846a9a0faf4fb5f7cd702ee1b43b5a11c4c64204
2 * Mesa 3-D graphics library
5 * Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
7 * Permission is hereby granted, free of charge, to any person obtaining a
8 * copy of this software and associated documentation files (the "Software"),
9 * to deal in the Software without restriction, including without limitation
10 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
11 * and/or sell copies of the Software, and to permit persons to whom the
12 * Software is furnished to do so, subject to the following conditions:
14 * The above copyright notice and this permission notice shall be included
15 * in all copies or substantial portions of the Software.
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
18 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
20 * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
21 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
22 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
28 * Standard C library function wrappers.
30 * This file provides wrappers for all the standard C library functions
31 * like malloc(), free(), printf(), getenv(), etc.
48 /**********************************************************************/
52 /** Allocate \p BYTES bytes */
53 #define MALLOC(BYTES) malloc(BYTES)
54 /** Allocate and zero \p BYTES bytes */
55 #define CALLOC(BYTES) calloc(1, BYTES)
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))
61 #define FREE(PTR) free(PTR)
67 * For GL_ARB_vertex_buffer_object we need to treat vertex array pointers
68 * as offsets into buffer stores. Since the vertex array pointer and
69 * buffer store pointer are both pointers and we need to add them, we use
71 * Both pointers/offsets are expressed in bytes.
73 #define ADD_POINTERS(A, B) ( (GLubyte *) (A) + (uintptr_t) (B) )
77 * Sometimes we treat GLfloats as GLints. On x86 systems, moving a float
78 * as a int (thereby using integer registers instead of FP registers) is
79 * a performance win. Typically, this can be done with ordinary casts.
80 * But with gcc's -fstrict-aliasing flag (which defaults to on in gcc 3.0)
81 * these casts generate warnings.
82 * The following union typedef is used to solve that.
84 typedef union { GLfloat f
; GLint i
; } fi_type
;
88 /**********************************************************************
92 #define MAX_GLUSHORT 0xffff
93 #define MAX_GLUINT 0xffffffff
95 /* Degrees to radians conversion: */
96 #define DEG2RAD (M_PI/180.0)
100 *** SQRTF: single-precision square root
102 #if 0 /* _mesa_sqrtf() not accurate enough - temporarily disabled */
103 # define SQRTF(X) _mesa_sqrtf(X)
105 # define SQRTF(X) (float) sqrt((float) (X))
110 *** INV_SQRTF: single-precision inverse square root
113 #define INV_SQRTF(X) _mesa_inv_sqrt(X)
115 #define INV_SQRTF(X) (1.0F / SQRTF(X)) /* this is faster on a P4 */
120 * \name Work-arounds for platforms that lack C99 math functions
123 #if (_XOPEN_SOURCE < 600) && !defined(_ISOC99_SOURCE) \
124 && (!defined(__STDC_VERSION__) || (__STDC_VERSION__ < 199901L))
125 #define acosf(f) ((float) acos(f))
126 #define asinf(f) ((float) asin(f))
127 #define atan2f(f) ((float) atan2(f))
128 #define atanf(f) ((float) atan(f))
129 #define cielf(f) ((float) ciel(f))
130 #define cosf(f) ((float) cos(f))
131 #define coshf(f) ((float) cosh(f))
132 #define expf(f) ((float) exp(f))
133 #define exp2f(f) ((float) exp2(f))
134 #define floorf(f) ((float) floor(f))
135 #define logf(f) ((float) log(f))
136 #define log2f(f) ((float) log2(f))
137 #define powf(f) ((float) pow(f))
138 #define sinf(f) ((float) sin(f))
139 #define sinhf(f) ((float) sinh(f))
140 #define sqrtf(f) ((float) sqrt(f))
141 #define tanf(f) ((float) tan(f))
142 #define tanhf(f) ((float) tanh(f))
143 #define truncf(f) ((float) trunc(f))
148 *** LOG2: Log base 2 of float
152 /* This is pretty fast, but not accurate enough (only 2 fractional bits).
153 * Based on code from http://www.stereopsis.com/log2.html
155 static INLINE GLfloat
LOG2(GLfloat x
)
157 const GLfloat y
= x
* x
* x
* x
;
158 const GLuint ix
= *((GLuint
*) &y
);
159 const GLuint exp
= (ix
>> 23) & 0xFF;
160 const GLint log2
= ((GLint
) exp
) - 127;
161 return (GLfloat
) log2
* (1.0 / 4.0); /* 4, because of x^4 above */
164 /* Pretty fast, and accurate.
165 * Based on code from http://www.flipcode.com/totd/
167 static INLINE GLfloat
LOG2(GLfloat val
)
172 log_2
= ((num
.i
>> 23) & 255) - 128;
173 num
.i
&= ~(255 << 23);
175 num
.f
= ((-1.0f
/3) * num
.f
+ 2) * num
.f
- 2.0f
/3;
176 return num
.f
+ log_2
;
180 * NOTE: log_base_2(x) = log(x) / log(2)
181 * NOTE: 1.442695 = 1/log(2).
