1 /* 2 Simple DirectMedia Layer 3 Copyright (C) 1997-2018 Sam Lantinga <slouken@libsdl.org> 4 5 This software is provided 'as-is', without any express or implied 6 warranty. In no event will the authors be held liable for any damages 7 arising from the use of this software. 8 9 Permission is granted to anyone to use this software for any purpose, 10 including commercial applications, and to alter it and redistribute it 11 freely, subject to the following restrictions: 12 13 1. The origin of this software must not be misrepresented; you must not 14 claim that you wrote the original software. If you use this software 15 in a product, an acknowledgment in the product documentation would be 16 appreciated but is not required. 17 2. Altered source versions must be plainly marked as such, and must not be 18 misrepresented as being the original software. 19 3. This notice may not be removed or altered from any source distribution. 20 */ 21 #include "../SDL_internal.h" 22 23 #ifndef SDL_blit_h_ 24 #define SDL_blit_h_ 25 26 #include "SDL_cpuinfo.h" 27 #include "SDL_endian.h" 28 #include "SDL_surface.h" 29 30 /* Table to do pixel byte expansion */ 31 extern Uint8* SDL_expand_byte[9]; 32 33 /* SDL blit copy flags */ 34 #define SDL_COPY_MODULATE_COLOR 0x00000001 35 #define SDL_COPY_MODULATE_ALPHA 0x00000002 36 #define SDL_COPY_BLEND 0x00000010 37 #define SDL_COPY_ADD 0x00000020 38 #define SDL_COPY_MOD 0x00000040 39 #define SDL_COPY_COLORKEY 0x00000100 40 #define SDL_COPY_NEAREST 0x00000200 41 #define SDL_COPY_RLE_DESIRED 0x00001000 42 #define SDL_COPY_RLE_COLORKEY 0x00002000 43 #define SDL_COPY_RLE_ALPHAKEY 0x00004000 44 #define SDL_COPY_RLE_MASK (SDL_COPY_RLE_DESIRED|SDL_COPY_RLE_COLORKEY|SDL_COPY_RLE_ALPHAKEY) 45 46 /* SDL blit CPU flags */ 47 #define SDL_CPU_ANY 0x00000000 48 #define SDL_CPU_MMX 0x00000001 49 #define SDL_CPU_3DNOW 0x00000002 50 #define SDL_CPU_SSE 0x00000004 51 #define SDL_CPU_SSE2 0x00000008 52 #define SDL_CPU_ALTIVEC_PREFETCH 0x00000010 53 #define SDL_CPU_ALTIVEC_NOPREFETCH 0x00000020 54 55 typedef struct 56 { 57 Uint8 *src; 58 int src_w, src_h; 59 int src_pitch; 60 int src_skip; 61 Uint8 *dst; 62 int dst_w, dst_h; 63 int dst_pitch; 64 int dst_skip; 65 SDL_PixelFormat *src_fmt; 66 SDL_PixelFormat *dst_fmt; 67 Uint8 *table; 68 int flags; 69 Uint32 colorkey; 70 Uint8 r, g, b, a; 71 } SDL_BlitInfo; 72 73 typedef void (*SDL_BlitFunc) (SDL_BlitInfo *info); 74 75 76 typedef struct 77 { 78 Uint32 src_format; 79 Uint32 dst_format; 80 int flags; 81 int cpu; 82 SDL_BlitFunc func; 83 } SDL_BlitFuncEntry; 84 85 /* Blit mapping