src/gallium/include/pipe/p_util.h

   1 /**************************************************************************
   2  *
   3  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
   4  * 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
   8  * "Software"), to deal in the Software without restriction, including
   9  * without limitation the rights to use, copy, modify, merge, publish,
  10  * distribute, sub license, and/or sell copies of the Software, and to
  11  * permit persons to whom the Software is furnished to do so, subject to
  12  * the following conditions:
  13  *
  14  * The above copyright notice and this permission notice (including the
  15  * next paragraph) shall be included in all copies or substantial portions
  16  * of the Software.
  17  *
  18  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  19  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  20  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
  21  * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
  22  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  23  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  24  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  25  *
  26  **************************************************************************/
  27
  28 #ifndef P_UTIL_H
  29 #define P_UTIL_H
  30
  31 #include "p_compiler.h"
  32 #include "p_debug.h"
  33 #include "p_pointer.h"
  34 #include <math.h>
  35 #include <stdarg.h>
  36
  37
  38 #ifdef __cplusplus
  39 extern "C" {
  40 #endif
  41
  42
  43 #if defined(WIN32) && defined(DEBUG) /* memory debugging */
  44
  45 #include "p_debug.h"
  46
  47 #define MALLOC( _size ) \
  48    debug_malloc( __FILE__, __LINE__, __FUNCTION__, _size )
  49 #define CALLOC( _count, _size ) \
  50    debug_calloc(__FILE__, __LINE__, __FUNCTION__, _count, _size )
  51 #define FREE( _ptr ) \
  52    debug_free( __FILE__, __LINE__, __FUNCTION__,  _ptr )
  53 #define REALLOC( _ptr, _old_size, _size ) \
  54    debug_realloc( __FILE__, __LINE__, __FUNCTION__,  _ptr, _old_size, _size )
  55
  56 #else
  57
  58 #ifdef WIN32
  59
  60 void * __stdcall
  61 EngAllocMem(
  62     unsigned long Flags,
  63     unsigned long MemSize,
  64     unsigned long Tag );
  65
  66 void __stdcall
  67 EngFreeMem(
  68     void *Mem );
  69
  70 static INLINE void *
  71 MALLOC( unsigned size )
  72 {
  73 #ifdef WINCE
  74    /* TODO: Need to abstract this */
  75    return malloc( size );
  76 #else
  77    return EngAllocMem( 0, size, 'D3AG' );
  78 #endif
  79 }
  80
  81 static INLINE void *
  82 CALLOC( unsigned count, unsigned size )
  83 {
  84    void *ptr = MALLOC( count * size );
  85    if( ptr ) {
  86       memset( ptr, 0, count * size );
  87    }
  88    return ptr;
  89 }
  90
  91 static INLINE void
  92 FREE( void *ptr )
  93 {
  94    if( ptr ) {
  95 #ifdef WINCE
  96       /* TODO: Need to abstract this */
  97       free( ptr );
  98 #else
  99       EngFreeMem( ptr );
 100 #endif
 101    }
 102 }
 103
 104 static INLINE void *
 105 REALLOC( void *old_ptr, unsigned old_size, unsigned new_size )
 106 {
 107    void *new_ptr = NULL;
 108
 109    if (new_size != 0) {
 110       unsigned copy_size = old_size < new_size ? old_size : new_size;
 111       new_ptr = MALLOC( new_size );
 112       if (new_ptr && old_ptr && copy_size) {
 113          memcpy( new_ptr, old_ptr, copy_size );
 114       }
 115    }
 116
 117    FREE( old_ptr );
 118    return new_ptr;
 119 }
 120
 121 #else /* !WIN32 */
 122
 123 #define MALLOC( SIZE )  malloc( SIZE )
 124
 125 #define CALLOC( COUNT, SIZE )   calloc( COUNT, SIZE )
 126
 127 #define FREE( PTR )  free( PTR )
 128
 129 #define REALLOC( OLDPTR, OLDSIZE, NEWSIZE )  realloc( OLDPTR, NEWSIZE )
 130
 131 #endif /* !WIN32 */
 132 #endif /* !DEBUG */
 133
 134 #define MALLOC_STRUCT(T)   (struct T *) MALLOC(sizeof(struct T))
 135
 136 #define CALLOC_STRUCT(T)   (struct T *) CALLOC(1, sizeof(struct T))
 137
 138 #define GETENV( X ) debug_get_option( X, NULL )
 139
 140
 141 /**
 142  * Return memory on given byte alignment
 143  */
 144 static INLINE void *
 145 align_malloc(size_t bytes, uint alignment)
 146 {
 147 #if defined(HAVE_POSIX_MEMALIGN)
 148    void *mem;
 149    (void) posix_memalign(& mem, alignment, bytes);
 150    return mem;
 151 #else
 152    char *ptr, *buf;
 153
 154    assert( alignment > 0 );
 155
 156    ptr = (char *) MALLOC(bytes + alignment + sizeof(void *));
 157    if (!ptr)
 158       return NULL;
 159
 160    buf = (char *) align_pointer( ptr + sizeof(void *), alignment );
 161    *(char **)(buf - sizeof(void *)) = ptr;
 162
 163    return buf;
 164 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 165 }
 166
 167 /**
 168  * Free memory returned by align_malloc().
