3 * $Id: bits.h,v 1.10 2004/04/08 01:36:11 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,
37 /*----- Header files ------------------------------------------------------*/
41 #if __STDC_VERSION__ >= 199900l
45 /*----- Decide on some types ----------------------------------------------*/
47 /* --- Make GNU C shut up --- */
49 #if __GNUC__ > 2 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 91)
50 # define MLIB_BITS_EXTENSION __extension__
52 # define MLIB_BITS_EXTENSION
55 /* --- Decide on a 32-bit type --- *
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.
63 #if UINT_MAX >= 0xffffffffu
64 typedef unsigned int uint32;
66 typedef unsigned long uint32;
69 /* --- Decide on a 64-bit type --- *
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@
76 #if defined(ULONG_LONG_MAX) && !defined(ULLONG_MAX)
77 # define ULLONG_MAX ULONG_LONG_MAX
80 #if UINT_MAX >> 31 > 0xffffffff
82 typedef unsigned int uint64;
83 #elif ULONG_MAX >> 31 > 0xffffffff
85 typedef unsigned long uint64;
86 #elif defined(ULLONG_MAX)
88 MLIB_BITS_EXTENSION typedef unsigned long long uint64;
96 typedef struct { uint64 i; } kludge64;
98 typedef struct { uint32 hi, lo; } kludge64;
101 /* --- Decide on a 24-bit type --- */
103 #if UINT_MAX >= 0x00ffffffu
104 typedef unsigned int uint24;
106 typedef unsigned long uint24;
109 /* --- Decide on 16-bit and 8-bit types --- *
111 * This is more for brevity than anything else.
114 typedef unsigned short uint16;
115 typedef unsigned char octet;
117 /* --- WARNING! --- *
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.
123 /*----- Macros ------------------------------------------------------------*/
125 /* --- Useful masks --- */
128 #define MASK16 0xffffu
129 #define MASK24 0xffffffu
130 #define MASK32 0xffffffffu
133 # define MASK64 MLIB_BITS_EXTENSION 0xffffffffffffffffu
136 /* --- Type coercions --- */
138 #define U8(x) ((octet)((x) & MASK8))
139 #define U16(x) ((uint16)((x) & MASK16))
140 #define U24(x) ((uint24)((x) & MASK24))
141 #define U32(x) ((uint32)((x) & MASK32))
144 # define U64(x) ((uint64)(x) & MASK64)
145 # define U64_(d, x) ((d).i = U64(x).i)
147 # define U64_(d, x) ((d).hi = U32((x).hi), (d).lo = U32((x).lo))
150 /* --- Safe shifting macros --- */
152 #define LSL8(v, s) U8(U8(v) << ((s) & 7u))
153 #define LSR8(v, s) U8(U8(v) >> ((s) & 7u))
154 #define LSL16(v, s) U16(U16(v) << ((s) & 15u))
155 #define LSR16(v, s) U16(U16(v) >> ((s) & 15u))
156 #define LSL24(v, s) U24(U24(v) << ((s) % 24u))
157 #define LSR24(v, s) U24(U24(v) >> ((s) % 24u))
158 #define LSL32(v, s) U32(U32(v) << ((s) & 31u))
159 #define LSR32(v, s) U32(U32(v) >> ((s) & 31u))
162 # define LSL64(v, s) U64(U64(v) << ((s) & 63u))
163 # define LSR64(v, s) U64(U64(v) >> ((s) & 63u))
164 # define LSL64_(d, v, s) ((d).i = LSL64((v).i, (s)))
165 # define LSR64_(d, v, s) ((d).i = LSR64((v).i, (s)))
