3 * $Id: bits.h,v 1.7 2000/07/22 09:48:26 mdw Exp $
5 * Portable bit-level manipulation macros
7 * (c) 1998 Straylight/Edgeware
10 /*----- Licensing notice --------------------------------------------------*
12 * This file is part of the mLib utilities library.
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.
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.
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,
30 /*----- Revision history --------------------------------------------------*
33 * Revision 1.7 2000/07/22 09:48:26 mdw
34 * Added macros for reading 64-bit values.
36 * Revision 1.6 2000/07/16 12:28:28 mdw
37 * Add 64-bit support, with faked arithmetic on 32-bit hosts.
39 * Revision 1.5 2000/06/17 10:36:06 mdw
40 * Support for 24-bit types.
42 * Revision 1.4 1999/12/10 23:42:04 mdw
43 * Change header file guard names.
45 * Revision 1.3 1999/06/20 23:31:52 mdw
46 * More portability enhancements.
48 * Revision 1.2 1999/06/17 00:12:46 mdw
49 * Improve portability for shift and rotate macros.
51 * Revision 1.1 1999/06/01 09:46:19 mdw
52 * New addition: bit manipulation macros.
63 /*----- Header files ------------------------------------------------------*/
67 #if __STDC_VERSION__ >= 199900l
71 /*----- Decide on some types ----------------------------------------------*/
73 /* --- Decide on a 32-bit type --- *
75 * I want a type which is capable of expressing 32-bit numbers. Because some
76 * implementations have 64-bit @long@s (infinitely preferable to the abortion
77 * that is @long long@), using @unsigned long@ regardless is wasteful. So,
78 * if @int@ appears to be good enough, then I'll go with that.
81 #if UINT_MAX >= 0xffffffffu
82 typedef unsigned int uint32
;
84 typedef unsigned long uint32
;
87 /* --- Decide on a 64-bit type --- *
89 * The test is quite subtle. Think about it. Note that (at least on my
90 * machine), the 32-bit macros are *much* faster than GCC's @long long@
94 #if defined(ULONG_LONG_MAX) && !defined(ULLONG_MAX)
95 # define ULLONG_MAX ULONG_LONG_MAX
98 #if UINT_MAX >> 31 > 0xffffffff
100 typedef unsigned int uint64
;
101 #elif ULONG_MAX >> 31 > 0xffffffff
103 typedef unsigned long uint64
;
104 #elif defined(ULLONG_MAX)
106 typedef unsigned long long uint64
;
114 typedef struct { uint64 i
; } kludge64
;
116 typedef struct { uint32 hi
, lo
; } kludge64
;
119 /* --- Decide on a 24-bit type --- */
121 #if UINT_MAX >= 0x00ffffffu
122 typedef unsigned int uint24
;
124 typedef unsigned long uint24
;
127 /* --- Decide on 16-bit and 8-bit types --- *
129 * This is more for brevity than anything else.
132 typedef unsigned short uint16
;
133 typedef unsigned char octet
;
135 /* --- WARNING! --- *
137 * Never lose sight of the fact that the above types may be wider than the
138 * names suggest. Some architectures have 32-bit @short@s for example.
141 /*----- Macros ------------------------------------------------------------*/
143 /* --- Useful masks --- */
146 #define MASK16 0xffffu
147 #define MASK24 0xffffffu
148 #define MASK32 0xffffffffu
151 # define MASK64 0xffffffffffffffffull
154 /* --- Type coercions --- */
