src/video/SDL_blit.h
author Sam Lantinga <slouken@libsdl.org>
Sun, 21 Dec 2008 08:55:06 +0000
changeset 2899 a0c837a16e4c
parent 2898 e40448bc7727
child 3035 ff602fdfdedc
permissions -rw-r--r--
Added ARGB optimized case for Mac OS X
slouken@0
     1
/*
slouken@0
     2
    SDL - Simple DirectMedia Layer
slouken@2859
     3
    Copyright (C) 1997-2009 Sam Lantinga
slouken@0
     4
slouken@0
     5
    This library is free software; you can redistribute it and/or
slouken@1312
     6
    modify it under the terms of the GNU Lesser General Public
slouken@0
     7
    License as published by the Free Software Foundation; either
slouken@1312
     8
    version 2.1 of the License, or (at your option) any later version.
slouken@0
     9
slouken@0
    10
    This library is distributed in the hope that it will be useful,
slouken@0
    11
    but WITHOUT ANY WARRANTY; without even the implied warranty of
slouken@0
    12
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
slouken@1312
    13
    Lesser General Public License for more details.
slouken@0
    14
slouken@1312
    15
    You should have received a copy of the GNU Lesser General Public
slouken@1312
    16
    License along with this library; if not, write to the Free Software
slouken@1312
    17
    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
slouken@0
    18
slouken@0
    19
    Sam Lantinga
slouken@252
    20
    slouken@libsdl.org
slouken@0
    21
*/
slouken@1402
    22
#include "SDL_config.h"
slouken@0
    23
slouken@0
    24
#ifndef _SDL_blit_h
slouken@0
    25
#define _SDL_blit_h
slouken@0
    26
slouken@2249
    27
#ifdef __MMX__
slouken@2249
    28
#include <mmintrin.h>
slouken@2249
    29
#endif
slouken@2261
    30
#ifdef __3dNOW__
slouken@2261
    31
#include <mm3dnow.h>
slouken@2261
    32
#endif
slouken@2249
    33
#ifdef __SSE__
slouken@2249
    34
#include <xmmintrin.h>
slouken@2249
    35
#endif
slouken@2262
    36
#ifdef __SSE2__
slouken@2262
    37
#include <emmintrin.h>
slouken@2262
    38
#endif
slouken@2249
    39
slouken@2250
    40
#include "SDL_cpuinfo.h"
slouken@0
    41
#include "SDL_endian.h"
slouken@2898
    42
#include "SDL_video.h"
slouken@0
    43
slouken@2262
    44
/* SDL blit copy flags */
slouken@2266
    45
#define SDL_COPY_MODULATE_COLOR     0x00000001
slouken@2266
    46
#define SDL_COPY_MODULATE_ALPHA     0x00000002
slouken@2266
    47
#define SDL_COPY_MASK               0x00000010
slouken@2266
    48
#define SDL_COPY_BLEND              0x00000020
slouken@2266
    49
#define SDL_COPY_ADD                0x00000040
slouken@2266
    50
#define SDL_COPY_MOD                0x00000080
slouken@2266
    51
#define SDL_COPY_COLORKEY           0x00000100
slouken@2266
    52
#define SDL_COPY_NEAREST            0x00000200
slouken@2266
    53
#define SDL_COPY_RLE_DESIRED        0x00001000
slouken@2266
    54
#define SDL_COPY_RLE_COLORKEY       0x00002000
slouken@2266
    55
#define SDL_COPY_RLE_ALPHAKEY       0x00004000
slouken@2853
    56
#define SDL_COPY_RLE_MASK           (SDL_COPY_RLE_DESIRED|SDL_COPY_RLE_COLORKEY|SDL_COPY_RLE_ALPHAKEY)
slouken@2262
    57
slouken@2262
    58
/* SDL blit CPU flags */
slouken@2266
    59
#define SDL_CPU_ANY                 0x00000000
slouken@2266
    60
#define SDL_CPU_MMX                 0x00000001
slouken@2266
    61
#define SDL_CPU_3DNOW               0x00000002
slouken@2266
    62
#define SDL_CPU_SSE                 0x00000004
slouken@2266