183 #define LOG2(x) ((GLfloat) (log(x) * 1.442695F))
188 *** IS_INF_OR_NAN: test if float is infinite or NaN
191 static INLINE
int IS_INF_OR_NAN( float x
)
195 return !(int)((unsigned int)((tmp
.i
& 0x7fffffff)-0x7f800000) >> 31);
197 #elif defined(isfinite)
198 #define IS_INF_OR_NAN(x) (!isfinite(x))
199 #elif defined(finite)
200 #define IS_INF_OR_NAN(x) (!finite(x))
202 #define IS_INF_OR_NAN(x) (!finite(x))
203 #elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
204 #define IS_INF_OR_NAN(x) (!isfinite(x))
206 #define IS_INF_OR_NAN(x) (!finite(x))
211 *** IS_NEGATIVE: test if float is negative
213 #if defined(USE_IEEE)
214 static INLINE
int GET_FLOAT_BITS( float x
)
220 #define IS_NEGATIVE(x) (GET_FLOAT_BITS(x) < 0)
222 #define IS_NEGATIVE(x) (x < 0.0F)
227 *** DIFFERENT_SIGNS: test if two floats have opposite signs
229 #if defined(USE_IEEE)
230 #define DIFFERENT_SIGNS(x,y) ((GET_FLOAT_BITS(x) ^ GET_FLOAT_BITS(y)) & (1<<31))
232 /* Could just use (x*y<0) except for the flatshading requirements.
233 * Maybe there's a better way?
235 #define DIFFERENT_SIGNS(x,y) ((x) * (y) <= 0.0F && (x) - (y) != 0.0F)
240 *** CEILF: ceiling of float
241 *** FLOORF: floor of float
242 *** FABSF: absolute value of float
243 *** LOGF: the natural logarithm (base e) of the value
244 *** EXPF: raise e to the value
245 *** LDEXPF: multiply value by an integral power of two
246 *** FREXPF: extract mantissa and exponent from value
248 #if defined(__gnu_linux__)
250 #define CEILF(x) ceilf(x)
251 #define FLOORF(x) floorf(x)
252 #define FABSF(x) fabsf(x)
253 #define LOGF(x) logf(x)
254 #define EXPF(x) expf(x)
255 #define LDEXPF(x,y) ldexpf(x,y)
256 #define FREXPF(x,y) frexpf(x,y)
258 #define CEILF(x) ((GLfloat) ceil(x))
259 #define FLOORF(x) ((GLfloat) floor(x))
260 #define FABSF(x) ((GLfloat) fabs(x))
261 #define LOGF(x) ((GLfloat) log(x))
262 #define EXPF(x) ((GLfloat) exp(x))
263 #define LDEXPF(x,y) ((GLfloat) ldexp(x,y))
264 #define FREXPF(x,y) ((GLfloat) frexp(x,y))
269 *** IROUND: return (as an integer) float rounded to nearest integer
271 #if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
272 static INLINE
int iround(float f
)
275 __asm__ ("fistpl %0" : "=m" (r
) : "t" (f
) : "st");
278 #define IROUND(x) iround(x)
279 #elif defined(USE_X86_ASM) && defined(_MSC_VER)
280 static INLINE
int iround(float f
)
289 #define IROUND(x) iround(x)
290 #elif defined(__WATCOMC__) && defined(__386__)
291 long iround(float f
);
292 #pragma aux iround = \
294 "fistp dword ptr [esp]" \
299 #define IROUND(x) iround(x)
301 #define IROUND(f) ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
304 #define IROUND64(f) ((GLint64) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
307 *** IROUND_POS: return (as an integer) positive float rounded to nearest int
310 #define IROUND_POS(f) (assert((f) >= 0.0F), IROUND(f))
312 #define IROUND_POS(f) (IROUND(f))
317 *** IFLOOR: return (as an integer) floor of float
319 #if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
321 * IEEE floor for computers that round to nearest or even.
322 * 'f' must be between -4194304 and 4194303.
323 * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1",
324 * but uses some IEEE specific tricks for better speed.
325 * Contributed by Josh Vanderhoof
327 static INLINE
int ifloor(float f
)
331 af
= (3 << 22) + 0.5 + (double)f
;
332 bf
= (3 << 22) + 0.5 - (double)f
;
333 /* GCC generates an extra fstp/fld without this. */
334 __asm__ ("fstps %0" : "=m" (ai
) : "t" (af
) : "st");
335 __asm__ ("fstps %0" : "=m" (bi
) : "t" (bf
) : "st");
336 return (ai
- bi
) >> 1;
338 #define IFLOOR(x) ifloor(x)
339 #elif defined(USE_IEEE)
340 static INLINE
int ifloor(float f
)
346 af
= (3 << 22) + 0.5 + (double)f
;
347 bf
= (3 << 22) + 0.5 - (double)f
;
348 u
.f
= (float) af
; ai
= u
.i
;
349 u
.f
= (float) bf
; bi
= u
.i
;
350 return (ai
- bi
) >> 1;
352 #define IFLOOR(x) ifloor(x)
354 static INLINE
int ifloor(float f
)
357 return (i
> f
) ? i
- 1 : i
;
359 #define IFLOOR(x) ifloor(x)
364 *** ICEIL: return (as an integer) ceiling of float
366 #if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
368 * IEEE ceil for computers that round to nearest or even.