 169  */
 170 static INLINE void
 171 align_free(void *ptr)
 172 {
 173 #if defined(HAVE_POSIX_MEMALIGN)
 174    FREE(ptr);
 175 #else
 176    void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
 177    void *realAddr = *cubbyHole;
 178    FREE(realAddr);
 179 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 180 }
 181
 182
 183
 184 /**
 185  * Duplicate a block of memory.
 186  */
 187 static INLINE void *
 188 mem_dup(const void *src, uint size)
 189 {
 190    void *dup = MALLOC(size);
 191    if (dup)
 192       memcpy(dup, src, size);
 193    return dup;
 194 }
 195
 196
 197
 198 #define CLAMP( X, MIN, MAX )  ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) )
 199 #define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
 200 #define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
 201
 202 #define Elements(x) (sizeof(x)/sizeof((x)[0]))
 203 #define Offset(TYPE, MEMBER) ((unsigned)&(((TYPE *)NULL)->MEMBER))
 204
 205 /**
 206  * Return a pointer aligned to next multiple of 16 bytes.
 207  */
 208 static INLINE void *
 209 align16( void *unaligned )
 210 {
 211    return align_pointer( unaligned, 16 );
 212 }
 213
 214
 215 static INLINE int align_int(int x, int align)
 216 {
 217    return (x + align - 1) & ~(align - 1);
 218 }
 219
 220
 221
 222 #if defined(__MSC__) && defined(__WIN32__)
 223 static INLINE unsigned ffs( unsigned u )
 224 {
 225    unsigned i;
 226
 227    if( u == 0 ) {
 228       return 0;
 229    }
 230
 231    __asm bsf eax, [u]
 232    __asm inc eax
 233    __asm mov [i], eax
 234
 235    return i;
 236 }
 237 #endif
 238
 239 union fi {
 240    float f;
 241    int i;
 242    unsigned ui;
 243 };
 244
 245 #define UBYTE_TO_FLOAT( ub ) ((float)(ub) / 255.0F)
 246
 247 #define IEEE_0996 0x3f7f0000    /* 0.996 or so */
 248
 249 /* This function/macro is sensitive to precision.  Test very carefully
 250  * if you change it!
 251  */
 252 #define UNCLAMPED_FLOAT_TO_UBYTE(UB, F)                                 \
 253         do {                                                            \
 254            union fi __tmp;                                              \
 255            __tmp.f = (F);                                               \
 256            if (__tmp.i < 0)                                             \
 257               UB = (ubyte) 0;                                           \
 258            else if (__tmp.i >= IEEE_0996)                               \
 259               UB = (ubyte) 255;                                 \
 260            else {                                                       \
 261               __tmp.f = __tmp.f * (255.0f/256.0f) + 32768.0f;           \
 262               UB = (ubyte) __tmp.i;                                     \
 263            }                                                            \
 264         } while (0)
 265
 266
 267
 268 static INLINE unsigned pack_ub4( unsigned char b0,
 269                                  unsigned char b1,
 270                                  unsigned char b2,
 271                                  unsigned char b3 )
 272 {
 273    return ((((unsigned int)b0) << 0) |
 274            (((unsigned int)b1) << 8) |
 275            (((unsigned int)b2) << 16) |
 276            (((unsigned int)b3) << 24));
 277 }
 278
 279 static INLINE unsigned fui( float f )
 280 {
 281    union fi fi;
 282    fi.f = f;
 283    return fi.ui;
 284 }
 285
 286 static INLINE unsigned char float_to_ubyte( float f )
 287 {
 288    unsigned char ub;
 289    UNCLAMPED_FLOAT_TO_UBYTE(ub, f);
 290    return ub;
 291 }
 292
 293 static INLINE unsigned pack_ui32_float4( float a,
 294                                          float b,
 295                                          float c,
 296                                          float d )
 297 {
 298    return pack_ub4( float_to_ubyte(a),
 299                     float_to_ubyte(b),
