src/video/SDL_stretch.c
author Sam Lantinga <slouken@libsdl.org>
Tue, 21 Feb 2006 08:46:50 +0000
changeset 1402 d910939febfa
parent 1361 19418e4422cb
child 1442 e3242177fe4a
permissions -rw-r--r--
Use consistent identifiers for the various platforms we support.
Make sure every source file includes SDL_config.h, so the proper system
headers are chosen.
     1 /*
     2     SDL - Simple DirectMedia Layer
     3     Copyright (C) 1997-2006 Sam Lantinga
     4 
     5     This library is free software; you can redistribute it and/or
     6     modify it under the terms of the GNU Lesser General Public
     7     License as published by the Free Software Foundation; either
     8     version 2.1 of the License, or (at your option) any later version.
     9 
    10     This library is distributed in the hope that it will be useful,
    11     but WITHOUT ANY WARRANTY; without even the implied warranty of
    12     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
    13     Lesser General Public License for more details.
    14 
    15     You should have received a copy of the GNU Lesser General Public
    16     License along with this library; if not, write to the Free Software
    17     Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
    18 
    19     Sam Lantinga
    20     slouken@libsdl.org
    21 */
    22 #include "SDL_config.h"
    23 
    24 /* This a stretch blit implementation based on ideas given to me by
    25    Tomasz Cejner - thanks! :)
    26 
    27    April 27, 2000 - Sam Lantinga
    28 */
    29 
    30 #include "SDL_video.h"
    31 #include "SDL_blit.h"
    32 
    33 /* This isn't ready for general consumption yet - it should be folded
    34    into the general blitting mechanism.
    35 */
    36 
    37 #if ((defined(_MFC_VER) && defined(_M_IX86)/* && !defined(_WIN32_WCE) still needed? */) || \
    38      (defined(__GNUC__) && defined(__i386__))) && SDL_ASSEMBLY_ROUTINES
    39 #define USE_ASM_STRETCH
    40 #endif
    41 
    42 #ifdef USE_ASM_STRETCH
    43 
    44 #if defined(_M_IX86) || defined(i386)
    45 #define PREFIX16	0x66
    46 #define STORE_BYTE	0xAA
    47 #define STORE_WORD	0xAB
    48 #define LOAD_BYTE	0xAC
    49 #define LOAD_WORD	0xAD
    50 #define RETURN		0xC3
    51 #else
    52 #error Need assembly opcodes for this architecture
    53 #endif
    54 
    55 static unsigned char copy_row[4096];
    56 
    57 static int generate_rowbytes(int src_w, int dst_w, int bpp)
    58 {
    59 	static struct {
    60 		int bpp;
    61 		int src_w;
    62 		int dst_w;
    63 	} last;
    64 
    65 	int i;
    66 	int pos, inc;
    67 	unsigned char *eip;
    68 	unsigned char load, store;
    69 
    70 	/* See if we need to regenerate the copy buffer */
    71 	if ( (src_w == last.src_w) &&
    72 	     (dst_w == last.dst_w) && (bpp == last.bpp) ) {
    73 		return(0);
    74 	}
    75 	last.bpp = bpp;
    76 	last.src_w = src_w;
    77 	last.dst_w = dst_w;
    78 
    79 	switch (bpp) {
    80 	    case 1:
    81 		load = LOAD_BYTE;
    82 		store = STORE_BYTE;
    83 		break;
    84 	    case 2:
    85 	    case 4:
    86 		load = LOAD_WORD;
    87 		store = STORE_WORD;
    88 		break;
    89 	    default:
    90 		SDL_SetError("ASM stretch of %d bytes isn't supported\n", bpp);
    91 		return(-1);
    92 	}
    93 	pos = 0x10000;
    94 	inc = (src_w << 16) / dst_w;
    95 	eip = copy_row;
