| 1 | /* |
| 2 | * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com> |
| 3 | * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com> |
| 4 | * |
| 5 | * This program is free software; you can redistribute it and/or |
| 6 | * modify it under the terms of the GNU General Public License as |
| 7 | * published by the Free Software Foundation; either version 2 of |
| 8 | * the License, or (at your option) any later version. |
| 9 | * |
| 10 | * This program 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 | * General Public License for more details. |
| 14 | */ |
| 15 | |
| 16 | #define MAX_SPANS 512 |
| 17 | #define MAX_BLOCKS 64 |
| 18 | #define MAX_BLOCKS_PER_ROW 128 |
| 19 | |
| 20 | #define RENDER_STATE_MASK_EVALUATE 0x20 |
| 21 | #define RENDER_FLAGS_MODULATE_TEXELS 0x1 |
| 22 | #define RENDER_FLAGS_BLEND 0x2 |
| 23 | #define RENDER_INTERLACE_ENABLED 0x1 |
| 24 | |
| 25 | #include "psx_gpu_offsets.h" |
| 26 | |
| 27 | #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4) |
| 28 | |
| 29 | #define edge_data_left_x_offset 0 |
| 30 | #define edge_data_num_blocks_offset 2 |
| 31 | #define edge_data_right_mask_offset 4 |
| 32 | #define edge_data_y_offset 6 |
| 33 | |
| 34 | .syntax unified |
| 35 | .text |
| 36 | |
| 37 | #define psx_gpu r0 |
| 38 | #define v_a r1 |
| 39 | #define v_b r2 |
| 40 | #define v_c r3 |
| 41 | |
| 42 | #define x0 r4 |
| 43 | #define x1 r5 |
| 44 | #define x2 r6 |
| 45 | #define x0_x1 r5 |
| 46 | #define x1_x2 r6 |
| 47 | #define y0 r7 |
| 48 | #define y1 r8 |
| 49 | #define y2 r9 |
| 50 | #define y0_y1 r7 |
| 51 | #define y1_y2 r8 |
| 52 | #define b0 r9 |
| 53 | #define b1 r10 |
| 54 | #define b2 r11 |
| 55 | #define b0_b1 r10 |
| 56 | #define b1_b2 r11 |
| 57 | |
| 58 | |
| 59 | #define area_r_s r5 |
| 60 | |
| 61 | #define g_bx0 r2 |
| 62 | #define g_bx r3 |
| 63 | #define g_bx2 r4 |
| 64 | #define g_bx3 r5 |
| 65 | #define b_base r6 |
| 66 | #define g_by r8 |
| 67 | |
| 68 | #define gs_bx r7 |
| 69 | #define gs_by r10 |
| 70 | |
| 71 | #define ga_bx g_bx |
| 72 | #define ga_by g_by |
| 73 | |
| 74 | #define gw_bx_h g_bx |
| 75 | #define gw_by_h g_by |
| 76 | |
| 77 | #define gw_bx_l r11 |
| 78 | #define gw_by_l gw_bx_l |
| 79 | |
| 80 | #define store_a r0 |
| 81 | #define store_b r1 |
| 82 | #define store_inc r5 |
| 83 | |
| 84 | |
| 85 | #define v0 q0 |
| 86 | #define uvrgb0 d0 |
| 87 | #define x0_y0 d1 |
| 88 | |
| 89 | #define v1 q1 |
| 90 | #define uvrgb1 d2 |
| 91 | #define x1_y1 d3 |
| 92 | |
| 93 | #define v2 q2 |
| 94 | #define uvrgb2 d4 |
| 95 | #define x2_y2 d5 |
| 96 | |
| 97 | #define x0_ab q3 |
| 98 | #define uvrg_xxxx0 q3 |
| 99 | #define uvrg0 d6 |
| 100 | #define xxxx0 d7 |
| 101 | |
| 102 | #define x1_ab q4 |
| 103 | #define uvrg_xxxx1 q4 |
| 104 | #define uvrg1 d8 |
| 105 | #define xxxx1 d9 |
| 106 | |
| 107 | #define x2_ab q5 |
| 108 | #define uvrg_xxxx2 q5 |
| 109 | #define uvrg2 d10 |
| 110 | #define xxxx2 d11 |
| 111 | |
| 112 | #define y0_ab q6 |
| 113 | #define yyyy_uvrg0 q6 |
| 114 | #define yyyy0 d12 |
| 115 | #define uvrg0b d13 |
| 116 | |
| 117 | #define y1_ab q7 |
| 118 | #define yyyy_uvrg1 q7 |
| 119 | #define yyyy1 d14 |
| 120 | #define uvrg1b d15 |
| 121 | |
| 122 | #define y2_ab q8 |
| 123 | #define yyyy_uvrg2 q8 |
| 124 | #define yyyy2 d16 |
| 125 | #define uvrg2b d17 |
| 126 | |
| 127 | #define d0_ab q9 |
| 128 | #define d0_a d18 |
| 129 | #define d0_b d19 |
| 130 | |
| 131 | #define d1_ab q10 |
| 132 | #define d1_a d20 |
| 133 | #define d1_b d21 |
| 134 | |
| 135 | #define d2_ab q11 |
| 136 | #define d2_a d22 |
| 137 | #define d2_b d23 |
| 138 | |
| 139 | #define d3_ab q12 |
| 140 | #define d3_a d24 |
| 141 | #define d3_b d25 |
| 142 | |
| 143 | #define ga_uvrg_x q1 |
| 144 | #define ga_uvrg_y q4 |
| 145 | |
| 146 | #define dx x0_x1 |
| 147 | #define dy y0_y1 |
| 148 | #define db b0_b1 |
| 149 | |
| 150 | #define uvrg_base q11 |
| 151 | |
| 152 | #define gs_uvrg_x q5 |
| 153 | #define gs_uvrg_y q6 |
| 154 | |
| 155 | #define g_uvrg_x q1 |
| 156 | #define ga_uv_x d2 |
| 157 | #define g_uv_x d2 |
| 158 | #define ga_rg_x d3 |
| 159 | #define g_rg_x d3 |
| 160 | |
| 161 | #define g_uvrg_y q4 |
| 162 | #define ga_uv_y d8 |
| 163 | #define g_uv_y d8 |
| 164 | #define ga_rg_y d9 |
| 165 | #define g_rg_y d9 |
| 166 | |
| 167 | #define gw_uv_x q1 |
| 168 | #define gw_rg_x q2 |
| 169 | #define gw_uv_y q4 |
| 170 | #define gw_rg_y q3 |
| 171 | |
| 172 | #define w_mask q9 |
| 173 | #define w_mask_l d18 |
| 174 | |
| 175 | #define r_shift q10 |
| 176 | |
| 177 | #define uvrg_dx0 q0 |
| 178 | #define uvrg_dx0l d0 |
| 179 | #define uvrg_dx0h d1 |
| 180 | |
| 181 | #define uvrg_dx1 q1 |
| 182 | #define uvrg_dx1l d2 |
| 183 | #define uvrg_dx1h d3 |
| 184 | |
| 185 | #define uvrg_dx2 q2 |
| 186 | #define uvrg_dx2l d4 |
| 187 | #define uvrg_dx2h d5 |
| 188 | |
| 189 | #define uvrg_dx3 q3 |
| 190 | #define uvrg_dx3l d6 |
| 191 | #define uvrg_dx3h d7 |
| 192 | |
| 193 | #define uvrgb_phase q13 |
| 194 | |
| 195 | .align 4 |
| 196 | |
| 197 | #include "arm_features.h" |
| 198 | |
| 199 | #define function(name) FUNCTION(name): |
| 200 | |
| 201 | #ifndef TEXRELS_FORBIDDEN |
| 202 | |
| 203 | #define JT_OP_REL(table_label, index_reg, temp) |
| 204 | #define JT_OP(x...) x |
| 205 | #define JTE(start, target) target |
| 206 | |
| 207 | #else |
| 208 | |
| 209 | #define JT_OP_REL(table_label, index_reg, temp) \ |
| 210 | adr temp, table_label; \ |
| 211 | ldr temp, [temp, index_reg, lsl #2]; \ |
| 212 | add pc, pc, temp \ |
| 213 | |
| 214 | #define JT_OP(x...) |
| 215 | #define JTE(start, target) (target - start) |
| 216 | |
| 217 | #endif |
| 218 | |
| 219 | #ifdef __MACH__ |
| 220 | #define flush_render_block_buffer _flush_render_block_buffer |
| 221 | #define setup_sprite_untextured_simple _setup_sprite_untextured_simple |
| 222 | #define update_texture_8bpp_cache _update_texture_8bpp_cache |
| 223 | #endif |
| 224 | |
| 225 | @ r0: psx_gpu |
| 226 | @ r1: v_a |
| 227 | @ r2: v_b |
| 228 | @ r3: v_c |
| 229 | |
| 230 | function(compute_all_gradients) |
| 231 | // First compute the triangle area reciprocal and shift. The division will |
| 232 | // happen concurrently with much of the work which follows. |
| 233 | @ r12 = psx_gpu->triangle_area |
| 234 | ldr r12, [psx_gpu, #psx_gpu_triangle_area_offset] |
| 235 | stmdb sp!, { r4 - r11, lr } |
| 236 | |
| 237 | @ load exponent of 62 into upper half of double |
| 238 | movw r4, #0 |
| 239 | clz r14, r12 @ r14 = shift |
| 240 | |
| 241 | movt r4, #((62 + 1023) << 4) |
| 242 | mov r12, r12, lsl r14 @ r12 = triangle_area_normalized |
| 243 | |
| 244 | @ load area normalized into lower half of double |
| 245 | mov r5, r12, lsr #10 |
| 246 | vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n |
| 247 | |
| 248 | movt r4, #((1022 + 31) << 4) |
| 249 | mov r5, r12, lsl #20 |
| 250 | |
| 251 | add r4, r4, r12, lsr #11 |
| 252 | vmov.f64 d31, r5, r4 |
| 253 | |
| 254 | vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n |
| 255 | |
| 256 | // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) = |
| 257 | // ( d0 * d1 ) - ( d2 * d3 ) = |
| 258 | // ( m0 ) - ( m1 ) = gradient |
| 259 | |
| 260 | // This is split to do 12 elements at a time over three sets: a, b, and c. |
| 261 | // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so |
| 262 | // two of the slots are unused. |
| 263 | |
| 264 | // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as |
| 265 | // is g. |
| 266 | |
| 267 | // First type is: uvrg bxxx xxxx |
| 268 | // Second type is: yyyy ybyy uvrg |
| 269 | // Since x_a and y_c are the same the same variable is used for both. |
| 270 | |
| 271 | vld1.u32 { v0 }, [v_a, :128] @ v0 = { uvrg0, b0, x0, y0 } |
| 272 | ldrsh x0, [v_a, #8] @ load x0 |
| 273 | |
| 274 | vld1.u32 { v1 }, [v_b, :128] @ v1 = { uvrg1, b1, x1, y1} |
| 275 | ldrh x1, [v_b, #8] @ load x1 |
| 276 | |
| 277 | vld1.u32 { v2 }, [v_c, :128] @ v2 = { uvrg2, b2, x2, y2 } |
| 278 | ldrh x2, [v_c, #8] @ load x2 |
| 279 | |
| 280 | vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- } |
| 281 | ldrh y0, [v_a, #10] @ load y0 |
| 282 | |
| 283 | vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- } |
| 284 | ldrh y1, [v_b, #10] @ load y1 |
| 285 | |
| 286 | vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- } |
| 287 | ldrh y2, [v_c, #10] @ load y2 |
| 288 | |
| 289 | vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 } |
| 290 | vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 } |
| 291 | |
| 292 | orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 } |
| 293 | pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 } |
| 294 | |
| 295 | vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 } |
| 296 | vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 } |
| 297 | |
| 298 | vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 } |
| 299 | vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 } |
| 300 | |
| 301 | ldrb b2, [v_c, #4] @ load b2 |
| 302 | orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 } |
| 303 | |
| 304 | ldrb b1, [v_b, #4] @ load b1 |
| 305 | orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 } |
| 306 | |
| 307 | vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 } |
| 308 | vsub.s16 d0_ab, x1_ab, x0_ab |
| 309 | |
| 310 | ldrb b0, [v_a, #4] @ load b0 |
| 311 | orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 } |
| 312 | |
| 313 | vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 } |
| 314 | vsub.s16 d2_ab, x2_ab, x1_ab |
| 315 | |
| 316 | vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 } |
| 317 | vsub.s16 d1_ab, y2_ab, y1_ab |
| 318 | |
| 319 | orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 } |
| 320 | ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 } |
| 321 | |
| 322 | ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 } |
| 323 | ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 } |
| 324 | |
| 325 | vsub.s16 d3_ab, y1_ab, y0_ab |
| 326 | smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) - |
| 327 | @ ((x2 - X1) * (b1 - b0)) |
| 328 | vmull.s16 ga_uvrg_x, d0_a, d1_a |
| 329 | smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) - |
| 330 | @ ((b2 - b1) * (y1 - y0)) |
| 331 | vmlsl.s16 ga_uvrg_x, d2_a, d3_a |
| 332 | movs gs_bx, ga_bx, asr #31 |
| 333 | |
| 334 | vmull.s16 ga_uvrg_y, d0_b, d1_b |
| 335 | rsbmi ga_bx, ga_bx, #0 |
| 336 | |
| 337 | @ r12 = psx_gpu->uvrgb_phase |
| 338 | ldr r12, [psx_gpu, #psx_gpu_uvrgb_phase_offset] |
| 339 | |
| 340 | vmlsl.s16 ga_uvrg_y, d2_b, d3_b |
| 341 | movs gs_by, ga_by, asr #31 |
| 342 | |
| 343 | vshr.u64 d0, d30, #22 |
| 344 | add b_base, r12, b0, lsl #16 |
| 345 | |
| 346 | vdup.u32 uvrgb_phase, r12 |
| 347 | |
| 348 | rsbmi ga_by, ga_by, #0 |
| 349 | vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0 |
| 350 | |
| 351 | @ r12 = psx_gpu->triangle_winding_offset |
| 352 | ldrb r12, [psx_gpu, #psx_gpu_triangle_winding_offset] |
| 353 | vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0 |
| 354 | |
| 355 | rsb r12, r12, #0 @ r12 = -(triangle->winding) |
| 356 | |
| 357 | vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w } |
| 358 | sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS) |
| 359 | |
| 360 | vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16 |
| 361 | vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift } |
| 362 | |
| 363 | vadd.u32 uvrg_base, uvrgb_phase |
| 364 | vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x) |
| 365 | |
| 366 | vmov area_r_s, s0 @ area_r_s = triangle_reciprocal |
| 367 | vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y) |
| 368 | |
| 369 | vmull.u32 gw_rg_x, ga_rg_x, d0[0] |
| 370 | vmull.u32 gw_uv_x, ga_uv_x, d0[0] |
| 371 | vmull.u32 gw_rg_y, ga_rg_y, d0[0] |
| 372 | vmull.u32 gw_uv_y, ga_uv_y, d0[0] |
| 373 | |
| 374 | vshl.u64 gw_rg_x, gw_rg_x, r_shift |
| 375 | vshl.u64 gw_uv_x, gw_uv_x, r_shift |
| 376 | vshl.u64 gw_rg_y, gw_rg_y, r_shift |
| 377 | vshl.u64 gw_uv_y, gw_uv_y, r_shift |
| 378 | |
| 379 | veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask |
| 380 | vmovn.u64 g_uv_x, gw_uv_x |
| 381 | |
| 382 | veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask |
| 383 | vmovn.u64 g_rg_x, gw_rg_x |
| 384 | |
| 385 | veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x |
| 386 | vmovn.u64 g_uv_y, gw_uv_y |
| 387 | |
| 388 | vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x |
| 389 | vmovn.u64 g_rg_y, gw_rg_y |
| 390 | |
| 391 | veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y |
| 392 | mov ga_bx, ga_bx, lsl #13 |
| 393 | |
| 394 | vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y |
| 395 | mov ga_by, ga_by, lsl #13 |
| 396 | |
| 397 | vdup.u32 x0_y0, x0 |
| 398 | umull gw_bx_l, gw_bx_h, ga_bx, area_r_s |
| 399 | |
| 400 | vshl.u32 g_uvrg_x, g_uvrg_x, #4 |
| 401 | vshl.u32 g_uvrg_y, g_uvrg_y, #4 |
| 402 | |
| 403 | umull gw_by_l, gw_by_h, ga_by, area_r_s |
| 404 | vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0] |
| 405 | |
| 406 | eor gs_bx, gs_bx, r12 |
| 407 | vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1 |
| 408 | |
| 409 | veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0 |
| 410 | eor gs_by, gs_by, r12 |
| 411 | |
| 412 | rsb r11, r14, #0 @ r11 = negative shift for scalar lsr |
| 413 | add store_a, psx_gpu, #psx_gpu_uvrg_offset |
| 414 | |
| 415 | sub r11, r11, #(32 - 13) |
| 416 | |
| 417 | add store_b, store_a, #16 |
| 418 | mov store_inc, #32 |
| 419 | |
| 420 | vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1 |
| 421 | vst1.u32 { uvrg_base }, [store_a, :128], store_inc |
| 422 | |
| 423 | vst1.u32 { uvrg_dx1 }, [store_b, :128], store_inc |
| 424 | mov g_bx, gw_bx_h, lsr r11 |
| 425 | |
| 426 | vst1.u32 { g_uvrg_y }, [store_a, :128], store_inc |
| 427 | mov g_by, gw_by_h, lsr r11 |
| 428 | |
| 429 | vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \ |
| 430 | [store_b, :128], store_inc |
| 431 | eor g_bx, g_bx, gs_bx |
| 432 | |
| 433 | vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \ |
| 434 | [store_b, :128], store_inc |
| 435 | sub g_bx, g_bx, gs_bx |
| 436 | |
| 437 | lsl g_bx, g_bx, #4 |
| 438 | eor g_by, g_by, gs_by |
| 439 | |
| 440 | mls b_base, g_bx, x0, b_base |
| 441 | sub g_by, g_by, gs_by |
| 442 | |
| 443 | lsl g_by, g_by, #4 |
| 444 | mov g_bx0, #0 |
| 445 | |
| 446 | add g_bx2, g_bx, g_bx |
| 447 | add g_bx3, g_bx, g_bx2 |
| 448 | |
| 449 | stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by } |
| 450 | |
| 451 | ldmia sp!, { r4 - r11, pc } |
| 452 | |
| 453 | |
| 454 | #define psx_gpu r0 |
| 455 | #define v_a r1 |
| 456 | #define v_b r2 |
| 457 | #define v_c r3 |
| 458 | |
| 459 | #define temp r14 |
| 460 | |
| 461 | #define x_a r4 |
| 462 | #define x_b r5 |
| 463 | #define x_c r6 |
| 464 | #define y_a r1 |
| 465 | #define y_b r2 |
| 466 | #define y_c r3 |
| 467 | |
| 468 | #define height_minor_a r7 |
| 469 | #define height_minor_b r8 |
| 470 | #define height_major r9 |
| 471 | #define height r9 |
| 472 | |
| 473 | #define reciprocal_table_ptr r10 |
| 474 | |
| 475 | #define edge_alt_low r4 |
| 476 | #define edge_alt_high r5 |
| 477 | #define edge_dx_dy_alt r6 |
| 478 | #define edge_shift_alt r10 |
| 479 | |
| 480 | #define edge_dx_dy_alt_low r4 |
| 481 | #define edge_dx_dy_alt_high r5 |
| 482 | |
| 483 | #define span_edge_data r4 |
| 484 | #define span_uvrg_offset r5 |
| 485 | #define span_b_offset r6 |
| 486 | |
| 487 | #define clip r14 |
| 488 | |
| 489 | #define b r11 |
| 490 | #define b_dy r12 |
| 491 | |
| 492 | |
| 493 | #define alternate_x q0 |
| 494 | #define alternate_dx_dy q1 |
| 495 | #define alternate_x_32 q2 |
| 496 | |
| 497 | #define alternate_x_low d0 |
| 498 | #define alternate_x_high d1 |
| 499 | #define alternate_dx_dy_low d2 |
| 500 | #define alternate_dx_dy_high d3 |
| 501 | #define alternate_x_32_low d4 |
| 502 | #define alternate_x_32_high d5 |
| 503 | |
| 504 | #define left_x q3 |
| 505 | #define right_x q4 |
| 506 | #define left_dx_dy q5 |
| 507 | #define right_dx_dy q6 |
| 508 | #define left_edge q7 |
| 509 | #define right_edge q8 |
| 510 | |
| 511 | #define left_x_low d6 |
| 512 | #define left_x_high d7 |
| 513 | #define right_x_low d8 |
| 514 | #define right_x_high d9 |
| 515 | #define left_dx_dy_low d10 |
| 516 | #define left_dx_dy_high d11 |
| 517 | #define right_dx_dy_low d12 |
| 518 | #define right_dx_dy_high d13 |
| 519 | #define left_edge_low d14 |
| 520 | #define left_edge_high d15 |
| 521 | #define right_edge_low d16 |
| 522 | #define right_edge_high d17 |
| 523 | |
| 524 | #define y_mid_point d18 |
| 525 | #define c_0x0004 d19 |
| 526 | |
| 527 | #define left_right_x_16 q11 |
| 528 | #define span_shifts_y q12 |
| 529 | #define c_0x0001 q13 |
| 530 | |
| 531 | #define span_shifts d24 |
| 532 | #define y_x4 d25 |
| 533 | #define c_0xFFFE d26 |
| 534 | #define c_0x0007 d27 |
| 535 | |
| 536 | #define left_right_x_16_low d22 |
| 537 | #define left_right_x_16_high d23 |
| 538 | |
| 539 | #define uvrg q14 |
| 540 | #define uvrg_dy q15 |
| 541 | |
| 542 | #define alternate_x_16 d4 |
| 543 | |
| 544 | #define v_clip q3 |
| 545 | #define v_clip_low d6 |
| 546 | |
| 547 | #define right_x_32 q10 |
| 548 | #define left_x_32 q11 |
| 549 | #define alternate_select d24 |
| 550 | |
| 551 | #define right_x_32_low d20 |
| 552 | #define right_x_32_high d21 |
| 553 | #define left_x_32_low d22 |
| 554 | #define left_x_32_high d23 |
| 555 | |
| 556 | #define edges_xy q0 |
| 557 | #define edges_dx_dy d2 |
| 558 | #define edge_shifts d3 |
| 559 | #define edge_shifts_64 q2 |
| 560 | |
| 561 | #define edges_xy_left d0 |
| 562 | #define edges_xy_right d1 |
| 563 | |
| 564 | #define height_reciprocals d6 |
| 565 | #define heights d7 |
| 566 | |
| 567 | #define widths d8 |
| 568 | #define c_0x01 d9 |
| 569 | #define x_starts d10 |
| 570 | #define x_ends d11 |
| 571 | |
| 572 | #define heights_b d12 |
| 573 | #define edges_dx_dy_64 q10 |
| 574 | |
| 575 | #define edges_dx_dy_64_left d20 |
| 576 | #define edges_dx_dy_64_right d21 |
| 577 | |
| 578 | |
| 579 | #define setup_spans_prologue() \ |
| 580 | stmdb sp!, { r4 - r11, lr }; \ |
| 581 | \ |
| 582 | ldrsh x_a, [v_a, #8]; \ |
| 583 | ldrsh x_b, [v_b, #8]; \ |
| 584 | ldrsh x_c, [v_c, #8]; \ |
| 585 | ldrsh y_a, [v_a, #10]; \ |
| 586 | ldrsh y_b, [v_b, #10]; \ |
| 587 | ldrsh y_c, [v_c, #10]; \ |
| 588 | \ |
| 589 | add temp, psx_gpu, #psx_gpu_uvrg_offset; \ |
| 590 | vld1.32 { uvrg }, [temp]; \ |
| 591 | add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \ |
| 592 | vld1.32 { uvrg_dy }, [temp]; \ |
| 593 | ldr reciprocal_table_ptr, [psx_gpu, #psx_gpu_reciprocal_table_ptr_offset]; \ |
| 594 | \ |
| 595 | vmov.u32 c_0x01, #0x01 \ |
| 596 | |
| 597 | #define setup_spans_load_b() \ |
| 598 | ldr b, [psx_gpu, #psx_gpu_b_offset]; \ |
| 599 | ldr b_dy, [psx_gpu, #psx_gpu_b_dy_offset] \ |
| 600 | |
| 601 | #define setup_spans_prologue_b() \ |
| 602 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 603 | add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \ |
| 604 | \ |
| 605 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 606 | vmov.u16 c_0x0004, #0x0004; \ |
| 607 | \ |
| 608 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 609 | vmov.u16 c_0x0001, #0x0001; \ |
| 610 | \ |
| 611 | vld1.u16 { left_edge_low[], left_edge_high[] }, [temp]; \ |
| 612 | add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \ |
| 613 | \ |
| 614 | vld1.u16 { right_edge_low[], right_edge_high[] }, [temp]; \ |
| 615 | vadd.u16 right_edge, right_edge, c_0x0001; \ |
| 616 | \ |
| 617 | vmov.u16 c_0x0007, #0x0007; \ |
| 618 | vmvn.u16 c_0xFFFE, #0x0001 \ |
| 619 | |
| 620 | |
| 621 | #define compute_edge_delta_x2() \ |
| 622 | ldr temp, [reciprocal_table_ptr, height, lsl #2]; \ |
| 623 | \ |
| 624 | vdup.u32 heights, height; \ |
| 625 | vsub.u32 widths, x_ends, x_starts; \ |
| 626 | \ |
| 627 | vdup.u32 edge_shifts, temp; \ |
| 628 | vsub.u32 heights_b, heights, c_0x01; \ |
| 629 | vshr.u32 height_reciprocals, edge_shifts, #10; \ |
| 630 | \ |
| 631 | vmla.s32 heights_b, x_starts, heights; \ |
| 632 | vbic.u16 edge_shifts, #0xE0; \ |
| 633 | vmul.s32 edges_dx_dy, widths, height_reciprocals; \ |
| 634 | vmull.s32 edges_xy, heights_b, height_reciprocals \ |
| 635 | |
| 636 | #define width_alt r6 |
| 637 | #define height_reciprocal_alt r11 |
| 638 | #define height_b_alt r12 |
| 639 | |
| 640 | #define compute_edge_delta_x3(start_c, height_a, height_b) \ |
| 641 | vmov heights, height_a, height_b; \ |
| 642 | ldr temp, [reciprocal_table_ptr, height_a, lsl #2]; \ |
| 643 | vmov.u32 edge_shifts[0], temp; \ |
| 644 | ldr temp, [reciprocal_table_ptr, height_b, lsl #2]; \ |
| 645 | vmov.u32 edge_shifts[1], temp; \ |
| 646 | ldr edge_shift_alt, [reciprocal_table_ptr, height_minor_b, lsl #2]; \ |
| 647 | \ |
| 648 | vsub.u32 widths, x_ends, x_starts; \ |
| 649 | sub width_alt, x_c, start_c; \ |
| 650 | \ |
| 651 | vsub.u32 heights_b, heights, c_0x01; \ |
| 652 | sub height_b_alt, height_minor_b, #1; \ |
| 653 | \ |
| 654 | vshr.u32 height_reciprocals, edge_shifts, #10; \ |
| 655 | lsr height_reciprocal_alt, edge_shift_alt, #10; \ |
| 656 | \ |
| 657 | vmla.s32 heights_b, x_starts, heights; \ |
| 658 | mla height_b_alt, height_minor_b, start_c, height_b_alt; \ |
| 659 | \ |
| 660 | vbic.u16 edge_shifts, #0xE0; \ |
| 661 | and edge_shift_alt, edge_shift_alt, #0x1F; \ |
| 662 | \ |
| 663 | vmul.s32 edges_dx_dy, widths, height_reciprocals; \ |
| 664 | mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \ |
| 665 | \ |
| 666 | vmull.s32 edges_xy, heights_b, height_reciprocals; \ |
| 667 | smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \ |
| 668 | |
| 669 | |
| 670 | #define setup_spans_adjust_y_up() \ |
| 671 | vsub.u32 y_x4, y_x4, c_0x0004 \ |
| 672 | |
| 673 | #define setup_spans_adjust_y_down() \ |
| 674 | vadd.u32 y_x4, y_x4, c_0x0004 \ |
| 675 | |
| 676 | #define setup_spans_adjust_interpolants_up() \ |
| 677 | vsub.u32 uvrg, uvrg, uvrg_dy; \ |
| 678 | sub b, b, b_dy \ |
| 679 | |
| 680 | #define setup_spans_adjust_interpolants_down() \ |
| 681 | vadd.u32 uvrg, uvrg, uvrg_dy; \ |
| 682 | add b, b, b_dy \ |
| 683 | |
| 684 | |
| 685 | #define setup_spans_clip_interpolants_increment() \ |
| 686 | mla b, b_dy, clip, b; \ |
| 687 | vmla.s32 uvrg, uvrg_dy, v_clip \ |
| 688 | |
| 689 | #define setup_spans_clip_interpolants_decrement() \ |
| 690 | mls b, b_dy, clip, b; \ |
| 691 | vmls.s32 uvrg, uvrg_dy, v_clip \ |
| 692 | |
| 693 | #define setup_spans_clip_alternate_yes() \ |
| 694 | smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \ |
| 695 | |
| 696 | #define setup_spans_clip_alternate_no() \ |
| 697 | |
| 698 | #define setup_spans_clip(direction, alternate_active) \ |
| 699 | vdup.u32 v_clip, clip; \ |
| 700 | setup_spans_clip_alternate_##alternate_active(); \ |
| 701 | setup_spans_clip_interpolants_##direction(); \ |
| 702 | vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \ |
| 703 | |
| 704 | |
| 705 | #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \ |
| 706 | vmovl.s32 edge_shifts_64, edge_shifts; \ |
| 707 | vmovl.s32 edges_dx_dy_64, edges_dx_dy; \ |
| 708 | \ |
| 709 | vshl.s64 edges_xy, edges_xy, edge_shifts_64; \ |
| 710 | vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \ |
| 711 | \ |
| 712 | vmov left_x_low, edges_xy_##left_index; \ |
| 713 | vmov right_x_low, edges_xy_##right_index; \ |
| 714 | \ |
| 715 | vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \ |
| 716 | vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \ |
| 717 | vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \ |
| 718 | vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \ |
| 719 | \ |
| 720 | vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \ |
| 721 | vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \ |
| 722 | \ |
| 723 | vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \ |
| 724 | vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \ |
| 725 | |
| 726 | |
| 727 | #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \ |
| 728 | setup_spans_adjust_edges_alternate_no(left_index, right_index); \ |
| 729 | \ |
| 730 | vdup.u16 y_mid_point, y_b; \ |
| 731 | rsb temp, edge_shift_alt, #32; \ |
| 732 | \ |
| 733 | lsl edge_alt_high, edge_alt_high, edge_shift_alt; \ |
| 734 | orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \ |
| 735 | lsl edge_alt_low, edge_alt_low, edge_shift_alt; \ |
| 736 | vmov alternate_x_low, edge_alt_low, edge_alt_high; \ |
| 737 | \ |
| 738 | asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \ |
| 739 | lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \ |
| 740 | vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \ |
| 741 | vmov alternate_dx_dy_high, alternate_dx_dy_low; \ |
| 742 | \ |
| 743 | vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \ |
| 744 | vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \ |
| 745 | |
| 746 | |
| 747 | #define setup_spans_y_select_up() \ |
| 748 | vclt.s16 alternate_select, y_x4, y_mid_point \ |
| 749 | |
| 750 | #define setup_spans_y_select_down() \ |
| 751 | vcgt.s16 alternate_select, y_x4, y_mid_point \ |
| 752 | |
| 753 | |
| 754 | #define setup_spans_alternate_select_left() \ |
| 755 | vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \ |
| 756 | |
| 757 | #define setup_spans_alternate_select_right() \ |
| 758 | vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \ |
| 759 | |
| 760 | |
| 761 | #define setup_spans_set_x4_alternate_yes(alternate, direction) \ |
| 762 | vshrn.s64 alternate_x_32_low, alternate_x, #32; \ |
| 763 | vshrn.s64 left_x_32_low, left_x, #32; \ |
| 764 | vshrn.s64 right_x_32_low, right_x, #32; \ |
| 765 | \ |
| 766 | vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \ |
| 767 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 768 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 769 | \ |
| 770 | vshrn.s64 alternate_x_32_high, alternate_x, #32; \ |
| 771 | vshrn.s64 left_x_32_high, left_x, #32; \ |
| 772 | vshrn.s64 right_x_32_high, right_x, #32; \ |
| 773 | \ |
| 774 | vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \ |
| 775 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 776 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 777 | \ |
| 778 | vmovn.u32 alternate_x_16, alternate_x_32; \ |
| 779 | setup_spans_y_select_##direction(); \ |
| 780 | vmovn.u32 left_right_x_16_low, left_x_32; \ |
| 781 | \ |
| 782 | vmovn.u32 left_right_x_16_high, right_x_32; \ |
| 783 | setup_spans_alternate_select_##alternate(); \ |
| 784 | \ |
| 785 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 786 | str b, [span_b_offset], #4; \ |
| 787 | setup_spans_adjust_interpolants_##direction(); \ |
| 788 | \ |
| 789 | vmax.s16 left_right_x_16, left_right_x_16, left_edge; \ |
| 790 | \ |
| 791 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 792 | str b, [span_b_offset], #4; \ |
| 793 | setup_spans_adjust_interpolants_##direction(); \ |
| 794 | \ |
| 795 | vmin.s16 left_right_x_16, left_right_x_16, right_edge; \ |
| 796 | \ |
| 797 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 798 | str b, [span_b_offset], #4; \ |
| 799 | setup_spans_adjust_interpolants_##direction(); \ |
| 800 | \ |
| 801 | vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \ |
| 802 | vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \ |
| 803 | vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \ |
