[BACK]Return to intprops.h CVS log [TXT] [DIR] Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / libiconv / srclib

Diff for /embedaddon/libiconv/srclib/intprops.h between versions 1.1 and 1.1.1.2

version 1.1, 2012/02/21 22:57:48 version 1.1.1.2, 2012/05/29 09:29:43
Line 1 Line 1
 /* intprops.h -- properties of integer types  /* intprops.h -- properties of integer types
   
   Copyright (C) 2001, 2002, 2003, 2004, 2005 Free Software Foundation, Inc.   Copyright (C) 2001-2005, 2009-2011 Free Software Foundation, Inc.
   
    This program is free software: you can redistribute it and/or modify     This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by     it under the terms of the GNU General Public License as published by
Line 17 Line 17
   
 /* Written by Paul Eggert.  */  /* Written by Paul Eggert.  */
   
   #ifndef _GL_INTPROPS_H
   #define _GL_INTPROPS_H
   
 #include <limits.h>  #include <limits.h>
   
   /* Return an integer value, converted to the same type as the integer
      expression E after integer type promotion.  V is the unconverted value.  */
   #define _GL_INT_CONVERT(e, v) (0 * (e) + (v))
   
   /* Act like _GL_INT_CONVERT (E, -V) but work around a bug in IRIX 6.5 cc; see
      <http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00406.html>.  */
   #define _GL_INT_NEGATE_CONVERT(e, v) (0 * (e) - (v))
   
 /* The extra casts in the following macros work around compiler bugs,  /* The extra casts in the following macros work around compiler bugs,
    e.g., in Cray C 5.0.3.0.  */     e.g., in Cray C 5.0.3.0.  */
   
Line 34 Line 45
 #define TYPE_ONES_COMPLEMENT(t) ((t) ~ (t) 0 == 0)  #define TYPE_ONES_COMPLEMENT(t) ((t) ~ (t) 0 == 0)
 #define TYPE_SIGNED_MAGNITUDE(t) ((t) ~ (t) 0 < (t) -1)  #define TYPE_SIGNED_MAGNITUDE(t) ((t) ~ (t) 0 < (t) -1)
   
   /* True if the signed integer expression E uses two's complement.  */
   #define _GL_INT_TWOS_COMPLEMENT(e) (~ _GL_INT_CONVERT (e, 0) == -1)
   
 /* True if the arithmetic type T is signed.  */  /* True if the arithmetic type T is signed.  */
 #define TYPE_SIGNED(t) (! ((t) 0 < (t) -1))  #define TYPE_SIGNED(t) (! ((t) 0 < (t) -1))
   
/* The maximum and minimum values for the integer type T.  These/* Return 1 if the integer expression E, after integer promotion, has
    a signed type.  */
 #define _GL_INT_SIGNED(e) (_GL_INT_NEGATE_CONVERT (e, 1) < 0)
 
 
 /* Minimum and maximum values for integer types and expressions.  These
    macros have undefined behavior if T is signed and has padding bits.     macros have undefined behavior if T is signed and has padding bits.
    If this is a problem for you, please let us know how to fix it for     If this is a problem for you, please let us know how to fix it for
    your host.  */     your host.  */
 #define TYPE_MINIMUM(t) \  
   ((t) (! TYPE_SIGNED (t) \  
         ? (t) 0 \  
         : TYPE_SIGNED_MAGNITUDE (t) \  
         ? ~ (t) 0 \  
         : ~ (t) 0 << (sizeof (t) * CHAR_BIT - 1)))  
 #define TYPE_MAXIMUM(t) \  
   ((t) (! TYPE_SIGNED (t) \  
         ? (t) -1 \  
         : ~ (~ (t) 0 << (sizeof (t) * CHAR_BIT - 1))))  
   
/* Return zero if T can be determined to be an unsigned type./* The maximum and minimum values for the integer type T.  */
   Otherwise, return 1.#define TYPE_MINIMUM(t)                                                 \
   When compiling with GCC, INT_STRLEN_BOUND uses this macro to obtain a  ((t) (! TYPE_SIGNED (t)                                               \
   tighter bound.  Otherwise, it overestimates the true bound by one byte        ? (t) 0                                                         \
   when applied to unsigned types of size 2, 4, 16, ... bytes.        : TYPE_SIGNED_MAGNITUDE (t)                                     \
   The symbol signed_type_or_expr__ is private to this header file.  */        ? ~ (t) 0                                                       \
#if __GNUC__ >= 2        : ~ TYPE_MAXIMUM (t)))
# define signed_type_or_expr__(t) TYPE_SIGNED (__typeof__ (t))#define TYPE_MAXIMUM(t)                                                 \
   ((t) (! TYPE_SIGNED (t)                                               \
         ? (t) -1                                                        \
         : ((((t) 1 << (sizeof (t) * CHAR_BIT - 2)) - 1) * 2 + 1)))
 
