src/video/SDL_blit.h
author Sam Lantinga
Sun, 13 Feb 2011 13:46:10 -0800
changeset 5288 d4381f3b0d1e
parent 5262 b530ef003506
child 5389 24903690f48a
permissions -rw-r--r--
A few fixes:
Fixed creating render texture framebuffer.
Removed the need for palette watch, added surface format caching.
Added an SDL_DONTFREE flag so you can't free the window and 1.2 shadow surfaces.
slouken@0
     1
/*
slouken@0
     2
    SDL - Simple DirectMedia Layer
slouken@5262
     3
    Copyright (C) 1997-2011 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@4426
    27
#ifdef __MINGW32__
slouken@4426
    28
#include <_mingw.h>
slouken@4426
    29
#endif
slouken@4426
    30
slouken@4426
    31
#if defined(__MINGW32__) && defined(__MINGW64_VERSION_MAJOR)
slouken@4426
    32
#include <intrin.h>
slouken@4426
    33
#else
slouken@2249
    34
#ifdef __MMX__
slouken@2249
    35
#include <mmintrin.h>
slouken@2249
    36
#endif
slouken@2249
    37
#ifdef __SSE__
slouken@2249
    38
#include <xmmintrin.h>
slouken@2249
    39
#endif
slouken@2262
    40
#ifdef __SSE2__
slouken@2262
    41
#include <emmintrin.h>
slouken@2262
    42
#endif
slouken@4426
    43
#endif
slouken@2249
    44
slouken@2250
    45
#include "SDL_cpuinfo.h"
slouken@0
    46
#include "SDL_endian.h"
slouken@5163
    47
#include "SDL_surface.h"
slouken@0
    48
slouken@2262
    49
/* SDL blit copy flags */
slouken@2266
    50
#define SDL_COPY_MODULATE_COLOR     0x00000001
slouken@2266
    51
#define SDL_COPY_MODULATE_ALPHA     0x00000002
slouken@5140
    52
#define SDL_COPY_BLEND              0x00000010
slouken@5140
    53
#define SDL_COPY_ADD                0x00000020
slouken@5184
    54
#define SDL_COPY_MOD                0x00000040
slouken@2266
    55
#define SDL_COPY_COLORKEY           0x00000100
slouken@2266
    56
#define SDL_COPY_NEAREST            0x00000200
slouken@2266
    57
#define SDL_COPY_RLE_DESIRED        0x00001000
slouken@2266
    58
#define SDL_COPY_RLE_COLORKEY       0x00002000
slouken@2266
    59
#define SDL_COPY_RLE_ALPHAKEY       0x00004000
slouken@2853
    60
#define SDL_COPY_RLE_MASK           (SDL_COPY_RLE_DESIRED|SDL_COPY_RLE_COLORKEY|SDL_COPY_RLE_ALPHAKEY)
slouken@2262
    61
slouken@2262
    62
/* SDL blit CPU flags */
slouken@2266
    63
#define SDL_CPU_ANY                 0x00000000
slouken@2266
    64
#define SDL_CPU_MMX                 0x00000001
slouken@2266
    65
#define SDL_CPU_SSE                 0x00000004
slouken@2266
    66
#define SDL_CPU_SSE2                0x00000008
slouken@2262
    67
slouken@2267
    68
typedef struct
slouken@2267
    69
{
slouken@2262
    70
    Uint8 *src;
slouken@2262
    71
    int src_w, src_h;
slouken@2262
    72
    int src_pitch;
slouken@2267
    73
    int src_skip;
slouken@2262
    74
    Uint8 *dst;
slouken@2262
    75
    int dst_w, dst_h;
slouken@2262
    76
    int dst_pitch;
slouken@2267
    77
    int dst_skip;
slouken@2262
    78
    SDL_PixelFormat *src_fmt;
slouken@2262
    79
    SDL_PixelFormat *dst_fmt;
slouken@1895
    80
    Uint8 *table;