    96 	for ( i=0; i<dst_w; ++i ) {
    97 		while ( pos >= 0x10000L ) {
    98 			if ( bpp == 2 ) {
    99 				*eip++ = PREFIX16;
   100 			}
   101 			*eip++ = load;
   102 			pos -= 0x10000L;
   103 		}
   104 		if ( bpp == 2 ) {
   105 			*eip++ = PREFIX16;
   106 		}
   107 		*eip++ = store;
   108 		pos += inc;
   109 	}
   110 	*eip++ = RETURN;
   111 
   112 	/* Verify that we didn't overflow (too late) */
   113 	if ( eip > (copy_row+sizeof(copy_row)) ) {
   114 		SDL_SetError("Copy buffer overflow");
   115 		return(-1);
   116 	}
   117 	return(0);
   118 }
   119 
   120 #else
   121 
   122 #define DEFINE_COPY_ROW(name, type)			\
   123 void name(type *src, int src_w, type *dst, int dst_w)	\
   124 {							\
   125 	int i;						\
   126 	int pos, inc;					\
   127 	type pixel = 0;					\
   128 							\
   129 	pos = 0x10000;					\
   130 	inc = (src_w << 16) / dst_w;			\
   131 	for ( i=dst_w; i>0; --i ) {			\
   132 		while ( pos >= 0x10000L ) {		\
   133 			pixel = *src++;			\
   134 			pos -= 0x10000L;		\
   135 		}					\
   136 		*dst++ = pixel;				\
   137 		pos += inc;				\
   138 	}						\
   139 }
   140 DEFINE_COPY_ROW(copy_row1, Uint8)
   141 DEFINE_COPY_ROW(copy_row2, Uint16)
   142 DEFINE_COPY_ROW(copy_row4, Uint32)
   143 
   144 #endif /* USE_ASM_STRETCH */
   145 
   146 /* The ASM code doesn't handle 24-bpp stretch blits */
   147 void copy_row3(Uint8 *src, int src_w, Uint8 *dst, int dst_w)
   148 {
   149 	int i;
   150 	int pos, inc;
   151 	Uint8 pixel[3];
   152 
   153 	pos = 0x10000;
   154 	inc = (src_w << 16) / dst_w;
   155 	for ( i=dst_w; i>0; --i ) {
   156 		while ( pos >= 0x10000L ) {
   157 			pixel[0] = *src++;
   158 			pixel[1] = *src++;
   159 			pixel[2] = *src++;
   160 			pos -= 0x10000L;
   161 		}
   162 		*dst++ = pixel[0];
   163 		*dst++ = pixel[1];
   164 		*dst++ = pixel[2];
   165 		pos += inc;
   166 	}
   167 }
   168 
   169 /* Perform a stretch blit between two surfaces of the same format.
   170    NOTE:  This function is not safe to call from multiple threads!
   171 */
   172 int SDL_SoftStretch(SDL_Surface *src, SDL_Rect *srcrect,
   173                     SDL_Surface *dst, SDL_Rect *dstrect)
   174 {
   175 	int src_locked;
   176 	int dst_locked;
   177 	int pos, inc;
   178 	int dst_width;
   179 	int dst_maxrow;
   180 	int src_row, dst_row;
   181 	Uint8 *srcp = NULL;
   182 	Uint8 *dstp;
   183 	SDL_Rect full_src;
   184 	SDL_Rect full_dst;
   185 #if defined(USE_ASM_STRETCH) && defined(__GNUC__)
   186 	int u1, u2;
   187 #endif
   188 	const int bpp = dst->format->BytesPerPixel;
   189 
   190 	if ( src->format->BitsPerPixel != dst->format->BitsPerPixel ) {
   191 		SDL_SetError("Only works with same format surfaces");
   192 		return(-1);
   193 	}
   194 
   195 	/* Verify the blit rectangles */
   196 	if ( srcrect ) {
   197 		if ( (srcrect->x < 0) || (srcrect->y < 0) ||
   198 		     ((srcrect->x+srcrect->w) > src->w) ||
   199 		     ((srcrect->y+srcrect->h) > src->h) ) {
   200 			SDL_SetError("Invalid source blit rectangle");
   201 			return(-1);
   202 		}
   203 	} else {
   204 		full_src.x = 0;
   205 		full_src.y = 0;
   206 		full_src.w = src->w;
   207 		full_src.h = src->h;
   208 		srcrect = &full_src;
   209 	}
   210 	if ( dstrect ) {
   211 		if ( (dstrect->x < 0) || (dstrect->y < 0) ||
   212 		     ((dstrect->x+dstrect->w) > dst->w) ||
   213 		     ((dstrect->y+dstrect->h) > dst->h) ) {
