| 1 | /*************************************************************************** |
| 2 | * Copyright (C) 2010 PCSX4ALL Team * |
| 3 | * Copyright (C) 2010 Unai * |
| 4 | * Copyright (C) 2011 notaz * |
| 5 | * Copyright (C) 2016 Senquack (dansilsby <AT> gmail <DOT> com) * |
| 6 | * * |
| 7 | * This program is free software; you can redistribute it and/or modify * |
| 8 | * it under the terms of the GNU General Public License as published by * |
| 9 | * the Free Software Foundation; either version 2 of the License, or * |
| 10 | * (at your option) any later version. * |
| 11 | * * |
| 12 | * This program is distributed in the hope that it will be useful, * |
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of * |
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * |
| 15 | * GNU General Public License for more details. * |
| 16 | * * |
| 17 | * You should have received a copy of the GNU General Public License * |
| 18 | * along with this program; if not, write to the * |
| 19 | * Free Software Foundation, Inc., * |
| 20 | * 51 Franklin Street, Fifth Floor, Boston, MA 02111-1307 USA. * |
| 21 | ***************************************************************************/ |
| 22 | |
| 23 | #include <stddef.h> |
| 24 | #include <stdio.h> |
| 25 | #include <stdlib.h> |
| 26 | #include <string.h> |
| 27 | #include "../gpulib/gpu.h" |
| 28 | |
| 29 | #ifdef THREAD_RENDERING |
| 30 | #include "../gpulib/gpulib_thread_if.h" |
| 31 | #define do_cmd_list real_do_cmd_list |
| 32 | #define renderer_init real_renderer_init |
| 33 | #define renderer_finish real_renderer_finish |
| 34 | #define renderer_sync_ecmds real_renderer_sync_ecmds |
| 35 | #define renderer_update_caches real_renderer_update_caches |
| 36 | #define renderer_flush_queues real_renderer_flush_queues |
| 37 | #define renderer_set_interlace real_renderer_set_interlace |
| 38 | #define renderer_set_config real_renderer_set_config |
| 39 | #define renderer_notify_res_change real_renderer_notify_res_change |
| 40 | #define renderer_notify_update_lace real_renderer_notify_update_lace |
| 41 | #define renderer_sync real_renderer_sync |
| 42 | #define ex_regs scratch_ex_regs |
| 43 | #endif |
| 44 | |
| 45 | //#include "port.h" |
| 46 | #include "gpu_unai.h" |
| 47 | |
| 48 | // GPU fixed point math |
| 49 | #include "gpu_fixedpoint.h" |
| 50 | |
| 51 | // Inner loop driver instantiation file |
| 52 | #include "gpu_inner.h" |
| 53 | |
| 54 | // GPU internal image drawing functions |
| 55 | #include "gpu_raster_image.h" |
| 56 | |
| 57 | // GPU internal line drawing functions |
| 58 | #include "gpu_raster_line.h" |
| 59 | |
| 60 | // GPU internal polygon drawing functions |
| 61 | #include "gpu_raster_polygon.h" |
| 62 | |
| 63 | // GPU internal sprite drawing functions |
| 64 | #include "gpu_raster_sprite.h" |
| 65 | |
| 66 | // GPU command buffer execution/store |
| 67 | #include "gpu_command.h" |
| 68 | |
| 69 | ///////////////////////////////////////////////////////////////////////////// |
| 70 | |
| 71 | #define DOWNSCALE_VRAM_SIZE (1024 * 512 * 2 * 2 + 4096) |
| 72 | |
| 73 | INLINE void scale_640_to_320(le16_t *dest, const le16_t *src, bool isRGB24) { |
| 74 | size_t uCount = 320; |
| 75 | |
| 76 | if(isRGB24) { |
| 77 | const uint8_t* src8 = (const uint8_t *)src; |
| 78 | uint8_t* dst8 = (uint8_t *)dest; |
| 79 | |
| 80 | do { |
| 81 | *dst8++ = *src8++; |
| 82 | *dst8++ = *src8++; |
| 83 | *dst8++ = *src8; |
| 84 | src8 += 4; |
| 85 | } while(--uCount); |
| 86 | } else { |
| 87 | const le16_t* src16 = src; |
| 88 | le16_t* dst16 = dest; |
| 89 | |
| 90 | do { |
| 91 | *dst16++ = *src16; |
| 92 | src16 += 2; |
| 93 | } while(--uCount); |
| 94 | } |
| 95 | } |
| 96 | |
| 97 | INLINE void scale_512_to_320(le16_t *dest, const le16_t *src, bool isRGB24) { |
| 98 | size_t uCount = 64; |
| 99 | |
| 100 | if(isRGB24) { |
| 101 | const uint8_t* src8 = (const uint8_t *)src; |
| 102 | uint8_t* dst8 = (uint8_t *)dest; |
| 103 | |
| 104 | do { |
| 105 | *dst8++ = *src8++; |
