| 1 |
/* |
| 2 |
* gfxaccel.cpp - Generic Native QuickDraw acceleration |
| 3 |
* |
| 4 |
* SheepShaver (C) 1997-2004 Marc Hellwig and Christian Bauer |
| 5 |
* |
| 6 |
* This program is free software; you can redistribute it and/or modify |
| 7 |
* it under the terms of the GNU General Public License as published by |
| 8 |
* the Free Software Foundation; either version 2 of the License, or |
| 9 |
* (at your option) any later version. |
| 10 |
* |
| 11 |
* This program is distributed in the hope that it will be useful, |
| 12 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 |
* GNU General Public License for more details. |
| 15 |
* |
| 16 |
* You should have received a copy of the GNU General Public License |
| 17 |
* along with this program; if not, write to the Free Software |
| 18 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 19 |
*/ |
| 20 |
|
| 21 |
#include "sysdeps.h" |
| 22 |
|
| 23 |
#include "prefs.h" |
| 24 |
#include "video.h" |
| 25 |
#include "video_defs.h" |
| 26 |
|
| 27 |
#define DEBUG 0 |
| 28 |
#include "debug.h" |
| 29 |
|
| 30 |
|
| 31 |
/* |
| 32 |
* Utility functions |
| 33 |
*/ |
| 34 |
|
| 35 |
// Return bytes per pixel for requested depth |
| 36 |
static inline int bytes_per_pixel(int depth) |
| 37 |
{ |
| 38 |
int bpp; |
| 39 |
switch (depth) { |
| 40 |
case 8: |
| 41 |
bpp = 1; |
| 42 |
break; |
| 43 |
case 15: case 16: |
| 44 |
bpp = 2; |
| 45 |
break; |
| 46 |
case 24: case 32: |
| 47 |
bpp = 4; |
| 48 |
break; |
| 49 |
default: |
| 50 |
abort(); |
| 51 |
} |
| 52 |
return bpp; |
| 53 |
} |
| 54 |
|
| 55 |
|
| 56 |
/* |
| 57 |
* Rectangle inversion |
| 58 |
*/ |
| 59 |
|
| 60 |
template< int bpp > |
| 61 |
static inline void do_invrect(uint8 *dest, uint32 length) |
| 62 |
{ |
| 63 |
#define INVERT_1(PTR, OFS) ((uint8 *)(PTR))[OFS] = ~((uint8 *)(PTR))[OFS] |
| 64 |
#define INVERT_2(PTR, OFS) ((uint16 *)(PTR))[OFS] = ~((uint16 *)(PTR))[OFS] |
| 65 |
#define INVERT_4(PTR, OFS) ((uint32 *)(PTR))[OFS] = ~((uint32 *)(PTR))[OFS] |
| 66 |
#define INVERT_8(PTR, OFS) ((uint64 *)(PTR))[OFS] = ~((uint64 *)(PTR))[OFS] |
| 67 |
|
| 68 |
#ifndef UNALIGNED_PROFITABLE |
| 69 |
// Align on 16-bit boundaries |
| 70 |
if (bpp < 16 && (((uintptr)dest) & 1)) { |
| 71 |
INVERT_1(dest, 0); |
| 72 |
dest += 1; length -= 1; |
| 73 |
} |
| 74 |
|
| 75 |
// Align on 32-bit boundaries |
| 76 |
if (bpp < 32 && (((uintptr)dest) & 2)) { |
| 77 |
INVERT_2(dest, 0); |
| 78 |
dest += 2; length -= 2; |
| 79 |
} |
| 80 |
#endif |
| 81 |
|
| 82 |
// Invert 8-byte words |
| 83 |
if (length >= 8) { |