    63
#define SDL_CPU_SSE2                0x00000008
slouken@2266
    64
#define SDL_CPU_ALTIVEC_PREFETCH    0x00000010
slouken@2266
    65
#define SDL_CPU_ALTIVEC_NOPREFETCH  0x00000020
slouken@2262
    66
slouken@2267
    67
typedef struct
slouken@2267
    68
{
slouken@2262
    69
    Uint8 *src;
slouken@2262
    70
    int src_w, src_h;
slouken@2262
    71
    int src_pitch;
slouken@2267
    72
    int src_skip;
slouken@2262
    73
    Uint8 *dst;
slouken@2262
    74
    int dst_w, dst_h;
slouken@2262
    75
    int dst_pitch;
slouken@2267
    76
    int dst_skip;
slouken@2262
    77
    SDL_PixelFormat *src_fmt;
slouken@2262
    78
    SDL_PixelFormat *dst_fmt;
slouken@1895
    79
    Uint8 *table;
slouken@2262
    80
    int flags;
slouken@2262
    81
    Uint32 colorkey;
slouken@2262
    82
    Uint8 r, g, b, a;
slouken@0
    83
} SDL_BlitInfo;
slouken@0
    84
slouken@2267
    85
typedef void (SDLCALL * SDL_BlitFunc) (SDL_BlitInfo * info);
slouken@2262
    86
slouken@2267
    87
typedef struct
slouken@2267
    88
{
slouken@2262
    89
    Uint32 src_format;
slouken@2262
    90
    Uint32 dst_format;
slouken@2262
    91
    int flags;
slouken@2262
    92
    int cpu;
slouken@2262
    93
    SDL_BlitFunc func;
slouken@2262
    94
} SDL_BlitFuncEntry;
slouken@0
    95
slouken@0
    96
/* Blit mapping definition */
slouken@1895
    97
typedef struct SDL_BlitMap
slouken@1895
    98
{
slouken@1895
    99
    SDL_Surface *dst;
slouken@1895
   100
    int identity;
slouken@2257
   101
    SDL_blit blit;
slouken@2257
   102
    void *data;
slouken@2262
   103
    SDL_BlitInfo info;
slouken@0
   104
slouken@1895
   105
    /* the version count matches the destination; mismatch indicates
slouken@1895
   106
       an invalid mapping */
slouken@1895
   107
    unsigned int format_version;
slouken@0
   108
} SDL_BlitMap;
slouken@0
   109
slouken@0
   110
/* Functions found in SDL_blit.c */
slouken@1895
   111
extern int SDL_CalculateBlit(SDL_Surface * surface);
slouken@0
   112
slouken@2267
   113
/* Functions found in SDL_blit_*.c */
slouken@2267
   114
extern SDL_BlitFunc SDL_CalculateBlit0(SDL_Surface * surface);
slouken@2267
   115
extern SDL_BlitFunc SDL_CalculateBlit1(SDL_Surface * surface);
slouken@2267
   116
extern SDL_BlitFunc SDL_CalculateBlitN(SDL_Surface * surface);
slouken@2267
   117
extern SDL_BlitFunc SDL_CalculateBlitA(SDL_Surface * surface);
slouken@2263
   118
slouken@0
   119
/*
slouken@0
   120
 * Useful macros for blitting routines
slouken@0
   121
 */
slouken@0
   122
slouken@2249
   123
#if defined(__GNUC__)
slouken@2249
   124
#define DECLARE_ALIGNED(t,v,a)  t __attribute__((aligned(a))) v
slouken@2249
   125
#elif defined(_MSC_VER)
slouken@2251
   126
#define DECLARE_ALIGNED(t,v,a)  __declspec(align(a)) t v
slouken@2249
   127
#else
slouken@2249
   128
#define DECLARE_ALIGNED(t,v,a)  t v
slouken@2249
   129
#endif
slouken@2249
   130
slouken@0
   131
#define FORMAT_EQUAL(A, B)						\
slouken@0
   132
    ((A)->BitsPerPixel == (B)->BitsPerPixel				\
slouken@0
   133
     && ((A)->Rmask == (B)->Rmask) && ((A)->Amask == (B)->Amask))
slouken@0
   134
slouken@0
   135
/* Load pixel of the specified format from a buffer and get its R-G-B values */
slouken@0
   136
/* FIXME: rescale values to 0..255 here? */
icculus@1162
   137
#define RGB_FROM_PIXEL(Pixel, fmt, r, g, b)				\
slouken@0
   138
{									\
icculus@1162
   139