| 106 | *dst8++ = *src8++; |
| 107 | *dst8++ = *src8++; |
| 108 | *dst8++ = *src8++; |
| 109 | *dst8++ = *src8++; |
| 110 | *dst8++ = *src8; |
| 111 | src8 += 4; |
| 112 | *dst8++ = *src8++; |
| 113 | *dst8++ = *src8++; |
| 114 | *dst8++ = *src8++; |
| 115 | *dst8++ = *src8++; |
| 116 | *dst8++ = *src8++; |
| 117 | *dst8++ = *src8; |
| 118 | src8 += 4; |
| 119 | *dst8++ = *src8++; |
| 120 | *dst8++ = *src8++; |
| 121 | *dst8++ = *src8; |
| 122 | src8 += 4; |
| 123 | } while(--uCount); |
| 124 | } else { |
| 125 | const le16_t* src16 = src; |
| 126 | le16_t* dst16 = dest; |
| 127 | |
| 128 | do { |
| 129 | *dst16++ = *src16++; |
| 130 | *dst16++ = *src16; |
| 131 | src16 += 2; |
| 132 | *dst16++ = *src16++; |
| 133 | *dst16++ = *src16; |
| 134 | src16 += 2; |
| 135 | *dst16++ = *src16; |
| 136 | src16 += 2; |
| 137 | } while(--uCount); |
| 138 | } |
| 139 | } |
| 140 | |
| 141 | static uint16_t *get_downscale_buffer(int *x, int *y, int *w, int *h, int *vram_h) |
| 142 | { |
| 143 | le16_t *dest = gpu_unai.downscale_vram; |
| 144 | const le16_t *src = gpu_unai.vram; |
| 145 | bool isRGB24 = (gpu_unai.GPU_GP1 & 0x00200000 ? true : false); |
| 146 | int stride = 1024, dstride = 1024, lines = *h, orig_w = *w; |
| 147 | |
| 148 | // PS1 fb read wraps around (fixes black screen in 'Tobal no. 1') |
| 149 | unsigned int fb_mask = 1024 * 512 - 1; |
| 150 | |
| 151 | if (*h > 240) { |
| 152 | *h /= 2; |
| 153 | stride *= 2; |
| 154 | lines = *h; |
| 155 | |
| 156 | // Ensure start at a non-skipped line |
| 157 | while (*y & gpu_unai.ilace_mask) ++*y; |
| 158 | } |
| 159 | |
| 160 | unsigned int fb_offset_src = (*y * dstride + *x) & fb_mask; |
| 161 | unsigned int fb_offset_dest = fb_offset_src; |
| 162 | |
| 163 | if (*w == 512 || *w == 640) { |
| 164 | *w = 320; |
| 165 | } |
| 166 | |
| 167 | switch(orig_w) { |
| 168 | case 640: |
| 169 | do { |
| 170 | scale_640_to_320(dest + fb_offset_dest, src + fb_offset_src, isRGB24); |
| 171 | fb_offset_src = (fb_offset_src + stride) & fb_mask; |
| 172 | fb_offset_dest = (fb_offset_dest + dstride) & fb_mask; |
| 173 | } while(--lines); |
| 174 | |
| 175 | break; |
| 176 | case 512: |
| 177 | do { |
| 178 | scale_512_to_320(dest + fb_offset_dest, src + fb_offset_src, isRGB24); |
| 179 | fb_offset_src = (fb_offset_src + stride) & fb_mask; |
| 180 | fb_offset_dest = (fb_offset_dest + dstride) & fb_mask; |
| 181 | } while(--lines); |
| 182 | break; |
| 183 | default: |
| 184 | size_t size = isRGB24 ? *w * 3 : *w * 2; |
| 185 | |
| 186 | do { |
| 187 | memcpy(dest + fb_offset_dest, src + fb_offset_src, size); |
| 188 | fb_offset_src = (fb_offset_src + stride) & fb_mask; |
| 189 | fb_offset_dest = (fb_offset_dest + dstride) & fb_mask; |
| 190 | } while(--lines); |
| 191 | break; |
| 192 | } |
| 193 | |
| 194 | return (uint16_t *)gpu_unai.downscale_vram; |
| 195 | } |
| 196 | |
| 197 | static void map_downscale_buffer(void) |
| 198 | { |
| 199 | if (gpu_unai.downscale_vram) |
| 200 | return; |
| 201 | |
| 202 | gpu_unai.downscale_vram = (le16_t*)gpu.mmap(DOWNSCALE_VRAM_SIZE); |
| 203 | |
| 204 | if (gpu_unai.downscale_vram == NULL) { |
| 205 | fprintf(stderr, "failed to map downscale buffer\n"); |
| 206 | gpu.get_downscale_buffer = NULL; |
| 207 | } |
| 208 | else { |
| 209 | gpu.get_downscale_buffer = get_downscale_buffer; |
| 210 | } |
| 211 | } |
| 212 | |
| 213 | static void unmap_downscale_buffer(void) |
| 214 | { |
| 215 | if (gpu_unai.downscale_vram == NULL) |
| 216 | return; |
| 217 | |
| 218 | gpu.munmap(gpu_unai.downscale_vram, DOWNSCALE_VRAM_SIZE); |
| 219 | gpu_unai.downscale_vram = NULL; |
| 220 | gpu.get_downscale_buffer = NULL; |
| 221 | } |
| 222 | |
| 223 | int renderer_init(void) |
| 224 | { |
| 225 | memset((void*)&gpu_unai, 0, sizeof(gpu_unai)); |
| 226 | gpu_unai.vram = (le16_t *)gpu.vram; |
| 227 | |
| 228 | // Original standalone gpu_unai initialized TextureWindow[]. I added the |
| 229 | // same behavior here, since it seems unsafe to leave [2],[3] unset when |