| 84 |
const int r = (length / 8) % 8; |
| 85 |
dest += r * 8; |
| 86 |
|
| 87 |
int n = ((length / 8) + 7) / 8; |
| 88 |
switch (r) { |
| 89 |
case 0: do { |
| 90 |
dest += 64; |
| 91 |
INVERT_8(dest, -8); |
| 92 |
case 7: INVERT_8(dest, -7); |
| 93 |
case 6: INVERT_8(dest, -6); |
| 94 |
case 5: INVERT_8(dest, -5); |
| 95 |
case 4: INVERT_8(dest, -4); |
| 96 |
case 3: INVERT_8(dest, -3); |
| 97 |
case 2: INVERT_8(dest, -2); |
| 98 |
case 1: INVERT_8(dest, -1); |
| 99 |
} while (--n > 0); |
| 100 |
} |
| 101 |
} |
| 102 |
|
| 103 |
// 32-bit cell to invert? |
| 104 |
if (length & 4) { |
| 105 |
INVERT_4(dest, 0); |
| 106 |
if (bpp <= 16) |
| 107 |
dest += 4; |
| 108 |
} |
| 109 |
|
| 110 |
// 16-bit cell to invert? |
| 111 |
if (bpp <= 16 && (length & 2)) { |
| 112 |
INVERT_2(dest, 0); |
| 113 |
if (bpp <= 8) |
| 114 |
dest += 2; |
| 115 |
} |
| 116 |
|
| 117 |
// 8-bit cell to invert? |
| 118 |
if (bpp <= 8 && (length & 1)) |
| 119 |
INVERT_1(dest, 0); |
| 120 |
|
| 121 |
#undef INVERT_1 |
| 122 |
#undef INVERT_2 |
| 123 |
#undef INVERT_4 |
| 124 |
#undef INVERT_8 |
| 125 |
} |
| 126 |
|
| 127 |
void NQD_invrect(uint32 p) |
| 128 |
{ |
| 129 |
D(bug("accl_invrect %08x\n", p)); |
| 130 |
|
| 131 |
// Get inversion parameters |
| 132 |
int16 dest_X = (int16)ReadMacInt16(p + acclDestRect + 2) - (int16)ReadMacInt16(p + acclDestBoundsRect + 2); |
| 133 |
int16 dest_Y = (int16)ReadMacInt16(p + acclDestRect + 0) - (int16)ReadMacInt16(p + acclDestBoundsRect + 0); |
| 134 |
int16 width = (int16)ReadMacInt16(p + acclDestRect + 6) - (int16)ReadMacInt16(p + acclDestRect + 2); |
| 135 |
int16 height = (int16)ReadMacInt16(p + acclDestRect + 4) - (int16)ReadMacInt16(p + acclDestRect + 0); |
| 136 |
D(bug(" dest X %d, dest Y %d\n", dest_X, dest_Y)); |
| 137 |
D(bug(" width %d, height %d, bytes_per_row %d\n", width, height, (int32)ReadMacInt32(p + acclDestRowBytes))); |
| 138 |
|
| 139 |
//!!?? pen_mode == 14 |
| 140 |
|
| 141 |
// And perform the inversion |
| 142 |
const int bpp = bytes_per_pixel(ReadMacInt32(p + acclDestPixelSize)); |
| 143 |
const int dest_row_bytes = (int32)ReadMacInt32(p + acclDestRowBytes); |
| 144 |
uint8 *dest = Mac2HostAddr(ReadMacInt32(p + acclDestBaseAddr) + (dest_Y * dest_row_bytes) + (dest_X * bpp)); |
| 145 |
width *= bpp; |
| 146 |
switch (bpp) { |
| 147 |
case 1: |
| 148 |
for (int i = 0; i < height; i++) { |
| 149 |
do_invrect<8>(dest, width); |
| 150 |
dest += dest_row_bytes; |
| 151 |
} |
| 152 |
break; |
| 153 |
case 2: |
| 154 |
for (int i = 0; i < height; i++) { |
| 155 |