   214 			SDL_SetError("Invalid destination blit rectangle");
   215 			return(-1);
   216 		}
   217 	} else {
   218 		full_dst.x = 0;
   219 		full_dst.y = 0;
   220 		full_dst.w = dst->w;
   221 		full_dst.h = dst->h;
   222 		dstrect = &full_dst;
   223 	}
   224 
   225 	/* Lock the destination if it's in hardware */
   226 	dst_locked = 0;
   227 	if ( SDL_MUSTLOCK(dst) ) {
   228 		if ( SDL_LockSurface(dst) < 0 ) {
   229 			SDL_SetError("Unable to lock destination surface");
   230 			return(-1);
   231 		}
   232 		dst_locked = 1;
   233 	}
   234 	/* Lock the source if it's in hardware */
   235 	src_locked = 0;
   236 	if ( SDL_MUSTLOCK(src) ) {
   237 		if ( SDL_LockSurface(src) < 0 ) {
   238 			if ( dst_locked ) {
   239 				SDL_UnlockSurface(dst);
   240 			}
   241 			SDL_SetError("Unable to lock source surface");
   242 			return(-1);
   243 		}
   244 		src_locked = 1;
   245 	}
   246 
   247 	/* Set up the data... */
   248 	pos = 0x10000;
   249 	inc = (srcrect->h << 16) / dstrect->h;
   250 	src_row = srcrect->y;
   251 	dst_row = dstrect->y;
   252 	dst_width = dstrect->w*bpp;
   253 
   254 #ifdef USE_ASM_STRETCH
   255 	/* Write the opcodes for this stretch */
   256 	if ( (bpp != 3) &&
   257 	     (generate_rowbytes(srcrect->w, dstrect->w, bpp) < 0) ) {
   258 		return(-1);
   259 	}
   260 #endif
   261 
   262 	/* Perform the stretch blit */
   263 	for ( dst_maxrow = dst_row+dstrect->h; dst_row<dst_maxrow; ++dst_row ) {
   264 		dstp = (Uint8 *)dst->pixels + (dst_row*dst->pitch)
   265 		                            + (dstrect->x*bpp);
   266 		while ( pos >= 0x10000L ) {
   267 			srcp = (Uint8 *)src->pixels + (src_row*src->pitch)
   268 			                            + (srcrect->x*bpp);
   269 			++src_row;
   270 			pos -= 0x10000L;
   271 		}
   272 #ifdef USE_ASM_STRETCH
   273 		switch (bpp) {
   274 		    case 3:
   275 			copy_row3(srcp, srcrect->w, dstp, dstrect->w);
   276 			break;
   277 		    default:
   278 #ifdef __GNUC__
   279 			__asm__ __volatile__ (
   280 			"call *%4"
   281 			: "=&D" (u1), "=&S" (u2)
   282 			: "0" (dstp), "1" (srcp), "r" (copy_row)
   283 			: "memory" );
   284 #else
   285 #ifdef _MSC_VER
   286 		{ void *code = copy_row;
   287 			__asm {
   288 				push edi
   289 				push esi
   290 	
   291 				mov edi, dstp
   292 				mov esi, srcp
   293 				call dword ptr code
   294 
   295 				pop esi
   296 				pop edi
   297 			}
   298 		}
   299 #else
   300 #error Need inline assembly for this compiler
   301 #endif
   302 #endif /* __GNUC__ */
   303 			break;
   304 		}
   305 #else
   306 		switch (bpp) {
   307 		    case 1:
   308 			copy_row1(srcp, srcrect->w, dstp, dstrect->w);
   309 			break;
   310 		    case 2:
   311 			copy_row2((Uint16 *)srcp, srcrect->w,
   312 			          (Uint16 *)dstp, dstrect->w);
   313 			break;
   314 		    case 3:
   315 			copy_row3(srcp, srcrect->w, dstp, dstrect->w);
   316 			break;
   317 		    case 4:
   318 			copy_row4((Uint32 *)srcp, srcrect->w,
   319 			          (Uint32 *)dstp, dstrect->w);
   320 			break;
   321 		}
   322 #endif
   323 		pos += inc;
   324 	}
   325 
   326 	/* We need to unlock the surfaces if they're locked */
   327 	if ( dst_locked ) {
   328 		SDL_UnlockSurface(dst);
   329 	}
   330 	if ( src_locked ) {
   331 		SDL_UnlockSurface(src);
   332 	}
   333 	return(0);
   334 }
   335