slouken@2262
    81
    int flags;
slouken@2262
    82
    Uint32 colorkey;
slouken@2262
    83
    Uint8 r, g, b, a;
slouken@0
    84
} SDL_BlitInfo;
slouken@0
    85
slouken@2267
    86
typedef void (SDLCALL * SDL_BlitFunc) (SDL_BlitInfo * info);
slouken@2262
    87
slouken@2267
    88
typedef struct
slouken@2267
    89
{
slouken@2262
    90
    Uint32 src_format;
slouken@2262
    91
    Uint32 dst_format;
slouken@2262
    92
    int flags;
slouken@2262
    93
    int cpu;
slouken@2262
    94
    SDL_BlitFunc func;
slouken@2262
    95
} SDL_BlitFuncEntry;
slouken@0
    96
slouken@0
    97
/* Blit mapping definition */
slouken@1895
    98
typedef struct SDL_BlitMap
slouken@1895
    99
{
slouken@1895
   100
    SDL_Surface *dst;
slouken@1895
   101
    int identity;
slouken@2257
   102
    SDL_blit blit;
slouken@2257
   103
    void *data;
slouken@2262
   104
    SDL_BlitInfo info;
slouken@0
   105
slouken@1895
   106
    /* the version count matches the destination; mismatch indicates
slouken@1895
   107
       an invalid mapping */
slouken@5288
   108
    Uint32 palette_version;
slouken@0
   109
} SDL_BlitMap;
slouken@0
   110
slouken@0
   111
/* Functions found in SDL_blit.c */
slouken@1895
   112
extern int SDL_CalculateBlit(SDL_Surface * surface);
slouken@0
   113
slouken@2267
   114
/* Functions found in SDL_blit_*.c */
slouken@2267
   115
extern SDL_BlitFunc SDL_CalculateBlit0(SDL_Surface * surface);
slouken@2267
   116
extern SDL_BlitFunc SDL_CalculateBlit1(SDL_Surface * surface);
slouken@2267
   117
extern SDL_BlitFunc SDL_CalculateBlitN(SDL_Surface * surface);
slouken@2267
   118
extern SDL_BlitFunc SDL_CalculateBlitA(SDL_Surface * surface);
slouken@2263
   119
slouken@0
   120
/*
slouken@0
   121
 * Useful macros for blitting routines
slouken@0
   122
 */
slouken@0
   123
slouken@2249
   124
#if defined(__GNUC__)
slouken@2249
   125
#define DECLARE_ALIGNED(t,v,a)  t __attribute__((aligned(a))) v
slouken@2249
   126
#elif defined(_MSC_VER)
slouken@2251
   127
#define DECLARE_ALIGNED(t,v,a)  __declspec(align(a)) t v
slouken@2249
   128
#else
slouken@2249
   129
#define DECLARE_ALIGNED(t,v,a)  t v
slouken@2249
   130
#endif
slouken@2249
   131
slouken@0
   132
/* Load pixel of the specified format from a buffer and get its R-G-B values */
slouken@0
   133
/* FIXME: rescale values to 0..255 here? */
icculus@1162
   134
#define RGB_FROM_PIXEL(Pixel, fmt, r, g, b)				\
slouken@0
   135
{									\
icculus@1162
   136
	r = (((Pixel&fmt->Rmask)>>fmt->Rshift)<<fmt->Rloss); 		\
icculus@1162
   137
	g = (((Pixel&fmt->Gmask)>>fmt->Gshift)<<fmt->Gloss); 		\
icculus@1162
   138
	b = (((Pixel&fmt->Bmask)>>fmt->Bshift)<<fmt->Bloss); 		\
slouken@0
   139
}
icculus@1162
   140
#define RGB_FROM_RGB565(Pixel, r, g, b)					\
slouken@0
   141
{									\
icculus@1162
   142
	r = (((Pixel&0xF800)>>11)<<3);		 			\
icculus@1162
   143
	g = (((Pixel&0x07E0)>>5)<<2); 					\
icculus@1162
   144
	b = ((Pixel&0x001F)<<3); 					\
slouken@0
   145
}
icculus@1162
   146