| 230 | // using HLE and Rearmed gpu_neon sets this similarly on init. -senquack |
| 231 | gpu_unai.TextureWindow[0] = 0; |
| 232 | gpu_unai.TextureWindow[1] = 0; |
| 233 | gpu_unai.TextureWindow[2] = 255; |
| 234 | gpu_unai.TextureWindow[3] = 255; |
| 235 | //senquack - new vars must be updated whenever texture window is changed: |
| 236 | // (used for polygon-drawing in gpu_inner.h, gpu_raster_polygon.h) |
| 237 | const u32 fb = FIXED_BITS; // # of fractional fixed-pt bits of u4/v4 |
| 238 | gpu_unai.u_msk = (((u32)gpu_unai.TextureWindow[2]) << fb) | ((1 << fb) - 1); |
| 239 | gpu_unai.v_msk = (((u32)gpu_unai.TextureWindow[3]) << fb) | ((1 << fb) - 1); |
| 240 | |
| 241 | // Configuration options |
| 242 | gpu_unai.config = gpu_unai_config_ext; |
| 243 | //senquack - disabled, not sure this is needed and would require modifying |
| 244 | // sprite-span functions, perhaps unnecessarily. No Abe Oddysey hack was |
| 245 | // present in latest PCSX4ALL sources we were using. |
| 246 | //gpu_unai.config.enableAbbeyHack = gpu_unai_config_ext.abe_hack; |
| 247 | gpu_unai.ilace_mask = gpu_unai.config.ilace_force; |
| 248 | |
| 249 | #ifdef GPU_UNAI_USE_INT_DIV_MULTINV |
| 250 | // s_invTable |
| 251 | for(int i=1;i<=(1<<TABLE_BITS);++i) |
| 252 | { |
| 253 | double v = 1.0 / double(i); |
| 254 | #ifdef GPU_TABLE_10_BITS |
| 255 | v *= double(0xffffffff>>1); |
| 256 | #else |
| 257 | v *= double(0x80000000); |
| 258 | #endif |
| 259 | s_invTable[i-1]=s32(v); |
| 260 | } |
| 261 | #endif |
| 262 | |
| 263 | SetupLightLUT(); |
| 264 | SetupDitheringConstants(); |
| 265 | |
| 266 | if (gpu_unai.config.scale_hires) { |
| 267 | map_downscale_buffer(); |
| 268 | } |
| 269 | |
| 270 | return 0; |
| 271 | } |
| 272 | |
| 273 | void renderer_finish(void) |
| 274 | { |
| 275 | unmap_downscale_buffer(); |
| 276 | } |
| 277 | |
| 278 | void renderer_notify_res_change(void) |
| 279 | { |
| 280 | if (PixelSkipEnabled()) { |
| 281 | // Set blit_mask for high horizontal resolutions. This allows skipping |
| 282 | // rendering pixels that would never get displayed on low-resolution |
| 283 | // platforms that use simple pixel-dropping scaler. |
| 284 | |
| 285 | switch (gpu.screen.hres) |
| 286 | { |
| 287 | case 512: gpu_unai.blit_mask = 0xa4; break; // GPU_BlitWWSWWSWS |
| 288 | case 640: gpu_unai.blit_mask = 0xaa; break; // GPU_BlitWS |
| 289 | default: gpu_unai.blit_mask = 0; break; |
| 290 | } |
| 291 | } else { |
| 292 | gpu_unai.blit_mask = 0; |
| 293 | } |
| 294 | |
| 295 | if (LineSkipEnabled()) { |
| 296 | // Set rendering line-skip (only render every other line in high-res |
| 297 | // 480 vertical mode, or, optionally, force it for all video modes) |
| 298 | |
| 299 | if (gpu.screen.vres == 480) { |
| 300 | if (gpu_unai.config.ilace_force) { |
| 301 | gpu_unai.ilace_mask = 3; // Only need 1/4 of lines |
| 302 | } else { |
| 303 | gpu_unai.ilace_mask = 1; // Only need 1/2 of lines |
| 304 | } |
| 305 | } else { |
| 306 | // Vert resolution changed from 480 to lower one |
| 307 | gpu_unai.ilace_mask = gpu_unai.config.ilace_force; |
| 308 | } |
| 309 | } else { |
| 310 | gpu_unai.ilace_mask = 0; |
| 311 | } |
| 312 | |
| 313 | /* |
| 314 | printf("res change hres: %d vres: %d depth: %d ilace_mask: %d\n", |
| 315 | gpu.screen.hres, gpu.screen.vres, (gpu.status & PSX_GPU_STATUS_RGB24) ? 24 : 15, |
| 316 | gpu_unai.ilace_mask); |
| 317 | */ |
| 318 | } |
| 319 | |
| 320 | void renderer_notify_scanout_change(int x, int y) |
| 321 | { |
| 322 | } |
| 323 | |
| 324 | #ifdef USE_GPULIB |
| 325 | // Handles GP0 draw settings commands 0xE1...0xE6 |
| 326 | static void gpuGP0Cmd_0xEx(gpu_unai_t &gpu_unai, u32 cmd_word) |
| 327 | { |
| 328 | // Assume incoming GP0 command is 0xE1..0xE6, convert to 1..6 |
| 329 | u8 num = (cmd_word >> 24) & 7; |
| 330 | gpu.ex_regs[num] = cmd_word; // Update gpulib register |
| 331 | switch (num) { |
| 332 | case 1: { |
| 333 | // GP0(E1h) - Draw Mode setting (aka "Texpage") |
| 334 | u32 cur_texpage = gpu_unai.GPU_GP1 & 0x7FF; |
