src/video/SDL_stretch.c
author Ryan C. Gordon <icculus@icculus.org>
Fri, 06 Jan 2006 00:56:12 +0000
changeset 1233 228c94f0b5dc
parent 1228 f4a3a4129d04
child 1234 73676c1f56ee
permissions -rw-r--r--
Patched to compile.
     1 /*
     2     SDL - Simple DirectMedia Layer
     3     Copyright (C) 1997-2004 Sam Lantinga
     4 
     5     This library is free software; you can redistribute it and/or
     6     modify it under the terms of the GNU Library General Public
     7     License as published by the Free Software Foundation; either
     8     version 2 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     Library General Public License for more details.
    14 
    15     You should have received a copy of the GNU Library General Public
    16     License along with this library; if not, write to the Free
    17     Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
    18 
    19     Sam Lantinga
    20     slouken@libsdl.org
    21 */
    22 
    23 #ifdef SAVE_RCSID
    24 static char rcsid =
    25  "@(#) $Id$";
    26 #endif
    27 
    28 /* This a stretch blit implementation based on ideas given to me by
    29    Tomasz Cejner - thanks! :)
    30 
    31    April 27, 2000 - Sam Lantinga
    32 */
    33 
    34 #include "SDL_error.h"
    35 #include "SDL_video.h"
    36 #include "SDL_blit.h"
    37 
    38 /* This isn't ready for general consumption yet - it should be folded
    39    into the general blitting mechanism.
    40 */
    41 
    42 #if (defined(WIN32) && !defined(_M_ALPHA) && !defined(_WIN32_WCE) && \
    43      !defined(__WATCOMC__) && !defined(__LCC__) && !defined(__FREEBCC__)) || \
    44     (defined(i386) && defined(__GNUC__) && defined(USE_ASMBLIT))
    45 #define USE_ASM_STRETCH
    46 #endif
    47 
    48 #ifdef USE_ASM_STRETCH
    49 
    50 #if defined(WIN32) || defined(i386)
    51 #define PREFIX16	0x66
    52 #define STORE_BYTE	0xAA
    53 #define STORE_WORD	0xAB
    54 #define LOAD_BYTE	0xAC
    55 #define LOAD_WORD	0xAD
    56 #define RETURN		0xC3
    57 #else
    58 #error Need assembly opcodes for this architecture
    59 #endif
    60 
    61 static unsigned char copy_row[4096];
    62 
    63 static int generate_rowbytes(int src_w, int dst_w, int bpp)
    64 {
    65 	static struct {
    66 		int bpp;
    67 		int src_w;
    68 		int dst_w;
    69 	} last;
    70 
    71 	int i;
    72 	int pos, inc;
    73 	unsigned char *eip;
    74 	unsigned char load, store;
    75 
    76 	/* See if we need to regenerate the copy buffer */
    77 	if ( (src_w == last.src_w) &&
    78 	     (dst_w == last.dst_w) && (bpp == last.bpp) ) {
    79 		return(0);
    80 	}
    81 	last.bpp = bpp;
    82 	last.src_w = src_w;
    83 	last.dst_w = dst_w;
    84 
    85 	switch (bpp) {
    86 	    case 1:
    87 		load = LOAD_BYTE;
    88 		store = STORE_BYTE;
    89 		break;
    90 	    case 2:
    91 	    case 4:
    92 		load = LOAD_WORD;
    93 		store = STORE_WORD;
    94 		break;
    95 	    default:
    96 		SDL_SetError("ASM stretch of %d bytes isn't supported\n", bpp);
    97 		return(-1);
    98 	}
    99 	pos = 0x10000;
   100 	inc = (src_w << 16) / dst_w;
   101 	eip = copy_row;
   102 	for ( i=0; i<dst_w; ++i ) {
   103 		while ( pos >= 0x10000L ) {
   104 			if ( bpp == 2 ) {
   105 				*eip++ = PREFIX16;
   106 			}
   107 			*eip++ = load;