| 804 | \ |
| 805 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 806 | str b, [span_b_offset], #4; \ |
| 807 | setup_spans_adjust_interpolants_##direction(); \ |
| 808 | \ |
| 809 | vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \ |
| 810 | vshl.u16 span_shifts, c_0xFFFE, span_shifts; \ |
| 811 | \ |
| 812 | vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \ |
| 813 | \ |
| 814 | setup_spans_adjust_y_##direction() \ |
| 815 | |
| 816 | |
| 817 | #define setup_spans_set_x4_alternate_no(alternate, direction) \ |
| 818 | vshrn.s64 left_x_32_low, left_x, #32; \ |
| 819 | vshrn.s64 right_x_32_low, right_x, #32; \ |
| 820 | \ |
| 821 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 822 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 823 | \ |
| 824 | vshrn.s64 left_x_32_high, left_x, #32; \ |
| 825 | vshrn.s64 right_x_32_high, right_x, #32; \ |
| 826 | \ |
| 827 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 828 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 829 | \ |
| 830 | vmovn.u32 left_right_x_16_low, left_x_32; \ |
| 831 | vmovn.u32 left_right_x_16_high, right_x_32; \ |
| 832 | \ |
| 833 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 834 | str b, [span_b_offset], #4; \ |
| 835 | setup_spans_adjust_interpolants_##direction(); \ |
| 836 | \ |
| 837 | vmax.s16 left_right_x_16, left_right_x_16, left_edge; \ |
| 838 | \ |
| 839 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 840 | str b, [span_b_offset], #4; \ |
| 841 | setup_spans_adjust_interpolants_##direction(); \ |
| 842 | \ |
| 843 | vmin.s16 left_right_x_16, left_right_x_16, right_edge; \ |
| 844 | \ |
| 845 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 846 | str b, [span_b_offset], #4; \ |
| 847 | setup_spans_adjust_interpolants_##direction(); \ |
| 848 | \ |
| 849 | vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \ |
| 850 | vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \ |
| 851 | vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \ |
| 852 | \ |
| 853 | vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \ |
| 854 | str b, [span_b_offset], #4; \ |
| 855 | setup_spans_adjust_interpolants_##direction(); \ |
| 856 | \ |
| 857 | vshl.u16 span_shifts, c_0xFFFE, span_shifts; \ |
| 858 | vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \ |
| 859 | \ |
| 860 | vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \ |
| 861 | \ |
| 862 | setup_spans_adjust_y_##direction() \ |
| 863 | |
| 864 | |
| 865 | #define edge_adjust_low r11 |
| 866 | #define edge_adjust_high r12 |
| 867 | |
| 868 | #define setup_spans_alternate_adjust_yes() \ |
| 869 | smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \ |
| 870 | subs edge_alt_low, edge_alt_low, edge_adjust_low; \ |
| 871 | sbc edge_alt_high, edge_alt_high, edge_adjust_high \ |
| 872 | |
| 873 | #define setup_spans_alternate_adjust_no() \ |
| 874 | |
| 875 | |
| 876 | #define setup_spans_down(left_index, right_index, alternate, alternate_active) \ |
| 877 | setup_spans_alternate_adjust_##alternate_active(); \ |
| 878 | setup_spans_load_b(); \ |
| 879 | \ |
| 880 | ldrsh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \ |
| 881 | subs y_c, y_c, temp; \ |
| 882 | subgt height, height, y_c; \ |
| 883 | addgt height, height, #1; \ |
| 884 | \ |
| 885 | ldrsh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \ |
| 886 | subs clip, temp, y_a; \ |
| 887 | ble 0f; \ |
| 888 | \ |
| 889 | sub height, height, clip; \ |
| 890 | add y_a, y_a, clip; \ |
| 891 | setup_spans_clip(increment, alternate_active); \ |
| 892 | \ |
| 893 | 0: \ |
| 894 | cmp height, #0; \ |
| 895 | ble 1f; \ |
| 896 | \ |
| 897 | orr temp, y_a, y_a, lsl #16; \ |
| 898 | add temp, temp, #(1 << 16); \ |
| 899 | add y_a, temp, #2; \ |
| 900 | add y_a, y_a, #(2 << 16); \ |
| 901 | vmov y_x4, temp, y_a; \ |
| 902 | \ |
| 903 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ |
| 904 | right_index); \ |
| 905 | setup_spans_prologue_b(); \ |
| 906 | \ |
| 907 | strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \ |
| 908 | \ |
| 909 | 2: \ |
| 910 | setup_spans_set_x4_alternate_##alternate_active(alternate, down); \ |
| 911 | subs height, height, #4; \ |
| 912 | bhi 2b; \ |
| 913 | \ |
| 914 | 1: \ |
| 915 | |
| 916 | |
| 917 | #define setup_spans_alternate_pre_increment_yes() \ |
| 918 | adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \ |
| 919 | adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \ |
| 920 | |
| 921 | #define setup_spans_alternate_pre_increment_no() \ |
| 922 | |
| 923 | |
| 924 | #define setup_spans_up_decrement_yes() \ |
| 925 | suble height, height, #1 \ |
| 926 | |
| 927 | #define setup_spans_up_decrement_no() \ |
| 928 | |
| 929 | |
| 930 | #define setup_spans_up(left_index, right_index, alternate, alternate_active) \ |
| 931 | setup_spans_alternate_adjust_##alternate_active(); \ |
| 932 | setup_spans_load_b(); \ |
| 933 | sub y_a, y_a, #1; \ |
| 934 | \ |
| 935 | ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \ |
| 936 | subs temp, temp, y_c; \ |
| 937 | subgt height, height, temp; \ |
| 938 | setup_spans_up_decrement_##alternate_active(); \ |
| 939 | \ |
| 940 | ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \ |
| 941 | subs clip, y_a, temp; \ |
| 942 | ble 0f; \ |
| 943 | \ |
| 944 | sub height, height, clip; \ |
| 945 | sub y_a, y_a, clip; \ |
| 946 | setup_spans_clip(decrement, alternate_active); \ |
| 947 | \ |
| 948 | 0: \ |
| 949 | cmp height, #0; \ |
| 950 | ble 1f; \ |
| 951 | \ |
| 952 | orr temp, y_a, y_a, lsl #16; \ |
| 953 | sub temp, temp, #(1 << 16); \ |
| 954 | sub y_a, temp, #2; \ |
| 955 | sub y_a, y_a, #(2 << 16); \ |
| 956 | vmov y_x4, temp, y_a; \ |
| 957 | \ |
| 958 | vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \ |
| 959 | \ |
| 960 | setup_spans_alternate_pre_increment_##alternate_active(); \ |
| 961 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ |
| 962 | right_index); \ |
| 963 | setup_spans_adjust_interpolants_up(); \ |
| 964 | setup_spans_prologue_b(); \ |
| 965 | \ |
| 966 | strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \ |
| 967 | \ |
| 968 | 2: \ |
| 969 | setup_spans_set_x4_alternate_##alternate_active(alternate, up); \ |
| 970 | subs height, height, #4; \ |
| 971 | bhi 2b; \ |
| 972 | \ |
| 973 | 1: \ |
| 974 | |
| 975 | |
| 976 | #define setup_spans_epilogue() \ |
| 977 | ldmia sp!, { r4 - r11, pc } \ |
| 978 | |
| 979 | |
| 980 | #define setup_spans_up_up(minor, major) \ |
| 981 | setup_spans_prologue(); \ |
| 982 | sub height_minor_a, y_a, y_b; \ |
| 983 | sub height_minor_b, y_b, y_c; \ |
| 984 | sub height, y_a, y_c; \ |
| 985 | \ |
| 986 | vdup.u32 x_starts, x_a; \ |
| 987 | vmov x_ends, x_c, x_b; \ |
| 988 | \ |
| 989 | compute_edge_delta_x3(x_b, height_major, height_minor_a); \ |
| 990 | setup_spans_up(major, minor, minor, yes); \ |
| 991 | setup_spans_epilogue() \ |
| 992 | |
| 993 | function(setup_spans_up_left) |
| 994 | setup_spans_up_up(left, right) |
| 995 | |
| 996 | function(setup_spans_up_right) |
| 997 | setup_spans_up_up(right, left) |
| 998 | |
| 999 | #define setup_spans_down_down(minor, major) \ |
| 1000 | setup_spans_prologue(); \ |
| 1001 | sub height_minor_a, y_b, y_a; \ |
| 1002 | sub height_minor_b, y_c, y_b; \ |
| 1003 | sub height, y_c, y_a; \ |
| 1004 | \ |
| 1005 | vdup.u32 x_starts, x_a; \ |
| 1006 | vmov x_ends, x_c, x_b; \ |
| 1007 | \ |
| 1008 | compute_edge_delta_x3(x_b, height_major, height_minor_a); \ |
| 1009 | setup_spans_down(major, minor, minor, yes); \ |
| 1010 | setup_spans_epilogue() \ |
| 1011 | |
| 1012 | function(setup_spans_down_left) |
| 1013 | setup_spans_down_down(left, right) |
| 1014 | |
| 1015 | function(setup_spans_down_right) |
| 1016 | setup_spans_down_down(right, left) |
| 1017 | |
| 1018 | |
| 1019 | #define setup_spans_up_flat() \ |
| 1020 | sub height, y_a, y_c; \ |
| 1021 | \ |
| 1022 | compute_edge_delta_x2(); \ |
| 1023 | setup_spans_up(left, right, none, no); \ |
| 1024 | setup_spans_epilogue() \ |
| 1025 | |
| 1026 | function(setup_spans_up_a) |
| 1027 | setup_spans_prologue() |
| 1028 | |
| 1029 | vmov x_starts, x_a, x_b |
| 1030 | vdup.u32 x_ends, x_c |
| 1031 | |
| 1032 | setup_spans_up_flat() |
| 1033 | |
| 1034 | function(setup_spans_up_b) |
| 1035 | setup_spans_prologue() |
| 1036 | |
| 1037 | vdup.u32 x_starts, x_a |
| 1038 | vmov x_ends, x_b, x_c |
| 1039 | |
| 1040 | setup_spans_up_flat() |
| 1041 | |
| 1042 | #define setup_spans_down_flat() \ |
| 1043 | sub height, y_c, y_a; \ |
| 1044 | \ |
| 1045 | compute_edge_delta_x2(); \ |
| 1046 | setup_spans_down(left, right, none, no); \ |
| 1047 | setup_spans_epilogue() \ |
| 1048 | |
| 1049 | function(setup_spans_down_a) |
| 1050 | setup_spans_prologue() |
| 1051 | |
| 1052 | vmov x_starts, x_a, x_b |
| 1053 | vdup.u32 x_ends, x_c |
| 1054 | |
| 1055 | setup_spans_down_flat() |
| 1056 | |
| 1057 | function(setup_spans_down_b) |
| 1058 | setup_spans_prologue() |
| 1059 | |
| 1060 | vdup.u32 x_starts, x_a |
| 1061 | vmov x_ends, x_b, x_c |
| 1062 | |
| 1063 | setup_spans_down_flat() |
| 1064 | |
| 1065 | |
| 1066 | #define middle_y r9 |
| 1067 | |
| 1068 | #define edges_xy_b q11 |
| 1069 | #define edges_dx_dy_b d26 |
| 1070 | #define edge_shifts_b d27 |
| 1071 | #define edges_dx_dy_and_shifts_b q13 |
| 1072 | #define height_increment d20 |
| 1073 | |
| 1074 | #define edges_dx_dy_and_shifts q1 |
| 1075 | |
| 1076 | #define edges_xy_b_left d22 |
| 1077 | #define edges_xy_b_right d23 |
| 1078 | |
| 1079 | #define setup_spans_up_down_load_edge_set_b() \ |
| 1080 | vmov edges_xy, edges_xy_b; \ |
| 1081 | vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \ |
| 1082 | |
| 1083 | |
| 1084 | function(setup_spans_up_down) |
| 1085 | setup_spans_prologue() |
| 1086 | |
| 1087 | // s32 middle_y = y_a; |
| 1088 | sub height_minor_a, y_a, y_b |
| 1089 | sub height_minor_b, y_c, y_a |
| 1090 | sub height_major, y_c, y_b |
| 1091 | |
| 1092 | vmov x_starts, x_a, x_c |
| 1093 | vdup.u32 x_ends, x_b |
| 1094 | |
| 1095 | compute_edge_delta_x3(x_a, height_minor_a, height_major) |
| 1096 | |
| 1097 | mov temp, #0 |
| 1098 | vmov height_increment, temp, height_minor_b |
| 1099 | vmlal.s32 edges_xy, edges_dx_dy, height_increment |
| 1100 | |
| 1101 | vmov edges_xy_b_left, edge_alt_low, edge_alt_high |
| 1102 | vmov edges_xy_b_right, edges_xy_right |
| 1103 | |
| 1104 | vmov edge_shifts_b, edge_shifts |
| 1105 | vmov.u32 edge_shifts_b[0], edge_shift_alt |
| 1106 | |
| 1107 | vneg.s32 edges_dx_dy_b, edges_dx_dy |
| 1108 | vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt |
| 1109 | |
| 1110 | mov middle_y, y_a |
| 1111 | |
| 1112 | setup_spans_load_b() |
| 1113 | sub y_a, y_a, #1 |
| 1114 | |
| 1115 | ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset] |
| 1116 | subs temp, temp, y_b |
| 1117 | subgt height_minor_a, height_minor_a, temp |
| 1118 | |
| 1119 | ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset] |
| 1120 | subs clip, y_a, temp |
| 1121 | ble 0f |
| 1122 | |
| 1123 | sub height_minor_a, height_minor_a, clip |
| 1124 | sub y_a, y_a, clip |
| 1125 | setup_spans_clip(decrement, no) |
| 1126 | |
| 1127 | 0: |
| 1128 | cmp height_minor_a, #0 |
| 1129 | ble 3f |
| 1130 | |
| 1131 | orr temp, y_a, y_a, lsl #16 |
| 1132 | sub temp, temp, #(1 << 16) |
| 1133 | sub y_a, temp, #2 |
| 1134 | sub y_a, y_a, #(2 << 16) |
| 1135 | vmov y_x4, temp, y_a |
| 1136 | |
| 1137 | vaddw.s32 edges_xy, edges_xy, edges_dx_dy |
| 1138 | |
| 1139 | strh height_minor_a, [psx_gpu, #psx_gpu_num_spans_offset] |
| 1140 | |
| 1141 | setup_spans_adjust_edges_alternate_no(left, right); |
| 1142 | setup_spans_adjust_interpolants_up() |
| 1143 | setup_spans_up_down_load_edge_set_b() |
| 1144 | |
| 1145 | setup_spans_prologue_b() |
| 1146 | |
| 1147 | |
| 1148 | 2: |
| 1149 | setup_spans_set_x4_alternate_no(none, up) |
| 1150 | subs height_minor_a, height_minor_a, #4 |
| 1151 | bhi 2b |
| 1152 | |
| 1153 | add span_edge_data, span_edge_data, height_minor_a, lsl #3 |
| 1154 | add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4 |
| 1155 | add span_b_offset, span_b_offset, height_minor_a, lsl #2 |
| 1156 | |
| 1157 | 4: |
| 1158 | add temp, psx_gpu, #psx_gpu_uvrg_offset |
| 1159 | vld1.32 { uvrg }, [temp] |
| 1160 | mov y_a, middle_y |
| 1161 | |
| 1162 | setup_spans_load_b() |
| 1163 | |
| 1164 | ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset] |
| 1165 | subs y_c, y_c, temp |
| 1166 | subgt height_minor_b, height_minor_b, y_c |
| 1167 | addgt height_minor_b, height_minor_b, #1 |
| 1168 | |
| 1169 | ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset] |
| 1170 | subs clip, temp, y_a |
| 1171 | ble 0f |
| 1172 | |
| 1173 | sub height_minor_b, height_minor_b, clip |
| 1174 | add y_a, y_a, clip |
| 1175 | setup_spans_clip(increment, no) |
| 1176 | |
| 1177 | 0: |
| 1178 | cmp height_minor_b, #0 |
| 1179 | ble 1f |
| 1180 | |
| 1181 | orr temp, y_a, y_a, lsl #16 |
| 1182 | add temp, temp, #(1 << 16) |
| 1183 | add y_a, temp, #2 |
| 1184 | add y_a, y_a, #(2 << 16) |
| 1185 | vmov y_x4, temp, y_a |
| 1186 | |
| 1187 | setup_spans_adjust_edges_alternate_no(left, right) |
| 1188 | |
| 1189 | ldrh temp, [psx_gpu, #psx_gpu_num_spans_offset] |
| 1190 | add temp, temp, height_minor_b |
| 1191 | |
| 1192 | cmp temp, #MAX_SPANS |
| 1193 | beq 5f |
| 1194 | |
| 1195 | strh temp, [psx_gpu, #psx_gpu_num_spans_offset] |
| 1196 | |
| 1197 | 2: |
| 1198 | setup_spans_set_x4_alternate_no(none, down) |
| 1199 | subs height_minor_b, height_minor_b, #4 |
| 1200 | bhi 2b |
| 1201 | |
| 1202 | 1: |
| 1203 | setup_spans_epilogue() |
| 1204 | |
| 1205 | 3: |
| 1206 | setup_spans_up_down_load_edge_set_b() |
| 1207 | setup_spans_prologue_b() |
| 1208 | bal 4b |
| 1209 | |
| 1210 | 5: |
| 1211 | // FIXME: overflow corner case |
| 1212 | sub temp, temp, height_minor_b |
| 1213 | bics height_minor_b, #3 |
| 1214 | add temp, temp, height_minor_b |
| 1215 | strh temp, [psx_gpu, #psx_gpu_num_spans_offset] |
| 1216 | bne 2b |
| 1217 | bal 1b |
| 1218 | |
| 1219 | #undef span_uvrg_offset |
| 1220 | #undef span_edge_data |
| 1221 | #undef span_b_offset |
| 1222 | #undef left_x |
| 1223 | #undef b |
| 1224 | |
| 1225 | #define psx_gpu r0 |
| 1226 | #define num_spans r1 |
| 1227 | #define span_uvrg_offset r2 |
| 1228 | #define span_edge_data r3 |
| 1229 | #define span_b_offset r4 |
| 1230 | #define b_dx r5 |
| 1231 | #define span_num_blocks r6 |
| 1232 | #define y r7 |
| 1233 | #define left_x r8 |
| 1234 | #define b r9 |
| 1235 | #define dither_offset_ptr r10 |
| 1236 | #define block_ptr_a r11 |
| 1237 | #define fb_ptr r12 |
| 1238 | #define num_blocks r14 |
| 1239 | |
| 1240 | #define uvrg_dx_ptr r2 |
| 1241 | #define texture_mask_ptr r3 |
| 1242 | #define dither_shift r8 |
| 1243 | #define dither_row r10 |
| 1244 | |
| 1245 | #define c_32 r7 |
| 1246 | #define b_dx4 r8 |
| 1247 | #define b_dx8 r9 |
| 1248 | #define block_ptr_b r10 |
| 1249 | |
| 1250 | #define block_span_ptr r10 |
| 1251 | #define right_mask r8 |
| 1252 | |
| 1253 | #define color r2 |
| 1254 | #define color_r r3 |
| 1255 | #define color_g r4 |
| 1256 | #define color_b r5 |
| 1257 | |
| 1258 | #undef uvrg |
| 1259 | |
| 1260 | #define u_block q0 |
| 1261 | #define v_block q1 |
| 1262 | #define r_block q2 |
| 1263 | #define g_block q3 |
| 1264 | #define b_block q4 |
| 1265 | |
| 1266 | #define uv_dx4 d10 |
| 1267 | #define rg_dx4 d11 |
| 1268 | #define uv_dx8 d12 |
| 1269 | #define rg_dx8 d13 |
| 1270 | #define b_whole_8 d14 |
| 1271 | #define fb_mask_ptrs d15 |
| 1272 | |
| 1273 | #define uvrg_dx4 q5 |
| 1274 | #define uvrg_dx8 q6 |
| 1275 | #define uv_dx8 d12 |
| 1276 | #define rg_dx8 d13 |
| 1277 | |
| 1278 | #define u_whole q8 |
| 1279 | #define v_whole q9 |
| 1280 | #define r_whole q10 |
| 1281 | #define g_whole q11 |
| 1282 | #define b_whole q12 |
| 1283 | |
| 1284 | #define u_whole_low d16 |
| 1285 | #define u_whole_high d17 |
| 1286 | #define v_whole_low d18 |
| 1287 | #define v_whole_high d19 |
| 1288 | #define r_whole_low d20 |
| 1289 | #define r_whole_high d21 |
| 1290 | #define g_whole_low d22 |
| 1291 | #define g_whole_high d23 |
| 1292 | #define b_whole_low d24 |
| 1293 | #define b_whole_high d25 |
| 1294 | |
| 1295 | #define dx4 q13 |
| 1296 | #define dx8 q13 |
| 1297 | |
| 1298 | #define u_whole_8 d26 |
| 1299 | #define v_whole_8 d27 |
| 1300 | #define u_whole_8b d24 |
| 1301 | #define r_whole_8 d24 |
| 1302 | #define g_whole_8 d25 |
| 1303 | |
| 1304 | #define uv_whole_8 q13 |
| 1305 | #define uv_whole_8b q14 |
| 1306 | |
| 1307 | #define dither_offsets q14 |
| 1308 | #define texture_mask q15 |
| 1309 | #define texture_mask_u d30 |
| 1310 | #define texture_mask_v d31 |
| 1311 | |
| 1312 | #define dither_offsets_short d28 |
| 1313 | |
| 1314 | #define v_left_x q8 |
| 1315 | #define uvrg q9 |
| 1316 | #define block_span q10 |
| 1317 | |
| 1318 | #define uv d18 |
| 1319 | #define rg d19 |
| 1320 | |
| 1321 | #define draw_mask q1 |
| 1322 | #define draw_mask_edge q13 |
| 1323 | #define test_mask q0 |
| 1324 | |
| 1325 | #define uvrg_dx q3 |
| 1326 | |
| 1327 | #define colors q2 |
| 1328 | |
| 1329 | #define setup_blocks_texture_swizzled() \ |
| 1330 | vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \ |
| 1331 | vsli.u8 u_whole_8, v_whole_8, #4; \ |
| 1332 | vsri.u8 v_whole_8, u_whole_8b, #4 \ |
| 1333 | |
| 1334 | #define setup_blocks_texture_unswizzled() \ |
| 1335 | |
| 1336 | |
| 1337 | #define setup_blocks_shaded_textured_builder(swizzling) \ |
| 1338 | .align 3; \ |
| 1339 | \ |
| 1340 | function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \ |
| 1341 | ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \ |
| 1342 | add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \ |
| 1343 | \ |
| 1344 | vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \ |
| 1345 | add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \ |
| 1346 | \ |
| 1347 | cmp num_spans, #0; \ |
| 1348 | bxeq lr; \ |
| 1349 | \ |
| 1350 | stmdb sp!, { r4 - r11, r14 }; \ |
| 1351 | vshl.u32 uvrg_dx4, uvrg_dx, #2; \ |
| 1352 | \ |
| 1353 | ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \ |
| 1354 | vshl.u32 uvrg_dx8, uvrg_dx, #3; \ |
| 1355 | \ |
| 1356 | vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \ |
| 1357 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 1358 | \ |
| 1359 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 1360 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 1361 | \ |
| 1362 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 1363 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1364 | \ |
| 1365 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \ |
| 1366 | \ |
| 1367 | 0: \ |
| 1368 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1369 | \ |
| 1370 | ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \ |
| 1371 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 1372 | \ |
| 1373 | ldrh y, [span_edge_data, #edge_data_y_offset]; \ |
| 1374 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \ |
| 1375 | \ |
| 1376 | cmp span_num_blocks, #0; \ |
| 1377 | beq 1f; \ |
| 1378 | \ |
| 1379 | ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \ |
| 1380 | add num_blocks, span_num_blocks, num_blocks; \ |
| 1381 | \ |
| 1382 | cmp num_blocks, #MAX_BLOCKS; \ |
| 1383 | bgt 2f; \ |
| 1384 | \ |
| 1385 | 3: \ |
| 1386 | ldr b, [span_b_offset]; \ |
| 1387 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 1388 | \ |
| 1389 | vdup.u32 v_left_x, left_x; \ |
| 1390 | and y, y, #0x3; \ |
| 1391 | \ |
| 1392 | ldr dither_row, [dither_offset_ptr, y, lsl #2]; \ |
| 1393 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 1394 | \ |
| 1395 | mla b, b_dx, left_x, b; \ |
| 1396 | and dither_shift, left_x, #0x03; \ |
| 1397 | \ |
| 1398 | vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \ |
| 1399 | vshr.u32 uvrg_dx, uvrg_dx4, #2; \ |
| 1400 | \ |
| 1401 | mov dither_shift, dither_shift, lsl #3; \ |
| 1402 | vmla.u32 uvrg, uvrg_dx, v_left_x; \ |
| 1403 | \ |
| 1404 | mov c_32, #32; \ |
| 1405 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1406 | \ |
| 1407 | mov dither_row, dither_row, ror dither_shift; \ |
| 1408 | mov b_dx4, b_dx, lsl #2; \ |
| 1409 | \ |
| 1410 | vdup.u32 dither_offsets_short, dither_row; \ |
| 1411 | add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \ |
| 1412 | \ |
| 1413 | vdup.u32 b_block, b; \ |
| 1414 | vshll.s8 dither_offsets, dither_offsets_short, #4; \ |
| 1415 | \ |
| 1416 | vdup.u32 u_block, uv[0]; \ |
| 1417 | mov b_dx8, b_dx, lsl #3; \ |
| 1418 | \ |
| 1419 | vdup.u32 v_block, uv[1]; \ |
| 1420 | vdup.u32 r_block, rg[0]; \ |
| 1421 | vdup.u32 g_block, rg[1]; \ |
| 1422 | \ |
| 1423 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 1424 | \ |
| 1425 | vadd.u32 u_block, u_block, block_span; \ |
| 1426 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 1427 | \ |
| 1428 | vadd.u32 v_block, v_block, block_span; \ |
| 1429 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 1430 | \ |
| 1431 | vadd.u32 r_block, r_block, block_span; \ |
| 1432 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 1433 | \ |
| 1434 | vadd.u32 g_block, g_block, block_span; \ |
| 1435 | vld1.u32 { block_span }, [block_span_ptr, :128]; \ |
| 1436 | \ |
| 1437 | vadd.u32 b_block, b_block, block_span; \ |
| 1438 | add block_ptr_b, block_ptr_a, #16; \ |
| 1439 | \ |
| 1440 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1441 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1442 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 1443 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 1444 | \ |
| 1445 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1446 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 1447 | \ |
| 1448 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1449 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1450 | \ |
| 1451 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1452 | vdup.u32 dx4, rg_dx4[0]; \ |
| 1453 | \ |
| 1454 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 1455 | vdup.u32 dx4, rg_dx4[1]; \ |
| 1456 | \ |
| 1457 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 1458 | vdup.u32 dx4, b_dx4; \ |
| 1459 | \ |
| 1460 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 1461 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1462 | \ |
| 1463 | vadd.u32 u_block, u_block, dx8; \ |
| 1464 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1465 | \ |
| 1466 | vadd.u32 v_block, v_block, dx8; \ |
| 1467 | vdup.u32 dx8, rg_dx8[0]; \ |
| 1468 | \ |
| 1469 | vadd.u32 r_block, r_block, dx8; \ |
| 1470 | vdup.u32 dx8, rg_dx8[1]; \ |
| 1471 | \ |
| 1472 | vadd.u32 g_block, g_block, dx8; \ |
| 1473 | vdup.u32 dx8, b_dx8; \ |
| 1474 | \ |
| 1475 | vadd.u32 b_block, b_block, dx8; \ |
| 1476 | vmovn.u16 u_whole_8, u_whole; \ |
| 1477 | \ |
| 1478 | vmovn.u16 v_whole_8, v_whole; \ |
| 1479 | \ |
| 1480 | vmovn.u16 b_whole_8, b_whole; \ |
| 1481 | pld [fb_ptr]; \ |
| 1482 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1483 | \ |
| 1484 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1485 | setup_blocks_texture_##swizzling(); \ |
| 1486 | \ |
| 1487 | vmovn.u16 r_whole_8, r_whole; \ |
| 1488 | beq 5f; \ |
| 1489 | \ |
| 1490 | 4: \ |
| 1491 | vmovn.u16 g_whole_8, g_whole; \ |
| 1492 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1493 | \ |
| 1494 | vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \ |
| 1495 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1496 | \ |
| 1497 | vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \ |
| 1498 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 1499 | \ |
| 1500 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \ |
| 1501 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 1502 | \ |
| 1503 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1504 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 1505 | \ |
| 1506 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1507 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1508 | \ |
| 1509 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1510 | vdup.u32 dx4, rg_dx4[0]; \ |
| 1511 | \ |
| 1512 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 1513 | vdup.u32 dx4, rg_dx4[1]; \ |
| 1514 | \ |
| 1515 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 1516 | vdup.u32 dx4, b_dx4; \ |
| 1517 | \ |
| 1518 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 1519 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1520 | \ |
| 1521 | vadd.u32 u_block, u_block, dx8; \ |
| 1522 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1523 | \ |
| 1524 | vadd.u32 v_block, v_block, dx8; \ |
| 1525 | vdup.u32 dx8, rg_dx8[0]; \ |
| 1526 | \ |
| 1527 | vadd.u32 r_block, r_block, dx8; \ |
| 1528 | vdup.u32 dx8, rg_dx8[1]; \ |
| 1529 | \ |
| 1530 | vadd.u32 g_block, g_block, dx8; \ |
| 1531 | vdup.u32 dx8, b_dx8; \ |
| 1532 | \ |
| 1533 | vadd.u32 b_block, b_block, dx8; \ |
| 1534 | vmovn.u16 u_whole_8, u_whole; \ |
| 1535 | \ |
| 1536 | add fb_ptr, fb_ptr, #16; \ |
| 1537 | vmovn.u16 v_whole_8, v_whole; \ |
| 1538 | \ |
| 1539 | vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \ |
| 1540 | vmovn.u16 b_whole_8, b_whole; \ |
| 1541 | \ |
| 1542 | pld [fb_ptr]; \ |
| 1543 | \ |
| 1544 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1545 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1546 | \ |
| 1547 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1548 | setup_blocks_texture_##swizzling(); \ |
| 1549 | \ |
| 1550 | vmovn.u16 r_whole_8, r_whole; \ |
| 1551 | bne 4b; \ |
| 1552 | \ |
| 1553 | 5: \ |
| 1554 | vmovn.u16 g_whole_8, g_whole; \ |
| 1555 | ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \ |
| 1556 | \ |
| 1557 | vld1.u32 { test_mask }, [psx_gpu, :128]; \ |
| 1558 | vdup.u8 draw_mask, right_mask; \ |
| 1559 | \ |
| 1560 | vmov.u32 fb_mask_ptrs[0], right_mask; \ |
| 1561 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 1562 | vzip.u8 u_whole_8, v_whole_8; \ |
| 1563 | \ |
| 1564 | vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \ |
| 1565 | vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \ |
| 1566 | vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \ |
| 1567 | vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \ |
| 1568 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \ |
| 1569 | \ |
| 1570 | 1: \ |
| 1571 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 1572 | add span_b_offset, span_b_offset, #4; \ |
| 1573 | \ |
| 1574 | add span_edge_data, span_edge_data, #8; \ |
| 1575 | subs num_spans, num_spans, #1; \ |
| 1576 | \ |
| 1577 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 1578 | bne 0b; \ |
| 1579 | \ |
| 1580 | ldmia sp!, { r4 - r11, pc }; \ |
| 1581 | \ |
| 1582 | 2: \ |
| 1583 | /* TODO: Load from psx_gpu instead of saving/restoring these */\ |
| 1584 | vpush { texture_mask }; \ |
| 1585 | vpush { uvrg_dx4 }; \ |
| 1586 | \ |
| 1587 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \ |
| 1588 | bl flush_render_block_buffer; \ |
| 1589 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \ |
| 1590 | \ |
| 1591 | vpop { uvrg_dx4 }; \ |
| 1592 | vpop { texture_mask }; \ |
| 1593 | \ |
| 1594 | vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \ |
| 1595 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1596 | \ |