do_invrect<16>(dest, width); |
| 156 |
dest += dest_row_bytes; |
| 157 |
} |
| 158 |
break; |
| 159 |
case 4: |
| 160 |
for (int i = 0; i < height; i++) { |
| 161 |
do_invrect<32>(dest, width); |
| 162 |
dest += dest_row_bytes; |
| 163 |
} |
| 164 |
break; |
| 165 |
} |
| 166 |
} |
| 167 |
|
| 168 |
|
| 169 |
/* |
| 170 |
* Rectangle filling |
| 171 |
*/ |
| 172 |
|
| 173 |
template< int bpp > |
| 174 |
static inline void do_fillrect(uint8 *dest, uint32 color, uint32 length) |
| 175 |
{ |
| 176 |
#define FILL_1(PTR, OFS, VAL) ((uint8 *)(PTR))[OFS] = (VAL) |
| 177 |
#define FILL_2(PTR, OFS, VAL) ((uint16 *)(PTR))[OFS] = (VAL) |
| 178 |
#define FILL_4(PTR, OFS, VAL) ((uint32 *)(PTR))[OFS] = (VAL) |
| 179 |
#define FILL_8(PTR, OFS, VAL) ((uint64 *)(PTR))[OFS] = (VAL) |
| 180 |
|
| 181 |
#ifndef UNALIGNED_PROFITABLE |
| 182 |
// Align on 16-bit boundaries |
| 183 |
if (bpp < 16 && (((uintptr)dest) & 1)) { |
| 184 |
FILL_1(dest, 0, color); |
| 185 |
dest += 1; length -= 1; |
| 186 |
} |
| 187 |
|
| 188 |
// Align on 32-bit boundaries |
| 189 |
if (bpp < 32 && (((uintptr)dest) & 2)) { |
| 190 |
FILL_2(dest, 0, color); |
| 191 |
dest += 2; length -= 2; |
| 192 |
} |
| 193 |
#endif |
| 194 |
|
| 195 |
// Fill 8-byte words |
| 196 |
if (length >= 8) { |
| 197 |
const uint64 c = (((uint64)color) << 32) | color; |
| 198 |
const int r = (length / 8) % 8; |
| 199 |
dest += r * 8; |
| 200 |
|
| 201 |
int n = ((length / 8) + 7) / 8; |
| 202 |
switch (r) { |
| 203 |
case 0: do { |
| 204 |
dest += 64; |
| 205 |
FILL_8(dest, -8, c); |
| 206 |
case 7: FILL_8(dest, -7, c); |
| 207 |
case 6: FILL_8(dest, -6, c); |
| 208 |
case 5: FILL_8(dest, -5, c); |
| 209 |
case 4: FILL_8(dest, -4, c); |
| 210 |
case 3: FILL_8(dest, -3, c); |
| 211 |
case 2: FILL_8(dest, -2, c); |
| 212 |
case 1: FILL_8(dest, -1, c); |
| 213 |
} while (--n > 0); |
| 214 |
} |
| 215 |
} |
| 216 |
|
| 217 |
// 32-bit cell to fill? |
| 218 |
if (length & 4) { |
| 219 |
FILL_4(dest, 0, color); |
| 220 |
if (bpp <= 16) |
| 221 |
dest += 4; |
| 222 |
} |
| 223 |
|
| 224 |
// 16-bit cell to fill? |
| 225 |
if (bpp <= 16 && (length & 2)) { |
| 226 |
FILL_2(dest, 0, color); |
| 227 |
if (bpp <= 8) |
| 228 |
dest += 2; |
| 229 |
} |
| 230 |
|
| 231 |
// 8-bit cell to fill? |
| 232 |
if (bpp <= 8 && (length & 1)) |
| 233 |
FILL_1(dest, 0, color); |
| 234 |
|
| 235 |
#undef FILL_1 |
| 236 |
#undef FILL_2 |
| 237 |
#undef FILL_4 |
| 238 |
#undef FILL_8 |
| 239 |
} |
| 240 |
|
| 241 |
void NQD_fillrect(uint32 p) |
| 242 |
{ |
| 243 |
D(bug("accl_fillrect %08x\n", p)); |
| 244 |
|