#define RGB_FROM_RGB555(Pixel, r, g, b)					\
slouken@0
   147
{									\
icculus@1162
   148
	r = (((Pixel&0x7C00)>>10)<<3);		 			\
icculus@1162
   149
	g = (((Pixel&0x03E0)>>5)<<3); 					\
icculus@1162
   150
	b = ((Pixel&0x001F)<<3); 					\
slouken@0
   151
}
icculus@1162
   152
#define RGB_FROM_RGB888(Pixel, r, g, b)					\
slouken@0
   153
{									\
icculus@1162
   154
	r = ((Pixel&0xFF0000)>>16);		 			\
icculus@1162
   155
	g = ((Pixel&0xFF00)>>8);		 			\
icculus@1162
   156
	b = (Pixel&0xFF);			 			\
slouken@0
   157
}
icculus@1162
   158
#define RETRIEVE_RGB_PIXEL(buf, bpp, Pixel)				   \
slouken@0
   159
do {									   \
slouken@0
   160
	switch (bpp) {							   \
slouken@0
   161
		case 2:							   \
icculus@1162
   162
			Pixel = *((Uint16 *)(buf));			   \
slouken@0
   163
		break;							   \
slouken@0
   164
									   \
slouken@0
   165
		case 3: {						   \
slouken@0
   166
		        Uint8 *B = (Uint8 *)(buf);			   \
slouken@2824
   167
			if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		   \
icculus@1162
   168
			        Pixel = B[0] + (B[1] << 8) + (B[2] << 16); \
slouken@0
   169
			} else {					   \
icculus@1162
   170
			        Pixel = (B[0] << 16) + (B[1] << 8) + B[2]; \
slouken@0
   171
			}						   \
slouken@0
   172
		}							   \
slouken@0
   173
		break;							   \
slouken@0
   174
									   \
slouken@0
   175
		case 4:							   \
icculus@1162
   176
			Pixel = *((Uint32 *)(buf));			   \
slouken@0
   177
		break;							   \
slouken@0
   178
									   \
slouken@0
   179
		default:						   \
slouken@2824
   180
		        Pixel; /* stop gcc complaints */		   \
slouken@0
   181
		break;							   \
slouken@0
   182
	}								   \
slouken@2824
   183
} while (0)
slouken@0
   184
icculus@1162
   185
#define DISEMBLE_RGB(buf, bpp, fmt, Pixel, r, g, b)			   \
slouken@0
   186
do {									   \
slouken@0
   187
	switch (bpp) {							   \
slouken@0
   188
		case 2:							   \
icculus@1162
   189
			Pixel = *((Uint16 *)(buf));			   \
slouken@2824
   190
			RGB_FROM_PIXEL(Pixel, fmt, r, g, b);		   \
slouken@0
   191
		break;							   \
slouken@0
   192
									   \
slouken@2824
   193
		case 3:	{						   \
slouken@2824
   194
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		   \
slouken@2824
   195
			        r = *((buf)+fmt->Rshift/8);		   \
slouken@2824
   196
				g = *((buf)+fmt->Gshift/8);		   \
slouken@2824
   197
				b = *((buf)+fmt->Bshift/8);		   \
slouken@0
   198
			} else {					   \
slouken@2824
   199
			        r = *((buf)+2-fmt->Rshift/8);		   \
slouken@2824
   200
				g = *((buf)+2-fmt->Gshift/8);		   \
slouken@2824
   201
				b = *((buf)+2-fmt->Bshift/8);		   \
slouken@0
   202
			}						   \
slouken@0
   203
		}							   \
slouken@0
   204
		break;							   \
slouken@0
   205
									   \
slouken@0
   206
		case 4:							   \
icculus@1162
   207
			Pixel = *((Uint32 *)(buf));			   \
slouken@2824
   208
			RGB_FROM_PIXEL(Pixel, fmt, r, g, b);		   \
slouken@0
   209
		break;							   \
slouken@0
   210
									   \
slouken@2824
   211