369 * 'f' must be between -4194304 and 4194303.
370 * This ceil operation is done by "(iround(f + .5) + iround(f - .5) + 1) >> 1",
371 * but uses some IEEE specific tricks for better speed.
372 * Contributed by Josh Vanderhoof
374 static INLINE
int iceil(float f
)
378 af
= (3 << 22) + 0.5 + (double)f
;
379 bf
= (3 << 22) + 0.5 - (double)f
;
380 /* GCC generates an extra fstp/fld without this. */
381 __asm__ ("fstps %0" : "=m" (ai
) : "t" (af
) : "st");
382 __asm__ ("fstps %0" : "=m" (bi
) : "t" (bf
) : "st");
383 return (ai
- bi
+ 1) >> 1;
385 #define ICEIL(x) iceil(x)
386 #elif defined(USE_IEEE)
387 static INLINE
int iceil(float f
)
392 af
= (3 << 22) + 0.5 + (double)f
;
393 bf
= (3 << 22) + 0.5 - (double)f
;
394 u
.f
= (float) af
; ai
= u
.i
;
395 u
.f
= (float) bf
; bi
= u
.i
;
396 return (ai
- bi
+ 1) >> 1;
398 #define ICEIL(x) iceil(x)
400 static INLINE
int iceil(float f
)
403 return (i
< f
) ? i
+ 1 : i
;
405 #define ICEIL(x) iceil(x)
410 * Is x a power of two?
413 _mesa_is_pow_two(int x
)
415 return !(x
& (x
- 1));
419 * Round given integer to next higer power of two
420 * If X is zero result is undefined.
422 * Source for the fallback implementation is
423 * Sean Eron Anderson's webpage "Bit Twiddling Hacks"
424 * http://graphics.stanford.edu/~seander/bithacks.html
426 * When using builtin function have to do some work
427 * for case when passed values 1 to prevent hiting
428 * undefined result from __builtin_clz. Undefined
429 * results would be different depending on optimization
430 * level used for build.
432 static INLINE
int32_t
433 _mesa_next_pow_two_32(uint32_t x
)
435 #if defined(__GNUC__) && \
436 ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
437 uint32_t y
= (x
!= 1);
438 return (1 + y
) << ((__builtin_clz(x
- y
) ^ 31) );
451 static INLINE
int64_t
452 _mesa_next_pow_two_64(uint64_t x
)
454 #if defined(__GNUC__) && \
455 ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
456 uint64_t y
= (x
!= 1);
457 if (sizeof(x
) == sizeof(long))
458 return (1 + y
) << ((__builtin_clzl(x
- y
) ^ 63));
460 return (1 + y
) << ((__builtin_clzll(x
- y
) ^ 63));
476 * Return 1 if this is a little endian machine, 0 if big endian.
478 static INLINE GLboolean
479 _mesa_little_endian(void)
481 const GLuint ui
= 1; /* intentionally not static */
482 return *((const GLubyte
*) &ui
);
487 /**********************************************************************
492 _mesa_align_malloc( size_t bytes
, unsigned long alignment
);
495 _mesa_align_calloc( size_t bytes
, unsigned long alignment
);
498 _mesa_align_free( void *ptr
);
501 _mesa_align_realloc(void *oldBuffer
, size_t oldSize
, size_t newSize
,
502 unsigned long alignment
);
505 _mesa_exec_malloc( GLuint size
);
508 _mesa_exec_free( void *addr
);
511 _mesa_realloc( void *oldBuffer
, size_t oldSize
, size_t newSize
);
514 _mesa_memset16( unsigned short *dst
, unsigned short val
, size_t n
);
517 _mesa_sqrtd(double x
);
520 _mesa_sqrtf(float x
);
523 _mesa_inv_sqrtf(float x
);
526 _mesa_init_sqrt_table(void);
529 _mesa_ffs(int32_t i
);
532 _mesa_ffsll(int64_t i
);
535 _mesa_bitcount(unsigned int n
);
538 _mesa_float_to_half(float f
);
541 _mesa_half_to_float(GLhalfARB h
);
545 _mesa_bsearch( const void *key
, const void *base
, size_t nmemb
, size_t size
,
546 int (*compar
)(const void *, const void *) );
549 _mesa_getenv( const char *var
);
552 _mesa_strdup( const char *s
);
555 _mesa_strtof( const char *s
, char **end
);
558 _mesa_str_checksum(const char *str
);
561 _mesa_snprintf( char *str
, size_t size
, const char *fmt
, ... );
564 _mesa_warning( __GLcontext
*gc
, const char *fmtString
, ... );
567 _mesa_problem( const __GLcontext
*ctx
, const char *fmtString
, ... );
570 _mesa_error( __GLcontext
*ctx
, GLenum error
, const char *fmtString
, ... );
573 _mesa_debug( const __GLcontext
*ctx
, const char *fmtString
, ... );
580 #endif /* IMPORTS_H */