167 # define LSL64_(d, v, s) do { \
168 unsigned _s = (s) & 63u; \
169 uint32 _l = (v).lo, _h = (v).hi; \
170 kludge64 *_d = &(d); \
172 _d->hi = LSL32(_l, _s - 32u); \
178 _d->hi = LSL32(_h, _s) | LSR32(_l, 32u - _s); \
179 _d->lo = LSL32(_l, _s); \
182 # define LSR64_(d, v, s) do { \
183 unsigned _s = (s) & 63u; \
184 uint32 _l = (v).lo, _h = (v).hi; \
185 kludge64 *_d = &(d); \
187 _d->lo = LSR32(_h, _s - 32u); \
193 _d->lo = LSR32(_l, _s) | LSL32(_h, 32u - _s); \
194 _d->hi = LSR32(_h, _s); \
199 /* --- Rotation macros --- */
201 #define ROL8(v, s) (LSL8((v), (s)) | (LSR8((v), 8u - (s))))
202 #define ROR8(v, s) (LSR8((v), (s)) | (LSL8((v), 8u - (s))))
203 #define ROL16(v, s) (LSL16((v), (s)) | (LSR16((v), 16u - (s))))
204 #define ROR16(v, s) (LSR16((v), (s)) | (LSL16((v), 16u - (s))))
205 #define ROL24(v, s) (LSL24((v), (s)) | (LSR24((v), 24u - (s))))
206 #define ROR24(v, s) (LSR24((v), (s)) | (LSL24((v), 24u - (s))))
207 #define ROL32(v, s) (LSL32((v), (s)) | (LSR32((v), 32u - (s))))
208 #define ROR32(v, s) (LSR32((v), (s)) | (LSL32((v), 32u - (s))))
211 # define ROL64(v, s) (LSL64((v), (s)) | (LSR64((v), 64u - (s))))
212 # define ROR64(v, s) (LSR64((v), (s)) | (LSL64((v), 64u - (s))))
213 # define ROL64_(d, v, s) ((d).i = ROL64((v).i, (s)))
214 # define ROR64_(d, v, s) ((d).i = ROR64((v).i, (s)))
216 # define ROL64_(d, v, s) do { \
217 unsigned _s = (s) & 63u; \
218 uint32 _l = (v).lo, _h = (v).hi; \
219 kludge64 *_d = &(d); \
221 _d->hi = LSL32(_l, _s - 32u) | LSR32(_h, 64u - _s); \
222 _d->lo = LSL32(_h, _s - 32u) | LSR32(_l, 64u - _s); \
227 _d->hi = LSL32(_h, _s) | LSR32(_l, 32u - _s); \
228 _d->lo = LSL32(_l, _s) | LSR32(_h, 32u - _s); \
231 # define ROR64_(d, v, s) do { \
232 unsigned _s = (s) & 63u; \
233 uint32 _l = (v).lo, _h = (v).hi; \
234 kludge64 *_d = &(d); \
236 _d->hi = LSR32(_l, _s - 32u) | LSL32(_h, 64u - _s); \
237 _d->lo = LSR32(_h, _s - 32u) | LSL32(_l, 64u - _s); \
242 _d->hi = LSR32(_h, _s) | LSL32(_l, 32u - _s); \
243 _d->lo = LSR32(_l, _s) | LSL32(_h, 32u - _s); \
248 /* --- Storage and retrieval --- */
250 #define GETBYTE(p, o) (((octet *)(p))[o] & MASK8)
251 #define PUTBYTE(p, o, v) (((octet *)(p))[o] = U8((v)))
253 #define LOAD8(p) (GETBYTE((p), 0))
254 #define STORE8(p, v) (PUTBYTE((p), 0, (v)))
256 #define LOAD16_B(p) \
257 (((uint16)GETBYTE((p), 0) << 8) | \
258 ((uint16)GETBYTE((p), 1) << 0))
259 #define LOAD16_L(p) \
260 (((uint16)GETBYTE((p), 0) << 0) | \
261 ((uint16)GETBYTE((p), 1) << 8))
262 #define LOAD16(p) LOAD16_B((p))
264 #define STORE16_B(p, v) \
265 (PUTBYTE((p), 0, (uint16)(v) >> 8), \
266 PUTBYTE((p), 1, (uint16)(v) >> 0))
267 #define STORE16_L(p, v) \
268 (PUTBYTE((p), 0, (uint16)(v) >> 0), \
269 PUTBYTE((p), 1, (uint16)(v) >> 8))
270 #define STORE16(p, v) STORE16_B((p), (v))
272 #define LOAD24_B(p) \
273 (((uint24)GETBYTE((p), 0) << 16) | \