| 335 | u32 new_texpage = cmd_word & 0x7FF; |
| 336 | if (cur_texpage != new_texpage) { |
| 337 | gpu_unai.GPU_GP1 = (gpu_unai.GPU_GP1 & ~0x7FF) | new_texpage; |
| 338 | gpuSetTexture(gpu_unai.GPU_GP1); |
| 339 | } |
| 340 | } break; |
| 341 | |
| 342 | case 2: { |
| 343 | // GP0(E2h) - Texture Window setting |
| 344 | if (cmd_word != gpu_unai.TextureWindowCur) { |
| 345 | static const u8 TextureMask[32] = { |
| 346 | 255, 7, 15, 7, 31, 7, 15, 7, 63, 7, 15, 7, 31, 7, 15, 7, |
| 347 | 127, 7, 15, 7, 31, 7, 15, 7, 63, 7, 15, 7, 31, 7, 15, 7 |
| 348 | }; |
| 349 | gpu_unai.TextureWindowCur = cmd_word; |
| 350 | gpu_unai.TextureWindow[0] = ((cmd_word >> 10) & 0x1F) << 3; |
| 351 | gpu_unai.TextureWindow[1] = ((cmd_word >> 15) & 0x1F) << 3; |
| 352 | gpu_unai.TextureWindow[2] = TextureMask[(cmd_word >> 0) & 0x1F]; |
| 353 | gpu_unai.TextureWindow[3] = TextureMask[(cmd_word >> 5) & 0x1F]; |
| 354 | gpu_unai.TextureWindow[0] &= ~gpu_unai.TextureWindow[2]; |
| 355 | gpu_unai.TextureWindow[1] &= ~gpu_unai.TextureWindow[3]; |
| 356 | |
| 357 | // Inner loop vars must be updated whenever texture window is changed: |
| 358 | const u32 fb = FIXED_BITS; // # of fractional fixed-pt bits of u4/v4 |
| 359 | gpu_unai.u_msk = (((u32)gpu_unai.TextureWindow[2]) << fb) | ((1 << fb) - 1); |
| 360 | gpu_unai.v_msk = (((u32)gpu_unai.TextureWindow[3]) << fb) | ((1 << fb) - 1); |
| 361 | |
| 362 | gpuSetTexture(gpu_unai.GPU_GP1); |
| 363 | } |
| 364 | } break; |
| 365 | |
| 366 | case 3: { |
| 367 | // GP0(E3h) - Set Drawing Area top left (X1,Y1) |
| 368 | gpu_unai.DrawingArea[0] = cmd_word & 0x3FF; |
| 369 | gpu_unai.DrawingArea[1] = (cmd_word >> 10) & 0x3FF; |
| 370 | } break; |
| 371 | |
| 372 | case 4: { |
| 373 | // GP0(E4h) - Set Drawing Area bottom right (X2,Y2) |
| 374 | gpu_unai.DrawingArea[2] = (cmd_word & 0x3FF) + 1; |
| 375 | gpu_unai.DrawingArea[3] = ((cmd_word >> 10) & 0x3FF) + 1; |
| 376 | } break; |
| 377 | |
| 378 | case 5: { |
| 379 | // GP0(E5h) - Set Drawing Offset (X,Y) |
| 380 | gpu_unai.DrawingOffset[0] = ((s32)cmd_word<<(32-11))>>(32-11); |
| 381 | gpu_unai.DrawingOffset[1] = ((s32)cmd_word<<(32-22))>>(32-11); |
| 382 | } break; |
| 383 | |
| 384 | case 6: { |
| 385 | // GP0(E6h) - Mask Bit Setting |
| 386 | gpu_unai.Masking = (cmd_word & 0x2) << 1; |
| 387 | gpu_unai.PixelMSB = (cmd_word & 0x1) << 8; |
| 388 | } break; |
| 389 | } |
| 390 | } |
| 391 | #endif |
| 392 | |
| 393 | #include "../gpulib/gpu_timing.h" |
| 394 | extern const unsigned char cmd_lengths[256]; |
| 395 | |
| 396 | int do_cmd_list(u32 *_list, int list_len, int *cpu_cycles_out, int *last_cmd) |
| 397 | { |
| 398 | u32 cmd = 0, len, i; |
| 399 | le32_t *list = (le32_t *)_list; |
| 400 | le32_t *list_start = list; |
| 401 | le32_t *list_end = list + list_len; |
| 402 | u32 cpu_cycles = 0; |
| 403 | |
| 404 | //TODO: set ilace_mask when resolution changes instead of every time, |
| 405 | // eliminate #ifdef below. |
| 406 | gpu_unai.ilace_mask = gpu_unai.config.ilace_force; |
| 407 | |
| 408 | #ifdef HAVE_PRE_ARMV7 /* XXX */ |
| 409 | gpu_unai.ilace_mask |= !!(gpu.status & PSX_GPU_STATUS_INTERLACE); |
| 410 | #endif |
| 411 | if (gpu_unai.config.scale_hires) { |
| 412 | gpu_unai.ilace_mask |= !!(gpu.status & PSX_GPU_STATUS_INTERLACE); |
| 413 | } |
| 414 | |
| 415 | for (; list < list_end; list += 1 + len) |
| 416 | { |
| 417 | cmd = le32_to_u32(*list) >> 24; |
| 418 | len = cmd_lengths[cmd]; |
| 419 | if (list + 1 + len > list_end) { |
| 420 | cmd = -1; |
| 421 | break; |
| 422 | } |
| 423 | |
| 424 | #define PRIM cmd |
| 425 | gpu_unai.PacketBuffer.U4[0] = list[0]; |
| 426 | for (i = 1; i <= len; i++) |
| 427 | gpu_unai.PacketBuffer.U4[i] = list[i]; |
| 428 | |
| 429 | PtrUnion packet = { .ptr = (void*)&gpu_unai.PacketBuffer }; |
| 430 | |
| 431 | switch (cmd) |
| 432 | { |
| 433 | case 0x02: |
| 434 | gpuClearImage(packet); |
| 435 | cpu_cycles += gput_fill(le16_to_s16(packet.U2[4]) & 0x3ff, |
| 436 | le16_to_s16(packet.U2[5]) & 0x1ff); |