		default:						   \
slouken@2824
   212
		        Pixel; /* stop gcc complaints */		   \
slouken@0
   213
		break;							   \
slouken@0
   214
	}								   \
slouken@2824
   215
} while (0)
slouken@0
   216
slouken@0
   217
/* Assemble R-G-B values into a specified pixel format and store them */
icculus@1162
   218
#define PIXEL_FROM_RGB(Pixel, fmt, r, g, b)				\
slouken@0
   219
{									\
icculus@1162
   220
	Pixel = ((r>>fmt->Rloss)<<fmt->Rshift)|				\
slouken@0
   221
		((g>>fmt->Gloss)<<fmt->Gshift)|				\
slouken@0
   222
		((b>>fmt->Bloss)<<fmt->Bshift);				\
slouken@0
   223
}
icculus@1162
   224
#define RGB565_FROM_RGB(Pixel, r, g, b)					\
slouken@0
   225
{									\
icculus@1162
   226
	Pixel = ((r>>3)<<11)|((g>>2)<<5)|(b>>3);			\
slouken@0
   227
}
icculus@1162
   228
#define RGB555_FROM_RGB(Pixel, r, g, b)					\
slouken@0
   229
{									\
icculus@1162
   230
	Pixel = ((r>>3)<<10)|((g>>3)<<5)|(b>>3);			\
slouken@0
   231
}
icculus@1162
   232
#define RGB888_FROM_RGB(Pixel, r, g, b)					\
slouken@0
   233
{									\
icculus@1162
   234
	Pixel = (r<<16)|(g<<8)|b;					\
slouken@0
   235
}
slouken@2899
   236
#define ARGB8888_FROM_RGBA(Pixel, r, g, b, a)				\
slouken@2899
   237
{									\
slouken@2899
   238
	Pixel = (a<<24)|(r<<16)|(g<<8)|b;				\
slouken@2899
   239
}
slouken@3054
   240
#define RGBA8888_FROM_RGBA(Pixel, r, g, b, a)				\
slouken@3054
   241
{									\
slouken@3054
   242
	Pixel = (r<<24)|(g<<16)|(b<<8)|a;				\
slouken@3054
   243
}
slouken@3054
   244
#define ABGR8888_FROM_RGBA(Pixel, r, g, b, a)				\
slouken@3054
   245
{									\
slouken@3054
   246
	Pixel = (a<<24)|(b<<16)|(g<<8)|r;				\
slouken@3054
   247
}
slouken@3054
   248
#define BGRA8888_FROM_RGBA(Pixel, r, g, b, a)				\
slouken@3054
   249
{									\
slouken@3054
   250
	Pixel = (b<<24)|(g<<16)|(r<<8)|a;				\
slouken@3054
   251
}
slouken@0
   252
#define ASSEMBLE_RGB(buf, bpp, fmt, r, g, b) 				\
slouken@0
   253
{									\
slouken@0
   254
	switch (bpp) {							\
slouken@0
   255
		case 2: {						\
icculus@1162
   256
			Uint16 Pixel;					\
slouken@0
   257
									\
icculus@1162
   258
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   259
			*((Uint16 *)(buf)) = Pixel;			\
slouken@0
   260
		}							\
slouken@0
   261
		break;							\
slouken@0
   262
									\
slouken@0
   263
		case 3: {						\
slouken@2824
   264
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		\
slouken@0
   265
			        *((buf)+fmt->Rshift/8) = r;		\
slouken@0
   266
				*((buf)+fmt->Gshift/8) = g;		\
slouken@0
   267
				*((buf)+fmt->Bshift/8) = b;		\
slouken@0
   268
			} else {					\
slouken@0
   269
			        *((buf)+2-fmt->Rshift/8) = r;		\
slouken@0
   270
				*((buf)+2-fmt->Gshift/8) = g;		\
slouken@0
   271
				*((buf)+2-fmt->Bshift/8) = b;		\
slouken@0
   272
			}						\
slouken@0
   273
		}							\
slouken@0
   274
		break;							\
slouken@0
   275
									\
slouken@0
   276
		case 4: {						\
icculus@1162
   277
			Uint32 Pixel;					\
slouken@0
   278
									\
icculus@1162