	r = (((Pixel&fmt->Rmask)>>fmt->Rshift)<<fmt->Rloss); 		\
icculus@1162
   140
	g = (((Pixel&fmt->Gmask)>>fmt->Gshift)<<fmt->Gloss); 		\
icculus@1162
   141
	b = (((Pixel&fmt->Bmask)>>fmt->Bshift)<<fmt->Bloss); 		\
slouken@0
   142
}
icculus@1162
   143
#define RGB_FROM_RGB565(Pixel, r, g, b)					\
slouken@0
   144
{									\
icculus@1162
   145
	r = (((Pixel&0xF800)>>11)<<3);		 			\
icculus@1162
   146
	g = (((Pixel&0x07E0)>>5)<<2); 					\
icculus@1162
   147
	b = ((Pixel&0x001F)<<3); 					\
slouken@0
   148
}
icculus@1162
   149
#define RGB_FROM_RGB555(Pixel, r, g, b)					\
slouken@0
   150
{									\
icculus@1162
   151
	r = (((Pixel&0x7C00)>>10)<<3);		 			\
icculus@1162
   152
	g = (((Pixel&0x03E0)>>5)<<3); 					\
icculus@1162
   153
	b = ((Pixel&0x001F)<<3); 					\
slouken@0
   154
}
icculus@1162
   155
#define RGB_FROM_RGB888(Pixel, r, g, b)					\
slouken@0
   156
{									\
icculus@1162
   157
	r = ((Pixel&0xFF0000)>>16);		 			\
icculus@1162
   158
	g = ((Pixel&0xFF00)>>8);		 			\
icculus@1162
   159
	b = (Pixel&0xFF);			 			\
slouken@0
   160
}
icculus@1162
   161
#define RETRIEVE_RGB_PIXEL(buf, bpp, Pixel)				   \
slouken@0
   162
do {									   \
slouken@0
   163
	switch (bpp) {							   \
slouken@0
   164
		case 2:							   \
icculus@1162
   165
			Pixel = *((Uint16 *)(buf));			   \
slouken@0
   166
		break;							   \
slouken@0
   167
									   \
slouken@0
   168
		case 3: {						   \
slouken@0
   169
		        Uint8 *B = (Uint8 *)(buf);			   \
slouken@2824
   170
			if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		   \
icculus@1162
   171
			        Pixel = B[0] + (B[1] << 8) + (B[2] << 16); \
slouken@0
   172
			} else {					   \
icculus@1162
   173
			        Pixel = (B[0] << 16) + (B[1] << 8) + B[2]; \
slouken@0
   174
			}						   \
slouken@0
   175
		}							   \
slouken@0
   176
		break;							   \
slouken@0
   177
									   \
slouken@0
   178
		case 4:							   \
icculus@1162
   179
			Pixel = *((Uint32 *)(buf));			   \
slouken@0
   180
		break;							   \
slouken@0
   181
									   \
slouken@0
   182
		default:						   \
slouken@2824
   183
		        Pixel; /* stop gcc complaints */		   \
slouken@0
   184
		break;							   \
slouken@0
   185
	}								   \
slouken@2824
   186
} while (0)
slouken@0
   187
icculus@1162
   188
#define DISEMBLE_RGB(buf, bpp, fmt, Pixel, r, g, b)			   \
slouken@0
   189
do {									   \
slouken@0
   190
	switch (bpp) {							   \
slouken@0
   191
		case 2:							   \
icculus@1162
   192
			Pixel = *((Uint16 *)(buf));			   \
slouken@2824
   193
			RGB_FROM_PIXEL(Pixel, fmt, r, g, b);		   \
slouken@0
   194
		break;							   \
slouken@0
   195
									   \
slouken@2824
   196
		case 3:	{						   \
slouken@2824
   197
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		   \
slouken@2824
   198
			        r = *((buf)+fmt->Rshift/8);		   \
slouken@2824
   199
				g = *((buf)+fmt->Gshift/8);		   \
slouken@2824
   200
				b = *((buf)+fmt->Bshift/8);		   \
slouken@0
   201
			} else {					   \
slouken@2824
   202
			        r = *((buf)+2-fmt->Rshift/8);		   \
slouken@2824
   203
				g = *((buf)+2-fmt->Gshift/8);		   \
slouken@2824
   204
				b = *((buf)+2-fmt->Bshift/8);		   \
slouken@0
   205
			}						   \
slouken@0
   206
		}							   \
slouken@0
   207
		break;							   \
slouken@0
   208
									   \
slouken@0
   209
		case 4:							   \
icculus@1162
   210
			Pixel = *((Uint32 *)(buf));			   \
slouken@2824