   108 			pos -= 0x10000L;
   109 		}
   110 		if ( bpp == 2 ) {
   111 			*eip++ = PREFIX16;
   112 		}
   113 		*eip++ = store;
   114 		pos += inc;
   115 	}
   116 	*eip++ = RETURN;
   117 
   118 	/* Verify that we didn't overflow (too late) */
   119 	if ( eip > (copy_row+sizeof(copy_row)) ) {
   120 		SDL_SetError("Copy buffer overflow");
   121 		return(-1);
   122 	}
   123 	return(0);
   124 }
   125 
   126 #else
   127 
   128 #define DEFINE_COPY_ROW(name, type)			\
   129 void name(type *src, int src_w, type *dst, int dst_w)	\
   130 {							\
   131 	int i;						\
   132 	int pos, inc;					\
   133 	type pixel = 0;					\
   134 							\
   135 	pos = 0x10000;					\
   136 	inc = (src_w << 16) / dst_w;			\
   137 	for ( i=dst_w; i>0; --i ) {			\
   138 		while ( pos >= 0x10000L ) {		\
   139 			pixel = *src++;			\
   140 			pos -= 0x10000L;		\
   141 		}					\
   142 		*dst++ = pixel;				\
   143 		pos += inc;				\
   144 	}						\
   145 }
   146 DEFINE_COPY_ROW(copy_row1, Uint8)
   147 DEFINE_COPY_ROW(copy_row2, Uint16)
   148 DEFINE_COPY_ROW(copy_row4, Uint32)
   149 
   150 #endif /* USE_ASM_STRETCH */
   151 
   152 /* The ASM code doesn't handle 24-bpp stretch blits */
   153 void copy_row3(Uint8 *src, int src_w, Uint8 *dst, int dst_w)
   154 {
   155 	int i;
   156 	int pos, inc;
   157 	Uint8 pixel[3];
   158 
   159 	pos = 0x10000;
   160 	inc = (src_w << 16) / dst_w;
   161 	for ( i=dst_w; i>0; --i ) {
   162 		while ( pos >= 0x10000L ) {
   163 			pixel[0] = *src++;
   164 			pixel[1] = *src++;
   165 			pixel[2] = *src++;
   166 			pos -= 0x10000L;
   167 		}
   168 		*dst++ = pixel[0];
   169 		*dst++ = pixel[1];
   170 		*dst++ = pixel[2];
   171 		pos += inc;
   172 	}
   173 }
   174 
   175 /* Perform a stretch blit between two surfaces of the same format.
   176    NOTE:  This function is not safe to call from multiple threads!
   177 */
   178 int SDL_SoftStretch(SDL_Surface *src, SDL_Rect *srcrect,
   179                     SDL_Surface *dst, SDL_Rect *dstrect)
   180 {
   181 	int src_locked;
   182 	int dst_locked;
   183 	int pos, inc;
   184 	int dst_width;
   185 	int dst_maxrow;
   186 	int src_row, dst_row;
   187 	Uint8 *srcp = NULL;
   188 	Uint8 *dstp;
   189 	SDL_Rect full_src;
   190 	SDL_Rect full_dst;
   191 #if defined(USE_ASM_STRETCH) && defined(__GNUC__)
   192 	int u1, u2;
   193 #endif
   194 	const int bpp = dst->format->BytesPerPixel;
   195 
   196 	if ( src->format->BitsPerPixel != dst->format->BitsPerPixel ) {
   197 		SDL_SetError("Only works with same format surfaces");
   198 		return(-1);
   199 	}
   200 
   201 	/* Verify the blit rectangles */
   202 	if ( srcrect ) {
   203 		if ( (srcrect->x < 0) || (srcrect->y < 0) ||
   204 		     ((srcrect->x+srcrect->w) > src->w) ||
   205 		     ((srcrect->y+srcrect->h) > src->h) ) {
   206 			SDL_SetError("Invalid source blit rectangle");
   207 			return(-1);
   208 		}
   209 	} else {
   210 		full_src.x = 0;
   211 		full_src.y = 0;
   212 		full_src.w = src->w;
   213 		full_src.h = src->h;
   214 		srcrect = &full_src;
   215 	}
   216 	if ( dstrect ) {
   217 		if ( (dstrect->x < 0) || (dstrect->y < 0) ||
   218 		     ((dstrect->x+dstrect->w) > dst->w) ||
   219 		     ((dstrect->y+dstrect->h) > dst->h) ) {
   220 			SDL_SetError("Invalid destination blit rectangle");
