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_config.h"
  32 #include "p_compiler.h"
  33 #include "p_debug.h"
  34 #include "p_format.h"
  35 #include "p_pointer.h"
  36
  37 #if defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT)
  38 __inline double ceil(double val)
  39 {
  40         double ceil_val;
  41
  42         if((val - (long) val) == 0) {
  43                 ceil_val = val;
  44         } else {
  45                 if(val > 0) {
  46                         ceil_val = (long) val + 1;
  47                 } else {
  48                         ceil_val = (long) val;
  49                 }
  50         }
  51
  52         return ceil_val;
  53 }
  54
  55 #ifndef PIPE_SUBSYSTEM_WINDOWS_CE
  56 __inline double floor(double val)
  57 {
  58         double floor_val;
  59
  60         if((val - (long) val) == 0) {
  61                 floor_val = val;
  62         } else {
  63                 if(val > 0) {
  64                         floor_val = (long) val;
  65                 } else {
  66                         floor_val = (long) val - 1;
  67                 }
  68         }
  69
  70         return floor_val;
  71 }
  72 #endif
  73
  74 #pragma function(pow)
  75 __inline double __cdecl pow(double val, double exponent)
  76 {
  77         /* XXX */
  78         assert(0);
  79         return 0;
  80 }
  81
  82 #pragma function(log)
  83 __inline double __cdecl log(double val)
  84 {
  85         /* XXX */
  86         assert(0);
  87         return 0;
  88 }
  89
  90 #pragma function(atan2)
  91 __inline double __cdecl atan2(double val)
  92 {
  93         /* XXX */
  94         assert(0);
  95         return 0;
  96 }
  97 #else
  98 #include <math.h>
  99 #include <stdarg.h>
 100 #endif
 101
 102
 103 #ifdef __cplusplus
 104 extern "C" {
 105 #endif
 106
 107
 108 #if defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) && defined(DEBUG)
 109
 110 /* memory debugging */
 111
 112 #include "p_debug.h"
 113
 114 #define MALLOC( _size ) \
 115    debug_malloc( __FILE__, __LINE__, __FUNCTION__, _size )
 116 #define CALLOC( _count, _size ) \
 117    debug_calloc(__FILE__, __LINE__, __FUNCTION__, _count, _size )
 118 #define FREE( _ptr ) \
 119    debug_free( __FILE__, __LINE__, __FUNCTION__,  _ptr )
 120 #define REALLOC( _ptr, _old_size, _size ) \
 121    debug_realloc( __FILE__, __LINE__, __FUNCTION__,  _ptr, _old_size, _size )
 122
 123 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
 124
 125 void * __stdcall
 126 EngAllocMem(
 127     unsigned long Flags,
 128     unsigned long MemSize,
 129     unsigned long Tag );
 130
 131 void __stdcall
 132 EngFreeMem(
 133     void *Mem );
 134
 135 #define MALLOC( _size ) EngAllocMem( 0, _size, 'D3AG' )
 136 #define _FREE( _ptr ) EngFreeMem( _ptr )
 137
 138 #elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT)
 139
 140 void *
 141 ExAllocatePool(
 142     unsigned long PoolType,
 143     size_t NumberOfBytes);
 144
 145 void
 146 ExFreePool(void *P);
 147
 148 #define MALLOC(_size) ExAllocatePool(0, _size)
 149 #define _FREE(_ptr) ExFreePool(_ptr)
 150
 151 #else
 152
 153 #define MALLOC( SIZE )  malloc( SIZE )
 154 #define CALLOC( COUNT, SIZE )   calloc( COUNT, SIZE )
 155 #define FREE( PTR )  free( PTR )
 156 #define REALLOC( OLDPTR, OLDSIZE, NEWSIZE )  realloc( OLDPTR, NEWSIZE )
 157
 158 #endif
 159
 160
 161 #ifndef CALLOC
 162 static INLINE void *
 163 CALLOC( unsigned count, unsigned size )
 164 {
 165    void *ptr = MALLOC( count * size );
 166    if( ptr ) {
 167       memset( ptr, 0, count * size );
 168    }
 169    return ptr;
 170 }
 171 #endif /* !CALLOC */
 172
 173 #ifndef FREE
 174 static INLINE void
 175 FREE( void *ptr )
 176 {
 177    if( ptr ) {
 178       _FREE( ptr );
 179    }
 180 }
 181 #endif /* !FREE */
 182
 183 #ifndef REALLOC
 184 static INLINE void *
 185 REALLOC( void *old_ptr, unsigned old_size, unsigned new_size )
 186 {
 187    void *new_ptr = NULL;
 188
 189    if (new_size != 0) {
 190       unsigned copy_size = old_size < new_size ? old_size : new_size;
 191       new_ptr = MALLOC( new_size );
 192       if (new_ptr && old_ptr && copy_size) {
 193          memcpy( new_ptr, old_ptr, copy_size );
 194       }
 195    }
 196
 197    FREE( old_ptr );
 198    return new_ptr;
 199 }
 200 #endif /* !REALLOC */
 201
 202
 203 #define MALLOC_STRUCT(T)   (struct T *) MALLOC(sizeof(struct T))
 204
 205 #define CALLOC_STRUCT(T)   (struct T *) CALLOC(1, sizeof(struct T))
 206
 207
 208 /**
 209  * Return memory on given byte alignment
 210  */
 211 static INLINE void *
 212 align_malloc(size_t bytes, uint alignment)
 213 {
 214 #if defined(HAVE_POSIX_MEMALIGN)
 215    void *mem;
 216    (void) posix_memalign(& mem, alignment, bytes);
 217    return mem;
 218 #else
 219    char *ptr, *buf;
 220
 221    assert( alignment > 0 );
 222
 223    ptr = (char *) MALLOC(bytes + alignment + sizeof(void *));
 224    if (!ptr)
 225       return NULL;
 226
 227    buf = (char *) align_pointer( ptr + sizeof(void *), alignment );
 228    *(char **)(buf - sizeof(void *)) = ptr;
 229
 230    return buf;
 231 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 232 }
 233
 234 /**
 235  * Free memory returned by align_malloc().