   211
			RGB_FROM_PIXEL(Pixel, fmt, r, g, b);		   \
slouken@0
   212
		break;							   \
slouken@0
   213
									   \
slouken@2824
   214
		default:						   \
slouken@2824
   215
		        Pixel; /* stop gcc complaints */		   \
slouken@0
   216
		break;							   \
slouken@0
   217
	}								   \
slouken@2824
   218
} while (0)
slouken@0
   219
slouken@0
   220
/* Assemble R-G-B values into a specified pixel format and store them */
icculus@1162
   221
#define PIXEL_FROM_RGB(Pixel, fmt, r, g, b)				\
slouken@0
   222
{									\
icculus@1162
   223
	Pixel = ((r>>fmt->Rloss)<<fmt->Rshift)|				\
slouken@0
   224
		((g>>fmt->Gloss)<<fmt->Gshift)|				\
slouken@0
   225
		((b>>fmt->Bloss)<<fmt->Bshift);				\
slouken@0
   226
}
icculus@1162
   227
#define RGB565_FROM_RGB(Pixel, r, g, b)					\
slouken@0
   228
{									\
icculus@1162
   229
	Pixel = ((r>>3)<<11)|((g>>2)<<5)|(b>>3);			\
slouken@0
   230
}
icculus@1162
   231
#define RGB555_FROM_RGB(Pixel, r, g, b)					\
slouken@0
   232
{									\
icculus@1162
   233
	Pixel = ((r>>3)<<10)|((g>>3)<<5)|(b>>3);			\
slouken@0
   234
}
icculus@1162
   235
#define RGB888_FROM_RGB(Pixel, r, g, b)					\
slouken@0
   236
{									\
icculus@1162
   237
	Pixel = (r<<16)|(g<<8)|b;					\
slouken@0
   238
}
slouken@2899
   239
#define ARGB8888_FROM_RGBA(Pixel, r, g, b, a)				\
slouken@2899
   240
{									\
slouken@2899
   241
	Pixel = (a<<24)|(r<<16)|(g<<8)|b;				\
slouken@2899
   242
}
slouken@0
   243
#define ASSEMBLE_RGB(buf, bpp, fmt, r, g, b) 				\
slouken@0
   244
{									\
slouken@0
   245
	switch (bpp) {							\
slouken@0
   246
		case 2: {						\
icculus@1162
   247
			Uint16 Pixel;					\
slouken@0
   248
									\
icculus@1162
   249
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   250
			*((Uint16 *)(buf)) = Pixel;			\
slouken@0
   251
		}							\
slouken@0
   252
		break;							\
slouken@0
   253
									\
slouken@0
   254
		case 3: {						\
slouken@2824
   255
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		\
slouken@0
   256
			        *((buf)+fmt->Rshift/8) = r;		\
slouken@0
   257
				*((buf)+fmt->Gshift/8) = g;		\
slouken@0
   258
				*((buf)+fmt->Bshift/8) = b;		\
slouken@0
   259
			} else {					\
slouken@0
   260
			        *((buf)+2-fmt->Rshift/8) = r;		\
slouken@0
   261
				*((buf)+2-fmt->Gshift/8) = g;		\
slouken@0
   262
				*((buf)+2-fmt->Bshift/8) = b;		\
slouken@0
   263
			}						\
slouken@0
   264
		}							\
slouken@0
   265
		break;							\
slouken@0
   266
									\
slouken@0
   267
		case 4: {						\
icculus@1162
   268
			Uint32 Pixel;					\
slouken@0
   269
									\
icculus@1162
   270
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   271
			*((Uint32 *)(buf)) = Pixel;			\
slouken@0
   272
		}							\
slouken@0
   273
		break;							\
slouken@0
   274
	}								\
slouken@0
   275
}
slouken@0
   276
#define ASSEMBLE_RGB_AMASK(buf, bpp, fmt, r, g, b, Amask)		\
slouken@0
   277
{									\
slouken@0
   278
	switch (bpp) {							\
slouken@0
   279
		case 2: {						\
slouken@0
   280
			Uint16 *bufp;					\
icculus@1162
   281
			Uint16 Pixel;					\
slouken@0
   282
									\
slouken@0
   283
			bufp = (Uint16 *)buf;				\
icculus@1162
   284
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   285
			*bufp = Pixel | (*bufp & Amask);		\
slouken@0
   286
		}							\
slouken@0
   287
		break;							\
slouken@0
   288
									\
slouken@0
   289
		case 3: {						\
slouken@2824
