[mLib] / bits.h

/* -*-c-*-
 *
 * $Id$
 *
 * Portable bit-level manipulation macros
 *
 * (c) 1998 Straylight/Edgeware
 */

/*----- Licensing notice --------------------------------------------------*
 *
 * This file is part of the mLib utilities library.
 *
 * mLib is free software; you can redistribute it and/or modify
 * it under the terms of the GNU Library General Public License as
 * published by the Free Software Foundation; either version 2 of the
 * License, or (at your option) any later version.
 *
 * mLib 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 Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public
 * License along with mLib; if not, write to the Free
 * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
 * MA 02111-1307, USA.
 */

#ifndef MLIB_BITS_H
#define MLIB_BITS_H

#ifdef __cplusplus
  extern "C" {
#endif

/*----- Header files ------------------------------------------------------*/

#include <limits.h>
#include <stddef.h>
#if __STDC_VERSION__ >= 199900l
#  include <stdint.h>
#endif

/*----- Decide on some types ----------------------------------------------*/

/* --- Make GNU C shut up --- */

#if __GNUC__ > 2 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 91)
#  define MLIB_BITS_EXTENSION __extension__
#else
#  define MLIB_BITS_EXTENSION
#endif

/* --- Decide on a 32-bit type --- *
 *
 * I want a type which is capable of expressing 32-bit numbers.  Because some
 * implementations have 64-bit @long@s (infinitely preferable to the abortion
 * that is @long long@), using @unsigned long@ regardless is wasteful.  So,
 * if @int@ appears to be good enough, then I'll go with that.
 */

#if UINT_MAX >= 0xffffffffu
  typedef unsigned int uint32;
#else
  typedef unsigned long uint32;
#endif

/* --- Decide on a 64-bit type --- *
 *
 * The test is quite subtle.  Think about it.  Note that (at least on my
 * machine), the 32-bit macros are *much* faster than GCC's @long long@
 * support.
 */

#if defined(ULONG_LONG_MAX) && !defined(ULLONG_MAX)
#  define ULLONG_MAX ULONG_LONG_MAX
#endif

#if UINT_MAX >> 31 > 0xffffffff
#  define HAVE_UINT64
   typedef unsigned int uint64;
#elif ULONG_MAX >> 31 > 0xffffffff
#  define HAVE_UINT64
   typedef unsigned long uint64;
#elif defined(ULLONG_MAX)
#  define HAVE_UINT64
   MLIB_BITS_EXTENSION typedef unsigned long long uint64;
#endif

#ifdef DEBUG64
#  undef HAVE_UINT64
#endif

#ifdef HAVE_UINT64
  typedef struct { uint64 i; } kludge64;
#else
  typedef struct { uint32 hi, lo; } kludge64;
#endif

/* --- Decide on a 24-bit type --- */

#if UINT_MAX >= 0x00ffffffu
  typedef unsigned int uint24;
#else
  typedef unsigned long uint24;
#endif

/* --- Decide on 16-bit and 8-bit types --- *
 *
 * This is more for brevity than anything else.
 */

typedef unsigned short uint16;
typedef unsigned char octet, uint8;

/* --- WARNING! --- *
 *
 * Never lose sight of the fact that the above types may be wider than the
 * names suggest.  Some architectures have 32-bit @short@s for example.
 */

/*----- Macros ------------------------------------------------------------*/

/* --- Useful masks --- */

#define MASK8 0xffu
#define MASK16 0xffffu
#define MASK16_L MASK16
#define MASK16_B MASK16
#define MASK24 0xffffffu
#define MASK24_L MASK24
#define MASK24_B MASK24
#define MASK32 0xffffffffu
#define MASK32_L MASK32
#define MASK32_B MASK32

#ifdef HAVE_UINT64
#  define MASK64 MLIB_BITS_EXTENSION 0xffffffffffffffffu
#  define MASK64_L MASK64
#  define MASK64_B MASK64
#endif

/* --- Sizes --- */

#define SZ_8 1
#define SZ_16 2
#define SZ_16_L 2
#define SZ_16_B 2
#define SZ_24 3
#define SZ_24_L 3
#define SZ_24_B 3
#define SZ_32 4
#define SZ_32_L 4
#define SZ_32_B 4

#ifdef HAVE_UINT64
#  define SZ_64 8
#  define SZ_64_L 8
#  define SZ_64_B 8
#endif

/* --- Type aliases --- */

#define TY_U8 octet
#define TY_U16 uint16
#define TY_U16_L uint16
#define TY_U16_B uint16
#define TY_U24 uint24
#define TY_U24_L uint24
#define TY_U24_B uint24
#define TY_U32 uint32
#define TY_U32_L uint32
#define TY_U32_B uint32

#ifdef HAVE_UINT64
#  define TY_U64 uint64
#  define TY_U64_L uint64
#  define TY_U64_B uint64
#endif

/* --- List macros --- */

#ifdef HAVE_UINT64
#  define  DOUINTCONV(_)						\
     _(8, 8, 8)								\
     _(16, 16, 16) _(16, 16_L, 16l) _(16, 16_B, 16b)			\
     _(24, 24, 24) _(24, 24_L, 24l) _(24, 24_B, 24b)			\
     _(32, 32, 32) _(32, 32_L, 32l) _(32, 32_B, 32b)			\
     _(64, 64, 64) _(64, 64_L, 64l) _(64, 64_B, 64b)
#  define DOUINTSZ(_) _(8) _(16) _(24) _(32) _(64)
#else
#  define  DOUINTCONV(_)						\
     _(8, 8, 8)								\
     _(16, 16, 16) _(16, 16_L, 16l) _(16, 16_B, 16b)			\
     _(24, 24, 24) _(24, 24_L, 24l) _(24, 24_B, 24b)			\
     _(32, 32, 32) _(32, 32_L, 32l) _(32, 32_B, 32b)
#  define DOUINTSZ(_) _(8) _(16) _(24) _(32)
#endif

/* --- Type coercions --- */

#define U8(x) ((octet)((x) & MASK8))
#define U16(x) ((uint16)((x) & MASK16))
#define U24(x) ((uint24)((x) & MASK24))
#define U32(x) ((uint32)((x) & MASK32))