   221 			return(-1);
   222 		}
   223 	} else {
   224 		full_dst.x = 0;
   225 		full_dst.y = 0;
   226 		full_dst.w = dst->w;
   227 		full_dst.h = dst->h;
   228 		dstrect = &full_dst;
   229 	}
   230 
   231 	/* Lock the destination if it's in hardware */
   232 	dst_locked = 0;
   233 	if ( SDL_MUSTLOCK(dst) ) {
   234 		if ( SDL_LockSurface(dst) < 0 ) {
   235 			SDL_SetError("Unable to lock destination surface");
   236 			return(-1);
   237 		}
   238 		dst_locked = 1;
   239 	}
   240 	/* Lock the source if it's in hardware */
   241 	src_locked = 0;
   242 	if ( SDL_MUSTLOCK(src) ) {
   243 		if ( SDL_LockSurface(src) < 0 ) {
   244 			if ( dst_locked ) {
   245 				SDL_UnlockSurface(dst);
   246 			}
   247 			SDL_SetError("Unable to lock source surface");
   248 			return(-1);
   249 		}
   250 		src_locked = 1;
   251 	}
   252 
   253 	/* Set up the data... */
   254 	pos = 0x10000;
   255 	inc = (srcrect->h << 16) / dstrect->h;
   256 	src_row = srcrect->y;
   257 	dst_row = dstrect->y;
   258 	dst_width = dstrect->w*bpp;
   259 
   260 #ifdef USE_ASM_STRETCH
   261 	/* Write the opcodes for this stretch */
   262 	if ( (bpp != 3) &&
   263 	     (generate_rowbytes(srcrect->w, dstrect->w, bpp) < 0) ) {
   264 		return(-1);
   265 	}
   266 #endif
   267 
   268 	/* Perform the stretch blit */
   269 	for ( dst_maxrow = dst_row+dstrect->h; dst_row<dst_maxrow; ++dst_row ) {
   270 		dstp = (Uint8 *)dst->pixels + (dst_row*dst->pitch)
   271 		                            + (dstrect->x*bpp);
   272 		while ( pos >= 0x10000L ) {
   273 			srcp = (Uint8 *)src->pixels + (src_row*src->pitch)
   274 			                            + (srcrect->x*bpp);
   275 			++src_row;
   276 			pos -= 0x10000L;
   277 		}
   278 #ifdef USE_ASM_STRETCH
   279 		switch (bpp) {
   280 		    case 3:
   281 			copy_row3(srcp, srcrect->w, dstp, dstrect->w);
   282 			break;
   283 		    default:
   284 #ifdef __GNUC__
   285 			__asm__ __volatile__ (
   286 			"call *%4"
   287 			: "=&D" (u1), "=&S" (u2)
   288 			: "0" (dstp), "1" (srcp), "r" (&copy_row)
   289 			: "memory" );
   290 #else
   291 #ifdef WIN32
   292 		{ void *code = &copy_row;
   293 			__asm {
   294 				push edi
   295 				push esi
   296 	
   297 				mov edi, dstp
   298 				mov esi, srcp
   299 				call dword ptr code
   300 
   301 				pop esi
   302 				pop edi
   303 			}
   304 		}
   305 #else
   306 #error Need inline assembly for this compiler
   307 #endif
   308 #endif /* __GNUC__ */
   309 			break;
   310 		}
   311 #else
   312 		switch (bpp) {
   313 		    case 1:
   314 			copy_row1(srcp, srcrect->w, dstp, dstrect->w);
   315 			break;
   316 		    case 2:
   317 			copy_row2((Uint16 *)srcp, srcrect->w,
   318 			          (Uint16 *)dstp, dstrect->w);
   319 			break;
   320 		    case 3:
   321 			copy_row3(srcp, srcrect->w, dstp, dstrect->w);
   322 			break;
   323 		    case 4:
   324 			copy_row4((Uint32 *)srcp, srcrect->w,
   325 			          (Uint32 *)dstp, dstrect->w);
   326 			break;
   327 		}
   328 #endif
   329 		pos += inc;
   330 	}
   331 
   332 	/* We need to unlock the surfaces if they're locked */
   333 	if ( dst_locked ) {
   334 		SDL_UnlockSurface(dst);
   335 	}
   336 	if ( src_locked ) {
   337 		SDL_UnlockSurface(src);
   338 	}
   339 	return(0);
   340 }
   341