   290
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		\
slouken@0
   291
			        *((buf)+fmt->Rshift/8) = r;		\
slouken@0
   292
				*((buf)+fmt->Gshift/8) = g;		\
slouken@0
   293
				*((buf)+fmt->Bshift/8) = b;		\
slouken@0
   294
			} else {					\
slouken@0
   295
			        *((buf)+2-fmt->Rshift/8) = r;		\
slouken@0
   296
				*((buf)+2-fmt->Gshift/8) = g;		\
slouken@0
   297
				*((buf)+2-fmt->Bshift/8) = b;		\
slouken@0
   298
			}						\
slouken@0
   299
		}							\
slouken@0
   300
		break;							\
slouken@0
   301
									\
slouken@0
   302
		case 4: {						\
slouken@0
   303
			Uint32 *bufp;					\
icculus@1162
   304
			Uint32 Pixel;					\
slouken@0
   305
									\
slouken@0
   306
			bufp = (Uint32 *)buf;				\
icculus@1162
   307
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   308
			*bufp = Pixel | (*bufp & Amask);		\
slouken@0
   309
		}							\
slouken@0
   310
		break;							\
slouken@0
   311
	}								\
slouken@0
   312
}
slouken@0
   313
slouken@0
   314
/* FIXME: Should we rescale alpha into 0..255 here? */
icculus@1162
   315
#define RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a)				\
slouken@0
   316
{									\
icculus@1162
   317
	r = ((Pixel&fmt->Rmask)>>fmt->Rshift)<<fmt->Rloss; 		\
icculus@1162
   318
	g = ((Pixel&fmt->Gmask)>>fmt->Gshift)<<fmt->Gloss; 		\
icculus@1162
   319
	b = ((Pixel&fmt->Bmask)>>fmt->Bshift)<<fmt->Bloss; 		\
icculus@1162
   320
	a = ((Pixel&fmt->Amask)>>fmt->Ashift)<<fmt->Aloss;	 	\
slouken@0
   321
}
icculus@1162
   322
#define RGBA_FROM_8888(Pixel, fmt, r, g, b, a)	\
slouken@0
   323
{						\
icculus@1162
   324
	r = (Pixel&fmt->Rmask)>>fmt->Rshift;	\
icculus@1162
   325
	g = (Pixel&fmt->Gmask)>>fmt->Gshift;	\
icculus@1162
   326
	b = (Pixel&fmt->Bmask)>>fmt->Bshift;	\
icculus@1162
   327
	a = (Pixel&fmt->Amask)>>fmt->Ashift;	\
slouken@0
   328
}
icculus@1162
   329
#define RGBA_FROM_RGBA8888(Pixel, r, g, b, a)				\
slouken@0
   330
{									\
icculus@1162
   331
	r = (Pixel>>24);						\
icculus@1162
   332
	g = ((Pixel>>16)&0xFF);						\
icculus@1162
   333
	b = ((Pixel>>8)&0xFF);						\
icculus@1162
   334
	a = (Pixel&0xFF);						\
slouken@0
   335
}
icculus@1162
   336
#define RGBA_FROM_ARGB8888(Pixel, r, g, b, a)				\
slouken@0
   337
{									\
icculus@1162
   338
	r = ((Pixel>>16)&0xFF);						\
icculus@1162
   339
	g = ((Pixel>>8)&0xFF);						\
icculus@1162
   340
	b = (Pixel&0xFF);						\
icculus@1162
   341
	a = (Pixel>>24);						\
slouken@0
   342
}
icculus@1162
   343
#define RGBA_FROM_ABGR8888(Pixel, r, g, b, a)				\
slouken@0
   344
{									\
icculus@1162
   345
	r = (Pixel&0xFF);						\
icculus@1162
   346
	g = ((Pixel>>8)&0xFF);						\
icculus@1162
   347
	b = ((Pixel>>16)&0xFF);						\
icculus@1162
   348
	a = (Pixel>>24);						\
slouken@0
   349
}
icculus@1162
   350
#define DISEMBLE_RGBA(buf, bpp, fmt, Pixel, r, g, b, a)			   \
slouken@0
   351
do {									   \
slouken@0
   352
	switch (bpp) {							   \
slouken@0
   353
		case 2:							   \
icculus@1162
   354
			Pixel = *((Uint16 *)(buf));			   \
slouken@2824
   355
			RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a);	   \
slouken@0
   356
		break;							   \
slouken@0
   357
									   \
slouken@2824
   358
		case 3:	{						   \
slouken@2824
   359
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		   \
slouken@2824
   360
			        r = *((buf)+fmt->Rshift/8);		   \
slouken@2824
   361