| 245 |
// Get filling parameters |
| 246 |
int16 dest_X = (int16)ReadMacInt16(p + acclDestRect + 2) - (int16)ReadMacInt16(p + acclDestBoundsRect + 2); |
| 247 |
int16 dest_Y = (int16)ReadMacInt16(p + acclDestRect + 0) - (int16)ReadMacInt16(p + acclDestBoundsRect + 0); |
| 248 |
int16 width = (int16)ReadMacInt16(p + acclDestRect + 6) - (int16)ReadMacInt16(p + acclDestRect + 2); |
| 249 |
int16 height = (int16)ReadMacInt16(p + acclDestRect + 4) - (int16)ReadMacInt16(p + acclDestRect + 0); |
| 250 |
uint32 color = htonl(ReadMacInt32(p + acclPenMode) == 8 ? ReadMacInt32(p + acclForePen) : ReadMacInt32(p + acclBackPen)); |
| 251 |
D(bug(" dest X %d, dest Y %d\n", dest_X, dest_Y)); |
| 252 |
D(bug(" width %d, height %d\n", width, height)); |
| 253 |
D(bug(" bytes_per_row %d color %08x\n", (int32)ReadMacInt32(p + acclDestRowBytes), color)); |
| 254 |
|
| 255 |
// And perform the fill |
| 256 |
const int bpp = bytes_per_pixel(ReadMacInt32(p + acclDestPixelSize)); |
| 257 |
const int dest_row_bytes = (int32)ReadMacInt32(p + acclDestRowBytes); |
| 258 |
uint8 *dest = Mac2HostAddr(ReadMacInt32(p + acclDestBaseAddr) + (dest_Y * dest_row_bytes) + (dest_X * bpp)); |
| 259 |
width *= bpp; |
| 260 |
switch (bpp) { |
| 261 |
case 1: |
| 262 |
for (int i = 0; i < height; i++) { |
| 263 |
memset(dest, color, width); |
| 264 |
dest += dest_row_bytes; |
| 265 |
} |
| 266 |
break; |
| 267 |
case 2: |
| 268 |
for (int i = 0; i < height; i++) { |
| 269 |
do_fillrect<16>(dest, color, width); |
| 270 |
dest += dest_row_bytes; |
| 271 |
} |
| 272 |
break; |
| 273 |
case 4: |
| 274 |
for (int i = 0; i < height; i++) { |
| 275 |
do_fillrect<32>(dest, color, width); |
| 276 |
dest += dest_row_bytes; |
| 277 |
} |
| 278 |
break; |
| 279 |
} |
| 280 |
} |
| 281 |
|
| 282 |
bool NQD_fillrect_hook(uint32 p) |
| 283 |
{ |
| 284 |
D(bug("accl_fillrect_hook %08x\n", p)); |
| 285 |
|
| 286 |
// Check if we can accelerate this fillrect |
| 287 |
if (ReadMacInt32(p + 0x284) != 0 && ReadMacInt32(p + acclDestPixelSize) >= 8) { |
| 288 |
const int transfer_mode = ReadMacInt32(p + acclTransferMode); |
| 289 |
if (transfer_mode == 8) { |
| 290 |
// Fill |
| 291 |
WriteMacInt32(p + acclDrawProc, NativeTVECT(NATIVE_FILLRECT)); |
| 292 |
return true; |
| 293 |
} |
| 294 |
else if (transfer_mode == 10) { |
| 295 |
// Invert |
| 296 |
WriteMacInt32(p + acclDrawProc, NativeTVECT(NATIVE_INVRECT)); |
| 297 |
return true; |
| 298 |
} |
| 299 |
} |
| 300 |
return false; |
| 301 |
} |
| 302 |
|
| 303 |
|
| 304 |
/* |
| 305 |
* Isomorphic rectangle blitting |