#ifdef HAVE_UINT64
#  define U64(x) ((uint64)(x) & MASK64)
#  define U64_(d, x) ((d).i = U64(x).i)
#else
#  define U64_(d, x) ((d).hi = U32((x).hi), (d).lo = U32((x).lo))
#endif

/* --- Safe shifting macros --- */

#define LSL8(v, s) U8(U8(v) << ((s) & 7u))
#define LSR8(v, s) U8(U8(v) >> ((s) & 7u))
#define LSL16(v, s) U16(U16(v) << ((s) & 15u))
#define LSR16(v, s) U16(U16(v) >> ((s) & 15u))
#define LSL24(v, s) U24(U24(v) << ((s) % 24u))
#define LSR24(v, s) U24(U24(v) >> ((s) % 24u))
#define LSL32(v, s) U32(U32(v) << ((s) & 31u))
#define LSR32(v, s) U32(U32(v) >> ((s) & 31u))

#ifdef HAVE_UINT64
#  define LSL64(v, s) U64(U64(v) << ((s) & 63u))
#  define LSR64(v, s) U64(U64(v) >> ((s) & 63u))
#  define LSL64_(d, v, s) ((d).i = LSL64((v).i, (s)))
#  define LSR64_(d, v, s) ((d).i = LSR64((v).i, (s)))
#else
#  define LSL64_(d, v, s) do {						\
     unsigned _s = (s) & 63u;						\
     uint32 _l = (v).lo, _h = (v).hi;					\
     kludge64 *_d = &(d);						\
     if (_s >= 32) {							\
       _d->hi = LSL32(_l, _s - 32u);					\
       _d->lo = 0;							\
     } else if (!_s) {							\
       _d->lo = _l;							\
       _d->hi = _h;							\
     } else {								\
       _d->hi = LSL32(_h, _s) | LSR32(_l, 32u - _s);			\
       _d->lo = LSL32(_l, _s);						\
     }									\
   } while (0)
#  define LSR64_(d, v, s) do {						\
     unsigned _s = (s) & 63u;						\
     uint32 _l = (v).lo, _h = (v).hi;					\
     kludge64 *_d = &(d);						\
     if (_s >= 32) {							\
       _d->lo = LSR32(_h, _s - 32u);					\
       _d->hi = 0;							\
     } else if (!_s) {							\
       _d->lo = _l;							\
       _d->hi = _h;							\
     } else {								\
       _d->lo = LSR32(_l, _s) | LSL32(_h, 32u - _s);			\
       _d->hi = LSR32(_h, _s);						\
     }									\
   } while (0)
#endif

/* --- Rotation macros --- */

#define ROL8(v, s) (LSL8((v), (s)) | (LSR8((v), 8u - (s))))
#define ROR8(v, s) (LSR8((v), (s)) | (LSL8((v), 8u - (s))))
#define ROL16(v, s) (LSL16((v), (s)) | (LSR16((v), 16u - (s))))
#define ROR16(v, s) (LSR16((v), (s)) | (LSL16((v), 16u - (s))))
#define ROL24(v, s) (LSL24((v), (s)) | (LSR24((v), 24u - (s))))
#define ROR24(v, s) (LSR24((v), (s)) | (LSL24((v), 24u - (s))))
#define ROL32(v, s) (LSL32((v), (s)) | (LSR32((v), 32u - (s))))
#define ROR32(v, s) (LSR32((v), (s)) | (LSL32((v), 32u - (s))))

#ifdef HAVE_UINT64
#  define ROL64(v, s) (LSL64((v), (s)) | (LSR64((v), 64u - (s))))
#  define ROR64(v, s) (LSR64((v), (s)) | (LSL64((v), 64u - (s))))
#  define ROL64_(d, v, s) ((d).i = ROL64((v).i, (s)))
#  define ROR64_(d, v, s) ((d).i = ROR64((v).i, (s)))
#else
#  define ROL64_(d, v, s) do {						\
     unsigned _s = (s) & 63u;						\
     uint32 _l = (v).lo, _h = (v).hi;					\
     kludge64 *_d = &(d);						\
     if (_s > 32) {							\
       _d->hi = LSL32(_l, _s - 32u) | LSR32(_h, 64u - _s);		\
       _d->lo = LSL32(_h, _s - 32u) | LSR32(_l, 64u - _s);		\
     } else if (!_s) {							\
       _d->lo = _l;							\
       _d->hi = _h;							\
     } else if (_s == 32) {						\
       _d->lo = _h;							\
       _d->hi = _l;							\
     } else {								\
       _d->hi = LSL32(_h, _s) | LSR32(_l, 32u - _s);			\
       _d->lo = LSL32(_l, _s) | LSR32(_h, 32u - _s);			\
     }									\
   } while (0)
#  define ROR64_(d, v, s) do {						\
     unsigned _s = (s) & 63u;						\
     uint32 _l = (v).lo, _h = (v).hi;					\
     kludge64 *_d = &(d);						\
     if (_s > 32) {							\
       _d->hi = LSR32(_l, _s - 32u) | LSL32(_h, 64u - _s);		\
       _d->lo = LSR32(_h, _s - 32u) | LSL32(_l, 64u - _s);		\
     } else if (!_s) {							\
       _d->lo = _l;							\
       _d->hi = _h;							\
     } else if (_s == 32) {						\
       _d->lo = _h;							\
       _d->hi = _l;							\
     } else {								\
       _d->hi = LSR32(_h, _s) | LSL32(_l, 32u - _s);			\
       _d->lo = LSR32(_l, _s) | LSL32(_h, 32u - _s);			\
     }									\
   } while (0)
#endif

/* --- Storage and retrieval --- */

#define GETBYTE(p, o) (((octet *)(p))[o] & MASK8)
#define PUTBYTE(p, o, v) (((octet *)(p))[o] = U8((v)))

#define LOAD8(p) (GETBYTE((p), 0))
#define STORE8(p, v) (PUTBYTE((p), 0, (v)))

#define LOAD16_B(p)							\
  (((uint16)GETBYTE((p), 0) << 8) |					\
   ((uint16)GETBYTE((p), 1) << 0))
#define LOAD16_L(p)							\
  (((uint16)GETBYTE((p), 0) << 0) |					\
   ((uint16)GETBYTE((p), 1) << 8))
#define LOAD16(p) LOAD16_B((p))

