0bfe8d59 |
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 | |
e223fa15 |
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 | |
0bfe8d59 |
45 | //#include "port.h" |
e223fa15 |
46 | #include "gpu_unai.h" |
0bfe8d59 |
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 | |
e223fa15 |
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 | |
0bfe8d59 |
223 | int renderer_init(void) |
224 | { |
e223fa15 |
225 | memset((void*)&gpu_unai, 0, sizeof(gpu_unai)); |
226 | gpu_unai.vram = (le16_t *)gpu.vram; |
0bfe8d59 |
227 | |
e223fa15 |
228 | // Original standalone gpu_unai initialized TextureWindow[]. I added the |
0bfe8d59 |
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 |
e223fa15 |
231 | gpu_unai.TextureWindow[0] = 0; |
232 | gpu_unai.TextureWindow[1] = 0; |
233 | gpu_unai.TextureWindow[2] = 255; |
234 | gpu_unai.TextureWindow[3] = 255; |
0bfe8d59 |
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 |
e223fa15 |
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); |
0bfe8d59 |
240 | |
241 | // Configuration options |
e223fa15 |
242 | gpu_unai.config = gpu_unai_config_ext; |
0bfe8d59 |
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. |
e223fa15 |
246 | //gpu_unai.config.enableAbbeyHack = gpu_unai_config_ext.abe_hack; |
247 | gpu_unai.ilace_mask = gpu_unai.config.ilace_force; |
0bfe8d59 |
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 | |
e223fa15 |
266 | if (gpu_unai.config.scale_hires) { |
267 | map_downscale_buffer(); |
268 | } |
269 | |
0bfe8d59 |
270 | return 0; |
271 | } |
272 | |
273 | void renderer_finish(void) |
274 | { |
e223fa15 |
275 | unmap_downscale_buffer(); |
0bfe8d59 |
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 | { |
e223fa15 |
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; |
0bfe8d59 |
290 | } |
291 | } else { |
e223fa15 |
292 | gpu_unai.blit_mask = 0; |
0bfe8d59 |
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) { |
e223fa15 |
300 | if (gpu_unai.config.ilace_force) { |
301 | gpu_unai.ilace_mask = 3; // Only need 1/4 of lines |
0bfe8d59 |
302 | } else { |
e223fa15 |
303 | gpu_unai.ilace_mask = 1; // Only need 1/2 of lines |
0bfe8d59 |
304 | } |
305 | } else { |
306 | // Vert resolution changed from 480 to lower one |
e223fa15 |
307 | gpu_unai.ilace_mask = gpu_unai.config.ilace_force; |
0bfe8d59 |
308 | } |
309 | } else { |
e223fa15 |
310 | gpu_unai.ilace_mask = 0; |
0bfe8d59 |
311 | } |
312 | |
313 | /* |
314 | printf("res change hres: %d vres: %d depth: %d ilace_mask: %d\n", |
e223fa15 |
315 | gpu.screen.hres, gpu.screen.vres, (gpu.status & PSX_GPU_STATUS_RGB24) ? 24 : 15, |
316 | gpu_unai.ilace_mask); |
0bfe8d59 |
317 | */ |
318 | } |
319 | |
2da2fc76 |
320 | void renderer_notify_scanout_change(int x, int y) |
0b4038f8 |
321 | { |
322 | } |
323 | |
0bfe8d59 |
324 | #ifdef USE_GPULIB |
325 | // Handles GP0 draw settings commands 0xE1...0xE6 |
e223fa15 |
326 | static void gpuGP0Cmd_0xEx(gpu_unai_t &gpu_unai, u32 cmd_word) |
0bfe8d59 |
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") |
e223fa15 |
334 | u32 cur_texpage = gpu_unai.GPU_GP1 & 0x7FF; |
0bfe8d59 |
335 | u32 new_texpage = cmd_word & 0x7FF; |
336 | if (cur_texpage != new_texpage) { |
e223fa15 |
337 | gpu_unai.GPU_GP1 = (gpu_unai.GPU_GP1 & ~0x7FF) | new_texpage; |
338 | gpuSetTexture(gpu_unai.GPU_GP1); |
0bfe8d59 |
339 | } |
340 | } break; |
341 | |
342 | case 2: { |
343 | // GP0(E2h) - Texture Window setting |
e223fa15 |