| 306 |
*/ |
| 307 |
|
| 308 |
// TODO: optimize for VOSF and target pixmap == screen |
| 309 |
void NQD_bitblt(uint32 p) |
| 310 |
{ |
| 311 |
D(bug("accl_bitblt %08x\n", p)); |
| 312 |
|
| 313 |
// Get blitting parameters |
| 314 |
int16 src_X = (int16)ReadMacInt16(p + acclSrcRect + 2) - (int16)ReadMacInt16(p + acclSrcBoundsRect + 2); |
| 315 |
int16 src_Y = (int16)ReadMacInt16(p + acclSrcRect + 0) - (int16)ReadMacInt16(p + acclSrcBoundsRect + 0); |
| 316 |
int16 dest_X = (int16)ReadMacInt16(p + acclDestRect + 2) - (int16)ReadMacInt16(p + acclDestBoundsRect + 2); |
| 317 |
int16 dest_Y = (int16)ReadMacInt16(p + acclDestRect + 0) - (int16)ReadMacInt16(p + acclDestBoundsRect + 0); |
| 318 |
int16 width = (int16)ReadMacInt16(p + acclDestRect + 6) - (int16)ReadMacInt16(p + acclDestRect + 2); |
| 319 |
int16 height = (int16)ReadMacInt16(p + acclDestRect + 4) - (int16)ReadMacInt16(p + acclDestRect + 0); |
| 320 |
D(bug(" src addr %08x, dest addr %08x\n", ReadMacInt32(p + acclSrcBaseAddr), ReadMacInt32(p + acclDestBaseAddr))); |
| 321 |
D(bug(" src X %d, src Y %d, dest X %d, dest Y %d\n", src_X, src_Y, dest_X, dest_Y)); |
| 322 |
D(bug(" width %d, height %d\n", width, height)); |
| 323 |
|
| 324 |
// And perform the blit |
| 325 |
const int bpp = bytes_per_pixel(ReadMacInt32(p + acclSrcPixelSize)); |
| 326 |
width *= bpp; |
| 327 |
if ((int32)ReadMacInt32(p + acclSrcRowBytes) > 0) { |
| 328 |
const int src_row_bytes = (int32)ReadMacInt32(p + acclSrcRowBytes); |
| 329 |
const int dst_row_bytes = (int32)ReadMacInt32(p + acclDestRowBytes); |
| 330 |
uint8 *src = Mac2HostAddr(ReadMacInt32(p + acclSrcBaseAddr) + (src_Y * src_row_bytes) + (src_X * bpp)); |
| 331 |
uint8 *dst = Mac2HostAddr(ReadMacInt32(p + acclDestBaseAddr) + (dest_Y * dst_row_bytes) + (dest_X * bpp)); |
| 332 |
for (int i = 0; i < height; i++) { |
| 333 |
memmove(dst, src, width); |
| 334 |
src += src_row_bytes; |
| 335 |
dst += dst_row_bytes; |
| 336 |
} |
| 337 |
} |
| 338 |
else { |
| 339 |
const int src_row_bytes = -(int32)ReadMacInt32(p + acclSrcRowBytes); |
| 340 |
const int dst_row_bytes = -(int32)ReadMacInt32(p + acclDestRowBytes); |
| 341 |
uint8 *src = Mac2HostAddr(ReadMacInt32(p + acclSrcBaseAddr) + ((src_Y + height - 1) * src_row_bytes) + (src_X * bpp)); |
| 342 |
uint8 *dst = Mac2HostAddr(ReadMacInt32(p + acclDestBaseAddr) + ((dest_Y + height - 1) * dst_row_bytes) + (dest_X * bpp)); |
| 343 |
for (int i = height - 1; i >= 0; i--) { |
| 344 |
memmove(dst, src, width); |
| 345 |
src -= src_row_bytes; |
| 346 |
dst -= dst_row_bytes; |
| 347 |