   279
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   280
			*((Uint32 *)(buf)) = Pixel;			\
slouken@0
   281
		}							\
slouken@0
   282
		break;							\
slouken@0
   283
	}								\
slouken@0
   284
}
slouken@0
   285
#define ASSEMBLE_RGB_AMASK(buf, bpp, fmt, r, g, b, Amask)		\
slouken@0
   286
{									\
slouken@0
   287
	switch (bpp) {							\
slouken@0
   288
		case 2: {						\
slouken@0
   289
			Uint16 *bufp;					\
icculus@1162
   290
			Uint16 Pixel;					\
slouken@0
   291
									\
slouken@0
   292
			bufp = (Uint16 *)buf;				\
icculus@1162
   293
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   294
			*bufp = Pixel | (*bufp & Amask);		\
slouken@0
   295
		}							\
slouken@0
   296
		break;							\
slouken@0
   297
									\
slouken@0
   298
		case 3: {						\
slouken@2824
   299
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		\
slouken@0
   300
			        *((buf)+fmt->Rshift/8) = r;		\
slouken@0
   301
				*((buf)+fmt->Gshift/8) = g;		\
slouken@0
   302
				*((buf)+fmt->Bshift/8) = b;		\
slouken@0
   303
			} else {					\
slouken@0
   304
			        *((buf)+2-fmt->Rshift/8) = r;		\
slouken@0
   305
				*((buf)+2-fmt->Gshift/8) = g;		\
slouken@0
   306
				*((buf)+2-fmt->Bshift/8) = b;		\
slouken@0
   307
			}						\
slouken@0
   308
		}							\
slouken@0
   309
		break;							\
slouken@0
   310
									\
slouken@0
   311
		case 4: {						\
slouken@0
   312
			Uint32 *bufp;					\
icculus@1162
   313
			Uint32 Pixel;					\
slouken@0
   314
									\
slouken@0
   315
			bufp = (Uint32 *)buf;				\
icculus@1162
   316
			PIXEL_FROM_RGB(Pixel, fmt, r, g, b);		\
icculus@1162
   317
			*bufp = Pixel | (*bufp & Amask);		\
slouken@0
   318
		}							\
slouken@0
   319
		break;							\
slouken@0
   320
	}								\
slouken@0
   321
}
slouken@0
   322
slouken@0
   323
/* FIXME: Should we rescale alpha into 0..255 here? */
icculus@1162
   324
#define RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a)				\
slouken@0
   325
{									\
icculus@1162
   326
	r = ((Pixel&fmt->Rmask)>>fmt->Rshift)<<fmt->Rloss; 		\
icculus@1162
   327
	g = ((Pixel&fmt->Gmask)>>fmt->Gshift)<<fmt->Gloss; 		\
icculus@1162
   328
	b = ((Pixel&fmt->Bmask)>>fmt->Bshift)<<fmt->Bloss; 		\
icculus@1162
   329
	a = ((Pixel&fmt->Amask)>>fmt->Ashift)<<fmt->Aloss;	 	\
slouken@0
   330
}
icculus@1162
   331
#define RGBA_FROM_8888(Pixel, fmt, r, g, b, a)	\
slouken@0
   332
{						\
icculus@1162
   333
	r = (Pixel&fmt->Rmask)>>fmt->Rshift;	\
icculus@1162
   334
	g = (Pixel&fmt->Gmask)>>fmt->Gshift;	\
icculus@1162
   335
	b = (Pixel&fmt->Bmask)>>fmt->Bshift;	\
icculus@1162
   336
	a = (Pixel&fmt->Amask)>>fmt->Ashift;	\
slouken@0
   337
}
icculus@1162
   338
#define RGBA_FROM_RGBA8888(Pixel, r, g, b, a)				\
slouken@0
   339
{									\
icculus@1162
   340
	r = (Pixel>>24);						\
icculus@1162
   341
	g = ((Pixel>>16)&0xFF);						\
icculus@1162
   342
	b = ((Pixel>>8)&0xFF);						\
icculus@1162
   343
	a = (Pixel&0xFF);						\
slouken@0
   344
}
icculus@1162
   345