| 1597 | mov num_blocks, span_num_blocks; \ |
| 1598 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1599 | bal 3b \ |
| 1600 | |
| 1601 | |
| 1602 | setup_blocks_shaded_textured_builder(swizzled) |
| 1603 | setup_blocks_shaded_textured_builder(unswizzled) |
| 1604 | |
| 1605 | |
| 1606 | #define setup_blocks_unshaded_textured_builder(swizzling) \ |
| 1607 | .align 3; \ |
| 1608 | \ |
| 1609 | function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \ |
| 1610 | ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \ |
| 1611 | add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \ |
| 1612 | \ |
| 1613 | vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \ |
| 1614 | add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \ |
| 1615 | \ |
| 1616 | cmp num_spans, #0; \ |
| 1617 | bxeq lr; \ |
| 1618 | \ |
| 1619 | stmdb sp!, { r4 - r11, r14 }; \ |
| 1620 | vshl.u32 uvrg_dx4, uvrg_dx, #2; \ |
| 1621 | \ |
| 1622 | vshl.u32 uvrg_dx8, uvrg_dx, #3; \ |
| 1623 | \ |
| 1624 | vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \ |
| 1625 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 1626 | \ |
| 1627 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 1628 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 1629 | \ |
| 1630 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1631 | \ |
| 1632 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \ |
| 1633 | \ |
| 1634 | 0: \ |
| 1635 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1636 | \ |
| 1637 | ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \ |
| 1638 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 1639 | \ |
| 1640 | ldrh y, [span_edge_data, #edge_data_y_offset]; \ |
| 1641 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \ |
| 1642 | \ |
| 1643 | cmp span_num_blocks, #0; \ |
| 1644 | beq 1f; \ |
| 1645 | \ |
| 1646 | ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \ |
| 1647 | add num_blocks, span_num_blocks, num_blocks; \ |
| 1648 | \ |
| 1649 | cmp num_blocks, #MAX_BLOCKS; \ |
| 1650 | bgt 2f; \ |
| 1651 | \ |
| 1652 | 3: \ |
| 1653 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 1654 | \ |
| 1655 | vdup.u32 v_left_x, left_x; \ |
| 1656 | and y, y, #0x3; \ |
| 1657 | \ |
| 1658 | ldr dither_row, [dither_offset_ptr, y, lsl #2]; \ |
| 1659 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 1660 | \ |
| 1661 | and dither_shift, left_x, #0x03; \ |
| 1662 | \ |
| 1663 | vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \ |
| 1664 | vshr.u32 uvrg_dx, uvrg_dx4, #2; \ |
| 1665 | \ |
| 1666 | mov dither_shift, dither_shift, lsl #3; \ |
| 1667 | vmla.u32 uvrg, uvrg_dx, v_left_x; \ |
| 1668 | \ |
| 1669 | mov c_32, #32; \ |
| 1670 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1671 | \ |
| 1672 | mov dither_row, dither_row, ror dither_shift; \ |
| 1673 | \ |
| 1674 | vdup.u32 dither_offsets_short, dither_row; \ |
| 1675 | add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \ |
| 1676 | \ |
| 1677 | vshll.s8 dither_offsets, dither_offsets_short, #4; \ |
| 1678 | \ |
| 1679 | vdup.u32 u_block, uv[0]; \ |
| 1680 | \ |
| 1681 | vdup.u32 v_block, uv[1]; \ |
| 1682 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 1683 | \ |
| 1684 | vadd.u32 u_block, u_block, block_span; \ |
| 1685 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 1686 | \ |
| 1687 | vadd.u32 v_block, v_block, block_span; \ |
| 1688 | add block_ptr_b, block_ptr_a, #16; \ |
| 1689 | \ |
| 1690 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1691 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1692 | \ |
| 1693 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1694 | \ |
| 1695 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1696 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1697 | \ |
| 1698 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1699 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1700 | \ |
| 1701 | vadd.u32 u_block, u_block, dx8; \ |
| 1702 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1703 | \ |
| 1704 | vadd.u32 v_block, v_block, dx8; \ |
| 1705 | vmovn.u16 u_whole_8, u_whole; \ |
| 1706 | \ |
| 1707 | vmovn.u16 v_whole_8, v_whole; \ |
| 1708 | \ |
| 1709 | pld [fb_ptr]; \ |
| 1710 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1711 | \ |
| 1712 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1713 | setup_blocks_texture_##swizzling(); \ |
| 1714 | \ |
| 1715 | beq 5f; \ |
| 1716 | \ |
| 1717 | 4: \ |
| 1718 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1719 | \ |
| 1720 | vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \ |
| 1721 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1722 | \ |
| 1723 | add block_ptr_b, block_ptr_b, #32; \ |
| 1724 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \ |
| 1725 | \ |
| 1726 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1727 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1728 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1729 | \ |
| 1730 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1731 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1732 | \ |
| 1733 | vadd.u32 u_block, u_block, dx8; \ |
| 1734 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1735 | \ |
| 1736 | vadd.u32 v_block, v_block, dx8; \ |
| 1737 | vmovn.u16 u_whole_8, u_whole; \ |
| 1738 | \ |
| 1739 | add fb_ptr, fb_ptr, #16; \ |
| 1740 | vmovn.u16 v_whole_8, v_whole; \ |
| 1741 | \ |
| 1742 | vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \ |
| 1743 | pld [fb_ptr]; \ |
| 1744 | \ |
| 1745 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1746 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1747 | \ |
| 1748 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1749 | setup_blocks_texture_##swizzling(); \ |
| 1750 | \ |
| 1751 | bne 4b; \ |
| 1752 | \ |
| 1753 | 5: \ |
| 1754 | ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \ |
| 1755 | \ |
| 1756 | vld1.u32 { test_mask }, [psx_gpu, :128]; \ |
| 1757 | vdup.u8 draw_mask, right_mask; \ |
| 1758 | \ |
| 1759 | vmov.u32 fb_mask_ptrs[0], right_mask; \ |
| 1760 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 1761 | vzip.u8 u_whole_8, v_whole_8; \ |
| 1762 | \ |
| 1763 | vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \ |
| 1764 | add block_ptr_b, block_ptr_b, #32; \ |
| 1765 | vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \ |
| 1766 | vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \ |
| 1767 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \ |
| 1768 | \ |
| 1769 | 1: \ |
| 1770 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 1771 | add span_edge_data, span_edge_data, #8; \ |
| 1772 | subs num_spans, num_spans, #1; \ |
| 1773 | \ |
| 1774 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 1775 | bne 0b; \ |
| 1776 | \ |
| 1777 | ldmia sp!, { r4 - r11, pc }; \ |
| 1778 | \ |
| 1779 | 2: \ |
| 1780 | /* TODO: Load from psx_gpu instead of saving/restoring these */\ |
| 1781 | vpush { texture_mask }; \ |
| 1782 | vpush { uvrg_dx4 }; \ |
| 1783 | \ |
| 1784 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \ |
| 1785 | bl flush_render_block_buffer; \ |
| 1786 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \ |
| 1787 | \ |
| 1788 | vpop { uvrg_dx4 }; \ |
| 1789 | vpop { texture_mask }; \ |
| 1790 | \ |
| 1791 | vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \ |
| 1792 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1793 | \ |
| 1794 | mov num_blocks, span_num_blocks; \ |
| 1795 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1796 | bal 3b \ |
| 1797 | |
| 1798 | |
| 1799 | setup_blocks_unshaded_textured_builder(swizzled) |
| 1800 | setup_blocks_unshaded_textured_builder(unswizzled) |
| 1801 | |
| 1802 | |
| 1803 | .align 3 |
| 1804 | |
| 1805 | function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect) |
| 1806 | ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset] |
| 1807 | veor.u32 draw_mask, draw_mask, draw_mask |
| 1808 | |
| 1809 | cmp num_spans, #0 |
| 1810 | bxeq lr |
| 1811 | |
| 1812 | stmdb sp!, { r4 - r11, r14 } |
| 1813 | vld1.u32 { test_mask }, [psx_gpu, :128] |
| 1814 | |
| 1815 | ldr color, [psx_gpu, #psx_gpu_triangle_color_offset] |
| 1816 | |
| 1817 | ubfx color_r, color, #3, #5 |
| 1818 | ubfx color_g, color, #11, #5 |
| 1819 | ubfx color_b, color, #19, #5 |
| 1820 | |
| 1821 | orr color, color_r, color_b, lsl #10 |
| 1822 | orr color, color, color_g, lsl #5 |
| 1823 | |
| 1824 | vdup.u16 colors, color |
| 1825 | |
| 1826 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 1827 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset |
| 1828 | |
| 1829 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset |
| 1830 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6 |
| 1831 | |
| 1832 | 0: |
| 1833 | ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset] |
| 1834 | ldrh y, [span_edge_data, #edge_data_y_offset] |
| 1835 | |
| 1836 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset] |
| 1837 | |
| 1838 | cmp span_num_blocks, #0 |
| 1839 | beq 1f |
| 1840 | |
| 1841 | ldrh left_x, [span_edge_data, #edge_data_left_x_offset] |
| 1842 | add num_blocks, span_num_blocks, num_blocks |
| 1843 | |
| 1844 | cmp num_blocks, #MAX_BLOCKS |
| 1845 | bgt 2f |
| 1846 | |
| 1847 | 3: |
| 1848 | add fb_ptr, fb_ptr, y, lsl #11 |
| 1849 | and y, y, #0x3 |
| 1850 | |
| 1851 | add fb_ptr, fb_ptr, left_x, lsl #1 |
| 1852 | mov c_32, #32 |
| 1853 | |
| 1854 | subs span_num_blocks, span_num_blocks, #1 |
| 1855 | |
| 1856 | add block_ptr_b, block_ptr_a, #16 |
| 1857 | pld [fb_ptr] |
| 1858 | |
| 1859 | vmov.u32 fb_mask_ptrs[1], fb_ptr |
| 1860 | beq 5f |
| 1861 | |
| 1862 | 4: |
| 1863 | vst1.u32 { draw_mask }, [block_ptr_a, :128], c_32 |
| 1864 | vst1.u32 { colors }, [block_ptr_b, :128], c_32 |
| 1865 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32 |
| 1866 | |
| 1867 | add fb_ptr, fb_ptr, #16 |
| 1868 | add block_ptr_b, block_ptr_b, #32 |
| 1869 | |
| 1870 | pld [fb_ptr] |
| 1871 | |
| 1872 | vmov.u32 fb_mask_ptrs[1], fb_ptr |
| 1873 | subs span_num_blocks, span_num_blocks, #1 |
| 1874 | |
| 1875 | bne 4b |
| 1876 | |
| 1877 | 5: |
| 1878 | ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset] |
| 1879 | |
| 1880 | vdup.u8 draw_mask_edge, right_mask |
| 1881 | vtst.u16 draw_mask_edge, draw_mask_edge, test_mask |
| 1882 | |
| 1883 | vst1.u32 { colors }, [block_ptr_b, :128], c_32 |
| 1884 | vst1.u32 { draw_mask_edge }, [block_ptr_a, :128], c_32 |
| 1885 | add block_ptr_b, block_ptr_b, #32 |
| 1886 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32 |
| 1887 | |
| 1888 | 1: |
| 1889 | add span_edge_data, span_edge_data, #8 |
| 1890 | subs num_spans, num_spans, #1 |
| 1891 | |
| 1892 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 1893 | bne 0b |
| 1894 | |
| 1895 | ldmia sp!, { r4 - r11, pc } |
| 1896 | |
| 1897 | 2: |
| 1898 | vpush { colors } |
| 1899 | |
| 1900 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 } |
| 1901 | bl flush_render_block_buffer |
| 1902 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 } |
| 1903 | |
| 1904 | vpop { colors } |
| 1905 | |
| 1906 | vld1.u32 { test_mask }, [psx_gpu, :128] |
| 1907 | veor.u32 draw_mask, draw_mask, draw_mask |
| 1908 | |
| 1909 | mov num_blocks, span_num_blocks |
| 1910 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset |
| 1911 | bal 3b |
| 1912 | |
| 1913 | |
| 1914 | #define mask_msb_scalar r14 |
| 1915 | |
| 1916 | #define msb_mask q15 |
| 1917 | |
| 1918 | #define pixels_low d16 |
| 1919 | |
| 1920 | #define msb_mask_low d30 |
| 1921 | #define msb_mask_high d31 |
| 1922 | |
| 1923 | |
| 1924 | .align 3 |
| 1925 | |
| 1926 | function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct) |
| 1927 | ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset] |
| 1928 | |
| 1929 | cmp num_spans, #0 |
| 1930 | bxeq lr |
| 1931 | |
| 1932 | stmdb sp!, { r4 - r11, r14 } |
| 1933 | |
| 1934 | ldr color, [psx_gpu, #psx_gpu_triangle_color_offset] |
| 1935 | |
| 1936 | ubfx color_r, color, #3, #5 |
| 1937 | ubfx color_g, color, #11, #5 |
| 1938 | |
| 1939 | ldrh mask_msb_scalar, [psx_gpu, #psx_gpu_mask_msb_offset] |
| 1940 | ubfx color_b, color, #19, #5 |
| 1941 | |
| 1942 | orr color, color_r, color_b, lsl #10 |
| 1943 | orr color, color, color_g, lsl #5 |
| 1944 | orr color, color, mask_msb_scalar |
| 1945 | |
| 1946 | vdup.u16 colors, color |
| 1947 | |
| 1948 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset |
| 1949 | orr color, color, color, lsl #16 |
| 1950 | |
| 1951 | |
| 1952 | 0: |
| 1953 | ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset] |
| 1954 | ldrh y, [span_edge_data, #edge_data_y_offset] |
| 1955 | |
| 1956 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset] |
| 1957 | |
| 1958 | cmp span_num_blocks, #0 |
| 1959 | beq 1f |
| 1960 | |
| 1961 | ldrh left_x, [span_edge_data, #edge_data_left_x_offset] |
| 1962 | |
| 1963 | add fb_ptr, fb_ptr, y, lsl #11 |
| 1964 | subs span_num_blocks, span_num_blocks, #1 |
| 1965 | |
| 1966 | add fb_ptr, fb_ptr, left_x, lsl #1 |
| 1967 | beq 3f |
| 1968 | |
| 1969 | 2: |
| 1970 | vst1.u32 { colors }, [fb_ptr]! |
| 1971 | subs span_num_blocks, span_num_blocks, #1 |
| 1972 | |
| 1973 | bne 2b |
| 1974 | |
| 1975 | 3: |
| 1976 | ldrb right_mask, [span_edge_data, #edge_data_right_mask_offset] |
| 1977 | |
| 1978 | cmp right_mask, #0x0 |
| 1979 | beq 5f |
| 1980 | |
| 1981 | tst right_mask, #0xF |
| 1982 | streq color, [fb_ptr], #4 |
| 1983 | moveq right_mask, right_mask, lsr #4 |
| 1984 | streq color, [fb_ptr], #4 |
| 1985 | |
| 1986 | tst right_mask, #0x3 |
| 1987 | streq color, [fb_ptr], #4 |
| 1988 | moveq right_mask, right_mask, lsr #2 |
| 1989 | |
| 1990 | tst right_mask, #0x1 |
| 1991 | strheq color, [fb_ptr] |
| 1992 | |
| 1993 | 1: |
| 1994 | add span_edge_data, span_edge_data, #8 |
| 1995 | subs num_spans, num_spans, #1 |
| 1996 | bne 0b |
| 1997 | |
| 1998 | ldmia sp!, { r4 - r11, pc } |
| 1999 | |
| 2000 | 5: |
| 2001 | vst1.u32 { colors }, [fb_ptr] |
| 2002 | bal 1b |
| 2003 | |
| 2004 | |
| 2005 | #undef c_64 |
| 2006 | |
| 2007 | #define c_64 r7 |
| 2008 | #define rg_dx_ptr r2 |
| 2009 | |
| 2010 | |
| 2011 | #undef r_block |
| 2012 | #undef g_block |
| 2013 | #undef b_block |
| 2014 | #undef r_whole |
| 2015 | #undef g_whole |
| 2016 | #undef b_whole |
| 2017 | #undef r_whole_low |
| 2018 | #undef r_whole_high |
| 2019 | #undef g_whole_low |
| 2020 | #undef g_whole_high |
| 2021 | #undef b_whole_low |
| 2022 | #undef b_whole_high |
| 2023 | #undef r_whole_8 |
| 2024 | #undef g_whole_8 |
| 2025 | #undef b_whole_8 |
| 2026 | #undef dither_offsets |
| 2027 | #undef rg_dx4 |
| 2028 | #undef rg_dx8 |
| 2029 | #undef dx4 |
| 2030 | #undef dx8 |
| 2031 | #undef v_left_x |
| 2032 | #undef uvrg |
| 2033 | #undef block_span |
| 2034 | #undef rg |
| 2035 | #undef draw_mask |
| 2036 | #undef test_mask |
| 2037 | |
| 2038 | #define r_block q0 |
| 2039 | #define g_block q1 |
| 2040 | #define b_block q2 |
| 2041 | |
| 2042 | #define r_whole q3 |
| 2043 | #define g_whole q4 |
| 2044 | #define b_whole q5 |
| 2045 | |
| 2046 | #define r_whole_low d6 |
| 2047 | #define r_whole_high d7 |
| 2048 | #define g_whole_low d8 |
| 2049 | #define g_whole_high d9 |
| 2050 | #define b_whole_low d10 |
| 2051 | #define b_whole_high d11 |
| 2052 | |
| 2053 | #define gb_whole_8 q6 |
| 2054 | |
| 2055 | #define g_whole_8 d12 |
| 2056 | #define b_whole_8 d13 |
| 2057 | |
| 2058 | #define r_whole_8 d14 |
| 2059 | |
| 2060 | #define pixels q8 |
| 2061 | |
| 2062 | #define rg_dx4 d18 |
| 2063 | #define rg_dx8 d19 |
| 2064 | |
| 2065 | #define dx4 q10 |
| 2066 | #define dx8 q10 |
| 2067 | |
| 2068 | #define v_left_x d6 |
| 2069 | #define uvrg q4 |
| 2070 | #define block_span q5 |
| 2071 | |
| 2072 | #define rg d9 |
| 2073 | |
| 2074 | #define d64_1 d22 |
| 2075 | #define d64_128 d23 |
| 2076 | |
| 2077 | #define d128_4 q12 |
| 2078 | #define d128_0x7 q13 |
| 2079 | |
| 2080 | #define d64_4 d24 |
| 2081 | |
| 2082 | #define dither_offsets q14 |
| 2083 | #define draw_mask q15 |
| 2084 | |
| 2085 | #define dither_offsets_low d28 |
| 2086 | |
| 2087 | #define rg_dx d0 |
| 2088 | #define test_mask q10 |
| 2089 | |
| 2090 | |
| 2091 | #define setup_blocks_shaded_untextured_dither_a_dithered() \ |
| 2092 | vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \ |
| 2093 | vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \ |
| 2094 | |
| 2095 | #define setup_blocks_shaded_untextured_dither_b_dithered() \ |
| 2096 | vqsub.u8 r_whole_8, r_whole_8, d64_4; \ |
| 2097 | vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \ |
| 2098 | |
| 2099 | #define setup_blocks_shaded_untextured_dither_a_undithered() \ |
| 2100 | |
| 2101 | #define setup_blocks_shaded_untextured_dither_b_undithered() \ |
| 2102 | |
| 2103 | |
| 2104 | #define setup_blocks_shaded_untextured_indirect_builder(dithering) \ |
| 2105 | .align 3; \ |
| 2106 | \ |
| 2107 | function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \ |
| 2108 | ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \ |
| 2109 | add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \ |
| 2110 | \ |
| 2111 | vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \ |
| 2112 | \ |
| 2113 | cmp num_spans, #0; \ |
| 2114 | bxeq lr; \ |
| 2115 | \ |
| 2116 | stmdb sp!, { r4 - r11, r14 }; \ |
| 2117 | vshl.u32 rg_dx4, rg_dx, #2; \ |
| 2118 | \ |
| 2119 | ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \ |
| 2120 | vshl.u32 rg_dx8, rg_dx, #3; \ |
| 2121 | \ |
| 2122 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 2123 | \ |
| 2124 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 2125 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 2126 | \ |
| 2127 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 2128 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 2129 | \ |
| 2130 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \ |
| 2131 | vmov.u8 d64_1, #1; \ |
| 2132 | \ |
| 2133 | vmov.u8 d128_4, #4; \ |
| 2134 | vmov.u8 d64_128, #128; \ |
| 2135 | \ |
| 2136 | vmov.u8 d128_0x7, #0x7; \ |
| 2137 | \ |
| 2138 | 0: \ |
| 2139 | ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \ |
| 2140 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 2141 | \ |
| 2142 | ldrh y, [span_edge_data, #edge_data_y_offset]; \ |
| 2143 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \ |
| 2144 | \ |
| 2145 | cmp span_num_blocks, #0; \ |
| 2146 | beq 1f; \ |
| 2147 | \ |
| 2148 | ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \ |
| 2149 | add num_blocks, span_num_blocks, num_blocks; \ |
| 2150 | \ |
| 2151 | cmp num_blocks, #MAX_BLOCKS; \ |
| 2152 | bgt 2f; \ |
| 2153 | \ |
| 2154 | 3: \ |
| 2155 | ldr b, [span_b_offset]; \ |
| 2156 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 2157 | \ |
| 2158 | vdup.u32 v_left_x, left_x; \ |
| 2159 | and y, y, #0x3; \ |
| 2160 | \ |
| 2161 | ldr dither_row, [dither_offset_ptr, y, lsl #2]; \ |
| 2162 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 2163 | \ |
| 2164 | mla b, b_dx, left_x, b; \ |
| 2165 | and dither_shift, left_x, #0x03; \ |
| 2166 | \ |
| 2167 | vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \ |
| 2168 | vshr.u32 rg_dx, rg_dx4, #2; \ |
| 2169 | \ |
| 2170 | mov dither_shift, dither_shift, lsl #3; \ |
| 2171 | vmla.u32 rg, rg_dx, v_left_x; \ |
| 2172 | \ |
| 2173 | mov c_64, #64; \ |
| 2174 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2175 | \ |
| 2176 | mov dither_row, dither_row, ror dither_shift; \ |
| 2177 | mov b_dx4, b_dx, lsl #2; \ |
| 2178 | \ |
| 2179 | vdup.u32 dither_offsets, dither_row; \ |
| 2180 | add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \ |
| 2181 | \ |
| 2182 | vdup.u32 b_block, b; \ |
| 2183 | vadd.u8 dither_offsets, dither_offsets, d128_4; \ |
| 2184 | \ |
| 2185 | mov b_dx8, b_dx, lsl #3; \ |
| 2186 | vdup.u32 r_block, rg[0]; \ |
| 2187 | vdup.u32 g_block, rg[1]; \ |
| 2188 | \ |
| 2189 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 2190 | \ |
| 2191 | vadd.u32 r_block, r_block, block_span; \ |
| 2192 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 2193 | \ |
| 2194 | vadd.u32 g_block, g_block, block_span; \ |
| 2195 | vld1.u32 { block_span }, [block_span_ptr, :128]; \ |
| 2196 | \ |
| 2197 | vadd.u32 b_block, b_block, block_span; \ |
| 2198 | add block_ptr_b, block_ptr_a, #16; \ |
| 2199 | \ |
| 2200 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2201 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2202 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2203 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2204 | \ |
| 2205 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2206 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2207 | \ |
| 2208 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2209 | vdup.u32 dx4, b_dx4; \ |
| 2210 | \ |
| 2211 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2212 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2213 | \ |
| 2214 | vadd.u32 r_block, r_block, dx8; \ |
| 2215 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2216 | \ |
| 2217 | vadd.u32 g_block, g_block, dx8; \ |
| 2218 | vdup.u32 dx8, b_dx8; \ |
| 2219 | \ |
| 2220 | vadd.u32 b_block, b_block, dx8; \ |
| 2221 | \ |
| 2222 | vmovn.u16 r_whole_8, r_whole; \ |
| 2223 | vmovn.u16 g_whole_8, g_whole; \ |
| 2224 | vmovn.u16 b_whole_8, b_whole; \ |
| 2225 | \ |
| 2226 | beq 5f; \ |
| 2227 | veor.u32 draw_mask, draw_mask, draw_mask; \ |
| 2228 | \ |
| 2229 | 4: \ |
| 2230 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2231 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2232 | \ |
| 2233 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2234 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2235 | \ |
| 2236 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2237 | str fb_ptr, [block_ptr_a, #44]; \ |
| 2238 | \ |
| 2239 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2240 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2241 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2242 | \ |
| 2243 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2244 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2245 | \ |
| 2246 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2247 | vdup.u32 dx4, b_dx4; \ |
| 2248 | \ |
| 2249 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2250 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2251 | \ |
| 2252 | vmull.u8 pixels, r_whole_8, d64_1; \ |
| 2253 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2254 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2255 | \ |
| 2256 | vadd.u32 r_block, r_block, dx8; \ |
| 2257 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2258 | \ |
| 2259 | vadd.u32 g_block, g_block, dx8; \ |
| 2260 | vdup.u32 dx8, b_dx8; \ |
| 2261 | \ |
| 2262 | vadd.u32 b_block, b_block, dx8; \ |
| 2263 | add fb_ptr, fb_ptr, #16; \ |
| 2264 | \ |
| 2265 | vmovn.u16 r_whole_8, r_whole; \ |
| 2266 | vmovn.u16 g_whole_8, g_whole; \ |
| 2267 | vmovn.u16 b_whole_8, b_whole; \ |
| 2268 | \ |
| 2269 | vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \ |
| 2270 | vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \ |
| 2271 | \ |
| 2272 | pld [fb_ptr]; \ |
| 2273 | \ |
| 2274 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2275 | bne 4b; \ |
| 2276 | \ |
| 2277 | 5: \ |
| 2278 | str fb_ptr, [block_ptr_a, #44]; \ |
| 2279 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2280 | \ |
| 2281 | ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \ |
| 2282 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2283 | \ |
| 2284 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2285 | vdup.u8 draw_mask, right_mask; \ |
| 2286 | \ |
| 2287 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2288 | vld1.u32 { test_mask }, [psx_gpu, :128]; \ |
| 2289 | \ |
| 2290 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 2291 | \ |
| 2292 | vmull.u8 pixels, r_whole_8, d64_1; \ |
| 2293 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2294 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2295 | \ |
| 2296 | vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \ |
| 2297 | vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \ |
| 2298 | \ |
| 2299 | 1: \ |
| 2300 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 2301 | add span_b_offset, span_b_offset, #4; \ |
| 2302 | \ |
| 2303 | add span_edge_data, span_edge_data, #8; \ |
| 2304 | subs num_spans, num_spans, #1; \ |
| 2305 | \ |
| 2306 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 2307 | bne 0b; \ |
| 2308 | \ |
| 2309 | ldmia sp!, { r4 - r11, pc }; \ |
| 2310 | \ |
| 2311 | 2: \ |
| 2312 | /* TODO: Load from psx_gpu instead of saving/restoring these */\ |
| 2313 | vpush { rg_dx4 }; \ |
| 2314 | \ |
| 2315 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \ |
| 2316 | bl flush_render_block_buffer; \ |
| 2317 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \ |
| 2318 | \ |
| 2319 | vpop { rg_dx4 }; \ |
| 2320 | \ |
| 2321 | vmov.u8 d64_1, #1; \ |
| 2322 | vmov.u8 d128_4, #4; \ |
| 2323 | vmov.u8 d64_128, #128; \ |
| 2324 | vmov.u8 d128_0x7, #0x7; \ |
| 2325 | \ |
| 2326 | vadd.u32 rg_dx8, rg_dx4, rg_dx4; \ |
| 2327 | \ |
| 2328 | mov num_blocks, span_num_blocks; \ |
| 2329 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 2330 | bal 3b \ |
| 2331 | |
| 2332 | |
| 2333 | setup_blocks_shaded_untextured_indirect_builder(undithered) |
| 2334 | setup_blocks_shaded_untextured_indirect_builder(dithered) |
| 2335 | |
| 2336 | |
| 2337 | #undef draw_mask |
| 2338 | |
| 2339 | #define mask_msb_ptr r14 |
| 2340 | |
| 2341 | #define draw_mask q0 |
| 2342 | #define pixels_low d16 |
| 2343 | #define pixels_high d17 |
| 2344 | |
| 2345 | |
| 2346 | |
| 2347 | #define setup_blocks_shaded_untextured_direct_builder(dithering) \ |
| 2348 | .align 3; \ |
| 2349 | \ |
| 2350 | function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \ |
| 2351 | ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \ |
| 2352 | add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \ |
| 2353 | \ |
| 2354 | vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \ |
| 2355 | \ |
| 2356 | cmp num_spans, #0; \ |
| 2357 | bxeq lr; \ |
| 2358 | \ |
| 2359 | stmdb sp!, { r4 - r11, r14 }; \ |
| 2360 | vshl.u32 rg_dx4, rg_dx, #2; \ |
| 2361 | \ |
| 2362 | ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \ |
| 2363 | vshl.u32 rg_dx8, rg_dx, #3; \ |
| 2364 | \ |
| 2365 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 2366 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 2367 | \ |
| 2368 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 2369 | vmov.u8 d64_1, #1; \ |
| 2370 | \ |
| 2371 | vmov.u8 d128_4, #4; \ |
| 2372 | vmov.u8 d64_128, #128; \ |
| 2373 | \ |
| 2374 | vmov.u8 d128_0x7, #0x7; \ |
| 2375 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 2376 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 2377 | \ |
| 2378 | 0: \ |
| 2379 | ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \ |
| 2380 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 2381 | \ |
| 2382 | ldrh y, [span_edge_data, #edge_data_y_offset]; \ |
| 2383 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \ |
| 2384 | \ |
| 2385 | cmp span_num_blocks, #0; \ |
| 2386 | beq 1f; \ |
| 2387 | \ |
| 2388 | ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \ |
| 2389 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 2390 | \ |
| 2391 | ldr b, [span_b_offset]; \ |
| 2392 | vdup.u32 v_left_x, left_x; \ |
| 2393 | and y, y, #0x3; \ |
| 2394 | \ |
| 2395 | ldr dither_row, [dither_offset_ptr, y, lsl #2]; \ |
| 2396 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 2397 | \ |
| 2398 | mla b, b_dx, left_x, b; \ |
| 2399 | and dither_shift, left_x, #0x03; \ |
| 2400 | \ |
| 2401 | vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \ |
| 2402 | vshr.u32 rg_dx, rg_dx4, #2; \ |
| 2403 | \ |
| 2404 | mov dither_shift, dither_shift, lsl #3; \ |
| 2405 | vmla.u32 rg, rg_dx, v_left_x; \ |
| 2406 | \ |
| 2407 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2408 | \ |
| 2409 | mov dither_row, dither_row, ror dither_shift; \ |
| 2410 | mov b_dx4, b_dx, lsl #2; \ |
| 2411 | \ |
| 2412 | vdup.u32 dither_offsets, dither_row; \ |
| 2413 | add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \ |
| 2414 | \ |
| 2415 | vdup.u32 b_block, b; \ |
| 2416 | vadd.u8 dither_offsets, dither_offsets, d128_4; \ |
| 2417 | \ |
| 2418 | mov b_dx8, b_dx, lsl #3; \ |
| 2419 | vdup.u32 r_block, rg[0]; \ |
| 2420 | vdup.u32 g_block, rg[1]; \ |
| 2421 | \ |
| 2422 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 2423 | \ |
| 2424 | vadd.u32 r_block, r_block, block_span; \ |
| 2425 | vld1.u32 { block_span }, [block_span_ptr, :128]!; \ |
| 2426 | \ |
| 2427 | vadd.u32 g_block, g_block, block_span; \ |
| 2428 | vld1.u32 { block_span }, [block_span_ptr, :128]; \ |
| 2429 | \ |
| 2430 | vadd.u32 b_block, b_block, block_span; \ |
| 2431 | add block_ptr_b, block_ptr_a, #16; \ |
| 2432 | \ |
| 2433 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2434 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2435 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2436 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2437 | \ |
| 2438 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2439 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2440 | \ |
| 2441 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2442 | vdup.u32 dx4, b_dx4; \ |
| 2443 | \ |
| 2444 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2445 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2446 | \ |
| 2447 | vadd.u32 r_block, r_block, dx8; \ |
| 2448 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2449 | \ |
| 2450 | vadd.u32 g_block, g_block, dx8; \ |
| 2451 | vdup.u32 dx8, b_dx8; \ |
| 2452 | \ |
| 2453 | vadd.u32 b_block, b_block, dx8; \ |
| 2454 | \ |
| 2455 | vmovn.u16 r_whole_8, r_whole; \ |
| 2456 | vmovn.u16 g_whole_8, g_whole; \ |
| 2457 | vmovn.u16 b_whole_8, b_whole; \ |
| 2458 | \ |
| 2459 | beq 3f; \ |
| 2460 | \ |
| 2461 | 2: \ |
| 2462 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2463 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2464 | \ |
| 2465 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2466 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2467 | \ |
| 2468 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2469 | \ |
| 2470 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2471 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2472 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2473 | \ |
| 2474 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2475 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2476 | \ |
| 2477 | vmov pixels, msb_mask; \ |
| 2478 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2479 | vdup.u32 dx4, b_dx4; \ |
| 2480 | \ |
| 2481 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2482 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2483 | \ |
| 2484 | vmlal.u8 pixels, r_whole_8, d64_1; \ |
| 2485 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2486 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2487 | \ |
| 2488 | vadd.u32 r_block, r_block, dx8; \ |
| 2489 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2490 | \ |
| 2491 | vadd.u32 g_block, g_block, dx8; \ |
| 2492 | vdup.u32 dx8, b_dx8; \ |
| 2493 | \ |
| 2494 | vadd.u32 b_block, b_block, dx8; \ |
| 2495 | \ |
| 2496 | vmovn.u16 r_whole_8, r_whole; \ |
| 2497 | vmovn.u16 g_whole_8, g_whole; \ |
| 2498 | vmovn.u16 b_whole_8, b_whole; \ |
| 2499 | \ |
| 2500 | vst1.u32 { pixels }, [fb_ptr]!; \ |
| 2501 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2502 | bne 2b; \ |
| 2503 | \ |
| 2504 | 3: \ |
| 2505 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2506 | \ |
| 2507 | ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \ |
| 2508 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2509 | \ |
| 2510 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2511 | rbit right_mask, right_mask; \ |
| 2512 | vmov pixels, msb_mask; \ |
| 2513 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2514 | clz right_mask, right_mask; \ |
| 2515 | \ |
| 2516 | vmlal.u8 pixels, r_whole_8, d64_1; \ |
| 2517 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2518 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2519 | \ |
| 2520 | JT_OP_REL(100f, right_mask, temp); \ |
| 2521 | JT_OP(ldr pc, [pc, right_mask, lsl #2]); \ |
| 2522 | nop; \ |
| 2523 | 100: \ |
| 2524 | nop; \ |
| 2525 | .word JTE(100b, 4f); \ |
| 2526 | .word JTE(100b, 5f); \ |
| 2527 | .word JTE(100b, 6f); \ |
| 2528 | .word JTE(100b, 7f); \ |
| 2529 | .word JTE(100b, 8f); \ |
| 2530 | .word JTE(100b, 9f); \ |
| 2531 | .word JTE(100b, 10f); \ |
| 2532 | .word JTE(100b, 11f); \ |
| 2533 | \ |
| 2534 | 4: \ |
| 2535 | vst1.u16 { pixels_low[0] }, [fb_ptr]; \ |
| 2536 | bal 1f; \ |
| 2537 | \ |
| 2538 | 5: \ |
| 2539 | vst1.u32 { pixels_low[0] }, [fb_ptr]; \ |
| 2540 | bal 1f; \ |
| 2541 | \ |
| 2542 | 6: \ |
| 2543 | vst1.u32 { pixels_low[0] }, [fb_ptr]!; \ |
| 2544 | vst1.u16 { pixels_low[2] }, [fb_ptr]; \ |
| 2545 | bal 1f; \ |
| 2546 | \ |
| 2547 | 7: \ |
| 2548 | vst1.u32 { pixels_low }, [fb_ptr]; \ |
| 2549 | bal 1f; \ |
| 2550 | \ |
| 2551 | 8: \ |
| 2552 | vst1.u32 { pixels_low }, [fb_ptr]!; \ |
| 2553 | vst1.u16 { pixels_high[0] }, [fb_ptr]; \ |
| 2554 | bal 1f; \ |
| 2555 | \ |
| 2556 | 9: \ |
| 2557 | vst1.u32 { pixels_low }, [fb_ptr]!; \ |
| 2558 | vst1.u32 { pixels_high[0] }, [fb_ptr]!; \ |
| 2559 | bal 1f; \ |
| 2560 | \ |
| 2561 | 10: \ |
| 2562 | vst1.u32 { pixels_low }, [fb_ptr]!; \ |
| 2563 | vst1.u32 { pixels_high[0] }, [fb_ptr]!; \ |
| 2564 | vst1.u16 { pixels_high[2] }, [fb_ptr]; \ |
| 2565 | bal 1f; \ |
| 2566 | \ |
| 2567 | 11: \ |
| 2568 | vst1.u32 { pixels }, [fb_ptr]; \ |
| 2569 | bal 1f; \ |
| 2570 | \ |
| 2571 | 1: \ |
| 2572 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 2573 | add span_b_offset, span_b_offset, #4; \ |
| 2574 | \ |
| 2575 | add span_edge_data, span_edge_data, #8; \ |
| 2576 | subs num_spans, num_spans, #1; \ |
| 2577 | \ |
| 2578 | bne 0b; \ |
| 2579 | \ |
| 2580 | ldmia sp!, { r4 - r11, pc } \ |
| 2581 | |
| 2582 | setup_blocks_shaded_untextured_direct_builder(undithered) |
| 2583 | setup_blocks_shaded_untextured_direct_builder(dithered) |
| 2584 | |
| 2585 | |
| 2586 | #undef psx_gpu |
| 2587 | #undef num_blocks |
| 2588 | #undef triangle |
| 2589 | #undef c_64 |
| 2590 | |
| 2591 | #define psx_gpu r0 |
| 2592 | #define block_ptr r1 |
| 2593 | #define num_blocks r2 |
| 2594 | #define uv_01 r3 |
| 2595 | #define uv_23 r4 |
| 2596 | #define uv_45 r5 |
| 2597 | #define uv_67 r6 |
| 2598 | #define uv_0 r7 |
| 2599 | #define uv_1 r3 |
| 2600 | #define uv_2 r8 |
| 2601 | #define uv_3 r4 |
| 2602 | #define uv_4 r9 |
| 2603 | #define uv_5 r5 |
| 2604 | #define uv_6 r10 |
| 2605 | #define uv_7 r6 |
| 2606 | #define texture_ptr r11 |
| 2607 | |
| 2608 | #define pixel_0 r7 |
| 2609 | #define pixel_1 r3 |
| 2610 | #define pixel_2 r8 |
| 2611 | #define pixel_3 r4 |
| 2612 | #define pixel_4 r9 |
| 2613 | #define pixel_5 r5 |
| 2614 | #define pixel_6 r10 |
| 2615 | #define pixel_7 r6 |
| 2616 | |
| 2617 | #define pixels_a r7 |
| 2618 | #define pixels_b r9 |
| 2619 | #define pixels_c r8 |
| 2620 | #define pixels_d r10 |
| 2621 | |
| 2622 | #define c_64 r0 |
| 2623 | |
| 2624 | #define clut_ptr r12 |
| 2625 | #define current_texture_mask r5 |
| 2626 | #define dirty_textures_mask r6 |
| 2627 | |
| 2628 | #define texels d0 |
| 2629 | |
| 2630 | #define clut_low_a d2 |
| 2631 | #define clut_low_b d3 |
| 2632 | #define clut_high_a d4 |
| 2633 | #define clut_high_b d5 |
| 2634 | |
| 2635 | #define clut_a q1 |
| 2636 | #define clut_b q2 |
| 2637 | |
| 2638 | #define texels_low d6 |
| 2639 | #define texels_high d7 |
| 2640 | |
| 2641 | .align 3 |
| 2642 | |
| 2643 | function(texture_blocks_untextured) |
| 2644 | bx lr |
| 2645 | |
| 2646 | |
| 2647 | .align 3 |
| 2648 | |
| 2649 | function(texture_blocks_4bpp) |
| 2650 | stmdb sp!, { r3 - r11, r14 } |
| 2651 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 2652 | |
| 2653 | ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] |
| 2654 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 2655 | |
| 2656 | ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset] |
| 2657 | vld1.u32 { clut_a, clut_b }, [clut_ptr, :128] |
| 2658 | |
| 2659 | ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset] |
| 2660 | vuzp.u8 clut_a, clut_b |
| 2661 | |
| 2662 | ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset] |
| 2663 | tst dirty_textures_mask, current_texture_mask |
| 2664 | |
| 2665 | bne 1f |
| 2666 | mov c_64, #64 |
| 2667 | |
| 2668 | 0: |
| 2669 | ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 } |
| 2670 | |
| 2671 | uxtah uv_0, texture_ptr, uv_01 |
| 2672 | uxtah uv_1, texture_ptr, uv_01, ror #16 |
| 2673 | |
| 2674 | uxtah uv_2, texture_ptr, uv_23 |
| 2675 | uxtah uv_3, texture_ptr, uv_23, ror #16 |
| 2676 | |
| 2677 | uxtah uv_4, texture_ptr, uv_45 |
| 2678 | ldrb pixel_0, [uv_0] |
| 2679 | |
| 2680 | uxtah uv_5, texture_ptr, uv_45, ror #16 |
| 2681 | ldrb pixel_1, [uv_1] |
| 2682 | |
| 2683 | uxtah uv_6, texture_ptr, uv_67 |
| 2684 | ldrb pixel_2, [uv_2] |
| 2685 | |
| 2686 | uxtah uv_7, texture_ptr, uv_67, ror #16 |
| 2687 | ldrb pixel_3, [uv_3] |
| 2688 | |
| 2689 | ldrb pixel_4, [uv_4] |
| 2690 | subs num_blocks, num_blocks, #1 |
| 2691 | |
| 2692 | ldrb pixel_5, [uv_5] |
| 2693 | orr pixels_a, pixel_0, pixel_1, lsl #8 |
| 2694 | |
| 2695 | ldrb pixel_6, [uv_6] |
| 2696 | orr pixels_b, pixel_4, pixel_5, lsl #8 |
| 2697 | |
| 2698 | ldrb pixel_7, [uv_7] |
| 2699 | orr pixels_a, pixels_a, pixel_2, lsl #16 |
| 2700 | |
| 2701 | orr pixels_b, pixels_b, pixel_6, lsl #16 |
| 2702 | orr pixels_a, pixels_a, pixel_3, lsl #24 |
| 2703 | |
| 2704 | orr pixels_b, pixels_b, pixel_7, lsl #24 |
| 2705 | vmov texels, pixels_a, pixels_b |
| 2706 | |
| 2707 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels |
| 2708 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels |
| 2709 | |
| 2710 | vst2.u8 { texels_low, texels_high }, [block_ptr, :128], c_64 |
| 2711 | bne 0b |
| 2712 | |
| 2713 | ldmia sp!, { r3 - r11, pc } |
| 2714 | |
| 2715 | 1: |
| 2716 | stmdb sp!, { r1 - r2 } |
| 2717 | bl update_texture_4bpp_cache |
| 2718 | |
| 2719 | mov c_64, #64 |
| 2720 | ldmia sp!, { r1 - r2 } |
| 2721 | bal 0b |
| 2722 | |
| 2723 | |
| 2724 | .align 3 |
| 2725 | |
| 2726 | function(texture_blocks_8bpp) |
| 2727 | stmdb sp!, { r3 - r11, r14 } |
| 2728 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 2729 | |
| 2730 | ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] |
| 2731 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 2732 | |
| 2733 | ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset] |
| 2734 | ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset] |
| 2735 | |
| 2736 | ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset] |
| 2737 | tst dirty_textures_mask, current_texture_mask |
| 2738 | |
| 2739 | bne 1f |
| 2740 | nop |
| 2741 | |
| 2742 | 0: |
| 2743 | ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 } |
| 2744 | |
| 2745 | uxtah uv_0, texture_ptr, uv_01 |
| 2746 | uxtah uv_1, texture_ptr, uv_01, ror #16 |
| 2747 | |
| 2748 | uxtah uv_2, texture_ptr, uv_23 |
| 2749 | uxtah uv_3, texture_ptr, uv_23, ror #16 |
| 2750 | |
| 2751 | uxtah uv_4, texture_ptr, uv_45 |
| 2752 | ldrb pixel_0, [uv_0] |
| 2753 | |
| 2754 | uxtah uv_5, texture_ptr, uv_45, ror #16 |
| 2755 | ldrb pixel_1, [uv_1] |
| 2756 | |
| 2757 | uxtah uv_6, texture_ptr, uv_67 |
| 2758 | ldrb pixel_2, [uv_2] |
| 2759 | |
| 2760 | uxtah uv_7, texture_ptr, uv_67, ror #16 |
| 2761 | ldrb pixel_3, [uv_3] |
| 2762 | |
| 2763 | ldrb pixel_4, [uv_4] |
| 2764 | add pixel_0, pixel_0, pixel_0 |
| 2765 | |
| 2766 | ldrb pixel_5, [uv_5] |
| 2767 | add pixel_1, pixel_1, pixel_1 |
| 2768 | |
| 2769 | ldrb pixel_6, [uv_6] |
| 2770 | add pixel_2, pixel_2, pixel_2 |
| 2771 | |
| 2772 | ldrb pixel_7, [uv_7] |
| 2773 | add pixel_3, pixel_3, pixel_3 |
| 2774 | |
| 2775 | ldrh pixel_0, [clut_ptr, pixel_0] |
| 2776 | add pixel_4, pixel_4, pixel_4 |
| 2777 | |
| 2778 | ldrh pixel_1, [clut_ptr, pixel_1] |
| 2779 | add pixel_5, pixel_5, pixel_5 |
| 2780 | |
| 2781 | ldrh pixel_2, [clut_ptr, pixel_2] |
| 2782 | add pixel_6, pixel_6, pixel_6 |
| 2783 | |
| 2784 | ldrh pixel_3, [clut_ptr, pixel_3] |
| 2785 | add pixel_7, pixel_7, pixel_7 |
| 2786 | |
| 2787 | ldrh pixel_4, [clut_ptr, pixel_4] |
| 2788 | orr pixels_a, pixel_0, pixel_1, lsl #16 |
| 2789 | |
| 2790 | ldrh pixel_5, [clut_ptr, pixel_5] |
| 2791 | orr pixels_c, pixel_2, pixel_3, lsl #16 |
| 2792 | |
| 2793 | ldrh pixel_6, [clut_ptr, pixel_6] |
| 2794 | subs num_blocks, num_blocks, #1 |
| 2795 | |
| 2796 | ldrh pixel_7, [clut_ptr, pixel_7] |
| 2797 | orr pixels_b, pixel_4, pixel_5, lsl #16 |
| 2798 | |
| 2799 | orr pixels_d, pixel_6, pixel_7, lsl #16 |
| 2800 | stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d } |
| 2801 | |
| 2802 | add block_ptr, block_ptr, #64 |
| 2803 | bne 0b |
| 2804 | |
| 2805 | ldmia sp!, { r3 - r11, pc } |
| 2806 | |
| 2807 | 1: |
| 2808 | stmdb sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 } |
| 2809 | |
| 2810 | bl update_texture_8bpp_cache |
| 2811 | |
| 2812 | ldmia sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 } |
| 2813 | bal 0b |
| 2814 | |
| 2815 | |
| 2816 | #undef uv_0 |
| 2817 | #undef uv_1 |
| 2818 | #undef uv_2 |
| 2819 | #undef uv_3 |
| 2820 | #undef uv_4 |
| 2821 | #undef uv_5 |
| 2822 | #undef uv_6 |
| 2823 | #undef uv_7 |
| 2824 | |
| 2825 | #undef pixel_0 |
| 2826 | #undef pixel_1 |
| 2827 | #undef pixel_2 |
| 2828 | #undef pixel_3 |
| 2829 | #undef pixel_4 |
| 2830 | #undef pixel_5 |
| 2831 | #undef pixel_6 |
| 2832 | #undef pixel_7 |
| 2833 | |
| 2834 | #undef texture_ptr |
| 2835 | |
| 2836 | #undef pixels_a |
| 2837 | #undef pixels_b |
| 2838 | #undef pixels_c |
| 2839 | #undef pixels_d |
| 2840 | |
| 2841 | #define psx_gpu r0 |
| 2842 | #define block_ptr r1 |
| 2843 | #define num_blocks r2 |
| 2844 | |
| 2845 | #define uv_0 r3 |
| 2846 | #define uv_1 r4 |
| 2847 | #define u_0 r3 |
| 2848 | #define u_1 r4 |
| 2849 | #define v_0 r5 |
| 2850 | #define v_1 r6 |
| 2851 | |
| 2852 | #define uv_2 r5 |
| 2853 | #define uv_3 r6 |
| 2854 | #define u_2 r5 |
| 2855 | #define u_3 r6 |
| 2856 | #define v_2 r7 |
| 2857 | #define v_3 r8 |
| 2858 | |
| 2859 | #define uv_4 r7 |
| 2860 | #define uv_5 r8 |
| 2861 | #define u_4 r7 |
| 2862 | #define u_5 r8 |
| 2863 | #define v_4 r9 |
| 2864 | #define v_5 r10 |
| 2865 | |
| 2866 | #define uv_6 r9 |
| 2867 | #define uv_7 r10 |
| 2868 | #define u_6 r9 |
| 2869 | #define u_7 r10 |
| 2870 | #define v_6 r11 |
| 2871 | #define v_7 r0 |
| 2872 | |
| 2873 | #define pixel_0 r3 |
| 2874 | #define pixel_1 r4 |
| 2875 | #define pixel_2 r5 |
| 2876 | #define pixel_3 r6 |
| 2877 | #define pixel_4 r7 |
| 2878 | #define pixel_5 r8 |
| 2879 | #define pixel_6 r9 |
| 2880 | #define pixel_7 r10 |
| 2881 | |
| 2882 | #define pixels_a r3 |
| 2883 | #define pixels_b r5 |
| 2884 | #define pixels_c r7 |
| 2885 | #define pixels_d r9 |
| 2886 | |
| 2887 | #define texture_ptr r12 |
| 2888 | |
| 2889 | |
| 2890 | .align 3 |
| 2891 | |
| 2892 | function(texture_blocks_16bpp) |
| 2893 | stmdb sp!, { r3 - r11, r14 } |
| 2894 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 2895 | |
| 2896 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 2897 | ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] |
| 2898 | |
| 2899 | 0: |
| 2900 | ldrh uv_0, [block_ptr] |
| 2901 | subs num_blocks, num_blocks, #1 |
| 2902 | |
| 2903 | ldrh uv_1, [block_ptr, #2] |
| 2904 | |
| 2905 | and v_0, uv_0, #0xFF00 |
| 2906 | and v_1, uv_1, #0xFF00 |
| 2907 | |
| 2908 | and u_0, uv_0, #0xFF |
| 2909 | and u_1, uv_1, #0xFF |
| 2910 | |
| 2911 | add uv_0, u_0, v_0, lsl #2 |
| 2912 | ldrh uv_2, [block_ptr, #4] |
| 2913 | |
| 2914 | add uv_1, u_1, v_1, lsl #2 |
| 2915 | ldrh uv_3, [block_ptr, #6] |
| 2916 | |
| 2917 | add uv_0, uv_0, uv_0 |
| 2918 | add uv_1, uv_1, uv_1 |
| 2919 | |
| 2920 | and v_2, uv_2, #0xFF00 |
| 2921 | and v_3, uv_3, #0xFF00 |
| 2922 | |
| 2923 | and u_2, uv_2, #0xFF |
| 2924 | and u_3, uv_3, #0xFF |
| 2925 | |
| 2926 | add uv_2, u_2, v_2, lsl #2 |
| 2927 | ldrh uv_4, [block_ptr, #8] |
| 2928 | |
| 2929 | add uv_3, u_3, v_3, lsl #2 |
| 2930 | ldrh uv_5, [block_ptr, #10] |
| 2931 | |
| 2932 | add uv_2, uv_2, uv_2 |
| 2933 | add uv_3, uv_3, uv_3 |
| 2934 | |
| 2935 | and v_4, uv_4, #0xFF00 |
| 2936 | and v_5, uv_5, #0xFF00 |
| 2937 | |
| 2938 | and u_4, uv_4, #0xFF |
| 2939 | and u_5, uv_5, #0xFF |
| 2940 | |
| 2941 | add uv_4, u_4, v_4, lsl #2 |
| 2942 | ldrh uv_6, [block_ptr, #12] |
| 2943 | |
| 2944 | add uv_5, u_5, v_5, lsl #2 |
| 2945 | ldrh uv_7, [block_ptr, #14] |
| 2946 | |
| 2947 | add uv_4, uv_4, uv_4 |
| 2948 | ldrh pixel_0, [texture_ptr, uv_0] |
| 2949 | |
| 2950 | add uv_5, uv_5, uv_5 |
| 2951 | ldrh pixel_1, [texture_ptr, uv_1] |
| 2952 | |
| 2953 | and v_6, uv_6, #0xFF00 |
| 2954 | ldrh pixel_2, [texture_ptr, uv_2] |
| 2955 | |
| 2956 | and v_7, uv_7, #0xFF00 |
| 2957 | ldrh pixel_3, [texture_ptr, uv_3] |
| 2958 | |
| 2959 | and u_6, uv_6, #0xFF |
| 2960 | ldrh pixel_4, [texture_ptr, uv_4] |
| 2961 | |
| 2962 | and u_7, uv_7, #0xFF |
| 2963 | ldrh pixel_5, [texture_ptr, uv_5] |
| 2964 | |
| 2965 | add uv_6, u_6, v_6, lsl #2 |
| 2966 | add uv_7, u_7, v_7, lsl #2 |
| 2967 | |
| 2968 | add uv_6, uv_6, uv_6 |
| 2969 | add uv_7, uv_7, uv_7 |
| 2970 | |
| 2971 | orr pixels_a, pixel_0, pixel_1, lsl #16 |
| 2972 | orr pixels_b, pixel_2, pixel_3, lsl #16 |
| 2973 | |
| 2974 | ldrh pixel_6, [texture_ptr, uv_6] |
| 2975 | orr pixels_c, pixel_4, pixel_5, lsl #16 |
| 2976 | |
| 2977 | ldrh pixel_7, [texture_ptr, uv_7] |
| 2978 | orr pixels_d, pixel_6, pixel_7, lsl #16 |
| 2979 | |
| 2980 | stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d } |
| 2981 | add block_ptr, block_ptr, #64 |
| 2982 | |
| 2983 | bne 0b |
| 2984 | |
| 2985 | ldmia sp!, { r3 - r11, pc } |
| 2986 | |
| 2987 | |
| 2988 | #undef num_blocks |
| 2989 | |
| 2990 | #undef test_mask |
| 2991 | #undef texels |
| 2992 | #undef pixels_b |
| 2993 | #undef pixels |
| 2994 | #undef d64_1 |
| 2995 | #undef d64_4 |
| 2996 | #undef d64_128 |
| 2997 | #undef draw_mask |
| 2998 | #undef msb_mask |
| 2999 | #undef msb_mask_low |
| 3000 | #undef msb_mask_high |
| 3001 | #undef fb_pixels |
| 3002 | |
| 3003 | #undef c_32 |
| 3004 | #undef fb_ptr |
| 3005 | #undef mask_msb_ptr |
| 3006 | |
| 3007 | #define psx_gpu r0 |
| 3008 | #define num_blocks r1 |
| 3009 | #define color_ptr r2 |
| 3010 | #define colors_scalar r2 |
| 3011 | #define colors_scalar_compare r3 |
| 3012 | #define mask_msb_ptr r2 |
| 3013 | |
| 3014 | #define block_ptr_load_a r0 |
| 3015 | #define block_ptr_store r3 |
| 3016 | #define block_ptr_load_b r12 |
| 3017 | #define c_32 r2 |
| 3018 | |
| 3019 | #define c_48 r4 |
| 3020 | #define fb_ptr r14 |
| 3021 | #define draw_mask_bits_scalar r5 |
| 3022 | |
| 3023 | #define d128_0x07 q0 |
| 3024 | #define d128_0x1F q1 |
| 3025 | #define d128_0x8000 q2 |
| 3026 | #define test_mask q3 |
| 3027 | #define texels q4 |
| 3028 | #define colors_rg q5 |
| 3029 | #define colors_b_dm_bits q6 |
| 3030 | #define texels_rg q7 |
| 3031 | #define pixels_r q8 |
| 3032 | #define pixels_g q9 |
| 3033 | #define pixels_b q10 |
| 3034 | #define pixels q11 |
| 3035 | #define zero_mask q4 |
| 3036 | #define draw_mask q12 |
| 3037 | #define msb_mask q13 |
| 3038 | |
| 3039 | #define fb_pixels q8 |
| 3040 | |
| 3041 | #define pixels_gb_low q9 |
| 3042 | |
| 3043 | #define colors_r d10 |
| 3044 | #define colors_g d11 |
| 3045 | #define colors_b d12 |
| 3046 | #define draw_mask_bits d13 |
| 3047 | #define texels_r d14 |
| 3048 | #define texels_g d15 |
| 3049 | #define pixels_r_low d16 |
| 3050 | #define pixels_g_low d18 |
| 3051 | #define pixels_b_low d19 |
| 3052 | #define msb_mask_low d26 |
| 3053 | #define msb_mask_high d27 |
| 3054 | |
| 3055 | #define d64_1 d28 |
| 3056 | #define d64_4 d29 |
| 3057 | #define d64_128 d30 |
| 3058 | #define texels_b d31 |
| 3059 | |
| 3060 | #define shade_blocks_textured_modulated_prologue_indirect() \ |
| 3061 | mov c_48, #48; \ |
| 3062 | add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \ |
| 3063 | |
| 3064 | #define shade_blocks_textured_modulated_prologue_direct() \ |
| 3065 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3066 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] \ |
| 3067 | |
| 3068 | |
| 3069 | #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \ |
| 3070 | |
| 3071 | #define shade_blocks_textured_false_modulation_check_undithered(target) \ |
| 3072 | ldr colors_scalar, [psx_gpu, #psx_gpu_triangle_color_offset]; \ |
| 3073 | movw colors_scalar_compare, #0x8080; \ |
| 3074 | \ |
| 3075 | movt colors_scalar_compare, #0x80; \ |
| 3076 | cmp colors_scalar, colors_scalar_compare; \ |
| 3077 | beq shade_blocks_textured_unmodulated_##target \ |
| 3078 | |
| 3079 | #define shade_blocks_textured_false_modulation_check_dithered(target) \ |
| 3080 | |
| 3081 | #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \ |
| 3082 | shade_blocks_textured_false_modulation_check_##dithering(target); \ |
| 3083 | add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \ |
| 3084 | vld1.u32 { colors_r[] }, [color_ptr, :32]; \ |
| 3085 | vdup.u8 colors_g, colors_r[1]; \ |
| 3086 | vdup.u8 colors_b, colors_r[2]; \ |
| 3087 | vdup.u8 colors_r, colors_r[0] \ |
| 3088 | |
| 3089 | |
| 3090 | #define shade_blocks_textured_modulated_load_dithered(target) \ |
| 3091 | vld1.u32 { target }, [block_ptr_load_b, :128] \ |
| 3092 | |
| 3093 | #define shade_blocks_textured_modulated_load_last_dithered(target) \ |
| 3094 | vld1.u32 { target }, [block_ptr_load_b, :128], c_32 \ |
| 3095 | |
| 3096 | #define shade_blocks_textured_modulated_load_undithered(target) \ |
| 3097 | |
| 3098 | #define shade_blocks_textured_modulated_load_last_undithered(target) \ |
| 3099 | add block_ptr_load_b, block_ptr_load_b, #32 \ |
| 3100 | |
| 3101 | #define shade_blocks_textured_modulate_dithered(channel) \ |
| 3102 | vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \ |
| 3103 | |
| 3104 | #define shade_blocks_textured_modulate_undithered(channel) \ |
| 3105 | vmull.u8 pixels_##channel, texels_##channel, colors_##channel \ |
| 3106 | |
| 3107 | |
| 3108 | #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \ |
| 3109 | vst1.u32 { draw_mask }, [block_ptr_store, :128]! \ |
| 3110 | |
| 3111 | #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \ |
| 3112 | ldr fb_ptr, [block_ptr_load_b, #(offset - 64)]; \ |
| 3113 | vld1.u32 { fb_pixels }, [fb_ptr]; \ |
| 3114 | vbit.u16 pixels, fb_pixels, draw_mask \ |
| 3115 | |
| 3116 | #define shade_blocks_textured_modulated_store_pixels_indirect() \ |
| 3117 | vst1.u32 { pixels }, [block_ptr_store, :128], c_48 \ |
| 3118 | |
| 3119 | #define shade_blocks_textured_modulated_store_pixels_direct() \ |
| 3120 | vst1.u32 { pixels }, [fb_ptr] \ |
| 3121 | |
| 3122 | |
| 3123 | #define shade_blocks_textured_modulated_load_rg_shaded() \ |
| 3124 | vld1.u32 { colors_r, colors_g }, [block_ptr_load_b, :128], c_32 \ |
| 3125 | |
| 3126 | #define shade_blocks_textured_modulated_load_rg_unshaded() \ |
| 3127 | add block_ptr_load_b, block_ptr_load_b, #32 \ |
| 3128 | |
| 3129 | #define shade_blocks_textured_modulated_load_bdm_shaded() \ |
| 3130 | vld1.u32 { colors_b, draw_mask_bits }, [block_ptr_load_a, :128], c_32 \ |
| 3131 | |
| 3132 | #define shade_blocks_textured_modulated_load_bdm_unshaded() \ |
| 3133 | ldr draw_mask_bits_scalar, [block_ptr_load_a, #8]; \ |
| 3134 | add block_ptr_load_a, block_ptr_load_a, #32 \ |
| 3135 | |
| 3136 | #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \ |
| 3137 | vdup.u16 draw_mask, draw_mask_bits[0] \ |
| 3138 | |
| 3139 | #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \ |
| 3140 | vdup.u16 draw_mask, draw_mask_bits_scalar \ |
| 3141 | |
| 3142 | |
| 3143 | #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \ |
| 3144 | |
| 3145 | #define shade_blocks_textured_modulated_apply_msb_mask_direct() \ |
| 3146 | vorr.u16 pixels, pixels, msb_mask \ |
| 3147 | |
| 3148 | |
| 3149 | #define shade_blocks_textured_modulated_builder(shading, dithering, target) \ |
| 3150 | .align 3; \ |
| 3151 | \ |
| 3152 | function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \ |
| 3153 | shade_blocks_textured_modulated_prologue_##shading(dithering, target); \ |
| 3154 | stmdb sp!, { r4 - r5, lr }; \ |
| 3155 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 3156 | \ |
| 3157 | vld1.u32 { test_mask }, [psx_gpu, :128]; \ |
| 3158 | \ |
| 3159 | shade_blocks_textured_modulated_prologue_##target(); \ |
| 3160 | \ |
| 3161 | add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3162 | mov c_32, #32; \ |
| 3163 | \ |
| 3164 | add block_ptr_load_b, block_ptr_load_a, #16; \ |
| 3165 | vmov.u8 d64_1, #1; \ |
| 3166 | vmov.u8 d64_4, #4; \ |
| 3167 | vmov.u8 d64_128, #128; \ |
| 3168 | \ |
| 3169 | vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \ |
| 3170 | vmov.u8 d128_0x07, #0x07; \ |
| 3171 | \ |
| 3172 | shade_blocks_textured_modulated_load_rg_##shading(); \ |
| 3173 | vmov.u8 d128_0x1F, #0x1F; \ |
| 3174 | \ |
| 3175 | shade_blocks_textured_modulated_load_bdm_##shading(); \ |
| 3176 | vmov.u16 d128_0x8000, #0x8000; \ |
| 3177 | \ |
| 3178 | vmovn.u16 texels_r, texels; \ |
| 3179 | vshrn.u16 texels_g, texels, #5; \ |
| 3180 | \ |
| 3181 | vshrn.u16 texels_b, texels, #7; \ |
| 3182 | shade_blocks_textured_modulated_expand_draw_mask_##shading(); \ |
| 3183 | \ |
| 3184 | shade_blocks_textured_modulated_load_##dithering(pixels_r); \ |
| 3185 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 3186 | \ |
| 3187 | shade_blocks_textured_modulated_load_##dithering(pixels_g); \ |
| 3188 | vand.u8 texels_rg, texels_rg, d128_0x1F; \ |
| 3189 | \ |
| 3190 | shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \ |
| 3191 | vshr.u8 texels_b, texels_b, #3; \ |
| 3192 | \ |
| 3193 | shade_blocks_textured_modulate_##dithering(r); \ |
| 3194 | shade_blocks_textured_modulate_##dithering(g); \ |
| 3195 | shade_blocks_textured_modulate_##dithering(b); \ |
| 3196 | \ |
| 3197 | vand.u16 pixels, texels, d128_0x8000; \ |
| 3198 | vceq.u16 zero_mask, texels, #0; \ |
| 3199 | \ |
| 3200 | vqshrun.s16 pixels_r_low, pixels_r, #4; \ |
| 3201 | vqshrun.s16 pixels_g_low, pixels_g, #4; \ |
| 3202 | vqshrun.s16 pixels_b_low, pixels_b, #4; \ |
| 3203 | \ |
| 3204 | shade_blocks_textured_modulated_apply_msb_mask_##target(); \ |
| 3205 | vorr.u16 draw_mask, draw_mask, zero_mask; \ |