} |
| 348 |
} |
| 349 |
} |
| 350 |
|
| 351 |
/* |
| 352 |
BitBlt transfer modes: |
| 353 |
0 : srcCopy |
| 354 |
1 : srcOr |
| 355 |
2 : srcXor |
| 356 |
3 : srcBic |
| 357 |
4 : notSrcCopy |
| 358 |
5 : notSrcOr |
| 359 |
6 : notSrcXor |
| 360 |
7 : notSrcBic |
| 361 |
32 : blend |
| 362 |
33 : addPin |
| 363 |
34 : addOver |
| 364 |
35 : subPin |
| 365 |
36 : transparent |
| 366 |
37 : adMax |
| 367 |
38 : subOver |
| 368 |
39 : adMin |
| 369 |
50 : hilite |
| 370 |
*/ |
| 371 |
|
| 372 |
bool NQD_bitblt_hook(uint32 p) |
| 373 |
{ |
| 374 |
D(bug("accl_draw_hook %08x\n", p)); |
| 375 |
|
| 376 |
// Check if we can accelerate this bitblt |
| 377 |
if (ReadMacInt32(p + 0x018) + ReadMacInt32(p + 0x128) == 0 && |
| 378 |
ReadMacInt32(p + 0x130) == 0 && |
| 379 |
ReadMacInt32(p + acclSrcPixelSize) >= 8 && |
| 380 |
ReadMacInt32(p + acclSrcPixelSize) == ReadMacInt32(p + acclDestPixelSize) && |
| 381 |
(ReadMacInt32(p + acclSrcRowBytes) ^ ReadMacInt32(p + acclDestRowBytes)) >= 0 && // same sign? |
| 382 |
ReadMacInt32(p + acclTransferMode) == 0 && // srcCopy? |
| 383 |
ReadMacInt32(p + 0x15c) > 0) { |
| 384 |
|
| 385 |
// Yes, set function pointer |
| 386 |
WriteMacInt32(p + acclDrawProc, NativeTVECT(NATIVE_BITBLT)); |
| 387 |
return true; |
| 388 |
} |
| 389 |
return false; |
| 390 |
} |
| 391 |
|
| 392 |
// Wait for graphics operation to finish |
| 393 |
bool NQD_sync_hook(uint32 arg) |
| 394 |
{ |
| 395 |
D(bug("accl_sync_hook %08x\n", arg)); |
| 396 |
return true; |
| 397 |
} |
| 398 |
|
| 399 |
|
| 400 |
/* |
| 401 |
* Install Native QuickDraw acceleration hooks |
| 402 |
*/ |
| 403 |
|
| 404 |
void VideoInstallAccel(void) |
| 405 |
{ |
| 406 |
// Install acceleration hooks |
| 407 |
if (PrefsFindBool("gfxaccel")) { |
| 408 |
D(bug("Video: Installing acceleration hooks\n")); |
| 409 |
uint32 base; |
| 410 |
|
| 411 |
SheepVar bitblt_hook_info(sizeof(accl_hook_info)); |
| 412 |
base = bitblt_hook_info.addr(); |
| 413 |
WriteMacInt32(base + 0, NativeTVECT(NATIVE_BITBLT_HOOK)); |
| 414 |
WriteMacInt32(base + 4, NativeTVECT(NATIVE_SYNC_HOOK)); |
| 415 |
WriteMacInt32(base + 8, ACCL_BITBLT); |
| 416 |
NQDMisc(6, bitblt_hook_info.ptr()); |
| 417 |
|
| 418 |
SheepVar fillrect_hook_info(sizeof(accl_hook_info)); |
| 419 |
base = fillrect_hook_info.addr(); |
| 420 |
WriteMacInt32(base + 0, NativeTVECT(NATIVE_FILLRECT_HOOK)); |
| 421 |
WriteMacInt32(base + 4, NativeTVECT(NATIVE_SYNC_HOOK)); |
| 422 |
WriteMacInt32(base + 8, ACCL_FILLRECT); |
| 423 |
NQDMisc(6, fillrect_hook_info.ptr()); |
| 424 |
} |
| 425 |
} |