 /* The maximum and minimum values for the type of the expression E,
    after integer promotion.  E should not have side effects.  */
 #define _GL_INT_MINIMUM(e)                                              \
   (_GL_INT_SIGNED (e)                                                   \
    ? - _GL_INT_TWOS_COMPLEMENT (e) - _GL_SIGNED_INT_MAXIMUM (e)         \
    : _GL_INT_CONVERT (e, 0))
 #define _GL_INT_MAXIMUM(e)                                              \
   (_GL_INT_SIGNED (e)                                                   \
    ? _GL_SIGNED_INT_MAXIMUM (e)                                         \
    : _GL_INT_NEGATE_CONVERT (e, 1))
 #define _GL_SIGNED_INT_MAXIMUM(e)                                       \
   (((_GL_INT_CONVERT (e, 1) << (sizeof ((e) + 0) * CHAR_BIT - 2)) - 1) * 2 + 1)
 
 
 /* Return 1 if the __typeof__ keyword works.  This could be done by
    'configure', but for now it's easier to do it by hand.  */
 #if 2 <= __GNUC__ || 0x5110 <= __SUNPRO_C
 # define _GL_HAVE___TYPEOF__ 1
 #else  #else
# define signed_type_or_expr__(t) 1# define _GL_HAVE___TYPEOF__ 0
 #endif  #endif
   
   /* Return 1 if the integer type or expression T might be signed.  Return 0
      if it is definitely unsigned.  This macro does not evaluate its argument,
      and expands to an integer constant expression.  */
   #if _GL_HAVE___TYPEOF__
   # define _GL_SIGNED_TYPE_OR_EXPR(t) TYPE_SIGNED (__typeof__ (t))
   #else
   # define _GL_SIGNED_TYPE_OR_EXPR(t) 1
   #endif
   
   /* Bound on length of the string representing an unsigned integer
      value representable in B bits.  log10 (2.0) < 146/485.  The
      smallest value of B where this bound is not tight is 2621.  */
   #define INT_BITS_STRLEN_BOUND(b) (((b) * 146 + 484) / 485)
   
 /* Bound on length of the string representing an integer type or expression T.  /* Bound on length of the string representing an integer type or expression T.
   Subtract 1 for the sign bit if T is signed; log10 (2.0) < 146/485;   Subtract 1 for the sign bit if T is signed, and then add 1 more for
   add 1 for integer division truncation; add 1 more for a minus sign   a minus sign if needed.
   if needed.  */ 
#define INT_STRLEN_BOUND(t) \ 
  ((sizeof (t) * CHAR_BIT - signed_type_or_expr__ (t)) * 146 / 485 \ 
   + signed_type_or_expr__ (t) + 1) 
   
      Because _GL_SIGNED_TYPE_OR_EXPR sometimes returns 0 when its argument is
      signed, this macro may overestimate the true bound by one byte when
      applied to unsigned types of size 2, 4, 16, ... bytes.  */
   #define INT_STRLEN_BOUND(t)                                     \
     (INT_BITS_STRLEN_BOUND (sizeof (t) * CHAR_BIT                 \
                             - _GL_SIGNED_TYPE_OR_EXPR (t))        \
      + _GL_SIGNED_TYPE_OR_EXPR (t))
   
 /* Bound on buffer size needed to represent an integer type or expression T,  /* Bound on buffer size needed to represent an integer type or expression T,
    including the terminating null.  */     including the terminating null.  */
 #define INT_BUFSIZE_BOUND(t) (INT_STRLEN_BOUND (t) + 1)  #define INT_BUFSIZE_BOUND(t) (INT_STRLEN_BOUND (t) + 1)
   
   
   /* Range overflow checks.
   
      The INT_<op>_RANGE_OVERFLOW macros return 1 if the corresponding C
      operators might not yield numerically correct answers due to
      arithmetic overflow.  They do not rely on undefined or
      implementation-defined behavior.  Their implementations are simple
      and straightforward, but they are a bit harder to use than the
      INT_<op>_OVERFLOW macros described below.
   
      Example usage:
   
        long int i = ...;
        long int j = ...;
        if (INT_MULTIPLY_RANGE_OVERFLOW (i, j, LONG_MIN, LONG_MAX))
          printf ("multiply would overflow");
        else
          printf ("product is %ld", i * j);
   
      Restrictions on *_RANGE_OVERFLOW macros:
   
      These macros do not check for all possible numerical problems or
      undefined or unspecified behavior: they do not check for division
      by zero, for bad shift counts, or for shifting negative numbers.
   