344 | if (cmd_word != gpu_unai.TextureWindowCur) { |
0bfe8d59 |
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 | }; |
e223fa15 |
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]; |
0bfe8d59 |
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 |
e223fa15 |
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); |
0bfe8d59 |
361 | |
e223fa15 |
362 | gpuSetTexture(gpu_unai.GPU_GP1); |
0bfe8d59 |
363 | } |
364 | } break; |
365 | |
366 | case 3: { |
367 | // GP0(E3h) - Set Drawing Area top left (X1,Y1) |
e223fa15 |
368 | gpu_unai.DrawingArea[0] = cmd_word & 0x3FF; |
369 | gpu_unai.DrawingArea[1] = (cmd_word >> 10) & 0x3FF; |
0bfe8d59 |
370 | } break; |
371 | |
372 | case 4: { |
373 | // GP0(E4h) - Set Drawing Area bottom right (X2,Y2) |
e223fa15 |
374 | gpu_unai.DrawingArea[2] = (cmd_word & 0x3FF) + 1; |
375 | gpu_unai.DrawingArea[3] = ((cmd_word >> 10) & 0x3FF) + 1; |
0bfe8d59 |
376 | } break; |
377 | |
378 | case 5: { |
379 | // GP0(E5h) - Set Drawing Offset (X,Y) |
e223fa15 |
380 | gpu_unai.DrawingOffset[0] = ((s32)cmd_word<<(32-11))>>(32-11); |
381 | gpu_unai.DrawingOffset[1] = ((s32)cmd_word<<(32-22))>>(32-11); |
0bfe8d59 |
382 | } break; |
383 | |
384 | case 6: { |
385 | // GP0(E6h) - Mask Bit Setting |
e223fa15 |
386 | gpu_unai.Masking = (cmd_word & 0x2) << 1; |
387 | gpu_unai.PixelMSB = (cmd_word & 0x1) << 8; |
0bfe8d59 |
388 | } break; |
389 | } |
390 | } |
391 | #endif |
392 | |
90ac6fed |
393 | #include "../gpulib/gpu_timing.h" |
0bfe8d59 |
394 | extern const unsigned char cmd_lengths[256]; |
395 | |
8412166f |
396 | int do_cmd_list(u32 *list_, int list_len, |
397 | int *cycles_sum_out, int *cycles_last, int *last_cmd) |
0bfe8d59 |
398 | { |
8412166f |
399 | int cpu_cycles_sum = 0, cpu_cycles = *cycles_last; |
0bfe8d59 |
400 | u32 cmd = 0, len, i; |
8412166f |
401 | le32_t *list = (le32_t *)list_; |
e223fa15 |
402 | le32_t *list_start = list; |
403 | le32_t *list_end = list + list_len; |
0bfe8d59 |
404 | |
405 | //TODO: set ilace_mask when resolution changes instead of every time, |
406 | // eliminate #ifdef below. |
e223fa15 |
407 | gpu_unai.ilace_mask = gpu_unai.config.ilace_force; |
0bfe8d59 |
408 | |
409 | #ifdef HAVE_PRE_ARMV7 /* XXX */ |
e223fa15 |
410 | gpu_unai.ilace_mask |= !!(gpu.status & PSX_GPU_STATUS_INTERLACE); |
0bfe8d59 |
411 | #endif |
e223fa15 |
412 | if (gpu_unai.config.scale_hires) { |
413 | gpu_unai.ilace_mask |= !!(gpu.status & PSX_GPU_STATUS_INTERLACE); |
0bfe8d59 |
414 | } |
415 | |
416 | for (; list < list_end; list += 1 + len) |
417 | { |
e223fa15 |
418 | cmd = le32_to_u32(*list) >> 24; |
0bfe8d59 |
419 | len = cmd_lengths[cmd]; |
420 | if (list + 1 + len > list_end) { |
421 | cmd = -1; |
422 | break; |
423 | } |
424 | |
425 | #define PRIM cmd |
e223fa15 |
426 | gpu_unai.PacketBuffer.U4[0] = list[0]; |
0bfe8d59 |
427 | for (i = 1; i <= len; i++) |
e223fa15 |
428 | gpu_unai.PacketBuffer.U4[i] = list[i]; |
0bfe8d59 |
429 | |
e223fa15 |
430 | PtrUnion packet = { .ptr = (void*)&gpu_unai.PacketBuffer }; |
0bfe8d59 |
431 | |
432 | switch (cmd) |
433 | { |
434 | case 0x02: |
435 | gpuClearImage(packet); |
8412166f |
436 | gput_sum(cpu_cycles_sum, cpu_cycles, |
437 | gput_fill(le16_to_s16(packet.U2[4]) & 0x3ff, le16_to_s16(packet.U2[5]) & 0x1ff)); |
0bfe8d59 |
438 | break; |
439 | |
440 | case 0x20: |
441 | case 0x21: |
442 | case 0x22: |
443 | case 0x23: { // Monochrome 3-pt poly |
444 | PP driver = gpuPolySpanDrivers[ |
e223fa15 |
445 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