 236  */
 237 static INLINE void
 238 align_free(void *ptr)
 239 {
 240 #if defined(HAVE_POSIX_MEMALIGN)
 241    FREE(ptr);
 242 #else
 243    void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
 244    void *realAddr = *cubbyHole;
 245    FREE(realAddr);
 246 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 247 }
 248
 249
 250
 251 /**
 252  * Duplicate a block of memory.
 253  */
 254 static INLINE void *
 255 mem_dup(const void *src, uint size)
 256 {
 257    void *dup = MALLOC(size);
 258    if (dup)
 259       memcpy(dup, src, size);
 260    return dup;
 261 }
 262
 263
 264
 265 #define CLAMP( X, MIN, MAX )  ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) )
 266 #define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
 267 #define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
 268
 269 #ifndef Elements
 270 #define Elements(x) (sizeof(x)/sizeof((x)[0]))
 271 #endif
 272 #define Offset(TYPE, MEMBER) ((unsigned)&(((TYPE *)NULL)->MEMBER))
 273
 274 /**
 275  * Return a pointer aligned to next multiple of 16 bytes.
 276  */
 277 static INLINE void *
 278 align16( void *unaligned )
 279 {
 280    return align_pointer( unaligned, 16 );
 281 }
 282
 283
 284 static INLINE int align_int(int x, int align)
 285 {
 286    return (x + align - 1) & ~(align - 1);
 287 }
 288
 289
 290
 291 #if defined(PIPE_CC_MSVC) && defined(PIPE_ARCH_X86)
 292 static INLINE unsigned ffs( unsigned u )
 293 {
 294    unsigned i;
 295
 296    if( u == 0 ) {
 297       return 0;
 298    }
 299
 300    __asm bsf eax, [u]
 301    __asm inc eax
 302    __asm mov [i], eax
 303
 304    return i;
 305 }
 306 #endif
 307
 308 union fi {
 309    float f;
 310    int i;
 311    unsigned ui;
 312 };
 313
 314 #define UBYTE_TO_FLOAT( ub ) ((float)(ub) / 255.0F)
 315
 316 #define IEEE_0996 0x3f7f0000    /* 0.996 or so */
 317
 318 /* This function/macro is sensitive to precision.  Test very carefully
 319  * if you change it!
 320  */
 321 #define UNCLAMPED_FLOAT_TO_UBYTE(UB, F)                                 \
 322         do {                                                            \
 323            union fi __tmp;                                              \
 324            __tmp.f = (F);                                               \
 325            if (__tmp.i < 0)                                             \
 326               UB = (ubyte) 0;                                           \
 327            else if (__tmp.i >= IEEE_0996)                               \
 328               UB = (ubyte) 255;                                 \
 329            else {                                                       \
 330               __tmp.f = __tmp.f * (255.0f/256.0f) + 32768.0f;           \
 331               UB = (ubyte) __tmp.i;                                     \
 332            }                                                            \
 333         } while (0)
 334
 335
 336
 337 static INLINE unsigned pack_ub4( unsigned char b0,
 338                                  unsigned char b1,
 339                                  unsigned char b2,
 340                                  unsigned char b3 )
 341 {
 342    return ((((unsigned int)b0) << 0) |
 343            (((unsigned int)b1) << 8) |
 344            (((unsigned int)b2) << 16) |
 345            (((unsigned int)b3) << 24));
 346 }
 347
 348 static INLINE unsigned fui( float f )
 349 {
 350    union fi fi;
 351    fi.f = f;
 352    return fi.ui;
 353 }
 354
 355 static INLINE unsigned char float_to_ubyte( float f )
 356 {
 357    unsigned char ub;
 358    UNCLAMPED_FLOAT_TO_UBYTE(ub, f);
 359    return ub;
 360 }
 361
 362 static INLINE unsigned pack_ui32_float4( float a,
 363                                          float b,
 364                                          float c,
 365                                          float d )
 366 {
 367    return pack_ub4( float_to_ubyte(a),
 368                     float_to_ubyte(b),
 369                     float_to_ubyte(c),
 370                     float_to_ubyte(d) );
 371 }
 372
 373 #define COPY_4V( DST, SRC )         \
 374 do {                                \
 375    (DST)[0] = (SRC)[0];             \
