mesa/gallium: Move u_bit_scan{,64} from gallium to util.
[mesa.git] / src / mesa / main / imports.h
1 /*
2 * Mesa 3-D graphics library
3 *
4 * Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the "Software"),
8 * to deal in the Software without restriction, including without limitation
9 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
10 * and/or sell copies of the Software, and to permit persons to whom the
11 * Software is furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included
14 * in all copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
17 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
20 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
21 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
22 * OTHER DEALINGS IN THE SOFTWARE.
23 */
24
25
26 /**
27 * \file imports.h
28 * Standard C library function wrappers.
29 *
30 * This file provides wrappers for all the standard C library functions
31 * like malloc(), free(), printf(), getenv(), etc.
32 */
33
34
35 #ifndef IMPORTS_H
36 #define IMPORTS_H
37
38
39 #include <stdlib.h>
40 #include <stdarg.h>
41 #include <string.h>
42 #include "compiler.h"
43 #include "glheader.h"
44 #include "errors.h"
45 #include "util/bitscan.h"
46
47 #ifdef __cplusplus
48 extern "C" {
49 #endif
50
51
52 /**********************************************************************/
53 /** Memory macros */
54 /*@{*/
55
56 /** Allocate a structure of type \p T */
57 #define MALLOC_STRUCT(T) (struct T *) malloc(sizeof(struct T))
58 /** Allocate and zero a structure of type \p T */
59 #define CALLOC_STRUCT(T) (struct T *) calloc(1, sizeof(struct T))
60
61 /*@}*/
62
63
64 /*
65 * For GL_ARB_vertex_buffer_object we need to treat vertex array pointers
66 * as offsets into buffer stores. Since the vertex array pointer and
67 * buffer store pointer are both pointers and we need to add them, we use
68 * this macro.
69 * Both pointers/offsets are expressed in bytes.
70 */
71 #define ADD_POINTERS(A, B) ( (GLubyte *) (A) + (uintptr_t) (B) )
72
73
74 /**
75 * Sometimes we treat GLfloats as GLints. On x86 systems, moving a float
76 * as a int (thereby using integer registers instead of FP registers) is
77 * a performance win. Typically, this can be done with ordinary casts.
78 * But with gcc's -fstrict-aliasing flag (which defaults to on in gcc 3.0)
79 * these casts generate warnings.
80 * The following union typedef is used to solve that.
81 */
82 typedef union { GLfloat f; GLint i; GLuint u; } fi_type;
83
84
85
86 #if defined(_MSC_VER)
87 #define strcasecmp(s1, s2) _stricmp(s1, s2)
88 #endif
89 /*@}*/
90
91
92 /***
93 *** LOG2: Log base 2 of float
94 ***/
95 static inline GLfloat LOG2(GLfloat x)
96 {
97 #if 0
98 /* This is pretty fast, but not accurate enough (only 2 fractional bits).
99 * Based on code from http://www.stereopsis.com/log2.html
100 */
101 const GLfloat y = x * x * x * x;
102 const GLuint ix = *((GLuint *) &y);
103 const GLuint exp = (ix >> 23) & 0xFF;
104 const GLint log2 = ((GLint) exp) - 127;
105 return (GLfloat) log2 * (1.0 / 4.0); /* 4, because of x^4 above */
106 #endif
107 /* Pretty fast, and accurate.
108 * Based on code from http://www.flipcode.com/totd/
109 */
110 fi_type num;
111 GLint log_2;
112 num.f = x;
113 log_2 = ((num.i >> 23) & 255) - 128;
114 num.i &= ~(255 << 23);
115 num.i += 127 << 23;
116 num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3;
117 return num.f + log_2;
118 }
119
120
121
122 /**
123 * finite macro.
124 */
125 #if defined(_MSC_VER)
126 # define finite _finite
127 #endif
128
129
130 /***
131 *** IS_INF_OR_NAN: test if float is infinite or NaN
132 ***/
133 #if defined(isfinite)
134 #define IS_INF_OR_NAN(x) (!isfinite(x))
135 #elif defined(finite)
136 #define IS_INF_OR_NAN(x) (!finite(x))
137 #elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
138 #define IS_INF_OR_NAN(x) (!isfinite(x))
139 #else
140 #define IS_INF_OR_NAN(x) (!finite(x))
141 #endif
142
143
144 /**
145 * Convert float to int by rounding to nearest integer, away from zero.