156 #define U8(x) ((octet)((x) & MASK8))
157 #define U16(x) ((uint16)((x) & MASK16))
158 #define U24(x) ((uint24)((x) & MASK24))
159 #define U32(x) ((uint32)((x) & MASK32))
162 # define U64(x) ((uint64)(x) & MASK64)
163 # define U64_(d, x) ((d).i = U64(x).i)
165 # define U64_(d, x) ((d).hi = U32((x).hi), (d).lo = U32((x).lo))
168 /* --- Safe shifting macros --- */
170 #define LSL8(v, s) U8(U8(v) << ((s) & 7u))
171 #define LSR8(v, s) U8(U8(v) >> ((s) & 7u))
172 #define LSL16(v, s) U16(U16(v) << ((s) & 15u))
173 #define LSR16(v, s) U16(U16(v) >> ((s) & 15u))
174 #define LSL24(v, s) U24(U24(v) << ((s) % 24u))
175 #define LSR24(v, s) U24(U24(v) >> ((s) % 24u))
176 #define LSL32(v, s) U32(U32(v) << ((s) & 31u))
177 #define LSR32(v, s) U32(U32(v) >> ((s) & 31u))
180 # define LSL64(v, s) U64(U64(v) << ((s) & 63u))
181 # define LSR64(v, s) U64(U64(v) >> ((s) & 63u))
182 # define LSL64_(d, v, s) ((d).i = LSL64((v).i, (s)))
183 # define LSR64_(d, v, s) ((d).i = LSR64((v).i, (s)))
185 # define LSL64_(d, v, s) do { \
186 unsigned _s = (s) & 63u; \
187 uint32 _l = (v).lo, _h = (v).hi; \
188 kludge64 *_d = &(d); \
190 _d->hi = LSL32(_l, _s - 32u); \
196 _d->hi = LSL32(_h, _s) | LSR32(_l, 32u - _s); \
197 _d->lo = LSL32(_l, _s); \
200 # define LSR64_(d, v, s) do { \
201 unsigned _s = (s) & 63u; \
202 uint32 _l = (v).lo, _h = (v).hi; \
203 kludge64 *_d = &(d); \
205 _d->lo = LSR32(_h, _s - 32u); \
211 _d->lo = LSR32(_l, _s) | LSL32(_h, 32u - _s); \
212 _d->hi = LSR32(_h, _s); \
217 /* --- Rotation macros --- */
219 #define ROL8(v, s) (LSL8((v), (s)) | (LSR8((v), 8u - (s))))
220 #define ROR8(v, s) (LSR8((v), (s)) | (LSL8((v), 8u - (s))))
221 #define ROL16(v, s) (LSL16((v), (s)) | (LSR16((v), 16u - (s))))
222 #define ROR16(v, s) (LSR16((v), (s)) | (LSL16((v), 16u - (s))))
223 #define ROL24(v, s) (LSL24((v), (s)) | (LSR24((v), 24u - (s))))
224 #define ROR24(v, s) (LSR24((v), (s)) | (LSL24((v), 24u - (s))))
225 #define ROL32(v, s) (LSL32((v), (s)) | (LSR32((v), 32u - (s))))
226 #define ROR32(v, s) (LSR32((v), (s)) | (LSL32((v), 32u - (s))))
229 # define ROL64(v, s) (LSL64((v), (s)) | (LSR64((v), 64u - (s))))
230 # define ROR64(v, s) (LSR64((v), (s)) | (LSL64((v), 64u - (s))))
231 # define ROL64_(d, v, s) ((d).i = ROL64((v).i, (s)))
232 # define ROR64_(d, v, s) ((d).i = ROR64((v).i, (s)))
234 # define ROL64_(d, v, s) do { \
235 unsigned _s = (s) & 63u; \
236 uint32 _l = (v).lo, _h = (v).hi; \
237 kludge64 *_d = &(d); \
239 _d->hi = LSL32(_l, _s - 32u) | LSR32(_h, 64u - _s); \
240 _d->lo = LSL32(_h, _s - 32u) | LSR32(_l, 64u - _s); \
245 _d->hi = LSL32(_h, _s) | LSR32(_l, 32u - _s); \
246 _d->lo = LSL32(_l, _s) | LSR32(_h, 32u - _s); \
249 # define ROR64_(d, v, s) do { \
250 unsigned _s = (s) & 63u; \
251 uint32 _l = (v).lo, _h = (v).hi; \
252 kludge64 *_d = &(d); \
254 _d->hi = LSR32(_l, _s - 32u) | LSL32(_h, 64u - _s); \
255 _d->lo = LSR32(_h, _s - 32u) | LSL32(_l, 64u - _s); \
260 _d->hi = LSR32(_h, _s) | LSL32(_l, 32u - _s); \
261 _d->lo = LSR32(_l, _s) | LSL32(_h, 32u - _s); \
266 /* --- Storage and retrieval --- */
268 #define GETBYTE(p, o) (((octet *)(p))[o] & MASK8)