 300                     float_to_ubyte(c),
 301                     float_to_ubyte(d) );
 302 }
 303
 304 #define COPY_4V( DST, SRC )         \
 305 do {                                \
 306    (DST)[0] = (SRC)[0];             \
 307    (DST)[1] = (SRC)[1];             \
 308    (DST)[2] = (SRC)[2];             \
 309    (DST)[3] = (SRC)[3];             \
 310 } while (0)
 311
 312
 313 #define COPY_4FV( DST, SRC )  COPY_4V(DST, SRC)
 314
 315
 316 #define ASSIGN_4V( DST, V0, V1, V2, V3 ) \
 317 do {                                     \
 318    (DST)[0] = (V0);                      \
 319    (DST)[1] = (V1);                      \
 320    (DST)[2] = (V2);                      \
 321    (DST)[3] = (V3);                      \
 322 } while (0)
 323
 324
 325 static INLINE int ifloor(float f)
 326 {
 327    int ai, bi;
 328    double af, bf;
 329    union fi u;
 330
 331    af = (3 << 22) + 0.5 + (double)f;
 332    bf = (3 << 22) + 0.5 - (double)f;
 333    u.f = (float) af;  ai = u.i;
 334    u.f = (float) bf;  bi = u.i;
 335    return (ai - bi) >> 1;
 336 }
 337
 338
 339 #if defined(__GNUC__) && defined(__i386__)
 340 static INLINE int iround(float f)
 341 {
 342    int r;
 343    __asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st");
 344    return r;
 345 }
 346 #elif defined(__MSC__) && defined(__WIN32__)
 347 static INLINE int iround(float f)
 348 {
 349    int r;
 350    _asm {
 351          fld f
 352          fistp r
 353         }
 354    return r;
 355 }
 356 #else
 357 #define IROUND(f)  ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
 358 #endif
 359
 360
 361 /* Could maybe have an inline version of this?
 362  */
 363 #if defined(__GNUC__)
 364 #define FABSF(x)   fabsf(x)
 365 #else
 366 #define FABSF(x)   ((float) fabs(x))
 367 #endif
 368
 369 /* Pretty fast, and accurate.
 370  * Based on code from http://www.flipcode.com/totd/
 371  */
 372 static INLINE float LOG2(float val)
 373 {
 374    union fi num;
 375    int log_2;
 376
 377    num.f = val;
 378    log_2 = ((num.i >> 23) & 255) - 128;
 379    num.i &= ~(255 << 23);
 380    num.i += 127 << 23;
 381    num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3;
 382    return num.f + log_2;
 383 }
 384
 385 #if defined(__GNUC__)
 386 #define CEILF(x)   ceilf(x)
 387 #else
 388 #define CEILF(x)   ((float) ceil(x))
 389 #endif
 390
 391 static INLINE int align(int value, int alignment)
 392 {
 393    return (value + alignment - 1) & ~(alignment - 1);
 394 }
 395
 396
 397 /* util/p_util.c
 398  */
 399 extern void pipe_copy_rect(ubyte * dst, unsigned cpp, unsigned dst_pitch,
 400                            unsigned dst_x, unsigned dst_y, unsigned width,
 401                            unsigned height, const ubyte * src,
 402                            int src_pitch, unsigned src_x, int src_y);
 403
 404
 405
 406 #ifdef WIN32
 407
 408 #if !defined(_INC_MATH) || !defined(__cplusplus)
 409
 410 static INLINE float cosf( float f )
 411 {
 412    return (float) cos( (double) f );
 413 }
 414
 415 static INLINE float sinf( float f )
 416 {
 417    return (float) sin( (double) f );
 418 }
 419
 420 static INLINE float ceilf( float f )
 421 {
 422    return (float) ceil( (double) f );
 423 }
 424
 425 static INLINE float floorf( float f )
 426 {
 427    return (float) floor( (double) f );
 428 }
 429
 430 static INLINE float powf( float f, float g )
 431 {
 432    return (float) pow( (double) f, (double) g );
 433 }
 434
 435 static INLINE float sqrtf( float f )
 436 {
 437    return (float) sqrt( (double) f );
 438 }
 439
 440 static INLINE float fabsf( float f )
 441 {
 442    return (float) fabs( (double) f );
 443 }
 444
 445 static INLINE float logf( float f )
 446 {
 447    return (float) cos( (double) f );
 448 }
 449 #endif  /* _INC_MATH */
 450 #endif
 451
 452
 453 #ifdef __cplusplus
 454 }
 455 #endif
 456
 457 #endif