIX, and FP_CAST. */ static void builtin_define_float_constants (const char *name_prefix, const char *fp_suffix, const char *fp_cast, tree type) { /* Used to convert radix-based values to base 10 values in several cases. In the max_exp -> max_10_exp conversion for 128-bit IEEE, we need at least 6 significant digits for correct results. Using the fraction formed by (log(2)*1e6)/(log(10)*1e6) overflows a 32-bit integer as an intermediate; perhaps someone can find a better approximation, in the mean time, I suspect using doubles won't harm the bootstrap here. */ const double log10_2 = .30102999566398119521; double log10_b; const struct real_format *fmt; char name[64], buf[128]; int dig, min_10_exp, max_10_exp; int decimal_dig; fmt = REAL_MODE_FORMAT (TYPE_MODE (type)); gcc_assert (fmt->b != 10); /* The radix of the exponent representation. */ if (type == float_type_node) builtin_define_with_int_value ("__FLT_RADIX__", fmt->b); log10_b = log10_2 * fmt->log2_b; /* The number of radix digits, p, in the floating-point significand. */ sprintf (name, "__%s_MANT_DIG__", name_prefix); builtin_define_with_int_value (name, fmt->p); /* The number of decimal digits, q, such that any floating-point number with q decimal digits can be rounded into a floating-point number with p radix b digits and back again without change to the q decimal digits, p log10 b if b is a power of 10 floor((p - 1) log10 b) otherwise */ dig = (fmt->p - 1) * log10_b; sprintf (name, "__%s_DIG__", name_prefix); builtin_define_with_int_value (name, dig); /* The minimum negative int x such that b**(x-1) is a normalized float. */ sprintf (name, "__%s_MIN_EXP__", name_prefix); sprintf (buf, "(%d)", fmt->emin); builtin_define_with_value (name, buf, 0); /* The minimum negative int x such that 10**x is a normalized float, ceil (log10 (b ** (emin - 1))) = ceil (log10 (b) * (emin - 1)) Recall that emin is negative, so the integer truncation calculates the ceiling, not the floor, in this case. */ min_10_exp = (fmt->emin - 1) * log10_b; sprintf (name, "__%s_MIN_10_EXP__", name_prefix); sprintf (buf, "(%d)", min_10_exp); builtin_define_with_value (name, buf, 0); /* The maximum int x such that b**(x-1) is a representable float. */ sprintf (name, "__%s_MAX_EXP__", name_prefix); builtin_define_with_int_value (name, fmt->emax); /* The maximum int x such that 10**x is in the range of representable finite floating-point numbers, floor (log10((1 - b**-p) * b**emax)) = floor (log10(1 - b**-p) + log10(b**emax)) = floor (log10(1 - b**-p) + log10(b)*emax) The safest thing to do here is to just compute this number. But since we don't link cc1 with libm, we cannot. We could implement log10 here a series expansion, but that seems too much effort because: Note that the first term, for all extant p, is a number exceedingly close to zero, but slightly negative. Note that the second term is an integer scaling an irrational number, and that because of the floor we are only interested in its integral portion. In order for the first term to have any effect on the integral portion of the second term, the second term has to be exceedingly close to an integer itself (e.g. 123.000000000001 or something). Getting a result that close to an integer requires that the irrational multiplicand have a long series of zeros in its expansion, which doesn't occur in the first 20 digits or so of log10(b). Hand-waving aside, crunching all of the sets of constants above by hand does not yield a case for which the first term is significant, which in the end is all that matters. */ max_10_exp = fmt->emax * log10_b; sprintf (name, "__%s_MAX_10_EXP__", name_prefix); builtin_define_with_int_value (name, max_10_exp); /* The number of decimal digits, n, such that any floating-point number can be rounded to n decimal digits and back again without change to the value. p * log10(b) if b is a power of 10 ceil(1 + p * log10(b)) otherwise The only macro we care about is this number for the widest supported floating type, but we want this value for rendering constants below. */ { double d_decimal_dig = 1 + fmt->p * log10_b; decimal_dig = d_decimal_dig; if (decimal_dig < d_decimal_dig) decimal_dig++; } if (type == long_double_type_node) builtin_define_with_int_value ("__DECIMAL_DIG__", decimal_dig); /* Since, for the supported formats, B is always a power of 2, we construct the following numbers directly as a hexadecimal constants. */ /* The maximum representable finite floating-point number, (1 - b**-p) * b**emax */ { int i, n; char *p; strcpy (buf, "0x0."); n = fmt->p * fmt->log2_b; for (i = 0, p = buf + 4; i + 3 < n; i += 4) *p++ = 'f'; if (i < n) *p++ = "08ce"[n - i]; sprintf (p, "p%d", fmt->emax * fmt->log2_b); if (fmt->pnan < fmt->p) { /* This is an IBM extended double format made up of two IEEE doubles. The value of the long double is the sum of the values of the two parts. The most significant part is required to be the value of the long double rounded to the nearest double. Rounding means we need a slightly smaller value for LDBL_MAX. */ buf[4 + fmt->pnan / 4] = "7bde"[fmt->pnan % 4]; } } sprintf (name, "__%s_MAX__", name_prefix); builtin_define_with_hex_fp_value (name, type, decimal_dig, buf, fp_suffix, fp_cast); /* The minimum normalized positive floating-point number, b**(emin-1). */ sprintf (name, "__%s_MIN__", name_prefix); sprintf (buf, "0x1p%d", (fmt->emin - 1) * fmt->log2_b); builtin_define_with_hex_fp_value (name, type, decimal_dig, buf, fp_suffix, fp_cast); /* The difference between 1 and the least value greater than 1 that is representable in the given floating point type, b**(1-p). */ sprintf (name, "__%s_EPSILON__", name_prefix); if (fmt->pnan < fmt->p) /* This is an IBM extended double format, so 1.0 + any double is representable precisely. */ sprintf (buf, "0x1p%d", (fmt->emin - fmt->p) * fmt->log2_b); else sprintf (buf, "0x1p%d", (1 - fmt->p) * fmt->log2_b); builtin_define_with_hex_fp_value (name, type, decimal_dig, buf, fp_suffix, fp_cast); /* For C++ std::numeric_limits::denorm_min. The minimum denormalized positive floating-point number, b**(emin-p). Zero for formats that don't support denormals. */ sprintf (name, "__%s_DENORM_MIN__", name_prefix); if (fmt->has_denorm) { sprintf (buf, "0x1p%d", (fmt->emin - fmt->p) * fmt->log2_b); builtin_define_with_hex_fp_value (name, type, decimal_dig, buf, fp_suffix, fp_cast); } else { sprintf (buf, "0.0%s", fp_suffix); builtin_define_with_value (name, buf, 0); } sprintf (name, "__%s_HAS_DENORM__", name_prefix); builtin_define_with_value (name, fmt->has_denorm ? "1" : "0", 0); /* For C++ std::numeric_limits::has_infinity. */ sprintf (name, "__%s_HAS_INFINITY__", name_prefix); builtin_define_with_int_value (name, MODE_HAS_INFINITIES (TYPE_MODE (type))); /* For C++ std::numeric_limits::has_quiet_NaN. We do not have a predicate to distinguish a target that has both quiet and signalling NaNs from a target that has only quiet NaNs or only signalling NaNs, so we assume that a target that has any kind of NaN has quiet NaNs. */ sprintf (name, "__%s_HAS_QUIET_NAN__", name_prefix); builtin_define_with_int_value (name, MODE_HAS_NANS (TYPE_MODE (type))); } /* Define __DECx__ constants for TYPE using NAME_PREFIX and SUFFIX. */ static void builtin_define_decimal_float_constants (const char *name_prefix, const char *suffix, tree type) { const struct real_format *fmt; char name[64], buf[128], *p; int digits; fmt = REAL_MODE_FORMAT (TYPE_MODE (type)); /* The number of radix digits, p, in the significand. */ sprintf (name, "__%s_MANT_DIG__", name_prefix); builtin_define_with_int_value (name, fmt->p); /* The minimum negative int x such that b**(x-1) is a normalized float. */ sprintf (name, "__%s_MIN_EXP__", name_prefix); sprintf (buf, "(%d)", fmt->emin); builtin_define_with_value (name, buf, 0); /* The maximum int x such that b**(x-1) is a representable float. */ sprintf (name, "__%s_MAX_EXP__", name_prefix); builtin_define_with_int_value (name, fmt->emax); /* Compute the minimum representable value. */ sprintf (name, "__%s_MIN__", name_prefix); sprintf (buf, "1E%d%s", fmt->emin, suffix); builtin_define_with_value (name, buf, 0); /* Compute the maximum representable value. */ sprintf (name, "__%s_MAX__", name_prefix); p = buf; for (digits = fmt->p; digits; digits--) { *p++ = '9'; if (digits == fmt->p) *p++ = '.'; } *p = 0; /* fmt->p plus 1, to account for the decimal point. */ sprintf (&buf[fmt->p + 1], "E%d%s", fmt->emax, suffix); builtin_define_with_value (name, buf, 0); /* Compute epsilon (the difference between 1 and least value greater than 1 representable). */ sprintf (name, "__%s_EPSILON__", name_prefix); sprintf (buf, "1E-%d%s", fmt->p - 1, suffix); builtin_define_with_value (name, buf, 0); /* Minimum denormalized postive decimal value. */ sprintf (name, "__%s_DEN__", name_prefix); p = buf; for (digits = fmt->p; digits > 1; digits--) { *p++ = '0'; if (digits == fmt->p) *p++ = '.'; } *p = 0; sprintf (&buf[fmt->p], "1E%d%s", fmt->emin, suffix); builtin_define_with_value (name, buf, 0); } /* Define __GNUC__, __GNUC_MINOR__ and __GNUC_PATCHLEVEL__. */ static void define__GNUC__ (void) { /* The format of the version string, enforced below, is ([^0-9]*-)?[0-9]+[.][0-9]+([.][0-9]+)?([- ].*)? */ const char *q, *v = version_string; while (*v && !ISDIGIT (*v)) v++; gcc_assert (*v && (v <= version_string || v[-1] == '-')); q = v; while (ISDIGIT (*v)) v++; builtin_define_with_value_n ("__GNUC__", q, v - q); if (c_dialect_cxx ()) builtin_define_with_value_n ("__GNUG__", q, v - q); gcc_assert (*v == '.' && ISDIGIT (v[1])); q = ++v; while (ISDIGIT (*v)) v++; builtin_define_with_value_n ("__GNUC_MINOR__", q, v - q); if (*v == '.') { gcc_assert (ISDIGIT (v[1])); q = ++v; while (ISDIGIT (*v)) v++; builtin_define_with_value_n ("__GNUC_PATCHLEVEL__", q, v - q); } else builtin_define_with_value_n ("__GNUC_PATCHLEVEL__", "0", 1); gcc_assert (!*v || *v == ' ' || *v == '-'); } /* Define macros used by . Currently only defines limits for intmax_t, used by the testsuite. */ static void builtin_define_stdint_macros (void) { int intmax_long; if (intmax_type_node == long_long_integer_type_node) intmax_long = 2; else if (intmax_type_node == long_integer_type_node) intmax_long = 1; else if (intmax_type_node == integer_type_node) intmax_long = 0; else gcc_unreachable (); builtin_define_type_max ("__INTMAX_MAX__", intmax_type_node, intmax_long); } /* Hook that registers front end and target-specific built-ins. */ void c_cpp_builtins (cpp_reader *pfile) { /* -undef turns off target-specific built-ins. */ if (flag_undef) return; define__GNUC__ (); /* For stddef.h. They require macros defined in c-common.c. */ c_stddef_cpp_builtins (); if (c_dialect_cxx ()) { if (flag_weak && SUPPORTS_ONE_ONLY) cpp_define (pfile, "__GXX_WEAK__=1"); else cpp_define (pfile, "__GXX_WEAK__=0"); if (warn_deprecated) cpp_define (pfile, "__DEPRECATED"); } /* ANDROID LOCAL BEGIN */ if (c_dialect_cxx () || c_dialect_objcxx ()) { if (!flag_rtti) cpp_define (pfile, "__NO_RTTI"); } /* ANDROID LOCAL END */ /* Note that we define this for C as well, so that we know if __attribute__((cleanup)) will interface with EH. */ if (flag_exceptions) cpp_define (pfile, "__EXCEPTIONS"); /* Represents the C++ ABI version, always defined so it can be used while preprocessing C and assembler. */ if (flag_abi_version == 0) /* Use a very large value so that: #if __GXX_ABI_VERSION >= will work whether the user explicitly says "-fabi-version=x" or "-fabi-version=0". Do not use INT_MAX because that will be different from system to system. */ builtin_define_with_int_value ("__GXX_ABI_VERSION", 999999); else if (flag_abi_version == 1) /* Due to a historical accident, this version had the value "102". */ builtin_define_with_int_value ("__GXX_ABI_VERSION", 102); else /* Newer versions have values 1002, 1003, .... */ builtin_define_with_int_value ("__GXX_ABI_VERSION", 1000 + flag_abi_version); /* libgcc needs to know this. */ if (USING_SJLJ_EXCEPTIONS) cpp_define (pfile, "__USING_SJLJ_EXCEPTIONS__"); /* limits.h needs to know these. */ builtin_define_type_max ("__SCHAR_MAX__", signed_char_type_node, 0); builtin_define_type_max ("__SHRT_MAX__", short_integer_type_node, 0); builtin_define_type_max ("__INT_MAX__", integer_type_node, 0); builtin_define_type_max ("__LONG_MAX__", long_integer_type_node, 1); builtin_define_type_max ("__LONG_LONG_MAX__", long_long_integer_type_node, 2); builtin_define_type_max ("__WCHAR_MAX__", wchar_type_node, 0); builtin_define_type_precision ("__CHAR_BIT__", char_type_node); /* stdint.h (eventually) and the testsuite need to know these. */ builtin_define_stdint_macros (); /* float.h needs to know these. */ builtin_define_with_int_value ("__FLT_EVAL_METHOD__", TARGET_FLT_EVAL_METHOD); /* And decfloat.h needs this. */ builtin_define_with_int_value ("__DEC_EVAL_METHOD__", TARGET_DEC_EVAL_METHOD); builtin_define_float_constants ("FLT", "F", "%s", float_type_node); /* Cast the double precision constants when single precision constants are specified. The correct result is computed by the compiler when using macros that include a cast. This has the side-effect of making the value unusable in const expressions. */ if (flag_single_precision_constant) builtin_define_float_constants ("DBL", "L", "((double)%s)", double_type_node); else builtin_define_float_constants ("DBL", "", "%s", double_type_node); builtin_define_float_constants ("LDBL", "L", "%s", long_double_type_node); /* For decfloat.h. */ builtin_define_decimal_float_constants ("DEC32", "DF", dfloat32_type_node); builtin_define_decimal_float_constants ("DEC64", "DD", dfloat64_type_node); builtin_define_decimal_float_constants ("DEC128", "DL", dfloat128_type_node); /* For use in assembly language. */ builtin_define_with_value ("__REGISTER_PREFIX__", REGISTER_PREFIX, 0); builtin_define_with_value ("__USER_LABEL_PREFIX__", user_label_prefix, 0); /* Misc. */ builtin_define_with_value ("__VERSION__", version_string, 1); cpp_define (pfile, "__GNUC_GNU_INLINE__"); /* 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__"); } /* Other target-independent built-ins determined by command-line options. */ 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_really_no_inline) cpp_define (pfile, "__NO_INLINE__"); if (flag_signaling_nans) cpp_define (pfile, "__SUPPORT_SNAN__"); if (flag_finite_math_only) cpp_define (pfile, "__FINITE_MATH_ONLY__=1"); else cpp_define (pfile, "__FINITE_MATH_ONLY__=0"); if (flag_pic) { builtin_define_with_int_value ("__pic__", flag_pic); builtin_define_with_int_value ("__PIC__", flag_pic); } if (flag_iso) cpp_define (pfile, "__STRICT_ANSI__"); if (!flag_signed_char) cpp_define (pfile, "__CHAR_UNSIGNED__"); if (c_dialect_cxx () && TYPE_UNSIGNED (wchar_type_node)) cpp_define (pfile, "__WCHAR_UNSIGNED__"); /* Make the choice of ObjC runtime visible to source code. */ if (c_dialect_objc () && flag_next_runtime) cpp_define (pfile, "__NEXT_RUNTIME__"); /* Show the availability of some target pragmas. */ if (flag_mudflap || targetm.handle_pragma_redefine_extname) cpp_define (pfile, "__PRAGMA_REDEFINE_EXTNAME"); if (targetm.handle_pragma_extern_prefix) cpp_define (pfile, "__PRAGMA_EXTERN_PREFIX"); /* Make the choice of the stack protector runtime visible to source code. The macro names and values here were chosen for compatibility with an earlier implementation, i.e. ProPolice. */ if (flag_stack_protect == 2) cpp_define (pfile, "__SSP_ALL__=2"); else if (flag_stack_protect == 1) cpp_define (pfile, "__SSP__=1"); if (flag_openmp) cpp_define (pfile, "_OPENMP=200505"); /* A straightforward target hook doesn't work, because of problems linking that hook's body when part of non-C front ends. */ # define preprocessing_asm_p() (cpp_get_options (pfile)->lang == CLK_ASM) # define preprocessing_trad_p() (cpp_get_options (pfile)->traditional) # define builtin_define(TXT) cpp_define (pfile, TXT) # define builtin_assert(TXT) cpp_assert (pfile, TXT) TARGET_CPU_CPP_BUILTINS (); TARGET_OS_CPP_BUILTINS (); TARGET_OBJFMT_CPP_BUILTINS (); /* Support the __declspec keyword by turning them into attributes. Note that the current way we do this may result in a collision with predefined attributes later on. This can be solved by using one attribute, say __declspec__, and passing args to it. The problem with that approach is that args are not accumulated: each new appearance would clobber any existing args. */ if (TARGET_DECLSPEC) builtin_define ("__declspec(x)=__attribute__((x))"); } /* Pass an object-like macro. If it doesn't lie in the user's namespace, defines it unconditionally. Otherwise define a version with two leading underscores, and another version with two leading and trailing underscores, and define the original only if an ISO standard was not nominated. e.g. passing "unix" defines "__unix", "__unix__" and possibly "unix". Passing "_mips" defines "__mips", "__mips__" and possibly "_mips". */ void builtin_define_std (const char *macro) { size_t len = strlen (macro); char *buff = (char *) alloca (len + 5); char *p = buff + 2; char *q = p + len; /* prepend __ (or maybe just _) if in user's namespace. */ memcpy (p, macro, len + 1); if (!( *p == '_' && (p[1] == '_' || ISUPPER (p[1])))) { if (*p != '_') *--p = '_'; if (p[1] != '_') *--p = '_'; } cpp_define (parse_in, p); /* If it was in user's namespace... */ if (p != buff + 2) { /* Define the macro with leading and following __. */ if (q[-1] != '_') *q++ = '_'; if (q[-2] != '_') *q++ = '_'; *q = '\0'; cpp_define (parse_in, p); /* Finally, define the original macro if permitted. */ if (!flag_iso) cpp_define (parse_in, macro); } } /* Pass an object-like macro and a value to define it to. The third parameter says whether or not to turn the value into a string constant. */ void builtin_define_with_value (const char *macro, const char *expansion, int is_str) { char *buf; size_t mlen = strlen (macro); size_t elen = strlen (expansion); size_t extra = 2; /* space for an = and a NUL */ if (is_str) extra += 2; /* space for two quote marks */ buf = (char *) alloca (mlen + elen + extra); if (is_str) sprintf (buf, "%s=\"%s\"", macro, expansion); else sprintf (buf, "%s=%s", macro, expansion); cpp_define (parse_in, buf); } /* Pass an object-like macro and a value to define it to. The third parameter is the length of the expansion. */ static void builtin_define_with_value_n (const char *macro, const char *expansion, size_t elen) { char *buf; size_t mlen = strlen (macro); /* Space for an = and a NUL. */ buf = (char *) alloca (mlen + elen + 2); memcpy (buf, macro, mlen); buf[mlen] = '='; memcpy (buf + mlen + 1, expansion, elen); buf[mlen + elen + 1] = '\0'; cpp_define (parse_in, buf); } /* Pass an object-like macro and an integer value to define it to. */ static void builtin_define_with_int_value (const char *macro, HOST_WIDE_INT value) { char *buf; size_t mlen = strlen (macro); size_t vlen = 18; size_t extra = 2; /* space for = and NUL. */ buf = (char *) alloca (mlen + vlen + extra); memcpy (buf, macro, mlen); buf[mlen] = '='; sprintf (buf + mlen + 1, HOST_WIDE_INT_PRINT_DEC, value); cpp_define (parse_in, buf); } /* Pass an object-like macro a hexadecimal floating-point value. */ static void builtin_define_with_hex_fp_value (const char *macro, tree type ATTRIBUTE_UNUSED, int digits, const char *hex_str, const char *fp_suffix, const char *fp_cast) { REAL_VALUE_TYPE real; char dec_str[64], buf1[256], buf2[256]; /* Hex values are really cool and convenient, except that they're not supported in strict ISO C90 mode. First, the "p-" sequence is not valid as part of a preprocessor number. Second, we get a pedwarn from the preprocessor, which has no context, so we can't suppress the warning with __extension__. So instead what we do is construct the number in hex (because it's easy to get the exact correct value), parse it as a real, then print it back out as decimal. */ real_from_string (&real, hex_str); real_to_decimal (dec_str, &real, sizeof (dec_str), digits, 0); /* Assemble the macro in the following fashion macro = fp_cast [dec_str fp_suffix] */ sprintf (buf1, "%s%s", dec_str, fp_suffix); sprintf (buf2, fp_cast, buf1); sprintf (buf1, "%s=%s", macro, buf2); cpp_define (parse_in, buf1); } /* Define MAX for TYPE based on the precision of the type. IS_LONG is 1 for type "long" and 2 for "long long". We have to handle unsigned types, since wchar_t might be unsigned. */ static void builtin_define_type_max (const char *macro, tree type, int is_long) { static const char *const values[] = { "127", "255", "32767", "65535", "2147483647", "4294967295", "9223372036854775807", "18446744073709551615", "170141183460469231731687303715884105727", "340282366920938463463374607431768211455" }; static const char *const suffixes[] = { "", "U", "L", "UL", "LL", "ULL" }; const char *value, *suffix; char *buf; size_t idx; /* Pre-rendering the values mean we don't have to futz with printing a multi-word decimal value. There are also a very limited number of precisions that we support, so it's really a waste of time. */ switch (TYPE_PRECISION (type)) { case 8: idx = 0; break; case 16: idx = 2; break; case 32: idx = 4; break; case 64: idx = 6; break; case 128: idx = 8; break; default: gcc_unreachable (); } value = values[idx + TYPE_UNSIGNED (type)]; suffix = suffixes[is_long * 2 + TYPE_UNSIGNED (type)]; buf = (char *) alloca (strlen (macro) + 1 + strlen (value) + strlen (suffix) + 1); sprintf (buf, "%s=%s%s", macro, value, suffix); cpp_define (parse_in, buf); }