| 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 | |
| 35 | #define psx_gpu r0 |
| 36 | #define v_a r1 |
| 37 | #define v_b r2 |
| 38 | #define v_c r3 |
| 39 | |
| 40 | #define x0 r4 |
| 41 | #define x1 r5 |
| 42 | #define x2 r6 |
| 43 | #define x0_x1 r5 |
| 44 | #define x1_x2 r6 |
| 45 | #define y0 r7 |
| 46 | #define y1 r8 |
| 47 | #define y2 r9 |
| 48 | #define y0_y1 r7 |
| 49 | #define y1_y2 r8 |
| 50 | #define b0 r9 |
| 51 | #define b1 r10 |
| 52 | #define b2 r11 |
| 53 | #define b0_b1 r10 |
| 54 | #define b1_b2 r11 |
| 55 | |
| 56 | |
| 57 | #define area_r_s r5 |
| 58 | |
| 59 | #define g_bx0 r2 |
| 60 | #define g_bx r3 |
| 61 | #define g_bx2 r4 |
| 62 | #define g_bx3 r5 |
| 63 | #define b_base r6 |
| 64 | #define g_by r8 |
| 65 | |
| 66 | #define gs_bx r7 |
| 67 | #define gs_by r10 |
| 68 | |
| 69 | #define ga_bx g_bx |
| 70 | #define ga_by g_by |
| 71 | |
| 72 | #define gw_bx_h g_bx |
| 73 | #define gw_by_h g_by |
| 74 | |
| 75 | #define gw_bx_l r11 |
| 76 | #define gw_by_l gw_bx_l |
| 77 | |
| 78 | #define store_a r0 |
| 79 | #define store_b r1 |
| 80 | #define store_inc r5 |
| 81 | |
| 82 | |
| 83 | #define v0 q0 |
| 84 | #define uvrgb0 d0 |
| 85 | #define x0_y0 d1 |
| 86 | |
| 87 | #define v1 q1 |
| 88 | #define uvrgb1 d2 |
| 89 | #define x1_y1 d3 |
| 90 | |
| 91 | #define v2 q2 |
| 92 | #define uvrgb2 d4 |
| 93 | #define x2_y2 d5 |
| 94 | |
| 95 | #define x0_ab q3 |
| 96 | #define uvrg_xxxx0 q3 |
| 97 | #define uvrg0 d6 |
| 98 | #define xxxx0 d7 |
| 99 | |
| 100 | #define x1_ab q4 |
| 101 | #define uvrg_xxxx1 q4 |
| 102 | #define uvrg1 d8 |
| 103 | #define xxxx1 d9 |
| 104 | |
| 105 | #define x2_ab q5 |
| 106 | #define uvrg_xxxx2 q5 |
| 107 | #define uvrg2 d10 |
| 108 | #define xxxx2 d11 |
| 109 | |
| 110 | #define y0_ab q6 |
| 111 | #define yyyy_uvrg0 q6 |
| 112 | #define yyyy0 d12 |
| 113 | #define uvrg0b d13 |
| 114 | |
| 115 | #define y1_ab q7 |
| 116 | #define yyyy_uvrg1 q7 |
| 117 | #define yyyy1 d14 |
| 118 | #define uvrg1b d15 |
| 119 | |
| 120 | #define y2_ab q8 |
| 121 | #define yyyy_uvrg2 q8 |
| 122 | #define yyyy2 d16 |
| 123 | #define uvrg2b d17 |
| 124 | |
| 125 | #define d0_ab q9 |
| 126 | #define d0_a d18 |
| 127 | #define d0_b d19 |
| 128 | |
| 129 | #define d1_ab q10 |
| 130 | #define d1_a d20 |
| 131 | #define d1_b d21 |
| 132 | |
| 133 | #define d2_ab q11 |
| 134 | #define d2_a d22 |
| 135 | #define d2_b d23 |
| 136 | |
| 137 | #define d3_ab q12 |
| 138 | #define d3_a d24 |
| 139 | #define d3_b d25 |
| 140 | |
| 141 | #define ga_uvrg_x q1 |
| 142 | #define ga_uvrg_y q4 |
| 143 | |
| 144 | #define dx x0_x1 |
| 145 | #define dy y0_y1 |
| 146 | #define db b0_b1 |
| 147 | |
| 148 | #define uvrg_base q11 |
| 149 | |
| 150 | #define gs_uvrg_x q5 |
| 151 | #define gs_uvrg_y q6 |
| 152 | |
| 153 | #define g_uvrg_x q1 |
| 154 | #define ga_uv_x d2 |
| 155 | #define g_uv_x d2 |
| 156 | #define ga_rg_x d3 |
| 157 | #define g_rg_x d3 |
| 158 | |
| 159 | #define g_uvrg_y q4 |
| 160 | #define ga_uv_y d8 |
| 161 | #define g_uv_y d8 |
| 162 | #define ga_rg_y d9 |
| 163 | #define g_rg_y d9 |
| 164 | |
| 165 | #define gw_uv_x q1 |
| 166 | #define gw_rg_x q2 |
| 167 | #define gw_uv_y q4 |
| 168 | #define gw_rg_y q3 |
| 169 | |
| 170 | #define w_mask q9 |
| 171 | #define w_mask_l d18 |
| 172 | |
| 173 | #define r_shift q10 |
| 174 | |
| 175 | #define uvrg_dx0 q0 |
| 176 | #define uvrg_dx0l d0 |
| 177 | #define uvrg_dx0h d1 |
| 178 | |
| 179 | #define uvrg_dx1 q1 |
| 180 | #define uvrg_dx1l d2 |
| 181 | #define uvrg_dx1h d3 |
| 182 | |
| 183 | #define uvrg_dx2 q2 |
| 184 | #define uvrg_dx2l d4 |
| 185 | #define uvrg_dx2h d5 |
| 186 | |
| 187 | #define uvrg_dx3 q3 |
| 188 | #define uvrg_dx3l d6 |
| 189 | #define uvrg_dx3h d7 |
| 190 | |
| 191 | #define uvrgb_phase q13 |
| 192 | |
| 193 | .align 4 |
| 194 | |
| 195 | /* FIXME: users of this should be in psx_gpu instead */ |
| 196 | #ifndef __PIC__ |
| 197 | #define load_pointer(register, pointer) \ |
| 198 | movw register, :lower16:pointer; \ |
| 199 | movt register, :upper16:pointer; \ |
| 200 | |
| 201 | #else |
| 202 | #define load_pointer(register, pointer) \ |
| 203 | ldr register, =pointer \ |
| 204 | |
| 205 | #endif |
| 206 | |
| 207 | #define function(name) \ |
| 208 | .global name; \ |
| 209 | name: \ |
| 210 | |
| 211 | @ r0: psx_gpu |
| 212 | @ r1: v_a |
| 213 | @ r2: v_b |
| 214 | @ r3: v_c |
| 215 | |
| 216 | function(compute_all_gradients) |
| 217 | // First compute the triangle area reciprocal and shift. The division will |
| 218 | // happen concurrently with much of the work which follows. |
| 219 | @ r12 = psx_gpu->triangle_area |
| 220 | ldr r12, [ psx_gpu, #psx_gpu_triangle_area_offset ] |
| 221 | stmdb sp!, { r4 - r11, lr } |
| 222 | |
| 223 | @ load exponent of 62 into upper half of double |
| 224 | movw r4, #0 |
| 225 | clz r14, r12 @ r14 = shift |
| 226 | |
| 227 | movt r4, #((62 + 1023) << 4) |
| 228 | mov r12, r12, lsl r14 @ r12 = triangle_area_normalized |
| 229 | |
| 230 | @ load area normalized into lower half of double |
| 231 | mov r5, r12, lsr #10 |
| 232 | vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n |
| 233 | |
| 234 | movt r4, #((1022 + 31) << 4) |
| 235 | mov r5, r12, lsl #20 |
| 236 | |
| 237 | add r4, r4, r12, lsr #11 |
| 238 | vmov.f64 d31, r5, r4 |
| 239 | |
| 240 | vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n |
| 241 | |
| 242 | // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) = |
| 243 | // ( d0 * d1 ) - ( d2 * d3 ) = |
| 244 | // ( m0 ) - ( m1 ) = gradient |
| 245 | |
| 246 | // This is split to do 12 elements at a time over three sets: a, b, and c. |
| 247 | // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so |
| 248 | // two of the slots are unused. |
| 249 | |
| 250 | // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as |
| 251 | // is g. |
| 252 | |
| 253 | // First type is: uvrg bxxx xxxx |
| 254 | // Second type is: yyyy ybyy uvrg |
| 255 | // Since x_a and y_c are the same the same variable is used for both. |
| 256 | |
| 257 | vld1.u32 { v0 }, [ v_a, : 128 ] @ v0 = { uvrg0, b0, x0, y0 } |
| 258 | ldrsh x0, [ v_a, #8 ] @ load x0 |
| 259 | |
| 260 | vld1.u32 { v1 }, [ v_b, : 128 ] @ v1 = { uvrg1, b1, x1, y1} |
| 261 | ldrh x1, [ v_b, #8 ] @ load x1 |
| 262 | |
| 263 | vld1.u32 { v2 }, [ v_c, : 128 ] @ v2 = { uvrg2, b2, x2, y2 } |
| 264 | ldrh x2, [ v_c, #8 ] @ load x2 |
| 265 | |
| 266 | vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- } |
| 267 | ldrh y0, [ v_a, #10 ] @ load y0 |
| 268 | |
| 269 | vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- } |
| 270 | ldrh y1, [ v_b, #10 ] @ load y1 |
| 271 | |
| 272 | vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- } |
| 273 | ldrh y2, [ v_c, #10 ] @ load y2 |
| 274 | |
| 275 | vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 } |
| 276 | vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 } |
| 277 | |
| 278 | orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 } |
| 279 | pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 } |
| 280 | |
| 281 | vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 } |
| 282 | vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 } |
| 283 | |
| 284 | vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 } |
| 285 | vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 } |
| 286 | |
| 287 | ldrb b2, [ v_c, #4 ] @ load b2 |
| 288 | orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 } |
| 289 | |
| 290 | ldrb b1, [ v_b, #4 ] @ load b1 |
| 291 | orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 } |
| 292 | |
| 293 | vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 } |
| 294 | vsub.s16 d0_ab, x1_ab, x0_ab |
| 295 | |
| 296 | ldrb b0, [ v_a, #4 ] @ load b0 |
| 297 | orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 } |
| 298 | |
| 299 | vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 } |
| 300 | vsub.s16 d2_ab, x2_ab, x1_ab |
| 301 | |
| 302 | vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 } |
| 303 | vsub.s16 d1_ab, y2_ab, y1_ab |
| 304 | |
| 305 | orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 } |
| 306 | ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 } |
| 307 | |
| 308 | ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 } |
| 309 | ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 } |
| 310 | |
| 311 | vsub.s16 d3_ab, y1_ab, y0_ab |
| 312 | smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) - |
| 313 | @ ((x2 - X1) * (b1 - b0)) |
| 314 | vmull.s16 ga_uvrg_x, d0_a, d1_a |
| 315 | smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) - |
| 316 | @ ((b2 - b1) * (y1 - y0)) |
| 317 | vmlsl.s16 ga_uvrg_x, d2_a, d3_a |
| 318 | movs gs_bx, ga_bx, asr #31 |
| 319 | |
| 320 | vmull.s16 ga_uvrg_y, d0_b, d1_b |
| 321 | rsbmi ga_bx, ga_bx, #0 |
| 322 | |
| 323 | @ r12 = psx_gpu->uvrgb_phase |
| 324 | ldr r12, [ psx_gpu, #psx_gpu_uvrgb_phase_offset ] |
| 325 | |
| 326 | vmlsl.s16 ga_uvrg_y, d2_b, d3_b |
| 327 | movs gs_by, ga_by, asr #31 |
| 328 | |
| 329 | vshr.u64 d0, d30, #22 |
| 330 | add b_base, r12, b0, lsl #16 |
| 331 | |
| 332 | vdup.u32 uvrgb_phase, r12 |
| 333 | |
| 334 | rsbmi ga_by, ga_by, #0 |
| 335 | vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0 |
| 336 | |
| 337 | @ r12 = psx_gpu->triangle_winding_offset |
| 338 | ldrb r12, [ psx_gpu, #psx_gpu_triangle_winding_offset ] |
| 339 | vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0 |
| 340 | |
| 341 | rsb r12, r12, #0 @ r12 = -(triangle->winding) |
| 342 | |
| 343 | vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w } |
| 344 | sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS) |
| 345 | |
| 346 | vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16 |
| 347 | vdup.u32 r_shift, r14 @ r_shift = { shift, shift, shift, shift } |
| 348 | |
| 349 | vadd.u32 uvrg_base, uvrgb_phase |
| 350 | vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x) |
| 351 | |
| 352 | vmov area_r_s, s0 @ area_r_s = triangle_reciprocal |
| 353 | vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y) |
| 354 | |
| 355 | vmull.u32 gw_rg_x, ga_rg_x, d0[0] |
| 356 | vmull.u32 gw_uv_x, ga_uv_x, d0[0] |
| 357 | vmull.u32 gw_rg_y, ga_rg_y, d0[0] |
| 358 | vmull.u32 gw_uv_y, ga_uv_y, d0[0] |
| 359 | |
| 360 | vshl.u64 gw_rg_x, gw_rg_x, r_shift |
| 361 | vshl.u64 gw_uv_x, gw_uv_x, r_shift |
| 362 | vshl.u64 gw_rg_y, gw_rg_y, r_shift |
| 363 | vshl.u64 gw_uv_y, gw_uv_y, r_shift |
| 364 | |
| 365 | veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask |
| 366 | vmovn.u64 g_uv_x, gw_uv_x |
| 367 | |
| 368 | veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask |
| 369 | vmovn.u64 g_rg_x, gw_rg_x |
| 370 | |
| 371 | veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x |
| 372 | vmovn.u64 g_uv_y, gw_uv_y |
| 373 | |
| 374 | vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x |
| 375 | vmovn.u64 g_rg_y, gw_rg_y |
| 376 | |
| 377 | veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y |
| 378 | mov ga_bx, ga_bx, lsl #13 |
| 379 | |
| 380 | vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y |
| 381 | mov ga_by, ga_by, lsl #13 |
| 382 | |
| 383 | vdup.u32 x0_y0, x0 |
| 384 | umull gw_bx_l, gw_bx_h, ga_bx, area_r_s |
| 385 | |
| 386 | vshl.u32 g_uvrg_x, g_uvrg_x, #4 |
| 387 | vshl.u32 g_uvrg_y, g_uvrg_y, #4 |
| 388 | |
| 389 | umull gw_by_l, gw_by_h, ga_by, area_r_s |
| 390 | vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0] |
| 391 | |
| 392 | eor gs_bx, gs_bx, r12 |
| 393 | vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1 |
| 394 | |
| 395 | veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0 |
| 396 | eor gs_by, gs_by, r12 |
| 397 | |
| 398 | rsb r11, r14, #0 @ r11 = negative shift for scalar lsr |
| 399 | add store_a, psx_gpu, #psx_gpu_uvrg_offset |
| 400 | |
| 401 | sub r11, r11, #(32 - 13) |
| 402 | |
| 403 | add store_b, store_a, #16 |
| 404 | mov store_inc, #32 |
| 405 | |
| 406 | vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1 |
| 407 | vst1.u32 { uvrg_base }, [ store_a, : 128 ], store_inc |
| 408 | |
| 409 | vst1.u32 { uvrg_dx1 }, [ store_b, : 128 ], store_inc |
| 410 | mov g_bx, gw_bx_h, lsr r11 |
| 411 | |
| 412 | vst1.u32 { g_uvrg_y }, [ store_a, : 128 ], store_inc |
| 413 | mov g_by, gw_by_h, lsr r11 |
| 414 | |
| 415 | vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \ |
| 416 | [ store_b, : 128 ], store_inc |
| 417 | eor g_bx, g_bx, gs_bx |
| 418 | |
| 419 | vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \ |
| 420 | [ store_b, : 128 ], store_inc |
| 421 | sub g_bx, g_bx, gs_bx |
| 422 | |
| 423 | lsl g_bx, g_bx, #4 |
| 424 | eor g_by, g_by, gs_by |
| 425 | |
| 426 | mls b_base, g_bx, x0, b_base |
| 427 | sub g_by, g_by, gs_by |
| 428 | |
| 429 | lsl g_by, g_by, #4 |
| 430 | mov g_bx0, #0 |
| 431 | |
| 432 | add g_bx2, g_bx, g_bx |
| 433 | add g_bx3, g_bx, g_bx2 |
| 434 | |
| 435 | stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by } |
| 436 | |
| 437 | ldmia sp!, { r4 - r11, pc } |
| 438 | |
| 439 | |
| 440 | #define psx_gpu r0 |
| 441 | #define v_a r1 |
| 442 | #define v_b r2 |
| 443 | #define v_c r3 |
| 444 | |
| 445 | #define temp r14 |
| 446 | |
| 447 | #define x_a r4 |
| 448 | #define x_b r5 |
| 449 | #define x_c r6 |
| 450 | #define y_a r1 |
| 451 | #define y_b r2 |
| 452 | #define y_c r3 |
| 453 | |
| 454 | #define height_minor_a r7 |
| 455 | #define height_minor_b r8 |
| 456 | #define height_major r9 |
| 457 | #define height r9 |
| 458 | |
| 459 | #define reciprocal_table_ptr r10 |
| 460 | |
| 461 | #define edge_alt_low r4 |
| 462 | #define edge_alt_high r5 |
| 463 | #define edge_dx_dy_alt r6 |
| 464 | #define edge_shift_alt r10 |
| 465 | |
| 466 | #define edge_dx_dy_alt_low r4 |
| 467 | #define edge_dx_dy_alt_high r5 |
| 468 | |
| 469 | #define span_edge_data r4 |
| 470 | #define span_uvrg_offset r5 |
| 471 | #define span_b_offset r6 |
| 472 | |
| 473 | #define clip r14 |
| 474 | |
| 475 | #define b r11 |
| 476 | #define b_dy r12 |
| 477 | |
| 478 | |
| 479 | #define alternate_x q0 |
| 480 | #define alternate_dx_dy q1 |
| 481 | #define alternate_x_32 q2 |
| 482 | |
| 483 | #define alternate_x_low d0 |
| 484 | #define alternate_x_high d1 |
| 485 | #define alternate_dx_dy_low d2 |
| 486 | #define alternate_dx_dy_high d3 |
| 487 | #define alternate_x_32_low d4 |
| 488 | #define alternate_x_32_high d5 |
| 489 | |
| 490 | #define left_x q3 |
| 491 | #define right_x q4 |
| 492 | #define left_dx_dy q5 |
| 493 | #define right_dx_dy q6 |
| 494 | #define left_edge q7 |
| 495 | #define right_edge q8 |
| 496 | |
| 497 | #define left_x_low d6 |
| 498 | #define left_x_high d7 |
| 499 | #define right_x_low d8 |
| 500 | #define right_x_high d9 |
| 501 | #define left_dx_dy_low d10 |
| 502 | #define left_dx_dy_high d11 |
| 503 | #define right_dx_dy_low d12 |
| 504 | #define right_dx_dy_high d13 |
| 505 | #define left_edge_low d14 |
| 506 | #define left_edge_high d15 |
| 507 | #define right_edge_low d16 |
| 508 | #define right_edge_high d17 |
| 509 | |
| 510 | #define y_mid_point d18 |
| 511 | #define c_0x0004 d19 |
| 512 | |
| 513 | #define left_right_x_16 q11 |
| 514 | #define span_shifts_y q12 |
| 515 | #define c_0x0001 q13 |
| 516 | |
| 517 | #define span_shifts d24 |
| 518 | #define y_x4 d25 |
| 519 | #define c_0xFFFE d26 |
| 520 | #define c_0x0007 d27 |
| 521 | |
| 522 | #define left_right_x_16_low d22 |
| 523 | #define left_right_x_16_high d23 |
| 524 | |
| 525 | #define uvrg q14 |
| 526 | #define uvrg_dy q15 |
| 527 | |
| 528 | #define alternate_x_16 d4 |
| 529 | |
| 530 | #define v_clip q3 |
| 531 | #define v_clip_low d6 |
| 532 | |
| 533 | #define right_x_32 q10 |
| 534 | #define left_x_32 q11 |
| 535 | #define alternate_select d24 |
| 536 | |
| 537 | #define right_x_32_low d20 |
| 538 | #define right_x_32_high d21 |
| 539 | #define left_x_32_low d22 |
| 540 | #define left_x_32_high d23 |
| 541 | |
| 542 | #define edges_xy q0 |
| 543 | #define edges_dx_dy d2 |
| 544 | #define edge_shifts d3 |
| 545 | #define edge_shifts_64 q2 |
| 546 | |
| 547 | #define edges_xy_left d0 |
| 548 | #define edges_xy_right d1 |
| 549 | |
| 550 | #define height_reciprocals d6 |
| 551 | #define heights d7 |
| 552 | |
| 553 | #define widths d8 |
| 554 | #define c_0x01 d9 |
| 555 | #define x_starts d10 |
| 556 | #define x_ends d11 |
| 557 | |
| 558 | #define heights_b d12 |
| 559 | #define edges_dx_dy_64 q10 |
| 560 | |
| 561 | #define edges_dx_dy_64_left d20 |
| 562 | #define edges_dx_dy_64_right d21 |
| 563 | |
| 564 | |
| 565 | #define setup_spans_prologue() \ |
| 566 | stmdb sp!, { r4 - r11, lr }; \ |
| 567 | \ |
| 568 | ldrsh x_a, [ v_a, #8 ]; \ |
| 569 | ldrsh x_b, [ v_b, #8 ]; \ |
| 570 | ldrsh x_c, [ v_c, #8 ]; \ |
| 571 | ldrsh y_a, [ v_a, #10 ]; \ |
| 572 | ldrsh y_b, [ v_b, #10 ]; \ |
| 573 | ldrsh y_c, [ v_c, #10 ]; \ |
| 574 | \ |
| 575 | add temp, psx_gpu, #psx_gpu_uvrg_offset; \ |
| 576 | vld1.32 { uvrg }, [ temp ]; \ |
| 577 | add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \ |
| 578 | vld1.32 { uvrg_dy }, [ temp ]; \ |
| 579 | load_pointer(reciprocal_table_ptr, reciprocal_table); \ |
| 580 | \ |
| 581 | vmov.u32 c_0x01, #0x01 \ |
| 582 | |
| 583 | #define setup_spans_load_b() \ |
| 584 | ldr b, [ psx_gpu, #psx_gpu_b_offset ]; \ |
| 585 | ldr b_dy, [ psx_gpu, #psx_gpu_b_dy_offset ] \ |
| 586 | |
| 587 | #define setup_spans_prologue_b() \ |
| 588 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 589 | add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \ |
| 590 | \ |
| 591 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 592 | vmov.u16 c_0x0004, #0x0004; \ |
| 593 | \ |
| 594 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 595 | vmov.u16 c_0x0001, #0x0001; \ |
| 596 | \ |
| 597 | vld1.u16 { left_edge_low[], left_edge_high[] }, [ temp ]; \ |
| 598 | add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \ |
| 599 | \ |
| 600 | vld1.u16 { right_edge_low[], right_edge_high[] }, [ temp ]; \ |
| 601 | vadd.u16 right_edge, right_edge, c_0x0001; \ |
| 602 | \ |
| 603 | vmov.u16 c_0x0007, #0x0007; \ |
| 604 | vmvn.u16 c_0xFFFE, #0x0001 \ |
| 605 | |
| 606 | |
| 607 | #define compute_edge_delta_x2() \ |
| 608 | ldr temp, [ reciprocal_table_ptr, height, lsl #2 ]; \ |
| 609 | \ |
| 610 | vdup.u32 heights, height; \ |
| 611 | vsub.u32 widths, x_ends, x_starts; \ |
| 612 | \ |
| 613 | vdup.u32 edge_shifts, temp; \ |
| 614 | vsub.u32 heights_b, heights, c_0x01; \ |
| 615 | vshr.u32 height_reciprocals, edge_shifts, #10; \ |
| 616 | \ |
| 617 | vmla.s32 heights_b, x_starts, heights; \ |
| 618 | vbic.u16 edge_shifts, #0xE0; \ |
| 619 | vmul.s32 edges_dx_dy, widths, height_reciprocals; \ |
| 620 | vmull.s32 edges_xy, heights_b, height_reciprocals \ |
| 621 | |
| 622 | #define width_alt r6 |
| 623 | #define height_reciprocal_alt r11 |
| 624 | #define height_b_alt r12 |
| 625 | |
| 626 | #define compute_edge_delta_x3(start_c, height_a, height_b) \ |
| 627 | vmov.u32 heights, height_a, height_b; \ |
| 628 | ldr temp, [ reciprocal_table_ptr, height_a, lsl #2 ]; \ |
| 629 | vmov.u32 edge_shifts[0], temp; \ |
| 630 | ldr temp, [ reciprocal_table_ptr, height_b, lsl #2 ]; \ |
| 631 | vmov.u32 edge_shifts[1], temp; \ |
| 632 | ldr edge_shift_alt, [ reciprocal_table_ptr, height_minor_b, lsl #2 ]; \ |
| 633 | \ |
| 634 | vsub.u32 widths, x_ends, x_starts; \ |
| 635 | sub width_alt, x_c, start_c; \ |
| 636 | \ |
| 637 | vsub.u32 heights_b, heights, c_0x01; \ |
| 638 | sub height_b_alt, height_minor_b, #1; \ |
| 639 | \ |
| 640 | vshr.u32 height_reciprocals, edge_shifts, #10; \ |
| 641 | lsr height_reciprocal_alt, edge_shift_alt, #10; \ |
| 642 | \ |
| 643 | vmla.s32 heights_b, x_starts, heights; \ |
| 644 | mla height_b_alt, height_minor_b, start_c, height_b_alt; \ |
| 645 | \ |
| 646 | vbic.u16 edge_shifts, #0xE0; \ |
| 647 | and edge_shift_alt, edge_shift_alt, #0x1F; \ |
| 648 | \ |
| 649 | vmul.s32 edges_dx_dy, widths, height_reciprocals; \ |
| 650 | mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \ |
| 651 | \ |
| 652 | vmull.s32 edges_xy, heights_b, height_reciprocals; \ |
| 653 | smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \ |
| 654 | |
| 655 | |
| 656 | #define setup_spans_adjust_y_up() \ |
| 657 | vsub.u32 y_x4, y_x4, c_0x0004 \ |
| 658 | |
| 659 | #define setup_spans_adjust_y_down() \ |
| 660 | vadd.u32 y_x4, y_x4, c_0x0004 \ |
| 661 | |
| 662 | #define setup_spans_adjust_interpolants_up() \ |
| 663 | vsub.u32 uvrg, uvrg, uvrg_dy; \ |
| 664 | sub b, b, b_dy \ |
| 665 | |
| 666 | #define setup_spans_adjust_interpolants_down() \ |
| 667 | vadd.u32 uvrg, uvrg, uvrg_dy; \ |
| 668 | add b, b, b_dy \ |
| 669 | |
| 670 | |
| 671 | #define setup_spans_clip_interpolants_increment() \ |
| 672 | mla b, b_dy, clip, b; \ |
| 673 | vmla.s32 uvrg, uvrg_dy, v_clip \ |
| 674 | |
| 675 | #define setup_spans_clip_interpolants_decrement() \ |
| 676 | mls b, b_dy, clip, b; \ |
| 677 | vmls.s32 uvrg, uvrg_dy, v_clip \ |
| 678 | |