definition */ 86 typedef struct SDL_BlitMap 87 { 88 SDL_Surface *dst; 89 int identity; 90 SDL_blit blit; 91 void *data; 92 SDL_BlitInfo info; 93 94 /* the version count matches the destination; mismatch indicates 95 an invalid mapping */ 96 Uint32 dst_palette_version; 97 Uint32 src_palette_version; 98 } SDL_BlitMap; 99 100 /* Functions found in SDL_blit.c */ 101 extern int SDL_CalculateBlit(SDL_Surface * surface); 102 103 /* Functions found in SDL_blit_*.c */ 104 extern SDL_BlitFunc SDL_CalculateBlit0(SDL_Surface * surface); 105 extern SDL_BlitFunc SDL_CalculateBlit1(SDL_Surface * surface); 106 extern SDL_BlitFunc SDL_CalculateBlitN(SDL_Surface * surface); 107 extern SDL_BlitFunc SDL_CalculateBlitA(SDL_Surface * surface); 108 109 /* 110 * Useful macros for blitting routines 111 */ 112 113 #if defined(__GNUC__) 114 #define DECLARE_ALIGNED(t,v,a) t __attribute__((aligned(a))) v 115 #elif defined(_MSC_VER) 116 #define DECLARE_ALIGNED(t,v,a) __declspec(align(a)) t v 117 #else 118 #define DECLARE_ALIGNED(t,v,a) t v 119 #endif 120 121 /* Load pixel of the specified format from a buffer and get its R-G-B values */ 122 #define RGB_FROM_PIXEL(Pixel, fmt, r, g, b) \ 123 { \ 124 r = SDL_expand_byte[fmt->Rloss][((Pixel&fmt->Rmask)>>fmt->Rshift)]; \ 125 g = SDL_expand_byte[fmt->Gloss][((Pixel&fmt->Gmask)>>fmt->Gshift)]; \ 126 b = SDL_expand_byte[fmt->Bloss][((Pixel&fmt->Bmask)>>fmt->Bshift)]; \ 127 } 128 #define RGB_FROM_RGB565(Pixel, r, g, b) \ 129 { \ 130 r = SDL_expand_byte[3][((Pixel&0xF800)>>11)]; \ 131 g = SDL_expand_byte[2][((Pixel&0x07E0)>>5)]; \ 132 b = SDL_expand_byte[3][(Pixel&0x001F)]; \ 133 } 134 #define RGB_FROM_RGB555(Pixel, r, g, b) \ 135 { \ 136 r = SDL_expand_byte[3][((Pixel&0x7C00)>>10)]; \ 137 g = SDL_expand_byte[3][((Pixel&0x03E0)>>5)]; \ 138 b = SDL_expand_byte[3][(Pixel&0x001F)]; \ 139 } 140 #define RGB_FROM_RGB888(Pixel, r, g, b) \ 141 { \ 142 r = ((Pixel&0xFF0000)>>16); \ 143 g = ((Pixel&0xFF00)>>8); \ 144 b = (Pixel&0xFF); \ 145 } 146 #define RETRIEVE_RGB_PIXEL(buf, bpp, Pixel) \ 147 do { \ 148 switch (bpp) { \ 149 case 1: \ 150 Pixel = *((Uint8 *)(buf)); \ 151 break; \ 152 \ 153 case 2: \ 154 Pixel = *((Uint16 *)(buf)); \ 155 break; \ 156 \ 157 case 3: { \ 158 Uint8 *B = (Uint8 *)(buf); \ 159 if (SDL_BYTEORDER == SDL_LIL_ENDIAN) { \ 160 Pixel = B[0] + (B[1] << 8) + (B[2] << 16); \ 161 } else { \ 162 Pixel = (B[0] << 16) + (B[1] << 8) + B[2]; \ 163 } \ 164 } \ 165 break; \ 166 \ 167 case 4: \ 168 Pixel = *((Uint32 *)(buf)); \ 169 break; \ 170 \ 171 default: \ 172 Pixel = 0; /* stop gcc complaints */ \ 173 