| 437 | break; |
| 438 | |
| 439 | case 0x20: |
| 440 | case 0x21: |
| 441 | case 0x22: |
| 442 | case 0x23: { // Monochrome 3-pt poly |
| 443 | PP driver = gpuPolySpanDrivers[ |
| 444 | (gpu_unai.blit_mask?1024:0) | |
| 445 | Blending_Mode | |
| 446 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB |
| 447 | ]; |
| 448 | gpuDrawPolyF(packet, driver, false); |
| 449 | cpu_cycles += gput_poly_base(); |
| 450 | } break; |
| 451 | |
| 452 | case 0x24: |
| 453 | case 0x25: |
| 454 | case 0x26: |
| 455 | case 0x27: { // Textured 3-pt poly |
| 456 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 457 | gpuSetTexture(le32_to_u32(gpu_unai.PacketBuffer.U4[4]) >> 16); |
| 458 | |
| 459 | u32 driver_idx = |
| 460 | (gpu_unai.blit_mask?1024:0) | |
| 461 | Dithering | |
| 462 | Blending_Mode | gpu_unai.TEXT_MODE | |
| 463 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB; |
| 464 | |
| 465 | if (!FastLightingEnabled()) { |
| 466 | driver_idx |= Lighting; |
| 467 | } else { |
| 468 | if (!((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F))) |
| 469 | driver_idx |= Lighting; |
| 470 | } |
| 471 | |
| 472 | PP driver = gpuPolySpanDrivers[driver_idx]; |
| 473 | gpuDrawPolyFT(packet, driver, false); |
| 474 | cpu_cycles += gput_poly_base_t(); |
| 475 | } break; |
| 476 | |
| 477 | case 0x28: |
| 478 | case 0x29: |
| 479 | case 0x2A: |
| 480 | case 0x2B: { // Monochrome 4-pt poly |
| 481 | PP driver = gpuPolySpanDrivers[ |
| 482 | (gpu_unai.blit_mask?1024:0) | |
| 483 | Blending_Mode | |
| 484 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB |
| 485 | ]; |
| 486 | gpuDrawPolyF(packet, driver, true); // is_quad = true |
| 487 | cpu_cycles += gput_quad_base(); |
| 488 | } break; |
| 489 | |
| 490 | case 0x2C: |
| 491 | case 0x2D: |
| 492 | case 0x2E: |
| 493 | case 0x2F: { // Textured 4-pt poly |
| 494 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 495 | gpuSetTexture(le32_to_u32(gpu_unai.PacketBuffer.U4[4]) >> 16); |
| 496 | |
| 497 | u32 driver_idx = |
| 498 | (gpu_unai.blit_mask?1024:0) | |
| 499 | Dithering | |
| 500 | Blending_Mode | gpu_unai.TEXT_MODE | |
| 501 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB; |
| 502 | |
| 503 | if (!FastLightingEnabled()) { |
| 504 | driver_idx |= Lighting; |
| 505 | } else { |
| 506 | if (!((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F))) |
| 507 | driver_idx |= Lighting; |
| 508 | } |
| 509 | |
| 510 | PP driver = gpuPolySpanDrivers[driver_idx]; |
| 511 | gpuDrawPolyFT(packet, driver, true); // is_quad = true |
| 512 | cpu_cycles += gput_quad_base_t(); |
| 513 | } break; |
| 514 | |
| 515 | case 0x30: |
| 516 | case 0x31: |
| 517 | case 0x32: |
| 518 | case 0x33: { // Gouraud-shaded 3-pt poly |
| 519 | //NOTE: The '129' here is CF_GOURAUD | CF_LIGHT, however |
| 520 | // this is an untextured poly, so CF_LIGHT (texture blend) |
| 521 | // shouldn't apply. Until the original array of template |
| 522 | // instantiation ptrs is fixed, we're stuck with this. (TODO) |
| 523 | PP driver = gpuPolySpanDrivers[ |
| 524 | (gpu_unai.blit_mask?1024:0) | |
| 525 | Dithering | |
| 526 | Blending_Mode | |
| 527 | gpu_unai.Masking | Blending | 129 | gpu_unai.PixelMSB |
| 528 | ]; |
| 529 | gpuDrawPolyG(packet, driver, false); |
| 530 | cpu_cycles += gput_poly_base_g(); |
| 531 | } break; |
| 532 | |
| 533 | case 0x34: |
| 534 | case 0x35: |
| 535 | case 0x36: |
| 536 | case 0x37: { // Gouraud-shaded, textured 3-pt poly |
| 537 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 538 | gpuSetTexture (le32_to_u32(gpu_unai.PacketBuffer.U4[5]) >> 16); |
| 539 | PP driver = gpuPolySpanDrivers[ |
| 540 | (gpu_unai.blit_mask?1024:0) | |
| 541 | Dithering | |
| 542 | Blending_Mode | gpu_unai.TEXT_MODE | |
| 543 | gpu_unai.Masking | Blending | ((Lighting)?129:0) | gpu_unai.PixelMSB |
| 544 | ]; |
| 545 | gpuDrawPolyGT(packet, driver, false); |
| 546 | cpu_cycles += gput_poly_base_gt(); |
| 547 | } break; |
| 548 | |
| 549 | case 0x38: |
| 550 | case 0x39: |