269 #define PUTBYTE(p, o, v) (((octet *)(p))[o] = U8((v)))
271 #define LOAD8(p) (GETBYTE((p), 0))
272 #define STORE8(p, v) (PUTBYTE((p), 0, (v)))
274 #define LOAD16_B(p) \
275 (((uint16)GETBYTE((p), 0) << 8) | \
276 ((uint16)GETBYTE((p), 1) << 0))
277 #define LOAD16_L(p) \
278 (((uint16)GETBYTE((p), 0) << 0) | \
279 ((uint16)GETBYTE((p), 1) << 8))
280 #define LOAD16(p) LOAD16_B((p))
282 #define STORE16_B(p, v) \
283 (PUTBYTE((p), 0, (uint16)(v) >> 8), \
284 PUTBYTE((p), 1, (uint16)(v) >> 0))
285 #define STORE16_L(p, v) \
286 (PUTBYTE((p), 0, (uint16)(v) >> 0), \
287 PUTBYTE((p), 1, (uint16)(v) >> 8))
288 #define STORE16(p, v) STORE16_B((p), (v))
290 #define LOAD24_B(p) \
291 (((uint24)GETBYTE((p), 0) << 16) | \
292 ((uint24)GETBYTE((p), 1) << 8) | \
293 ((uint24)GETBYTE((p), 2) << 0))
294 #define LOAD24_L(p) \
295 (((uint24)GETBYTE((p), 0) << 0) | \
296 ((uint24)GETBYTE((p), 1) << 8) | \
297 ((uint24)GETBYTE((p), 2) << 16))
298 #define LOAD24(p) LOAD24_B((p))
300 #define STORE24_B(p, v) \
301 (PUTBYTE((p), 0, (uint24)(v) >> 16), \
302 PUTBYTE((p), 1, (uint24)(v) >> 8), \
303 PUTBYTE((p), 2, (uint24)(v) >> 0))
304 #define STORE24_L(p, v) \
305 (PUTBYTE((p), 0, (uint24)(v) >> 0), \
306 PUTBYTE((p), 1, (uint24)(v) >> 8), \
307 PUTBYTE((p), 2, (uint24)(v) >> 16))
308 #define STORE24(p, v) STORE24_B((p), (v))
310 #define LOAD32_B(p) \
311 (((uint32)GETBYTE((p), 0) << 24) | \
312 ((uint32)GETBYTE((p), 1) << 16) | \
313 ((uint32)GETBYTE((p), 2) << 8) | \
314 ((uint32)GETBYTE((p), 3) << 0))
315 #define LOAD32_L(p) \
316 (((uint32)GETBYTE((p), 0) << 0) | \
317 ((uint32)GETBYTE((p), 1) << 8) | \
318 ((uint32)GETBYTE((p), 2) << 16) | \
319 ((uint32)GETBYTE((p), 3) << 24))
320 #define LOAD32(p) LOAD32_B((p))
322 #define STORE32_B(p, v) \
323 (PUTBYTE((p), 0, (uint32)(v) >> 24), \
324 PUTBYTE((p), 1, (uint32)(v) >> 16), \
325 PUTBYTE((p), 2, (uint32)(v) >> 8), \
326 PUTBYTE((p), 3, (uint32)(v) >> 0))
327 #define STORE32_L(p, v) \
328 (PUTBYTE((p), 0, (uint32)(v) >> 0), \
329 PUTBYTE((p), 1, (uint32)(v) >> 8), \
330 PUTBYTE((p), 2, (uint32)(v) >> 16), \
331 PUTBYTE((p), 3, (uint32)(v) >> 24))
332 #define STORE32(p, v) STORE32_B((p), (v))
336 # define LOAD64_B(p) \
337 (((uint64)GETBYTE((p), 0) << 56) | \
338 ((uint64)GETBYTE((p), 1) << 48) | \
339 ((uint64)GETBYTE((p), 2) << 40) | \
340 ((uint64)GETBYTE((p), 3) << 32) | \
341 ((uint64)GETBYTE((p), 4) << 24) | \
342 ((uint64)GETBYTE((p), 5) << 16) | \
343 ((uint64)GETBYTE((p), 6) << 8) | \
344 ((uint64)GETBYTE((p), 7) << 0))
345 # define LOAD64_L(p) \
346 (((uint64)GETBYTE((p), 0) << 0) | \
347 ((uint64)GETBYTE((p), 1) << 8) | \
348 ((uint64)GETBYTE((p), 2) << 16) | \
349 ((uint64)GETBYTE((p), 3) << 24) | \
350 ((uint64)GETBYTE((p), 4) << 32) | \
351 ((uint64)GETBYTE((p), 5) << 40) | \
352 ((uint64)GETBYTE((p), 6) << 48) | \
353 ((uint64)GETBYTE((p), 7) << 56))
354 # define LOAD64(p) LOAD64_B((p))
355 # define LOAD64_B_(d, p) ((d).i = LOAD64_B((p)))
356 # define LOAD64_L_(d, p) ((d).i = LOAD64_L((p)))
357 # define LOAD64_(d, p) LOAD64_B_((d), (p))
359 # define STORE64_B(p, v) \
360 (PUTBYTE((p), 0, (uint64)(v) >> 56), \
361 PUTBYTE((p), 1, (uint64)(v) >> 48), \