break; \ 174 } \ 175 } while (0) 176 177 #define DISEMBLE_RGB(buf, bpp, fmt, Pixel, r, g, b) \ 178 do { \ 179 switch (bpp) { \ 180 case 1: \ 181 Pixel = *((Uint8 *)(buf)); \ 182 RGB_FROM_PIXEL(Pixel, fmt, r, g, b); \ 183 break; \ 184 \ 185 case 2: \ 186 Pixel = *((Uint16 *)(buf)); \ 187 RGB_FROM_PIXEL(Pixel, fmt, r, g, b); \ 188 break; \ 189 \ 190 case 3: { \ 191 Pixel = 0; \ 192 if (SDL_BYTEORDER == SDL_LIL_ENDIAN) { \ 193 r = *((buf)+fmt->Rshift/8); \ 194 g = *((buf)+fmt->Gshift/8); \ 195 b = *((buf)+fmt->Bshift/8); \ 196 } else { \ 197 r = *((buf)+2-fmt->Rshift/8); \ 198 g = *((buf)+2-fmt->Gshift/8); \ 199 b = *((buf)+2-fmt->Bshift/8); \ 200 } \ 201 } \ 202 break; \ 203 \ 204 case 4: \ 205 Pixel = *((Uint32 *)(buf)); \ 206 RGB_FROM_PIXEL(Pixel, fmt, r, g, b); \ 207 break; \ 208 \ 209 default: \ 210 /* stop gcc complaints */ \ 211 Pixel = 0; \ 212 r = g = b = 0; \ 213 break; \ 214 } \ 215 } while (0) 216 217 /* Assemble R-G-B values into a specified pixel format and store them */ 218 #define PIXEL_FROM_RGB(Pixel, fmt, r, g, b) \ 219 { \ 220 Pixel = ((r>>fmt->Rloss)<<fmt->Rshift)| \ 221 ((g>>fmt->Gloss)<<fmt->Gshift)| \ 222 ((b>>fmt->Bloss)<<fmt->Bshift)| \ 223 fmt->Amask; \ 224 } 225 #define RGB565_FROM_RGB(Pixel, r, g, b) \ 226 { \ 227 Pixel = ((r>>3)<<11)|((g>>2)<<5)|(b>>3); \ 228 } 229 #define RGB555_FROM_RGB(Pixel, r, g, b) \ 230 { \ 231 Pixel = ((r>>3)<<10)|((g>>3)<<5)|(b>>3); \ 232 } 233 #define RGB888_FROM_RGB(Pixel, r, g, b) \ 234 { \ 235 Pixel = (r<<16)|(g<<8)|b; \ 236 } 237 #define ARGB8888_FROM_RGBA(Pixel, r, g, b, a) \ 238 { \ 239 Pixel = (a<<24)|(r<<16)|(g<<8)|b; \ 240 } 241 #define RGBA8888_FROM_RGBA(Pixel, r, g, b, a) \ 242 { \ 243 Pixel = (r<<24)|(g<<16)|(b<<8)|a; \ 244 } 245 #define ABGR8888_FROM_RGBA(Pixel, r, g, b, a) \ 246 { \ 247 Pixel = (a<<24)|(b<<16)|(g<<8)|r; \ 248 } 249 #define BGRA8888_FROM_RGBA(Pixel, r, g, b, a) \ 250 { \ 251 Pixel = (b<<24)|(g<<16)|(r<<8)|a; \ 252 } 253 #define ARGB2101010_FROM_RGBA(Pixel, r, g, b, a) \ 254 { \ 255 r = r ? ((r << 2) | 0x3) : 0; \ 256 g = g ? ((g << 2) | 0x3) : 0; \ 257 b = b ? ((b << 2) | 0x3) : 0; \ 258 a = (a * 3) / 255; \ 259 Pixel = (a<<30)|(r<<20)|(g<<10)|b; \ 260 } 261 #define ASSEMBLE_RGB(buf, bpp, fmt, r, g, b) \ 262 { \ 263 switch (bpp) { \ 264 case 1: { \ 265 Uint8 Pixel; \ 266 \ 267 PIXEL_FROM_RGB(Pixel, fmt, r, g, b); \ 268 *((Uint8 *)(buf)) = Pixel; \ 269 } \ 270 break; \ 271 \ 272 case 2: { \ 273 Uint16 Pixel; \ 274 \ 275 PIXEL_FROM_RGB(Pixel, fmt, r, g, b); \ 276 *((Uint16 *)(buf)) = Pixel; \ 277 } \ 278 break; \ 279 \ 280 case 3: { \ 281 if (SDL_BYTEORDER == SDL_LIL_ENDIAN) { \ 282 *((buf)+fmt->Rshift/8) = r; \ 283 *((buf)+fmt->Gshift/8) = g; \ 284 *((buf)+fmt->Bshift/8) = b; \ 285 } else { \ 286 *((buf)+2-fmt->Rshift/8) = r; \ 287 *((buf)+2-fmt->Gshift/8) = g; \ 288 *((buf)+2-fmt->Bshift/8) = b; \ 289 } \ 290 } \ 291 break; \ 292 \ 293 case 4: { \ 294 Uint32 Pixel; \ 295 \ 296 PIXEL_FROM_RGB(Pixel, fmt, r, g, b); \ 297 *((Uint32 *)(buf)) = Pixel; \ 298 } \ 299 break; \ 300 } \ 301 } 302 303 /* FIXME: Should we rescale alpha into 0..255 here? */ 304 #define RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a) \ 305 { \ 306 r = SDL_expand_byte[fmt->Rloss][((Pixel&fmt->Rmask)>>fmt->Rshift)]; \ 307 g = SDL_expand_byte[fmt->Gloss][((Pixel&fmt->Gmask)>>fmt->Gshift)]; \ 308 b = SDL_expand_byte[fmt->Bloss][((Pixel&fmt->Bmask)>>fmt->Bshift)]; \ 309 a = SDL_expand_byte[fmt->Aloss][((Pixel&fmt->Amask)>>fmt->Ashift)]; \ 310 } 311 #define RGBA_FROM_8888(Pixel, fmt, r, g, b, a) \ 312 { \ 313 r = (Pixel&fmt->Rmask)>>fmt->Rshift; \ 314 g = (Pixel&fmt->Gmask)>>fmt->Gshift; \ 315 b = (Pixel&fmt->Bmask)>>fmt->Bshift; \ 316 a = (Pixel&fmt->Amask)>>fmt->Ashift; \ 317 } 318 #define RGBA_FROM_RGBA8888(Pixel, r, g, b, a) \ 319 { \ 320 r = (Pixel>>24); \ 321 g = ((Pixel>>16)&0xFF); \ 322 b = ((Pixel>>8)&0xFF); \ 323 a = (Pixel&0xFF); \ 324 } 325 #define RGBA_FROM_ARGB8888(Pixel, r, g, b, a) \ 326 { \ 327 r = ((Pixel>>16)&0xFF); \ 328 g = ((Pixel>>8)&0xFF); \ 329 b = (Pixel&0xFF); \ 330 a = (Pixel>>24); \ 331 } 332 #define RGBA_FROM_ABGR8888(Pixel, r, g, b, a) \ 333 { \ 334 r = (Pixel&0xFF); \ 335 g = ((Pixel>>8)&0xFF); \ 336 b = ((Pixel>>16)&0xFF); \ 337 a = (Pixel>>24); \ 338 } 339 #define RGBA_FROM_BGRA8888(Pixel, r, g, b, a) \ 340 { \ 341 r = ((Pixel>>8)&0xFF); \ 342 g = ((Pixel>>16)&0xFF); \ 343 b = (Pixel>>24); \ 344 a = (Pixel&0xFF); \ 345 } 346 #define RGBA_FROM_ARGB2101010(Pixel, r, g, b, a) \ 347 { \ 348 r = ((Pixel>>22)&0xFF); \ 349 g = ((Pixel>>12)&0xFF); \ 350 b = ((Pixel>>2)&0xFF); \ 351 a = SDL_expand_byte[6][(Pixel>>30)]; \ 352 } 353 #define DISEMBLE_RGBA(buf, bpp, fmt, Pixel, r, g, b, a) \ 354 do { \ 355 switch (bpp) { \ 356 case 1: \ 357 Pixel = *((Uint8 *)(buf)); \ 358 RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a); \ 359 break; \ 360 \ 361 case 2: \ 362 Pixel = *((Uint16 *)(buf)); \ 363 RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a); \ 364 break; \ 365 \ 366 case 3: { \ 367 Pixel = 0; \ 368 if (SDL_BYTEORDER == SDL_LIL_ENDIAN) { \ 369 r = *((buf)+fmt->Rshift/8); \ 370 g = *((buf)+fmt->Gshift/8); \ 371 b = *((buf)+fmt->Bshift/8); \ 372 } else { \ 373 r = *((buf)+2-fmt->Rshift/8); \ 374 g = *((buf)+2-fmt->Gshift/8); \ 375 b = *((buf)+2-fmt->Bshift/8); \ 376 } \ 377 a = 0xFF; \ 378 } \ 379 break; \ 380 \ 381 case 4: \ 382 Pixel = *((Uint32 *)(buf)); \ 383 RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a); \ 384 break; \ 385 \ 386 default: \ 387 /* stop gcc complaints */ \ 388 Pixel = 0; \ 389 r = g = b = a = 0; \ 390 break; \ 391 } \ 392 } while (0) 393 394 /* FIXME: this isn't correct, especially for Alpha (maximum != 255) */ 395 #define PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a) \ 396 { \ 397 Pixel = ((r>>fmt->Rloss)<<fmt->Rshift)| \ 398 ((g>>fmt->Gloss)<<fmt->Gshift)| \ 399 ((b>>fmt->Bloss)<<fmt->Bshift)| \ 400 ((a>>fmt->Aloss)<<fmt->Ashift); \ 401 } 402 #define ASSEMBLE_RGBA(buf, bpp, fmt, r, g, b, a) \ 403 { \ 404 switch (bpp) { \ 405 case 1: { \ 406 Uint8 _pixel; \ 407 \ 408 PIXEL_FROM_RGBA(_pixel, fmt, r, g, b, a); \ 409 *((Uint8 *)(buf)) = _pixel; \ 410 } \ 411 break; \ 412 \ 413 case 2: { \ 414 Uint16 _pixel; \ 415 \ 416 PIXEL_FROM_RGBA(_pixel, fmt, r, g, b, a); \ 417 *((Uint16 *)(buf)) = _pixel; \ 418 } \ 419 break; \ 420 \ 421 case 3: { \ 422 if (SDL_BYTEORDER == SDL_LIL_ENDIAN) { \ 423 *((buf)+fmt->Rshift/8) = r; \ 424 *((buf)+fmt->Gshift/8) = g; \ 425 *((buf)+fmt->Bshift/8) = b; \ 426 } else { \ 427 *((buf)+2-fmt->Rshift/8) = r; \ 428 *((buf)+2-fmt->Gshift/8) = g; \ 429 *((buf)+2-fmt->Bshift/8) = b; \ 430 } \ 431 } \ 432 break; \ 433 \ 434 case 4: { \ 435 Uint32 _pixel; \ 436 \ 437 PIXEL_FROM_RGBA(_pixel, fmt, r, g, b, a); \ 438 *((Uint32 *)(buf)) = _pixel; \ 439 } \ 440 break; \ 441 } \ 442 } 443 444 /* Blend the RGB values of two pixels with an alpha value */ 445 #define ALPHA_BLEND_RGB(sR, sG, sB, A, dR, dG, dB) \ 446 do { \ 447 dR = (Uint8)((((int)(sR-dR)*(int)A)/255)+dR); \ 448 dG = (Uint8)((((int)(sG-dG)*(int)A)/255)+dG); \ 449 dB = (Uint8)((((int)(sB-dB)*(int)A)/255)+dB); \ 450 } while(0) 451 452 453 /* Blend the RGBA values of two pixels */ 454 #define ALPHA_BLEND_RGBA(sR, sG, sB, sA, dR, dG, dB, dA) \ 455 do { \ 456 dR = (Uint8)((((int)(sR-dR)*(int)sA)/255)+dR); \ 457 dG = (Uint8)((((int)(sG-dG)*(int)sA)/255)+dG); \ 458 dB = (Uint8)((((int)(sB-dB)*(int)sA)/255)+dB); \ 459 dA = (Uint8)((int)sA+dA-((int)sA*dA)/255); \ 460 } while(0) 461 462 463 /* This is a very useful loop for optimizing blitters */ 464 #if defined(_MSC_VER) && (_MSC_VER == 1300) 465 /* There's a bug in the Visual C++ 7 optimizer when compiling this code */ 466 #else 467 #define USE_DUFFS_LOOP 468 #endif 469 #ifdef USE_DUFFS_LOOP 470 471 /* 8-times