| 679 | #define setup_spans_clip_alternate_yes() \ |
| 680 | smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \ |
| 681 | |
| 682 | #define setup_spans_clip_alternate_no() \ |
| 683 | |
| 684 | #define setup_spans_clip(direction, alternate_active) \ |
| 685 | vdup.u32 v_clip, clip; \ |
| 686 | setup_spans_clip_alternate_##alternate_active(); \ |
| 687 | setup_spans_clip_interpolants_##direction(); \ |
| 688 | vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \ |
| 689 | |
| 690 | |
| 691 | #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \ |
| 692 | vmovl.s32 edge_shifts_64, edge_shifts; \ |
| 693 | vmovl.s32 edges_dx_dy_64, edges_dx_dy; \ |
| 694 | \ |
| 695 | vshl.s64 edges_xy, edges_xy, edge_shifts_64; \ |
| 696 | vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \ |
| 697 | \ |
| 698 | vmov left_x_low, edges_xy_##left_index; \ |
| 699 | vmov right_x_low, edges_xy_##right_index; \ |
| 700 | \ |
| 701 | vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \ |
| 702 | vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \ |
| 703 | vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \ |
| 704 | vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \ |
| 705 | \ |
| 706 | vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \ |
| 707 | vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \ |
| 708 | \ |
| 709 | vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \ |
| 710 | vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \ |
| 711 | |
| 712 | |
| 713 | #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \ |
| 714 | setup_spans_adjust_edges_alternate_no(left_index, right_index); \ |
| 715 | \ |
| 716 | vdup.u16 y_mid_point, y_b; \ |
| 717 | rsb temp, edge_shift_alt, #32; \ |
| 718 | \ |
| 719 | lsl edge_alt_high, edge_alt_high, edge_shift_alt; \ |
| 720 | orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \ |
| 721 | lsl edge_alt_low, edge_alt_low, edge_shift_alt; \ |
| 722 | vmov alternate_x_low, edge_alt_low, edge_alt_high; \ |
| 723 | \ |
| 724 | asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \ |
| 725 | lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \ |
| 726 | vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \ |
| 727 | vmov alternate_dx_dy_high, alternate_dx_dy_low; \ |
| 728 | \ |
| 729 | vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \ |
| 730 | vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \ |
| 731 | |
| 732 | |
| 733 | #define setup_spans_y_select_up() \ |
| 734 | vclt.s16 alternate_select, y_x4, y_mid_point \ |
| 735 | |
| 736 | #define setup_spans_y_select_down() \ |
| 737 | vcgt.s16 alternate_select, y_x4, y_mid_point \ |
| 738 | |
| 739 | |
| 740 | #define setup_spans_alternate_select_left() \ |
| 741 | vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \ |
| 742 | |
| 743 | #define setup_spans_alternate_select_right() \ |
| 744 | vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \ |
| 745 | |
| 746 | |
| 747 | #define setup_spans_set_x4_alternate_yes(alternate, direction) \ |
| 748 | vshrn.s64 alternate_x_32_low, alternate_x, #32; \ |
| 749 | vshrn.s64 left_x_32_low, left_x, #32; \ |
| 750 | vshrn.s64 right_x_32_low, right_x, #32; \ |
| 751 | \ |
| 752 | vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \ |
| 753 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 754 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 755 | \ |
| 756 | vshrn.s64 alternate_x_32_high, alternate_x, #32; \ |
| 757 | vshrn.s64 left_x_32_high, left_x, #32; \ |
| 758 | vshrn.s64 right_x_32_high, right_x, #32; \ |
| 759 | \ |
| 760 | vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \ |
| 761 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 762 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 763 | \ |
| 764 | vmovn.u32 alternate_x_16, alternate_x_32; \ |
| 765 | setup_spans_y_select_##direction(); \ |
| 766 | vmovn.u32 left_right_x_16_low, left_x_32; \ |
| 767 | \ |
| 768 | vmovn.u32 left_right_x_16_high, right_x_32; \ |
| 769 | setup_spans_alternate_select_##alternate(); \ |
| 770 | \ |
| 771 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 772 | str b, [ span_b_offset ], #4; \ |
| 773 | setup_spans_adjust_interpolants_##direction(); \ |
| 774 | \ |
| 775 | vmax.s16 left_right_x_16, left_right_x_16, left_edge; \ |
| 776 | \ |
| 777 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 778 | str b, [ span_b_offset ], #4; \ |
| 779 | setup_spans_adjust_interpolants_##direction(); \ |
| 780 | \ |
| 781 | vmin.s16 left_right_x_16, left_right_x_16, right_edge; \ |
| 782 | \ |
| 783 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 784 | str b, [ span_b_offset ], #4; \ |
| 785 | setup_spans_adjust_interpolants_##direction(); \ |
| 786 | \ |
| 787 | vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \ |
| 788 | vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \ |
| 789 | vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \ |
| 790 | \ |
| 791 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 792 | str b, [ span_b_offset ], #4; \ |
| 793 | setup_spans_adjust_interpolants_##direction(); \ |
| 794 | \ |
| 795 | vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \ |
| 796 | vshl.u16 span_shifts, c_0xFFFE, span_shifts; \ |
| 797 | \ |
| 798 | vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \ |
| 799 | \ |
| 800 | setup_spans_adjust_y_##direction() \ |
| 801 | |
| 802 | |
| 803 | #define setup_spans_set_x4_alternate_no(alternate, direction) \ |
| 804 | vshrn.s64 left_x_32_low, left_x, #32; \ |
| 805 | vshrn.s64 right_x_32_low, right_x, #32; \ |
| 806 | \ |
| 807 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 808 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 809 | \ |
| 810 | vshrn.s64 left_x_32_high, left_x, #32; \ |
| 811 | vshrn.s64 right_x_32_high, right_x, #32; \ |
| 812 | \ |
| 813 | vadd.u64 left_x, left_x, left_dx_dy; \ |
| 814 | vadd.u64 right_x, right_x, right_dx_dy; \ |
| 815 | \ |
| 816 | vmovn.u32 left_right_x_16_low, left_x_32; \ |
| 817 | vmovn.u32 left_right_x_16_high, right_x_32; \ |
| 818 | \ |
| 819 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 820 | str b, [ span_b_offset ], #4; \ |
| 821 | setup_spans_adjust_interpolants_##direction(); \ |
| 822 | \ |
| 823 | vmax.s16 left_right_x_16, left_right_x_16, left_edge; \ |
| 824 | \ |
| 825 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 826 | str b, [ span_b_offset ], #4; \ |
| 827 | setup_spans_adjust_interpolants_##direction(); \ |
| 828 | \ |
| 829 | vmin.s16 left_right_x_16, left_right_x_16, right_edge; \ |
| 830 | \ |
| 831 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 832 | str b, [ span_b_offset ], #4; \ |
| 833 | setup_spans_adjust_interpolants_##direction(); \ |
| 834 | \ |
| 835 | vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \ |
| 836 | vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \ |
| 837 | vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \ |
| 838 | \ |
| 839 | vst1.u32 { uvrg }, [ span_uvrg_offset, :128 ]!; \ |
| 840 | str b, [ span_b_offset ], #4; \ |
| 841 | setup_spans_adjust_interpolants_##direction(); \ |
| 842 | \ |
| 843 | vshl.u16 span_shifts, c_0xFFFE, span_shifts; \ |
| 844 | vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \ |
| 845 | \ |
| 846 | vst4.u16 { left_right_x_16, span_shifts_y }, [ span_edge_data ]!; \ |
| 847 | \ |
| 848 | setup_spans_adjust_y_##direction() \ |
| 849 | |
| 850 | |
| 851 | #define edge_adjust_low r11 |
| 852 | #define edge_adjust_high r12 |
| 853 | |
| 854 | #define setup_spans_alternate_adjust_yes() \ |
| 855 | smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \ |
| 856 | subs edge_alt_low, edge_alt_low, edge_adjust_low; \ |
| 857 | sbc edge_alt_high, edge_alt_high, edge_adjust_high \ |
| 858 | |
| 859 | #define setup_spans_alternate_adjust_no() \ |
| 860 | |
| 861 | |
| 862 | #define setup_spans_down(left_index, right_index, alternate, alternate_active) \ |
| 863 | setup_spans_alternate_adjust_##alternate_active(); \ |
| 864 | setup_spans_load_b(); \ |
| 865 | \ |
| 866 | ldrsh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \ |
| 867 | subs y_c, y_c, temp; \ |
| 868 | subgt height, height, y_c; \ |
| 869 | addgt height, height, #1; \ |
| 870 | \ |
| 871 | ldrsh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \ |
| 872 | subs clip, temp, y_a; \ |
| 873 | ble 0f; \ |
| 874 | \ |
| 875 | sub height, height, clip; \ |
| 876 | add y_a, y_a, clip; \ |
| 877 | setup_spans_clip(increment, alternate_active); \ |
| 878 | \ |
| 879 | 0: \ |
| 880 | cmp height, #0; \ |
| 881 | ble 1f; \ |
| 882 | \ |
| 883 | orr temp, y_a, y_a, lsl #16; \ |
| 884 | add temp, temp, #(1 << 16); \ |
| 885 | add y_a, temp, #2; \ |
| 886 | add y_a, y_a, #(2 << 16); \ |
| 887 | vmov.u32 y_x4, temp, y_a; \ |
| 888 | \ |
| 889 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ |
| 890 | right_index); \ |
| 891 | setup_spans_prologue_b(); \ |
| 892 | \ |
| 893 | strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \ |
| 894 | \ |
| 895 | 2: \ |
| 896 | setup_spans_set_x4_alternate_##alternate_active(alternate, down); \ |
| 897 | subs height, height, #4; \ |
| 898 | bhi 2b; \ |
| 899 | \ |
| 900 | 1: \ |
| 901 | |
| 902 | |
| 903 | #define setup_spans_alternate_pre_increment_yes() \ |
| 904 | adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \ |
| 905 | adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \ |
| 906 | |
| 907 | #define setup_spans_alternate_pre_increment_no() \ |
| 908 | |
| 909 | |
| 910 | #define setup_spans_up_decrement_yes() \ |
| 911 | suble height, height, #1 \ |
| 912 | |
| 913 | #define setup_spans_up_decrement_no() \ |
| 914 | |
| 915 | |
| 916 | #define setup_spans_up(left_index, right_index, alternate, alternate_active) \ |
| 917 | setup_spans_alternate_adjust_##alternate_active(); \ |
| 918 | setup_spans_load_b(); \ |
| 919 | sub y_a, y_a, #1; \ |
| 920 | \ |
| 921 | ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ]; \ |
| 922 | subs temp, temp, y_c; \ |
| 923 | subgt height, height, temp; \ |
| 924 | setup_spans_up_decrement_##alternate_active(); \ |
| 925 | \ |
| 926 | ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ]; \ |
| 927 | subs clip, y_a, temp; \ |
| 928 | ble 0f; \ |
| 929 | \ |
| 930 | sub height, height, clip; \ |
| 931 | sub y_a, y_a, clip; \ |
| 932 | setup_spans_clip(decrement, alternate_active); \ |
| 933 | \ |
| 934 | 0: \ |
| 935 | cmp height, #0; \ |
| 936 | ble 1f; \ |
| 937 | \ |
| 938 | orr temp, y_a, y_a, lsl #16; \ |
| 939 | sub temp, temp, #(1 << 16); \ |
| 940 | sub y_a, temp, #2; \ |
| 941 | sub y_a, y_a, #(2 << 16); \ |
| 942 | vmov.u32 y_x4, temp, y_a; \ |
| 943 | \ |
| 944 | vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \ |
| 945 | \ |
| 946 | setup_spans_alternate_pre_increment_##alternate_active(); \ |
| 947 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ |
| 948 | right_index); \ |
| 949 | setup_spans_adjust_interpolants_up(); \ |
| 950 | setup_spans_prologue_b(); \ |
| 951 | \ |
| 952 | strh height, [ psx_gpu, #psx_gpu_num_spans_offset ]; \ |
| 953 | \ |
| 954 | 2: \ |
| 955 | setup_spans_set_x4_alternate_##alternate_active(alternate, up); \ |
| 956 | subs height, height, #4; \ |
| 957 | bhi 2b; \ |
| 958 | \ |
| 959 | 1: \ |
| 960 | |
| 961 | |
| 962 | #define setup_spans_epilogue() \ |
| 963 | ldmia sp!, { r4 - r11, pc } \ |
| 964 | |
| 965 | |
| 966 | #define setup_spans_up_up(minor, major) \ |
| 967 | setup_spans_prologue(); \ |
| 968 | sub height_minor_a, y_a, y_b; \ |
| 969 | sub height_minor_b, y_b, y_c; \ |
| 970 | sub height, y_a, y_c; \ |
| 971 | \ |
| 972 | vdup.u32 x_starts, x_a; \ |
| 973 | vmov.u32 x_ends, x_c, x_b; \ |
| 974 | \ |
| 975 | compute_edge_delta_x3(x_b, height_major, height_minor_a); \ |
| 976 | setup_spans_up(major, minor, minor, yes); \ |
| 977 | setup_spans_epilogue() \ |
| 978 | |
| 979 | function(setup_spans_up_left) |
| 980 | setup_spans_up_up(left, right) |
| 981 | |
| 982 | function(setup_spans_up_right) |
| 983 | setup_spans_up_up(right, left) |
| 984 | |
| 985 | .pool |
| 986 | |
| 987 | #define setup_spans_down_down(minor, major) \ |
| 988 | setup_spans_prologue(); \ |
| 989 | sub height_minor_a, y_b, y_a; \ |
| 990 | sub height_minor_b, y_c, y_b; \ |
| 991 | sub height, y_c, y_a; \ |
| 992 | \ |
| 993 | vdup.u32 x_starts, x_a; \ |
| 994 | vmov.u32 x_ends, x_c, x_b; \ |
| 995 | \ |
| 996 | compute_edge_delta_x3(x_b, height_major, height_minor_a); \ |
| 997 | setup_spans_down(major, minor, minor, yes); \ |
| 998 | setup_spans_epilogue() \ |
| 999 | |
| 1000 | function(setup_spans_down_left) |
| 1001 | setup_spans_down_down(left, right) |
| 1002 | |
| 1003 | function(setup_spans_down_right) |
| 1004 | setup_spans_down_down(right, left) |
| 1005 | |
| 1006 | |
| 1007 | #define setup_spans_up_flat() \ |
| 1008 | sub height, y_a, y_c; \ |
| 1009 | \ |
| 1010 | compute_edge_delta_x2(); \ |
| 1011 | setup_spans_up(left, right, none, no); \ |
| 1012 | setup_spans_epilogue() \ |
| 1013 | |
| 1014 | function(setup_spans_up_a) |
| 1015 | setup_spans_prologue() |
| 1016 | |
| 1017 | vmov.u32 x_starts, x_a, x_b |
| 1018 | vdup.u32 x_ends, x_c |
| 1019 | |
| 1020 | setup_spans_up_flat() |
| 1021 | |
| 1022 | function(setup_spans_up_b) |
| 1023 | setup_spans_prologue() |
| 1024 | |
| 1025 | vdup.u32 x_starts, x_a |
| 1026 | vmov.u32 x_ends, x_b, x_c |
| 1027 | |
| 1028 | setup_spans_up_flat() |
| 1029 | |
| 1030 | #define setup_spans_down_flat() \ |
| 1031 | sub height, y_c, y_a; \ |
| 1032 | \ |
| 1033 | compute_edge_delta_x2(); \ |
| 1034 | setup_spans_down(left, right, none, no); \ |
| 1035 | setup_spans_epilogue() \ |
| 1036 | |
| 1037 | function(setup_spans_down_a) |
| 1038 | setup_spans_prologue() |
| 1039 | |
| 1040 | vmov.u32 x_starts, x_a, x_b |
| 1041 | vdup.u32 x_ends, x_c |
| 1042 | |
| 1043 | setup_spans_down_flat() |
| 1044 | |
| 1045 | function(setup_spans_down_b) |
| 1046 | setup_spans_prologue() |
| 1047 | |
| 1048 | vdup.u32 x_starts, x_a |
| 1049 | vmov.u32 x_ends, x_b, x_c |
| 1050 | |
| 1051 | setup_spans_down_flat() |
| 1052 | |
| 1053 | |
| 1054 | #define middle_y r9 |
| 1055 | |
| 1056 | #define edges_xy_b q11 |
| 1057 | #define edges_dx_dy_b d26 |
| 1058 | #define edge_shifts_b d27 |
| 1059 | #define edges_dx_dy_and_shifts_b q13 |
| 1060 | #define height_increment d20 |
| 1061 | |
| 1062 | #define edges_dx_dy_and_shifts q1 |
| 1063 | |
| 1064 | #define edges_xy_b_left d22 |
| 1065 | #define edges_xy_b_right d23 |
| 1066 | |
| 1067 | #define setup_spans_up_down_load_edge_set_b() \ |
| 1068 | vmov edges_xy, edges_xy_b; \ |
| 1069 | vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \ |
| 1070 | |
| 1071 | |
| 1072 | function(setup_spans_up_down) |
| 1073 | setup_spans_prologue() |
| 1074 | |
| 1075 | // s32 middle_y = y_a; |
| 1076 | sub height_minor_a, y_a, y_b |
| 1077 | sub height_minor_b, y_c, y_a |
| 1078 | sub height_major, y_c, y_b |
| 1079 | |
| 1080 | vmov.u32 x_starts, x_a, x_c |
| 1081 | vdup.u32 x_ends, x_b |
| 1082 | |
| 1083 | compute_edge_delta_x3(x_a, height_minor_a, height_major) |
| 1084 | |
| 1085 | mov temp, #0 |
| 1086 | vmov.u32 height_increment, temp, height_minor_b |
| 1087 | vmlal.s32 edges_xy, edges_dx_dy, height_increment |
| 1088 | |
| 1089 | vmov edges_xy_b_left, edge_alt_low, edge_alt_high |
| 1090 | vmov edges_xy_b_right, edges_xy_right |
| 1091 | |
| 1092 | vmov edge_shifts_b, edge_shifts |
| 1093 | vmov.u32 edge_shifts_b[0], edge_shift_alt |
| 1094 | |
| 1095 | vneg.s32 edges_dx_dy_b, edges_dx_dy |
| 1096 | vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt |
| 1097 | |
| 1098 | mov middle_y, y_a |
| 1099 | |
| 1100 | setup_spans_load_b() |
| 1101 | sub y_a, y_a, #1 |
| 1102 | |
| 1103 | ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ] |
| 1104 | subs temp, temp, y_b |
| 1105 | subgt height_minor_a, height_minor_a, temp |
| 1106 | |
| 1107 | ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ] |
| 1108 | subs clip, y_a, temp |
| 1109 | ble 0f |
| 1110 | |
| 1111 | sub height_minor_a, height_minor_a, clip |
| 1112 | sub y_a, y_a, clip |
| 1113 | setup_spans_clip(decrement, no) |
| 1114 | |
| 1115 | 0: |
| 1116 | cmp height_minor_a, #0 |
| 1117 | ble 3f |
| 1118 | |
| 1119 | orr temp, y_a, y_a, lsl #16 |
| 1120 | sub temp, temp, #(1 << 16) |
| 1121 | sub y_a, temp, #2 |
| 1122 | sub y_a, y_a, #(2 << 16) |
| 1123 | vmov.u32 y_x4, temp, y_a |
| 1124 | |
| 1125 | vaddw.s32 edges_xy, edges_xy, edges_dx_dy |
| 1126 | |
| 1127 | strh height_minor_a, [ psx_gpu, #psx_gpu_num_spans_offset ] |
| 1128 | |
| 1129 | setup_spans_adjust_edges_alternate_no(left, right); |
| 1130 | setup_spans_adjust_interpolants_up() |
| 1131 | setup_spans_up_down_load_edge_set_b() |
| 1132 | |
| 1133 | setup_spans_prologue_b() |
| 1134 | |
| 1135 | |
| 1136 | 2: |
| 1137 | setup_spans_set_x4_alternate_no(none, up) |
| 1138 | subs height_minor_a, height_minor_a, #4 |
| 1139 | bhi 2b |
| 1140 | |
| 1141 | add span_edge_data, span_edge_data, height_minor_a, lsl #3 |
| 1142 | add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4 |
| 1143 | add span_b_offset, span_b_offset, height_minor_a, lsl #2 |
| 1144 | |
| 1145 | 4: |
| 1146 | add temp, psx_gpu, #psx_gpu_uvrg_offset |
| 1147 | vld1.32 { uvrg }, [ temp ] |
| 1148 | mov y_a, middle_y |
| 1149 | |
| 1150 | setup_spans_load_b() |
| 1151 | |
| 1152 | ldrh temp, [ psx_gpu, #psx_gpu_viewport_end_y_offset ] |
| 1153 | subs y_c, y_c, temp |
| 1154 | subgt height_minor_b, height_minor_b, y_c |
| 1155 | addgt height_minor_b, height_minor_b, #1 |
| 1156 | |
| 1157 | ldrh temp, [ psx_gpu, #psx_gpu_viewport_start_y_offset ] |
| 1158 | subs clip, temp, y_a |
| 1159 | ble 0f |
| 1160 | |
| 1161 | sub height_minor_b, height_minor_b, clip |
| 1162 | add y_a, y_a, clip |
| 1163 | setup_spans_clip(increment, no) |
| 1164 | |
| 1165 | 0: |
| 1166 | cmp height_minor_b, #0 |
| 1167 | ble 1f |
| 1168 | |
| 1169 | orr temp, y_a, y_a, lsl #16 |
| 1170 | add temp, temp, #(1 << 16) |
| 1171 | add y_a, temp, #2 |
| 1172 | add y_a, y_a, #(2 << 16) |
| 1173 | vmov.u32 y_x4, temp, y_a |
| 1174 | |
| 1175 | setup_spans_adjust_edges_alternate_no(left, right) |
| 1176 | |
| 1177 | ldrh temp, [ psx_gpu, #psx_gpu_num_spans_offset ] |
| 1178 | add temp, temp, height_minor_b |
| 1179 | |
| 1180 | cmp temp, #MAX_SPANS |
| 1181 | beq 5f |
| 1182 | |
| 1183 | strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ] |
| 1184 | |
| 1185 | 2: |
| 1186 | setup_spans_set_x4_alternate_no(none, down) |
| 1187 | subs height_minor_b, height_minor_b, #4 |
| 1188 | bhi 2b |
| 1189 | |
| 1190 | 1: |
| 1191 | setup_spans_epilogue() |
| 1192 | |
| 1193 | 3: |
| 1194 | setup_spans_up_down_load_edge_set_b() |
| 1195 | setup_spans_prologue_b() |
| 1196 | bal 4b |
| 1197 | |
| 1198 | 5: |
| 1199 | // FIXME: overflow corner case |
| 1200 | sub temp, temp, height_minor_b |
| 1201 | bics height_minor_b, #3 |
| 1202 | add temp, temp, height_minor_b |
| 1203 | strh temp, [ psx_gpu, #psx_gpu_num_spans_offset ] |
| 1204 | bne 2b |
| 1205 | bal 1b |
| 1206 | |
| 1207 | .pool |
| 1208 | |
| 1209 | #undef span_uvrg_offset |
| 1210 | #undef span_edge_data |
| 1211 | #undef span_b_offset |
| 1212 | #undef left_x |
| 1213 | #undef b |
| 1214 | |
| 1215 | #define psx_gpu r0 |
| 1216 | #define num_spans r1 |
| 1217 | #define span_uvrg_offset r2 |
| 1218 | #define span_edge_data r3 |
| 1219 | #define span_b_offset r4 |
| 1220 | #define b_dx r5 |
| 1221 | #define span_num_blocks r6 |
| 1222 | #define y r7 |
| 1223 | #define left_x r8 |
| 1224 | #define b r9 |
| 1225 | #define dither_offset_ptr r10 |
| 1226 | #define block_ptr_a r11 |
| 1227 | #define fb_ptr r12 |
| 1228 | #define num_blocks r14 |
| 1229 | |
| 1230 | #define uvrg_dx_ptr r2 |
| 1231 | #define texture_mask_ptr r3 |
| 1232 | #define dither_shift r8 |
| 1233 | #define dither_row r10 |
| 1234 | |
| 1235 | #define c_32 r7 |
| 1236 | #define b_dx4 r8 |
| 1237 | #define b_dx8 r9 |
| 1238 | #define block_ptr_b r10 |
| 1239 | |
| 1240 | #define block_span_ptr r10 |
| 1241 | #define right_mask r8 |
| 1242 | |
| 1243 | #define color r2 |
| 1244 | #define color_r r3 |
| 1245 | #define color_g r4 |
| 1246 | #define color_b r5 |
| 1247 | |
| 1248 | #undef uvrg |
| 1249 | |
| 1250 | #define u_block q0 |
| 1251 | #define v_block q1 |
| 1252 | #define r_block q2 |
| 1253 | #define g_block q3 |
| 1254 | #define b_block q4 |
| 1255 | |
| 1256 | #define uv_dx4 d10 |
| 1257 | #define rg_dx4 d11 |
| 1258 | #define uv_dx8 d12 |
| 1259 | #define rg_dx8 d13 |
| 1260 | #define b_whole_8 d14 |
| 1261 | #define fb_mask_ptrs d15 |
| 1262 | |
| 1263 | #define uvrg_dx4 q5 |
| 1264 | #define uvrg_dx8 q6 |
| 1265 | #define uv_dx8 d12 |
| 1266 | #define rg_dx8 d13 |
| 1267 | |
| 1268 | #define u_whole q8 |
| 1269 | #define v_whole q9 |
| 1270 | #define r_whole q10 |
| 1271 | #define g_whole q11 |
| 1272 | #define b_whole q12 |