| 551 | case 0x3A: |
| 552 | case 0x3B: { // Gouraud-shaded 4-pt poly |
| 553 | // See notes regarding '129' for 0x30..0x33 further above -senquack |
| 554 | PP driver = gpuPolySpanDrivers[ |
| 555 | (gpu_unai.blit_mask?1024:0) | |
| 556 | Dithering | |
| 557 | Blending_Mode | |
| 558 | gpu_unai.Masking | Blending | 129 | gpu_unai.PixelMSB |
| 559 | ]; |
| 560 | gpuDrawPolyG(packet, driver, true); // is_quad = true |
| 561 | cpu_cycles += gput_quad_base_g(); |
| 562 | } break; |
| 563 | |
| 564 | case 0x3C: |
| 565 | case 0x3D: |
| 566 | case 0x3E: |
| 567 | case 0x3F: { // Gouraud-shaded, textured 4-pt poly |
| 568 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 569 | gpuSetTexture (le32_to_u32(gpu_unai.PacketBuffer.U4[5]) >> 16); |
| 570 | PP driver = gpuPolySpanDrivers[ |
| 571 | (gpu_unai.blit_mask?1024:0) | |
| 572 | Dithering | |
| 573 | Blending_Mode | gpu_unai.TEXT_MODE | |
| 574 | gpu_unai.Masking | Blending | ((Lighting)?129:0) | gpu_unai.PixelMSB |
| 575 | ]; |
| 576 | gpuDrawPolyGT(packet, driver, true); // is_quad = true |
| 577 | cpu_cycles += gput_quad_base_gt(); |
| 578 | } break; |
| 579 | |
| 580 | case 0x40: |
| 581 | case 0x41: |
| 582 | case 0x42: |
| 583 | case 0x43: { // Monochrome line |
| 584 | // Shift index right by one, as untextured prims don't use lighting |
| 585 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
| 586 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
| 587 | gpuDrawLineF(packet, driver); |
| 588 | cpu_cycles += gput_line(0); |
| 589 | } break; |
| 590 | |
| 591 | case 0x48 ... 0x4F: { // Monochrome line strip |
| 592 | u32 num_vertexes = 1; |
| 593 | le32_t *list_position = &list[2]; |
| 594 | |
| 595 | // Shift index right by one, as untextured prims don't use lighting |
| 596 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
| 597 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
| 598 | gpuDrawLineF(packet, driver); |
| 599 | |
| 600 | while(1) |
| 601 | { |
| 602 | gpu_unai.PacketBuffer.U4[1] = gpu_unai.PacketBuffer.U4[2]; |
| 603 | gpu_unai.PacketBuffer.U4[2] = *list_position++; |
| 604 | gpuDrawLineF(packet, driver); |
| 605 | cpu_cycles += gput_line(0); |
| 606 | |
| 607 | num_vertexes++; |
| 608 | if(list_position >= list_end) { |
| 609 | cmd = -1; |
| 610 | goto breakloop; |
| 611 | } |
| 612 | if((le32_raw(*list_position) & HTOLE32(0xf000f000)) == HTOLE32(0x50005000)) |
| 613 | break; |
| 614 | } |
| 615 | |
| 616 | len += (num_vertexes - 2); |
| 617 | } break; |
| 618 | |
| 619 | case 0x50: |
| 620 | case 0x51: |
| 621 | case 0x52: |
| 622 | case 0x53: { // Gouraud-shaded line |
| 623 | // Shift index right by one, as untextured prims don't use lighting |
| 624 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
| 625 | // Index MSB selects Gouraud-shaded PixelSpanDriver: |
| 626 | driver_idx |= (1 << 5); |
| 627 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
| 628 | gpuDrawLineG(packet, driver); |
| 629 | cpu_cycles += gput_line(0); |
| 630 | } break; |
| 631 | |
| 632 | case 0x58 ... 0x5F: { // Gouraud-shaded line strip |
| 633 | u32 num_vertexes = 1; |
| 634 | le32_t *list_position = &list[2]; |
| 635 | |
| 636 | // Shift index right by one, as untextured prims don't use lighting |
| 637 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
| 638 | // Index MSB selects Gouraud-shaded PixelSpanDriver: |
| 639 | driver_idx |= (1 << 5); |
| 640 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
| 641 | gpuDrawLineG(packet, driver); |
| 642 | |
| 643 | while(1) |
| 644 | { |
| 645 | gpu_unai.PacketBuffer.U4[0] = gpu_unai.PacketBuffer.U4[2]; |
| 646 | gpu_unai.PacketBuffer.U4[1] = gpu_unai.PacketBuffer.U4[3]; |
| 647 | gpu_unai.PacketBuffer.U4[2] = *list_position++; |
| 648 | gpu_unai.PacketBuffer.U4[3] = *list_position++; |
| 649 | gpuDrawLineG(packet, driver); |
| 650 | cpu_cycles += gput_line(0); |
| 651 | |
| 652 | num_vertexes++; |
| 653 | if(list_position >= list_end) { |
| 654 | cmd = -1; |
| 655 | goto breakloop; |