| 3206 | vshr.u8 pixels_r_low, pixels_r_low, #3; \ |
| 3207 | vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \ |
| 3208 | \ |
| 3209 | subs num_blocks, num_blocks, #1; \ |
| 3210 | beq 1f; \ |
| 3211 | \ |
| 3212 | .align 3; \ |
| 3213 | \ |
| 3214 | 0: \ |
| 3215 | vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \ |
| 3216 | shade_blocks_textured_modulated_load_rg_##shading(); \ |
| 3217 | vshrn.u16 texels_g, texels, #5; \ |
| 3218 | \ |
| 3219 | shade_blocks_textured_modulated_load_bdm_##shading(); \ |
| 3220 | vshrn.u16 texels_b, texels, #7; \ |
| 3221 | \ |
| 3222 | pld [block_ptr_load_a]; \ |
| 3223 | vmovn.u16 texels_r, texels; \ |
| 3224 | vmlal.u8 pixels, pixels_r_low, d64_1; \ |
| 3225 | \ |
| 3226 | vmlal.u8 pixels, pixels_g_low, d64_4; \ |
| 3227 | vmlal.u8 pixels, pixels_b_low, d64_128; \ |
| 3228 | shade_blocks_textured_modulated_store_draw_mask_##target(-4); \ |
| 3229 | \ |
| 3230 | shade_blocks_textured_modulated_load_##dithering(pixels_r); \ |
| 3231 | shade_blocks_textured_modulated_expand_draw_mask_##shading(); \ |
| 3232 | \ |
| 3233 | shade_blocks_textured_modulated_load_##dithering(pixels_g); \ |
| 3234 | vand.u8 texels_rg, texels_rg, d128_0x1F; \ |
| 3235 | \ |
| 3236 | shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \ |
| 3237 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 3238 | \ |
| 3239 | shade_blocks_textured_modulated_store_pixels_##target(); \ |
| 3240 | vshr.u8 texels_b, texels_b, #3; \ |
| 3241 | \ |
| 3242 | shade_blocks_textured_modulate_##dithering(r); \ |
| 3243 | shade_blocks_textured_modulate_##dithering(g); \ |
| 3244 | shade_blocks_textured_modulate_##dithering(b); \ |
| 3245 | \ |
| 3246 | vand.u16 pixels, texels, d128_0x8000; \ |
| 3247 | vceq.u16 zero_mask, texels, #0; \ |
| 3248 | \ |
| 3249 | subs num_blocks, num_blocks, #1; \ |
| 3250 | \ |
| 3251 | vqshrun.s16 pixels_r_low, pixels_r, #4; \ |
| 3252 | vqshrun.s16 pixels_g_low, pixels_g, #4; \ |
| 3253 | vqshrun.s16 pixels_b_low, pixels_b, #4; \ |
| 3254 | \ |
| 3255 | shade_blocks_textured_modulated_apply_msb_mask_##target(); \ |
| 3256 | vorr.u16 draw_mask, draw_mask, zero_mask; \ |
| 3257 | vshr.u8 pixels_r_low, pixels_r_low, #3; \ |
| 3258 | vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \ |
| 3259 | \ |
| 3260 | bne 0b; \ |
| 3261 | \ |
| 3262 | 1: \ |
| 3263 | vmlal.u8 pixels, pixels_r_low, d64_1; \ |
| 3264 | vmlal.u8 pixels, pixels_g_low, d64_4; \ |
| 3265 | vmlal.u8 pixels, pixels_b_low, d64_128; \ |
| 3266 | \ |
| 3267 | shade_blocks_textured_modulated_store_draw_mask_##target(28); \ |
| 3268 | shade_blocks_textured_modulated_store_pixels_##target(); \ |
| 3269 | \ |
| 3270 | ldmia sp!, { r4 - r5, pc } \ |
| 3271 | |
| 3272 | |
| 3273 | shade_blocks_textured_modulated_builder(shaded, dithered, direct); |
| 3274 | shade_blocks_textured_modulated_builder(shaded, undithered, direct); |
| 3275 | shade_blocks_textured_modulated_builder(unshaded, dithered, direct); |
| 3276 | shade_blocks_textured_modulated_builder(unshaded, undithered, direct); |
| 3277 | |
| 3278 | shade_blocks_textured_modulated_builder(shaded, dithered, indirect); |
| 3279 | shade_blocks_textured_modulated_builder(shaded, undithered, indirect); |
| 3280 | shade_blocks_textured_modulated_builder(unshaded, dithered, indirect); |
| 3281 | shade_blocks_textured_modulated_builder(unshaded, undithered, indirect); |
| 3282 | |
| 3283 | |
| 3284 | #undef c_64 |
| 3285 | #undef fb_ptr |
| 3286 | #undef color_ptr |
| 3287 | |
| 3288 | #undef color_r |
| 3289 | #undef color_g |
| 3290 | #undef color_b |
| 3291 | |
| 3292 | #undef test_mask |
| 3293 | #undef pixels |
| 3294 | #undef draw_mask |
| 3295 | #undef zero_mask |
| 3296 | #undef fb_pixels |
| 3297 | #undef msb_mask |
| 3298 | #undef msb_mask_low |
| 3299 | #undef msb_mask_high |
| 3300 | |
| 3301 | #define psx_gpu r0 |
| 3302 | #define num_blocks r1 |
| 3303 | #define mask_msb_ptr r2 |
| 3304 | #define color_ptr r3 |
| 3305 | |
| 3306 | #define block_ptr_load r0 |
| 3307 | #define draw_mask_store_ptr r3 |
| 3308 | #define draw_mask_bits_ptr r12 |
| 3309 | #define draw_mask_ptr r12 |
| 3310 | #define pixel_store_ptr r14 |
| 3311 | |
| 3312 | #define fb_ptr_cmp r4 |
| 3313 | |
| 3314 | #define fb_ptr r3 |
| 3315 | #define fb_ptr_next r14 |
| 3316 | |
| 3317 | #define c_64 r2 |
| 3318 | |
| 3319 | #define test_mask q0 |
| 3320 | #define pixels q1 |
| 3321 | #define draw_mask q2 |
| 3322 | #define zero_mask q3 |
| 3323 | #define draw_mask_combined q4 |
| 3324 | #define fb_pixels q5 |
| 3325 | #define fb_pixels_next q6 |
| 3326 | #define msb_mask q7 |
| 3327 | |
| 3328 | #define draw_mask_low d4 |
| 3329 | #define draw_mask_high d5 |
| 3330 | #define msb_mask_low d14 |
| 3331 | #define msb_mask_high d15 |
| 3332 | |
| 3333 | .align 3 |
| 3334 | function(shade_blocks_textured_unmodulated_indirect) |
| 3335 | str r14, [sp, #-4] |
| 3336 | add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40) |
| 3337 | |
| 3338 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 3339 | add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16) |
| 3340 | |
| 3341 | vld1.u32 { test_mask }, [psx_gpu, :128] |
| 3342 | add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 3343 | |
| 3344 | mov c_64, #64 |
| 3345 | add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset |
| 3346 | |
| 3347 | vld1.u32 { pixels }, [block_ptr_load, :128], c_64 |
| 3348 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3349 | [draw_mask_bits_ptr, :16], c_64 |
| 3350 | vceq.u16 zero_mask, pixels, #0 |
| 3351 | |
| 3352 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3353 | vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64 |
| 3354 | |
| 3355 | subs num_blocks, num_blocks, #1 |
| 3356 | beq 1f |
| 3357 | |
| 3358 | 0: |
| 3359 | vld1.u32 { pixels }, [block_ptr_load, :128], c_64 |
| 3360 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3361 | |
| 3362 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3363 | [draw_mask_bits_ptr, :16], c_64 |
| 3364 | vceq.u16 zero_mask, pixels, #0 |
| 3365 | |
| 3366 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3367 | vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64 |
| 3368 | |
| 3369 | vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64 |
| 3370 | subs num_blocks, num_blocks, #1 |
| 3371 | |
| 3372 | bne 0b |
| 3373 | |
| 3374 | 1: |
| 3375 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3376 | vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64 |
| 3377 | |
| 3378 | ldr pc, [sp, #-4] |
| 3379 | |
| 3380 | |
| 3381 | .align 3 |
| 3382 | |
| 3383 | function(shade_blocks_textured_unmodulated_direct) |
| 3384 | stmdb sp!, { r4, r14 } |
| 3385 | add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40) |
| 3386 | |
| 3387 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 3388 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset |
| 3389 | |
| 3390 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] |
| 3391 | mov c_64, #64 |
| 3392 | |
| 3393 | vld1.u32 { test_mask }, [psx_gpu, :128] |
| 3394 | add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset |
| 3395 | |
| 3396 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3397 | [draw_mask_bits_ptr, :16], c_64 |
| 3398 | ldr fb_ptr_next, [block_ptr_load, #44] |
| 3399 | |
| 3400 | vld1.u32 { pixels }, [block_ptr_load, :128], c_64 |
| 3401 | vld1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3402 | vceq.u16 zero_mask, pixels, #0 |
| 3403 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3404 | |
| 3405 | subs num_blocks, num_blocks, #1 |
| 3406 | beq 1f |
| 3407 | |
| 3408 | 0: |
| 3409 | mov fb_ptr, fb_ptr_next |
| 3410 | ldr fb_ptr_next, [block_ptr_load, #44] |
| 3411 | |
| 3412 | vorr.u16 pixels, pixels, msb_mask |
| 3413 | |
| 3414 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3415 | vmov fb_pixels, fb_pixels_next |
| 3416 | |
| 3417 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3418 | [draw_mask_bits_ptr, :16], c_64 |
| 3419 | vbif.u16 fb_pixels, pixels, draw_mask_combined |
| 3420 | |
| 3421 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr |
| 3422 | pld [fb_ptr_next, #64] |
| 3423 | |
| 3424 | add fb_ptr_cmp, fb_ptr_cmp, #14 |
| 3425 | vld1.u32 { pixels }, [block_ptr_load, :128], c_64 |
| 3426 | |
| 3427 | cmp fb_ptr_cmp, #28 |
| 3428 | bls 4f |
| 3429 | |
| 3430 | vld1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3431 | vceq.u16 zero_mask, pixels, #0 |
| 3432 | |
| 3433 | vst1.u16 { fb_pixels }, [fb_ptr] |
| 3434 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3435 | |
| 3436 | 3: |
| 3437 | subs num_blocks, num_blocks, #1 |
| 3438 | bne 0b |
| 3439 | |
| 3440 | 1: |
| 3441 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3442 | vbif.u16 fb_pixels_next, pixels, draw_mask_combined |
| 3443 | |
| 3444 | vst1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3445 | |
| 3446 | ldmia sp!, { r4, pc } |
| 3447 | |
| 3448 | 4: |
| 3449 | vst1.u16 { fb_pixels }, [fb_ptr] |
| 3450 | vceq.u16 zero_mask, pixels, #0 |
| 3451 | |
| 3452 | vld1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3453 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3454 | |
| 3455 | bal 3b |
| 3456 | |
| 3457 | |
| 3458 | function(shade_blocks_unshaded_untextured_indirect) |
| 3459 | bx lr |
| 3460 | |
| 3461 | .align 3 |
| 3462 | |
| 3463 | function(shade_blocks_unshaded_untextured_direct) |
| 3464 | stmdb sp!, { r4, r14 } |
| 3465 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 3466 | |
| 3467 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 3468 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset |
| 3469 | |
| 3470 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] |
| 3471 | add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16) |
| 3472 | |
| 3473 | add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44) |
| 3474 | vld1.u16 { pixels }, [color_ptr, :128] |
| 3475 | |
| 3476 | mov c_64, #64 |
| 3477 | vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64 |
| 3478 | |
| 3479 | vorr.u16 pixels, pixels, msb_mask |
| 3480 | subs num_blocks, num_blocks, #1 |
| 3481 | |
| 3482 | ldr fb_ptr_next, [block_ptr_load], #64 |
| 3483 | |
| 3484 | vld1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3485 | beq 1f |
| 3486 | |
| 3487 | 0: |
| 3488 | vmov fb_pixels, fb_pixels_next |
| 3489 | mov fb_ptr, fb_ptr_next |
| 3490 | ldr fb_ptr_next, [block_ptr_load], #64 |
| 3491 | |
| 3492 | vbif.u16 fb_pixels, pixels, draw_mask |
| 3493 | vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64 |
| 3494 | |
| 3495 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr |
| 3496 | add fb_ptr_cmp, fb_ptr_cmp, #14 |
| 3497 | cmp fb_ptr_cmp, #28 |
| 3498 | bls 4f |
| 3499 | |
| 3500 | vld1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3501 | vst1.u16 { fb_pixels }, [fb_ptr] |
| 3502 | |
| 3503 | 3: |
| 3504 | subs num_blocks, num_blocks, #1 |
| 3505 | bne 0b |
| 3506 | |
| 3507 | 1: |
| 3508 | vbif.u16 fb_pixels_next, pixels, draw_mask |
| 3509 | vst1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3510 | |
| 3511 | ldmia sp!, { r4, pc } |
| 3512 | |
| 3513 | 4: |
| 3514 | vst1.u16 { fb_pixels }, [fb_ptr] |
| 3515 | vld1.u16 { fb_pixels_next }, [fb_ptr_next] |
| 3516 | bal 3b |
| 3517 | |
| 3518 | |
| 3519 | #undef draw_mask_ptr |
| 3520 | #undef c_64 |
| 3521 | #undef fb_ptr |
| 3522 | #undef fb_ptr_next |
| 3523 | #undef fb_ptr_cmp |
| 3524 | |
| 3525 | #define psx_gpu r0 |
| 3526 | #define num_blocks r1 |
| 3527 | #define msb_mask_ptr r2 |
| 3528 | #define pixel_ptr r3 |
| 3529 | #define draw_mask_ptr r0 |
| 3530 | #define c_64 r2 |
| 3531 | #define fb_ptr r12 |
| 3532 | #define fb_ptr_next r14 |
| 3533 | #define fb_ptr_cmp r4 |
| 3534 | |
| 3535 | #undef msb_mask |
| 3536 | #undef draw_mask |
| 3537 | #undef pixels |
| 3538 | #undef fb_pixels |
| 3539 | #undef d128_0x8000 |
| 3540 | #undef msb_mask_low |
| 3541 | #undef msb_mask_high |
| 3542 | #undef draw_mask_next |
| 3543 | #undef pixels_g |
| 3544 | #undef blend_pixels |
| 3545 | #undef fb_pixels_next |
| 3546 | |
| 3547 | #define msb_mask q0 |
| 3548 | #define draw_mask q1 |
| 3549 | #define pixels q2 |
| 3550 | #define fb_pixels q3 |
| 3551 | #define blend_pixels q4 |
| 3552 | #define pixels_no_msb q5 |
| 3553 | #define blend_mask q6 |
| 3554 | #define fb_pixels_no_msb q7 |
| 3555 | #define d128_0x8000 q8 |
| 3556 | #define d128_0x0421 q9 |
| 3557 | #define fb_pixels_next q10 |
| 3558 | #define blend_pixels_next q11 |
| 3559 | #define pixels_next q12 |
| 3560 | #define draw_mask_next q13 |
| 3561 | #define write_mask q14 |
| 3562 | |
| 3563 | #define pixels_rb q5 |
| 3564 | #define pixels_mg q7 |
| 3565 | #define pixels_g q7 |
| 3566 | #define d128_0x7C1F q8 |
| 3567 | #define d128_0x03E0 q9 |
| 3568 | #define fb_pixels_rb q10 |
| 3569 | #define fb_pixels_g q11 |
| 3570 | #define fb_pixels_masked q11 |
| 3571 | #define d128_0x83E0 q15 |
| 3572 | #define pixels_fourth q7 |
| 3573 | #define d128_0x1C07 q12 |
| 3574 | #define d128_0x00E0 q13 |
| 3575 | #define d128_0x80E0 q13 |
| 3576 | |
| 3577 | #define msb_mask_low d0 |
| 3578 | #define msb_mask_high d1 |
| 3579 | |
| 3580 | #define blend_blocks_average_set_blend_mask_textured(source) \ |
| 3581 | vclt.s16 blend_mask, source, #0 \ |
| 3582 | |
| 3583 | #define blend_blocks_average_set_stp_bit_textured() \ |
| 3584 | vorr.u16 blend_pixels, #0x8000 \ |
| 3585 | |
| 3586 | #define blend_blocks_average_combine_textured(source) \ |
| 3587 | vbif.u16 blend_pixels, source, blend_mask \ |
| 3588 | |
| 3589 | #define blend_blocks_average_set_blend_mask_untextured(source) \ |
| 3590 | |
| 3591 | #define blend_blocks_average_set_stp_bit_untextured() \ |
| 3592 | |
| 3593 | #define blend_blocks_average_combine_untextured(source) \ |
| 3594 | |
| 3595 | #define blend_blocks_average_mask_set_on() \ |
| 3596 | vclt.s16 write_mask, fb_pixels_next, #0 \ |
| 3597 | |
| 3598 | #define blend_blocks_average_mask_copy_on() \ |
| 3599 | vorr.u16 draw_mask, draw_mask_next, write_mask \ |
| 3600 | |
| 3601 | #define blend_blocks_average_mask_copy_b_on() \ |
| 3602 | vorr.u16 draw_mask_next, draw_mask_next, write_mask \ |
| 3603 | |
| 3604 | #define blend_blocks_average_mask_set_off() \ |
| 3605 | |
| 3606 | #define blend_blocks_average_mask_copy_off() \ |
| 3607 | vmov draw_mask, draw_mask_next \ |
| 3608 | |
| 3609 | #define blend_blocks_average_mask_copy_b_off() \ |
| 3610 | |
| 3611 | #define blend_blocks_average_builder(texturing, mask_evaluate) \ |
| 3612 | .align 3; \ |
| 3613 | \ |
| 3614 | function(blend_blocks_##texturing##_average_##mask_evaluate) \ |
| 3615 | stmdb sp!, { r4, r14 }; \ |
| 3616 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3617 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 3618 | \ |
| 3619 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3620 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 3621 | \ |
| 3622 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3623 | mov c_64, #64; \ |
| 3624 | \ |
| 3625 | vmov.u16 d128_0x8000, #0x8000; \ |
| 3626 | vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \ |
| 3627 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3628 | \ |
| 3629 | vmov.u16 d128_0x0421, #0x0400; \ |
| 3630 | vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \ |
| 3631 | \ |
| 3632 | vorr.u16 d128_0x0421, #0x0021; \ |
| 3633 | vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \ |
| 3634 | \ |
| 3635 | veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \ |
| 3636 | vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \ |
| 3637 | vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \ |
| 3638 | vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \ |
| 3639 | blend_blocks_average_mask_set_##mask_evaluate(); \ |
| 3640 | vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \ |
| 3641 | \ |
| 3642 | subs num_blocks, num_blocks, #1; \ |
| 3643 | beq 1f; \ |
| 3644 | \ |
| 3645 | 0: \ |
| 3646 | mov fb_ptr, fb_ptr_next; \ |
| 3647 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3648 | \ |
| 3649 | vmov pixels, pixels_next; \ |
| 3650 | vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \ |
| 3651 | \ |
| 3652 | vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \ |
| 3653 | \ |
| 3654 | blend_blocks_average_mask_copy_##mask_evaluate(); \ |
| 3655 | vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \ |
| 3656 | \ |
| 3657 | blend_blocks_average_set_blend_mask_##texturing(pixels); \ |
| 3658 | blend_blocks_average_set_stp_bit_##texturing(); \ |
| 3659 | vmov fb_pixels, fb_pixels_next; \ |
| 3660 | blend_blocks_average_combine_##texturing(pixels); \ |
| 3661 | \ |
| 3662 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3663 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 3664 | cmp fb_ptr_cmp, #28; \ |
| 3665 | bls 2f; \ |
| 3666 | \ |
| 3667 | vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \ |
| 3668 | veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \ |
| 3669 | \ |
| 3670 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3671 | vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \ |
| 3672 | \ |
| 3673 | vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \ |
| 3674 | vbif.u16 fb_pixels, blend_pixels, draw_mask; \ |
| 3675 | \ |
| 3676 | vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \ |
| 3677 | vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \ |
| 3678 | blend_blocks_average_mask_set_##mask_evaluate(); \ |
| 3679 | vst1.u16 { fb_pixels }, [fb_ptr]; \ |
| 3680 | \ |
| 3681 | 3: \ |
| 3682 | subs num_blocks, num_blocks, #1; \ |
| 3683 | bne 0b; \ |
| 3684 | \ |
| 3685 | 1: \ |
| 3686 | blend_blocks_average_mask_copy_b_##mask_evaluate(); \ |
| 3687 | vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \ |
| 3688 | \ |
| 3689 | blend_blocks_average_set_blend_mask_##texturing(pixels_next); \ |
| 3690 | blend_blocks_average_set_stp_bit_##texturing(); \ |
| 3691 | blend_blocks_average_combine_##texturing(pixels_next); \ |
| 3692 | \ |
| 3693 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3694 | vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \ |
| 3695 | vst1.u16 { fb_pixels_next }, [fb_ptr_next]; \ |
| 3696 | \ |
| 3697 | ldmia sp!, { r4, pc }; \ |
| 3698 | \ |
| 3699 | 2: \ |
| 3700 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3701 | vbif.u16 fb_pixels, blend_pixels, draw_mask; \ |
| 3702 | vst1.u16 { fb_pixels }, [fb_ptr]; \ |
| 3703 | \ |
| 3704 | vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \ |
| 3705 | veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \ |
| 3706 | vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \ |
| 3707 | vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \ |
| 3708 | vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \ |
| 3709 | vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \ |
| 3710 | \ |
| 3711 | bal 3b \ |
| 3712 | |
| 3713 | blend_blocks_average_builder(textured, off) |
| 3714 | blend_blocks_average_builder(untextured, off) |
| 3715 | blend_blocks_average_builder(textured, on) |
| 3716 | blend_blocks_average_builder(untextured, on) |
| 3717 | |
| 3718 | |
| 3719 | #define blend_blocks_add_mask_set_on() \ |
| 3720 | vclt.s16 write_mask, fb_pixels, #0 \ |
| 3721 | |
| 3722 | #define blend_blocks_add_mask_copy_on() \ |
| 3723 | vorr.u16 draw_mask, draw_mask, write_mask \ |
| 3724 | |
| 3725 | #define blend_blocks_add_mask_set_off() \ |
| 3726 | |
| 3727 | #define blend_blocks_add_mask_copy_off() \ |
| 3728 | |
| 3729 | |
| 3730 | #define blend_blocks_add_textured_builder(mask_evaluate) \ |
| 3731 | .align 3; \ |
| 3732 | \ |
| 3733 | function(blend_blocks_textured_add_##mask_evaluate) \ |
| 3734 | stmdb sp!, { r4, r14 }; \ |
| 3735 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3736 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 3737 | \ |
| 3738 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3739 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 3740 | \ |
| 3741 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3742 | mov c_64, #64; \ |
| 3743 | \ |
| 3744 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 3745 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 3746 | vmov.u16 d128_0x83E0, #0x8000; \ |
| 3747 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 3748 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 3749 | vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \ |
| 3750 | \ |
| 3751 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 3752 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3753 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 3754 | vclt.s16 blend_mask, pixels, #0; \ |
| 3755 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3756 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3757 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3758 | \ |
| 3759 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3760 | vorr.u16 pixels, pixels, msb_mask; \ |
| 3761 | vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \ |
| 3762 | vand.u16 pixels_mg, pixels, d128_0x83E0; \ |
| 3763 | vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \ |
| 3764 | vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \ |
| 3765 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3766 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \ |
| 3767 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3768 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \ |
| 3769 | \ |
| 3770 | subs num_blocks, num_blocks, #1; \ |
| 3771 | beq 1f; \ |
| 3772 | \ |
| 3773 | 0: \ |
| 3774 | mov fb_ptr, fb_ptr_next; \ |
| 3775 | \ |
| 3776 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3777 | \ |
| 3778 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 3779 | vclt.s16 blend_mask, pixels, #0; \ |
| 3780 | \ |
| 3781 | vorr.u16 pixels, pixels, msb_mask; \ |
| 3782 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3783 | vand.u16 pixels_mg, pixels, d128_0x83E0; \ |
| 3784 | \ |
| 3785 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3786 | pld [fb_ptr_next, #64]; \ |
| 3787 | \ |
| 3788 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3789 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3790 | \ |
| 3791 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 3792 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 3793 | \ |
| 3794 | cmp fb_ptr_cmp, #28; \ |
| 3795 | bls 2f; \ |
| 3796 | \ |
| 3797 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3798 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3799 | vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \ |
| 3800 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3801 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3802 | vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \ |
| 3803 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 3804 | \ |
| 3805 | 3: \ |
| 3806 | vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \ |
| 3807 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3808 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \ |
| 3809 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3810 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \ |
| 3811 | \ |
| 3812 | subs num_blocks, num_blocks, #1; \ |
| 3813 | bne 0b; \ |
| 3814 | \ |
| 3815 | 1: \ |
| 3816 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3817 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3818 | vst1.u16 { blend_pixels }, [fb_ptr_next]; \ |
| 3819 | \ |
| 3820 | ldmia sp!, { r4, pc }; \ |
| 3821 | \ |
| 3822 | 2: \ |
| 3823 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 3824 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3825 | \ |
| 3826 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3827 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3828 | vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \ |
| 3829 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3830 | vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \ |
| 3831 | bal 3b \ |
| 3832 | |
| 3833 | |
| 3834 | #define blend_blocks_add_untextured_builder(mask_evaluate) \ |
| 3835 | .align 3; \ |
| 3836 | \ |
| 3837 | function(blend_blocks_untextured_add_##mask_evaluate) \ |
| 3838 | stmdb sp!, { r4, r14 }; \ |
| 3839 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3840 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 3841 | \ |
| 3842 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3843 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 3844 | \ |
| 3845 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3846 | mov c_64, #64; \ |
| 3847 | \ |
| 3848 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 3849 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 3850 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 3851 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 3852 | \ |
| 3853 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 3854 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3855 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 3856 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3857 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3858 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3859 | \ |
| 3860 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3861 | vand.u16 pixels_g, pixels, d128_0x03E0; \ |
| 3862 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3863 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 3864 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3865 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 3866 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3867 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 3868 | \ |
| 3869 | subs num_blocks, num_blocks, #1; \ |
| 3870 | beq 1f; \ |
| 3871 | \ |
| 3872 | 0: \ |
| 3873 | mov fb_ptr, fb_ptr_next; \ |
| 3874 | \ |
| 3875 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3876 | \ |
| 3877 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 3878 | \ |
| 3879 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3880 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3881 | vand.u16 pixels_g, pixels, d128_0x03E0; \ |
| 3882 | \ |
| 3883 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3884 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 3885 | \ |
| 3886 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3887 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 3888 | cmp fb_ptr_cmp, #28; \ |
| 3889 | bls 2f; \ |
| 3890 | \ |
| 3891 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3892 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3893 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3894 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3895 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3896 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 3897 | \ |
| 3898 | 3: \ |
| 3899 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 3900 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3901 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 3902 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3903 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 3904 | \ |
| 3905 | subs num_blocks, num_blocks, #1; \ |
| 3906 | bne 0b; \ |
| 3907 | \ |
| 3908 | 1: \ |