#define STORE16_B(p, v)							\
  (PUTBYTE((p), 0, (uint16)(v) >> 8),					\
   PUTBYTE((p), 1, (uint16)(v) >> 0))
#define STORE16_L(p, v)							\
  (PUTBYTE((p), 0, (uint16)(v) >> 0),					\
   PUTBYTE((p), 1, (uint16)(v) >> 8))
#define STORE16(p, v) STORE16_B((p), (v))

#define LOAD24_B(p)							\
  (((uint24)GETBYTE((p), 0) << 16) |					\
   ((uint24)GETBYTE((p), 1) <<	8) |					\
   ((uint24)GETBYTE((p), 2) <<	0))
#define LOAD24_L(p)							\
  (((uint24)GETBYTE((p), 0) <<	0) |					\
   ((uint24)GETBYTE((p), 1) <<	8) |					\
   ((uint24)GETBYTE((p), 2) << 16))
#define LOAD24(p) LOAD24_B((p))

#define STORE24_B(p, v)							\
  (PUTBYTE((p), 0, (uint24)(v) >> 16),					\
   PUTBYTE((p), 1, (uint24)(v) >>  8),					\
   PUTBYTE((p), 2, (uint24)(v) >>  0))
#define STORE24_L(p, v)							\
  (PUTBYTE((p), 0, (uint24)(v) >>  0),					\
   PUTBYTE((p), 1, (uint24)(v) >>  8),					\
   PUTBYTE((p), 2, (uint24)(v) >> 16))
#define STORE24(p, v) STORE24_B((p), (v))

#define LOAD32_B(p)							\
  (((uint32)GETBYTE((p), 0) << 24) |					\
   ((uint32)GETBYTE((p), 1) << 16) |					\
   ((uint32)GETBYTE((p), 2) <<	8) |					\
   ((uint32)GETBYTE((p), 3) <<	0))
#define LOAD32_L(p)							\
  (((uint32)GETBYTE((p), 0) <<	0) |					\
   ((uint32)GETBYTE((p), 1) <<	8) |					\
   ((uint32)GETBYTE((p), 2) << 16) |					\
   ((uint32)GETBYTE((p), 3) << 24))
#define LOAD32(p) LOAD32_B((p))

#define STORE32_B(p, v)							\
  (PUTBYTE((p), 0, (uint32)(v) >> 24),					\
   PUTBYTE((p), 1, (uint32)(v) >> 16),					\
   PUTBYTE((p), 2, (uint32)(v) >>  8),					\
   PUTBYTE((p), 3, (uint32)(v) >>  0))
#define STORE32_L(p, v)							\
  (PUTBYTE((p), 0, (uint32)(v) >>  0),					\
   PUTBYTE((p), 1, (uint32)(v) >>  8),					\
   PUTBYTE((p), 2, (uint32)(v) >> 16),					\
   PUTBYTE((p), 3, (uint32)(v) >> 24))
#define STORE32(p, v) STORE32_B((p), (v))

#ifdef HAVE_UINT64

#  define LOAD64_B(p)							\
    (((uint64)GETBYTE((p), 0) << 56) |					\
     ((uint64)GETBYTE((p), 1) << 48) |					\
     ((uint64)GETBYTE((p), 2) << 40) |					\
     ((uint64)GETBYTE((p), 3) << 32) |					\
     ((uint64)GETBYTE((p), 4) << 24) |					\
     ((uint64)GETBYTE((p), 5) << 16) |					\
     ((uint64)GETBYTE((p), 6) <<  8) |					\
     ((uint64)GETBYTE((p), 7) <<  0))
#  define LOAD64_L(p)							\
    (((uint64)GETBYTE((p), 0) <<  0) |					\
     ((uint64)GETBYTE((p), 1) <<  8) |					\
     ((uint64)GETBYTE((p), 2) << 16) |					\
     ((uint64)GETBYTE((p), 3) << 24) |					\
     ((uint64)GETBYTE((p), 4) << 32) |					\
     ((uint64)GETBYTE((p), 5) << 40) |					\
     ((uint64)GETBYTE((p), 6) << 48) |					\
     ((uint64)GETBYTE((p), 7) << 56))
#  define LOAD64(p) LOAD64_B((p))
#  define LOAD64_B_(d, p) ((d).i = LOAD64_B((p)))
#  define LOAD64_L_(d, p) ((d).i = LOAD64_L((p)))
#  define LOAD64_(d, p) LOAD64_B_((d), (p))

#  define STORE64_B(p, v)						\
    (PUTBYTE((p), 0, (uint64)(v) >> 56),				\
     PUTBYTE((p), 1, (uint64)(v) >> 48),				\
     PUTBYTE((p), 2, (uint64)(v) >> 40),				\
     PUTBYTE((p), 3, (uint64)(v) >> 32),				\
     PUTBYTE((p), 4, (uint64)(v) >> 24),				\
     PUTBYTE((p), 5, (uint64)(v) >> 16),				\
     PUTBYTE((p), 6, (uint64)(v) >>  8),				\
     PUTBYTE((p), 7, (uint64)(v) >>  0))
#  define STORE64_L(p, v)						\
    (PUTBYTE((p), 0, (uint64)(v) >>  0),				\
     PUTBYTE((p), 1, (uint64)(v) >>  8),				\
     PUTBYTE((p), 2, (uint64)(v) >> 16),				\
     PUTBYTE((p), 3, (uint64)(v) >> 24),				\
     PUTBYTE((p), 4, (uint64)(v) >> 32),				\
     PUTBYTE((p), 5, (uint64)(v) >> 40),				\
     PUTBYTE((p), 6, (uint64)(v) >> 48),				\
     PUTBYTE((p), 7, (uint64)(v) >> 56))
#  define STORE64(p, v) STORE64_B((p), (v))
#  define STORE64_B_(p, v) STORE64_B((p), (v).i)
#  define STORE64_L_(p, v) STORE64_L((p), (v).i)
#  define STORE64_(p, v) STORE64_B_((p), (v))