| 1273 | |
| 1274 | #define u_whole_low d16 |
| 1275 | #define u_whole_high d17 |
| 1276 | #define v_whole_low d18 |
| 1277 | #define v_whole_high d19 |
| 1278 | #define r_whole_low d20 |
| 1279 | #define r_whole_high d21 |
| 1280 | #define g_whole_low d22 |
| 1281 | #define g_whole_high d23 |
| 1282 | #define b_whole_low d24 |
| 1283 | #define b_whole_high d25 |
| 1284 | |
| 1285 | #define dx4 q13 |
| 1286 | #define dx8 q13 |
| 1287 | |
| 1288 | #define u_whole_8 d26 |
| 1289 | #define v_whole_8 d27 |
| 1290 | #define u_whole_8b d24 |
| 1291 | #define r_whole_8 d24 |
| 1292 | #define g_whole_8 d25 |
| 1293 | |
| 1294 | #define uv_whole_8 q13 |
| 1295 | #define uv_whole_8b q14 |
| 1296 | |
| 1297 | #define dither_offsets q14 |
| 1298 | #define texture_mask q15 |
| 1299 | #define texture_mask_u d30 |
| 1300 | #define texture_mask_v d31 |
| 1301 | |
| 1302 | #define dither_offsets_short d28 |
| 1303 | |
| 1304 | #define v_left_x q8 |
| 1305 | #define uvrg q9 |
| 1306 | #define block_span q10 |
| 1307 | |
| 1308 | #define uv d18 |
| 1309 | #define rg d19 |
| 1310 | |
| 1311 | #define draw_mask q1 |
| 1312 | #define draw_mask_edge q13 |
| 1313 | #define test_mask q0 |
| 1314 | |
| 1315 | #define uvrg_dx q3 |
| 1316 | |
| 1317 | #define colors q2 |
| 1318 | |
| 1319 | #define setup_blocks_texture_swizzled() \ |
| 1320 | vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \ |
| 1321 | vsli.u8 u_whole_8, v_whole_8, #4; \ |
| 1322 | vsri.u8 v_whole_8, u_whole_8b, #4 \ |
| 1323 | |
| 1324 | #define setup_blocks_texture_unswizzled() \ |
| 1325 | |
| 1326 | |
| 1327 | #define setup_blocks_shaded_textured_builder(swizzling) \ |
| 1328 | .align 3; \ |
| 1329 | \ |
| 1330 | function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \ |
| 1331 | ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \ |
| 1332 | add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \ |
| 1333 | \ |
| 1334 | vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \ |
| 1335 | add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \ |
| 1336 | \ |
| 1337 | cmp num_spans, #0; \ |
| 1338 | bxeq lr; \ |
| 1339 | \ |
| 1340 | stmdb sp!, { r4 - r11, r14 }; \ |
| 1341 | vshl.u32 uvrg_dx4, uvrg_dx, #2; \ |
| 1342 | \ |
| 1343 | ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \ |
| 1344 | vshl.u32 uvrg_dx8, uvrg_dx, #3; \ |
| 1345 | \ |
| 1346 | vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \ |
| 1347 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 1348 | \ |
| 1349 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 1350 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 1351 | \ |
| 1352 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 1353 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1354 | \ |
| 1355 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \ |
| 1356 | \ |
| 1357 | 0: \ |
| 1358 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1359 | \ |
| 1360 | ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \ |
| 1361 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 1362 | \ |
| 1363 | ldrh y, [ span_edge_data, #edge_data_y_offset ]; \ |
| 1364 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \ |
| 1365 | \ |
| 1366 | cmp span_num_blocks, #0; \ |
| 1367 | beq 1f; \ |
| 1368 | \ |
| 1369 | ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \ |
| 1370 | add num_blocks, span_num_blocks, num_blocks; \ |
| 1371 | \ |
| 1372 | cmp num_blocks, #MAX_BLOCKS; \ |
| 1373 | bgt 2f; \ |
| 1374 | \ |
| 1375 | 3: \ |
| 1376 | ldr b, [ span_b_offset ]; \ |
| 1377 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 1378 | \ |
| 1379 | vdup.u32 v_left_x, left_x; \ |
| 1380 | and y, y, #0x3; \ |
| 1381 | \ |
| 1382 | ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \ |
| 1383 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 1384 | \ |
| 1385 | mla b, b_dx, left_x, b; \ |
| 1386 | and dither_shift, left_x, #0x03; \ |
| 1387 | \ |
| 1388 | vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \ |
| 1389 | vshr.u32 uvrg_dx, uvrg_dx4, #2; \ |
| 1390 | \ |
| 1391 | mov dither_shift, dither_shift, lsl #3; \ |
| 1392 | vmla.u32 uvrg, uvrg_dx, v_left_x; \ |
| 1393 | \ |
| 1394 | mov c_32, #32; \ |
| 1395 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1396 | \ |
| 1397 | mov dither_row, dither_row, ror dither_shift; \ |
| 1398 | mov b_dx4, b_dx, lsl #2; \ |
| 1399 | \ |
| 1400 | vdup.u32 dither_offsets_short, dither_row; \ |
| 1401 | add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \ |
| 1402 | \ |
| 1403 | vdup.u32 b_block, b; \ |
| 1404 | vshll.s8 dither_offsets, dither_offsets_short, #4; \ |
| 1405 | \ |
| 1406 | vdup.u32 u_block, uv[0]; \ |
| 1407 | mov b_dx8, b_dx, lsl #3; \ |
| 1408 | \ |
| 1409 | vdup.u32 v_block, uv[1]; \ |
| 1410 | vdup.u32 r_block, rg[0]; \ |
| 1411 | vdup.u32 g_block, rg[1]; \ |
| 1412 | \ |
| 1413 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 1414 | \ |
| 1415 | vadd.u32 u_block, u_block, block_span; \ |
| 1416 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 1417 | \ |
| 1418 | vadd.u32 v_block, v_block, block_span; \ |
| 1419 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 1420 | \ |
| 1421 | vadd.u32 r_block, r_block, block_span; \ |
| 1422 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 1423 | \ |
| 1424 | vadd.u32 g_block, g_block, block_span; \ |
| 1425 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \ |
| 1426 | \ |
| 1427 | vadd.u32 b_block, b_block, block_span; \ |
| 1428 | add block_ptr_b, block_ptr_a, #16; \ |
| 1429 | \ |
| 1430 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1431 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1432 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 1433 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 1434 | \ |
| 1435 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1436 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 1437 | \ |
| 1438 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1439 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1440 | \ |
| 1441 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1442 | vdup.u32 dx4, rg_dx4[0]; \ |
| 1443 | \ |
| 1444 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 1445 | vdup.u32 dx4, rg_dx4[1]; \ |
| 1446 | \ |
| 1447 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 1448 | vdup.u32 dx4, b_dx4; \ |
| 1449 | \ |
| 1450 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 1451 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1452 | \ |
| 1453 | vadd.u32 u_block, u_block, dx8; \ |
| 1454 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1455 | \ |
| 1456 | vadd.u32 v_block, v_block, dx8; \ |
| 1457 | vdup.u32 dx8, rg_dx8[0]; \ |
| 1458 | \ |
| 1459 | vadd.u32 r_block, r_block, dx8; \ |
| 1460 | vdup.u32 dx8, rg_dx8[1]; \ |
| 1461 | \ |
| 1462 | vadd.u32 g_block, g_block, dx8; \ |
| 1463 | vdup.u32 dx8, b_dx8; \ |
| 1464 | \ |
| 1465 | vadd.u32 b_block, b_block, dx8; \ |
| 1466 | vmovn.u16 u_whole_8, u_whole; \ |
| 1467 | \ |
| 1468 | vmovn.u16 v_whole_8, v_whole; \ |
| 1469 | \ |
| 1470 | vmovn.u16 b_whole_8, b_whole; \ |
| 1471 | pld [ fb_ptr ]; \ |
| 1472 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1473 | \ |
| 1474 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1475 | setup_blocks_texture_##swizzling(); \ |
| 1476 | \ |
| 1477 | vmovn.u16 r_whole_8, r_whole; \ |
| 1478 | beq 5f; \ |
| 1479 | \ |
| 1480 | 4: \ |
| 1481 | vmovn.u16 g_whole_8, g_whole; \ |
| 1482 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1483 | \ |
| 1484 | vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \ |
| 1485 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1486 | \ |
| 1487 | vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \ |
| 1488 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 1489 | \ |
| 1490 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \ |
| 1491 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 1492 | \ |
| 1493 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1494 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 1495 | \ |
| 1496 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1497 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1498 | \ |
| 1499 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1500 | vdup.u32 dx4, rg_dx4[0]; \ |
| 1501 | \ |
| 1502 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 1503 | vdup.u32 dx4, rg_dx4[1]; \ |
| 1504 | \ |
| 1505 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 1506 | vdup.u32 dx4, b_dx4; \ |
| 1507 | \ |
| 1508 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 1509 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1510 | \ |
| 1511 | vadd.u32 u_block, u_block, dx8; \ |
| 1512 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1513 | \ |
| 1514 | vadd.u32 v_block, v_block, dx8; \ |
| 1515 | vdup.u32 dx8, rg_dx8[0]; \ |
| 1516 | \ |
| 1517 | vadd.u32 r_block, r_block, dx8; \ |
| 1518 | vdup.u32 dx8, rg_dx8[1]; \ |
| 1519 | \ |
| 1520 | vadd.u32 g_block, g_block, dx8; \ |
| 1521 | vdup.u32 dx8, b_dx8; \ |
| 1522 | \ |
| 1523 | vadd.u32 b_block, b_block, dx8; \ |
| 1524 | vmovn.u16 u_whole_8, u_whole; \ |
| 1525 | \ |
| 1526 | add fb_ptr, fb_ptr, #16; \ |
| 1527 | vmovn.u16 v_whole_8, v_whole; \ |
| 1528 | \ |
| 1529 | vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \ |
| 1530 | vmovn.u16 b_whole_8, b_whole; \ |
| 1531 | \ |
| 1532 | pld [ fb_ptr ]; \ |
| 1533 | \ |
| 1534 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1535 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1536 | \ |
| 1537 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1538 | setup_blocks_texture_##swizzling(); \ |
| 1539 | \ |
| 1540 | vmovn.u16 r_whole_8, r_whole; \ |
| 1541 | bne 4b; \ |
| 1542 | \ |
| 1543 | 5: \ |
| 1544 | vmovn.u16 g_whole_8, g_whole; \ |
| 1545 | ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \ |
| 1546 | \ |
| 1547 | vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \ |
| 1548 | vdup.u8 draw_mask, right_mask; \ |
| 1549 | \ |
| 1550 | vmov.u32 fb_mask_ptrs[0], right_mask; \ |
| 1551 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 1552 | vzip.u8 u_whole_8, v_whole_8; \ |
| 1553 | \ |
| 1554 | vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \ |
| 1555 | vst1.u32 { r_whole_8, g_whole_8 }, [ block_ptr_b, :128 ], c_32; \ |
| 1556 | vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \ |
| 1557 | vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \ |
| 1558 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \ |
| 1559 | \ |
| 1560 | 1: \ |
| 1561 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 1562 | add span_b_offset, span_b_offset, #4; \ |
| 1563 | \ |
| 1564 | add span_edge_data, span_edge_data, #8; \ |
| 1565 | subs num_spans, num_spans, #1; \ |
| 1566 | \ |
| 1567 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 1568 | bne 0b; \ |
| 1569 | \ |
| 1570 | ldmia sp!, { r4 - r11, pc }; \ |
| 1571 | \ |
| 1572 | 2: \ |
| 1573 | /* TODO: Load from psx_gpu instead of saving/restoring these */\ |
| 1574 | vpush { texture_mask }; \ |
| 1575 | vpush { uvrg_dx4 }; \ |
| 1576 | \ |
| 1577 | stmdb sp!, { r0 - r3, r12, r14 }; \ |
| 1578 | bl flush_render_block_buffer; \ |
| 1579 | ldmia sp!, { r0 - r3, r12, r14 }; \ |
| 1580 | \ |
| 1581 | vpop { uvrg_dx4 }; \ |
| 1582 | vpop { texture_mask }; \ |
| 1583 | \ |
| 1584 | vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \ |
| 1585 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1586 | \ |
| 1587 | mov num_blocks, span_num_blocks; \ |
| 1588 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1589 | bal 3b \ |
| 1590 | |
| 1591 | |
| 1592 | setup_blocks_shaded_textured_builder(swizzled) |
| 1593 | setup_blocks_shaded_textured_builder(unswizzled) |
| 1594 | |
| 1595 | |
| 1596 | #define setup_blocks_unshaded_textured_builder(swizzling) \ |
| 1597 | .align 3; \ |
| 1598 | \ |
| 1599 | function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \ |
| 1600 | ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \ |
| 1601 | add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \ |
| 1602 | \ |
| 1603 | vld1.u32 { uvrg_dx }, [ uvrg_dx_ptr, :128 ]; \ |
| 1604 | add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \ |
| 1605 | \ |
| 1606 | cmp num_spans, #0; \ |
| 1607 | bxeq lr; \ |
| 1608 | \ |
| 1609 | stmdb sp!, { r4 - r11, r14 }; \ |
| 1610 | vshl.u32 uvrg_dx4, uvrg_dx, #2; \ |
| 1611 | \ |
| 1612 | vshl.u32 uvrg_dx8, uvrg_dx, #3; \ |
| 1613 | \ |
| 1614 | vld2.u8 { texture_mask_u[], texture_mask_v[] }, [ texture_mask_ptr, :16 ]; \ |
| 1615 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 1616 | \ |
| 1617 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 1618 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 1619 | \ |
| 1620 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1621 | \ |
| 1622 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \ |
| 1623 | \ |
| 1624 | 0: \ |
| 1625 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1626 | \ |
| 1627 | ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \ |
| 1628 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 1629 | \ |
| 1630 | ldrh y, [ span_edge_data, #edge_data_y_offset ]; \ |
| 1631 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \ |
| 1632 | \ |
| 1633 | cmp span_num_blocks, #0; \ |
| 1634 | beq 1f; \ |
| 1635 | \ |
| 1636 | ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \ |
| 1637 | add num_blocks, span_num_blocks, num_blocks; \ |
| 1638 | \ |
| 1639 | cmp num_blocks, #MAX_BLOCKS; \ |
| 1640 | bgt 2f; \ |
| 1641 | \ |
| 1642 | 3: \ |
| 1643 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 1644 | \ |
| 1645 | vdup.u32 v_left_x, left_x; \ |
| 1646 | and y, y, #0x3; \ |
| 1647 | \ |
| 1648 | ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \ |
| 1649 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 1650 | \ |
| 1651 | and dither_shift, left_x, #0x03; \ |
| 1652 | \ |
| 1653 | vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \ |
| 1654 | vshr.u32 uvrg_dx, uvrg_dx4, #2; \ |
| 1655 | \ |
| 1656 | mov dither_shift, dither_shift, lsl #3; \ |
| 1657 | vmla.u32 uvrg, uvrg_dx, v_left_x; \ |
| 1658 | \ |
| 1659 | mov c_32, #32; \ |
| 1660 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1661 | \ |
| 1662 | mov dither_row, dither_row, ror dither_shift; \ |
| 1663 | \ |
| 1664 | vdup.u32 dither_offsets_short, dither_row; \ |
| 1665 | add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \ |
| 1666 | \ |
| 1667 | vshll.s8 dither_offsets, dither_offsets_short, #4; \ |
| 1668 | \ |
| 1669 | vdup.u32 u_block, uv[0]; \ |
| 1670 | \ |
| 1671 | vdup.u32 v_block, uv[1]; \ |
| 1672 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 1673 | \ |
| 1674 | vadd.u32 u_block, u_block, block_span; \ |
| 1675 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 1676 | \ |
| 1677 | vadd.u32 v_block, v_block, block_span; \ |
| 1678 | add block_ptr_b, block_ptr_a, #16; \ |
| 1679 | \ |
| 1680 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1681 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1682 | \ |
| 1683 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1684 | \ |
| 1685 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1686 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1687 | \ |
| 1688 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1689 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1690 | \ |
| 1691 | vadd.u32 u_block, u_block, dx8; \ |
| 1692 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1693 | \ |
| 1694 | vadd.u32 v_block, v_block, dx8; \ |
| 1695 | vmovn.u16 u_whole_8, u_whole; \ |
| 1696 | \ |
| 1697 | vmovn.u16 v_whole_8, v_whole; \ |
| 1698 | \ |
| 1699 | pld [ fb_ptr ]; \ |
| 1700 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1701 | \ |
| 1702 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1703 | setup_blocks_texture_##swizzling(); \ |
| 1704 | \ |
| 1705 | beq 5f; \ |
| 1706 | \ |
| 1707 | 4: \ |
| 1708 | vshrn.u32 u_whole_low, u_block, #16; \ |
| 1709 | \ |
| 1710 | vst2.u8 { u_whole_8, v_whole_8 }, [ block_ptr_a, :128 ], c_32; \ |
| 1711 | vshrn.u32 v_whole_low, v_block, #16; \ |
| 1712 | \ |
| 1713 | add block_ptr_b, block_ptr_b, #32; \ |
| 1714 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \ |
| 1715 | \ |
| 1716 | vdup.u32 dx4, uv_dx4[0]; \ |
| 1717 | vaddhn.u32 u_whole_high, u_block, dx4; \ |
| 1718 | vdup.u32 dx4, uv_dx4[1]; \ |
| 1719 | \ |
| 1720 | vaddhn.u32 v_whole_high, v_block, dx4; \ |
| 1721 | vdup.u32 dx8, uv_dx8[0]; \ |
| 1722 | \ |
| 1723 | vadd.u32 u_block, u_block, dx8; \ |
| 1724 | vdup.u32 dx8, uv_dx8[1]; \ |
| 1725 | \ |
| 1726 | vadd.u32 v_block, v_block, dx8; \ |
| 1727 | vmovn.u16 u_whole_8, u_whole; \ |
| 1728 | \ |
| 1729 | add fb_ptr, fb_ptr, #16; \ |
| 1730 | vmovn.u16 v_whole_8, v_whole; \ |
| 1731 | \ |
| 1732 | vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \ |
| 1733 | pld [ fb_ptr ]; \ |
| 1734 | \ |
| 1735 | vmov.u32 fb_mask_ptrs[1], fb_ptr; \ |
| 1736 | subs span_num_blocks, span_num_blocks, #1; \ |
| 1737 | \ |
| 1738 | vand.u8 uv_whole_8, uv_whole_8, texture_mask; \ |
| 1739 | setup_blocks_texture_##swizzling(); \ |
| 1740 | \ |
| 1741 | bne 4b; \ |
| 1742 | \ |
| 1743 | 5: \ |
| 1744 | ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \ |
| 1745 | \ |
| 1746 | vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \ |
| 1747 | vdup.u8 draw_mask, right_mask; \ |
| 1748 | \ |
| 1749 | vmov.u32 fb_mask_ptrs[0], right_mask; \ |
| 1750 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 1751 | vzip.u8 u_whole_8, v_whole_8; \ |
| 1752 | \ |
| 1753 | vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \ |
| 1754 | add block_ptr_b, block_ptr_b, #32; \ |
| 1755 | vst1.u32 { uv_whole_8 }, [ block_ptr_a, :128 ], c_32; \ |
| 1756 | vst1.u32 { dither_offsets }, [ block_ptr_b, :128 ], c_32; \ |
| 1757 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32; \ |
| 1758 | \ |
| 1759 | 1: \ |
| 1760 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 1761 | add span_edge_data, span_edge_data, #8; \ |
| 1762 | subs num_spans, num_spans, #1; \ |
| 1763 | \ |
| 1764 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 1765 | bne 0b; \ |
| 1766 | \ |
| 1767 | ldmia sp!, { r4 - r11, pc }; \ |
| 1768 | \ |
| 1769 | 2: \ |
| 1770 | /* TODO: Load from psx_gpu instead of saving/restoring these */\ |
| 1771 | vpush { texture_mask }; \ |
| 1772 | vpush { uvrg_dx4 }; \ |
| 1773 | \ |
| 1774 | stmdb sp!, { r0 - r3, r12, r14 }; \ |
| 1775 | bl flush_render_block_buffer; \ |
| 1776 | ldmia sp!, { r0 - r3, r12, r14 }; \ |
| 1777 | \ |
| 1778 | vpop { uvrg_dx4 }; \ |
| 1779 | vpop { texture_mask }; \ |
| 1780 | \ |
| 1781 | vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \ |
| 1782 | vmov.u8 fb_mask_ptrs, #0; \ |
| 1783 | \ |
| 1784 | mov num_blocks, span_num_blocks; \ |
| 1785 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 1786 | bal 3b \ |
| 1787 | |
| 1788 | |
| 1789 | setup_blocks_unshaded_textured_builder(swizzled) |
| 1790 | setup_blocks_unshaded_textured_builder(unswizzled) |
| 1791 | |
| 1792 | |
| 1793 | .align 3 |
| 1794 | |
| 1795 | function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect) |
| 1796 | ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ] |
| 1797 | veor.u32 draw_mask, draw_mask, draw_mask |
| 1798 | |
| 1799 | cmp num_spans, #0 |
| 1800 | bxeq lr |
| 1801 | |
| 1802 | stmdb sp!, { r4 - r11, r14 } |
| 1803 | vld1.u32 { test_mask }, [ psx_gpu, :128 ] |
| 1804 | |
| 1805 | ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ] |
| 1806 | |
| 1807 | ubfx color_r, color, #3, #5 |
| 1808 | ubfx color_g, color, #11, #5 |
| 1809 | ubfx color_b, color, #19, #5 |
| 1810 | |
| 1811 | orr color, color_r, color_b, lsl #10 |
| 1812 | orr color, color, color_g, lsl #5 |
| 1813 | |
| 1814 | vdup.u16 colors, color |
| 1815 | |
| 1816 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 1817 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset |
| 1818 | |
| 1819 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset |
| 1820 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6 |
| 1821 | |
| 1822 | 0: |
| 1823 | ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ] |
| 1824 | ldrh y, [ span_edge_data, #edge_data_y_offset ] |
| 1825 | |
| 1826 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ] |
| 1827 | |
| 1828 | cmp span_num_blocks, #0 |
| 1829 | beq 1f |
| 1830 | |
| 1831 | ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ] |
| 1832 | add num_blocks, span_num_blocks, num_blocks |
| 1833 | |
| 1834 | cmp num_blocks, #MAX_BLOCKS |
| 1835 | bgt 2f |
| 1836 | |
| 1837 | 3: |
| 1838 | add fb_ptr, fb_ptr, y, lsl #11 |
| 1839 | and y, y, #0x3 |
| 1840 | |
| 1841 | add fb_ptr, fb_ptr, left_x, lsl #1 |
| 1842 | mov c_32, #32 |
| 1843 | |
| 1844 | subs span_num_blocks, span_num_blocks, #1 |
| 1845 | |