| 656 | } |
| 657 | if((le32_raw(*list_position) & HTOLE32(0xf000f000)) == HTOLE32(0x50005000)) |
| 658 | break; |
| 659 | } |
| 660 | |
| 661 | len += (num_vertexes - 2) * 2; |
| 662 | } break; |
| 663 | |
| 664 | case 0x60: |
| 665 | case 0x61: |
| 666 | case 0x62: |
| 667 | case 0x63: { // Monochrome rectangle (variable size) |
| 668 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
| 669 | s32 w = 0, h = 0; |
| 670 | gpuDrawT(packet, driver, &w, &h); |
| 671 | cpu_cycles += gput_sprite(w, h); |
| 672 | } break; |
| 673 | |
| 674 | case 0x64: |
| 675 | case 0x65: |
| 676 | case 0x66: |
| 677 | case 0x67: { // Textured rectangle (variable size) |
| 678 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 679 | u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); |
| 680 | s32 w = 0, h = 0; |
| 681 | |
| 682 | //senquack - Only color 808080h-878787h allows skipping lighting calculation: |
| 683 | // This fixes Silent Hill running animation on loading screens: |
| 684 | // (On PSX, color values 0x00-0x7F darken the source texture's color, |
| 685 | // 0x81-FF lighten textures (ultimately clamped to 0x1F), |
| 686 | // 0x80 leaves source texture color unchanged, HOWEVER, |
| 687 | // gpu_unai uses a simple lighting LUT whereby only the upper |
| 688 | // 5 bits of an 8-bit color are used, so 0x80-0x87 all behave as |
| 689 | // 0x80. |
| 690 | // |
| 691 | // NOTE: I've changed all textured sprite draw commands here and |
| 692 | // elsewhere to use proper behavior, but left poly commands |
| 693 | // alone, I don't want to slow rendering down too much. (TODO) |
| 694 | //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) |
| 695 | // Strip lower 3 bits of each color and determine if lighting should be used: |
| 696 | if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) |
| 697 | driver_idx |= Lighting; |
| 698 | PS driver = gpuSpriteSpanDrivers[driver_idx]; |
| 699 | gpuDrawS(packet, driver, &w, &h); |
| 700 | cpu_cycles += gput_sprite(w, h); |
| 701 | } break; |
| 702 | |
| 703 | case 0x68: |
| 704 | case 0x69: |
| 705 | case 0x6A: |
| 706 | case 0x6B: { // Monochrome rectangle (1x1 dot) |
| 707 | gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00010001); |
| 708 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
| 709 | s32 w = 0, h = 0; |
| 710 | gpuDrawT(packet, driver, &w, &h); |
| 711 | cpu_cycles += gput_sprite(1, 1); |
| 712 | } break; |
| 713 | |
| 714 | case 0x70: |
| 715 | case 0x71: |
| 716 | case 0x72: |
| 717 | case 0x73: { // Monochrome rectangle (8x8) |
| 718 | gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00080008); |
| 719 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
| 720 | s32 w = 0, h = 0; |
| 721 | gpuDrawT(packet, driver, &w, &h); |
| 722 | cpu_cycles += gput_sprite(w, h); |
| 723 | } break; |
| 724 | |
| 725 | case 0x74: |
| 726 | case 0x75: |
| 727 | case 0x76: |
| 728 | case 0x77: { // Textured rectangle (8x8) |
| 729 | gpu_unai.PacketBuffer.U4[3] = u32_to_le32(0x00080008); |
| 730 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 731 | u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); |
| 732 | s32 w = 0, h = 0; |
| 733 | |
| 734 | //senquack - Only color 808080h-878787h allows skipping lighting calculation: |
| 735 | //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) |
| 736 | // Strip lower 3 bits of each color and determine if lighting should be used: |
| 737 | if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) |
| 738 | driver_idx |= Lighting; |
| 739 | PS driver = gpuSpriteSpanDrivers[driver_idx]; |
| 740 | gpuDrawS(packet, driver, &w, &h); |
| 741 | cpu_cycles += gput_sprite(w, h); |
| 742 | } break; |
| 743 | |
| 744 | case 0x78: |
| 745 | case 0x79: |
| 746 | case 0x7A: |
| 747 | case 0x7B: { // Monochrome rectangle (16x16) |
| 748 | gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00100010); |
| 749 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
| 750 | s32 w = 0, h = 0; |
| 751 | gpuDrawT(packet, driver, &w, &h); |