#else

#  define LOAD64_B_(d, p)						\
    ((d).hi = LOAD32_B((octet *)(p) + 0),				\
     (d).lo = LOAD32_B((octet *)(p) + 4))
#  define LOAD64_L_(d, p)						\
    ((d).lo = LOAD32_L((octet *)(p) + 0),				\
     (d).hi = LOAD32_L((octet *)(p) + 4))
#  define LOAD64_(d, p) LOAD64_B_((d), (p))

#  define STORE64_B_(p, v)						\
    (STORE32_B((octet *)(p) + 0, (v).hi),				\
     STORE32_B((octet *)(p) + 4, (v).lo))
#  define STORE64_L_(p, v)						\
    (STORE32_L((octet *)(p) + 0, (v).lo),				\
     STORE32_L((octet *)(p) + 4, (v).hi))
#  define STORE64_(p, v) STORE64_B_((p), (v))

#endif

/* --- Other operations on 64-bit integers --- */

#ifdef HAVE_UINT64
#  define SET64(d, h, l) ((d).i = (U64((h)) << 32) | U64((l)))
#  define ASSIGN64(d, x) ((d).i = U64((x)))
#  define HI64(x) U32((x).i >> 32)
#  define LO64(x) U32((x).i)
#  define GET64(t, x) ((t)(x).i)
#else
#  define SET64(d, h, l) ((d).hi = U32(h), (d).lo = U32(l))
#  define ASSIGN64(d, x)						\
    ((d).hi = ((x & ~MASK32) >> 16) >> 16, (d).lo = U32(x))
#  define HI64(x) U32((x).hi)
#  define LO64(x) U32((x).lo)
#  define GET64(t, x) (((((t)HI64(x) << 16) << 16) & ~MASK32) | (t)LO64(x))
#endif

#ifdef HAVE_UINT64
#  define AND64(d, x, y) ((d).i = (x).i & (y).i)
#  define OR64(d, x, y) ((d).i = (x).i | (y).i)
#  define XOR64(d, x, y) ((d).i = (x).i ^ (y).i)
#  define CPL64(d, x) ((d).i = ~(x).i)
#  define ADD64(d, x, y) ((d).i = (x).i + (y).i)
#  define SUB64(d, x, y) ((d).i = (x).i - (y).i)
#  define CMP64(x, op, y) ((x).i op (y).i)
#  define ZERO64(x) ((x) == 0)
#else
#  define AND64(d, x, y) ((d).lo = (x).lo & (y).lo, (d).hi = (x).hi & (y).hi)
#  define OR64(d, x, y) ((d).lo = (x).lo | (y).lo, (d).hi = (x).hi | (y).hi)
#  define XOR64(d, x, y) ((d).lo = (x).lo ^ (y).lo, (d).hi = (x).hi ^ (y).hi)
#  define CPL64(d, x) ((d).lo = ~(x).lo, (d).hi = ~(x).hi)
#  define ADD64(d, x, y) do {						\
     uint32 _x = U32((x).lo + (y).lo);					\
     (d).hi = (x).hi + (y).hi + (_x < (x).lo);				\
     (d).lo = _x;							\
   } while (0)
#  define SUB64(d, x, y) do {						\
     uint32 _x = U32((x).lo - (y).lo);					\
     (d).hi = (x).hi - (y).hi - (_x > (x).lo);				\
     (d).lo = _x;							\
   } while (0)
#  define CMP64(x, op, y)						\
    ((x).hi == (y).hi ? (x).lo op (y).lo : (x).hi op (y).hi)
#  define ZERO64(x) ((x).lo == 0 && (x).hi == 0)
#endif

/* --- Storing integers in tables --- */

#ifdef HAVE_UINT64
#  define X64(x, y) { 0x##x##y }
#else
#  define X64(x, y) { 0x##x, 0x##y }
#endif