274 ((uint24)GETBYTE((p), 1) << 8) | \
275 ((uint24)GETBYTE((p), 2) << 0))
276 #define LOAD24_L(p) \
277 (((uint24)GETBYTE((p), 0) << 0) | \
278 ((uint24)GETBYTE((p), 1) << 8) | \
279 ((uint24)GETBYTE((p), 2) << 16))
280 #define LOAD24(p) LOAD24_B((p))
282 #define STORE24_B(p, v) \
283 (PUTBYTE((p), 0, (uint24)(v) >> 16), \
284 PUTBYTE((p), 1, (uint24)(v) >> 8), \
285 PUTBYTE((p), 2, (uint24)(v) >> 0))
286 #define STORE24_L(p, v) \
287 (PUTBYTE((p), 0, (uint24)(v) >> 0), \
288 PUTBYTE((p), 1, (uint24)(v) >> 8), \
289 PUTBYTE((p), 2, (uint24)(v) >> 16))
290 #define STORE24(p, v) STORE24_B((p), (v))
292 #define LOAD32_B(p) \
293 (((uint32)GETBYTE((p), 0) << 24) | \
294 ((uint32)GETBYTE((p), 1) << 16) | \
295 ((uint32)GETBYTE((p), 2) << 8) | \
296 ((uint32)GETBYTE((p), 3) << 0))
297 #define LOAD32_L(p) \
298 (((uint32)GETBYTE((p), 0) << 0) | \
299 ((uint32)GETBYTE((p), 1) << 8) | \
300 ((uint32)GETBYTE((p), 2) << 16) | \
301 ((uint32)GETBYTE((p), 3) << 24))
302 #define LOAD32(p) LOAD32_B((p))
304 #define STORE32_B(p, v) \
305 (PUTBYTE((p), 0, (uint32)(v) >> 24), \
306 PUTBYTE((p), 1, (uint32)(v) >> 16), \
307 PUTBYTE((p), 2, (uint32)(v) >> 8), \
308 PUTBYTE((p), 3, (uint32)(v) >> 0))
309 #define STORE32_L(p, v) \
310 (PUTBYTE((p), 0, (uint32)(v) >> 0), \
311 PUTBYTE((p), 1, (uint32)(v) >> 8), \
312 PUTBYTE((p), 2, (uint32)(v) >> 16), \
313 PUTBYTE((p), 3, (uint32)(v) >> 24))
314 #define STORE32(p, v) STORE32_B((p), (v))
318 # define LOAD64_B(p) \
319 (((uint64)GETBYTE((p), 0) << 56) | \
320 ((uint64)GETBYTE((p), 1) << 48) | \
321 ((uint64)GETBYTE((p), 2) << 40) | \
322 ((uint64)GETBYTE((p), 3) << 32) | \
323 ((uint64)GETBYTE((p), 4) << 24) | \
324 ((uint64)GETBYTE((p), 5) << 16) | \
325 ((uint64)GETBYTE((p), 6) << 8) | \
326 ((uint64)GETBYTE((p), 7) << 0))
327 # define LOAD64_L(p) \
328 (((uint64)GETBYTE((p), 0) << 0) | \
329 ((uint64)GETBYTE((p), 1) << 8) | \
330 ((uint64)GETBYTE((p), 2) << 16) | \
331 ((uint64)GETBYTE((p), 3) << 24) | \
332 ((uint64)GETBYTE((p), 4) << 32) | \
333 ((uint64)GETBYTE((p), 5) << 40) | \
334 ((uint64)GETBYTE((p), 6) << 48) | \
335 ((uint64)GETBYTE((p), 7) << 56))
336 # define LOAD64(p) LOAD64_B((p))
337 # define LOAD64_B_(d, p) ((d).i = LOAD64_B((p)))
338 # define LOAD64_L_(d, p) ((d).i = LOAD64_L((p)))
339 # define LOAD64_(d, p) LOAD64_B_((d), (p))
341 # define STORE64_B(p, v) \
342 (PUTBYTE((p), 0, (uint64)(v) >> 56), \
343 PUTBYTE((p), 1, (uint64)(v) >> 48), \
344 PUTBYTE((p), 2, (uint64)(v) >> 40), \
345 PUTBYTE((p), 3, (uint64)(v) >> 32), \
346 PUTBYTE((p), 4, (uint64)(v) >> 24), \
347 PUTBYTE((p), 5, (uint64)(v) >> 16), \
348 PUTBYTE((p), 6, (uint64)(v) >> 8), \
349 PUTBYTE((p), 7, (uint64)(v) >> 0))
350 # define STORE64_L(p, v) \
351 (PUTBYTE((p), 0, (uint64)(v) >> 0), \
352 PUTBYTE((p), 1, (uint64)(v) >> 8), \
353 PUTBYTE((p), 2, (uint64)(v) >> 16), \
