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