      These macros may evaluate their arguments zero or multiple times,
      so the arguments should not have side effects.  The arithmetic
      arguments (including the MIN and MAX arguments) must be of the same
      integer type after the usual arithmetic conversions, and the type
      must have minimum value MIN and maximum MAX.  Unsigned types should
      use a zero MIN of the proper type.
   
      These macros are tuned for constant MIN and MAX.  For commutative
      operations such as A + B, they are also tuned for constant B.  */
   
   /* Return 1 if A + B would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  */
   #define INT_ADD_RANGE_OVERFLOW(a, b, min, max)          \
     ((b) < 0                                              \
      ? (a) < (min) - (b)                                  \
      : (max) - (b) < (a))
   
   /* Return 1 if A - B would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  */
   #define INT_SUBTRACT_RANGE_OVERFLOW(a, b, min, max)     \
     ((b) < 0                                              \
      ? (max) + (b) < (a)                                  \
      : (a) < (min) + (b))
   
   /* Return 1 if - A would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  */
   #define INT_NEGATE_RANGE_OVERFLOW(a, min, max)          \
     ((min) < 0                                            \
      ? (a) < - (max)                                      \
      : 0 < (a))
   
   /* Return 1 if A * B would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  Avoid && and || as they tickle
      bugs in Sun C 5.11 2010/08/13 and other compilers; see
      <http://lists.gnu.org/archive/html/bug-gnulib/2011-05/msg00401.html>.  */
   #define INT_MULTIPLY_RANGE_OVERFLOW(a, b, min, max)     \
     ((b) < 0                                              \
      ? ((a) < 0                                           \
         ? (a) < (max) / (b)                               \
         : (b) == -1                                       \
         ? 0                                               \
         : (min) / (b) < (a))                              \
      : (b) == 0                                           \
      ? 0                                                  \
      : ((a) < 0                                           \
         ? (a) < (min) / (b)                               \
         : (max) / (b) < (a)))
   
   /* Return 1 if A / B would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  Do not check for division by zero.  */
   #define INT_DIVIDE_RANGE_OVERFLOW(a, b, min, max)       \
     ((min) < 0 && (b) == -1 && (a) < - (max))
   
   /* Return 1 if A % B would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  Do not check for division by zero.
      Mathematically, % should never overflow, but on x86-like hosts
      INT_MIN % -1 traps, and the C standard permits this, so treat this
      as an overflow too.  */
   #define INT_REMAINDER_RANGE_OVERFLOW(a, b, min, max)    \
     INT_DIVIDE_RANGE_OVERFLOW (a, b, min, max)
   
   /* Return 1 if A << B would overflow in [MIN,MAX] arithmetic.
      See above for restrictions.  Here, MIN and MAX are for A only, and B need
      not be of the same type as the other arguments.  The C standard says that
      behavior is undefined for shifts unless 0 <= B < wordwidth, and that when
      A is negative then A << B has undefined behavior and A >> B has
      implementation-defined behavior, but do not check these other
      restrictions.  */
   #define INT_LEFT_SHIFT_RANGE_OVERFLOW(a, b, min, max)   \
     ((a) < 0                                              \
      ? (a) < (min) >> (b)                                 \
      : (max) >> (b) < (a))
   
   
   /* The _GL*_OVERFLOW macros have the same restrictions as the
      *_RANGE_OVERFLOW macros, except that they do not assume that operands
      (e.g., A and B) have the same type as MIN and MAX.  Instead, they assume
      that the result (e.g., A + B) has that type.  */
   #define _GL_ADD_OVERFLOW(a, b, min, max)                                \
     ((min) < 0 ? INT_ADD_RANGE_OVERFLOW (a, b, min, max)                  \
      : (a) < 0 ? (b) <= (a) + (b)                                         \
      : (b) < 0 ? (a) <= (a) + (b)                                         \
      : (a) + (b) < (b))
   #define _GL_SUBTRACT_OVERFLOW(a, b, min, max)                           \
     ((min) < 0 ? INT_SUBTRACT_RANGE_OVERFLOW (a, b, min, max)             \
      : (a) < 0 ? 1                                                        \
      : (b) < 0 ? (a) - (b) <= (a)                                         \
      : (a) < (b))
   #define _GL_MULTIPLY_OVERFLOW(a, b, min, max)                           \
     (((min) == 0 && (((a) < 0 && 0 < (b)) || ((b) < 0 && 0 < (a))))       \
      || INT_MULTIPLY_RANGE_OVERFLOW (a, b, min, max))
   #define _GL_DIVIDE_OVERFLOW(a, b, min, max)                             \
     ((min) < 0 ? (b) == _GL_INT_NEGATE_CONVERT (min, 1) && (a) < - (max)  \
      : (a) < 0 ? (b) <= (a) + (b) - 1                                     \
      : (b) < 0 && (a) + (b) <= (a))
   #define _GL_REMAINDER_OVERFLOW(a, b, min, max)                          \
     ((min) < 0 ? (b) == _GL_INT_NEGATE_CONVERT (min, 1) && (a) < - (max)  \
      : (a) < 0 ? (a) % (b) != ((max) - (b) + 1) % (b)                     \
      : (b) < 0 && ! _GL_UNSIGNED_NEG_MULTIPLE (a, b, max))
   