| 3909 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3910 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3911 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3912 | vst1.u16 { blend_pixels }, [fb_ptr_next]; \ |
| 3913 | \ |
| 3914 | ldmia sp!, { r4, pc }; \ |
| 3915 | \ |
| 3916 | 2: \ |
| 3917 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 3918 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3919 | \ |
| 3920 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3921 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3922 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3923 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3924 | bal 3b \ |
| 3925 | |
| 3926 | |
| 3927 | blend_blocks_add_textured_builder(off) |
| 3928 | blend_blocks_add_textured_builder(on) |
| 3929 | blend_blocks_add_untextured_builder(off) |
| 3930 | blend_blocks_add_untextured_builder(on) |
| 3931 | |
| 3932 | #define blend_blocks_subtract_set_blend_mask_textured() \ |
| 3933 | vclt.s16 blend_mask, pixels_next, #0 \ |
| 3934 | |
| 3935 | #define blend_blocks_subtract_combine_textured() \ |
| 3936 | vbif.u16 blend_pixels, pixels, blend_mask \ |
| 3937 | |
| 3938 | #define blend_blocks_subtract_set_stb_textured() \ |
| 3939 | vorr.u16 blend_pixels, #0x8000 \ |
| 3940 | |
| 3941 | #define blend_blocks_subtract_msb_mask_textured() \ |
| 3942 | vorr.u16 pixels, pixels_next, msb_mask \ |
| 3943 | |
| 3944 | #define blend_blocks_subtract_set_blend_mask_untextured() \ |
| 3945 | |
| 3946 | #define blend_blocks_subtract_combine_untextured() \ |
| 3947 | |
| 3948 | #define blend_blocks_subtract_set_stb_untextured() \ |
| 3949 | vorr.u16 blend_pixels, blend_pixels, msb_mask \ |
| 3950 | |
| 3951 | #define blend_blocks_subtract_msb_mask_untextured() \ |
| 3952 | |
| 3953 | |
| 3954 | #define blend_blocks_subtract_mask_set_on() \ |
| 3955 | vclt.s16 write_mask, fb_pixels, #0 \ |
| 3956 | |
| 3957 | #define blend_blocks_subtract_mask_copy_on() \ |
| 3958 | vorr.u16 draw_mask, draw_mask_next, write_mask \ |
| 3959 | |
| 3960 | #define blend_blocks_subtract_mask_set_off() \ |
| 3961 | |
| 3962 | #define blend_blocks_subtract_mask_copy_off() \ |
| 3963 | vmov draw_mask, draw_mask_next \ |
| 3964 | |
| 3965 | |
| 3966 | #define blend_blocks_subtract_builder(texturing, mask_evaluate) \ |
| 3967 | .align 3; \ |
| 3968 | \ |
| 3969 | function(blend_blocks_##texturing##_subtract_##mask_evaluate) \ |
| 3970 | stmdb sp!, { r4, r14 }; \ |
| 3971 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3972 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 3973 | \ |
| 3974 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3975 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 3976 | \ |
| 3977 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3978 | mov c_64, #64; \ |
| 3979 | \ |
| 3980 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 3981 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 3982 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 3983 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 3984 | \ |
| 3985 | vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \ |
| 3986 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 3987 | vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \ |
| 3988 | blend_blocks_subtract_set_blend_mask_##texturing(); \ |
| 3989 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 3990 | blend_blocks_subtract_mask_set_##mask_evaluate(); \ |
| 3991 | vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \ |
| 3992 | \ |
| 3993 | vand.u16 pixels_g, pixels_next, d128_0x03E0; \ |
| 3994 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3995 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 3996 | vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3997 | vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 3998 | \ |
| 3999 | subs num_blocks, num_blocks, #1; \ |
| 4000 | beq 1f; \ |
| 4001 | \ |
| 4002 | 0: \ |
| 4003 | blend_blocks_subtract_mask_copy_##mask_evaluate(); \ |
| 4004 | mov fb_ptr, fb_ptr_next; \ |
| 4005 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 4006 | \ |
| 4007 | vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \ |
| 4008 | blend_blocks_subtract_msb_mask_##texturing(); \ |
| 4009 | \ |
| 4010 | vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \ |
| 4011 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4012 | vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \ |
| 4013 | blend_blocks_subtract_set_stb_##texturing(); \ |
| 4014 | vand.u16 pixels_g, pixels_next, d128_0x03E0; \ |
| 4015 | blend_blocks_subtract_combine_##texturing(); \ |
| 4016 | blend_blocks_subtract_set_blend_mask_##texturing(); \ |
| 4017 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4018 | \ |
| 4019 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 4020 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 4021 | cmp fb_ptr_cmp, #28; \ |
| 4022 | bls 2f; \ |
| 4023 | \ |
| 4024 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4025 | blend_blocks_subtract_mask_set_##mask_evaluate(); \ |
| 4026 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4027 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4028 | vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4029 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 4030 | vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4031 | \ |
| 4032 | 3: \ |
| 4033 | subs num_blocks, num_blocks, #1; \ |
| 4034 | bne 0b; \ |
| 4035 | \ |
| 4036 | 1: \ |
| 4037 | blend_blocks_subtract_mask_copy_##mask_evaluate(); \ |
| 4038 | \ |
| 4039 | blend_blocks_subtract_msb_mask_##texturing(); \ |
| 4040 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4041 | blend_blocks_subtract_set_stb_##texturing(); \ |
| 4042 | blend_blocks_subtract_combine_##texturing(); \ |
| 4043 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4044 | vst1.u16 { blend_pixels }, [fb_ptr_next]; \ |
| 4045 | \ |
| 4046 | ldmia sp!, { r4, pc }; \ |
| 4047 | \ |
| 4048 | 2: \ |
| 4049 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 4050 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4051 | blend_blocks_subtract_mask_set_##mask_evaluate(); \ |
| 4052 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4053 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4054 | vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4055 | vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4056 | bal 3b \ |
| 4057 | |
| 4058 | |
| 4059 | blend_blocks_subtract_builder(textured, off) |
| 4060 | blend_blocks_subtract_builder(textured, on) |
| 4061 | blend_blocks_subtract_builder(untextured, off) |
| 4062 | blend_blocks_subtract_builder(untextured, on) |
| 4063 | |
| 4064 | |
| 4065 | #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \ |
| 4066 | .align 3; \ |
| 4067 | \ |
| 4068 | function(blend_blocks_textured_add_fourth_##mask_evaluate) \ |
| 4069 | stmdb sp!, { r4, r14 }; \ |
| 4070 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 4071 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 4072 | \ |
| 4073 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 4074 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 4075 | \ |
| 4076 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 4077 | mov c_64, #64; \ |
| 4078 | \ |
| 4079 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 4080 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 4081 | vmov.u16 d128_0x1C07, #0x1C00; \ |
| 4082 | vmov.u16 d128_0x00E0, #0x00E0; \ |
| 4083 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 4084 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 4085 | vorr.u16 d128_0x1C07, #0x0007; \ |
| 4086 | \ |
| 4087 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 4088 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 4089 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 4090 | vclt.s16 blend_mask, pixels, #0; \ |
| 4091 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4092 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4093 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4094 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4095 | \ |
| 4096 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4097 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4098 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4099 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4100 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4101 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4102 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4103 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4104 | \ |
| 4105 | subs num_blocks, num_blocks, #1; \ |
| 4106 | beq 1f; \ |
| 4107 | \ |
| 4108 | 0: \ |
| 4109 | mov fb_ptr, fb_ptr_next; \ |
| 4110 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 4111 | \ |
| 4112 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4113 | vbif.u16 blend_pixels, pixels, blend_mask; \ |
| 4114 | \ |
| 4115 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 4116 | vclt.s16 blend_mask, pixels, #0; \ |
| 4117 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4118 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4119 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4120 | \ |
| 4121 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4122 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 4123 | \ |
| 4124 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 4125 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 4126 | cmp fb_ptr_cmp, #28; \ |
| 4127 | bls 2f; \ |
| 4128 | \ |
| 4129 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4130 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4131 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4132 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4133 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4134 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 4135 | \ |
| 4136 | 3: \ |
| 4137 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4138 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4139 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4140 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4141 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4142 | \ |
| 4143 | subs num_blocks, num_blocks, #1; \ |
| 4144 | bne 0b; \ |
| 4145 | \ |
| 4146 | 1: \ |
| 4147 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4148 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4149 | vbif.u16 blend_pixels, pixels, blend_mask; \ |
| 4150 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4151 | vst1.u16 { blend_pixels }, [fb_ptr_next]; \ |
| 4152 | \ |
| 4153 | ldmia sp!, { r4, pc }; \ |
| 4154 | \ |
| 4155 | 2: \ |
| 4156 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 4157 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4158 | \ |
| 4159 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4160 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4161 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4162 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4163 | bal 3b \ |
| 4164 | |
| 4165 | |
| 4166 | |
| 4167 | #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \ |
| 4168 | .align 3; \ |
| 4169 | \ |
| 4170 | function(blend_blocks_untextured_add_fourth_##mask_evaluate) \ |
| 4171 | stmdb sp!, { r4, r14 }; \ |
| 4172 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 4173 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 4174 | \ |
| 4175 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 4176 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \ |
| 4177 | \ |
| 4178 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 4179 | mov c_64, #64; \ |
| 4180 | \ |
| 4181 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 4182 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 4183 | vmov.u16 d128_0x1C07, #0x1C00; \ |
| 4184 | vmov.u16 d128_0x00E0, #0x00E0; \ |
| 4185 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 4186 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 4187 | vorr.u16 d128_0x1C07, #0x0007; \ |
| 4188 | \ |
| 4189 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 4190 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 4191 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 4192 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4193 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4194 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4195 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4196 | \ |
| 4197 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4198 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4199 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4200 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4201 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4202 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4203 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4204 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4205 | \ |
| 4206 | subs num_blocks, num_blocks, #1; \ |
| 4207 | beq 1f; \ |
| 4208 | \ |
| 4209 | 0: \ |
| 4210 | mov fb_ptr, fb_ptr_next; \ |
| 4211 | ldr fb_ptr_next, [pixel_ptr, #28]; \ |
| 4212 | \ |
| 4213 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \ |
| 4214 | \ |
| 4215 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4216 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4217 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4218 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4219 | \ |
| 4220 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4221 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \ |
| 4222 | \ |
| 4223 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 4224 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 4225 | cmp fb_ptr_cmp, #28; \ |
| 4226 | bls 2f; \ |
| 4227 | \ |
| 4228 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4229 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4230 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4231 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4232 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4233 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 4234 | \ |
| 4235 | 3: \ |
| 4236 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4237 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4238 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4239 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4240 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4241 | \ |
| 4242 | subs num_blocks, num_blocks, #1; \ |
| 4243 | bne 0b; \ |
| 4244 | \ |
| 4245 | 1: \ |
| 4246 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4247 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4248 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4249 | vst1.u16 { blend_pixels }, [fb_ptr_next]; \ |
| 4250 | \ |
| 4251 | ldmia sp!, { r4, pc }; \ |
| 4252 | \ |
| 4253 | 2: \ |
| 4254 | vst1.u16 { blend_pixels }, [fb_ptr]; \ |
| 4255 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4256 | \ |
| 4257 | vld1.u16 { fb_pixels }, [fb_ptr_next]; \ |
| 4258 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4259 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4260 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4261 | bal 3b \ |
| 4262 | |
| 4263 | |
| 4264 | blend_blocks_add_fourth_textured_builder(off) |
| 4265 | blend_blocks_add_fourth_textured_builder(on) |
| 4266 | blend_blocks_add_fourth_untextured_builder(off) |
| 4267 | blend_blocks_add_fourth_untextured_builder(on) |
| 4268 | |
| 4269 | // TODO: Optimize this more. Need a scene that actually uses it for |
| 4270 | // confirmation.. |
| 4271 | |
| 4272 | .align 3 |
| 4273 | |
| 4274 | function(blend_blocks_textured_unblended_on) |
| 4275 | stmdb sp!, { r4, r14 } |
| 4276 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset |
| 4277 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 4278 | |
| 4279 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16) |
| 4280 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] |
| 4281 | |
| 4282 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 4283 | mov c_64, #64 |
| 4284 | |
| 4285 | ldr fb_ptr, [pixel_ptr, #28] |
| 4286 | vld1.u16 { fb_pixels }, [fb_ptr] |
| 4287 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64 |
| 4288 | vclt.s16 write_mask, fb_pixels, #0 |
| 4289 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64 |
| 4290 | |
| 4291 | subs num_blocks, num_blocks, #1 |
| 4292 | beq 1f |
| 4293 | |
| 4294 | 0: |
| 4295 | vorr.u16 pixels, pixels, msb_mask |
| 4296 | vorr.u16 draw_mask, draw_mask, write_mask |
| 4297 | vbif.u16 fb_pixels, pixels, draw_mask |
| 4298 | vst1.u16 { fb_pixels }, [fb_ptr] |
| 4299 | |
| 4300 | ldr fb_ptr, [pixel_ptr, #28] |
| 4301 | vld1.u16 { fb_pixels }, [fb_ptr] |
| 4302 | vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64 |
| 4303 | vclt.s16 write_mask, fb_pixels, #0 |
| 4304 | vld1.u32 { pixels }, [pixel_ptr, :128], c_64 |
| 4305 | |
| 4306 | subs num_blocks, num_blocks, #1 |
| 4307 | bne 0b |
| 4308 | |
| 4309 | 1: |
| 4310 | vorr.u16 pixels, pixels, msb_mask |
| 4311 | vorr.u16 draw_mask, draw_mask, write_mask |
| 4312 | vbif.u16 fb_pixels, pixels, draw_mask |
| 4313 | vst1.u16 { fb_pixels }, [fb_ptr] |
| 4314 | |
| 4315 | ldmia sp!, { r4, pc } |
| 4316 | |
| 4317 | |
| 4318 | function(blend_blocks_textured_unblended_off) |
| 4319 | bx lr |
| 4320 | |
| 4321 | |
| 4322 | function(warmup) |
| 4323 | mov r3, #64 |
| 4324 | cmp r0, #0 |
| 4325 | bxeq lr |
| 4326 | |
| 4327 | 0: |
| 4328 | vld1.u32 { u_whole_8, v_whole_8 }, [r1, :128], r3 |
| 4329 | |
| 4330 | subs r0, r0, #1 |
| 4331 | bne 0b |
| 4332 | |
| 4333 | bx lr |
| 4334 | |
| 4335 | #undef vram_ptr |
| 4336 | #undef color |
| 4337 | #undef width |
| 4338 | #undef height |
| 4339 | #undef pitch |
| 4340 | |
| 4341 | #define vram_ptr r0 |
| 4342 | #define color r1 |
| 4343 | #define width r2 |
| 4344 | #define height r3 |
| 4345 | |
| 4346 | #define pitch r1 |
| 4347 | |
| 4348 | #define num_width r12 |
| 4349 | |
| 4350 | #undef colors_a |
| 4351 | #undef colors_b |
| 4352 | |
| 4353 | #define colors_a q0 |
| 4354 | #define colors_b q1 |
| 4355 | |
| 4356 | .align 3 |
| 4357 | |
| 4358 | function(render_block_fill_body) |
| 4359 | vdup.u16 colors_a, color |
| 4360 | mov pitch, #2048 |
| 4361 | |
| 4362 | vmov colors_b, colors_a |
| 4363 | sub pitch, pitch, width, lsl #1 |
| 4364 | |
| 4365 | mov num_width, width |
| 4366 | |
| 4367 | 0: |
| 4368 | vst1.u32 { colors_a, colors_b }, [vram_ptr, :256]! |
| 4369 | |
| 4370 | subs num_width, num_width, #16 |
| 4371 | bne 0b |
| 4372 | |
| 4373 | add vram_ptr, vram_ptr, pitch |
| 4374 | mov num_width, width |
| 4375 | |
| 4376 | subs height, height, #1 |
| 4377 | bne 0b |
| 4378 | |
| 4379 | bx lr |
| 4380 | |
| 4381 | |
| 4382 | #undef x |
| 4383 | #undef y |
| 4384 | #undef width |
| 4385 | #undef height |
| 4386 | #undef fb_ptr |
| 4387 | #undef texture_mask |
| 4388 | #undef num_blocks |
| 4389 | #undef temp |
| 4390 | #undef dirty_textures_mask |
| 4391 | #undef clut_ptr |
| 4392 | #undef current_texture_mask |
| 4393 | |
| 4394 | #define psx_gpu r0 |
| 4395 | #define x r1 |
| 4396 | #define y r2 |
| 4397 | #define u r3 |
| 4398 | #define v r4 |
| 4399 | #define width r5 |
| 4400 | #define height r6 |
| 4401 | #define offset_u r8 |
| 4402 | #define offset_v r9 |
| 4403 | #define offset_u_right r10 |
| 4404 | #define width_rounded r11 |
| 4405 | #define height_rounded r12 |
| 4406 | |
| 4407 | #define texture_offset_base r1 |
| 4408 | #define tile_width r2 |
| 4409 | #define tile_height r3 |
| 4410 | #define num_blocks r4 |
| 4411 | #define block r5 |
| 4412 | #define sub_tile_height r6 |
| 4413 | #define fb_ptr r7 |
| 4414 | #define texture_mask r8 |
| 4415 | #define column_data r9 |
| 4416 | #define texture_offset r10 |
| 4417 | #define tiles_remaining r11 |
| 4418 | #define fb_ptr_advance_column r12 |
| 4419 | #define texture_block_ptr r14 |
| 4420 | |
| 4421 | #define temp r14 |
| 4422 | |
| 4423 | #define texture_page_ptr r3 |
| 4424 | #define left_block_mask r4 |
| 4425 | #define right_block_mask r5 |
| 4426 | #define texture_mask_rev r10 |
| 4427 | #define control_mask r11 |
| 4428 | |
| 4429 | #define dirty_textures_mask r4 |
| 4430 | #define clut_ptr r5 |
| 4431 | #define current_texture_mask r6 |
| 4432 | |
| 4433 | |
| 4434 | #undef texels |
| 4435 | #undef clut_low_a |
| 4436 | #undef clut_low_b |
| 4437 | #undef clut_high_a |
| 4438 | #undef clut_high_b |
| 4439 | #undef clut_a |
| 4440 | #undef clut_b |
| 4441 | #undef texels_low |
| 4442 | #undef texels_high |
| 4443 | |
| 4444 | #define texels d0 |
| 4445 | #define draw_masks_fb_ptrs q1 |
| 4446 | |
| 4447 | #define draw_mask_fb_ptr_left d2 |
| 4448 | #define draw_mask_fb_ptr_right d3 |
| 4449 | |
| 4450 | #define draw_mask_fb_ptr_left_a d2 |
| 4451 | #define draw_mask_fb_ptr_left_b d3 |
| 4452 | #define draw_mask_fb_ptr_right_a d10 |
| 4453 | #define draw_mask_fb_ptr_right_b d11 |
| 4454 | #define draw_masks_fb_ptrs2 q5 |
| 4455 | |
| 4456 | #define clut_low_a d4 |
| 4457 | #define clut_low_b d5 |
| 4458 | #define clut_high_a d6 |
| 4459 | #define clut_high_b d7 |
| 4460 | |
| 4461 | #define block_masks d8 |
| 4462 | #define block_masks_shifted d9 |
| 4463 | |
| 4464 | #define clut_a q2 |
| 4465 | #define clut_b q3 |
| 4466 | |
| 4467 | #define texels_low d12 |
| 4468 | #define texels_high d13 |
| 4469 | |
| 4470 | #define texels_wide_low d14 |
| 4471 | #define texels_wide_high d15 |
| 4472 | #define texels_wide q7 |
| 4473 | |
| 4474 | |
| 4475 | setup_sprite_flush_blocks: |
| 4476 | vpush { q1 - q5 } |
| 4477 | |
| 4478 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 } |
| 4479 | bl flush_render_block_buffer |
| 4480 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 } |
| 4481 | |
| 4482 | vpop { q1 - q5 } |
| 4483 | |
| 4484 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 4485 | bx lr |
| 4486 | |
| 4487 | |
| 4488 | setup_sprite_update_texture_4bpp_cache: |
| 4489 | stmdb sp!, { r0 - r3, r14 } |
| 4490 | bl update_texture_4bpp_cache |
| 4491 | ldmia sp!, { r0 - r3, pc } |
| 4492 | |
| 4493 | |
| 4494 | setup_sprite_update_texture_8bpp_cache: |
| 4495 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r14 } |
| 4496 | bl update_texture_8bpp_cache |
| 4497 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS pc } |
| 4498 | |
| 4499 | |
| 4500 | #define setup_sprite_tiled_initialize_4bpp() \ |
| 4501 | ldr dirty_textures_mask, \ |
| 4502 | [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]; \ |
| 4503 | ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \ |
| 4504 | \ |
| 4505 | ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \ |
| 4506 | vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \ |
| 4507 | \ |
| 4508 | tst current_texture_mask, dirty_textures_mask; \ |
| 4509 | vuzp.u8 clut_a, clut_b; \ |
| 4510 | \ |
| 4511 | blne setup_sprite_update_texture_4bpp_cache \ |
| 4512 | |
| 4513 | #define setup_sprite_tiled_initialize_8bpp() \ |
| 4514 | ldr dirty_textures_mask, \ |
| 4515 | [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]; \ |
| 4516 | ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \ |
| 4517 | \ |
| 4518 | tst current_texture_mask, dirty_textures_mask; \ |
| 4519 | blne setup_sprite_update_texture_8bpp_cache \ |
| 4520 | |
| 4521 | |
| 4522 | #define setup_sprite_block_count_single() \ |
| 4523 | sub_tile_height \ |
| 4524 | |
| 4525 | #define setup_sprite_block_count_double() \ |
| 4526 | sub_tile_height, lsl #1 \ |
| 4527 | |
| 4528 | #define setup_sprite_tile_add_blocks(type) \ |
| 4529 | add num_blocks, num_blocks, setup_sprite_block_count_##type(); \ |
| 4530 | cmp num_blocks, #MAX_BLOCKS; \ |
| 4531 | \ |
| 4532 | movgt num_blocks, setup_sprite_block_count_##type(); \ |
| 4533 | blgt setup_sprite_flush_blocks \ |
| 4534 | |
| 4535 | |
| 4536 | #define setup_sprite_tile_full_4bpp(edge) \ |
| 4537 | setup_sprite_tile_add_blocks(double); \ |
| 4538 | \ |
| 4539 | 4: \ |
| 4540 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4541 | vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \ |
| 4542 | \ |
| 4543 | pld [fb_ptr]; \ |
| 4544 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4545 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4546 | \ |
| 4547 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4548 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4549 | \ |
| 4550 | vst2.u8 { texels_low, texels_high }, [block, :128]; \ |
| 4551 | add texture_block_ptr, texture_offset, #8; \ |
| 4552 | \ |
| 4553 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 4554 | add block, block, #40; \ |
| 4555 | \ |
| 4556 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4557 | add fb_ptr, fb_ptr, #16; \ |
| 4558 | \ |
| 4559 | vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \ |
| 4560 | add block, block, #24; \ |
| 4561 | \ |
| 4562 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4563 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4564 | \ |
| 4565 | pld [fb_ptr]; \ |
| 4566 | vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \ |
| 4567 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4568 | \ |
| 4569 | vst2.u8 { texels_low, texels_high }, [block, :128]; \ |
| 4570 | add block, block, #40; \ |
| 4571 | \ |
| 4572 | add texture_offset, texture_offset, #0x10; \ |
| 4573 | add fb_ptr, fb_ptr, #(2048 - 16); \ |
| 4574 | \ |
| 4575 | vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \ |
| 4576 | add block, block, #24; \ |
| 4577 | \ |
| 4578 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4579 | bne 4b; \ |
| 4580 | \ |
| 4581 | add texture_offset, texture_offset, #0xF00; \ |
| 4582 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 4583 | |
| 4584 | |
| 4585 | #define setup_sprite_tile_half_4bpp(edge) \ |
| 4586 | setup_sprite_tile_add_blocks(single); \ |
| 4587 | \ |
| 4588 | 4: \ |
| 4589 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4590 | vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \ |
| 4591 | \ |
| 4592 | pld [fb_ptr]; \ |
| 4593 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4594 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4595 | \ |
| 4596 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4597 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4598 | \ |
| 4599 | vst2.u8 { texels_low, texels_high }, [block, :128]; \ |
| 4600 | add block, block, #40; \ |
| 4601 | \ |
| 4602 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4603 | vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \ |
| 4604 | \ |
| 4605 | add block, block, #24; \ |
| 4606 | add texture_offset, texture_offset, #0x10; \ |
| 4607 | \ |
| 4608 | add fb_ptr, fb_ptr, #2048; \ |
| 4609 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4610 | \ |
| 4611 | bne 4b; \ |
| 4612 | \ |
| 4613 | add texture_offset, texture_offset, #0xF00; \ |
| 4614 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 4615 | |
| 4616 | |
| 4617 | #define setup_sprite_tile_full_8bpp(edge) \ |
| 4618 | setup_sprite_tile_add_blocks(double); \ |
| 4619 | add block, block, #16; \ |
| 4620 | \ |
| 4621 | 4: \ |
| 4622 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4623 | vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \ |
| 4624 | \ |
| 4625 | pld [fb_ptr]; \ |
| 4626 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4627 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4628 | \ |