446 | Blending_Mode | |
e223fa15 |
447 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB |
0bfe8d59 |
448 | ]; |
449 | gpuDrawPolyF(packet, driver, false); |
8412166f |
450 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base()); |
0bfe8d59 |
451 | } break; |
452 | |
453 | case 0x24: |
454 | case 0x25: |
455 | case 0x26: |
456 | case 0x27: { // Textured 3-pt poly |
e223fa15 |
457 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
458 | gpuSetTexture(le32_to_u32(gpu_unai.PacketBuffer.U4[4]) >> 16); |
0bfe8d59 |
459 | |
460 | u32 driver_idx = |
e223fa15 |
461 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
462 | Dithering | |
e223fa15 |
463 | Blending_Mode | gpu_unai.TEXT_MODE | |
464 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB; |
0bfe8d59 |
465 | |
466 | if (!FastLightingEnabled()) { |
467 | driver_idx |= Lighting; |
468 | } else { |
e223fa15 |
469 | if (!((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F))) |
0bfe8d59 |
470 | driver_idx |= Lighting; |
471 | } |
472 | |
473 | PP driver = gpuPolySpanDrivers[driver_idx]; |
474 | gpuDrawPolyFT(packet, driver, false); |
8412166f |
475 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_t()); |
0bfe8d59 |
476 | } break; |
477 | |
478 | case 0x28: |
479 | case 0x29: |
480 | case 0x2A: |
481 | case 0x2B: { // Monochrome 4-pt poly |
482 | PP driver = gpuPolySpanDrivers[ |
e223fa15 |
483 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
484 | Blending_Mode | |
e223fa15 |
485 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB |
0bfe8d59 |
486 | ]; |
487 | gpuDrawPolyF(packet, driver, true); // is_quad = true |
8412166f |
488 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base()); |
0bfe8d59 |
489 | } break; |
490 | |
491 | case 0x2C: |
492 | case 0x2D: |
493 | case 0x2E: |
494 | case 0x2F: { // Textured 4-pt poly |
e223fa15 |
495 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
496 | gpuSetTexture(le32_to_u32(gpu_unai.PacketBuffer.U4[4]) >> 16); |
0bfe8d59 |
497 | |
498 | u32 driver_idx = |
e223fa15 |
499 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
500 | Dithering | |
e223fa15 |
501 | Blending_Mode | gpu_unai.TEXT_MODE | |
502 | gpu_unai.Masking | Blending | gpu_unai.PixelMSB; |
0bfe8d59 |
503 | |
504 | if (!FastLightingEnabled()) { |
505 | driver_idx |= Lighting; |
506 | } else { |
e223fa15 |
507 | if (!((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F))) |
0bfe8d59 |
508 | driver_idx |= Lighting; |
509 | } |
510 | |
511 | PP driver = gpuPolySpanDrivers[driver_idx]; |
512 | gpuDrawPolyFT(packet, driver, true); // is_quad = true |
8412166f |
513 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_t()); |
0bfe8d59 |
514 | } break; |
515 | |
516 | case 0x30: |
517 | case 0x31: |
518 | case 0x32: |
519 | case 0x33: { // Gouraud-shaded 3-pt poly |
520 | //NOTE: The '129' here is CF_GOURAUD | CF_LIGHT, however |
521 | // this is an untextured poly, so CF_LIGHT (texture blend) |
522 | // shouldn't apply. Until the original array of template |
523 | // instantiation ptrs is fixed, we're stuck with this. (TODO) |
524 | PP driver = gpuPolySpanDrivers[ |
e223fa15 |
525 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
526 | Dithering | |
527 | Blending_Mode | |
e223fa15 |
528 | gpu_unai.Masking | Blending | 129 | gpu_unai.PixelMSB |
0bfe8d59 |
529 | ]; |
530 | gpuDrawPolyG(packet, driver, false); |
8412166f |
531 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_g()); |
