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    if(posix_memalign(& mem, alignment, bytes) != 0)
 217       return NULL;
 218    return mem;
 219 #else
 220    char *ptr, *buf;
 221
 222    assert( alignment > 0 );
 223
 224    ptr = (char *) MALLOC(bytes + alignment + sizeof(void *));
 225    if (!ptr)
 226       return NULL;
 227
 228    buf = (char *) align_pointer( ptr + sizeof(void *), alignment );
 229    *(char **)(buf - sizeof(void *)) = ptr;
 230
 231    return buf;
 232 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 233 }
 234
 235 /**
 236  * Free memory returned by align_malloc().
 237  */
 238 static INLINE void
 239 align_free(void *ptr)
 240 {
 241 #if defined(HAVE_POSIX_MEMALIGN)
 242    FREE(ptr);
 243 #else
 244    void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
 245    void *realAddr = *cubbyHole;
 246    FREE(realAddr);
 247 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 248 }
 249
 250
 251
 252 /**
 253  * Duplicate a block of memory.
 254  */
 255 static INLINE void *
 256 mem_dup(const void *src, uint size)
 257 {
 258    void *dup = MALLOC(size);
 259    if (dup)
 260       memcpy(dup, src, size);
 261    return dup;
 262 }
 263
 264
 265
 266 #define CLAMP( X, MIN, MAX )  ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) )
 267 #define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
 268 #define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
 269
 270 #ifndef Elements
 271 #define Elements(x) (sizeof(x)/sizeof((x)[0]))
 272 #endif
 273 #define Offset(TYPE, MEMBER) ((unsigned)&(((TYPE *)NULL)->MEMBER))
 274
 275 /**
 276  * Return a pointer aligned to next multiple of 16 bytes.
 277  */
 278 static INLINE void *
 279 align16( void *unaligned )
 280 {
 281    return align_pointer( unaligned, 16 );
 282 }
 283
 284
 285 static INLINE int align_int(int x, int align)
 286 {
 287    return (x + align - 1) & ~(align - 1);
 288 }
 289
 290
 291
 292 #if defined(PIPE_CC_MSVC) && defined(PIPE_ARCH_X86)
 293 static INLINE unsigned ffs( unsigned u )
 294 {
 295    unsigned i;
 296
 297    if( u == 0 ) {
 298       return 0;
 299    }
 300
 301    __asm bsf eax, [u]
 302    __asm inc eax
 303    __asm mov [i], eax
 304
 305    return i;
 306 }
 307 #endif
 308
 309 union fi {
 310    float f;
 311    int i;
 312    unsigned ui;
 313 };
 314
 315 #define UBYTE_TO_FLOAT( ub ) ((float)(ub) / 255.0F)
 316
 317 #define IEEE_0996 0x3f7f0000    /* 0.996 or so */
 318
 319 /* This function/macro is sensitive to precision.  Test very carefully
 320  * if you change it!
 321  */
 322 #define UNCLAMPED_FLOAT_TO_UBYTE(UB, F)                                 \
 323         do {                                                            \
 324            union fi __tmp;                                              \
 325            __tmp.f = (F);                                               \
 326            if (__tmp.i < 0)                                             \
 327               UB = (ubyte) 0;                                           \
 328            else if (__tmp.i >= IEEE_0996)                               \
 329               UB = (ubyte) 255;                                 \
 330            else {                                                       \
 331               __tmp.f = __tmp.f * (255.0f/256.0f) + 32768.0f;           \
 332               UB = (ubyte) __tmp.i;                                     \
 333            }                                                            \
 334         } while (0)
 335
 336
 337
 338 static INLINE unsigned pack_ub4( unsigned char b0,
 339                                  unsigned char b1,
 340                                  unsigned char b2,
 341                                  unsigned char b3 )
 342 {
 343    return ((((unsigned int)b0) << 0) |
 344            (((unsigned int)b1) << 8) |
 345            (((unsigned int)b2) << 16) |
 346            (((unsigned int)b3) << 24));
 347 }
 348
 349 static INLINE unsigned fui( float f )
 350 {
 351    union fi fi;
 352    fi.f = f;
 353    return fi.ui;
 354 }
 355
 356 static INLINE unsigned char float_to_ubyte( float f )
 357 {
 358    unsigned char ub;
 359    UNCLAMPED_FLOAT_TO_UBYTE(ub, f);
 360    return ub;
 361 }
 362
 363 static INLINE unsigned pack_ui32_float4( float a,
 364                                          float b,
 365                                          float c,
 366                                          float d )
 367 {