#define RGBA_FROM_ARGB8888(Pixel, r, g, b, a)				\
slouken@0
   346
{									\
icculus@1162
   347
	r = ((Pixel>>16)&0xFF);						\
icculus@1162
   348
	g = ((Pixel>>8)&0xFF);						\
icculus@1162
   349
	b = (Pixel&0xFF);						\
icculus@1162
   350
	a = (Pixel>>24);						\
slouken@0
   351
}
icculus@1162
   352
#define RGBA_FROM_ABGR8888(Pixel, r, g, b, a)				\
slouken@0
   353
{									\
icculus@1162
   354
	r = (Pixel&0xFF);						\
icculus@1162
   355
	g = ((Pixel>>8)&0xFF);						\
icculus@1162
   356
	b = ((Pixel>>16)&0xFF);						\
icculus@1162
   357
	a = (Pixel>>24);						\
slouken@0
   358
}
slouken@3054
   359
#define RGBA_FROM_BGRA8888(Pixel, r, g, b, a)				\
slouken@3054
   360
{									\
slouken@3054
   361
	r = ((Pixel>>8)&0xFF);						\
slouken@3054
   362
	g = ((Pixel>>16)&0xFF);						\
slouken@3054
   363
	b = (Pixel>>24);						\
slouken@3054
   364
	a = (Pixel&0xFF);						\
slouken@3054
   365
}
icculus@1162
   366
#define DISEMBLE_RGBA(buf, bpp, fmt, Pixel, r, g, b, a)			   \
slouken@0
   367
do {									   \
slouken@0
   368
	switch (bpp) {							   \
slouken@0
   369
		case 2:							   \
icculus@1162
   370
			Pixel = *((Uint16 *)(buf));			   \
slouken@2824
   371
			RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a);	   \
slouken@0
   372
		break;							   \
slouken@0
   373
									   \
slouken@2824
   374
		case 3:	{						   \
slouken@2824
   375
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		   \
slouken@2824
   376
			        r = *((buf)+fmt->Rshift/8);		   \
slouken@2824
   377
				g = *((buf)+fmt->Gshift/8);		   \
slouken@2824
   378
				b = *((buf)+fmt->Bshift/8);		   \
slouken@0
   379
			} else {					   \
slouken@2824
   380
			        r = *((buf)+2-fmt->Rshift/8);		   \
slouken@2824
   381
				g = *((buf)+2-fmt->Gshift/8);		   \
slouken@2824
   382
				b = *((buf)+2-fmt->Bshift/8);		   \
slouken@0
   383
			}						   \
slouken@2824
   384
			a = 0xFF;					   \
slouken@0
   385
		}							   \
slouken@0
   386
		break;							   \
slouken@0
   387
									   \
slouken@0
   388
		case 4:							   \
icculus@1162
   389
			Pixel = *((Uint32 *)(buf));			   \
slouken@2824
   390
			RGBA_FROM_PIXEL(Pixel, fmt, r, g, b, a);	   \
slouken@0
   391
		break;							   \
slouken@0
   392
									   \
slouken@0
   393
		default:						   \
slouken@2824
   394
		        Pixel; /* stop gcc complaints */		   \
slouken@0
   395
		break;							   \
slouken@0
   396
	}								   \
slouken@2824
   397
} while (0)
slouken@0
   398
slouken@0
   399
/* FIXME: this isn't correct, especially for Alpha (maximum != 255) */
icculus@1162
   400
#define PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a)				\
slouken@0
   401
{									\
icculus@1162
   402
	Pixel = ((r>>fmt->Rloss)<<fmt->Rshift)|				\
slouken@0
   403
		((g>>fmt->Gloss)<<fmt->Gshift)|				\
slouken@0
   404
		((b>>fmt->Bloss)<<fmt->Bshift)|				\
slouken@535
   405
		((a>>fmt->Aloss)<<fmt->Ashift);				\
slouken@0
   406
}
slouken@0
   407
#define ASSEMBLE_RGBA(buf, bpp, fmt, r, g, b, a)			\