0bfe8d59 |
532 | } break; |
533 | |
534 | case 0x34: |
535 | case 0x35: |
536 | case 0x36: |
537 | case 0x37: { // Gouraud-shaded, textured 3-pt poly |
e223fa15 |
538 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
539 | gpuSetTexture (le32_to_u32(gpu_unai.PacketBuffer.U4[5]) >> 16); |
0bfe8d59 |
540 | PP driver = gpuPolySpanDrivers[ |
e223fa15 |
541 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
542 | Dithering | |
e223fa15 |
543 | Blending_Mode | gpu_unai.TEXT_MODE | |
544 | gpu_unai.Masking | Blending | ((Lighting)?129:0) | gpu_unai.PixelMSB |
0bfe8d59 |
545 | ]; |
546 | gpuDrawPolyGT(packet, driver, false); |
8412166f |
547 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_poly_base_gt()); |
0bfe8d59 |
548 | } break; |
549 | |
550 | case 0x38: |
551 | case 0x39: |
552 | case 0x3A: |
553 | case 0x3B: { // Gouraud-shaded 4-pt poly |
554 | // See notes regarding '129' for 0x30..0x33 further above -senquack |
555 | PP driver = gpuPolySpanDrivers[ |
e223fa15 |
556 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
557 | Dithering | |
558 | Blending_Mode | |
e223fa15 |
559 | gpu_unai.Masking | Blending | 129 | gpu_unai.PixelMSB |
0bfe8d59 |
560 | ]; |
561 | gpuDrawPolyG(packet, driver, true); // is_quad = true |
8412166f |
562 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_g()); |
0bfe8d59 |
563 | } break; |
564 | |
565 | case 0x3C: |
566 | case 0x3D: |
567 | case 0x3E: |
568 | case 0x3F: { // Gouraud-shaded, textured 4-pt poly |
e223fa15 |
569 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
570 | gpuSetTexture (le32_to_u32(gpu_unai.PacketBuffer.U4[5]) >> 16); |
0bfe8d59 |
571 | PP driver = gpuPolySpanDrivers[ |
e223fa15 |
572 | (gpu_unai.blit_mask?1024:0) | |
0bfe8d59 |
573 | Dithering | |
e223fa15 |
574 | Blending_Mode | gpu_unai.TEXT_MODE | |
575 | gpu_unai.Masking | Blending | ((Lighting)?129:0) | gpu_unai.PixelMSB |
0bfe8d59 |
576 | ]; |
577 | gpuDrawPolyGT(packet, driver, true); // is_quad = true |
8412166f |
578 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_quad_base_gt()); |
0bfe8d59 |
579 | } break; |
580 | |
581 | case 0x40: |
582 | case 0x41: |
583 | case 0x42: |
584 | case 0x43: { // Monochrome line |
585 | // Shift index right by one, as untextured prims don't use lighting |
e223fa15 |
586 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
0bfe8d59 |
587 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
588 | gpuDrawLineF(packet, driver); |
8412166f |
589 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0)); |
0bfe8d59 |
590 | } break; |
591 | |
592 | case 0x48 ... 0x4F: { // Monochrome line strip |
593 | u32 num_vertexes = 1; |
e223fa15 |
594 | le32_t *list_position = &list[2]; |
0bfe8d59 |
595 | |
596 | // Shift index right by one, as untextured prims don't use lighting |
e223fa15 |
597 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
0bfe8d59 |
598 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
599 | gpuDrawLineF(packet, driver); |
600 | |
601 | while(1) |
602 | { |
e223fa15 |
603 | gpu_unai.PacketBuffer.U4[1] = gpu_unai.PacketBuffer.U4[2]; |
604 | gpu_unai.PacketBuffer.U4[2] = *list_position++; |
0bfe8d59 |
605 | gpuDrawLineF(packet, driver); |
8412166f |
606 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0)); |
0bfe8d59 |
607 | |
608 | num_vertexes++; |
609 | if(list_position >= list_end) { |
610 | cmd = -1; |
611 | goto breakloop; |
612 | } |
e223fa15 |