 368    return pack_ub4( float_to_ubyte(a),
 369                     float_to_ubyte(b),
 370                     float_to_ubyte(c),
 371                     float_to_ubyte(d) );
 372 }
 373
 374 #define COPY_4V( DST, SRC )         \
 375 do {                                \
 376    (DST)[0] = (SRC)[0];             \
 377    (DST)[1] = (SRC)[1];             \
 378    (DST)[2] = (SRC)[2];             \
 379    (DST)[3] = (SRC)[3];             \
 380 } while (0)
 381
 382
 383 #define COPY_4FV( DST, SRC )  COPY_4V(DST, SRC)
 384
 385
 386 #define ASSIGN_4V( DST, V0, V1, V2, V3 ) \
 387 do {                                     \
 388    (DST)[0] = (V0);                      \
 389    (DST)[1] = (V1);                      \
 390    (DST)[2] = (V2);                      \
 391    (DST)[3] = (V3);                      \
 392 } while (0)
 393
 394
 395 static INLINE int ifloor(float f)
 396 {
 397    int ai, bi;
 398    double af, bf;
 399    union fi u;
 400
 401    af = (3 << 22) + 0.5 + (double)f;
 402    bf = (3 << 22) + 0.5 - (double)f;
 403    u.f = (float) af;  ai = u.i;
 404    u.f = (float) bf;  bi = u.i;
 405    return (ai - bi) >> 1;
 406 }
 407
 408
 409 #if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
 410 static INLINE int iround(float f)
 411 {
 412    int r;
 413    __asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st");
 414    return r;
 415 }
 416 #elif defined(PIPE_CC_MSVC) && defined(PIPE_ARCH_X86)
 417 static INLINE int iround(float f)
 418 {
 419    int r;
 420    _asm {
 421          fld f
 422          fistp r
 423         }
 424    return r;
 425 }
 426 #else
 427 #define IROUND(f)  ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
 428 #endif
 429
 430
 431 /* Could maybe have an inline version of this?
 432  */
 433 #if defined(__GNUC__)
 434 #define FABSF(x)   fabsf(x)
 435 #else
 436 #define FABSF(x)   ((float) fabs(x))
 437 #endif
 438
 439 /* Pretty fast, and accurate.
 440  * Based on code from http://www.flipcode.com/totd/
 441  */
 442 static INLINE float LOG2(float val)
 443 {
 444    union fi num;
 445    int log_2;
 446
 447    num.f = val;
 448    log_2 = ((num.i >> 23) & 255) - 128;
 449    num.i &= ~(255 << 23);
 450    num.i += 127 << 23;
 451    num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3;
 452    return num.f + log_2;
 453 }
 454
 455 #if defined(__GNUC__)
 456 #define CEILF(x)   ceilf(x)
 457 #else
 458 #define CEILF(x)   ((float) ceil(x))
 459 #endif
 460
 461 static INLINE int align(int value, int alignment)
 462 {
 463    return (value + alignment - 1) & ~(alignment - 1);
 464 }
 465
 466
 467 /* util/p_util.c
 468  */
 469 extern void pipe_copy_rect(ubyte * dst, const struct pipe_format_block *block,
 470                            unsigned dst_stride, unsigned dst_x, unsigned dst_y,
 471                            unsigned width, unsigned height, const ubyte * src,
 472                            int src_stride, unsigned src_x, int src_y);
 473
 474 extern void
 475 pipe_fill_rect(ubyte * dst, const struct pipe_format_block *block,
 476                unsigned dst_stride, unsigned dst_x, unsigned dst_y,
 477                unsigned width, unsigned height, uint32_t value);
 478
 479
 480 #if defined(_MSC_VER)
 481 #if _MSC_VER < 1400 && !defined(__cplusplus) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
 482
 483 static INLINE float cosf( float f )
 484 {
 485    return (float) cos( (double) f );
 486 }
 487
 488 static INLINE float sinf( float f )
 489 {
 490    return (float) sin( (double) f );
 491 }
 492
 493 static INLINE float ceilf( float f )
 494 {
 495    return (float) ceil( (double) f );
 496 }
 497
 498 static INLINE float floorf( float f )
 499 {
 500    return (float) floor( (double) f );
 501 }
 502
 503 static INLINE float powf( float f, float g )
 504 {
 505    return (float) pow( (double) f, (double) g );
 506 }
 507
 508 static INLINE float sqrtf( float f )
 509 {
 510    return (float) sqrt( (double) f );
 511 }
 512
 513 static INLINE float fabsf( float f )
 514 {
 515    return (float) fabs( (double) f );
 516 }
 517
 518 static INLINE float logf( float f )
 519 {
 520    return (float) log( (double) f );
 521 }
 522
 523 #else
 524 /* Work-around an extra semi-colon in VS 2005 logf definition */
 525 #ifdef logf
 526 #undef logf
 527 #define logf(x) ((float)log((double)(x)))
 528 #endif /* logf */
 529 #endif
 530 #endif /* _MSC_VER */
 531
 532
 533 #ifdef __cplusplus
 534 }
 535 #endif
 536
 537 #endif