   /* Return a nonzero value if A is a mathematical multiple of B, where
      A is unsigned, B is negative, and MAX is the maximum value of A's
      type.  A's type must be the same as (A % B)'s type.  Normally (A %
      -B == 0) suffices, but things get tricky if -B would overflow.  */
   #define _GL_UNSIGNED_NEG_MULTIPLE(a, b, max)                            \
     (((b) < -_GL_SIGNED_INT_MAXIMUM (b)                                   \
       ? (_GL_SIGNED_INT_MAXIMUM (b) == (max)                              \
          ? (a)                                                            \
          : (a) % (_GL_INT_CONVERT (a, _GL_SIGNED_INT_MAXIMUM (b)) + 1))   \
       : (a) % - (b))                                                      \
      == 0)
   
   
   /* Integer overflow checks.
   
      The INT_<op>_OVERFLOW macros return 1 if the corresponding C operators
      might not yield numerically correct answers due to arithmetic overflow.
      They work correctly on all known practical hosts, and do not rely
      on undefined behavior due to signed arithmetic overflow.
   
      Example usage:
   
        long int i = ...;
        long int j = ...;
        if (INT_MULTIPLY_OVERFLOW (i, j))
          printf ("multiply would overflow");
        else
          printf ("product is %ld", i * j);
   
      These macros do not check for all possible numerical problems or
      undefined or unspecified behavior: they do not check for division
      by zero, for bad shift counts, or for shifting negative numbers.
   
      These macros may evaluate their arguments zero or multiple times, so the
      arguments should not have side effects.
   
      These macros are tuned for their last argument being a constant.
   
      Return 1 if the integer expressions A * B, A - B, -A, A * B, A / B,
      A % B, and A << B would overflow, respectively.  */
   
   #define INT_ADD_OVERFLOW(a, b) \
     _GL_BINARY_OP_OVERFLOW (a, b, _GL_ADD_OVERFLOW)
   #define INT_SUBTRACT_OVERFLOW(a, b) \
     _GL_BINARY_OP_OVERFLOW (a, b, _GL_SUBTRACT_OVERFLOW)
   #define INT_NEGATE_OVERFLOW(a) \
     INT_NEGATE_RANGE_OVERFLOW (a, _GL_INT_MINIMUM (a), _GL_INT_MAXIMUM (a))
   #define INT_MULTIPLY_OVERFLOW(a, b) \
     _GL_BINARY_OP_OVERFLOW (a, b, _GL_MULTIPLY_OVERFLOW)
   #define INT_DIVIDE_OVERFLOW(a, b) \
     _GL_BINARY_OP_OVERFLOW (a, b, _GL_DIVIDE_OVERFLOW)
   #define INT_REMAINDER_OVERFLOW(a, b) \
     _GL_BINARY_OP_OVERFLOW (a, b, _GL_REMAINDER_OVERFLOW)
   #define INT_LEFT_SHIFT_OVERFLOW(a, b) \
     INT_LEFT_SHIFT_RANGE_OVERFLOW (a, b, \
                                    _GL_INT_MINIMUM (a), _GL_INT_MAXIMUM (a))
   
   /* Return 1 if the expression A <op> B would overflow,
      where OP_RESULT_OVERFLOW (A, B, MIN, MAX) does the actual test,
      assuming MIN and MAX are the minimum and maximum for the result type.
      Arguments should be free of side effects.  */
   #define _GL_BINARY_OP_OVERFLOW(a, b, op_result_overflow)        \
     op_result_overflow (a, b,                                     \
                         _GL_INT_MINIMUM (0 * (b) + (a)),          \
                         _GL_INT_MAXIMUM (0 * (b) + (a)))
   
   #endif /* _GL_INTPROPS_H */
Removed from v.1.1  
changed lines
  Added in v.1.1.1.2

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>