| 1846 | add block_ptr_b, block_ptr_a, #16 |
| 1847 | pld [ fb_ptr ] |
| 1848 | |
| 1849 | vmov.u32 fb_mask_ptrs[1], fb_ptr |
| 1850 | beq 5f |
| 1851 | |
| 1852 | 4: |
| 1853 | vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_32 |
| 1854 | vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32 |
| 1855 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32 |
| 1856 | |
| 1857 | add fb_ptr, fb_ptr, #16 |
| 1858 | add block_ptr_b, block_ptr_b, #32 |
| 1859 | |
| 1860 | pld [ fb_ptr ] |
| 1861 | |
| 1862 | vmov.u32 fb_mask_ptrs[1], fb_ptr |
| 1863 | subs span_num_blocks, span_num_blocks, #1 |
| 1864 | |
| 1865 | bne 4b |
| 1866 | |
| 1867 | 5: |
| 1868 | ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ] |
| 1869 | |
| 1870 | vdup.u8 draw_mask_edge, right_mask |
| 1871 | vtst.u16 draw_mask_edge, draw_mask_edge, test_mask |
| 1872 | |
| 1873 | vst1.u32 { colors }, [ block_ptr_b, :128 ], c_32 |
| 1874 | vst1.u32 { draw_mask_edge }, [ block_ptr_a, :128 ], c_32 |
| 1875 | add block_ptr_b, block_ptr_b, #32 |
| 1876 | vst1.u32 { b_whole_8, fb_mask_ptrs }, [ block_ptr_a, :128 ], c_32 |
| 1877 | |
| 1878 | 1: |
| 1879 | add span_edge_data, span_edge_data, #8 |
| 1880 | subs num_spans, num_spans, #1 |
| 1881 | |
| 1882 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 1883 | bne 0b |
| 1884 | |
| 1885 | ldmia sp!, { r4 - r11, pc } |
| 1886 | |
| 1887 | 2: |
| 1888 | vpush { colors } |
| 1889 | |
| 1890 | stmdb sp!, { r0 - r3, r12, r14 } |
| 1891 | bl flush_render_block_buffer |
| 1892 | ldmia sp!, { r0 - r3, r12, r14 } |
| 1893 | |
| 1894 | vpop { colors } |
| 1895 | |
| 1896 | vld1.u32 { test_mask }, [ psx_gpu, :128 ] |
| 1897 | veor.u32 draw_mask, draw_mask, draw_mask |
| 1898 | |
| 1899 | mov num_blocks, span_num_blocks |
| 1900 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset |
| 1901 | bal 3b |
| 1902 | |
| 1903 | |
| 1904 | #define mask_msb_scalar r14 |
| 1905 | |
| 1906 | #define msb_mask q15 |
| 1907 | |
| 1908 | #define pixels_low d16 |
| 1909 | |
| 1910 | #define msb_mask_low d30 |
| 1911 | #define msb_mask_high d31 |
| 1912 | |
| 1913 | |
| 1914 | .align 3 |
| 1915 | |
| 1916 | function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct) |
| 1917 | ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ] |
| 1918 | |
| 1919 | cmp num_spans, #0 |
| 1920 | bxeq lr |
| 1921 | |
| 1922 | stmdb sp!, { r4 - r11, r14 } |
| 1923 | |
| 1924 | ldr color, [ psx_gpu, #psx_gpu_triangle_color_offset ] |
| 1925 | |
| 1926 | ubfx color_r, color, #3, #5 |
| 1927 | ubfx color_g, color, #11, #5 |
| 1928 | |
| 1929 | ldrh mask_msb_scalar, [ psx_gpu, #psx_gpu_mask_msb_offset ] |
| 1930 | ubfx color_b, color, #19, #5 |
| 1931 | |
| 1932 | orr color, color_r, color_b, lsl #10 |
| 1933 | orr color, color, color_g, lsl #5 |
| 1934 | orr color, color, mask_msb_scalar |
| 1935 | |
| 1936 | vdup.u16 colors, color |
| 1937 | |
| 1938 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset |
| 1939 | orr color, color, lsl #16 |
| 1940 | |
| 1941 | |
| 1942 | 0: |
| 1943 | ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ] |
| 1944 | ldrh y, [ span_edge_data, #edge_data_y_offset ] |
| 1945 | |
| 1946 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ] |
| 1947 | |
| 1948 | cmp span_num_blocks, #0 |
| 1949 | beq 1f |
| 1950 | |
| 1951 | ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ] |
| 1952 | |
| 1953 | add fb_ptr, fb_ptr, y, lsl #11 |
| 1954 | subs span_num_blocks, span_num_blocks, #1 |
| 1955 | |
| 1956 | add fb_ptr, fb_ptr, left_x, lsl #1 |
| 1957 | beq 3f |
| 1958 | |
| 1959 | 2: |
| 1960 | vst1.u32 { colors }, [ fb_ptr ]! |
| 1961 | subs span_num_blocks, span_num_blocks, #1 |
| 1962 | |
| 1963 | bne 2b |
| 1964 | |
| 1965 | 3: |
| 1966 | ldrb right_mask, [ span_edge_data, #edge_data_right_mask_offset ] |
| 1967 | |
| 1968 | cmp right_mask, #0x0 |
| 1969 | beq 5f |
| 1970 | |
| 1971 | tst right_mask, #0xF |
| 1972 | streq color, [ fb_ptr ], #4 |
| 1973 | moveq right_mask, right_mask, lsr #4 |
| 1974 | streq color, [ fb_ptr ], #4 |
| 1975 | |
| 1976 | tst right_mask, #0x3 |
| 1977 | streq color, [ fb_ptr ], #4 |
| 1978 | moveq right_mask, right_mask, lsr #2 |
| 1979 | |
| 1980 | tst right_mask, #0x1 |
| 1981 | streqh color, [ fb_ptr ] |
| 1982 | |
| 1983 | 1: |
| 1984 | add span_edge_data, span_edge_data, #8 |
| 1985 | subs num_spans, num_spans, #1 |
| 1986 | bne 0b |
| 1987 | |
| 1988 | ldmia sp!, { r4 - r11, pc } |
| 1989 | |
| 1990 | 5: |
| 1991 | vst1.u32 { colors }, [ fb_ptr ] |
| 1992 | bal 1b |
| 1993 | |
| 1994 | |
| 1995 | #undef c_64 |
| 1996 | |
| 1997 | #define c_64 r7 |
| 1998 | #define rg_dx_ptr r2 |
| 1999 | |
| 2000 | |
| 2001 | #undef r_block |
| 2002 | #undef g_block |
| 2003 | #undef b_block |
| 2004 | #undef r_whole |
| 2005 | #undef g_whole |
| 2006 | #undef b_whole |
| 2007 | #undef r_whole_low |
| 2008 | #undef r_whole_high |
| 2009 | #undef g_whole_low |
| 2010 | #undef g_whole_high |
| 2011 | #undef b_whole_low |
| 2012 | #undef b_whole_high |
| 2013 | #undef r_whole_8 |
| 2014 | #undef g_whole_8 |
| 2015 | #undef b_whole_8 |
| 2016 | #undef dither_offsets |
| 2017 | #undef rg_dx4 |
| 2018 | #undef rg_dx8 |
| 2019 | #undef dx4 |
| 2020 | #undef dx8 |
| 2021 | #undef v_left_x |
| 2022 | #undef uvrg |
| 2023 | #undef block_span |
| 2024 | #undef rg |
| 2025 | #undef draw_mask |
| 2026 | #undef test_mask |
| 2027 | |
| 2028 | #define r_block q0 |
| 2029 | #define g_block q1 |
| 2030 | #define b_block q2 |
| 2031 | |
| 2032 | #define r_whole q3 |
| 2033 | #define g_whole q4 |
| 2034 | #define b_whole q5 |
| 2035 | |
| 2036 | #define r_whole_low d6 |
| 2037 | #define r_whole_high d7 |
| 2038 | #define g_whole_low d8 |
| 2039 | #define g_whole_high d9 |
| 2040 | #define b_whole_low d10 |
| 2041 | #define b_whole_high d11 |
| 2042 | |
| 2043 | #define gb_whole_8 q6 |
| 2044 | |
| 2045 | #define g_whole_8 d12 |
| 2046 | #define b_whole_8 d13 |
| 2047 | |
| 2048 | #define r_whole_8 d14 |
| 2049 | |
| 2050 | #define pixels q8 |
| 2051 | |
| 2052 | #define rg_dx4 d18 |
| 2053 | #define rg_dx8 d19 |
| 2054 | |
| 2055 | #define dx4 q10 |
| 2056 | #define dx8 q10 |
| 2057 | |
| 2058 | #define v_left_x d6 |
| 2059 | #define uvrg q4 |
| 2060 | #define block_span q5 |
| 2061 | |
| 2062 | #define rg d9 |
| 2063 | |
| 2064 | #define d64_1 d22 |
| 2065 | #define d64_128 d23 |
| 2066 | |
| 2067 | #define d128_4 q12 |
| 2068 | #define d128_0x7 q13 |
| 2069 | |
| 2070 | #define d64_4 d24 |
| 2071 | |
| 2072 | #define dither_offsets q14 |
| 2073 | #define draw_mask q15 |
| 2074 | |
| 2075 | #define dither_offsets_low d28 |
| 2076 | |
| 2077 | #define rg_dx d0 |
| 2078 | #define test_mask q10 |
| 2079 | |
| 2080 | |
| 2081 | #define setup_blocks_shaded_untextured_dither_a_dithered() \ |
| 2082 | vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \ |
| 2083 | vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \ |
| 2084 | |
| 2085 | #define setup_blocks_shaded_untextured_dither_b_dithered() \ |
| 2086 | vqsub.u8 r_whole_8, r_whole_8, d64_4; \ |
| 2087 | vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \ |
| 2088 | |
| 2089 | #define setup_blocks_shaded_untextured_dither_a_undithered() \ |
| 2090 | |
| 2091 | #define setup_blocks_shaded_untextured_dither_b_undithered() \ |
| 2092 | |
| 2093 | |
| 2094 | #define setup_blocks_shaded_untextured_indirect_builder(dithering) \ |
| 2095 | .align 3; \ |
| 2096 | \ |
| 2097 | function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \ |
| 2098 | ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \ |
| 2099 | add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \ |
| 2100 | \ |
| 2101 | vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \ |
| 2102 | \ |
| 2103 | cmp num_spans, #0; \ |
| 2104 | bxeq lr; \ |
| 2105 | \ |
| 2106 | stmdb sp!, { r4 - r11, r14 }; \ |
| 2107 | vshl.u32 rg_dx4, rg_dx, #2; \ |
| 2108 | \ |
| 2109 | ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \ |
| 2110 | vshl.u32 rg_dx8, rg_dx, #3; \ |
| 2111 | \ |
| 2112 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 2113 | \ |
| 2114 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 2115 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 2116 | \ |
| 2117 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 2118 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 2119 | \ |
| 2120 | add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \ |
| 2121 | vmov.u8 d64_1, #1; \ |
| 2122 | \ |
| 2123 | vmov.u8 d128_4, #4; \ |
| 2124 | vmov.u8 d64_128, #128; \ |
| 2125 | \ |
| 2126 | vmov.u8 d128_0x7, #0x7; \ |
| 2127 | \ |
| 2128 | 0: \ |
| 2129 | ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \ |
| 2130 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 2131 | \ |
| 2132 | ldrh y, [ span_edge_data, #edge_data_y_offset ]; \ |
| 2133 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \ |
| 2134 | \ |
| 2135 | cmp span_num_blocks, #0; \ |
| 2136 | beq 1f; \ |
| 2137 | \ |
| 2138 | ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \ |
| 2139 | add num_blocks, span_num_blocks, num_blocks; \ |
| 2140 | \ |
| 2141 | cmp num_blocks, #MAX_BLOCKS; \ |
| 2142 | bgt 2f; \ |
| 2143 | \ |
| 2144 | 3: \ |
| 2145 | ldr b, [ span_b_offset ]; \ |
| 2146 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 2147 | \ |
| 2148 | vdup.u32 v_left_x, left_x; \ |
| 2149 | and y, y, #0x3; \ |
| 2150 | \ |
| 2151 | ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \ |
| 2152 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 2153 | \ |
| 2154 | mla b, b_dx, left_x, b; \ |
| 2155 | and dither_shift, left_x, #0x03; \ |
| 2156 | \ |
| 2157 | vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \ |
| 2158 | vshr.u32 rg_dx, rg_dx4, #2; \ |
| 2159 | \ |
| 2160 | mov dither_shift, dither_shift, lsl #3; \ |
| 2161 | vmla.u32 rg, rg_dx, v_left_x; \ |
| 2162 | \ |
| 2163 | mov c_64, #64; \ |
| 2164 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2165 | \ |
| 2166 | mov dither_row, dither_row, ror dither_shift; \ |
| 2167 | mov b_dx4, b_dx, lsl #2; \ |
| 2168 | \ |
| 2169 | vdup.u32 dither_offsets, dither_row; \ |
| 2170 | add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \ |
| 2171 | \ |
| 2172 | vdup.u32 b_block, b; \ |
| 2173 | vadd.u8 dither_offsets, dither_offsets, d128_4; \ |
| 2174 | \ |
| 2175 | mov b_dx8, b_dx, lsl #3; \ |
| 2176 | vdup.u32 r_block, rg[0]; \ |
| 2177 | vdup.u32 g_block, rg[1]; \ |
| 2178 | \ |
| 2179 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 2180 | \ |
| 2181 | vadd.u32 r_block, r_block, block_span; \ |
| 2182 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 2183 | \ |
| 2184 | vadd.u32 g_block, g_block, block_span; \ |
| 2185 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \ |
| 2186 | \ |
| 2187 | vadd.u32 b_block, b_block, block_span; \ |
| 2188 | add block_ptr_b, block_ptr_a, #16; \ |
| 2189 | \ |
| 2190 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2191 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2192 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2193 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2194 | \ |
| 2195 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2196 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2197 | \ |
| 2198 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2199 | vdup.u32 dx4, b_dx4; \ |
| 2200 | \ |
| 2201 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2202 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2203 | \ |
| 2204 | vadd.u32 r_block, r_block, dx8; \ |
| 2205 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2206 | \ |
| 2207 | vadd.u32 g_block, g_block, dx8; \ |
| 2208 | vdup.u32 dx8, b_dx8; \ |
| 2209 | \ |
| 2210 | vadd.u32 b_block, b_block, dx8; \ |
| 2211 | \ |
| 2212 | vmovn.u16 r_whole_8, r_whole; \ |
| 2213 | vmovn.u16 g_whole_8, g_whole; \ |
| 2214 | vmovn.u16 b_whole_8, b_whole; \ |
| 2215 | \ |
| 2216 | beq 5f; \ |
| 2217 | veor.u32 draw_mask, draw_mask, draw_mask; \ |
| 2218 | \ |
| 2219 | 4: \ |
| 2220 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2221 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2222 | \ |
| 2223 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2224 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2225 | \ |
| 2226 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2227 | str fb_ptr, [ block_ptr_a, #44 ]; \ |
| 2228 | \ |
| 2229 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2230 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2231 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2232 | \ |
| 2233 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2234 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2235 | \ |
| 2236 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2237 | vdup.u32 dx4, b_dx4; \ |
| 2238 | \ |
| 2239 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2240 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2241 | \ |
| 2242 | vmull.u8 pixels, r_whole_8, d64_1; \ |
| 2243 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2244 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2245 | \ |
| 2246 | vadd.u32 r_block, r_block, dx8; \ |
| 2247 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2248 | \ |
| 2249 | vadd.u32 g_block, g_block, dx8; \ |
| 2250 | vdup.u32 dx8, b_dx8; \ |
| 2251 | \ |
| 2252 | vadd.u32 b_block, b_block, dx8; \ |
| 2253 | add fb_ptr, fb_ptr, #16; \ |
| 2254 | \ |
| 2255 | vmovn.u16 r_whole_8, r_whole; \ |
| 2256 | vmovn.u16 g_whole_8, g_whole; \ |
| 2257 | vmovn.u16 b_whole_8, b_whole; \ |
| 2258 | \ |
| 2259 | vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \ |
| 2260 | vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \ |
| 2261 | \ |
| 2262 | pld [ fb_ptr ]; \ |
| 2263 | \ |
| 2264 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2265 | bne 4b; \ |
| 2266 | \ |
| 2267 | 5: \ |
| 2268 | str fb_ptr, [ block_ptr_a, #44 ]; \ |
| 2269 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2270 | \ |
| 2271 | ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \ |
| 2272 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2273 | \ |
| 2274 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2275 | vdup.u8 draw_mask, right_mask; \ |
| 2276 | \ |
| 2277 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2278 | vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \ |
| 2279 | \ |
| 2280 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 2281 | \ |
| 2282 | vmull.u8 pixels, r_whole_8, d64_1; \ |
| 2283 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2284 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2285 | \ |
| 2286 | vst1.u32 { draw_mask }, [ block_ptr_a, :128 ], c_64; \ |
| 2287 | vst1.u32 { pixels }, [ block_ptr_b, :128 ], c_64; \ |
| 2288 | \ |
| 2289 | 1: \ |
| 2290 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 2291 | add span_b_offset, span_b_offset, #4; \ |
| 2292 | \ |
| 2293 | add span_edge_data, span_edge_data, #8; \ |
| 2294 | subs num_spans, num_spans, #1; \ |
| 2295 | \ |
| 2296 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 2297 | bne 0b; \ |
| 2298 | \ |
| 2299 | ldmia sp!, { r4 - r11, pc }; \ |
| 2300 | \ |
| 2301 | 2: \ |
| 2302 | /* TODO: Load from psx_gpu instead of saving/restoring these */\ |
| 2303 | vpush { rg_dx4 }; \ |
| 2304 | \ |
| 2305 | stmdb sp!, { r0 - r3, r12, r14 }; \ |
| 2306 | bl flush_render_block_buffer; \ |
| 2307 | ldmia sp!, { r0 - r3, r12, r14 }; \ |
| 2308 | \ |
| 2309 | vpop { rg_dx4 }; \ |
| 2310 | \ |
| 2311 | vmov.u8 d64_1, #1; \ |
| 2312 | vmov.u8 d128_4, #4; \ |
| 2313 | vmov.u8 d64_128, #128; \ |
| 2314 | vmov.u8 d128_0x7, #0x7; \ |
| 2315 | \ |
| 2316 | vadd.u32 rg_dx8, rg_dx4, rg_dx4; \ |
| 2317 | \ |
| 2318 | mov num_blocks, span_num_blocks; \ |
| 2319 | add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 2320 | bal 3b \ |
| 2321 | |
| 2322 | |
| 2323 | setup_blocks_shaded_untextured_indirect_builder(undithered) |
| 2324 | setup_blocks_shaded_untextured_indirect_builder(dithered) |
| 2325 | |
| 2326 | |
| 2327 | #undef draw_mask |
| 2328 | |
| 2329 | #define mask_msb_ptr r14 |
| 2330 | |
| 2331 | #define draw_mask q0 |
| 2332 | #define pixels_low d16 |
| 2333 | #define pixels_high d17 |
| 2334 | |
| 2335 | |
| 2336 | |
| 2337 | #define setup_blocks_shaded_untextured_direct_builder(dithering) \ |
| 2338 | .align 3; \ |
| 2339 | \ |
| 2340 | function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \ |
| 2341 | ldrh num_spans, [ psx_gpu, #psx_gpu_num_spans_offset ]; \ |
| 2342 | add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \ |
| 2343 | \ |
| 2344 | vld1.u32 { rg_dx }, [ rg_dx_ptr, :64 ]; \ |
| 2345 | \ |
| 2346 | cmp num_spans, #0; \ |
| 2347 | bxeq lr; \ |
| 2348 | \ |
| 2349 | stmdb sp!, { r4 - r11, r14 }; \ |
| 2350 | vshl.u32 rg_dx4, rg_dx, #2; \ |
| 2351 | \ |
| 2352 | ldr b_dx, [ psx_gpu, #psx_gpu_b_dx_offset ]; \ |
| 2353 | vshl.u32 rg_dx8, rg_dx, #3; \ |
| 2354 | \ |
| 2355 | add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \ |
| 2356 | add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \ |
| 2357 | \ |
| 2358 | add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \ |
| 2359 | vmov.u8 d64_1, #1; \ |
| 2360 | \ |
| 2361 | vmov.u8 d128_4, #4; \ |
| 2362 | vmov.u8 d64_128, #128; \ |
| 2363 | \ |
| 2364 | vmov.u8 d128_0x7, #0x7; \ |
| 2365 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 2366 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 2367 | \ |
| 2368 | 0: \ |
| 2369 | ldrh span_num_blocks, [ span_edge_data, #edge_data_num_blocks_offset ]; \ |
| 2370 | add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \ |
| 2371 | \ |
| 2372 | ldrh y, [ span_edge_data, #edge_data_y_offset ]; \ |
| 2373 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \ |
| 2374 | \ |
| 2375 | cmp span_num_blocks, #0; \ |
| 2376 | beq 1f; \ |
| 2377 | \ |
| 2378 | ldrh left_x, [ span_edge_data, #edge_data_left_x_offset ]; \ |
| 2379 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 2380 | \ |
| 2381 | ldr b, [ span_b_offset ]; \ |
| 2382 | vdup.u32 v_left_x, left_x; \ |
| 2383 | and y, y, #0x3; \ |
| 2384 | \ |
| 2385 | ldr dither_row, [ dither_offset_ptr, y, lsl #2 ]; \ |
| 2386 | add fb_ptr, fb_ptr, left_x, lsl #1; \ |
| 2387 | \ |
| 2388 | mla b, b_dx, left_x, b; \ |
| 2389 | and dither_shift, left_x, #0x03; \ |
| 2390 | \ |
| 2391 | vld1.u32 { uvrg }, [ span_uvrg_offset, :128 ]; \ |
| 2392 | vshr.u32 rg_dx, rg_dx4, #2; \ |
| 2393 | \ |
| 2394 | mov dither_shift, dither_shift, lsl #3; \ |
| 2395 | vmla.u32 rg, rg_dx, v_left_x; \ |
| 2396 | \ |
| 2397 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2398 | \ |
| 2399 | mov dither_row, dither_row, ror dither_shift; \ |
| 2400 | mov b_dx4, b_dx, lsl #2; \ |
| 2401 | \ |
| 2402 | vdup.u32 dither_offsets, dither_row; \ |
| 2403 | add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \ |
| 2404 | \ |
| 2405 | vdup.u32 b_block, b; \ |
| 2406 | vadd.u8 dither_offsets, dither_offsets, d128_4; \ |
| 2407 | \ |
| 2408 | mov b_dx8, b_dx, lsl #3; \ |
| 2409 | vdup.u32 r_block, rg[0]; \ |
| 2410 | vdup.u32 g_block, rg[1]; \ |
| 2411 | \ |
| 2412 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 2413 | \ |
| 2414 | vadd.u32 r_block, r_block, block_span; \ |
| 2415 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]!; \ |
| 2416 | \ |
| 2417 | vadd.u32 g_block, g_block, block_span; \ |
| 2418 | vld1.u32 { block_span }, [ block_span_ptr, :128 ]; \ |
| 2419 | \ |
| 2420 | vadd.u32 b_block, b_block, block_span; \ |
| 2421 | add block_ptr_b, block_ptr_a, #16; \ |
| 2422 | \ |
| 2423 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2424 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2425 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2426 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2427 | \ |
| 2428 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2429 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2430 | \ |
| 2431 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2432 | vdup.u32 dx4, b_dx4; \ |
| 2433 | \ |
| 2434 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2435 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2436 | \ |
| 2437 | vadd.u32 r_block, r_block, dx8; \ |
| 2438 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2439 | \ |
| 2440 | vadd.u32 g_block, g_block, dx8; \ |
| 2441 | vdup.u32 dx8, b_dx8; \ |
| 2442 | \ |
| 2443 | vadd.u32 b_block, b_block, dx8; \ |
| 2444 | \ |
| 2445 | vmovn.u16 r_whole_8, r_whole; \ |
| 2446 | vmovn.u16 g_whole_8, g_whole; \ |
| 2447 | vmovn.u16 b_whole_8, b_whole; \ |
| 2448 | \ |
| 2449 | beq 3f; \ |
| 2450 | \ |
| 2451 | 2: \ |
| 2452 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2453 | vshrn.u32 r_whole_low, r_block, #16; \ |
| 2454 | \ |
| 2455 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2456 | vshrn.u32 g_whole_low, g_block, #16; \ |
| 2457 | \ |
| 2458 | vshrn.u32 b_whole_low, b_block, #16; \ |