/*----- That's all, folks -------------------------------------------------*/

#ifdef __cplusplus
  }
#endif

#endif
Commit	Line	Data
6b033bc4	1	/* --c--
6b033bc4	2	*
5892fd39	3	* $Id$
6b033bc4	4	*
	5	* Portable bit-level manipulation macros
	6	*
	7	* (c) 1998 Straylight/Edgeware
	8	*/
	9
d4efbcd9	10	/----- Licensing notice --------------------------------------------------
6b033bc4	11	*
	12	* This file is part of the mLib utilities library.
	13	*
	14	* mLib is free software; you can redistribute it and/or modify
	15	* it under the terms of the GNU Library General Public License as
	16	* published by the Free Software Foundation; either version 2 of the
	17	* License, or (at your option) any later version.
d4efbcd9	18	*
6b033bc4	19	* mLib is distributed in the hope that it will be useful,
	20	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	21	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	22	* GNU Library General Public License for more details.
d4efbcd9	23	*
6b033bc4	24	* You should have received a copy of the GNU Library General Public
	25	* License along with mLib; if not, write to the Free
	26	* Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
	27	* MA 02111-1307, USA.
	28	*/
	29
c6e0eaf0	30	#ifndef MLIB_BITS_H
c6e0eaf0	31	#define MLIB_BITS_H
6b033bc4	32
	33	#ifdef __cplusplus
	34	extern "C" {
	35	#endif
	36
	37	/----- Header files ------------------------------------------------------/
	38
	39	#include <limits.h>
	40	#include <stddef.h>
a6f4a484	41	#if __STDC_VERSION__ >= 199900l
	42	# include <stdint.h>
	43	#endif
6b033bc4	44
	45	/----- Decide on some types ----------------------------------------------/
	46
1f175471	47	/* --- Make GNU C shut up --- */
	48
	49	#if __GNUC__ > 2 \|\| (__GNUC__ == 2 && __GNUC_MINOR__ >= 91)
	50	# define MLIB_BITS_EXTENSION __extension__
	51	#else
	52	# define MLIB_BITS_EXTENSION
	53	#endif
	54
6b033bc4	55	/* --- Decide on a 32-bit type --- *
	56	*
	57	* I want a type which is capable of expressing 32-bit numbers. Because some
	58	* implementations have 64-bit @long@s (infinitely preferable to the abortion
	59	* that is @long long@), using @unsigned long@ regardless is wasteful. So,
	60	* if @int@ appears to be good enough, then I'll go with that.
	61	*/
	62
	63	#if UINT_MAX >= 0xffffffffu
	64	typedef unsigned int uint32;
	65	#else
	66	typedef unsigned long uint32;
	67	#endif
	68
a6f4a484	69	/* --- Decide on a 64-bit type --- *
	70	*
	71	* The test is quite subtle. Think about it. Note that (at least on my
	72	* machine), the 32-bit macros are much faster than GCC's @long long@
	73	* support.
	74	*/
	75
	76	#if defined(ULONG_LONG_MAX) && !defined(ULLONG_MAX)
	77	# define ULLONG_MAX ULONG_LONG_MAX
	78	#endif
	79
	80	#if UINT_MAX >> 31 > 0xffffffff
1f175471	81	# define HAVE_UINT64
1f175471	82	typedef unsigned int uint64;
a6f4a484	83	#elif ULONG_MAX >> 31 > 0xffffffff
1f175471	84	# define HAVE_UINT64
1f175471	85	typedef unsigned long uint64;
a6f4a484	86	#elif defined(ULLONG_MAX)
1f175471	87	# define HAVE_UINT64
1f175471	88	MLIB_BITS_EXTENSION typedef unsigned long long uint64;
a6f4a484	89	#endif
	90
	91	#ifdef DEBUG64
	92	# undef HAVE_UINT64
	93	#endif
	94
	95	#ifdef HAVE_UINT64
	96	typedef struct { uint64 i; } kludge64;
	97	#else
	98	typedef struct { uint32 hi, lo; } kludge64;
	99	#endif
	100
6a0129ea	101	/* --- Decide on a 24-bit type --- */
	102
	103	#if UINT_MAX >= 0x00ffffffu
	104	typedef unsigned int uint24;
	105	#else
	106	typedef unsigned long uint24;
	107	#endif
	108
6b033bc4	109	/* --- Decide on 16-bit and 8-bit types --- *
	110	*
	111	* This is more for brevity than anything else.
	112	*/
	113
	114	typedef unsigned short uint16;
9b5ac6ff	115	typedef unsigned char octet, uint8;
6b033bc4	116
	117	/* --- WARNING! --- *
	118	*
	119	* Never lose sight of the fact that the above types may be wider than the
	120	* names suggest. Some architectures have 32-bit @short@s for example.
	121	*/
	122
	123	/----- Macros ------------------------------------------------------------/
	124
	125	/* --- Useful masks --- */
	126
	127	#define MASK8 0xffu
	128	#define MASK16 0xffffu
9b5ac6ff	129	#define MASK16_L MASK16
9b5ac6ff	130	#define MASK16_B MASK16
6a0129ea	131	#define MASK24 0xffffffu
9b5ac6ff	132	#define MASK24_L MASK24
9b5ac6ff	133	#define MASK24_B MASK24
6b033bc4	134	#define MASK32 0xffffffffu
9b5ac6ff	135	#define MASK32_L MASK32
9b5ac6ff	136	#define MASK32_B MASK32
6b033bc4	137
a6f4a484	138	#ifdef HAVE_UINT64