slouken@0
   408
{									\
slouken@0
   409
	switch (bpp) {							\
slouken@0
   410
		case 2: {						\
icculus@1162
   411
			Uint16 Pixel;					\
slouken@0
   412
									\
icculus@1162
   413
			PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a);	\
icculus@1162
   414
			*((Uint16 *)(buf)) = Pixel;			\
slouken@0
   415
		}							\
slouken@0
   416
		break;							\
slouken@0
   417
									\
slouken@2824
   418
		case 3: {						\
slouken@2824
   419
                        if (SDL_BYTEORDER == SDL_LIL_ENDIAN) {		\
slouken@0
   420
			        *((buf)+fmt->Rshift/8) = r;		\
slouken@0
   421
				*((buf)+fmt->Gshift/8) = g;		\
slouken@0
   422
				*((buf)+fmt->Bshift/8) = b;		\
slouken@0
   423
			} else {					\
slouken@0
   424
			        *((buf)+2-fmt->Rshift/8) = r;		\
slouken@0
   425
				*((buf)+2-fmt->Gshift/8) = g;		\
slouken@0
   426
				*((buf)+2-fmt->Bshift/8) = b;		\
slouken@0
   427
			}						\
slouken@0
   428
		}							\
slouken@0
   429
		break;							\
slouken@0
   430
									\
slouken@0
   431
		case 4: {						\
icculus@1162
   432
			Uint32 Pixel;					\
slouken@0
   433
									\
icculus@1162
   434
			PIXEL_FROM_RGBA(Pixel, fmt, r, g, b, a);	\
icculus@1162
   435
			*((Uint32 *)(buf)) = Pixel;			\
slouken@0
   436
		}							\
slouken@0
   437
		break;							\
slouken@0
   438
	}								\
slouken@0
   439
}
slouken@0
   440
icculus@1162
   441
/* Blend the RGB values of two Pixels based on a source alpha value */
slouken@0
   442
#define ALPHA_BLEND(sR, sG, sB, A, dR, dG, dB)	\
slouken@0
   443
do {						\
icculus@3631
   444
	dR = ((((int)(sR-dR)*(int)A)/255)+dR);	\
icculus@3631
   445
	dG = ((((int)(sG-dG)*(int)A)/255)+dG);	\
icculus@3631
   446
	dB = ((((int)(sB-dB)*(int)A)/255)+dB);	\
icculus@1047
   447
} while(0)
icculus@1047
   448
icculus@1047
   449
slouken@0
   450
/* This is a very useful loop for optimizing blitters */
slouken@553
   451
#if defined(_MSC_VER) && (_MSC_VER == 1300)
slouken@553
   452
/* There's a bug in the Visual C++ 7 optimizer when compiling this code */
slouken@553
   453
#else
slouken@0
   454
#define USE_DUFFS_LOOP
slouken@553
   455
#endif
slouken@0
   456
#ifdef USE_DUFFS_LOOP
slouken@0
   457
slouken@0
   458
/* 8-times unrolled loop */
slouken@0
   459
#define DUFFS_LOOP8(pixel_copy_increment, width)			\
slouken@0
   460
{ int n = (width+7)/8;							\
slouken@91
   461
	switch (width & 7) {						\
slouken@0
   462
	case 0: do {	pixel_copy_increment;				\
slouken@0
   463
	case 7:		pixel_copy_increment;				\
slouken@0
   464
	case 6:		pixel_copy_increment;				\
slouken@0
   465
	case 5:		pixel_copy_increment;				\
slouken@0
   466
	case 4:		pixel_copy_increment;				\
slouken@0
   467
	case 3:		pixel_copy_increment;				\
slouken@0
   468
	case 2:		pixel_copy_increment;				\
slouken@0
   469
	case 1:		pixel_copy_increment;				\
slouken@0
   470
		} while ( --n > 0 );					\
slouken@0
   471
	}								\
slouken@0
   472
}
slouken@0
   473
slouken@0
   474
/* 4-times unrolled loop */
slouken@0
   475