| 2459 | \ |
| 2460 | vdup.u32 dx4, rg_dx4[0]; \ |
| 2461 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2462 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2463 | \ |
| 2464 | vaddhn.u32 r_whole_high, r_block, dx4; \ |
| 2465 | vdup.u32 dx4, rg_dx4[1]; \ |
| 2466 | \ |
| 2467 | vmov pixels, msb_mask; \ |
| 2468 | vaddhn.u32 g_whole_high, g_block, dx4; \ |
| 2469 | vdup.u32 dx4, b_dx4; \ |
| 2470 | \ |
| 2471 | vaddhn.u32 b_whole_high, b_block, dx4; \ |
| 2472 | vdup.u32 dx8, rg_dx8[0]; \ |
| 2473 | \ |
| 2474 | vmlal.u8 pixels, r_whole_8, d64_1; \ |
| 2475 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2476 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2477 | \ |
| 2478 | vadd.u32 r_block, r_block, dx8; \ |
| 2479 | vdup.u32 dx8, rg_dx8[1]; \ |
| 2480 | \ |
| 2481 | vadd.u32 g_block, g_block, dx8; \ |
| 2482 | vdup.u32 dx8, b_dx8; \ |
| 2483 | \ |
| 2484 | vadd.u32 b_block, b_block, dx8; \ |
| 2485 | \ |
| 2486 | vmovn.u16 r_whole_8, r_whole; \ |
| 2487 | vmovn.u16 g_whole_8, g_whole; \ |
| 2488 | vmovn.u16 b_whole_8, b_whole; \ |
| 2489 | \ |
| 2490 | vst1.u32 { pixels }, [ fb_ptr ]!; \ |
| 2491 | subs span_num_blocks, span_num_blocks, #1; \ |
| 2492 | bne 2b; \ |
| 2493 | \ |
| 2494 | 3: \ |
| 2495 | setup_blocks_shaded_untextured_dither_a_##dithering(); \ |
| 2496 | \ |
| 2497 | ldrh right_mask, [ span_edge_data, #edge_data_right_mask_offset ]; \ |
| 2498 | setup_blocks_shaded_untextured_dither_b_##dithering(); \ |
| 2499 | \ |
| 2500 | vshr.u8 r_whole_8, r_whole_8, #3; \ |
| 2501 | rbit right_mask, right_mask; \ |
| 2502 | vmov pixels, msb_mask; \ |
| 2503 | vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \ |
| 2504 | clz right_mask, right_mask; \ |
| 2505 | \ |
| 2506 | vmlal.u8 pixels, r_whole_8, d64_1; \ |
| 2507 | vmlal.u8 pixels, g_whole_8, d64_4; \ |
| 2508 | vmlal.u8 pixels, b_whole_8, d64_128; \ |
| 2509 | \ |
| 2510 | ldr pc, [ pc, right_mask, lsl #2 ]; \ |
| 2511 | nop; \ |
| 2512 | nop; \ |
| 2513 | .word 4f; \ |
| 2514 | .word 5f; \ |
| 2515 | .word 6f; \ |
| 2516 | .word 7f; \ |
| 2517 | .word 8f; \ |
| 2518 | .word 9f; \ |
| 2519 | .word 10f; \ |
| 2520 | .word 11f; \ |
| 2521 | \ |
| 2522 | 4: \ |
| 2523 | vst1.u16 { pixels_low[0] }, [ fb_ptr ]; \ |
| 2524 | bal 1f; \ |
| 2525 | \ |
| 2526 | 5: \ |
| 2527 | vst1.u32 { pixels_low[0] }, [ fb_ptr ]; \ |
| 2528 | bal 1f; \ |
| 2529 | \ |
| 2530 | 6: \ |
| 2531 | vst1.u32 { pixels_low[0] }, [ fb_ptr ]!; \ |
| 2532 | vst1.u16 { pixels_low[2] }, [ fb_ptr ]; \ |
| 2533 | bal 1f; \ |
| 2534 | \ |
| 2535 | 7: \ |
| 2536 | vst1.u32 { pixels_low }, [ fb_ptr ]; \ |
| 2537 | bal 1f; \ |
| 2538 | \ |
| 2539 | 8: \ |
| 2540 | vst1.u32 { pixels_low }, [ fb_ptr ]!; \ |
| 2541 | vst1.u16 { pixels_high[0] }, [ fb_ptr ]; \ |
| 2542 | bal 1f; \ |
| 2543 | \ |
| 2544 | 9: \ |
| 2545 | vst1.u32 { pixels_low }, [ fb_ptr ]!; \ |
| 2546 | vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \ |
| 2547 | bal 1f; \ |
| 2548 | \ |
| 2549 | 10: \ |
| 2550 | vst1.u32 { pixels_low }, [ fb_ptr ]!; \ |
| 2551 | vst1.u32 { pixels_high[0] }, [ fb_ptr ]!; \ |
| 2552 | vst1.u16 { pixels_high[2] }, [ fb_ptr ]; \ |
| 2553 | bal 1f; \ |
| 2554 | \ |
| 2555 | 11: \ |
| 2556 | vst1.u32 { pixels }, [ fb_ptr ]; \ |
| 2557 | bal 1f; \ |
| 2558 | \ |
| 2559 | 1: \ |
| 2560 | add span_uvrg_offset, span_uvrg_offset, #16; \ |
| 2561 | add span_b_offset, span_b_offset, #4; \ |
| 2562 | \ |
| 2563 | add span_edge_data, span_edge_data, #8; \ |
| 2564 | subs num_spans, num_spans, #1; \ |
| 2565 | \ |
| 2566 | bne 0b; \ |
| 2567 | \ |
| 2568 | ldmia sp!, { r4 - r11, pc } \ |
| 2569 | |
| 2570 | setup_blocks_shaded_untextured_direct_builder(undithered) |
| 2571 | setup_blocks_shaded_untextured_direct_builder(dithered) |
| 2572 | |
| 2573 | |
| 2574 | #undef psx_gpu |
| 2575 | #undef num_blocks |
| 2576 | #undef triangle |
| 2577 | #undef c_64 |
| 2578 | |
| 2579 | #define psx_gpu r0 |
| 2580 | #define block_ptr r1 |
| 2581 | #define num_blocks r2 |
| 2582 | #define uv_01 r3 |
| 2583 | #define uv_23 r4 |
| 2584 | #define uv_45 r5 |
| 2585 | #define uv_67 r6 |
| 2586 | #define uv_0 r7 |
| 2587 | #define uv_1 r3 |
| 2588 | #define uv_2 r8 |
| 2589 | #define uv_3 r4 |
| 2590 | #define uv_4 r9 |
| 2591 | #define uv_5 r5 |
| 2592 | #define uv_6 r10 |
| 2593 | #define uv_7 r6 |
| 2594 | #define texture_ptr r11 |
| 2595 | |
| 2596 | #define pixel_0 r7 |
| 2597 | #define pixel_1 r3 |
| 2598 | #define pixel_2 r8 |
| 2599 | #define pixel_3 r4 |
| 2600 | #define pixel_4 r9 |
| 2601 | #define pixel_5 r5 |
| 2602 | #define pixel_6 r10 |
| 2603 | #define pixel_7 r6 |
| 2604 | |
| 2605 | #define pixels_a r7 |
| 2606 | #define pixels_b r9 |
| 2607 | #define pixels_c r8 |
| 2608 | #define pixels_d r10 |
| 2609 | |
| 2610 | #define c_64 r0 |
| 2611 | |
| 2612 | #define clut_ptr r12 |
| 2613 | #define current_texture_mask r5 |
| 2614 | #define dirty_textures_mask r6 |
| 2615 | |
| 2616 | #define texels d0 |
| 2617 | |
| 2618 | #define clut_low_a d2 |
| 2619 | #define clut_low_b d3 |
| 2620 | #define clut_high_a d4 |
| 2621 | #define clut_high_b d5 |
| 2622 | |
| 2623 | #define clut_a q1 |
| 2624 | #define clut_b q2 |
| 2625 | |
| 2626 | #define texels_low d6 |
| 2627 | #define texels_high d7 |
| 2628 | |
| 2629 | .align 3 |
| 2630 | |
| 2631 | function(texture_blocks_untextured) |
| 2632 | bx lr |
| 2633 | |
| 2634 | |
| 2635 | .align 3 |
| 2636 | |
| 2637 | function(texture_blocks_4bpp) |
| 2638 | stmdb sp!, { r3 - r11, r14 } |
| 2639 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 2640 | |
| 2641 | ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] |
| 2642 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 2643 | |
| 2644 | ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ] |
| 2645 | vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ] |
| 2646 | |
| 2647 | ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ] |
| 2648 | vuzp.u8 clut_a, clut_b |
| 2649 | |
| 2650 | ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ] |
| 2651 | tst dirty_textures_mask, current_texture_mask |
| 2652 | |
| 2653 | bne 1f |
| 2654 | mov c_64, #64 |
| 2655 | |
| 2656 | 0: |
| 2657 | ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 } |
| 2658 | |
| 2659 | uxtah uv_0, texture_ptr, uv_01 |
| 2660 | uxtah uv_1, texture_ptr, uv_01, ror #16 |
| 2661 | |
| 2662 | uxtah uv_2, texture_ptr, uv_23 |
| 2663 | uxtah uv_3, texture_ptr, uv_23, ror #16 |
| 2664 | |
| 2665 | uxtah uv_4, texture_ptr, uv_45 |
| 2666 | ldrb pixel_0, [ uv_0 ] |
| 2667 | |
| 2668 | uxtah uv_5, texture_ptr, uv_45, ror #16 |
| 2669 | ldrb pixel_1, [ uv_1 ] |
| 2670 | |
| 2671 | uxtah uv_6, texture_ptr, uv_67 |
| 2672 | ldrb pixel_2, [ uv_2 ] |
| 2673 | |
| 2674 | uxtah uv_7, texture_ptr, uv_67, ror #16 |
| 2675 | ldrb pixel_3, [ uv_3 ] |
| 2676 | |
| 2677 | ldrb pixel_4, [ uv_4 ] |
| 2678 | subs num_blocks, num_blocks, #1 |
| 2679 | |
| 2680 | ldrb pixel_5, [ uv_5 ] |
| 2681 | orr pixels_a, pixel_0, pixel_1, lsl #8 |
| 2682 | |
| 2683 | ldrb pixel_6, [ uv_6 ] |
| 2684 | orr pixels_b, pixel_4, pixel_5, lsl #8 |
| 2685 | |
| 2686 | ldrb pixel_7, [ uv_7 ] |
| 2687 | orr pixels_a, pixels_a, pixel_2, lsl #16 |
| 2688 | |
| 2689 | orr pixels_b, pixels_b, pixel_6, lsl #16 |
| 2690 | orr pixels_a, pixels_a, pixel_3, lsl #24 |
| 2691 | |
| 2692 | orr pixels_b, pixels_b, pixel_7, lsl #24 |
| 2693 | vmov.u32 texels, pixels_a, pixels_b |
| 2694 | |
| 2695 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels |
| 2696 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels |
| 2697 | |
| 2698 | vst2.u8 { texels_low, texels_high }, [ block_ptr, :128 ], c_64 |
| 2699 | bne 0b |
| 2700 | |
| 2701 | ldmia sp!, { r3 - r11, pc } |
| 2702 | |
| 2703 | 1: |
| 2704 | stmdb sp!, { r1 - r2 } |
| 2705 | bl update_texture_4bpp_cache |
| 2706 | |
| 2707 | mov c_64, #64 |
| 2708 | ldmia sp!, { r1 - r2 } |
| 2709 | bal 0b |
| 2710 | |
| 2711 | |
| 2712 | .align 3 |
| 2713 | |
| 2714 | function(texture_blocks_8bpp) |
| 2715 | stmdb sp!, { r3 - r11, r14 } |
| 2716 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 2717 | |
| 2718 | ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] |
| 2719 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 2720 | |
| 2721 | ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ] |
| 2722 | ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ] |
| 2723 | |
| 2724 | ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ] |
| 2725 | tst dirty_textures_mask, current_texture_mask |
| 2726 | |
| 2727 | bne 1f |
| 2728 | nop |
| 2729 | |
| 2730 | 0: |
| 2731 | ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 } |
| 2732 | |
| 2733 | uxtah uv_0, texture_ptr, uv_01 |
| 2734 | uxtah uv_1, texture_ptr, uv_01, ror #16 |
| 2735 | |
| 2736 | uxtah uv_2, texture_ptr, uv_23 |
| 2737 | uxtah uv_3, texture_ptr, uv_23, ror #16 |
| 2738 | |
| 2739 | uxtah uv_4, texture_ptr, uv_45 |
| 2740 | ldrb pixel_0, [ uv_0 ] |
| 2741 | |
| 2742 | uxtah uv_5, texture_ptr, uv_45, ror #16 |
| 2743 | ldrb pixel_1, [ uv_1 ] |
| 2744 | |
| 2745 | uxtah uv_6, texture_ptr, uv_67 |
| 2746 | ldrb pixel_2, [ uv_2 ] |
| 2747 | |
| 2748 | uxtah uv_7, texture_ptr, uv_67, ror #16 |
| 2749 | ldrb pixel_3, [ uv_3 ] |
| 2750 | |
| 2751 | ldrb pixel_4, [ uv_4 ] |
| 2752 | add pixel_0, pixel_0, pixel_0 |
| 2753 | |
| 2754 | ldrb pixel_5, [ uv_5 ] |
| 2755 | add pixel_1, pixel_1, pixel_1 |
| 2756 | |
| 2757 | ldrb pixel_6, [ uv_6 ] |
| 2758 | add pixel_2, pixel_2, pixel_2 |
| 2759 | |
| 2760 | ldrb pixel_7, [ uv_7 ] |
| 2761 | add pixel_3, pixel_3, pixel_3 |
| 2762 | |
| 2763 | ldrh pixel_0, [ clut_ptr, pixel_0 ] |
| 2764 | add pixel_4, pixel_4, pixel_4 |
| 2765 | |
| 2766 | ldrh pixel_1, [ clut_ptr, pixel_1 ] |
| 2767 | add pixel_5, pixel_5, pixel_5 |
| 2768 | |
| 2769 | ldrh pixel_2, [ clut_ptr, pixel_2 ] |
| 2770 | add pixel_6, pixel_6, pixel_6 |
| 2771 | |
| 2772 | ldrh pixel_3, [ clut_ptr, pixel_3 ] |
| 2773 | add pixel_7, pixel_7, pixel_7 |
| 2774 | |
| 2775 | ldrh pixel_4, [ clut_ptr, pixel_4 ] |
| 2776 | orr pixels_a, pixel_0, pixel_1, lsl #16 |
| 2777 | |
| 2778 | ldrh pixel_5, [ clut_ptr, pixel_5 ] |
| 2779 | orr pixels_c, pixel_2, pixel_3, lsl #16 |
| 2780 | |
| 2781 | ldrh pixel_6, [ clut_ptr, pixel_6 ] |
| 2782 | subs num_blocks, num_blocks, #1 |
| 2783 | |
| 2784 | ldrh pixel_7, [ clut_ptr, pixel_7 ] |
| 2785 | orr pixels_b, pixel_4, pixel_5, lsl #16 |
| 2786 | |
| 2787 | orr pixels_d, pixel_6, pixel_7, lsl #16 |
| 2788 | stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d } |
| 2789 | |
| 2790 | add block_ptr, block_ptr, #64 |
| 2791 | bne 0b |
| 2792 | |
| 2793 | ldmia sp!, { r3 - r11, pc } |
| 2794 | |
| 2795 | 1: |
| 2796 | stmdb sp!, { r1 - r2, r12 } |
| 2797 | |
| 2798 | bl update_texture_8bpp_cache |
| 2799 | |
| 2800 | ldmia sp!, { r1 - r2, r12 } |
| 2801 | bal 0b |
| 2802 | |
| 2803 | |
| 2804 | #undef uv_0 |
| 2805 | #undef uv_1 |
| 2806 | #undef uv_2 |
| 2807 | #undef uv_3 |
| 2808 | #undef uv_4 |
| 2809 | #undef uv_5 |
| 2810 | #undef uv_6 |
| 2811 | #undef uv_7 |
| 2812 | |
| 2813 | #undef pixel_0 |
| 2814 | #undef pixel_1 |
| 2815 | #undef pixel_2 |
| 2816 | #undef pixel_3 |
| 2817 | #undef pixel_4 |
| 2818 | #undef pixel_5 |
| 2819 | #undef pixel_6 |
| 2820 | #undef pixel_7 |
| 2821 | |
| 2822 | #undef texture_ptr |
| 2823 | |
| 2824 | #undef pixels_a |
| 2825 | #undef pixels_b |
| 2826 | #undef pixels_c |
| 2827 | #undef pixels_d |
| 2828 | |
| 2829 | #define psx_gpu r0 |
| 2830 | #define block_ptr r1 |
| 2831 | #define num_blocks r2 |
| 2832 | |
| 2833 | #define uv_0 r3 |
| 2834 | #define uv_1 r4 |
| 2835 | #define u_0 r3 |
| 2836 | #define u_1 r4 |
| 2837 | #define v_0 r5 |
| 2838 | #define v_1 r6 |
| 2839 | |
| 2840 | #define uv_2 r5 |
| 2841 | #define uv_3 r6 |
| 2842 | #define u_2 r5 |
| 2843 | #define u_3 r6 |
| 2844 | #define v_2 r7 |
| 2845 | #define v_3 r8 |
| 2846 | |
| 2847 | #define uv_4 r7 |
| 2848 | #define uv_5 r8 |
| 2849 | #define u_4 r7 |
| 2850 | #define u_5 r8 |
| 2851 | #define v_4 r9 |
| 2852 | #define v_5 r10 |
| 2853 | |
| 2854 | #define uv_6 r9 |
| 2855 | #define uv_7 r10 |
| 2856 | #define u_6 r9 |
| 2857 | #define u_7 r10 |
| 2858 | #define v_6 r11 |
| 2859 | #define v_7 r0 |
| 2860 | |
| 2861 | #define pixel_0 r3 |
| 2862 | #define pixel_1 r4 |
| 2863 | #define pixel_2 r5 |
| 2864 | #define pixel_3 r6 |
| 2865 | #define pixel_4 r7 |
| 2866 | #define pixel_5 r8 |
| 2867 | #define pixel_6 r9 |
| 2868 | #define pixel_7 r10 |
| 2869 | |
| 2870 | #define pixels_a r3 |
| 2871 | #define pixels_b r5 |
| 2872 | #define pixels_c r7 |
| 2873 | #define pixels_d r9 |
| 2874 | |
| 2875 | #define texture_ptr r12 |
| 2876 | |
| 2877 | |
| 2878 | .align 3 |
| 2879 | |
| 2880 | function(texture_blocks_16bpp) |
| 2881 | stmdb sp!, { r3 - r11, r14 } |
| 2882 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 2883 | |
| 2884 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 2885 | ldr texture_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] |
| 2886 | |
| 2887 | 0: |
| 2888 | ldrh uv_0, [ block_ptr ] |
| 2889 | subs num_blocks, num_blocks, #1 |
| 2890 | |
| 2891 | ldrh uv_1, [ block_ptr, #2 ] |
| 2892 | |
| 2893 | and v_0, uv_0, #0xFF00 |
| 2894 | and v_1, uv_1, #0xFF00 |
| 2895 | |
| 2896 | and u_0, uv_0, #0xFF |
| 2897 | and u_1, uv_1, #0xFF |
| 2898 | |
| 2899 | add uv_0, u_0, v_0, lsl #2 |
| 2900 | ldrh uv_2, [ block_ptr, #4 ] |
| 2901 | |
| 2902 | add uv_1, u_1, v_1, lsl #2 |
| 2903 | ldrh uv_3, [ block_ptr, #6 ] |
| 2904 | |
| 2905 | add uv_0, uv_0, uv_0 |
| 2906 | add uv_1, uv_1, uv_1 |
| 2907 | |
| 2908 | and v_2, uv_2, #0xFF00 |
| 2909 | and v_3, uv_3, #0xFF00 |
| 2910 | |
| 2911 | and u_2, uv_2, #0xFF |
| 2912 | and u_3, uv_3, #0xFF |
| 2913 | |
| 2914 | add uv_2, u_2, v_2, lsl #2 |
| 2915 | ldrh uv_4, [ block_ptr, #8 ] |
| 2916 | |
| 2917 | add uv_3, u_3, v_3, lsl #2 |
| 2918 | ldrh uv_5, [ block_ptr, #10 ] |
| 2919 | |
| 2920 | add uv_2, uv_2, uv_2 |
| 2921 | add uv_3, uv_3, uv_3 |
| 2922 | |
| 2923 | and v_4, uv_4, #0xFF00 |
| 2924 | and v_5, uv_5, #0xFF00 |
| 2925 | |
| 2926 | and u_4, uv_4, #0xFF |
| 2927 | and u_5, uv_5, #0xFF |
| 2928 | |
| 2929 | add uv_4, u_4, v_4, lsl #2 |
| 2930 | ldrh uv_6, [ block_ptr, #12 ] |
| 2931 | |
| 2932 | add uv_5, u_5, v_5, lsl #2 |
| 2933 | ldrh uv_7, [ block_ptr, #14 ] |
| 2934 | |
| 2935 | add uv_4, uv_4, uv_4 |
| 2936 | ldrh pixel_0, [ texture_ptr, uv_0 ] |
| 2937 | |
| 2938 | add uv_5, uv_5, uv_5 |
| 2939 | ldrh pixel_1, [ texture_ptr, uv_1 ] |
| 2940 | |
| 2941 | and v_6, uv_6, #0xFF00 |
| 2942 | ldrh pixel_2, [ texture_ptr, uv_2 ] |
| 2943 | |
| 2944 | and v_7, uv_7, #0xFF00 |
| 2945 | ldrh pixel_3, [ texture_ptr, uv_3 ] |
| 2946 | |
| 2947 | and u_6, uv_6, #0xFF |
| 2948 | ldrh pixel_4, [ texture_ptr, uv_4 ] |
| 2949 | |
| 2950 | and u_7, uv_7, #0xFF |
| 2951 | ldrh pixel_5, [ texture_ptr, uv_5 ] |
| 2952 | |
| 2953 | add uv_6, u_6, v_6, lsl #2 |
| 2954 | add uv_7, u_7, v_7, lsl #2 |
| 2955 | |
| 2956 | add uv_6, uv_6, uv_6 |
| 2957 | add uv_7, uv_7, uv_7 |
| 2958 | |
| 2959 | orr pixels_a, pixel_0, pixel_1, lsl #16 |
| 2960 | orr pixels_b, pixel_2, pixel_3, lsl #16 |
| 2961 | |
| 2962 | ldrh pixel_6, [ texture_ptr, uv_6 ] |
| 2963 | orr pixels_c, pixel_4, pixel_5, lsl #16 |
| 2964 | |
| 2965 | ldrh pixel_7, [ texture_ptr, uv_7 ] |
| 2966 | orr pixels_d, pixel_6, pixel_7, lsl #16 |
| 2967 | |
| 2968 | stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d } |
| 2969 | add block_ptr, block_ptr, #64 |
| 2970 | |
| 2971 | bne 0b |
| 2972 | |
| 2973 | ldmia sp!, { r3 - r11, pc } |
| 2974 | |
| 2975 | |
| 2976 | #undef num_blocks |
| 2977 | |
| 2978 | #undef test_mask |
| 2979 | #undef texels |
| 2980 | #undef pixels_b |
| 2981 | #undef pixels |
| 2982 | #undef d64_1 |
| 2983 | #undef d64_4 |
| 2984 | #undef d64_128 |
| 2985 | #undef draw_mask |
| 2986 | #undef msb_mask |
| 2987 | #undef msb_mask_low |
| 2988 | #undef msb_mask_high |
| 2989 | #undef fb_pixels |
| 2990 | |
| 2991 | #undef c_32 |
| 2992 | #undef fb_ptr |
| 2993 | #undef mask_msb_ptr |
| 2994 | |
| 2995 | #define psx_gpu r0 |
| 2996 | #define num_blocks r1 |
| 2997 | #define color_ptr r2 |
| 2998 | #define colors_scalar r2 |
| 2999 | #define colors_scalar_compare r3 |
| 3000 | #define mask_msb_ptr r2 |
| 3001 | |
| 3002 | #define block_ptr_load_a r0 |
| 3003 | #define block_ptr_store r3 |
| 3004 | #define block_ptr_load_b r12 |
| 3005 | #define c_32 r2 |
| 3006 | |
| 3007 | #define c_48 r4 |
| 3008 | #define fb_ptr r14 |
| 3009 | #define draw_mask_bits_scalar r5 |
| 3010 | |
| 3011 | #define d128_0x07 q0 |
| 3012 | #define d128_0x1F q1 |
| 3013 | #define d128_0x8000 q2 |
| 3014 | #define test_mask q3 |
| 3015 | #define texels q4 |
| 3016 | #define colors_rg q5 |
| 3017 | #define colors_b_dm_bits q6 |
| 3018 | #define texels_rg q7 |
| 3019 | #define pixels_r q8 |
| 3020 | #define pixels_g q9 |
| 3021 | #define pixels_b q10 |
| 3022 | #define pixels q11 |
| 3023 | #define zero_mask q4 |
| 3024 | #define draw_mask q12 |
| 3025 | #define msb_mask q13 |
| 3026 | |
| 3027 | #define fb_pixels q8 |
| 3028 | |
| 3029 | #define pixels_gb_low q9 |
| 3030 | |
| 3031 | #define colors_r d10 |
| 3032 | #define colors_g d11 |
| 3033 | #define colors_b d12 |
| 3034 | #define draw_mask_bits d13 |
| 3035 | #define texels_r d14 |
| 3036 | #define texels_g d15 |
| 3037 | #define pixels_r_low d16 |
| 3038 | #define pixels_g_low d18 |
| 3039 | #define pixels_b_low d19 |
| 3040 | #define msb_mask_low d26 |
| 3041 | #define msb_mask_high d27 |
| 3042 | |
| 3043 | #define d64_1 d28 |
| 3044 | #define d64_4 d29 |
| 3045 | #define d64_128 d30 |
| 3046 | #define texels_b d31 |
| 3047 | |
| 3048 | #define shade_blocks_textured_modulated_prologue_indirect() \ |
| 3049 | mov c_48, #48; \ |
| 3050 | add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \ |
| 3051 | |
| 3052 | #define shade_blocks_textured_modulated_prologue_direct() \ |
| 3053 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3054 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] \ |
| 3055 | |
| 3056 | |
| 3057 | #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \ |
| 3058 | |
| 3059 | #define shade_blocks_textured_false_modulation_check_undithered(target) \ |
| 3060 | ldr colors_scalar, [ psx_gpu, #psx_gpu_triangle_color_offset ]; \ |
| 3061 | movw colors_scalar_compare, #0x8080; \ |
| 3062 | \ |
| 3063 | movt colors_scalar_compare, #0x80; \ |
| 3064 | cmp colors_scalar, colors_scalar_compare; \ |
| 3065 | beq shade_blocks_textured_unmodulated_##target \ |
| 3066 | |
| 3067 | #define shade_blocks_textured_false_modulation_check_dithered(target) \ |
| 3068 | |
| 3069 | #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \ |
| 3070 | shade_blocks_textured_false_modulation_check_##dithering(target); \ |
| 3071 | add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \ |
| 3072 | vld1.u32 { colors_r[] }, [ color_ptr, :32 ]; \ |
| 3073 | vdup.u8 colors_g, colors_r[1]; \ |
| 3074 | vdup.u8 colors_b, colors_r[2]; \ |
| 3075 | vdup.u8 colors_r, colors_r[0] \ |
| 3076 | |
| 3077 | |
| 3078 | #define shade_blocks_textured_modulated_load_dithered(target) \ |
| 3079 | vld1.u32 { target }, [ block_ptr_load_b, :128 ] \ |
| 3080 | |
| 3081 | #define shade_blocks_textured_modulated_load_last_dithered(target) \ |
| 3082 | vld1.u32 { target }, [ block_ptr_load_b, :128 ], c_32 \ |
| 3083 | |
| 3084 | #define shade_blocks_textured_modulated_load_undithered(target) \ |
| 3085 | |
| 3086 | #define shade_blocks_textured_modulated_load_last_undithered(target) \ |
| 3087 | add block_ptr_load_b, block_ptr_load_b, #32 \ |
| 3088 | |
| 3089 | #define shade_blocks_textured_modulate_dithered(channel) \ |
| 3090 | vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \ |
| 3091 | |
| 3092 | #define shade_blocks_textured_modulate_undithered(channel) \ |
| 3093 | vmull.u8 pixels_##channel, texels_##channel, colors_##channel \ |
| 3094 | |
| 3095 | |
| 3096 | #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \ |
| 3097 | vst1.u32 { draw_mask }, [ block_ptr_store, :128 ]! \ |
| 3098 | |
| 3099 | #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \ |
| 3100 | ldr fb_ptr, [ block_ptr_load_b, #(offset - 64) ]; \ |
| 3101 | vld1.u32 { fb_pixels }, [ fb_ptr ]; \ |
| 3102 | vbit.u16 pixels, fb_pixels, draw_mask \ |
| 3103 | |
| 3104 | #define shade_blocks_textured_modulated_store_pixels_indirect() \ |
| 3105 | vst1.u32 { pixels }, [ block_ptr_store, :128 ], c_48 \ |
| 3106 | |
| 3107 | #define shade_blocks_textured_modulated_store_pixels_direct() \ |
| 3108 | vst1.u32 { pixels }, [ fb_ptr ] \ |
| 3109 | |
| 3110 | |
| 3111 | #define shade_blocks_textured_modulated_load_rg_shaded() \ |
| 3112 | vld1.u32 { colors_r, colors_g }, [ block_ptr_load_b, :128 ], c_32 \ |
| 3113 | |
| 3114 | #define shade_blocks_textured_modulated_load_rg_unshaded() \ |
| 3115 | add block_ptr_load_b, block_ptr_load_b, #32 \ |
| 3116 | |