1f175471	139	# define MASK64 MLIB_BITS_EXTENSION 0xffffffffffffffffu
9b5ac6ff	140	# define MASK64_L MASK64
9b5ac6ff	141	# define MASK64_B MASK64
a6f4a484	142	#endif
a6f4a484	143
76a7638e	144	/* --- Sizes --- */
	145
	146	#define SZ_8 1
	147	#define SZ_16 2
	148	#define SZ_16_L 2
	149	#define SZ_16_B 2
	150	#define SZ_24 3
	151	#define SZ_24_L 3
	152	#define SZ_24_B 3
	153	#define SZ_32 4
	154	#define SZ_32_L 4
	155	#define SZ_32_B 4
	156
9b5ac6ff	157	#ifdef HAVE_UINT64
	158	# define SZ_64 8
	159	# define SZ_64_L 8
	160	# define SZ_64_B 8
	161	#endif
	162
	163	/* --- Type aliases --- */
	164
	165	#define TY_U8 octet
	166	#define TY_U16 uint16
	167	#define TY_U16_L uint16
	168	#define TY_U16_B uint16
	169	#define TY_U24 uint24
	170	#define TY_U24_L uint24
	171	#define TY_U24_B uint24
	172	#define TY_U32 uint32
	173	#define TY_U32_L uint32
	174	#define TY_U32_B uint32
	175
	176	#ifdef HAVE_UINT64
	177	# define TY_U64 uint64
	178	# define TY_U64_L uint64
	179	# define TY_U64_B uint64
	180	#endif
	181
	182	/* --- List macros --- */
	183
	184	#ifdef HAVE_UINT64
	185	# define DOUINTCONV(_) \
	186	_(8, 8, 8) \
	187	_(16, 16, 16) _(16, 16_L, 16l) _(16, 16_B, 16b) \
	188	_(24, 24, 24) _(24, 24_L, 24l) _(24, 24_B, 24b) \
	189	_(32, 32, 32) _(32, 32_L, 32l) _(32, 32_B, 32b) \
	190	_(64, 64, 64) _(64, 64_L, 64l) _(64, 64_B, 64b)
5ce3df29	191	# define DOUINTSZ(_) _(8) _(16) _(24) _(32) _(64)
9b5ac6ff	192	#else
	193	# define DOUINTCONV(_) \
	194	_(8, 8, 8) \
	195	_(16, 16, 16) _(16, 16_L, 16l) _(16, 16_B, 16b) \
	196	_(24, 24, 24) _(24, 24_L, 24l) _(24, 24_B, 24b) \
	197	_(32, 32, 32) _(32, 32_L, 32l) _(32, 32_B, 32b)
5ce3df29	198	# define DOUINTSZ(_) _(8) _(16) _(24) _(32)
9b5ac6ff	199	#endif
9b5ac6ff	200
6b033bc4	201	/* --- Type coercions --- */
	202
	203	#define U8(x) ((octet)((x) & MASK8))
	204	#define U16(x) ((uint16)((x) & MASK16))
6a0129ea	205	#define U24(x) ((uint24)((x) & MASK24))
6b033bc4	206	#define U32(x) ((uint32)((x) & MASK32))
6b033bc4	207
a6f4a484	208	#ifdef HAVE_UINT64
	209	# define U64(x) ((uint64)(x) & MASK64)
	210	# define U64_(d, x) ((d).i = U64(x).i)
	211	#else
	212	# define U64_(d, x) ((d).hi = U32((x).hi), (d).lo = U32((x).lo))
	213	#endif
	214
6b033bc4	215	/* --- Safe shifting macros --- */
6b033bc4	216
bc9f60b6	217	#define LSL8(v, s) U8(U8(v) << ((s) & 7u))
	218	#define LSR8(v, s) U8(U8(v) >> ((s) & 7u))
	219	#define LSL16(v, s) U16(U16(v) << ((s) & 15u))
	220	#define LSR16(v, s) U16(U16(v) >> ((s) & 15u))
6a0129ea	221	#define LSL24(v, s) U24(U24(v) << ((s) % 24u))
6a0129ea	222	#define LSR24(v, s) U24(U24(v) >> ((s) % 24u))
bc9f60b6	223	#define LSL32(v, s) U32(U32(v) << ((s) & 31u))
bc9f60b6	224	#define LSR32(v, s) U32(U32(v) >> ((s) & 31u))
6b033bc4	225
a6f4a484	226	#ifdef HAVE_UINT64
	227	# define LSL64(v, s) U64(U64(v) << ((s) & 63u))
	228	# define LSR64(v, s) U64(U64(v) >> ((s) & 63u))
	229	# define LSL64_(d, v, s) ((d).i = LSL64((v).i, (s)))
	230	# define LSR64_(d, v, s) ((d).i = LSR64((v).i, (s)))
	231	#else
	232	# define LSL64_(d, v, s) do { \
	233	unsigned _s = (s) & 63u; \
	234	uint32 _l = (v).lo, _h = (v).hi; \
	235	kludge64 *_d = &(d); \
	236	if (_s >= 32) { \
	237	_d->hi = LSL32(_l, _s - 32u); \
	238	_d->lo = 0; \
	239	} else if (!_s) { \
	240	_d->lo = _l; \
	241	_d->hi = _h; \
	242	} else { \
	243	_d->hi = LSL32(_h, _s) \| LSR32(_l, 32u - _s); \
	244	_d->lo = LSL32(_l, _s); \
	245	} \
	246	} while (0)
	247	# define LSR64_(d, v, s) do { \
	248	unsigned _s = (s) & 63u; \
	249	uint32 _l = (v).lo, _h = (v).hi; \
	250	kludge64 *_d = &(d); \
	251	if (_s >= 32) { \
	252	_d->lo = LSR32(_h, _s - 32u); \
	253	_d->hi = 0; \
	254	} else if (!_s) { \
	255	_d->lo = _l; \
	256	_d->hi = _h; \
	257	} else { \
	258	_d->lo = LSR32(_l, _s) \| LSL32(_h, 32u - _s); \
	259	_d->hi = LSR32(_h, _s); \
	260	} \
	261	} while (0)
	262	#endif
	263
6b033bc4	264	/* --- Rotation macros --- */
6b033bc4	265
bc9f60b6	266	#define ROL8(v, s) (LSL8((v), (s)) \| (LSR8((v), 8u - (s))))
	267	#define ROR8(v, s) (LSR8((v), (s)) \| (LSL8((v), 8u - (s))))
	268	#define ROL16(v, s) (LSL16((v), (s)) \| (LSR16((v), 16u - (s))))
	269	#define ROR16(v, s) (LSR16((v), (s)) \| (LSL16((v), 16u - (s))))
6a0129ea	270	#define ROL24(v, s) (LSL24((v), (s)) \| (LSR24((v), 24u - (s))))
6a0129ea	271	#define ROR24(v, s) (LSR24((v), (s)) \| (LSL24((v), 24u - (s))))
bc9f60b6	272	#define ROL32(v, s) (LSL32((v), (s)) \| (LSR32((v), 32u - (s))))
bc9f60b6	273	#define ROR32(v, s) (LSR32((v), (s)) \| (LSL32((v), 32u - (s))))
6b033bc4	274