#define DUFFS_LOOP4(pixel_copy_increment, width)			\
slouken@0
   476
{ int n = (width+3)/4;							\
slouken@91
   477
	switch (width & 3) {						\
slouken@0
   478
	case 0: do {	pixel_copy_increment;				\
slouken@0
   479
	case 3:		pixel_copy_increment;				\
slouken@0
   480
	case 2:		pixel_copy_increment;				\
slouken@0
   481
	case 1:		pixel_copy_increment;				\
slouken@3035
   482
		} while (--n > 0);					\
slouken@689
   483
	}								\
slouken@689
   484
}
slouken@689
   485
slouken@0
   486
/* Use the 8-times version of the loop by default */
slouken@0
   487
#define DUFFS_LOOP(pixel_copy_increment, width)				\
slouken@0
   488
	DUFFS_LOOP8(pixel_copy_increment, width)
slouken@0
   489
slouken@3035
   490
/* Special version of Duff's device for even more optimization */
slouken@3035
   491
#define DUFFS_LOOP_124(pixel_copy_increment1,				\
slouken@3035
   492
                       pixel_copy_increment2,				\
slouken@3035
   493
                       pixel_copy_increment4, width)			\
slouken@3035
   494
{ int n = width;							\
slouken@3035
   495
	if (n & 1) {							\
slouken@3035
   496
		pixel_copy_increment1; n -= 1;				\
slouken@3035
   497
	}								\
slouken@3035
   498
	if (n & 2) {							\
slouken@3035
   499
		pixel_copy_increment2; n -= 2;				\
slouken@3035
   500
	}								\
slouken@3035
   501
	if (n) {							\
slouken@3035
   502
		n = (n+7)/ 8;						\
slouken@3035
   503
		switch (n & 4) {					\
slouken@3035
   504
		case 0: do {	pixel_copy_increment4;			\
slouken@3035
   505
		case 4:		pixel_copy_increment4;			\
slouken@3035
   506
			} while (--n > 0);				\
slouken@3035
   507
		}							\
slouken@3035
   508
	}								\
slouken@689
   509
}
slouken@689
   510
slouken@3035
   511
#else
slouken@689
   512
slouken@689
   513
/* Don't use Duff's device to unroll loops */
slouken@0
   514
#define DUFFS_LOOP(pixel_copy_increment, width)				\
slouken@0
   515
{ int n;								\
slouken@0
   516
	for ( n=width; n > 0; --n ) {					\
slouken@0
   517
		pixel_copy_increment;					\
slouken@0
   518
	}								\
slouken@0
   519
}
slouken@0
   520
#define DUFFS_LOOP8(pixel_copy_increment, width)			\
slouken@0
   521
	DUFFS_LOOP(pixel_copy_increment, width)
slouken@0
   522
#define DUFFS_LOOP4(pixel_copy_increment, width)			\
slouken@0
   523
	DUFFS_LOOP(pixel_copy_increment, width)
slouken@3035
   524
#define DUFFS_LOOP_124(pixel_copy_increment1,				\
slouken@3035
   525
                       pixel_copy_increment2,				\
slouken@3035
   526
                       pixel_copy_increment4, width)			\
slouken@3035
   527
	DUFFS_LOOP(pixel_copy_increment1, width)
slouken@0
   528
slouken@0
   529
#endif /* USE_DUFFS_LOOP */
slouken@0
   530
slouken@0
   531
/* Prevent Visual C++ 6.0 from printing out stupid warnings */
slouken@0
   532
#if defined(_MSC_VER) && (_MSC_VER >= 600)
slouken@0
   533
#pragma warning(disable: 4550)
slouken@0
   534
#endif
slouken@0
   535
slouken@0
   536
#endif /* _SDL_blit_h */
slouken@2898
   537
slouken@1895
   538
/* vi: set ts=4 sw=4 expandtab: */