| 3117 | #define shade_blocks_textured_modulated_load_bdm_shaded() \ |
| 3118 | vld1.u32 { colors_b, draw_mask_bits }, [ block_ptr_load_a, :128 ], c_32 \ |
| 3119 | |
| 3120 | #define shade_blocks_textured_modulated_load_bdm_unshaded() \ |
| 3121 | ldr draw_mask_bits_scalar, [ block_ptr_load_a, #8 ]; \ |
| 3122 | add block_ptr_load_a, block_ptr_load_a, #32 \ |
| 3123 | |
| 3124 | #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \ |
| 3125 | vdup.u16 draw_mask, draw_mask_bits[0] \ |
| 3126 | |
| 3127 | #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \ |
| 3128 | vdup.u16 draw_mask, draw_mask_bits_scalar \ |
| 3129 | |
| 3130 | |
| 3131 | #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \ |
| 3132 | |
| 3133 | #define shade_blocks_textured_modulated_apply_msb_mask_direct() \ |
| 3134 | vorr.u16 pixels, pixels, msb_mask \ |
| 3135 | |
| 3136 | |
| 3137 | #define shade_blocks_textured_modulated_builder(shading, dithering, target) \ |
| 3138 | .align 3; \ |
| 3139 | \ |
| 3140 | function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \ |
| 3141 | shade_blocks_textured_modulated_prologue_##shading(dithering, target); \ |
| 3142 | stmdb sp!, { r4 - r5, lr }; \ |
| 3143 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 3144 | \ |
| 3145 | vld1.u32 { test_mask }, [ psx_gpu, :128 ]; \ |
| 3146 | \ |
| 3147 | shade_blocks_textured_modulated_prologue_##target(); \ |
| 3148 | \ |
| 3149 | add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3150 | mov c_32, #32; \ |
| 3151 | \ |
| 3152 | add block_ptr_load_b, block_ptr_load_a, #16; \ |
| 3153 | vmov.u8 d64_1, #1; \ |
| 3154 | vmov.u8 d64_4, #4; \ |
| 3155 | vmov.u8 d64_128, #128; \ |
| 3156 | \ |
| 3157 | vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \ |
| 3158 | vmov.u8 d128_0x07, #0x07; \ |
| 3159 | \ |
| 3160 | shade_blocks_textured_modulated_load_rg_##shading(); \ |
| 3161 | vmov.u8 d128_0x1F, #0x1F; \ |
| 3162 | \ |
| 3163 | shade_blocks_textured_modulated_load_bdm_##shading(); \ |
| 3164 | vmov.u16 d128_0x8000, #0x8000; \ |
| 3165 | \ |
| 3166 | vmovn.u16 texels_r, texels; \ |
| 3167 | vshrn.u16 texels_g, texels, #5; \ |
| 3168 | \ |
| 3169 | vshrn.u16 texels_b, texels, #7; \ |
| 3170 | shade_blocks_textured_modulated_expand_draw_mask_##shading(); \ |
| 3171 | \ |
| 3172 | shade_blocks_textured_modulated_load_##dithering(pixels_r); \ |
| 3173 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 3174 | \ |
| 3175 | shade_blocks_textured_modulated_load_##dithering(pixels_g); \ |
| 3176 | vand.u8 texels_rg, texels_rg, d128_0x1F; \ |
| 3177 | \ |
| 3178 | shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \ |
| 3179 | vshr.u8 texels_b, texels_b, #3; \ |
| 3180 | \ |
| 3181 | shade_blocks_textured_modulate_##dithering(r); \ |
| 3182 | shade_blocks_textured_modulate_##dithering(g); \ |
| 3183 | shade_blocks_textured_modulate_##dithering(b); \ |
| 3184 | \ |
| 3185 | vand.u16 pixels, texels, d128_0x8000; \ |
| 3186 | vceq.u16 zero_mask, texels, #0; \ |
| 3187 | \ |
| 3188 | vqshrun.s16 pixels_r_low, pixels_r, #4; \ |
| 3189 | vqshrun.s16 pixels_g_low, pixels_g, #4; \ |
| 3190 | vqshrun.s16 pixels_b_low, pixels_b, #4; \ |
| 3191 | \ |
| 3192 | shade_blocks_textured_modulated_apply_msb_mask_##target(); \ |
| 3193 | vorr.u16 draw_mask, draw_mask, zero_mask; \ |
| 3194 | vshr.u8 pixels_r_low, pixels_r_low, #3; \ |
| 3195 | vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \ |
| 3196 | \ |
| 3197 | subs num_blocks, num_blocks, #1; \ |
| 3198 | beq 1f; \ |
| 3199 | \ |
| 3200 | .align 3; \ |
| 3201 | \ |
| 3202 | 0: \ |
| 3203 | vld1.u32 { texels }, [ block_ptr_load_a, :128 ], c_32; \ |
| 3204 | shade_blocks_textured_modulated_load_rg_##shading(); \ |
| 3205 | vshrn.u16 texels_g, texels, #5; \ |
| 3206 | \ |
| 3207 | shade_blocks_textured_modulated_load_bdm_##shading(); \ |
| 3208 | vshrn.u16 texels_b, texels, #7; \ |
| 3209 | \ |
| 3210 | pld [ block_ptr_load_a ]; \ |
| 3211 | vmovn.u16 texels_r, texels; \ |
| 3212 | vmlal.u8 pixels, pixels_r_low, d64_1; \ |
| 3213 | \ |
| 3214 | vmlal.u8 pixels, pixels_g_low, d64_4; \ |
| 3215 | vmlal.u8 pixels, pixels_b_low, d64_128; \ |
| 3216 | shade_blocks_textured_modulated_store_draw_mask_##target(-4); \ |
| 3217 | \ |
| 3218 | shade_blocks_textured_modulated_load_##dithering(pixels_r); \ |
| 3219 | shade_blocks_textured_modulated_expand_draw_mask_##shading(); \ |
| 3220 | \ |
| 3221 | shade_blocks_textured_modulated_load_##dithering(pixels_g); \ |
| 3222 | vand.u8 texels_rg, texels_rg, d128_0x1F; \ |
| 3223 | \ |
| 3224 | shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \ |
| 3225 | vtst.u16 draw_mask, draw_mask, test_mask; \ |
| 3226 | \ |
| 3227 | shade_blocks_textured_modulated_store_pixels_##target(); \ |
| 3228 | vshr.u8 texels_b, texels_b, #3; \ |
| 3229 | \ |
| 3230 | shade_blocks_textured_modulate_##dithering(r); \ |
| 3231 | shade_blocks_textured_modulate_##dithering(g); \ |
| 3232 | shade_blocks_textured_modulate_##dithering(b); \ |
| 3233 | \ |
| 3234 | vand.u16 pixels, texels, d128_0x8000; \ |
| 3235 | vceq.u16 zero_mask, texels, #0; \ |
| 3236 | \ |
| 3237 | subs num_blocks, num_blocks, #1; \ |
| 3238 | \ |
| 3239 | vqshrun.s16 pixels_r_low, pixels_r, #4; \ |
| 3240 | vqshrun.s16 pixels_g_low, pixels_g, #4; \ |
| 3241 | vqshrun.s16 pixels_b_low, pixels_b, #4; \ |
| 3242 | \ |
| 3243 | shade_blocks_textured_modulated_apply_msb_mask_##target(); \ |
| 3244 | vorr.u16 draw_mask, draw_mask, zero_mask; \ |
| 3245 | vshr.u8 pixels_r_low, pixels_r_low, #3; \ |
| 3246 | vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \ |
| 3247 | \ |
| 3248 | bne 0b; \ |
| 3249 | \ |
| 3250 | 1: \ |
| 3251 | vmlal.u8 pixels, pixels_r_low, d64_1; \ |
| 3252 | vmlal.u8 pixels, pixels_g_low, d64_4; \ |
| 3253 | vmlal.u8 pixels, pixels_b_low, d64_128; \ |
| 3254 | \ |
| 3255 | shade_blocks_textured_modulated_store_draw_mask_##target(28); \ |
| 3256 | shade_blocks_textured_modulated_store_pixels_##target(); \ |
| 3257 | \ |
| 3258 | ldmia sp!, { r4 - r5, pc } \ |
| 3259 | |
| 3260 | |
| 3261 | shade_blocks_textured_modulated_builder(shaded, dithered, direct); |
| 3262 | shade_blocks_textured_modulated_builder(shaded, undithered, direct); |
| 3263 | shade_blocks_textured_modulated_builder(unshaded, dithered, direct); |
| 3264 | shade_blocks_textured_modulated_builder(unshaded, undithered, direct); |
| 3265 | |
| 3266 | shade_blocks_textured_modulated_builder(shaded, dithered, indirect); |
| 3267 | shade_blocks_textured_modulated_builder(shaded, undithered, indirect); |
| 3268 | shade_blocks_textured_modulated_builder(unshaded, dithered, indirect); |
| 3269 | shade_blocks_textured_modulated_builder(unshaded, undithered, indirect); |
| 3270 | |
| 3271 | |
| 3272 | #undef c_64 |
| 3273 | #undef fb_ptr |
| 3274 | #undef color_ptr |
| 3275 | |
| 3276 | #undef color_r |
| 3277 | #undef color_g |
| 3278 | #undef color_b |
| 3279 | |
| 3280 | #undef test_mask |
| 3281 | #undef pixels |
| 3282 | #undef draw_mask |
| 3283 | #undef zero_mask |
| 3284 | #undef fb_pixels |
| 3285 | #undef msb_mask |
| 3286 | #undef msb_mask_low |
| 3287 | #undef msb_mask_high |
| 3288 | |
| 3289 | #define psx_gpu r0 |
| 3290 | #define num_blocks r1 |
| 3291 | #define mask_msb_ptr r2 |
| 3292 | #define color_ptr r3 |
| 3293 | |
| 3294 | #define block_ptr_load r0 |
| 3295 | #define draw_mask_store_ptr r3 |
| 3296 | #define draw_mask_bits_ptr r12 |
| 3297 | #define draw_mask_ptr r12 |
| 3298 | #define pixel_store_ptr r14 |
| 3299 | |
| 3300 | #define fb_ptr_cmp r4 |
| 3301 | |
| 3302 | #define fb_ptr r3 |
| 3303 | #define fb_ptr_next r14 |
| 3304 | |
| 3305 | #define c_64 r2 |
| 3306 | |
| 3307 | #define test_mask q0 |
| 3308 | #define pixels q1 |
| 3309 | #define draw_mask q2 |
| 3310 | #define zero_mask q3 |
| 3311 | #define draw_mask_combined q4 |
| 3312 | #define fb_pixels q5 |
| 3313 | #define fb_pixels_next q6 |
| 3314 | #define msb_mask q7 |
| 3315 | |
| 3316 | #define draw_mask_low d4 |
| 3317 | #define draw_mask_high d5 |
| 3318 | #define msb_mask_low d14 |
| 3319 | #define msb_mask_high d15 |
| 3320 | |
| 3321 | .align 3 |
| 3322 | function(shade_blocks_textured_unmodulated_indirect) |
| 3323 | str r14, [ sp, #-4 ] |
| 3324 | add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40) |
| 3325 | |
| 3326 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 3327 | add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16) |
| 3328 | |
| 3329 | vld1.u32 { test_mask }, [ psx_gpu, :128 ] |
| 3330 | add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 3331 | |
| 3332 | mov c_64, #64 |
| 3333 | add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset |
| 3334 | |
| 3335 | vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64 |
| 3336 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3337 | [ draw_mask_bits_ptr, :16 ], c_64 |
| 3338 | vceq.u16 zero_mask, pixels, #0 |
| 3339 | |
| 3340 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3341 | vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64 |
| 3342 | |
| 3343 | subs num_blocks, num_blocks, #1 |
| 3344 | beq 1f |
| 3345 | |
| 3346 | 0: |
| 3347 | vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64 |
| 3348 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3349 | |
| 3350 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3351 | [ draw_mask_bits_ptr, :16 ], c_64 |
| 3352 | vceq.u16 zero_mask, pixels, #0 |
| 3353 | |
| 3354 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3355 | vst1.u32 { pixels }, [ pixel_store_ptr, :128 ], c_64 |
| 3356 | |
| 3357 | vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64 |
| 3358 | subs num_blocks, num_blocks, #1 |
| 3359 | |
| 3360 | bne 0b |
| 3361 | |
| 3362 | 1: |
| 3363 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3364 | vst1.u32 { draw_mask_combined }, [ draw_mask_store_ptr, :128 ], c_64 |
| 3365 | |
| 3366 | ldr pc, [ sp, #-4 ] |
| 3367 | |
| 3368 | |
| 3369 | .align 3 |
| 3370 | |
| 3371 | function(shade_blocks_textured_unmodulated_direct) |
| 3372 | stmdb sp!, { r4, r14 } |
| 3373 | add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40) |
| 3374 | |
| 3375 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 3376 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset |
| 3377 | |
| 3378 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] |
| 3379 | mov c_64, #64 |
| 3380 | |
| 3381 | vld1.u32 { test_mask }, [ psx_gpu, :128 ] |
| 3382 | add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset |
| 3383 | |
| 3384 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3385 | [ draw_mask_bits_ptr, :16 ], c_64 |
| 3386 | ldr fb_ptr_next, [ block_ptr_load, #44 ] |
| 3387 | |
| 3388 | vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64 |
| 3389 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3390 | vceq.u16 zero_mask, pixels, #0 |
| 3391 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3392 | |
| 3393 | subs num_blocks, num_blocks, #1 |
| 3394 | beq 1f |
| 3395 | |
| 3396 | 0: |
| 3397 | mov fb_ptr, fb_ptr_next |
| 3398 | ldr fb_ptr_next, [ block_ptr_load, #44 ] |
| 3399 | |
| 3400 | vorr.u16 pixels, pixels, msb_mask |
| 3401 | |
| 3402 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3403 | vmov fb_pixels, fb_pixels_next |
| 3404 | |
| 3405 | vld1.u16 { draw_mask_low[], draw_mask_high[] }, \ |
| 3406 | [ draw_mask_bits_ptr, :16 ], c_64 |
| 3407 | vbif.u16 fb_pixels, pixels, draw_mask_combined |
| 3408 | |
| 3409 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr |
| 3410 | pld [ fb_ptr_next, #64 ] |
| 3411 | |
| 3412 | add fb_ptr_cmp, fb_ptr_cmp, #14 |
| 3413 | vld1.u32 { pixels }, [ block_ptr_load, :128 ], c_64 |
| 3414 | |
| 3415 | cmp fb_ptr_cmp, #28 |
| 3416 | bls 4f |
| 3417 | |
| 3418 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3419 | vceq.u16 zero_mask, pixels, #0 |
| 3420 | |
| 3421 | vst1.u16 { fb_pixels }, [ fb_ptr ] |
| 3422 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3423 | |
| 3424 | 3: |
| 3425 | subs num_blocks, num_blocks, #1 |
| 3426 | bne 0b |
| 3427 | |
| 3428 | 1: |
| 3429 | vorr.u16 draw_mask_combined, draw_mask, zero_mask |
| 3430 | vbif.u16 fb_pixels_next, pixels, draw_mask_combined |
| 3431 | |
| 3432 | vst1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3433 | |
| 3434 | ldmia sp!, { r4, pc } |
| 3435 | |
| 3436 | 4: |
| 3437 | vst1.u16 { fb_pixels }, [ fb_ptr ] |
| 3438 | vceq.u16 zero_mask, pixels, #0 |
| 3439 | |
| 3440 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3441 | vtst.u16 draw_mask, draw_mask, test_mask |
| 3442 | |
| 3443 | bal 3b |
| 3444 | |
| 3445 | |
| 3446 | function(shade_blocks_unshaded_untextured_indirect) |
| 3447 | bx lr |
| 3448 | |
| 3449 | .align 3 |
| 3450 | |
| 3451 | function(shade_blocks_unshaded_untextured_direct) |
| 3452 | stmdb sp!, { r4, r14 } |
| 3453 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 3454 | |
| 3455 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 3456 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset |
| 3457 | |
| 3458 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] |
| 3459 | add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16) |
| 3460 | |
| 3461 | add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44) |
| 3462 | vld1.u16 { pixels }, [ color_ptr, :128 ] |
| 3463 | |
| 3464 | mov c_64, #64 |
| 3465 | vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64 |
| 3466 | |
| 3467 | vorr.u16 pixels, pixels, msb_mask |
| 3468 | subs num_blocks, num_blocks, #1 |
| 3469 | |
| 3470 | ldr fb_ptr_next, [ block_ptr_load ], #64 |
| 3471 | |
| 3472 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3473 | beq 1f |
| 3474 | |
| 3475 | 0: |
| 3476 | vmov fb_pixels, fb_pixels_next |
| 3477 | mov fb_ptr, fb_ptr_next |
| 3478 | ldr fb_ptr_next, [ block_ptr_load ], #64 |
| 3479 | |
| 3480 | vbif.u16 fb_pixels, pixels, draw_mask |
| 3481 | vld1.u16 { draw_mask }, [ draw_mask_ptr, :128 ], c_64 |
| 3482 | |
| 3483 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr |
| 3484 | add fb_ptr_cmp, fb_ptr_cmp, #14 |
| 3485 | cmp fb_ptr_cmp, #28 |
| 3486 | bls 4f |
| 3487 | |
| 3488 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3489 | vst1.u16 { fb_pixels }, [ fb_ptr ] |
| 3490 | |
| 3491 | 3: |
| 3492 | subs num_blocks, num_blocks, #1 |
| 3493 | bne 0b |
| 3494 | |
| 3495 | 1: |
| 3496 | vbif.u16 fb_pixels_next, pixels, draw_mask |
| 3497 | vst1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3498 | |
| 3499 | ldmia sp!, { r4, pc } |
| 3500 | |
| 3501 | 4: |
| 3502 | vst1.u16 { fb_pixels }, [ fb_ptr ] |
| 3503 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ] |
| 3504 | bal 3b |
| 3505 | |
| 3506 | |
| 3507 | #undef draw_mask_ptr |
| 3508 | #undef c_64 |
| 3509 | #undef fb_ptr |
| 3510 | #undef fb_ptr_next |
| 3511 | #undef fb_ptr_cmp |
| 3512 | |
| 3513 | #define psx_gpu r0 |
| 3514 | #define num_blocks r1 |
| 3515 | #define msb_mask_ptr r2 |
| 3516 | #define pixel_ptr r3 |
| 3517 | #define draw_mask_ptr r0 |
| 3518 | #define c_64 r2 |
| 3519 | #define fb_ptr r12 |
| 3520 | #define fb_ptr_next r14 |
| 3521 | #define fb_ptr_cmp r4 |
| 3522 | |
| 3523 | #undef msb_mask |
| 3524 | #undef draw_mask |
| 3525 | #undef pixels |
| 3526 | #undef fb_pixels |
| 3527 | #undef d128_0x8000 |
| 3528 | #undef msb_mask_low |
| 3529 | #undef msb_mask_high |
| 3530 | #undef draw_mask_next |
| 3531 | #undef pixels_g |
| 3532 | #undef blend_pixels |
| 3533 | #undef fb_pixels_next |
| 3534 | |
| 3535 | #define msb_mask q0 |
| 3536 | #define draw_mask q1 |
| 3537 | #define pixels q2 |
| 3538 | #define fb_pixels q3 |
| 3539 | #define blend_pixels q4 |
| 3540 | #define pixels_no_msb q5 |
| 3541 | #define blend_mask q6 |
| 3542 | #define fb_pixels_no_msb q7 |
| 3543 | #define d128_0x8000 q8 |
| 3544 | #define d128_0x0421 q9 |
| 3545 | #define fb_pixels_next q10 |
| 3546 | #define blend_pixels_next q11 |
| 3547 | #define pixels_next q12 |
| 3548 | #define draw_mask_next q13 |
| 3549 | #define write_mask q14 |
| 3550 | |
| 3551 | #define pixels_rb q5 |
| 3552 | #define pixels_mg q7 |
| 3553 | #define pixels_g q7 |
| 3554 | #define d128_0x7C1F q8 |
| 3555 | #define d128_0x03E0 q9 |
| 3556 | #define fb_pixels_rb q10 |
| 3557 | #define fb_pixels_g q11 |
| 3558 | #define fb_pixels_masked q11 |
| 3559 | #define d128_0x83E0 q15 |
| 3560 | #define pixels_fourth q7 |
| 3561 | #define d128_0x1C07 q12 |
| 3562 | #define d128_0x00E0 q13 |
| 3563 | #define d128_0x80E0 q13 |
| 3564 | |
| 3565 | #define msb_mask_low d0 |
| 3566 | #define msb_mask_high d1 |
| 3567 | |
| 3568 | #define blend_blocks_average_set_blend_mask_textured(source) \ |
| 3569 | vclt.s16 blend_mask, source, #0 \ |
| 3570 | |
| 3571 | #define blend_blocks_average_set_stp_bit_textured() \ |
| 3572 | vorr.u16 blend_pixels, #0x8000 \ |
| 3573 | |
| 3574 | #define blend_blocks_average_combine_textured(source) \ |
| 3575 | vbif.u16 blend_pixels, source, blend_mask \ |
| 3576 | |
| 3577 | #define blend_blocks_average_set_blend_mask_untextured(source) \ |
| 3578 | |
| 3579 | #define blend_blocks_average_set_stp_bit_untextured() \ |
| 3580 | |
| 3581 | #define blend_blocks_average_combine_untextured(source) \ |
| 3582 | |
| 3583 | #define blend_blocks_average_mask_set_on() \ |
| 3584 | vclt.s16 write_mask, fb_pixels_next, #0 \ |
| 3585 | |
| 3586 | #define blend_blocks_average_mask_copy_on() \ |
| 3587 | vorr.u16 draw_mask, draw_mask_next, write_mask \ |
| 3588 | |
| 3589 | #define blend_blocks_average_mask_copy_b_on() \ |
| 3590 | vorr.u16 draw_mask_next, draw_mask_next, write_mask \ |
| 3591 | |
| 3592 | #define blend_blocks_average_mask_set_off() \ |
| 3593 | |
| 3594 | #define blend_blocks_average_mask_copy_off() \ |
| 3595 | vmov draw_mask, draw_mask_next \ |
| 3596 | |
| 3597 | #define blend_blocks_average_mask_copy_b_off() \ |
| 3598 | |
| 3599 | #define blend_blocks_average_builder(texturing, mask_evaluate) \ |
| 3600 | .align 3; \ |
| 3601 | \ |
| 3602 | function(blend_blocks_##texturing##_average_##mask_evaluate) \ |
| 3603 | stmdb sp!, { r4, r14 }; \ |
| 3604 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3605 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 3606 | \ |
| 3607 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3608 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 3609 | \ |
| 3610 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3611 | mov c_64, #64; \ |
| 3612 | \ |
| 3613 | vmov.u16 d128_0x8000, #0x8000; \ |
| 3614 | vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3615 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3616 | \ |
| 3617 | vmov.u16 d128_0x0421, #0x0400; \ |
| 3618 | vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \ |
| 3619 | \ |
| 3620 | vorr.u16 d128_0x0421, #0x0021; \ |
| 3621 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \ |
| 3622 | \ |
| 3623 | veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \ |
| 3624 | vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \ |
| 3625 | vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \ |
| 3626 | vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \ |
| 3627 | blend_blocks_average_mask_set_##mask_evaluate(); \ |
| 3628 | vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \ |
| 3629 | \ |
| 3630 | subs num_blocks, num_blocks, #1; \ |
| 3631 | beq 1f; \ |
| 3632 | \ |
| 3633 | 0: \ |
| 3634 | mov fb_ptr, fb_ptr_next; \ |
| 3635 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3636 | \ |
| 3637 | vmov pixels, pixels_next; \ |
| 3638 | vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \ |
| 3639 | \ |
| 3640 | vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \ |
| 3641 | \ |
| 3642 | blend_blocks_average_mask_copy_##mask_evaluate(); \ |
| 3643 | vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3644 | \ |
| 3645 | blend_blocks_average_set_blend_mask_##texturing(pixels); \ |
| 3646 | blend_blocks_average_set_stp_bit_##texturing(); \ |
| 3647 | vmov fb_pixels, fb_pixels_next; \ |
| 3648 | blend_blocks_average_combine_##texturing(pixels); \ |
| 3649 | \ |
| 3650 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3651 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 3652 | cmp fb_ptr_cmp, #28; \ |
| 3653 | bls 2f; \ |
| 3654 | \ |
| 3655 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \ |
| 3656 | veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \ |
| 3657 | \ |
| 3658 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3659 | vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \ |
| 3660 | \ |
| 3661 | vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \ |
| 3662 | vbif.u16 fb_pixels, blend_pixels, draw_mask; \ |
| 3663 | \ |
| 3664 | vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \ |
| 3665 | vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \ |
| 3666 | blend_blocks_average_mask_set_##mask_evaluate(); \ |
| 3667 | vst1.u16 { fb_pixels }, [ fb_ptr ]; \ |
| 3668 | \ |
| 3669 | 3: \ |