				g = *((buf)+fmt->Gshift/8);		   \
slouken@2824
   362
				b = *((buf)+fmt->Bshift/8);		   \
slouken@0
   363
			} else {					   \
slouken@2824
   364
			        r = *((buf)+2-fmt->Rshift/8);		   \
slouken@2824
   365
				g = *((buf)+2-fmt->Gshift/8);		   \
slouken@2824
   366
				b = *((buf)+2-fmt->Bshift/8);		   \
slouken@0
   367
			}						   \
slouken@2824
   368
			a = 0xFF;					   \
slouken@0
   369
		}							   \
slouken@0
   370
		break;							   \
slouken@0
   371
									   \
slouken@0
   372
		case 4:							   \
icculus@1162
   373
			Pixel = *((Uint32 *)(buf));			   \
slouken@2824
   374
			RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a);	   \
slouken@0
   375
		break;							   \
slouken@0
   376
									   \
slouken@0
   377
		default:						   \
slouken@2824
   378
		        Pixel; /* stop gcc complaints */		   \
slouken@0
   379
		break;							   \
slouken@0
   380
	}								   \
slouken@2824
   381
} while (0)
slouken@0
   382
slouken@0
   383
/* FIXME: this isn't correct, especially for Alpha (maximum != 255) */
icculus@1162
   384
#define PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a)				\
slouken@0
   385
{									\
icculus@1162
   386
	Pixel = ((r>>fmt->Rloss)<<fmt->Rshift)|				\
slouken@0
   387
		((g>>fmt->Gloss)<<fmt->Gshift)|				\
slouken@0
   388
		((b>>fmt->Bloss)<<fmt->Bshift)|				\
slouken@535
   389
		((a>>fmt->Aloss)<<fmt->Ashift);				\
slouken@0
   390
}
slouken@0
   391
#define ASSEMBLE_RGBA(buf, bpp, fmt, r, g, b, a)			\
slouken@0
   392
{									\
slouken@0
   393
	switch (bpp) {							\
slouken@0
   394
		case 2: {						\
icculus@1162
   395
			Uint16 Pixel;					\
slouken@0
   396
									\
icculus@1162
   397
			PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a);	\
icculus@1162
   398
			*((Uint16 *)(buf)) = Pixel;			\
slouken@0
   399
		}							\
slouken@0
   400
		break;							\
slouken@0
   401
									\
slouken@2824
   402
		case 3: {						\
slouken@2824
   403
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		\
slouken@0
   404
			        *((buf)+fmt->Rshift/8) = r;		\
slouken@0
   405
				*((buf)+fmt->Gshift/8) = g;		\
slouken@0
   406
				*((buf)+fmt->Bshift/8) = b;		\
slouken@0
   407
			} else {					\
slouken@0
   408
			        *((buf)+2-fmt->Rshift/8) = r;		\
slouken@0
   409
				*((buf)+2-fmt->Gshift/8) = g;		\
slouken@0
   410
				*((buf)+2-fmt->Bshift/8) = b;		\
slouken@0
   411
			}						\
slouken@0
   412
		}							\
slouken@0
   413
		break;							\
slouken@0
   414
									\
slouken@0
   415
		case 4: {						\
icculus@1162
   416
			Uint32 Pixel;					\
slouken@0
   417
									\
icculus@1162
   418
			PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a);	\
icculus@1162
   419
			*((Uint32 *)(buf)) = Pixel;			\
slouken@0
   420
		}							\
slouken@0
   421
		break;							\
slouken@0
   422
	}								\
slouken@0
   423
}
slouken@0
   424
icculus@1162
   425
/* Blend the RGB values of two Pixels based on a source alpha value */
slouken@0
   426
#define ALPHA_BLEND(sR, sG, sB, A, dR, dG, dB)	\
slouken@0
   427
do {						\
slouken@0
   428
	dR = (((sR-dR)*(A))>>8)+dR;		\
slouken@0
   429
	dG = (((sG-dG)*(A))>>8)+dG;		\
slouken@0
   430
	dB = (((sB-dB)*(A))>>8)+dB;		\
slouken@0
   431
} while(0)
slouken@0
   432
icculus@1162
   433
/* Blend the RGB values of two Pixels based on a source alpha value */
icculus@1047
   434
#define ACCURATE_ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB)	\