146 */
147 static inline int IROUND(float f)
148 {
149 return (int) ((f >= 0.0F) ? (f + 0.5F) : (f - 0.5F));
150 }
151
152 /**
153 * Convert double to int by rounding to nearest integer, away from zero.
154 */
155 static inline int IROUNDD(double d)
156 {
157 return (int) ((d >= 0.0) ? (d + 0.5) : (d - 0.5));
158 }
159
160 /**
161 * Convert float to int64 by rounding to nearest integer.
162 */
163 static inline GLint64 IROUND64(float f)
164 {
165 return (GLint64) ((f >= 0.0F) ? (f + 0.5F) : (f - 0.5F));
166 }
167
168
169 /**
170 * Convert positive float to int by rounding to nearest integer.
171 */
172 static inline int IROUND_POS(float f)
173 {
174 assert(f >= 0.0F);
175 return (int) (f + 0.5F);
176 }
177
178 /** Return (as an integer) floor of float */
179 static inline int IFLOOR(float f)
180 {
181 #if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
182 /*
183 * IEEE floor for computers that round to nearest or even.
184 * 'f' must be between -4194304 and 4194303.
185 * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1",
186 * but uses some IEEE specific tricks for better speed.
187 * Contributed by Josh Vanderhoof
188 */
189 int ai, bi;
190 double af, bf;
191 af = (3 << 22) + 0.5 + (double)f;
192 bf = (3 << 22) + 0.5 - (double)f;
193 /* GCC generates an extra fstp/fld without this. */
194 __asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st");
195 __asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st");
196 return (ai - bi) >> 1;
197 #else
198 int ai, bi;
199 double af, bf;
200 fi_type u;
201 af = (3 << 22) + 0.5 + (double)f;
202 bf = (3 << 22) + 0.5 - (double)f;
203 u.f = (float) af; ai = u.i;
204 u.f = (float) bf; bi = u.i;
205 return (ai - bi) >> 1;
206 #endif
207 }
208
209
210 /**
211 * Is x a power of two?
212 */
213 static inline int
214 _mesa_is_pow_two(int x)
215 {
216 return !(x & (x - 1));
217 }
218
219 /**
220 * Round given integer to next higer power of two
221 * If X is zero result is undefined.
222 *
223 * Source for the fallback implementation is
224 * Sean Eron Anderson's webpage "Bit Twiddling Hacks"
225 * http://graphics.stanford.edu/~seander/bithacks.html
226 *
227 * When using builtin function have to do some work
228 * for case when passed values 1 to prevent hiting
229 * undefined result from __builtin_clz. Undefined
230 * results would be different depending on optimization
231 * level used for build.
232 */
233 static inline int32_t
234 _mesa_next_pow_two_32(uint32_t x)
235 {
236 #ifdef HAVE___BUILTIN_CLZ
237 uint32_t y = (x != 1);
238 return (1 + y) << ((__builtin_clz(x - y) ^ 31) );
239 #else
240 x--;
241 x |= x >> 1;
242 x |= x >> 2;
243 x |= x >> 4;
244 x |= x >> 8;
245 x |= x >> 16;
246 x++;
247 return x;
248 #endif
249 }
250
251 static inline int64_t
252 _mesa_next_pow_two_64(uint64_t x)
253 {
254 #ifdef HAVE___BUILTIN_CLZLL
255 uint64_t y = (x != 1);
256 STATIC_ASSERT(sizeof(x) == sizeof(long long));
257 return (1 + y) << ((__builtin_clzll(x - y) ^ 63));
258 #else
259 x--;
260 x |= x >> 1;
261 x |= x >> 2;
262 x |= x >> 4;
263 x |= x >> 8;
264 x |= x >> 16;
265 x |= x >> 32;
266 x++;
267 return x;
268 #endif
269 }
270
271
272 /*
273 * Returns the floor form of binary logarithm for a 32-bit integer.