| 4629 | add texture_block_ptr, texture_offset, #8; \ |
| 4630 | vst1.u32 { texels }, [block, :64]; \ |
| 4631 | \ |
| 4632 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 4633 | add block, block, #24; \ |
| 4634 | \ |
| 4635 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4636 | \ |
| 4637 | add fb_ptr, fb_ptr, #16; \ |
| 4638 | vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \ |
| 4639 | \ |
| 4640 | add block, block, #40; \ |
| 4641 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4642 | pld [fb_ptr]; \ |
| 4643 | \ |
| 4644 | vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \ |
| 4645 | vst1.u32 { texels }, [block, :64]; \ |
| 4646 | add block, block, #24; \ |
| 4647 | \ |
| 4648 | add texture_offset, texture_offset, #0x10; \ |
| 4649 | add fb_ptr, fb_ptr, #(2048 - 16); \ |
| 4650 | \ |
| 4651 | vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \ |
| 4652 | add block, block, #40; \ |
| 4653 | \ |
| 4654 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4655 | bne 4b; \ |
| 4656 | \ |
| 4657 | sub block, block, #16; \ |
| 4658 | add texture_offset, texture_offset, #0xF00; \ |
| 4659 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 4660 | |
| 4661 | |
| 4662 | #define setup_sprite_tile_half_8bpp(edge) \ |
| 4663 | setup_sprite_tile_add_blocks(single); \ |
| 4664 | add block, block, #16; \ |
| 4665 | \ |
| 4666 | 4: \ |
| 4667 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4668 | vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \ |
| 4669 | pld [fb_ptr]; \ |
| 4670 | \ |
| 4671 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4672 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4673 | \ |
| 4674 | vst1.u32 { texels }, [block, :64]; \ |
| 4675 | add block, block, #24; \ |
| 4676 | \ |
| 4677 | vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \ |
| 4678 | add block, block, #40; \ |
| 4679 | \ |
| 4680 | add texture_offset, texture_offset, #0x10; \ |
| 4681 | add fb_ptr, fb_ptr, #2048; \ |
| 4682 | \ |
| 4683 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4684 | bne 4b; \ |
| 4685 | \ |
| 4686 | sub block, block, #16; \ |
| 4687 | add texture_offset, texture_offset, #0xF00; \ |
| 4688 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 4689 | |
| 4690 | |
| 4691 | #define setup_sprite_tile_column_edge_pre_adjust_half_right() \ |
| 4692 | add texture_offset, texture_offset_base, #8; \ |
| 4693 | add fb_ptr, fb_ptr, #16 \ |
| 4694 | |
| 4695 | #define setup_sprite_tile_column_edge_pre_adjust_half_left() \ |
| 4696 | mov texture_offset, texture_offset_base \ |
| 4697 | |
| 4698 | #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \ |
| 4699 | setup_sprite_tile_column_edge_pre_adjust_half_##edge() \ |
| 4700 | |
| 4701 | #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \ |
| 4702 | mov texture_offset, texture_offset_base \ |
| 4703 | |
| 4704 | #define setup_sprite_tile_column_edge_post_adjust_half_right() \ |
| 4705 | sub fb_ptr, fb_ptr, #16 \ |
| 4706 | |
| 4707 | #define setup_sprite_tile_column_edge_post_adjust_half_left() \ |
| 4708 | |
| 4709 | #define setup_sprite_tile_column_edge_post_adjust_half(edge) \ |
| 4710 | setup_sprite_tile_column_edge_post_adjust_half_##edge() \ |
| 4711 | |
| 4712 | #define setup_sprite_tile_column_edge_post_adjust_full(edge) \ |
| 4713 | |
| 4714 | |
| 4715 | #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \ |
| 4716 | x4mode) \ |
| 4717 | mov sub_tile_height, column_data; \ |
| 4718 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 4719 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4720 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \ |
| 4721 | |
| 4722 | #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \ |
| 4723 | x4mode) \ |
| 4724 | and sub_tile_height, column_data, #0xFF; \ |
| 4725 | mov tiles_remaining, column_data, lsr #16; \ |
| 4726 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 4727 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4728 | \ |
| 4729 | subs tiles_remaining, tiles_remaining, #1; \ |
| 4730 | beq 2f; \ |
| 4731 | \ |
| 4732 | 3: \ |
| 4733 | mov sub_tile_height, #16; \ |
| 4734 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4735 | subs tiles_remaining, tiles_remaining, #1; \ |
| 4736 | bne 3b; \ |
| 4737 | \ |
| 4738 | 2: \ |
| 4739 | uxtb sub_tile_height, column_data, ror #8; \ |
| 4740 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4741 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \ |
| 4742 | |
| 4743 | |
| 4744 | #define setup_sprite_column_data_single() \ |
| 4745 | mov column_data, height; \ |
| 4746 | ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] \ |
| 4747 | |
| 4748 | #define setup_sprite_column_data_multi() \ |
| 4749 | and height_rounded, height_rounded, #0xF; \ |
| 4750 | rsb column_data, offset_v, #16; \ |
| 4751 | \ |
| 4752 | add height_rounded, height_rounded, #1; \ |
| 4753 | sub tile_height, tile_height, #1; \ |
| 4754 | \ |
| 4755 | orr column_data, column_data, tile_height, lsl #16; \ |
| 4756 | ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]; \ |
| 4757 | \ |
| 4758 | orr column_data, column_data, height_rounded, lsl #8 \ |
| 4759 | |
| 4760 | #define setup_sprite_setup_left_draw_mask_fb_ptr() \ |
| 4761 | vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \ |
| 4762 | vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \ |
| 4763 | |
| 4764 | #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \ |
| 4765 | mov fb_ptr_advance_column, #32; \ |
| 4766 | vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \ |
| 4767 | \ |
| 4768 | sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \ |
| 4769 | vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \ |
| 4770 | |
| 4771 | #define setup_sprite_setup_right_draw_mask_fb_ptr() \ |
| 4772 | vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \ |
| 4773 | vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \ |
| 4774 | |
| 4775 | #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \ |
| 4776 | edge, x4mode) \ |
| 4777 | setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \ |
| 4778 | setup_sprite_column_data_##multi_height(); \ |
| 4779 | vext.32 block_masks_shifted, block_masks, block_masks, #1; \ |
| 4780 | vorr.u32 block_masks, block_masks, block_masks_shifted; \ |
| 4781 | setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \ |
| 4782 | \ |
| 4783 | setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \ |
| 4784 | ldmia sp!, { r4 - r11, pc } \ |
| 4785 | |
| 4786 | #define setup_sprite_tiled_advance_column() \ |
| 4787 | add texture_offset_base, texture_offset_base, #0x100; \ |
| 4788 | tst texture_offset_base, #0xF00; \ |
| 4789 | subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \ |
| 4790 | |
| 4791 | #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \ |
| 4792 | right_mode, x4mode) \ |
| 4793 | setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\ |
| 4794 | setup_sprite_column_data_##multi_height(); \ |
| 4795 | \ |
| 4796 | setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \ |
| 4797 | \ |
| 4798 | setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\ |
| 4799 | \ |
| 4800 | subs tile_width, tile_width, #2; \ |
| 4801 | add fb_ptr, fb_ptr, fb_ptr_advance_column; \ |
| 4802 | \ |
| 4803 | beq 1f; \ |
| 4804 | \ |
| 4805 | vmov.u8 draw_masks_fb_ptrs, #0; \ |
| 4806 | vmov.u8 draw_masks_fb_ptrs2, #0; \ |
| 4807 | \ |
| 4808 | 0: \ |
| 4809 | setup_sprite_tiled_advance_column(); \ |
| 4810 | setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \ |
| 4811 | add fb_ptr, fb_ptr, fb_ptr_advance_column; \ |
| 4812 | subs tile_width, tile_width, #1; \ |
| 4813 | bne 0b; \ |
| 4814 | \ |
| 4815 | 1: \ |
| 4816 | setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \ |
| 4817 | \ |
| 4818 | setup_sprite_tiled_advance_column(); \ |
| 4819 | setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\ |
| 4820 | ldmia sp!, { r4 - r11, pc } \ |
| 4821 | |
| 4822 | |
| 4823 | #define setup_sprite_offset_u_adjust() \ |
| 4824 | |
| 4825 | #define setup_sprite_get_left_block_mask() \ |
| 4826 | and left_block_mask, left_block_mask, #0xFF \ |
| 4827 | |
| 4828 | #define setup_sprite_compare_left_block_mask() \ |
| 4829 | cmp left_block_mask, #0xFF \ |
| 4830 | |
| 4831 | #define setup_sprite_get_right_block_mask() \ |
| 4832 | uxtb right_block_mask, right_block_mask, ror #8 \ |
| 4833 | |
| 4834 | #define setup_sprite_compare_right_block_mask() \ |
| 4835 | cmp right_block_mask, #0xFF \ |
| 4836 | |
| 4837 | |
| 4838 | |
| 4839 | /* 4x stuff */ |
| 4840 | #define fb_ptr2 column_data |
| 4841 | |
| 4842 | #define setup_sprite_offset_u_adjust_4x() \ |
| 4843 | sub fb_ptr, fb_ptr, offset_u, lsl #1; \ |
| 4844 | lsl offset_u_right, #1; \ |
| 4845 | lsl offset_u, #1; \ |
| 4846 | add offset_u_right, #1 \ |
| 4847 | |
| 4848 | #define setup_sprite_get_left_block_mask_4x() \ |
| 4849 | sxth left_block_mask, left_block_mask \ |
| 4850 | |
| 4851 | #define setup_sprite_compare_left_block_mask_4x() \ |
| 4852 | cmp left_block_mask, #0xFFFFFFFF \ |
| 4853 | |
| 4854 | #define setup_sprite_get_right_block_mask_4x() \ |
| 4855 | sxth right_block_mask, right_block_mask, ror #16 \ |
| 4856 | |
| 4857 | #define setup_sprite_compare_right_block_mask_4x() \ |
| 4858 | cmp right_block_mask, #0xFFFFFFFF \ |
| 4859 | |
| 4860 | |
| 4861 | #define widen_texels_16bpp(texels_) \ |
| 4862 | vmov texels_wide_low, texels_; \ |
| 4863 | vmov texels_wide_high, texels_; \ |
| 4864 | vzip.16 texels_wide_low, texels_wide_high \ |
| 4865 | |
| 4866 | #define widen_texels_8bpp(texels_) \ |
| 4867 | vmov texels_wide_low, texels_; \ |
| 4868 | vmov texels_wide_high, texels_; \ |
| 4869 | vzip.8 texels_wide_low, texels_wide_high \ |
| 4870 | |
| 4871 | #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \ |
| 4872 | vst1.u32 { texels_ }, [block_, :128]; \ |
| 4873 | add block_, block_, #40; \ |
| 4874 | \ |
| 4875 | vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \ |
| 4876 | vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \ |
| 4877 | add block_, block_, #24 \ |
| 4878 | |
| 4879 | /* assumes 16-byte offset already added to block_ */ |
| 4880 | #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \ |
| 4881 | vst1.u32 { texels_ }, [block_, :64]; \ |
| 4882 | add block_, block_, #24; \ |
| 4883 | \ |
| 4884 | vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \ |
| 4885 | vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \ |
| 4886 | add block_, block_, #40 \ |
| 4887 | |
| 4888 | #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \ |
| 4889 | draw_mask_fb_ptr_b_) \ |
| 4890 | widen_texels_16bpp(texels_low); \ |
| 4891 | add fb_ptr_tmp, fb_ptr, #1024*2; \ |
| 4892 | \ |
| 4893 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \ |
| 4894 | \ |
| 4895 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \ |
| 4896 | widen_texels_16bpp(texels_high); \ |
| 4897 | \ |
| 4898 | add fb_ptr_tmp, fb_ptr, #8*2; \ |
| 4899 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \ |
| 4900 | \ |
| 4901 | add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \ |
| 4902 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \ |
| 4903 | |
| 4904 | #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \ |
| 4905 | draw_mask_fb_ptr_b_) \ |
| 4906 | widen_texels_8bpp(texels); \ |
| 4907 | add fb_ptr_tmp, fb_ptr, #1024*2; \ |
| 4908 | \ |
| 4909 | write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \ |
| 4910 | write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \ |
| 4911 | \ |
| 4912 | add fb_ptr_tmp, fb_ptr, #8*2; \ |
| 4913 | write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \ |
| 4914 | \ |
| 4915 | add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \ |
| 4916 | write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \ |
| 4917 | |
| 4918 | |
| 4919 | #define setup_sprite_tiled_initialize_4bpp_4x() \ |
| 4920 | ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \ |
| 4921 | vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \ |
| 4922 | \ |
| 4923 | vuzp.u8 clut_a, clut_b \ |
| 4924 | |
| 4925 | #define setup_sprite_tiled_initialize_8bpp_4x() \ |
| 4926 | |
| 4927 | |
| 4928 | #define setup_sprite_block_count_single_4x() \ |
| 4929 | sub_tile_height, lsl #2 \ |
| 4930 | |
| 4931 | #define setup_sprite_block_count_double_4x() \ |
| 4932 | sub_tile_height, lsl #(1+2) \ |
| 4933 | |
| 4934 | #define setup_sprite_tile_full_4bpp_4x(edge) \ |
| 4935 | setup_sprite_tile_add_blocks(double_4x); \ |
| 4936 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 4937 | \ |
| 4938 | 4: \ |
| 4939 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4940 | pld [fb_ptr]; \ |
| 4941 | \ |
| 4942 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4943 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4944 | \ |
| 4945 | add texture_block_ptr, texture_offset, #8; \ |
| 4946 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4947 | \ |
| 4948 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 4949 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4950 | \ |
| 4951 | vzip.8 texels_low, texels_high; \ |
| 4952 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \ |
| 4953 | draw_mask_fb_ptr_left_b); \ |
| 4954 | \ |
| 4955 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4956 | pld [fb_ptr, #2048]; \ |
| 4957 | \ |
| 4958 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4959 | add fb_ptr, fb_ptr, #16*2; \ |
| 4960 | \ |
| 4961 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4962 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4963 | \ |
| 4964 | vzip.8 texels_low, texels_high; \ |
| 4965 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \ |
| 4966 | draw_mask_fb_ptr_right_b); \ |
| 4967 | \ |
| 4968 | add texture_offset, texture_offset, #0x10; \ |
| 4969 | add fb_ptr, fb_ptr, #(2048 - 16) * 2; \ |
| 4970 | \ |
| 4971 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4972 | bne 4b; \ |
| 4973 | \ |
| 4974 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 4975 | add texture_offset, texture_offset, #0xF00; \ |
| 4976 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 4977 | |
| 4978 | |
| 4979 | #define setup_sprite_tile_half_4bpp_4x(edge) \ |
| 4980 | setup_sprite_tile_add_blocks(single_4x); \ |
| 4981 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 4982 | \ |
| 4983 | 4: \ |
| 4984 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4985 | pld [fb_ptr]; \ |
| 4986 | \ |
| 4987 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4988 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 4989 | \ |
| 4990 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4991 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4992 | \ |
| 4993 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4994 | add texture_offset, texture_offset, #0x10; \ |
| 4995 | \ |
| 4996 | vzip.8 texels_low, texels_high; \ |
| 4997 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \ |
| 4998 | draw_mask_fb_ptr_##edge##_b); \ |
| 4999 | \ |
| 5000 | pld [fb_ptr, #2048]; \ |
| 5001 | add fb_ptr, fb_ptr, #2048 * 2; \ |
| 5002 | \ |
| 5003 | subs sub_tile_height, sub_tile_height, #1; \ |
| 5004 | bne 4b; \ |
| 5005 | \ |
| 5006 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 5007 | add texture_offset, texture_offset, #0xF00; \ |
| 5008 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 5009 | |
| 5010 | |
| 5011 | #define setup_sprite_tile_full_8bpp_4x(edge) \ |
| 5012 | setup_sprite_tile_add_blocks(double_4x); \ |
| 5013 | add block, block, #16; \ |
| 5014 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 5015 | \ |
| 5016 | 4: \ |
| 5017 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 5018 | pld [fb_ptr]; \ |
| 5019 | \ |
| 5020 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 5021 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 5022 | \ |
| 5023 | add texture_block_ptr, texture_offset, #8; \ |
| 5024 | do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \ |
| 5025 | draw_mask_fb_ptr_left_b); \ |
| 5026 | \ |
| 5027 | pld [fb_ptr, #2048]; \ |
| 5028 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 5029 | \ |
| 5030 | add fb_ptr, fb_ptr, #16*2; \ |
| 5031 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 5032 | \ |
| 5033 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 5034 | \ |
| 5035 | do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \ |
| 5036 | draw_mask_fb_ptr_right_b); \ |
| 5037 | \ |
| 5038 | add texture_offset, texture_offset, #0x10; \ |
| 5039 | add fb_ptr, fb_ptr, #(2048 - 16) * 2; \ |
| 5040 | \ |
| 5041 | subs sub_tile_height, sub_tile_height, #1; \ |
| 5042 | bne 4b; \ |
| 5043 | \ |
| 5044 | sub block, block, #16; \ |
| 5045 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 5046 | add texture_offset, texture_offset, #0xF00; \ |
| 5047 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 5048 | |
| 5049 | |
| 5050 | #define setup_sprite_tile_half_8bpp_4x(edge) \ |
| 5051 | setup_sprite_tile_add_blocks(single_4x); \ |
| 5052 | add block, block, #16; \ |
| 5053 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 5054 | \ |
| 5055 | 4: \ |
| 5056 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 5057 | pld [fb_ptr]; \ |
| 5058 | \ |
| 5059 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 5060 | vld1.u32 { texels }, [texture_block_ptr, :64]; \ |
| 5061 | \ |
| 5062 | pld [fb_ptr, #2048]; \ |
| 5063 | do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \ |
| 5064 | draw_mask_fb_ptr_##edge##_b); \ |
| 5065 | \ |
| 5066 | add texture_offset, texture_offset, #0x10; \ |
| 5067 | add fb_ptr, fb_ptr, #2048 * 2; \ |
| 5068 | \ |
| 5069 | subs sub_tile_height, sub_tile_height, #1; \ |
| 5070 | bne 4b; \ |
| 5071 | \ |
| 5072 | sub block, block, #16; \ |
| 5073 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 5074 | add texture_offset, texture_offset, #0xF00; \ |
| 5075 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \ |
| 5076 | |
| 5077 | |
| 5078 | #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \ |
| 5079 | add texture_offset, texture_offset_base, #8; \ |
| 5080 | add fb_ptr, fb_ptr, #16 * 2 \ |
| 5081 | |
| 5082 | #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \ |
| 5083 | mov texture_offset, texture_offset_base \ |
| 5084 | |
| 5085 | #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \ |
| 5086 | setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \ |
| 5087 | |
| 5088 | #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \ |
| 5089 | mov texture_offset, texture_offset_base \ |
| 5090 | |
| 5091 | #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \ |
| 5092 | sub fb_ptr, fb_ptr, #16 * 2 \ |
| 5093 | |
| 5094 | #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \ |
| 5095 | |
| 5096 | #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \ |
| 5097 | setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \ |
| 5098 | |
| 5099 | #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \ |
| 5100 | |
| 5101 | |
| 5102 | #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \ |
| 5103 | vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \ |
| 5104 | vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \ |
| 5105 | vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \ |
| 5106 | vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \ |
| 5107 | |
| 5108 | #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \ |
| 5109 | mov fb_ptr_advance_column, #32 * 2; \ |
| 5110 | vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \ |
| 5111 | vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \ |
| 5112 | sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \ |
| 5113 | vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \ |
| 5114 | vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \ |
| 5115 | |
| 5116 | #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \ |
| 5117 | vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \ |
| 5118 | vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \ |
| 5119 | vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \ |
| 5120 | vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \ |
| 5121 | |
| 5122 | |
| 5123 | // r0: psx_gpu |
| 5124 | // r1: x |
| 5125 | // r2: y |
| 5126 | // r3: u |
| 5127 | // [sp]: v |
| 5128 | // [sp + 4]: width |
| 5129 | // [sp + 8]: height |
| 5130 | // [sp + 12]: color (unused) |
| 5131 | |
| 5132 | #define setup_sprite_tiled_builder(texture_mode, x4mode) \ |
| 5133 | \ |
| 5134 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \ |
| 5135 | x4mode); \ |
| 5136 | setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \ |
| 5137 | x4mode); \ |
| 5138 | setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \ |
| 5139 | x4mode); \ |
| 5140 | setup_sprite_tile_column_width_single(texture_mode, single, full, none, \ |
| 5141 | x4mode); \ |
| 5142 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \ |
| 5143 | x4mode); \ |
| 5144 | setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \ |
| 5145 | x4mode); \ |
| 5146 | setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \ |
| 5147 | x4mode); \ |
| 5148 | setup_sprite_tile_column_width_single(texture_mode, single, half, right, \ |
| 5149 | x4mode); \ |
| 5150 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \ |
| 5151 | x4mode); \ |
| 5152 | setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \ |
| 5153 | x4mode); \ |
| 5154 | setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \ |
| 5155 | x4mode); \ |
| 5156 | setup_sprite_tile_column_width_single(texture_mode, single, half, left, \ |
| 5157 | x4mode); \ |
| 5158 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \ |
| 5159 | x4mode); \ |
| 5160 | setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \ |
| 5161 | x4mode); \ |
| 5162 | \ |
| 5163 | .align 4; \ |
| 5164 | \ |
| 5165 | function(setup_sprite_##texture_mode##x4mode) \ |
| 5166 | stmdb sp!, { r4 - r11, r14 }; \ |
| 5167 | setup_sprite_tiled_initialize_##texture_mode##x4mode(); \ |
| 5168 | \ |
| 5169 | ldr v, [sp, #36]; \ |
| 5170 | and offset_u, u, #0xF; \ |
| 5171 | \ |
| 5172 | ldr width, [sp, #40]; \ |
| 5173 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \ |
| 5174 | \ |
| 5175 | ldr height, [sp, #44]; \ |
| 5176 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 5177 | \ |
| 5178 | add fb_ptr, fb_ptr, x, lsl #1; \ |
| 5179 | and offset_v, v, #0xF; \ |
| 5180 | \ |
| 5181 | sub fb_ptr, fb_ptr, offset_u, lsl #1; \ |
| 5182 | add width_rounded, offset_u, width; \ |
| 5183 | \ |
| 5184 | add height_rounded, offset_v, height; \ |
| 5185 | add width_rounded, width_rounded, #15; \ |
| 5186 | \ |
| 5187 | add height_rounded, height_rounded, #15; \ |
| 5188 | mov tile_width, width_rounded, lsr #4; \ |
| 5189 | \ |
| 5190 | /* texture_offset_base = VH-VL-00-00 */\ |
| 5191 | mov texture_offset_base, v, lsl #8; \ |
| 5192 | and offset_u_right, width_rounded, #0xF; \ |
| 5193 | \ |
| 5194 | /* texture_offset_base = VH-UH-UL-00 */\ |
| 5195 | bfi texture_offset_base, u, #4, #8; \ |
| 5196 | mov right_block_mask, #0xFFFFFFFE; \ |
| 5197 | \ |
| 5198 | setup_sprite_offset_u_adjust##x4mode(); \ |
| 5199 | \ |
| 5200 | /* texture_offset_base = VH-UH-VL-00 */\ |
| 5201 | bfi texture_offset_base, v, #4, #4; \ |
| 5202 | mov left_block_mask, #0xFFFFFFFF; \ |
| 5203 | \ |
| 5204 | mov tile_height, height_rounded, lsr #4; \ |
| 5205 | mvn left_block_mask, left_block_mask, lsl offset_u; \ |
| 5206 | \ |
| 5207 | /* texture_mask = HH-HL-WH-WL */\ |
| 5208 | ldrh texture_mask, [psx_gpu, #psx_gpu_texture_mask_width_offset]; \ |
| 5209 | mov right_block_mask, right_block_mask, lsl offset_u_right; \ |
| 5210 | \ |
| 5211 | /* texture_mask_rev = WH-WL-HH-HL */\ |
| 5212 | rev16 texture_mask_rev, texture_mask; \ |
| 5213 | vmov block_masks, left_block_mask, right_block_mask; \ |
| 5214 | \ |
| 5215 | /* texture_mask = HH-HL-HL-WL */\ |
| 5216 | bfi texture_mask, texture_mask_rev, #4, #4; \ |
| 5217 | /* texture_mask_rev = 00-00-00-WH */\ |
| 5218 | mov texture_mask_rev, texture_mask_rev, lsr #12; \ |
| 5219 | \ |
| 5220 | /* texture_mask = HH-WH-HL-WL */\ |
| 5221 | bfi texture_mask, texture_mask_rev, #8, #4; \ |
| 5222 | setup_sprite_get_left_block_mask##x4mode(); \ |
| 5223 | \ |
| 5224 | mov control_mask, #0; \ |
| 5225 | setup_sprite_compare_left_block_mask##x4mode(); \ |
| 5226 | \ |
| 5227 | setup_sprite_get_right_block_mask##x4mode(); \ |
| 5228 | orreq control_mask, control_mask, #0x4; \ |
| 5229 | \ |
| 5230 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \ |
| 5231 | setup_sprite_compare_right_block_mask##x4mode(); \ |
| 5232 | \ |
| 5233 | orreq control_mask, control_mask, #0x8; \ |
| 5234 | cmp tile_width, #1; \ |
| 5235 | \ |
| 5236 | add block, psx_gpu, #psx_gpu_blocks_offset; \ |
| 5237 | orreq control_mask, control_mask, #0x1; \ |
| 5238 | \ |
| 5239 | cmp tile_height, #1; \ |
| 5240 | add block, block, num_blocks, lsl #6; \ |
| 5241 | \ |
| 5242 | orreq control_mask, control_mask, #0x2; \ |
| 5243 | JT_OP_REL(9f, control_mask, temp); \ |
| 5244 | JT_OP(ldr pc, [pc, control_mask, lsl #2]); \ |
| 5245 | nop; \ |
| 5246 | \ |
| 5247 | 9: \ |
| 5248 | .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \ |
| 5249 | .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \ |
| 5250 | .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \ |
| 5251 | .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \ |
| 5252 | .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \ |
| 5253 | .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \ |
| 5254 | .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \ |
| 5255 | .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\ |
| 5256 | .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \ |
| 5257 | .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \ |
| 5258 | .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \ |
| 5259 | .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \ |
| 5260 | .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \ |
| 5261 | .word 0x00000000; \ |
| 5262 | .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \ |
| 5263 | |
| 5264 | |
| 5265 | setup_sprite_tiled_builder(4bpp,); |
| 5266 | setup_sprite_tiled_builder(8bpp,); |
| 5267 | |
| 5268 | #undef draw_mask_fb_ptr_left |
| 5269 | #undef draw_mask_fb_ptr_right |
| 5270 | |
| 5271 | setup_sprite_tiled_builder(4bpp, _4x); |
| 5272 | setup_sprite_tiled_builder(8bpp, _4x); |
| 5273 | |
| 5274 | |
| 5275 | #undef block_ptr |
| 5276 | #undef num_blocks |
| 5277 | #undef clut_ptr |
| 5278 | |
| 5279 | #define psx_gpu r0 |
| 5280 | #define block_ptr r0 |
| 5281 | #define num_blocks r1 |
| 5282 | #define clut_ptr r2 |
| 5283 | #define texel_shift_mask r3 |
| 5284 | #define block_pixels_a r4 |
| 5285 | #define block_pixels_b r5 |
| 5286 | #define texel_0 r6 |
| 5287 | #define texel_2 r7 |
| 5288 | #define texel_4 r8 |
| 5289 | #define texel_6 r9 |
| 5290 | #define texel_1 r10 |
| 5291 | #define texel_3 r11 |
| 5292 | #define texel_5 r12 |
| 5293 | #define texel_7 r14 |
| 5294 | #define texels_01 r6 |
| 5295 | #define texels_23 r7 |
| 5296 | #define texels_45 r8 |
| 5297 | #define texels_67 r9 |
| 5298 | |
| 5299 | function(texture_sprite_blocks_8bpp) |
| 5300 | stmdb sp!, { r4 - r11, r14 } |
| 5301 | movw texel_shift_mask, #(0xFF << 1) |