| 752 | cpu_cycles += gput_sprite(w, h); |
| 753 | } break; |
| 754 | |
| 755 | case 0x7C: |
| 756 | case 0x7D: |
| 757 | #ifdef __arm__ |
| 758 | if ((gpu_unai.GPU_GP1 & 0x180) == 0 && (gpu_unai.Masking | gpu_unai.PixelMSB) == 0) |
| 759 | { |
| 760 | s32 w = 0, h = 0; |
| 761 | gpuSetCLUT(le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 762 | gpuDrawS16(packet, &w, &h); |
| 763 | cpu_cycles += gput_sprite(w, h); |
| 764 | break; |
| 765 | } |
| 766 | // fallthrough |
| 767 | #endif |
| 768 | case 0x7E: |
| 769 | case 0x7F: { // Textured rectangle (16x16) |
| 770 | gpu_unai.PacketBuffer.U4[3] = u32_to_le32(0x00100010); |
| 771 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
| 772 | u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); |
| 773 | s32 w = 0, h = 0; |
| 774 | //senquack - Only color 808080h-878787h allows skipping lighting calculation: |
| 775 | //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) |
| 776 | // Strip lower 3 bits of each color and determine if lighting should be used: |
| 777 | if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) |
| 778 | driver_idx |= Lighting; |
| 779 | PS driver = gpuSpriteSpanDrivers[driver_idx]; |
| 780 | gpuDrawS(packet, driver, &w, &h); |
| 781 | cpu_cycles += gput_sprite(w, h); |
| 782 | } break; |
| 783 | |
| 784 | #ifdef TEST |
| 785 | case 0x80: // vid -> vid |
| 786 | gpuMoveImage(packet); |
| 787 | break; |
| 788 | |
| 789 | case 0xA0: // sys -> vid |
| 790 | { |
| 791 | u32 load_width = list[2] & 0xffff; |
| 792 | u32 load_height = list[2] >> 16; |
| 793 | u32 load_size = load_width * load_height; |
| 794 | |
| 795 | len += load_size / 2; |
| 796 | } break; |
| 797 | |
| 798 | case 0xC0: |
| 799 | break; |
| 800 | #else |
| 801 | case 0x1F: // irq? |
| 802 | case 0x80 ... 0x9F: // vid -> vid |
| 803 | case 0xA0 ... 0xBF: // sys -> vid |
| 804 | case 0xC0 ... 0xDF: // vid -> sys |
| 805 | // Handled by gpulib |
| 806 | goto breakloop; |
| 807 | #endif |
| 808 | case 0xE1 ... 0xE6: { // Draw settings |
| 809 | gpuGP0Cmd_0xEx(gpu_unai, le32_to_u32(gpu_unai.PacketBuffer.U4[0])); |
| 810 | } break; |
| 811 | } |
| 812 | } |
| 813 | |
| 814 | breakloop: |
| 815 | gpu.ex_regs[1] &= ~0x1ff; |
| 816 | gpu.ex_regs[1] |= gpu_unai.GPU_GP1 & 0x1ff; |
| 817 | |
| 818 | *cpu_cycles_out += cpu_cycles; |
| 819 | *last_cmd = cmd; |
| 820 | return list - list_start; |
| 821 | } |
| 822 | |
| 823 | void renderer_sync_ecmds(u32 *ecmds) |
| 824 | { |
| 825 | int dummy; |
| 826 | do_cmd_list(&ecmds[1], 6, &dummy, &dummy); |
| 827 | } |
| 828 | |
| 829 | void renderer_update_caches(int x, int y, int w, int h, int state_changed) |
| 830 | { |
| 831 | } |
| 832 | |
| 833 | void renderer_flush_queues(void) |
| 834 | { |
| 835 | } |
| 836 | |
| 837 | void renderer_set_interlace(int enable, int is_odd) |
| 838 | { |
| 839 | } |
| 840 | |
| 841 | #include "../../frontend/plugin_lib.h" |
| 842 | // Handle any gpulib settings applicable to gpu_unai: |
| 843 | void renderer_set_config(const struct rearmed_cbs *cbs) |
| 844 | { |
| 845 | gpu_unai.vram = (le16_t *)gpu.vram; |
| 846 | gpu_unai.config.ilace_force = cbs->gpu_unai.ilace_force; |
| 847 | gpu_unai.config.pixel_skip = cbs->gpu_unai.pixel_skip; |
| 848 | gpu_unai.config.lighting = cbs->gpu_unai.lighting; |
| 849 | gpu_unai.config.fast_lighting = cbs->gpu_unai.fast_lighting; |
| 850 | gpu_unai.config.blending = cbs->gpu_unai.blending; |
| 851 | gpu_unai.config.dithering = cbs->gpu_unai.dithering; |
| 852 | gpu_unai.config.scale_hires = cbs->gpu_unai.scale_hires; |
| 853 | |
| 854 | gpu.state.downscale_enable = gpu_unai.config.scale_hires; |
| 855 | if (gpu_unai.config.scale_hires) { |
| 856 | map_downscale_buffer(); |
| 857 | } else { |
| 858 | unmap_downscale_buffer(); |
| 859 | } |
| 860 | } |
| 861 | |
| 862 | void renderer_sync(void) |
| 863 | { |
| 864 | } |
| 865 | |
| 866 | void renderer_notify_update_lace(int updated) |
| 867 | { |
| 868 | } |
| 869 | |
| 870 | // vim:shiftwidth=2:expandtab |