613 | if((le32_raw(*list_position) & HTOLE32(0xf000f000)) == HTOLE32(0x50005000)) |
0bfe8d59 |
614 | break; |
615 | } |
616 | |
617 | len += (num_vertexes - 2); |
618 | } break; |
619 | |
620 | case 0x50: |
621 | case 0x51: |
622 | case 0x52: |
623 | case 0x53: { // Gouraud-shaded line |
624 | // Shift index right by one, as untextured prims don't use lighting |
e223fa15 |
625 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
0bfe8d59 |
626 | // Index MSB selects Gouraud-shaded PixelSpanDriver: |
627 | driver_idx |= (1 << 5); |
628 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
629 | gpuDrawLineG(packet, driver); |
8412166f |
630 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0)); |
0bfe8d59 |
631 | } break; |
632 | |
633 | case 0x58 ... 0x5F: { // Gouraud-shaded line strip |
634 | u32 num_vertexes = 1; |
e223fa15 |
635 | le32_t *list_position = &list[2]; |
0bfe8d59 |
636 | |
637 | // Shift index right by one, as untextured prims don't use lighting |
e223fa15 |
638 | u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; |
0bfe8d59 |
639 | // Index MSB selects Gouraud-shaded PixelSpanDriver: |
640 | driver_idx |= (1 << 5); |
641 | PSD driver = gpuPixelSpanDrivers[driver_idx]; |
642 | gpuDrawLineG(packet, driver); |
643 | |
644 | while(1) |
645 | { |
e223fa15 |
646 | gpu_unai.PacketBuffer.U4[0] = gpu_unai.PacketBuffer.U4[2]; |
647 | gpu_unai.PacketBuffer.U4[1] = gpu_unai.PacketBuffer.U4[3]; |
648 | gpu_unai.PacketBuffer.U4[2] = *list_position++; |
649 | gpu_unai.PacketBuffer.U4[3] = *list_position++; |
0bfe8d59 |
650 | gpuDrawLineG(packet, driver); |
8412166f |
651 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_line(0)); |
0bfe8d59 |
652 | |
653 | num_vertexes++; |
654 | if(list_position >= list_end) { |
655 | cmd = -1; |
656 | goto breakloop; |
657 | } |
e223fa15 |
658 | if((le32_raw(*list_position) & HTOLE32(0xf000f000)) == HTOLE32(0x50005000)) |
0bfe8d59 |
659 | break; |
660 | } |
661 | |
662 | len += (num_vertexes - 2) * 2; |
663 | } break; |
664 | |
665 | case 0x60: |
666 | case 0x61: |
667 | case 0x62: |
668 | case 0x63: { // Monochrome rectangle (variable size) |
e223fa15 |
669 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
1cec4719 |
670 | s32 w = 0, h = 0; |
671 | gpuDrawT(packet, driver, &w, &h); |
8412166f |
672 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
673 | } break; |
674 | |
675 | case 0x64: |
676 | case 0x65: |
677 | case 0x66: |
678 | case 0x67: { // Textured rectangle (variable size) |
e223fa15 |
679 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
680 | u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); |
1cec4719 |
681 | s32 w = 0, h = 0; |
0bfe8d59 |
682 | |
683 | //senquack - Only color 808080h-878787h allows skipping lighting calculation: |
684 | // This fixes Silent Hill running animation on loading screens: |
685 | // (On PSX, color values 0x00-0x7F darken the source texture's color, |
686 | // 0x81-FF lighten textures (ultimately clamped to 0x1F), |
687 | // 0x80 leaves source texture color unchanged, HOWEVER, |
e223fa15 |
688 | // gpu_unai uses a simple lighting LUT whereby only the upper |
0bfe8d59 |
689 | // 5 bits of an 8-bit color are used, so 0x80-0x87 all behave as |
690 | // 0x80. |
691 | // |
692 | // NOTE: I've changed all textured sprite draw commands here and |
693 | // elsewhere to use proper behavior, but left poly commands |
694 | // alone, I don't want to slow rendering down too much. (TODO) |
e223fa15 |