| 3670 | subs num_blocks, num_blocks, #1; \ |
| 3671 | bne 0b; \ |
| 3672 | \ |
| 3673 | 1: \ |
| 3674 | blend_blocks_average_mask_copy_b_##mask_evaluate(); \ |
| 3675 | vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \ |
| 3676 | \ |
| 3677 | blend_blocks_average_set_blend_mask_##texturing(pixels_next); \ |
| 3678 | blend_blocks_average_set_stp_bit_##texturing(); \ |
| 3679 | blend_blocks_average_combine_##texturing(pixels_next); \ |
| 3680 | \ |
| 3681 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3682 | vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \ |
| 3683 | vst1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \ |
| 3684 | \ |
| 3685 | ldmia sp!, { r4, pc }; \ |
| 3686 | \ |
| 3687 | 2: \ |
| 3688 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3689 | vbif.u16 fb_pixels, blend_pixels, draw_mask; \ |
| 3690 | vst1.u16 { fb_pixels }, [ fb_ptr ]; \ |
| 3691 | \ |
| 3692 | vld1.u16 { fb_pixels_next }, [ fb_ptr_next ]; \ |
| 3693 | veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \ |
| 3694 | vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \ |
| 3695 | vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \ |
| 3696 | vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \ |
| 3697 | vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \ |
| 3698 | \ |
| 3699 | bal 3b \ |
| 3700 | |
| 3701 | blend_blocks_average_builder(textured, off) |
| 3702 | blend_blocks_average_builder(untextured, off) |
| 3703 | blend_blocks_average_builder(textured, on) |
| 3704 | blend_blocks_average_builder(untextured, on) |
| 3705 | |
| 3706 | |
| 3707 | #define blend_blocks_add_mask_set_on() \ |
| 3708 | vclt.s16 write_mask, fb_pixels, #0 \ |
| 3709 | |
| 3710 | #define blend_blocks_add_mask_copy_on() \ |
| 3711 | vorr.u16 draw_mask, draw_mask, write_mask \ |
| 3712 | |
| 3713 | #define blend_blocks_add_mask_set_off() \ |
| 3714 | |
| 3715 | #define blend_blocks_add_mask_copy_off() \ |
| 3716 | |
| 3717 | |
| 3718 | #define blend_blocks_add_textured_builder(mask_evaluate) \ |
| 3719 | .align 3; \ |
| 3720 | \ |
| 3721 | function(blend_blocks_textured_add_##mask_evaluate) \ |
| 3722 | stmdb sp!, { r4, r14 }; \ |
| 3723 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3724 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 3725 | \ |
| 3726 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3727 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 3728 | \ |
| 3729 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3730 | mov c_64, #64; \ |
| 3731 | \ |
| 3732 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 3733 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 3734 | vmov.u16 d128_0x83E0, #0x8000; \ |
| 3735 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 3736 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 3737 | vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \ |
| 3738 | \ |
| 3739 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3740 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3741 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 3742 | vclt.s16 blend_mask, pixels, #0; \ |
| 3743 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3744 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3745 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3746 | \ |
| 3747 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3748 | vorr.u16 pixels, pixels, msb_mask; \ |
| 3749 | vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \ |
| 3750 | vand.u16 pixels_mg, pixels, d128_0x83E0; \ |
| 3751 | vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \ |
| 3752 | vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \ |
| 3753 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3754 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \ |
| 3755 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3756 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \ |
| 3757 | \ |
| 3758 | subs num_blocks, num_blocks, #1; \ |
| 3759 | beq 1f; \ |
| 3760 | \ |
| 3761 | 0: \ |
| 3762 | mov fb_ptr, fb_ptr_next; \ |
| 3763 | \ |
| 3764 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3765 | \ |
| 3766 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 3767 | vclt.s16 blend_mask, pixels, #0; \ |
| 3768 | \ |
| 3769 | vorr.u16 pixels, pixels, msb_mask; \ |
| 3770 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3771 | vand.u16 pixels_mg, pixels, d128_0x83E0; \ |
| 3772 | \ |
| 3773 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3774 | pld [ fb_ptr_next, #64 ]; \ |
| 3775 | \ |
| 3776 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3777 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3778 | \ |
| 3779 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 3780 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3781 | \ |
| 3782 | cmp fb_ptr_cmp, #28; \ |
| 3783 | bls 2f; \ |
| 3784 | \ |
| 3785 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3786 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3787 | vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \ |
| 3788 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3789 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3790 | vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \ |
| 3791 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 3792 | \ |
| 3793 | 3: \ |
| 3794 | vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \ |
| 3795 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3796 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \ |
| 3797 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3798 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \ |
| 3799 | \ |
| 3800 | subs num_blocks, num_blocks, #1; \ |
| 3801 | bne 0b; \ |
| 3802 | \ |
| 3803 | 1: \ |
| 3804 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3805 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3806 | vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \ |
| 3807 | \ |
| 3808 | ldmia sp!, { r4, pc }; \ |
| 3809 | \ |
| 3810 | 2: \ |
| 3811 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 3812 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3813 | \ |
| 3814 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3815 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3816 | vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \ |
| 3817 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3818 | vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \ |
| 3819 | bal 3b \ |
| 3820 | |
| 3821 | |
| 3822 | #define blend_blocks_add_untextured_builder(mask_evaluate) \ |
| 3823 | .align 3; \ |
| 3824 | \ |
| 3825 | function(blend_blocks_untextured_add_##mask_evaluate) \ |
| 3826 | stmdb sp!, { r4, r14 }; \ |
| 3827 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3828 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 3829 | \ |
| 3830 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3831 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 3832 | \ |
| 3833 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3834 | mov c_64, #64; \ |
| 3835 | \ |
| 3836 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 3837 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 3838 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 3839 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 3840 | \ |
| 3841 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3842 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3843 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 3844 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3845 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3846 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3847 | \ |
| 3848 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3849 | vand.u16 pixels_g, pixels, d128_0x03E0; \ |
| 3850 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3851 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 3852 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3853 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 3854 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3855 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 3856 | \ |
| 3857 | subs num_blocks, num_blocks, #1; \ |
| 3858 | beq 1f; \ |
| 3859 | \ |
| 3860 | 0: \ |
| 3861 | mov fb_ptr, fb_ptr_next; \ |
| 3862 | \ |
| 3863 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3864 | \ |
| 3865 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 3866 | \ |
| 3867 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3868 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3869 | vand.u16 pixels_g, pixels, d128_0x03E0; \ |
| 3870 | \ |
| 3871 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3872 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3873 | \ |
| 3874 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 3875 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 3876 | cmp fb_ptr_cmp, #28; \ |
| 3877 | bls 2f; \ |
| 3878 | \ |
| 3879 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3880 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3881 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3882 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3883 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3884 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 3885 | \ |
| 3886 | 3: \ |
| 3887 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 3888 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3889 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 3890 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 3891 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 3892 | \ |
| 3893 | subs num_blocks, num_blocks, #1; \ |
| 3894 | bne 0b; \ |
| 3895 | \ |
| 3896 | 1: \ |
| 3897 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 3898 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 3899 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 3900 | vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \ |
| 3901 | \ |
| 3902 | ldmia sp!, { r4, pc }; \ |
| 3903 | \ |
| 3904 | 2: \ |
| 3905 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 3906 | vand.u16 pixels_rb, pixels, d128_0x7C1F; \ |
| 3907 | \ |
| 3908 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3909 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 3910 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 3911 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3912 | bal 3b \ |
| 3913 | |
| 3914 | |
| 3915 | blend_blocks_add_textured_builder(off) |
| 3916 | blend_blocks_add_textured_builder(on) |
| 3917 | blend_blocks_add_untextured_builder(off) |
| 3918 | blend_blocks_add_untextured_builder(on) |
| 3919 | |
| 3920 | #define blend_blocks_subtract_set_blend_mask_textured() \ |
| 3921 | vclt.s16 blend_mask, pixels_next, #0 \ |
| 3922 | |
| 3923 | #define blend_blocks_subtract_combine_textured() \ |
| 3924 | vbif.u16 blend_pixels, pixels, blend_mask \ |
| 3925 | |
| 3926 | #define blend_blocks_subtract_set_stb_textured() \ |
| 3927 | vorr.u16 blend_pixels, #0x8000 \ |
| 3928 | |
| 3929 | #define blend_blocks_subtract_msb_mask_textured() \ |
| 3930 | vorr.u16 pixels, pixels_next, msb_mask \ |
| 3931 | |
| 3932 | #define blend_blocks_subtract_set_blend_mask_untextured() \ |
| 3933 | |
| 3934 | #define blend_blocks_subtract_combine_untextured() \ |
| 3935 | |
| 3936 | #define blend_blocks_subtract_set_stb_untextured() \ |
| 3937 | vorr.u16 blend_pixels, blend_pixels, msb_mask \ |
| 3938 | |
| 3939 | #define blend_blocks_subtract_msb_mask_untextured() \ |
| 3940 | |
| 3941 | |
| 3942 | #define blend_blocks_subtract_mask_set_on() \ |
| 3943 | vclt.s16 write_mask, fb_pixels, #0 \ |
| 3944 | |
| 3945 | #define blend_blocks_subtract_mask_copy_on() \ |
| 3946 | vorr.u16 draw_mask, draw_mask_next, write_mask \ |
| 3947 | |
| 3948 | #define blend_blocks_subtract_mask_set_off() \ |
| 3949 | |
| 3950 | #define blend_blocks_subtract_mask_copy_off() \ |
| 3951 | vmov draw_mask, draw_mask_next \ |
| 3952 | |
| 3953 | |
| 3954 | #define blend_blocks_subtract_builder(texturing, mask_evaluate) \ |
| 3955 | .align 3; \ |
| 3956 | \ |
| 3957 | function(blend_blocks_##texturing##_subtract_##mask_evaluate) \ |
| 3958 | stmdb sp!, { r4, r14 }; \ |
| 3959 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 3960 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 3961 | \ |
| 3962 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 3963 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 3964 | \ |
| 3965 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 3966 | mov c_64, #64; \ |
| 3967 | \ |
| 3968 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 3969 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 3970 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 3971 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 3972 | \ |
| 3973 | vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3974 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3975 | vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \ |
| 3976 | blend_blocks_subtract_set_blend_mask_##texturing(); \ |
| 3977 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 3978 | blend_blocks_subtract_mask_set_##mask_evaluate(); \ |
| 3979 | vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \ |
| 3980 | \ |
| 3981 | vand.u16 pixels_g, pixels_next, d128_0x03E0; \ |
| 3982 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 3983 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 3984 | vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 3985 | vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 3986 | \ |
| 3987 | subs num_blocks, num_blocks, #1; \ |
| 3988 | beq 1f; \ |
| 3989 | \ |
| 3990 | 0: \ |
| 3991 | blend_blocks_subtract_mask_copy_##mask_evaluate(); \ |
| 3992 | mov fb_ptr, fb_ptr_next; \ |
| 3993 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 3994 | \ |
| 3995 | vld1.u32 { draw_mask_next }, [ draw_mask_ptr, :128 ], c_64; \ |
| 3996 | blend_blocks_subtract_msb_mask_##texturing(); \ |
| 3997 | \ |
| 3998 | vld1.u32 { pixels_next }, [ pixel_ptr, :128 ], c_64; \ |
| 3999 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4000 | vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \ |
| 4001 | blend_blocks_subtract_set_stb_##texturing(); \ |
| 4002 | vand.u16 pixels_g, pixels_next, d128_0x03E0; \ |
| 4003 | blend_blocks_subtract_combine_##texturing(); \ |
| 4004 | blend_blocks_subtract_set_blend_mask_##texturing(); \ |
| 4005 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4006 | \ |
| 4007 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 4008 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 4009 | cmp fb_ptr_cmp, #28; \ |
| 4010 | bls 2f; \ |
| 4011 | \ |
| 4012 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4013 | blend_blocks_subtract_mask_set_##mask_evaluate(); \ |
| 4014 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4015 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4016 | vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4017 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 4018 | vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4019 | \ |
| 4020 | 3: \ |
| 4021 | subs num_blocks, num_blocks, #1; \ |
| 4022 | bne 0b; \ |
| 4023 | \ |
| 4024 | 1: \ |
| 4025 | blend_blocks_subtract_mask_copy_##mask_evaluate(); \ |
| 4026 | \ |
| 4027 | blend_blocks_subtract_msb_mask_##texturing(); \ |
| 4028 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4029 | blend_blocks_subtract_set_stb_##texturing(); \ |
| 4030 | blend_blocks_subtract_combine_##texturing(); \ |
| 4031 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4032 | vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \ |
| 4033 | \ |
| 4034 | ldmia sp!, { r4, pc }; \ |
| 4035 | \ |
| 4036 | 2: \ |
| 4037 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 4038 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4039 | blend_blocks_subtract_mask_set_##mask_evaluate(); \ |
| 4040 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4041 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4042 | vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4043 | vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4044 | bal 3b \ |
| 4045 | |
| 4046 | |
| 4047 | blend_blocks_subtract_builder(textured, off) |
| 4048 | blend_blocks_subtract_builder(textured, on) |
| 4049 | blend_blocks_subtract_builder(untextured, off) |
| 4050 | blend_blocks_subtract_builder(untextured, on) |
| 4051 | |
| 4052 | |
| 4053 | #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \ |
| 4054 | .align 3; \ |
| 4055 | \ |
| 4056 | function(blend_blocks_textured_add_fourth_##mask_evaluate) \ |
| 4057 | stmdb sp!, { r4, r14 }; \ |
| 4058 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 4059 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 4060 | \ |
| 4061 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 4062 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 4063 | \ |
| 4064 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 4065 | mov c_64, #64; \ |
| 4066 | \ |
| 4067 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 4068 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 4069 | vmov.u16 d128_0x1C07, #0x1C00; \ |
| 4070 | vmov.u16 d128_0x00E0, #0x00E0; \ |
| 4071 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 4072 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 4073 | vorr.u16 d128_0x1C07, #0x0007; \ |
| 4074 | \ |
| 4075 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 4076 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 4077 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 4078 | vclt.s16 blend_mask, pixels, #0; \ |
| 4079 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4080 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4081 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4082 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4083 | \ |
| 4084 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4085 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4086 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4087 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4088 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4089 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4090 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4091 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4092 | \ |
| 4093 | subs num_blocks, num_blocks, #1; \ |
| 4094 | beq 1f; \ |
| 4095 | \ |
| 4096 | 0: \ |
| 4097 | mov fb_ptr, fb_ptr_next; \ |
| 4098 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 4099 | \ |
| 4100 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4101 | vbif.u16 blend_pixels, pixels, blend_mask; \ |
| 4102 | \ |
| 4103 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 4104 | vclt.s16 blend_mask, pixels, #0; \ |
| 4105 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4106 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4107 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4108 | \ |
| 4109 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4110 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 4111 | \ |
| 4112 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 4113 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 4114 | cmp fb_ptr_cmp, #28; \ |
| 4115 | bls 2f; \ |
| 4116 | \ |
| 4117 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4118 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4119 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4120 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4121 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4122 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 4123 | \ |
| 4124 | 3: \ |
| 4125 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4126 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4127 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4128 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4129 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4130 | \ |
| 4131 | subs num_blocks, num_blocks, #1; \ |
| 4132 | bne 0b; \ |
| 4133 | \ |
| 4134 | 1: \ |
| 4135 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4136 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4137 | vbif.u16 blend_pixels, pixels, blend_mask; \ |
| 4138 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4139 | vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \ |
| 4140 | \ |
| 4141 | ldmia sp!, { r4, pc }; \ |
| 4142 | \ |
| 4143 | 2: \ |
| 4144 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 4145 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4146 | \ |
| 4147 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4148 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4149 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4150 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4151 | bal 3b \ |