icculus@1047
   435
do {						\
icculus@1047
   436
    unsigned tR, tG, tB, tA; \
icculus@1047
   437
    tA = 255 - sA; \
icculus@1047
   438
    tR = 1 + (sR * sA) + (dR * tA); \
icculus@1047
   439
    dR = (tR + (tR >> 8)) >> 8; \
icculus@1047
   440
    tG = 1 + (sG * sA) + (dG * tA); \
icculus@1047
   441
    dG = (tG + (tG >> 8)) >> 8; \
icculus@1047
   442
    tB = 1 + (sB * sA) + (dB * tA); \
icculus@1047
   443
    dB = (tB + (tB >> 8)) >> 8; \
icculus@1047
   444
} while(0)
icculus@1047
   445
icculus@1047
   446
slouken@0
   447
/* This is a very useful loop for optimizing blitters */
slouken@553
   448
#if defined(_MSC_VER) && (_MSC_VER == 1300)
slouken@553
   449
/* There's a bug in the Visual C++ 7 optimizer when compiling this code */
slouken@553
   450
#else
slouken@0
   451
#define USE_DUFFS_LOOP
slouken@553
   452
#endif
slouken@0
   453
#ifdef USE_DUFFS_LOOP
slouken@0
   454
slouken@0
   455
/* 8-times unrolled loop */
slouken@0
   456
#define DUFFS_LOOP8(pixel_copy_increment, width)			\
slouken@0
   457
{ int n = (width+7)/8;							\
slouken@91
   458
	switch (width & 7) {						\
slouken@0
   459
	case 0: do {	pixel_copy_increment;				\
slouken@0
   460
	case 7:		pixel_copy_increment;				\
slouken@0
   461
	case 6:		pixel_copy_increment;				\
slouken@0
   462
	case 5:		pixel_copy_increment;				\
slouken@0
   463
	case 4:		pixel_copy_increment;				\
slouken@0
   464
	case 3:		pixel_copy_increment;				\
slouken@0
   465
	case 2:		pixel_copy_increment;				\
slouken@0
   466
	case 1:		pixel_copy_increment;				\
slouken@0
   467
		} while ( --n > 0 );					\
slouken@0
   468
	}								\
slouken@0
   469
}
slouken@0
   470
slouken@0
   471
/* 4-times unrolled loop */
slouken@0
   472
#define DUFFS_LOOP4(pixel_copy_increment, width)			\
slouken@0
   473
{ int n = (width+3)/4;							\
slouken@91
   474
	switch (width & 3) {						\
slouken@0
   475
	case 0: do {	pixel_copy_increment;				\
slouken@0
   476
	case 3:		pixel_copy_increment;				\
slouken@0
   477
	case 2:		pixel_copy_increment;				\
slouken@0
   478
	case 1:		pixel_copy_increment;				\
slouken@0
   479
		} while ( --n > 0 );					\
slouken@0
   480
	}								\
slouken@0
   481
}
slouken@0
   482
slouken@689
   483
/* 2 - times unrolled loop */
slouken@689
   484
#define DUFFS_LOOP_DOUBLE2(pixel_copy_increment,			\
slouken@689
   485
				double_pixel_copy_increment, width)	\
slouken@689
   486
{ int n, w = width;							\
slouken@689
   487
	if( w & 1 ) {							\
slouken@689
   488
	    pixel_copy_increment;					\
slouken@689
   489
	    w--;							\
slouken@689
   490
	}								\
slouken@689
   491
	if ( w > 0 )	{						\
slouken@689
   492
	    n = ( w + 2) / 4;						\
slouken@689
   493
	    switch( w & 2 ) {						\
slouken@689
   494
	    case 0: do {	double_pixel_copy_increment;		\
slouken@689
   495
	    case 2:		double_pixel_copy_increment;		\
slouken@689
   496
		    } while ( --n > 0 );					\
slouken@689
   497
	    }								\
slouken@689
   498
	}								\
slouken@689
   499
}
slouken@689
   500
slouken@689
   501
/* 2 - times unrolled loop 4 pixels */
slouken@689
   502
#define DUFFS_LOOP_QUATRO2(pixel_copy_increment,			\
slouken@689
   503
				double_pixel_copy_increment,		\
slouken@689
   504
				quatro_pixel_copy_increment, width)	\
slouken@689
   505
{ int n, w = width;								\
slouken@689
   506
        if(w & 1) {							\
slouken@689
   507