| 5302 | |
| 5303 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5304 | ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset] |
| 5305 | |
| 5306 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 5307 | ldr block_pixels_a, [block_ptr, #16] |
| 5308 | |
| 5309 | 0: |
| 5310 | and texel_0, texel_shift_mask, block_pixels_a, lsl #1 |
| 5311 | ldr block_pixels_b, [block_ptr, #20] |
| 5312 | |
| 5313 | and texel_1, texel_shift_mask, block_pixels_a, lsr #7 |
| 5314 | ldrh texel_0, [clut_ptr, texel_0] |
| 5315 | |
| 5316 | and texel_2, texel_shift_mask, block_pixels_a, lsr #15 |
| 5317 | ldrh texel_1, [clut_ptr, texel_1] |
| 5318 | |
| 5319 | and texel_3, texel_shift_mask, block_pixels_a, lsr #23 |
| 5320 | ldr block_pixels_a, [block_ptr, #(64 + 16)] |
| 5321 | |
| 5322 | ldrh texel_2, [clut_ptr, texel_2] |
| 5323 | and texel_4, texel_shift_mask, block_pixels_b, lsl #1 |
| 5324 | |
| 5325 | ldrh texel_3, [clut_ptr, texel_3] |
| 5326 | and texel_5, texel_shift_mask, block_pixels_b, lsr #7 |
| 5327 | |
| 5328 | ldrh texel_4, [clut_ptr, texel_4] |
| 5329 | and texel_6, texel_shift_mask, block_pixels_b, lsr #15 |
| 5330 | |
| 5331 | ldrh texel_5, [clut_ptr, texel_5] |
| 5332 | and texel_7, texel_shift_mask, block_pixels_b, lsr #23 |
| 5333 | |
| 5334 | ldrh texel_6, [clut_ptr, texel_6] |
| 5335 | orr texels_01, texel_0, texel_1, lsl #16 |
| 5336 | |
| 5337 | ldrh texel_7, [clut_ptr, texel_7] |
| 5338 | orr texels_23, texel_2, texel_3, lsl #16 |
| 5339 | |
| 5340 | orr texels_45, texel_4, texel_5, lsl #16 |
| 5341 | str texels_01, [block_ptr, #0] |
| 5342 | |
| 5343 | orr texels_67, texel_6, texel_7, lsl #16 |
| 5344 | str texels_23, [block_ptr, #4] |
| 5345 | |
| 5346 | subs num_blocks, num_blocks, #1 |
| 5347 | str texels_45, [block_ptr, #8] |
| 5348 | |
| 5349 | str texels_67, [block_ptr, #12] |
| 5350 | add block_ptr, block_ptr, #64 |
| 5351 | |
| 5352 | bne 0b |
| 5353 | |
| 5354 | ldmia sp!, { r4 - r11, pc } |
| 5355 | |
| 5356 | |
| 5357 | #undef width_rounded |
| 5358 | #undef texture_mask |
| 5359 | #undef num_blocks |
| 5360 | #undef texture_offset |
| 5361 | #undef texels_low |
| 5362 | #undef texels_high |
| 5363 | #undef texels_wide_low |
| 5364 | #undef texels_wide_high |
| 5365 | #undef texels_wide |
| 5366 | #undef fb_ptr2 |
| 5367 | #undef temp |
| 5368 | |
| 5369 | #define psx_gpu r0 |
| 5370 | #define x r1 |
| 5371 | #define y r2 |
| 5372 | #define u r3 |
| 5373 | #define v r4 |
| 5374 | #define width r5 |
| 5375 | #define height r6 |
| 5376 | #define left_offset r8 |
| 5377 | #define width_rounded r9 |
| 5378 | #define right_width r10 |
| 5379 | |
| 5380 | #define block_width r11 |
| 5381 | |
| 5382 | #define texture_offset_base r1 |
| 5383 | #define texture_mask r2 |
| 5384 | #define texture_page_ptr r3 |
| 5385 | #define num_blocks r4 |
| 5386 | #define block r5 |
| 5387 | #define fb_ptr r7 |
| 5388 | #define texture_offset r8 |
| 5389 | #define blocks_remaining r9 |
| 5390 | #define fb_ptr2 r10 |
| 5391 | #define fb_ptr_pitch r12 |
| 5392 | #define texture_block_ptr r14 |
| 5393 | |
| 5394 | #define texture_mask_width r2 |
| 5395 | #define texture_mask_height r3 |
| 5396 | #define left_mask_bits r4 |
| 5397 | #define right_mask_bits r5 |
| 5398 | |
| 5399 | |
| 5400 | #undef block_masks |
| 5401 | #undef block_masks_shifted |
| 5402 | #undef texels |
| 5403 | |
| 5404 | #define block_masks d0 |
| 5405 | #define block_masks_shifted d1 |
| 5406 | #define draw_mask_fb_ptr d2 |
| 5407 | #define texels q2 |
| 5408 | |
| 5409 | #define draw_mask_fb_ptr_a d2 |
| 5410 | #define draw_mask_fb_ptr_b d3 |
| 5411 | #define texels_low d4 |
| 5412 | #define texels_high d5 |
| 5413 | #define texels_wide_low d6 |
| 5414 | #define texels_wide_high d7 |
| 5415 | #define texels_wide q3 |
| 5416 | |
| 5417 | |
| 5418 | setup_sprites_16bpp_flush: |
| 5419 | vpush { d0 - d3 } |
| 5420 | |
| 5421 | stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 } |
| 5422 | bl flush_render_block_buffer |
| 5423 | ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 } |
| 5424 | |
| 5425 | vpop { d0 - d3 } |
| 5426 | |
| 5427 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5428 | mov num_blocks, block_width |
| 5429 | |
| 5430 | bx lr |
| 5431 | |
| 5432 | function(setup_sprite_16bpp) |
| 5433 | stmdb sp!, { r4 - r11, r14 } |
| 5434 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset] |
| 5435 | |
| 5436 | ldr v, [sp, #36] |
| 5437 | add fb_ptr, fb_ptr, y, lsl #11 |
| 5438 | |
| 5439 | ldr width, [sp, #40] |
| 5440 | add fb_ptr, fb_ptr, x, lsl #1 |
| 5441 | |
| 5442 | ldr height, [sp, #44] |
| 5443 | and left_offset, u, #0x7 |
| 5444 | |
| 5445 | add texture_offset_base, u, u |
| 5446 | add width_rounded, width, #7 |
| 5447 | |
| 5448 | add texture_offset_base, texture_offset_base, v, lsl #11 |
| 5449 | mov left_mask_bits, #0xFF |
| 5450 | |
| 5451 | ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset] |
| 5452 | add width_rounded, width_rounded, left_offset |
| 5453 | |
| 5454 | ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset] |
| 5455 | sub fb_ptr, fb_ptr, left_offset, lsl #1 |
| 5456 | |
| 5457 | add texture_mask, texture_mask_width, texture_mask_width |
| 5458 | mov right_mask_bits, #0xFE |
| 5459 | |
| 5460 | and right_width, width_rounded, #0x7 |
| 5461 | mvn left_mask_bits, left_mask_bits, lsl left_offset |
| 5462 | |
| 5463 | add texture_mask, texture_mask, texture_mask_height, lsl #11 |
| 5464 | mov block_width, width_rounded, lsr #3 |
| 5465 | |
| 5466 | mov right_mask_bits, right_mask_bits, lsl right_width |
| 5467 | movw fb_ptr_pitch, #(2048 + 16) |
| 5468 | |
| 5469 | sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4 |
| 5470 | vmov block_masks, left_mask_bits, right_mask_bits |
| 5471 | |
| 5472 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5473 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5474 | |
| 5475 | bic texture_offset_base, texture_offset_base, #0xF |
| 5476 | cmp block_width, #1 |
| 5477 | |
| 5478 | ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] |
| 5479 | add block, block, num_blocks, lsl #6 |
| 5480 | |
| 5481 | bne 0f |
| 5482 | |
| 5483 | vext.32 block_masks_shifted, block_masks, block_masks, #1 |
| 5484 | vorr.u32 block_masks, block_masks, block_masks_shifted |
| 5485 | vdup.u8 draw_mask_fb_ptr, block_masks[0] |
| 5486 | |
| 5487 | 1: |
| 5488 | add num_blocks, num_blocks, #1 |
| 5489 | cmp num_blocks, #MAX_BLOCKS |
| 5490 | blgt setup_sprites_16bpp_flush |
| 5491 | |
| 5492 | and texture_block_ptr, texture_offset_base, texture_mask |
| 5493 | subs height, height, #1 |
| 5494 | |
| 5495 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5496 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5497 | |
| 5498 | vst1.u32 { texels }, [block, :128] |
| 5499 | add block, block, #40 |
| 5500 | |
| 5501 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5502 | pld [fb_ptr] |
| 5503 | |
| 5504 | vst1.u32 { draw_mask_fb_ptr }, [block, :64] |
| 5505 | |
| 5506 | add block, block, #24 |
| 5507 | add texture_offset_base, texture_offset_base, #2048 |
| 5508 | add fb_ptr, fb_ptr, #2048 |
| 5509 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5510 | bne 1b |
| 5511 | |
| 5512 | ldmia sp!, { r4 - r11, pc } |
| 5513 | |
| 5514 | 0: |
| 5515 | add num_blocks, num_blocks, block_width |
| 5516 | mov texture_offset, texture_offset_base |
| 5517 | |
| 5518 | cmp num_blocks, #MAX_BLOCKS |
| 5519 | blgt setup_sprites_16bpp_flush |
| 5520 | |
| 5521 | add texture_offset_base, texture_offset_base, #2048 |
| 5522 | and texture_block_ptr, texture_offset, texture_mask |
| 5523 | |
| 5524 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5525 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5526 | |
| 5527 | vst1.u32 { texels }, [block, :128] |
| 5528 | add block, block, #40 |
| 5529 | |
| 5530 | vdup.u8 draw_mask_fb_ptr, block_masks[0] |
| 5531 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5532 | pld [fb_ptr] |
| 5533 | |
| 5534 | vst1.u32 { draw_mask_fb_ptr }, [block, :64] |
| 5535 | subs blocks_remaining, block_width, #2 |
| 5536 | |
| 5537 | add texture_offset, texture_offset, #16 |
| 5538 | add fb_ptr, fb_ptr, #16 |
| 5539 | |
| 5540 | vmov.u8 draw_mask_fb_ptr, #0 |
| 5541 | |
| 5542 | add block, block, #24 |
| 5543 | beq 2f |
| 5544 | |
| 5545 | 1: |
| 5546 | and texture_block_ptr, texture_offset, texture_mask |
| 5547 | subs blocks_remaining, blocks_remaining, #1 |
| 5548 | |
| 5549 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5550 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5551 | |
| 5552 | vst1.u32 { texels }, [block, :128] |
| 5553 | add block, block, #40 |
| 5554 | |
| 5555 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5556 | pld [fb_ptr] |
| 5557 | |
| 5558 | vst1.u32 { draw_mask_fb_ptr }, [block, :64] |
| 5559 | |
| 5560 | add texture_offset, texture_offset, #16 |
| 5561 | add fb_ptr, fb_ptr, #16 |
| 5562 | |
| 5563 | add block, block, #24 |
| 5564 | bne 1b |
| 5565 | |
| 5566 | 2: |
| 5567 | and texture_block_ptr, texture_offset, texture_mask |
| 5568 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5569 | |
| 5570 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5571 | vdup.u8 draw_mask_fb_ptr, block_masks[4] |
| 5572 | |
| 5573 | vst1.u32 { texels }, [block, :128] |
| 5574 | add block, block, #40 |
| 5575 | |
| 5576 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5577 | vst1.u32 { draw_mask_fb_ptr }, [block, :64] |
| 5578 | |
| 5579 | add block, block, #24 |
| 5580 | subs height, height, #1 |
| 5581 | |
| 5582 | add fb_ptr, fb_ptr, fb_ptr_pitch |
| 5583 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5584 | |
| 5585 | bne 0b |
| 5586 | |
| 5587 | ldmia sp!, { r4 - r11, pc } |
| 5588 | |
| 5589 | |
| 5590 | // 4x version |
| 5591 | // FIXME: duplicate code with normal version :( |
| 5592 | #undef draw_mask_fb_ptr |
| 5593 | |
| 5594 | function(setup_sprite_16bpp_4x) |
| 5595 | stmdb sp!, { r4 - r11, r14 } |
| 5596 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset] |
| 5597 | |
| 5598 | ldr v, [sp, #36] |
| 5599 | add fb_ptr, fb_ptr, y, lsl #11 |
| 5600 | |
| 5601 | ldr width, [sp, #40] |
| 5602 | add fb_ptr, fb_ptr, x, lsl #1 |
| 5603 | |
| 5604 | ldr height, [sp, #44] |
| 5605 | and left_offset, u, #0x7 |
| 5606 | |
| 5607 | add texture_offset_base, u, u |
| 5608 | add width_rounded, width, #7 |
| 5609 | |
| 5610 | add texture_offset_base, texture_offset_base, v, lsl #11 |
| 5611 | movw left_mask_bits, #0xFFFF |
| 5612 | |
| 5613 | ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset] |
| 5614 | add width_rounded, width_rounded, left_offset |
| 5615 | |
| 5616 | lsl left_offset, #1 |
| 5617 | |
| 5618 | ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset] |
| 5619 | sub fb_ptr, fb_ptr, left_offset, lsl #1 |
| 5620 | |
| 5621 | add texture_mask, texture_mask_width, texture_mask_width |
| 5622 | movw right_mask_bits, #0xFFFC |
| 5623 | |
| 5624 | and right_width, width_rounded, #0x7 |
| 5625 | mvn left_mask_bits, left_mask_bits, lsl left_offset |
| 5626 | |
| 5627 | lsl right_width, #1 |
| 5628 | |
| 5629 | add texture_mask, texture_mask, texture_mask_height, lsl #11 |
| 5630 | mov block_width, width_rounded, lsr #3 |
| 5631 | |
| 5632 | mov right_mask_bits, right_mask_bits, lsl right_width |
| 5633 | movw fb_ptr_pitch, #(2048 + 16) * 2 |
| 5634 | |
| 5635 | sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1 |
| 5636 | vmov block_masks, left_mask_bits, right_mask_bits |
| 5637 | |
| 5638 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5639 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5640 | |
| 5641 | bic texture_offset_base, texture_offset_base, #0xF |
| 5642 | cmp block_width, #1 |
| 5643 | |
| 5644 | ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] |
| 5645 | add block, block, num_blocks, lsl #6 |
| 5646 | |
| 5647 | lsl block_width, #2 |
| 5648 | bne 0f |
| 5649 | |
| 5650 | vext.32 block_masks_shifted, block_masks, block_masks, #1 |
| 5651 | vorr.u32 block_masks, block_masks, block_masks_shifted |
| 5652 | vdup.u8 draw_mask_fb_ptr_a, block_masks[0] |
| 5653 | vdup.u8 draw_mask_fb_ptr_b, block_masks[1] |
| 5654 | |
| 5655 | 1: |
| 5656 | add num_blocks, num_blocks, block_width |
| 5657 | cmp num_blocks, #MAX_BLOCKS |
| 5658 | blgt setup_sprites_16bpp_flush |
| 5659 | |
| 5660 | and texture_block_ptr, texture_offset_base, texture_mask |
| 5661 | subs height, height, #1 |
| 5662 | |
| 5663 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5664 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5665 | |
| 5666 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5667 | |
| 5668 | add texture_offset_base, texture_offset_base, #2048 |
| 5669 | add fb_ptr, fb_ptr, #2048*2 |
| 5670 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5671 | bne 1b |
| 5672 | |
| 5673 | ldmia sp!, { r4 - r11, pc } |
| 5674 | |
| 5675 | 0: |
| 5676 | add num_blocks, num_blocks, block_width |
| 5677 | mov texture_offset, texture_offset_base |
| 5678 | |
| 5679 | vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits |
| 5680 | vdup.u8 draw_mask_fb_ptr_b, block_masks[1] |
| 5681 | |
| 5682 | cmp num_blocks, #MAX_BLOCKS |
| 5683 | blgt setup_sprites_16bpp_flush |
| 5684 | |
| 5685 | add texture_offset_base, texture_offset_base, #2048 |
| 5686 | and texture_block_ptr, texture_offset, texture_mask |
| 5687 | |
| 5688 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5689 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5690 | |
| 5691 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5692 | |
| 5693 | subs blocks_remaining, block_width, #2*4 |
| 5694 | add texture_offset, texture_offset, #16 |
| 5695 | |
| 5696 | vmov.u8 draw_mask_fb_ptr_a, #0 |
| 5697 | vmov.u8 draw_mask_fb_ptr_b, #0 |
| 5698 | |
| 5699 | add fb_ptr, fb_ptr, #16*2 |
| 5700 | beq 2f |
| 5701 | |
| 5702 | 1: |
| 5703 | and texture_block_ptr, texture_offset, texture_mask |
| 5704 | subs blocks_remaining, blocks_remaining, #4 |
| 5705 | |
| 5706 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5707 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5708 | |
| 5709 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5710 | add texture_offset, texture_offset, #16 |
| 5711 | |
| 5712 | add fb_ptr, fb_ptr, #16*2 |
| 5713 | bgt 1b |
| 5714 | |
| 5715 | 2: |
| 5716 | vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits |
| 5717 | vdup.u8 draw_mask_fb_ptr_b, block_masks[5] |
| 5718 | |
| 5719 | and texture_block_ptr, texture_offset, texture_mask |
| 5720 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5721 | |
| 5722 | vld1.u32 { texels }, [texture_block_ptr, :128] |
| 5723 | |
| 5724 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5725 | subs height, height, #1 |
| 5726 | |
| 5727 | add fb_ptr, fb_ptr, fb_ptr_pitch |
| 5728 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5729 | |
| 5730 | bne 0b |
| 5731 | |
| 5732 | ldmia sp!, { r4 - r11, pc } |
| 5733 | |
| 5734 | |
| 5735 | #undef width |
| 5736 | #undef right_width |
| 5737 | #undef right_mask_bits |
| 5738 | #undef color |
| 5739 | #undef height |
| 5740 | #undef blocks_remaining |
| 5741 | #undef colors |
| 5742 | #undef right_mask |
| 5743 | #undef test_mask |
| 5744 | #undef draw_mask |
| 5745 | |
| 5746 | #define psx_gpu r0 |
| 5747 | #define x r1 |
| 5748 | #define y r2 |
| 5749 | #define width r3 |
| 5750 | #define right_width r5 |
| 5751 | #define right_mask_bits r6 |
| 5752 | #define fb_ptr r7 |
| 5753 | #define color r8 |
| 5754 | #define height r9 |
| 5755 | #define fb_ptr_pitch r12 |
| 5756 | |
| 5757 | // referenced by setup_sprites_16bpp_flush |
| 5758 | #define num_blocks r4 |
| 5759 | #define block r5 |
| 5760 | #define block_width r11 |
| 5761 | |
| 5762 | #define color_r r1 |
| 5763 | #define color_g r2 |
| 5764 | #define color_b r8 |
| 5765 | #define blocks_remaining r6 |
| 5766 | |
| 5767 | #define colors q0 |
| 5768 | #define right_mask q1 |
| 5769 | #define test_mask q2 |
| 5770 | #define draw_mask q2 |
| 5771 | #define draw_mask_bits_fb_ptr d6 |
| 5772 | |
| 5773 | |
| 5774 | .align 3 |
| 5775 | |
| 5776 | function(setup_sprite_untextured) |
| 5777 | ldrh r12, [psx_gpu, #psx_gpu_render_state_offset] |
| 5778 | tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \ |
| 5779 | | RENDER_FLAGS_BLEND) |
| 5780 | ldrbeq r12, [psx_gpu, #psx_gpu_render_mode_offset] |
| 5781 | tsteq r12, #RENDER_INTERLACE_ENABLED |
| 5782 | beq setup_sprite_untextured_simple |
| 5783 | |
| 5784 | stmdb sp!, { r4 - r11, r14 } |
| 5785 | |
| 5786 | ldr width, [sp, #40] |
| 5787 | ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset] |
| 5788 | |
| 5789 | ldr height, [sp, #44] |
| 5790 | add fb_ptr, fb_ptr, y, lsl #11 |
| 5791 | |
| 5792 | add fb_ptr, fb_ptr, x, lsl #1 |
| 5793 | sub right_width, width, #1 |
| 5794 | |
| 5795 | ldr color, [sp, #48] |
| 5796 | and right_width, #7 |
| 5797 | |
| 5798 | add block_width, width, #7 |
| 5799 | add right_width, #1 |
| 5800 | |
| 5801 | lsr block_width, #3 |
| 5802 | mov right_mask_bits, #0xff |
| 5803 | |
| 5804 | sub fb_ptr_pitch, block_width, #1 |
| 5805 | lsl right_mask_bits, right_width |
| 5806 | |
| 5807 | lsl fb_ptr_pitch, #3+1 |
| 5808 | ubfx color_r, color, #3, #5 |
| 5809 | |
| 5810 | rsb fb_ptr_pitch, #1024*2 |
| 5811 | ubfx color_g, color, #11, #5 |
| 5812 | |
| 5813 | vld1.u32 { test_mask }, [psx_gpu, :128] |
| 5814 | ubfx color_b, color, #19, #5 |
| 5815 | |
| 5816 | vdup.u16 right_mask, right_mask_bits |
| 5817 | orr color, color_r, color_b, lsl #10 |
| 5818 | |
| 5819 | ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5820 | orr color, color, color_g, lsl #5 |
| 5821 | |
| 5822 | vtst.u16 right_mask, right_mask, test_mask |
| 5823 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5824 | |
| 5825 | vdup.u16 colors, color |
| 5826 | add block, block, num_blocks, lsl #6 |
| 5827 | |
| 5828 | |
| 5829 | setup_sprite_untextured_height_loop: |
| 5830 | add num_blocks, block_width |
| 5831 | sub blocks_remaining, block_width, #1 |
| 5832 | |
| 5833 | cmp num_blocks, #MAX_BLOCKS |
| 5834 | blgt setup_sprites_16bpp_flush |
| 5835 | |
| 5836 | cmp blocks_remaining, #0 |
| 5837 | ble 1f |
| 5838 | |
| 5839 | vmov.u8 draw_mask, #0 /* zero_mask */ |
| 5840 | vmov.u8 draw_mask_bits_fb_ptr, #0 |
| 5841 | |
| 5842 | 0: |
| 5843 | vst1.u32 { draw_mask }, [block, :128]! |
| 5844 | subs blocks_remaining, #1 |
| 5845 | |
| 5846 | vst1.u32 { colors }, [block, :128] |
| 5847 | add block, block, #24 |
| 5848 | |
| 5849 | vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr |
| 5850 | vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64] |
| 5851 | |
| 5852 | add block, block, #24 |
| 5853 | add fb_ptr, #8*2 |
| 5854 | bgt 0b |
| 5855 | |
| 5856 | 1: |
| 5857 | vst1.u32 { right_mask }, [block, :128]! |
| 5858 | subs height, #1 |
| 5859 | |
| 5860 | vst1.u32 { colors }, [block, :128] |
| 5861 | add block, block, #24 |
| 5862 | |
| 5863 | vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr |
| 5864 | vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64] |
| 5865 | |
| 5866 | add block, block, #24 |
| 5867 | add fb_ptr, fb_ptr_pitch |
| 5868 | |
| 5869 | strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] |
| 5870 | bgt setup_sprite_untextured_height_loop |
| 5871 | |
| 5872 | ldmia sp!, { r4 - r11, pc } |
| 5873 | |
| 5874 | |
| 5875 | |
| 5876 | #undef texture_page_ptr |
| 5877 | #undef vram_ptr |
| 5878 | #undef dirty_textures_mask |
| 5879 | #undef current_texture_mask |
| 5880 | |
| 5881 | #define psx_gpu r0 |
| 5882 | #define current_texture_page r1 |
| 5883 | #define texture_page_ptr r2 |
| 5884 | #define vram_ptr_a r3 |
| 5885 | #define current_texture_page_x r12 |
| 5886 | #define current_texture_page_y r4 |
| 5887 | #define dirty_textures_mask r5 |
| 5888 | #define tile_y r6 |
| 5889 | #define tile_x r7 |
| 5890 | #define sub_y r8 |
| 5891 | #define current_texture_mask r9 |
| 5892 | #define c_4096 r10 |
| 5893 | #define vram_ptr_b r11 |
| 5894 | |
| 5895 | #define texel_block_a d0 |
| 5896 | #define texel_block_b d1 |
| 5897 | #define texel_block_expanded_a q1 |
| 5898 | #define texel_block_expanded_b q2 |
| 5899 | #define texel_block_expanded_ab q2 |
| 5900 | #define texel_block_expanded_c q3 |
| 5901 | #define texel_block_expanded_d q4 |
| 5902 | #define texel_block_expanded_cd q3 |
| 5903 | |
| 5904 | function(update_texture_4bpp_cache) |
| 5905 | stmdb sp!, { r4 - r11, r14 } |
| 5906 | vpush { q0 - q3 } |
| 5907 | |
| 5908 | ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset] |
| 5909 | |
| 5910 | ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset] |
| 5911 | ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset] |
| 5912 | |
| 5913 | and current_texture_page_x, current_texture_page, #0xF |
| 5914 | ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset] |
| 5915 | |
| 5916 | mov current_texture_page_y, current_texture_page, lsr #4 |
| 5917 | ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset] |
| 5918 | |
| 5919 | add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19 |
| 5920 | mov tile_y, #16 |
| 5921 | |
| 5922 | add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7 |
| 5923 | bic dirty_textures_mask, current_texture_mask |
| 5924 | |
| 5925 | mov tile_x, #16 |
| 5926 | str dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset] |
| 5927 | |
| 5928 | mov sub_y, #8 |
| 5929 | movw c_4096, #4096 |
| 5930 | |
| 5931 | add vram_ptr_b, vram_ptr_a, #2048 |
| 5932 | |
| 5933 | 0: |
| 5934 | vld1.u32 { texel_block_a }, [vram_ptr_a, :64], c_4096 |
| 5935 | vld1.u32 { texel_block_b }, [vram_ptr_b, :64], c_4096 |
| 5936 | |
| 5937 | vmovl.u8 texel_block_expanded_a, texel_block_a |
| 5938 | vshll.u8 texel_block_expanded_b, texel_block_a, #4 |
| 5939 | vmovl.u8 texel_block_expanded_c, texel_block_b |
| 5940 | vshll.u8 texel_block_expanded_d, texel_block_b, #4 |
| 5941 | |
| 5942 | vbic.u16 texel_block_expanded_a, #0x00F0 |
| 5943 | vbic.u16 texel_block_expanded_b, #0x00F0 |
| 5944 | vbic.u16 texel_block_expanded_c, #0x00F0 |
| 5945 | vbic.u16 texel_block_expanded_d, #0x00F0 |
| 5946 | |
| 5947 | vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \ |
| 5948 | texel_block_expanded_b |
| 5949 | vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \ |
| 5950 | texel_block_expanded_d |
| 5951 | |
| 5952 | vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \ |
| 5953 | [texture_page_ptr, :256]! |
| 5954 | |
| 5955 | subs sub_y, sub_y, #1 |
| 5956 | bne 0b |
| 5957 | |
| 5958 | mov sub_y, #8 |
| 5959 | add vram_ptr_a, vram_ptr_a, #8 |
| 5960 | add vram_ptr_b, vram_ptr_b, #8 |
| 5961 | |
| 5962 | sub vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 5963 | sub vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 5964 | |
| 5965 | subs tile_x, tile_x, #1 |
| 5966 | bne 0b |
| 5967 | |
| 5968 | mov tile_x, #16 |
| 5969 | add vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 5970 | add vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 5971 | |
| 5972 | sub vram_ptr_a, vram_ptr_a, #(8 * 16) |
| 5973 | sub vram_ptr_b, vram_ptr_b, #(8 * 16) |
| 5974 | |
| 5975 | subs tile_y, tile_y, #1 |
| 5976 | bne 0b |
| 5977 | |
| 5978 | vpop { q0 - q3 } |
| 5979 | ldmia sp!, { r4 - r11, pc } |
| 5980 | |
| 5981 | |
| 5982 | #undef current_texture_page |
| 5983 | |
| 5984 | #define psx_gpu r0 |
| 5985 | #define texture_page r1 |
| 5986 | #define texture_page_ptr r2 |
| 5987 | #define vram_ptr_a r3 |
| 5988 | #define texture_page_x r12 |
| 5989 | #define texture_page_y r4 |
| 5990 | #define current_texture_page r5 |
| 5991 | #define tile_y r6 |
| 5992 | #define tile_x r7 |
| 5993 | #define sub_y r8 |
| 5994 | #define c_4096 r10 |
| 5995 | #define vram_ptr_b r11 |
| 5996 | |
| 5997 | |
| 5998 | #undef texels_a |
| 5999 | #undef texels_b |
| 6000 | |
| 6001 | #define texels_a q0 |
| 6002 | #define texels_b q1 |
| 6003 | #define texels_c q2 |
| 6004 | #define texels_d q3 |
| 6005 | |
| 6006 | |
| 6007 | function(update_texture_8bpp_cache_slice) |
| 6008 | stmdb sp!, { r4 - r11, r14 } |
| 6009 | vpush { q0 - q3 } |
| 6010 | |
| 6011 | ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset] |
| 6012 | ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset] |
| 6013 | |
| 6014 | ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset] |
| 6015 | mov tile_y, #16 |
| 6016 | |
| 6017 | and texture_page_x, texture_page, #0xF |
| 6018 | mov texture_page_y, texture_page, lsr #4 |
| 6019 | |
| 6020 | add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7 |
| 6021 | mov tile_x, #8 |
| 6022 | |
| 6023 | add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19 |
| 6024 | eor current_texture_page, current_texture_page, texture_page |
| 6025 | |
| 6026 | ands current_texture_page, current_texture_page, #0x1 |
| 6027 | mov sub_y, #4 |
| 6028 | |
| 6029 | addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16) |
| 6030 | movw c_4096, #4096 |
| 6031 | |
| 6032 | add vram_ptr_b, vram_ptr_a, #2048 |
| 6033 | |
| 6034 | 0: |
| 6035 | vld1.u32 { texels_a }, [vram_ptr_a, :128], c_4096 |
| 6036 | vld1.u32 { texels_b }, [vram_ptr_b, :128], c_4096 |
| 6037 | vld1.u32 { texels_c }, [vram_ptr_a, :128], c_4096 |
| 6038 | vld1.u32 { texels_d }, [vram_ptr_b, :128], c_4096 |
| 6039 | |
| 6040 | vst1.u32 { texels_a, texels_b }, [texture_page_ptr, :256]! |
| 6041 | vst1.u32 { texels_c, texels_d }, [texture_page_ptr, :256]! |
| 6042 | |
| 6043 | subs sub_y, sub_y, #1 |
| 6044 | bne 0b |
| 6045 | |
| 6046 | mov sub_y, #4 |
| 6047 | |
| 6048 | add vram_ptr_a, vram_ptr_a, #16 |
| 6049 | add vram_ptr_b, vram_ptr_b, #16 |
| 6050 | |
| 6051 | sub vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 6052 | sub vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 6053 | |
| 6054 | subs tile_x, tile_x, #1 |
| 6055 | bne 0b |
| 6056 | |
| 6057 | mov tile_x, #8 |
| 6058 | |
| 6059 | add vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 6060 | add vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 6061 | |
| 6062 | sub vram_ptr_a, vram_ptr_a, #(8 * 16) |
| 6063 | sub vram_ptr_b, vram_ptr_b, #(8 * 16) |
| 6064 | |
| 6065 | subs tile_y, tile_y, #1 |
| 6066 | add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16) |
| 6067 | |
| 6068 | bne 0b |
| 6069 | |
| 6070 | vpop { q0 - q3 } |
| 6071 | ldmia sp!, { r4 - r11, pc } |
| 6072 | |
| 6073 | |
| 6074 | /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */ |
| 6075 | function(scale2x_tiles8) |
| 6076 | push { r4, r14 } |
| 6077 | |
| 6078 | mov r4, r1 |
| 6079 | add r12, r0, #1024*2 |
| 6080 | mov r14, r2 |
| 6081 | |
| 6082 | 0: |
| 6083 | vld1.u16 { q0 }, [r1, :128]! |
| 6084 | vld1.u16 { q2 }, [r1, :128]! |
| 6085 | vmov q1, q0 |
| 6086 | vmov q3, q2 |
| 6087 | vzip.16 q0, q1 |
| 6088 | vzip.16 q2, q3 |
| 6089 | subs r14, #2 |
| 6090 | vst1.u16 { q0, q1 }, [r0, :128]! |
| 6091 | vst1.u16 { q0, q1 }, [r12, :128]! |
| 6092 | blt 1f |
| 6093 | vst1.u16 { q2, q3 }, [r0, :128]! |
| 6094 | vst1.u16 { q2, q3 }, [r12, :128]! |
| 6095 | bgt 0b |
| 6096 | 1: |
| 6097 | subs r3, #1 |
| 6098 | mov r14, r2 |
| 6099 | add r0, #1024*2*2 |
| 6100 | add r4, #1024*2 |
| 6101 | sub r0, r0, r2, lsl #4+1 |
| 6102 | mov r1, r4 |
| 6103 | add r12, r0, #1024*2 |
| 6104 | bgt 0b |
| 6105 | nop |
| 6106 | |
| 6107 | pop { r4, pc } |
| 6108 | |
| 6109 | // vim:filetype=armasm |