 376    (DST)[1] = (SRC)[1];             \
 377    (DST)[2] = (SRC)[2];             \
 378    (DST)[3] = (SRC)[3];             \
 379 } while (0)
 380
 381
 382 #define COPY_4FV( DST, SRC )  COPY_4V(DST, SRC)
 383
 384
 385 #define ASSIGN_4V( DST, V0, V1, V2, V3 ) \
 386 do {                                     \
 387    (DST)[0] = (V0);                      \
 388    (DST)[1] = (V1);                      \
 389    (DST)[2] = (V2);                      \
 390    (DST)[3] = (V3);                      \
 391 } while (0)
 392
 393
 394 static INLINE int ifloor(float f)
 395 {
 396    int ai, bi;
 397    double af, bf;
 398    union fi u;
 399
 400    af = (3 << 22) + 0.5 + (double)f;
 401    bf = (3 << 22) + 0.5 - (double)f;
 402    u.f = (float) af;  ai = u.i;
 403    u.f = (float) bf;  bi = u.i;
 404    return (ai - bi) >> 1;
 405 }
 406
 407
 408 #if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
 409 static INLINE int iround(float f)
 410 {
 411    int r;
 412    __asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st");
 413    return r;
 414 }
 415 #elif defined(PIPE_CC_MSVC) && defined(PIPE_ARCH_X86)
 416 static INLINE int iround(float f)
 417 {
 418    int r;
 419    _asm {
 420          fld f
 421          fistp r
 422         }
 423    return r;
 424 }
 425 #else
 426 #define IROUND(f)  ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
 427 #endif
 428
 429
 430 /* Could maybe have an inline version of this?
 431  */
 432 #if defined(__GNUC__)
 433 #define FABSF(x)   fabsf(x)
 434 #else
 435 #define FABSF(x)   ((float) fabs(x))
 436 #endif
 437
 438 /* Pretty fast, and accurate.
 439  * Based on code from http://www.flipcode.com/totd/
 440  */
 441 static INLINE float LOG2(float val)
 442 {
 443    union fi num;
 444    int log_2;
 445
 446    num.f = val;
 447    log_2 = ((num.i >> 23) & 255) - 128;
 448    num.i &= ~(255 << 23);
 449    num.i += 127 << 23;
 450    num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3;
 451    return num.f + log_2;
 452 }
 453
 454 #if defined(__GNUC__)
 455 #define CEILF(x)   ceilf(x)
 456 #else
 457 #define CEILF(x)   ((float) ceil(x))
 458 #endif
 459
 460 static INLINE int align(int value, int alignment)
 461 {
 462    return (value + alignment - 1) & ~(alignment - 1);
 463 }
 464
 465
 466 /* util/p_util.c
 467  */
 468 extern void pipe_copy_rect(ubyte * dst, const struct pipe_format_block *block,
 469                            unsigned dst_stride, unsigned dst_x, unsigned dst_y,
 470                            unsigned width, unsigned height, const ubyte * src,
 471                            int src_stride, unsigned src_x, int src_y);
 472
 473 extern void
 474 pipe_fill_rect(ubyte * dst, const struct pipe_format_block *block,
 475                unsigned dst_stride, unsigned dst_x, unsigned dst_y,
 476                unsigned width, unsigned height, uint32_t value);
 477
 478
 479 #if defined(_MSC_VER)
 480 #if _MSC_VER < 1400 && !defined(__cplusplus) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
 481
 482 static INLINE float cosf( float f )
 483 {
 484    return (float) cos( (double) f );
 485 }
 486
 487 static INLINE float sinf( float f )
 488 {
 489    return (float) sin( (double) f );
 490 }
 491
 492 static INLINE float ceilf( float f )
 493 {
 494    return (float) ceil( (double) f );
 495 }
 496
 497 static INLINE float floorf( float f )
 498 {
 499    return (float) floor( (double) f );
 500 }
 501
 502 static INLINE float powf( float f, float g )
 503 {
 504    return (float) pow( (double) f, (double) g );
 505 }
 506
 507 static INLINE float sqrtf( float f )
 508 {
 509    return (float) sqrt( (double) f );
 510 }
 511
 512 static INLINE float fabsf( float f )
 513 {
 514    return (float) fabs( (double) f );
 515 }
 516
 517 static INLINE float logf( float f )
 518 {
 519    return (float) log( (double) f );
 520 }
 521
 522 #else
 523 /* Work-around an extra semi-colon in VS 2005 logf definition */
 524 #ifdef logf
 525 #undef logf
 526 #define logf(x) ((float)log((double)(x)))
 527 #endif /* logf */
 528 #endif
 529 #endif /* _MSC_VER */
 530
 531
 532 #ifdef __cplusplus
 533 }
 534 #endif
 535
 536 #endif