	  pixel_copy_increment;						\
slouken@689
   508
	  w--;								\
slouken@689
   509
	}								\
slouken@689
   510
	if(w & 2) {							\
slouken@689
   511
	  double_pixel_copy_increment;					\
slouken@689
   512
	  w -= 2;							\
slouken@689
   513
	}								\
slouken@689
   514
	if ( w > 0 ) {							\
slouken@689
   515
	    n = ( w + 7 ) / 8;						\
slouken@689
   516
	    switch( w & 4 ) {						\
slouken@689
   517
	    case 0: do {	quatro_pixel_copy_increment;		\
slouken@689
   518
	    case 4:		quatro_pixel_copy_increment;		\
slouken@689
   519
		    } while ( --n > 0 );					\
slouken@689
   520
	    }								\
slouken@689
   521
	}								\
slouken@689
   522
}
slouken@689
   523
slouken@0
   524
/* Use the 8-times version of the loop by default */
slouken@0
   525
#define DUFFS_LOOP(pixel_copy_increment, width)				\
slouken@0
   526
	DUFFS_LOOP8(pixel_copy_increment, width)
slouken@0
   527
slouken@0
   528
#else
slouken@0
   529
slouken@0
   530
/* Don't use Duff's device to unroll loops */
slouken@689
   531
#define DUFFS_LOOP_DOUBLE2(pixel_copy_increment,			\
slouken@689
   532
			 double_pixel_copy_increment, width)		\
slouken@689
   533
{ int n = width;								\
slouken@689
   534
    if( n & 1 ) {							\
slouken@689
   535
	pixel_copy_increment;						\
slouken@689
   536
	n--;								\
slouken@689
   537
    }									\
slouken@689
   538
    n=n>>1;								\
slouken@689
   539
    for(; n > 0; --n) {   						\
slouken@689
   540
	double_pixel_copy_increment;					\
slouken@689
   541
    }									\
slouken@689
   542
}
slouken@689
   543
slouken@689
   544
/* Don't use Duff's device to unroll loops */
slouken@689
   545
#define DUFFS_LOOP_QUATRO2(pixel_copy_increment,			\
slouken@689
   546
				double_pixel_copy_increment,		\
slouken@689
   547
				quatro_pixel_copy_increment, width)	\
slouken@689
   548
{ int n = width;								\
slouken@689
   549
        if(n & 1) {							\
slouken@689
   550
	  pixel_copy_increment;						\
slouken@689
   551
	  n--;								\
slouken@689
   552
	}								\
slouken@689
   553
	if(n & 2) {							\
slouken@689
   554
	  double_pixel_copy_increment;					\
slouken@689
   555
	  n -= 2;							\
slouken@689
   556
	}								\
slouken@689
   557
	n=n>>2;								\
slouken@689
   558
	for(; n > 0; --n) {   						\
slouken@689
   559
	  quatro_pixel_copy_increment;					\
slouken@689
   560
        }								\
slouken@689
   561
}
slouken@689
   562
slouken@689
   563
/* Don't use Duff's device to unroll loops */
slouken@0
   564
#define DUFFS_LOOP(pixel_copy_increment, width)				\
slouken@0
   565
{ int n;								\
slouken@0
   566
	for ( n=width; n > 0; --n ) {					\
slouken@0
   567
		pixel_copy_increment;					\
slouken@0
   568
	}								\
slouken@0
   569
}
slouken@0
   570
#define DUFFS_LOOP8(pixel_copy_increment, width)			\
slouken@0
   571
	DUFFS_LOOP(pixel_copy_increment, width)
slouken@0
   572
#define DUFFS_LOOP4(pixel_copy_increment, width)			\
slouken@0
   573
	DUFFS_LOOP(pixel_copy_increment, width)
slouken@0
   574
slouken@0
   575
#endif /* USE_DUFFS_LOOP */
slouken@0
   576
slouken@0
   577
/* Prevent Visual C++ 6.0 from printing out stupid warnings */
slouken@0
   578
#if defined(_MSC_VER) && (_MSC_VER >= 600)
slouken@0
   579
#pragma warning(disable: 4550)
slouken@0
   580
#endif
slouken@0
   581
slouken@0
   582
#endif /* _SDL_blit_h */
slouken@2898
   583
slouken@1895
   584
/* vi: set ts=4 sw=4 expandtab: */