354 PUTBYTE((p), 3, (uint64)(v) >> 24), \
355 PUTBYTE((p), 4, (uint64)(v) >> 32), \
356 PUTBYTE((p), 5, (uint64)(v) >> 40), \
357 PUTBYTE((p), 6, (uint64)(v) >> 48), \
358 PUTBYTE((p), 7, (uint64)(v) >> 56))
359 # define STORE64(p, v) STORE64_B((p), (v))
360 # define STORE64_B_(p, v) STORE64_B((p), (v).i)
361 # define STORE64_L_(p, v) STORE64_L((p), (v).i)
362 # define STORE64_(p, v) STORE64_B_((p), (v))
366 # define LOAD64_B_(d, p) \
367 ((d).hi = LOAD32_B((octet *)(p) + 0), \
368 (d).lo = LOAD32_B((octet *)(p) + 4))
369 # define LOAD64_L_(d, p) \
370 ((d).lo = LOAD32_L((octet *)(p) + 0), \
371 (d).hi = LOAD32_L((octet *)(p) + 4))
372 # define LOAD64_(d, p) LOAD64_B_((d), (p))
374 # define STORE64_B_(p, v) \
375 (STORE32_B((octet *)(p) + 0, (v).hi), \
376 STORE32_B((octet *)(p) + 4, (v).lo))
377 # define STORE64_L_(p, v) \
378 (STORE32_L((octet *)(p) + 0, (v).lo), \
379 STORE32_L((octet *)(p) + 4, (v).hi))
380 # define STORE64_(p, v) STORE64_B_((p), (v))
384 /* --- Other operations on 64-bit integers --- */
387 # define SET64(d, h, l) ((d).i = (U64((h)) << 32) | U64((l)))
388 # define ASSIGN64(d, x) ((d).i = U64((x)))
389 # define HI64(x) U32((x).i >> 32)
390 # define LO64(x) U32((x).i)
391 # define GET64(t, x) ((t)(x).i)
393 # define SET64(d, h, l) ((d).hi = U32(h), (d).lo = U32(l))
394 # define ASSIGN64(d, x) \
395 ((d).hi = ((x & ~MASK32) >> 16) >> 16, (d).lo = U32(x))
396 # define HI64(x) U32((x).hi)
397 # define LO64(x) U32((x).lo)
398 # define GET64(t, x) (((((t)HI64(x) << 16) << 16) & ~MASK32) | (t)LO64(x))
402 # define AND64(d, x, y) ((d).i = (x).i & (y).i)
403 # define OR64(d, x, y) ((d).i = (x).i | (y).i)
404 # define XOR64(d, x, y) ((d).i = (x).i ^ (y).i)
405 # define CPL64(d, x) ((d).i = ~(x).i)
406 # define ADD64(d, x, y) ((d).i = (x).i + (y).i)
407 # define SUB64(d, x, y) ((d).i = (x).i - (y).i)
408 # define CMP64(x, op, y) ((x).i op (y).i)
409 # define ZERO64(x) ((x) == 0)
411 # define AND64(d, x, y) ((d).lo = (x).lo & (y).lo, (d).hi = (x).hi & (y).hi)
412 # define OR64(d, x, y) ((d).lo = (x).lo | (y).lo, (d).hi = (x).hi | (y).hi)
413 # define XOR64(d, x, y) ((d).lo = (x).lo ^ (y).lo, (d).hi = (x).hi ^ (y).hi)
414 # define CPL64(d, x) ((d).lo = ~(x).lo, (d).hi = ~(x).hi)
415 # define ADD64(d, x, y) do { \
416 uint32 _x = U32((x).lo + (y).lo); \
417 (d).hi = (x).hi + (y).hi + (_x < (x).lo); \
420 # define SUB64(d, x, y) do { \
421 uint32 _x = U32((x).lo - (y).lo); \
422 (d).hi = (x).hi - (y).hi - (_x > (x).lo); \
425 # define CMP64(x, op, y) \
426 ((x).hi == (y).hi ? (x).lo op (y).lo : (x).hi op (y).hi)
427 # define ZERO64(x) ((x).lo == 0 && (x).hi == 0)
430 /* --- Storing integers in tables --- */
433 # define X64(x, y) { 0x##x##y }
435 # define X64(x, y) { 0x##x, 0x##y }
438 /*----- That's all, folks -------------------------------------------------*/