a6f4a484	275	#ifdef HAVE_UINT64
	276	# define ROL64(v, s) (LSL64((v), (s)) \| (LSR64((v), 64u - (s))))
	277	# define ROR64(v, s) (LSR64((v), (s)) \| (LSL64((v), 64u - (s))))
	278	# define ROL64_(d, v, s) ((d).i = ROL64((v).i, (s)))
	279	# define ROR64_(d, v, s) ((d).i = ROR64((v).i, (s)))
	280	#else
	281	# define ROL64_(d, v, s) do { \
	282	unsigned _s = (s) & 63u; \
	283	uint32 _l = (v).lo, _h = (v).hi; \
	284	kludge64 *_d = &(d); \
5892fd39	285	if (_s > 32) { \
a6f4a484	286	_d->hi = LSL32(_l, _s - 32u) \| LSR32(_h, 64u - _s); \
	287	_d->lo = LSL32(_h, _s - 32u) \| LSR32(_l, 64u - _s); \
	288	} else if (!_s) { \
	289	_d->lo = _l; \
	290	_d->hi = _h; \
5892fd39	291	} else if (_s == 32) { \
	292	_d->lo = _h; \
	293	_d->hi = _l; \
a6f4a484	294	} else { \
	295	_d->hi = LSL32(_h, _s) \| LSR32(_l, 32u - _s); \
	296	_d->lo = LSL32(_l, _s) \| LSR32(_h, 32u - _s); \
	297	} \
	298	} while (0)
	299	# define ROR64_(d, v, s) do { \
	300	unsigned _s = (s) & 63u; \
	301	uint32 _l = (v).lo, _h = (v).hi; \
	302	kludge64 *_d = &(d); \
5892fd39	303	if (_s > 32) { \
a6f4a484	304	_d->hi = LSR32(_l, _s - 32u) \| LSL32(_h, 64u - _s); \
	305	_d->lo = LSR32(_h, _s - 32u) \| LSL32(_l, 64u - _s); \
	306	} else if (!_s) { \
	307	_d->lo = _l; \
	308	_d->hi = _h; \
5892fd39	309	} else if (_s == 32) { \
	310	_d->lo = _h; \
	311	_d->hi = _l; \
a6f4a484	312	} else { \
	313	_d->hi = LSR32(_h, _s) \| LSL32(_l, 32u - _s); \
	314	_d->lo = LSR32(_l, _s) \| LSL32(_h, 32u - _s); \
	315	} \
	316	} while (0)
	317	#endif
	318
6b033bc4	319	/* --- Storage and retrieval --- */
	320
	321	#define GETBYTE(p, o) (((octet *)(p))[o] & MASK8)
	322	#define PUTBYTE(p, o, v) (((octet *)(p))[o] = U8((v)))
	323
	324	#define LOAD8(p) (GETBYTE((p), 0))
	325	#define STORE8(p, v) (PUTBYTE((p), 0, (v)))
	326
0d4e268a	327	#define LOAD16_B(p) \
	328	(((uint16)GETBYTE((p), 0) << 8) \| \
	329	((uint16)GETBYTE((p), 1) << 0))
	330	#define LOAD16_L(p) \
	331	(((uint16)GETBYTE((p), 0) << 0) \| \
	332	((uint16)GETBYTE((p), 1) << 8))
6b033bc4	333	#define LOAD16(p) LOAD16_B((p))
6b033bc4	334
0d4e268a	335	#define STORE16_B(p, v) \
	336	(PUTBYTE((p), 0, (uint16)(v) >> 8), \
	337	PUTBYTE((p), 1, (uint16)(v) >> 0))
	338	#define STORE16_L(p, v) \
	339	(PUTBYTE((p), 0, (uint16)(v) >> 0), \
	340	PUTBYTE((p), 1, (uint16)(v) >> 8))
6b033bc4	341	#define STORE16(p, v) STORE16_B((p), (v))
6b033bc4	342
6a0129ea	343	#define LOAD24_B(p) \
6a0129ea	344	(((uint24)GETBYTE((p), 0) << 16) \| \
d4efbcd9 MW	345	((uint24)GETBYTE((p), 1) << 8) \| \
d4efbcd9 MW	346	((uint24)GETBYTE((p), 2) << 0))
6a0129ea	347	#define LOAD24_L(p) \
d4efbcd9 MW	348	(((uint24)GETBYTE((p), 0) << 0) \| \
d4efbcd9 MW	349	((uint24)GETBYTE((p), 1) << 8) \| \
6a0129ea	350	((uint24)GETBYTE((p), 2) << 16))
	351	#define LOAD24(p) LOAD24_B((p))
	352
	353	#define STORE24_B(p, v) \
	354	(PUTBYTE((p), 0, (uint24)(v) >> 16), \
	355	PUTBYTE((p), 1, (uint24)(v) >> 8), \
	356	PUTBYTE((p), 2, (uint24)(v) >> 0))
	357	#define STORE24_L(p, v) \
	358	(PUTBYTE((p), 0, (uint24)(v) >> 0), \
	359	PUTBYTE((p), 1, (uint24)(v) >> 8), \
	360	PUTBYTE((p), 2, (uint24)(v) >> 16))
	361	#define STORE24(p, v) STORE24_B((p), (v))
	362
6b033bc4	363	#define LOAD32_B(p) \
0d4e268a	364	(((uint32)GETBYTE((p), 0) << 24) \| \
0d4e268a	365	((uint32)GETBYTE((p), 1) << 16) \| \
d4efbcd9 MW	366	((uint32)GETBYTE((p), 2) << 8) \| \
d4efbcd9 MW	367	((uint32)GETBYTE((p), 3) << 0))
6b033bc4	368	#define LOAD32_L(p) \
d4efbcd9 MW	369	(((uint32)GETBYTE((p), 0) << 0) \| \
d4efbcd9 MW	370	((uint32)GETBYTE((p), 1) << 8) \| \
0d4e268a	371	((uint32)GETBYTE((p), 2) << 16) \| \
0d4e268a	372	((uint32)GETBYTE((p), 3) << 24))
6b033bc4	373	#define LOAD32(p) LOAD32_B((p))
	374
	375	#define STORE32_B(p, v) \
0d4e268a	376	(PUTBYTE((p), 0, (uint32)(v) >> 24), \
	377	PUTBYTE((p), 1, (uint32)(v) >> 16), \
	378	PUTBYTE((p), 2, (uint32)(v) >> 8), \
	379	PUTBYTE((p), 3, (uint32)(v) >> 0))
6b033bc4	380	#define STORE32_L(p, v) \
0d4e268a	381	(PUTBYTE((p), 0, (uint32)(v) >> 0), \
	382	PUTBYTE((p), 1, (uint32)(v) >> 8), \
	383	PUTBYTE((p), 2, (uint32)(v) >> 16), \
	384	PUTBYTE((p), 3, (uint32)(v) >> 24))
6b033bc4	385	#define STORE32(p, v) STORE32_B((p), (v))
6b033bc4	386
a6f4a484	387	#ifdef HAVE_UINT64
	388
	389	# define LOAD64_B(p) \
	390	(((uint64)GETBYTE((p), 0) << 56) \| \
	391	((uint64)GETBYTE((p), 1) << 48) \| \
	392	((uint64)GETBYTE((p), 2) << 40) \| \
	393	((uint64)GETBYTE((p), 3) << 32) \| \
	394	((uint64)GETBYTE((p), 4) << 24) \| \
	395	((uint64)GETBYTE((p), 5) << 16) \| \