695 | //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) |
0bfe8d59 |
696 | // Strip lower 3 bits of each color and determine if lighting should be used: |
e223fa15 |
697 | if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) |
0bfe8d59 |
698 | driver_idx |= Lighting; |
699 | PS driver = gpuSpriteSpanDrivers[driver_idx]; |
1cec4719 |
700 | gpuDrawS(packet, driver, &w, &h); |
8412166f |
701 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
702 | } break; |
703 | |
704 | case 0x68: |
705 | case 0x69: |
706 | case 0x6A: |
707 | case 0x6B: { // Monochrome rectangle (1x1 dot) |
e223fa15 |
708 | gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00010001); |
709 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
1cec4719 |
710 | s32 w = 0, h = 0; |
711 | gpuDrawT(packet, driver, &w, &h); |
8412166f |
712 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(1, 1)); |
0bfe8d59 |
713 | } break; |
714 | |
715 | case 0x70: |
716 | case 0x71: |
717 | case 0x72: |
718 | case 0x73: { // Monochrome rectangle (8x8) |
e223fa15 |
719 | gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00080008); |
720 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
1cec4719 |
721 | s32 w = 0, h = 0; |
722 | gpuDrawT(packet, driver, &w, &h); |
8412166f |
723 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
724 | } break; |
725 | |
726 | case 0x74: |
727 | case 0x75: |
728 | case 0x76: |
729 | case 0x77: { // Textured rectangle (8x8) |
e223fa15 |
730 | gpu_unai.PacketBuffer.U4[3] = u32_to_le32(0x00080008); |
731 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
732 | u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); |
1cec4719 |
733 | s32 w = 0, h = 0; |
0bfe8d59 |
734 | |
735 | //senquack - Only color 808080h-878787h allows skipping lighting calculation: |
e223fa15 |
736 | //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) |
0bfe8d59 |
737 | // Strip lower 3 bits of each color and determine if lighting should be used: |
e223fa15 |
738 | if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) |
0bfe8d59 |
739 | driver_idx |= Lighting; |
740 | PS driver = gpuSpriteSpanDrivers[driver_idx]; |
1cec4719 |
741 | gpuDrawS(packet, driver, &w, &h); |
8412166f |
742 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
743 | } break; |
744 | |
745 | case 0x78: |
746 | case 0x79: |
747 | case 0x7A: |
748 | case 0x7B: { // Monochrome rectangle (16x16) |
e223fa15 |
749 | gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00100010); |
750 | PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; |
1cec4719 |
751 | s32 w = 0, h = 0; |
752 | gpuDrawT(packet, driver, &w, &h); |
8412166f |
753 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
754 | } break; |
755 | |
756 | case 0x7C: |
757 | case 0x7D: |
758 | #ifdef __arm__ |
e223fa15 |
759 | if ((gpu_unai.GPU_GP1 & 0x180) == 0 && (gpu_unai.Masking | gpu_unai.PixelMSB) == 0) |
0bfe8d59 |
760 | { |
1cec4719 |
761 | s32 w = 0, h = 0; |
762 | gpuSetCLUT(le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
763 | gpuDrawS16(packet, &w, &h); |
8412166f |
764 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
765 | break; |
766 | } |
767 | // fallthrough |
768 | #endif |
769 | case 0x7E: |
770 | case 0x7F: { // Textured rectangle (16x16) |
e223fa15 |
771 | gpu_unai.PacketBuffer.U4[3] = u32_to_le32(0x00100010); |
772 | gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); |
773 | u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); |