274 */
275 static inline GLuint
276 _mesa_logbase2(GLuint n)
277 {
278 #ifdef HAVE___BUILTIN_CLZ
279 return (31 - __builtin_clz(n | 1));
280 #else
281 GLuint pos = 0;
282 if (n >= 1<<16) { n >>= 16; pos += 16; }
283 if (n >= 1<< 8) { n >>= 8; pos += 8; }
284 if (n >= 1<< 4) { n >>= 4; pos += 4; }
285 if (n >= 1<< 2) { n >>= 2; pos += 2; }
286 if (n >= 1<< 1) { pos += 1; }
287 return pos;
288 #endif
289 }
290
291
292 /**
293 * Return 1 if this is a little endian machine, 0 if big endian.
294 */
295 static inline GLboolean
296 _mesa_little_endian(void)
297 {
298 const GLuint ui = 1; /* intentionally not static */
299 return *((const GLubyte *) &ui);
300 }
301
302
303
304 /**********************************************************************
305 * Functions
306 */
307
308 extern void *
309 _mesa_align_malloc( size_t bytes, unsigned long alignment );
310
311 extern void *
312 _mesa_align_calloc( size_t bytes, unsigned long alignment );
313
314 extern void
315 _mesa_align_free( void *ptr );
316
317 extern void *
318 _mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize,
319 unsigned long alignment);
320
321 extern void *
322 _mesa_exec_malloc( GLuint size );
323
324 extern void
325 _mesa_exec_free( void *addr );
326
327
328 #ifdef HAVE___BUILTIN_POPCOUNT
329 #define _mesa_bitcount(i) __builtin_popcount(i)
330 #else
331 extern unsigned int
332 _mesa_bitcount(unsigned int n);
333 #endif
334
335 #ifdef HAVE___BUILTIN_POPCOUNTLL
336 #define _mesa_bitcount_64(i) __builtin_popcountll(i)
337 #else
338 extern unsigned int
339 _mesa_bitcount_64(uint64_t n);
340 #endif
341
342 /**
343 * Find the last (most significant) bit set in a word.
344 *
345 * Essentially ffs() in the reverse direction.
346 */
347 static inline unsigned int
348 _mesa_fls(unsigned int n)
349 {
350 #ifdef HAVE___BUILTIN_CLZ
351 return n == 0 ? 0 : 32 - __builtin_clz(n);
352 #else
353 unsigned int v = 1;
354
355 if (n == 0)
356 return 0;
357
358 while (n >>= 1)
359 v++;
360
361 return v;
362 #endif
363 }
364
365 /**
366 * Find the last (most significant) bit set in a uint64_t value.
367 *
368 * Essentially ffsll() in the reverse direction.
369 */
370 static inline unsigned int
371 _mesa_flsll(uint64_t n)
372 {
373 #ifdef HAVE___BUILTIN_CLZLL
374 return n == 0 ? 0 : 64 - __builtin_clzll(n);
375 #else
376 unsigned int v = 1;
377
378 if (n == 0)
379 return 0;
380
381 while (n >>= 1)
382 v++;
383
384 return v;
385 #endif
386 }
387
388 static inline bool
389 _mesa_half_is_negative(GLhalfARB h)
390 {
391 return h & 0x8000;
392 }
393
394 extern unsigned int
395 _mesa_str_checksum(const char *str);
396
397 extern int
398 _mesa_snprintf( char *str, size_t size, const char *fmt, ... ) PRINTFLIKE(3, 4);
399
400 extern int
401 _mesa_vsnprintf(char *str, size_t size, const char *fmt, va_list arg);
402
403
404 #if defined(_MSC_VER) && !defined(snprintf)
405 #define snprintf _snprintf
406 #endif
407
408 #if defined(_WIN32) && !defined(strtok_r)
409 #define strtok_r strtok_s
410 #endif
411
412 #ifdef __cplusplus
413 }
414 #endif
415
416
417 #endif /* IMPORTS_H */