unrolled loop */ 472 #define DUFFS_LOOP8(pixel_copy_increment, width) \ 473 { int n = (width+7)/8; \ 474 switch (width & 7) { \ 475 case 0: do { pixel_copy_increment; /* fallthrough */ \ 476 case 7: pixel_copy_increment; /* fallthrough */ \ 477 case 6: pixel_copy_increment; /* fallthrough */ \ 478 case 5: pixel_copy_increment; /* fallthrough */ \ 479 case 4: pixel_copy_increment; /* fallthrough */ \ 480 case 3: pixel_copy_increment; /* fallthrough */ \ 481 case 2: pixel_copy_increment; /* fallthrough */ \ 482 case 1: pixel_copy_increment; /* fallthrough */ \ 483 } while ( --n > 0 ); \ 484 } \ 485 } 486 487 /* 4-times unrolled loop */ 488 #define DUFFS_LOOP4(pixel_copy_increment, width) \ 489 { int n = (width+3)/4; \ 490 switch (width & 3) { \ 491 case 0: do { pixel_copy_increment; /* fallthrough */ \ 492 case 3: pixel_copy_increment; /* fallthrough */ \ 493 case 2: pixel_copy_increment; /* fallthrough */ \ 494 case 1: pixel_copy_increment; /* fallthrough */ \ 495 } while (--n > 0); \ 496 } \ 497 } 498 499 /* Use the 8-times version of the loop by default */ 500 #define DUFFS_LOOP(pixel_copy_increment, width) \ 501 DUFFS_LOOP8(pixel_copy_increment, width) 502 503 /* Special version of Duff's device for even more optimization */ 504 #define DUFFS_LOOP_124(pixel_copy_increment1, \ 505 pixel_copy_increment2, \ 506 pixel_copy_increment4, width) \ 507 { int n = width; \ 508 if (n & 1) { \ 509 pixel_copy_increment1; n -= 1; \ 510 } \ 511 if (n & 2) { \ 512 pixel_copy_increment2; n -= 2; \ 513 } \ 514 if (n & 4) { \ 515 pixel_copy_increment4; n -= 4; \ 516 } \ 517 if (n) { \ 518 n /= 8; \ 519 do { \ 520 pixel_copy_increment4; \ 521 pixel_copy_increment4; \ 522 } while (--n > 0); \ 523 } \ 524 } 525 526 #else 527 528 /* Don't use Duff's device to unroll loops */ 529 #define DUFFS_LOOP(pixel_copy_increment, width) \ 530 { int n; \ 531 for ( n=width; n > 0; --n ) { \ 532 pixel_copy_increment; \ 533 } \ 534 } 535 #define DUFFS_LOOP8(pixel_copy_increment, width) \ 536 DUFFS_LOOP(pixel_copy_increment, width) 537 #define DUFFS_LOOP4(pixel_copy_increment, width) \ 538 DUFFS_LOOP(pixel_copy_increment, width) 539 #define DUFFS_LOOP_124(pixel_copy_increment1, \ 540 pixel_copy_increment2, \ 541 pixel_copy_increment4, width) \ 542 DUFFS_LOOP(pixel_copy_increment1, width) 543 544 #endif /* USE_DUFFS_LOOP */ 545 546 /* Prevent Visual C++ 6.0 from printing out stupid warnings */ 547 #if defined(_MSC_VER) && (_MSC_VER >= 600) 548 #pragma warning(disable: 4550) 549 #endif 550 551 #endif /* SDL_blit_h_ */ 552 553 /* vi: set ts=4 sw=4 expandtab: */ 554