1cec4719 |
774 | s32 w = 0, h = 0; |
0bfe8d59 |
775 | //senquack - Only color 808080h-878787h allows skipping lighting calculation: |
e223fa15 |
776 | //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) |
0bfe8d59 |
777 | // Strip lower 3 bits of each color and determine if lighting should be used: |
e223fa15 |
778 | if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) |
0bfe8d59 |
779 | driver_idx |= Lighting; |
780 | PS driver = gpuSpriteSpanDrivers[driver_idx]; |
1cec4719 |
781 | gpuDrawS(packet, driver, &w, &h); |
8412166f |
782 | gput_sum(cpu_cycles_sum, cpu_cycles, gput_sprite(w, h)); |
0bfe8d59 |
783 | } break; |
784 | |
72583812 |
785 | #ifdef TEST |
0bfe8d59 |
786 | case 0x80: // vid -> vid |
787 | gpuMoveImage(packet); |
788 | break; |
789 | |
0bfe8d59 |
790 | case 0xA0: // sys -> vid |
791 | { |
792 | u32 load_width = list[2] & 0xffff; |
793 | u32 load_height = list[2] >> 16; |
794 | u32 load_size = load_width * load_height; |
795 | |
796 | len += load_size / 2; |
797 | } break; |
798 | |
799 | case 0xC0: |
800 | break; |
801 | #else |
1cec4719 |
802 | case 0x1F: // irq? |
72583812 |
803 | case 0x80 ... 0x9F: // vid -> vid |
804 | case 0xA0 ... 0xBF: // sys -> vid |
805 | case 0xC0 ... 0xDF: // vid -> sys |
0bfe8d59 |
806 | // Handled by gpulib |
807 | goto breakloop; |
808 | #endif |
809 | case 0xE1 ... 0xE6: { // Draw settings |
e223fa15 |
810 | gpuGP0Cmd_0xEx(gpu_unai, le32_to_u32(gpu_unai.PacketBuffer.U4[0])); |
0bfe8d59 |
811 | } break; |
812 | } |
813 | } |
814 | |
815 | breakloop: |
816 | gpu.ex_regs[1] &= ~0x1ff; |
e223fa15 |
817 | gpu.ex_regs[1] |= gpu_unai.GPU_GP1 & 0x1ff; |
0bfe8d59 |
818 | |
8412166f |
819 | *cycles_sum_out += cpu_cycles_sum; |
820 | *cycles_last = cpu_cycles; |
0bfe8d59 |
821 | *last_cmd = cmd; |
822 | return list - list_start; |
823 | } |
824 | |
e223fa15 |
825 | void renderer_sync_ecmds(u32 *ecmds) |
0bfe8d59 |
826 | { |
827 | int dummy; |
8412166f |
828 | do_cmd_list(&ecmds[1], 6, &dummy, &dummy, &dummy); |
0bfe8d59 |
829 | } |
830 | |
0b4038f8 |
831 | void renderer_update_caches(int x, int y, int w, int h, int state_changed) |
0bfe8d59 |
832 | { |
833 | } |
834 | |
835 | void renderer_flush_queues(void) |
836 | { |
837 | } |
838 | |
839 | void renderer_set_interlace(int enable, int is_odd) |
840 | { |
841 | } |
842 | |
843 | #include "../../frontend/plugin_lib.h" |
e223fa15 |
844 | // Handle any gpulib settings applicable to gpu_unai: |
0bfe8d59 |
845 | void renderer_set_config(const struct rearmed_cbs *cbs) |
846 | { |
e223fa15 |
847 | gpu_unai.vram = (le16_t *)gpu.vram; |
848 | gpu_unai.config.ilace_force = cbs->gpu_unai.ilace_force; |
849 | gpu_unai.config.pixel_skip = cbs->gpu_unai.pixel_skip; |
850 | gpu_unai.config.lighting = cbs->gpu_unai.lighting; |
851 | gpu_unai.config.fast_lighting = cbs->gpu_unai.fast_lighting; |
852 | gpu_unai.config.blending = cbs->gpu_unai.blending; |
853 | gpu_unai.config.dithering = cbs->gpu_unai.dithering; |
854 | gpu_unai.config.scale_hires = cbs->gpu_unai.scale_hires; |
855 | |
856 | gpu.state.downscale_enable = gpu_unai.config.scale_hires; |
857 | if (gpu_unai.config.scale_hires) { |
858 | map_downscale_buffer(); |
859 | } else { |
860 | unmap_downscale_buffer(); |
861 | } |
862 | } |
863 | |
864 | void renderer_sync(void) |
865 | { |
866 | } |
867 | |
868 | void renderer_notify_update_lace(int updated) |
869 | { |
0bfe8d59 |
870 | } |
871 | |
872 | // vim:shiftwidth=2:expandtab |