	396	((uint64)GETBYTE((p), 6) << 8) \| \
	397	((uint64)GETBYTE((p), 7) << 0))
	398	# define LOAD64_L(p) \
	399	(((uint64)GETBYTE((p), 0) << 0) \| \
	400	((uint64)GETBYTE((p), 1) << 8) \| \
	401	((uint64)GETBYTE((p), 2) << 16) \| \
	402	((uint64)GETBYTE((p), 3) << 24) \| \
	403	((uint64)GETBYTE((p), 4) << 32) \| \
	404	((uint64)GETBYTE((p), 5) << 40) \| \
	405	((uint64)GETBYTE((p), 6) << 48) \| \
	406	((uint64)GETBYTE((p), 7) << 56))
	407	# define LOAD64(p) LOAD64_B((p))
	408	# define LOAD64_B_(d, p) ((d).i = LOAD64_B((p)))
	409	# define LOAD64_L_(d, p) ((d).i = LOAD64_L((p)))
	410	# define LOAD64_(d, p) LOAD64_B_((d), (p))
	411
	412	# define STORE64_B(p, v) \
	413	(PUTBYTE((p), 0, (uint64)(v) >> 56), \
	414	PUTBYTE((p), 1, (uint64)(v) >> 48), \
	415	PUTBYTE((p), 2, (uint64)(v) >> 40), \
	416	PUTBYTE((p), 3, (uint64)(v) >> 32), \
	417	PUTBYTE((p), 4, (uint64)(v) >> 24), \
	418	PUTBYTE((p), 5, (uint64)(v) >> 16), \
	419	PUTBYTE((p), 6, (uint64)(v) >> 8), \
	420	PUTBYTE((p), 7, (uint64)(v) >> 0))
	421	# define STORE64_L(p, v) \
	422	(PUTBYTE((p), 0, (uint64)(v) >> 0), \
	423	PUTBYTE((p), 1, (uint64)(v) >> 8), \
	424	PUTBYTE((p), 2, (uint64)(v) >> 16), \
	425	PUTBYTE((p), 3, (uint64)(v) >> 24), \
	426	PUTBYTE((p), 4, (uint64)(v) >> 32), \
	427	PUTBYTE((p), 5, (uint64)(v) >> 40), \
	428	PUTBYTE((p), 6, (uint64)(v) >> 48), \
	429	PUTBYTE((p), 7, (uint64)(v) >> 56))
	430	# define STORE64(p, v) STORE64_B((p), (v))
	431	# define STORE64_B_(p, v) STORE64_B((p), (v).i)
	432	# define STORE64_L_(p, v) STORE64_L((p), (v).i)
	433	# define STORE64_(p, v) STORE64_B_((p), (v))
	434
	435	#else
	436
	437	# define LOAD64_B_(d, p) \
	438	((d).hi = LOAD32_B((octet *)(p) + 0), \
	439	(d).lo = LOAD32_B((octet *)(p) + 4))
	440	# define LOAD64_L_(d, p) \
	441	((d).lo = LOAD32_L((octet *)(p) + 0), \
	442	(d).hi = LOAD32_L((octet *)(p) + 4))
	443	# define LOAD64_(d, p) LOAD64_B_((d), (p))
	444
	445	# define STORE64_B_(p, v) \
	446	(STORE32_B((octet *)(p) + 0, (v).hi), \
	447	STORE32_B((octet *)(p) + 4, (v).lo))
	448	# define STORE64_L_(p, v) \
	449	(STORE32_L((octet *)(p) + 0, (v).lo), \
	450	STORE32_L((octet *)(p) + 4, (v).hi))
451	# define STORE64_(p, v) STORE64_B_((p), (v))
452
453	#endif
454
455	/* --- Other operations on 64-bit integers --- */
456
457	#ifdef HAVE_UINT64
458	# define SET64(d, h, l) ((d).i = (U64((h)) << 32) \| U64((l)))
459	# define ASSIGN64(d, x) ((d).i = U64((x)))
460	# define HI64(x) U32((x).i >> 32)
461	# define LO64(x) U32((x).i)
3b7cea93	462	# define GET64(t, x) ((t)(x).i)
a6f4a484	463	#else
	464	# define SET64(d, h, l) ((d).hi = U32(h), (d).lo = U32(l))
	465	# define ASSIGN64(d, x) \
	466	((d).hi = ((x & ~MASK32) >> 16) >> 16, (d).lo = U32(x))
	467	# define HI64(x) U32((x).hi)
	468	# define LO64(x) U32((x).lo)
3b7cea93	469	# define GET64(t, x) (((((t)HI64(x) << 16) << 16) & ~MASK32) \| (t)LO64(x))
a6f4a484	470	#endif
	471
	472	#ifdef HAVE_UINT64
	473	# define AND64(d, x, y) ((d).i = (x).i & (y).i)
	474	# define OR64(d, x, y) ((d).i = (x).i \| (y).i)
	475	# define XOR64(d, x, y) ((d).i = (x).i ^ (y).i)
	476	# define CPL64(d, x) ((d).i = ~(x).i)
	477	# define ADD64(d, x, y) ((d).i = (x).i + (y).i)
	478	# define SUB64(d, x, y) ((d).i = (x).i - (y).i)
	479	# define CMP64(x, op, y) ((x).i op (y).i)
	480	# define ZERO64(x) ((x) == 0)
	481	#else
	482	# define AND64(d, x, y) ((d).lo = (x).lo & (y).lo, (d).hi = (x).hi & (y).hi)
	483	# define OR64(d, x, y) ((d).lo = (x).lo \| (y).lo, (d).hi = (x).hi \| (y).hi)
	484	# define XOR64(d, x, y) ((d).lo = (x).lo ^ (y).lo, (d).hi = (x).hi ^ (y).hi)
	485	# define CPL64(d, x) ((d).lo = ~(x).lo, (d).hi = ~(x).hi)
	486	# define ADD64(d, x, y) do { \
	487	uint32 _x = U32((x).lo + (y).lo); \
	488	(d).hi = (x).hi + (y).hi + (_x < (x).lo); \
	489	(d).lo = _x; \
	490	} while (0)
	491	# define SUB64(d, x, y) do { \
	492	uint32 _x = U32((x).lo - (y).lo); \
	493	(d).hi = (x).hi - (y).hi - (_x > (x).lo); \
	494	(d).lo = _x; \
	495	} while (0)
	496	# define CMP64(x, op, y) \
	497	((x).hi == (y).hi ? (x).lo op (y).lo : (x).hi op (y).hi)
	498	# define ZERO64(x) ((x).lo == 0 && (x).hi == 0)
	499	#endif
	500
91928c79	501	/* --- Storing integers in tables --- */
	502
	503	#ifdef HAVE_UINT64
	504	# define X64(x, y) { 0x##x##y }
	505	#else
	506	# define X64(x, y) { 0x##x, 0x##y }
	507	#endif
d4efbcd9	508
6b033bc4	509	/----- That's all, folks -------------------------------------------------/
	510
	511	#ifdef __cplusplus
	512	}
	513	#endif
	514
	515	#endif