362 PUTBYTE((p), 2, (uint64)(v) >> 40), \
363 PUTBYTE((p), 3, (uint64)(v) >> 32), \
364 PUTBYTE((p), 4, (uint64)(v) >> 24), \
365 PUTBYTE((p), 5, (uint64)(v) >> 16), \
366 PUTBYTE((p), 6, (uint64)(v) >> 8), \
367 PUTBYTE((p), 7, (uint64)(v) >> 0))
368 # define STORE64_L(p, v) \
369 (PUTBYTE((p), 0, (uint64)(v) >> 0), \
370 PUTBYTE((p), 1, (uint64)(v) >> 8), \
371 PUTBYTE((p), 2, (uint64)(v) >> 16), \
372 PUTBYTE((p), 3, (uint64)(v) >> 24), \
373 PUTBYTE((p), 4, (uint64)(v) >> 32), \
374 PUTBYTE((p), 5, (uint64)(v) >> 40), \
375 PUTBYTE((p), 6, (uint64)(v) >> 48), \
376 PUTBYTE((p), 7, (uint64)(v) >> 56))
377 # define STORE64(p, v) STORE64_B((p), (v))
378 # define STORE64_B_(p, v) STORE64_B((p), (v).i)
379 # define STORE64_L_(p, v) STORE64_L((p), (v).i)
380 # define STORE64_(p, v) STORE64_B_((p), (v))
384 # define LOAD64_B_(d, p) \
385 ((d).hi = LOAD32_B((octet *)(p) + 0), \
386 (d).lo = LOAD32_B((octet *)(p) + 4))
387 # define LOAD64_L_(d, p) \
388 ((d).lo = LOAD32_L((octet *)(p) + 0), \
389 (d).hi = LOAD32_L((octet *)(p) + 4))
390 # define LOAD64_(d, p) LOAD64_B_((d), (p))
392 # define STORE64_B_(p, v) \
393 (STORE32_B((octet *)(p) + 0, (v).hi), \
394 STORE32_B((octet *)(p) + 4, (v).lo))
395 # define STORE64_L_(p, v) \
396 (STORE32_L((octet *)(p) + 0, (v).lo), \
397 STORE32_L((octet *)(p) + 4, (v).hi))
398 # define STORE64_(p, v) STORE64_B_((p), (v))
402 /* --- Other operations on 64-bit integers --- */
405 # define SET64(d, h, l) ((d).i = (U64((h)) << 32) | U64((l)))
406 # define ASSIGN64(d, x) ((d).i = U64((x)))
407 # define HI64(x) U32((x).i >> 32)
408 # define LO64(x) U32((x).i)
409 # define GET64(t, x) ((t)(x).i)
411 # define SET64(d, h, l) ((d).hi = U32(h), (d).lo = U32(l))
412 # define ASSIGN64(d, x) \
413 ((d).hi = ((x & ~MASK32) >> 16) >> 16, (d).lo = U32(x))
414 # define HI64(x) U32((x).hi)
415 # define LO64(x) U32((x).lo)
416 # define GET64(t, x) (((((t)HI64(x) << 16) << 16) & ~MASK32) | (t)LO64(x))
420 # define AND64(d, x, y) ((d).i = (x).i & (y).i)
421 # define OR64(d, x, y) ((d).i = (x).i | (y).i)
422 # define XOR64(d, x, y) ((d).i = (x).i ^ (y).i)
423 # define CPL64(d, x) ((d).i = ~(x).i)
424 # define ADD64(d, x, y) ((d).i = (x).i + (y).i)
425 # define SUB64(d, x, y) ((d).i = (x).i - (y).i)
426 # define CMP64(x, op, y) ((x).i op (y).i)
427 # define ZERO64(x) ((x) == 0)
429 # define AND64(d, x, y) ((d).lo = (x).lo & (y).lo, (d).hi = (x).hi & (y).hi)
430 # define OR64(d, x, y) ((d).lo = (x).lo | (y).lo, (d).hi = (x).hi | (y).hi)
431 # define XOR64(d, x, y) ((d).lo = (x).lo ^ (y).lo, (d).hi = (x).hi ^ (y).hi)
432 # define CPL64(d, x) ((d).lo = ~(x).lo, (d).hi = ~(x).hi)
433 # define ADD64(d, x, y) do { \
434 uint32 _x = U32((x).lo + (y).lo); \
435 (d).hi = (x).hi + (y).hi + (_x < (x).lo); \
438 # define SUB64(d, x, y) do { \
439 uint32 _x = U32((x).lo - (y).lo); \
440 (d).hi = (x).hi - (y).hi - (_x > (x).lo); \
443 # define CMP64(x, op, y) \
444 ((x).hi == (y).hi ? (x).lo op (y).lo : (x).hi op (y).hi)
445 # define ZERO64(x) ((x).lo == 0 && (x).hi == 0)
448 /*----- That's all, folks -------------------------------------------------*/