| 4152 | |
| 4153 | |
| 4154 | |
| 4155 | #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \ |
| 4156 | .align 3; \ |
| 4157 | \ |
| 4158 | function(blend_blocks_untextured_add_fourth_##mask_evaluate) \ |
| 4159 | stmdb sp!, { r4, r14 }; \ |
| 4160 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \ |
| 4161 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 4162 | \ |
| 4163 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \ |
| 4164 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ]; \ |
| 4165 | \ |
| 4166 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \ |
| 4167 | mov c_64, #64; \ |
| 4168 | \ |
| 4169 | vmov.u16 d128_0x7C1F, #0x7C00; \ |
| 4170 | vmov.u16 d128_0x03E0, #0x0300; \ |
| 4171 | vmov.u16 d128_0x1C07, #0x1C00; \ |
| 4172 | vmov.u16 d128_0x00E0, #0x00E0; \ |
| 4173 | vorr.u16 d128_0x7C1F, #0x001F; \ |
| 4174 | vorr.u16 d128_0x03E0, #0x00E0; \ |
| 4175 | vorr.u16 d128_0x1C07, #0x0007; \ |
| 4176 | \ |
| 4177 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 4178 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 4179 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 4180 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4181 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4182 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4183 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4184 | \ |
| 4185 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4186 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4187 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4188 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4189 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4190 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4191 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4192 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4193 | \ |
| 4194 | subs num_blocks, num_blocks, #1; \ |
| 4195 | beq 1f; \ |
| 4196 | \ |
| 4197 | 0: \ |
| 4198 | mov fb_ptr, fb_ptr_next; \ |
| 4199 | ldr fb_ptr_next, [ pixel_ptr, #28 ]; \ |
| 4200 | \ |
| 4201 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64; \ |
| 4202 | \ |
| 4203 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4204 | vshr.s16 pixels_fourth, pixels, #2; \ |
| 4205 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4206 | vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \ |
| 4207 | \ |
| 4208 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4209 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64; \ |
| 4210 | \ |
| 4211 | sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \ |
| 4212 | add fb_ptr_cmp, fb_ptr_cmp, #14; \ |
| 4213 | cmp fb_ptr_cmp, #28; \ |
| 4214 | bls 2f; \ |
| 4215 | \ |
| 4216 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4217 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4218 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4219 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4220 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4221 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 4222 | \ |
| 4223 | 3: \ |
| 4224 | vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \ |
| 4225 | vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \ |
| 4226 | vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \ |
| 4227 | vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \ |
| 4228 | vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \ |
| 4229 | \ |
| 4230 | subs num_blocks, num_blocks, #1; \ |
| 4231 | bne 0b; \ |
| 4232 | \ |
| 4233 | 1: \ |
| 4234 | vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \ |
| 4235 | vorr.u16 blend_pixels, blend_pixels, msb_mask; \ |
| 4236 | vbit.u16 blend_pixels, fb_pixels, draw_mask; \ |
| 4237 | vst1.u16 { blend_pixels }, [ fb_ptr_next ]; \ |
| 4238 | \ |
| 4239 | ldmia sp!, { r4, pc }; \ |
| 4240 | \ |
| 4241 | 2: \ |
| 4242 | vst1.u16 { blend_pixels }, [ fb_ptr ]; \ |
| 4243 | vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \ |
| 4244 | \ |
| 4245 | vld1.u16 { fb_pixels }, [ fb_ptr_next ]; \ |
| 4246 | blend_blocks_add_mask_set_##mask_evaluate(); \ |
| 4247 | blend_blocks_add_mask_copy_##mask_evaluate(); \ |
| 4248 | vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \ |
| 4249 | bal 3b \ |
| 4250 | |
| 4251 | |
| 4252 | blend_blocks_add_fourth_textured_builder(off) |
| 4253 | blend_blocks_add_fourth_textured_builder(on) |
| 4254 | blend_blocks_add_fourth_untextured_builder(off) |
| 4255 | blend_blocks_add_fourth_untextured_builder(on) |
| 4256 | |
| 4257 | // TODO: Optimize this more. Need a scene that actually uses it for |
| 4258 | // confirmation.. |
| 4259 | |
| 4260 | .align 3 |
| 4261 | |
| 4262 | function(blend_blocks_textured_unblended_on) |
| 4263 | stmdb sp!, { r4, r14 } |
| 4264 | add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset |
| 4265 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 4266 | |
| 4267 | add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16) |
| 4268 | vld1.u16 { msb_mask_low[], msb_mask_high[] }, [ mask_msb_ptr, :16 ] |
| 4269 | |
| 4270 | add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 4271 | mov c_64, #64 |
| 4272 | |
| 4273 | ldr fb_ptr, [ pixel_ptr, #28 ] |
| 4274 | vld1.u16 { fb_pixels }, [ fb_ptr ] |
| 4275 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64 |
| 4276 | vclt.s16 write_mask, fb_pixels, #0 |
| 4277 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64 |
| 4278 | |
| 4279 | subs num_blocks, num_blocks, #1 |
| 4280 | beq 1f |
| 4281 | |
| 4282 | 0: |
| 4283 | vorr.u16 pixels, pixels, msb_mask |
| 4284 | vorr.u16 draw_mask, draw_mask, write_mask |
| 4285 | vbif.u16 fb_pixels, pixels, draw_mask |
| 4286 | vst1.u16 { fb_pixels }, [ fb_ptr ] |
| 4287 | |
| 4288 | ldr fb_ptr, [ pixel_ptr, #28 ] |
| 4289 | vld1.u16 { fb_pixels }, [ fb_ptr ] |
| 4290 | vld1.u32 { draw_mask }, [ draw_mask_ptr, :128 ], c_64 |
| 4291 | vclt.s16 write_mask, fb_pixels, #0 |
| 4292 | vld1.u32 { pixels }, [ pixel_ptr, :128 ], c_64 |
| 4293 | |
| 4294 | subs num_blocks, num_blocks, #1 |
| 4295 | bne 0b |
| 4296 | |
| 4297 | 1: |
| 4298 | vorr.u16 pixels, pixels, msb_mask |
| 4299 | vorr.u16 draw_mask, draw_mask, write_mask |
| 4300 | vbif.u16 fb_pixels, pixels, draw_mask |
| 4301 | vst1.u16 { fb_pixels }, [ fb_ptr ] |
| 4302 | |
| 4303 | ldmia sp!, { r4, pc } |
| 4304 | |
| 4305 | |
| 4306 | function(blend_blocks_textured_unblended_off) |
| 4307 | bx lr |
| 4308 | |
| 4309 | |
| 4310 | function(warmup) |
| 4311 | mov r3, #64 |
| 4312 | cmp r0, #0 |
| 4313 | bxeq lr |
| 4314 | |
| 4315 | 0: |
| 4316 | vld1.u32 { u_whole_8, v_whole_8 }, [ r1, :128 ], r3 |
| 4317 | |
| 4318 | subs r0, r0, #1 |
| 4319 | bne 0b |
| 4320 | |
| 4321 | bx lr |
| 4322 | |
| 4323 | #undef vram_ptr |
| 4324 | #undef color |
| 4325 | #undef width |
| 4326 | #undef height |
| 4327 | #undef pitch |
| 4328 | |
| 4329 | #define vram_ptr r0 |
| 4330 | #define color r1 |
| 4331 | #define width r2 |
| 4332 | #define height r3 |
| 4333 | |
| 4334 | #define pitch r1 |
| 4335 | |
| 4336 | #define num_width r12 |
| 4337 | |
| 4338 | #undef colors_a |
| 4339 | #undef colors_b |
| 4340 | |
| 4341 | #define colors_a q0 |
| 4342 | #define colors_b q1 |
| 4343 | |
| 4344 | .align 3 |
| 4345 | |
| 4346 | function(render_block_fill_body) |
| 4347 | vdup.u16 colors_a, color |
| 4348 | mov pitch, #2048 |
| 4349 | |
| 4350 | vmov colors_b, colors_a |
| 4351 | sub pitch, pitch, width, lsl #1 |
| 4352 | |
| 4353 | mov num_width, width |
| 4354 | |
| 4355 | 0: |
| 4356 | vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]! |
| 4357 | |
| 4358 | subs num_width, num_width, #16 |
| 4359 | bne 0b |
| 4360 | |
| 4361 | add vram_ptr, vram_ptr, pitch |
| 4362 | mov num_width, width |
| 4363 | |
| 4364 | subs height, height, #1 |
| 4365 | bne 0b |
| 4366 | |
| 4367 | bx lr |
| 4368 | |
| 4369 | |
| 4370 | #undef x |
| 4371 | #undef y |
| 4372 | #undef width |
| 4373 | #undef height |
| 4374 | #undef fb_ptr |
| 4375 | #undef texture_mask |
| 4376 | #undef num_blocks |
| 4377 | #undef temp |
| 4378 | #undef dirty_textures_mask |
| 4379 | #undef clut_ptr |
| 4380 | #undef current_texture_mask |
| 4381 | |
| 4382 | #define psx_gpu r0 |
| 4383 | #define x r1 |
| 4384 | #define y r2 |
| 4385 | #define u r3 |
| 4386 | #define v r4 |
| 4387 | #define width r5 |
| 4388 | #define height r6 |
| 4389 | #define offset_u r8 |
| 4390 | #define offset_v r9 |
| 4391 | #define offset_u_right r10 |
| 4392 | #define width_rounded r11 |
| 4393 | #define height_rounded r12 |
| 4394 | |
| 4395 | #define texture_offset_base r1 |
| 4396 | #define tile_width r2 |
| 4397 | #define tile_height r3 |
| 4398 | #define num_blocks r4 |
| 4399 | #define block r5 |
| 4400 | #define sub_tile_height r6 |
| 4401 | #define fb_ptr r7 |
| 4402 | #define texture_mask r8 |
| 4403 | #define column_data r9 |
| 4404 | #define texture_offset r10 |
| 4405 | #define tiles_remaining r11 |
| 4406 | #define fb_ptr_advance_column r12 |
| 4407 | #define texture_block_ptr r14 |
| 4408 | |
| 4409 | #define texture_page_ptr r3 |
| 4410 | #define left_block_mask r4 |
| 4411 | #define right_block_mask r5 |
| 4412 | #define texture_mask_rev r10 |
| 4413 | #define control_mask r11 |
| 4414 | |
| 4415 | #define dirty_textures_mask r4 |
| 4416 | #define clut_ptr r5 |
| 4417 | #define current_texture_mask r6 |
| 4418 | |
| 4419 | |
| 4420 | #undef texels |
| 4421 | #undef clut_low_a |
| 4422 | #undef clut_low_b |
| 4423 | #undef clut_high_a |
| 4424 | #undef clut_high_b |
| 4425 | #undef clut_a |
| 4426 | #undef clut_b |
| 4427 | #undef texels_low |
| 4428 | #undef texels_high |
| 4429 | |
| 4430 | #define texels d0 |
| 4431 | #define draw_masks_fb_ptrs q1 |
| 4432 | |
| 4433 | #define draw_mask_fb_ptr_left d2 |
| 4434 | #define draw_mask_fb_ptr_right d3 |
| 4435 | |
| 4436 | #define draw_mask_fb_ptr_left_a d2 |
| 4437 | #define draw_mask_fb_ptr_left_b d3 |
| 4438 | #define draw_mask_fb_ptr_right_a d10 |
| 4439 | #define draw_mask_fb_ptr_right_b d11 |
| 4440 | #define draw_masks_fb_ptrs2 q5 |
| 4441 | |
| 4442 | #define clut_low_a d4 |
| 4443 | #define clut_low_b d5 |
| 4444 | #define clut_high_a d6 |
| 4445 | #define clut_high_b d7 |
| 4446 | |
| 4447 | #define block_masks d8 |
| 4448 | #define block_masks_shifted d9 |
| 4449 | |
| 4450 | #define clut_a q2 |
| 4451 | #define clut_b q3 |
| 4452 | |
| 4453 | #define texels_low d12 |
| 4454 | #define texels_high d13 |
| 4455 | |
| 4456 | #define texels_wide_low d14 |
| 4457 | #define texels_wide_high d15 |
| 4458 | #define texels_wide q7 |
| 4459 | |
| 4460 | |
| 4461 | setup_sprite_flush_blocks: |
| 4462 | vpush { q1 - q5 } |
| 4463 | |
| 4464 | stmdb sp!, { r0 - r3, r12, r14 } |
| 4465 | bl flush_render_block_buffer |
| 4466 | ldmia sp!, { r0 - r3, r12, r14 } |
| 4467 | |
| 4468 | vpop { q1 - q5 } |
| 4469 | |
| 4470 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 4471 | bx lr |
| 4472 | |
| 4473 | |
| 4474 | setup_sprite_update_texture_4bpp_cache: |
| 4475 | stmdb sp!, { r0 - r3, r14 } |
| 4476 | bl update_texture_4bpp_cache |
| 4477 | ldmia sp!, { r0 - r3, pc } |
| 4478 | |
| 4479 | |
| 4480 | setup_sprite_update_texture_8bpp_cache: |
| 4481 | stmdb sp!, { r0 - r3, r14 } |
| 4482 | bl update_texture_8bpp_cache |
| 4483 | ldmia sp!, { r0 - r3, pc } |
| 4484 | |
| 4485 | |
| 4486 | #define setup_sprite_tiled_initialize_4bpp() \ |
| 4487 | ldr dirty_textures_mask, \ |
| 4488 | [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ]; \ |
| 4489 | ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \ |
| 4490 | \ |
| 4491 | ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \ |
| 4492 | vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \ |
| 4493 | \ |
| 4494 | tst current_texture_mask, dirty_textures_mask; \ |
| 4495 | vuzp.u8 clut_a, clut_b; \ |
| 4496 | \ |
| 4497 | blne setup_sprite_update_texture_4bpp_cache \ |
| 4498 | |
| 4499 | #define setup_sprite_tiled_initialize_8bpp() \ |
| 4500 | ldr dirty_textures_mask, \ |
| 4501 | [ psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset ]; \ |
| 4502 | ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ]; \ |
| 4503 | \ |
| 4504 | tst current_texture_mask, dirty_textures_mask; \ |
| 4505 | blne setup_sprite_update_texture_8bpp_cache \ |
| 4506 | |
| 4507 | |
| 4508 | #define setup_sprite_block_count_single() \ |
| 4509 | sub_tile_height \ |
| 4510 | |
| 4511 | #define setup_sprite_block_count_double() \ |
| 4512 | sub_tile_height, lsl #1 \ |
| 4513 | |
| 4514 | #define setup_sprite_tile_add_blocks(type) \ |
| 4515 | add num_blocks, num_blocks, setup_sprite_block_count_##type(); \ |
| 4516 | cmp num_blocks, #MAX_BLOCKS; \ |
| 4517 | \ |
| 4518 | movgt num_blocks, setup_sprite_block_count_##type(); \ |
| 4519 | blgt setup_sprite_flush_blocks \ |
| 4520 | |
| 4521 | |
| 4522 | #define setup_sprite_tile_full_4bpp(edge) \ |
| 4523 | setup_sprite_tile_add_blocks(double); \ |
| 4524 | \ |
| 4525 | 4: \ |
| 4526 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4527 | vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \ |
| 4528 | \ |
| 4529 | pld [ fb_ptr ]; \ |
| 4530 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4531 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4532 | \ |
| 4533 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4534 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4535 | \ |
| 4536 | vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \ |
| 4537 | add texture_block_ptr, texture_offset, #8; \ |
| 4538 | \ |
| 4539 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 4540 | add block, block, #40; \ |
| 4541 | \ |
| 4542 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4543 | add fb_ptr, fb_ptr, #16; \ |
| 4544 | \ |
| 4545 | vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \ |
| 4546 | add block, block, #24; \ |
| 4547 | \ |
| 4548 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4549 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4550 | \ |
| 4551 | pld [ fb_ptr ]; \ |
| 4552 | vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \ |
| 4553 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4554 | \ |
| 4555 | vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \ |
| 4556 | add block, block, #40; \ |
| 4557 | \ |
| 4558 | add texture_offset, texture_offset, #0x10; \ |
| 4559 | add fb_ptr, fb_ptr, #(2048 - 16); \ |
| 4560 | \ |
| 4561 | vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \ |
| 4562 | add block, block, #24; \ |
| 4563 | \ |
| 4564 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4565 | bne 4b; \ |
| 4566 | \ |
| 4567 | add texture_offset, texture_offset, #0xF00; \ |
| 4568 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 4569 | |
| 4570 | |
| 4571 | #define setup_sprite_tile_half_4bpp(edge) \ |
| 4572 | setup_sprite_tile_add_blocks(single); \ |
| 4573 | \ |
| 4574 | 4: \ |
| 4575 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4576 | vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \ |
| 4577 | \ |
| 4578 | pld [ fb_ptr ]; \ |
| 4579 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4580 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4581 | \ |
| 4582 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4583 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4584 | \ |
| 4585 | vst2.u8 { texels_low, texels_high }, [ block, :128 ]; \ |
| 4586 | add block, block, #40; \ |
| 4587 | \ |
| 4588 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4589 | vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \ |
| 4590 | \ |
| 4591 | add block, block, #24; \ |
| 4592 | add texture_offset, texture_offset, #0x10; \ |
| 4593 | \ |
| 4594 | add fb_ptr, fb_ptr, #2048; \ |
| 4595 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4596 | \ |
| 4597 | bne 4b; \ |
| 4598 | \ |
| 4599 | add texture_offset, texture_offset, #0xF00; \ |
| 4600 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 4601 | |
| 4602 | |
| 4603 | #define setup_sprite_tile_full_8bpp(edge) \ |
| 4604 | setup_sprite_tile_add_blocks(double); \ |
| 4605 | add block, block, #16; \ |
| 4606 | \ |
| 4607 | 4: \ |
| 4608 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4609 | vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \ |
| 4610 | \ |
| 4611 | pld [ fb_ptr ]; \ |
| 4612 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4613 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4614 | \ |
| 4615 | add texture_block_ptr, texture_offset, #8; \ |
| 4616 | vst1.u32 { texels }, [ block, :64 ]; \ |
| 4617 | \ |
| 4618 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 4619 | add block, block, #24; \ |
| 4620 | \ |
| 4621 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4622 | \ |
| 4623 | add fb_ptr, fb_ptr, #16; \ |
| 4624 | vst1.u32 { draw_mask_fb_ptr_left }, [ block, :64 ]; \ |
| 4625 | \ |
| 4626 | add block, block, #40; \ |
| 4627 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4628 | pld [ fb_ptr ]; \ |
| 4629 | \ |
| 4630 | vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \ |
| 4631 | vst1.u32 { texels }, [ block, :64 ]; \ |
| 4632 | add block, block, #24; \ |
| 4633 | \ |
| 4634 | add texture_offset, texture_offset, #0x10; \ |
| 4635 | add fb_ptr, fb_ptr, #(2048 - 16); \ |
| 4636 | \ |
| 4637 | vst1.u32 { draw_mask_fb_ptr_right }, [ block, :64 ]; \ |
| 4638 | add block, block, #40; \ |
| 4639 | \ |
| 4640 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4641 | bne 4b; \ |
| 4642 | \ |
| 4643 | sub block, block, #16; \ |
| 4644 | add texture_offset, texture_offset, #0xF00; \ |
| 4645 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 4646 | |
| 4647 | |
| 4648 | #define setup_sprite_tile_half_8bpp(edge) \ |
| 4649 | setup_sprite_tile_add_blocks(single); \ |
| 4650 | add block, block, #16; \ |
| 4651 | \ |
| 4652 | 4: \ |
| 4653 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4654 | vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \ |
| 4655 | pld [ fb_ptr ]; \ |
| 4656 | \ |
| 4657 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4658 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4659 | \ |
| 4660 | vst1.u32 { texels }, [ block, :64 ]; \ |
| 4661 | add block, block, #24; \ |
| 4662 | \ |
| 4663 | vst1.u32 { draw_mask_fb_ptr_##edge }, [ block, :64 ]; \ |
| 4664 | add block, block, #40; \ |
| 4665 | \ |
| 4666 | add texture_offset, texture_offset, #0x10; \ |
| 4667 | add fb_ptr, fb_ptr, #2048; \ |
| 4668 | \ |
| 4669 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4670 | bne 4b; \ |
| 4671 | \ |
| 4672 | sub block, block, #16; \ |
| 4673 | add texture_offset, texture_offset, #0xF00; \ |
| 4674 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 4675 | |
| 4676 | |
| 4677 | #define setup_sprite_tile_column_edge_pre_adjust_half_right() \ |
| 4678 | add texture_offset, texture_offset_base, #8; \ |
| 4679 | add fb_ptr, fb_ptr, #16 \ |
| 4680 | |
| 4681 | #define setup_sprite_tile_column_edge_pre_adjust_half_left() \ |
| 4682 | mov texture_offset, texture_offset_base \ |
| 4683 | |
| 4684 | #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \ |
| 4685 | setup_sprite_tile_column_edge_pre_adjust_half_##edge() \ |
| 4686 | |
| 4687 | #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \ |
| 4688 | mov texture_offset, texture_offset_base \ |
| 4689 | |
| 4690 | #define setup_sprite_tile_column_edge_post_adjust_half_right() \ |
| 4691 | sub fb_ptr, fb_ptr, #16 \ |
| 4692 | |
| 4693 | #define setup_sprite_tile_column_edge_post_adjust_half_left() \ |
| 4694 | |
| 4695 | #define setup_sprite_tile_column_edge_post_adjust_half(edge) \ |
| 4696 | setup_sprite_tile_column_edge_post_adjust_half_##edge() \ |
| 4697 | |
| 4698 | #define setup_sprite_tile_column_edge_post_adjust_full(edge) \ |
| 4699 | |
| 4700 | |
| 4701 | #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \ |
| 4702 | x4mode) \ |
| 4703 | mov sub_tile_height, column_data; \ |
| 4704 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 4705 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4706 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \ |
| 4707 | |
| 4708 | #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \ |
| 4709 | x4mode) \ |
| 4710 | and sub_tile_height, column_data, #0xFF; \ |
| 4711 | mov tiles_remaining, column_data, lsr #16; \ |
| 4712 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 4713 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4714 | \ |
| 4715 | subs tiles_remaining, tiles_remaining, #1; \ |
| 4716 | beq 2f; \ |
| 4717 | \ |
| 4718 | 3: \ |
| 4719 | mov sub_tile_height, #16; \ |
| 4720 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4721 | subs tiles_remaining, tiles_remaining, #1; \ |
| 4722 | bne 3b; \ |
| 4723 | \ |
| 4724 | 2: \ |
| 4725 | uxtb sub_tile_height, column_data, ror #8; \ |
| 4726 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 4727 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \ |
| 4728 | |
| 4729 | |
| 4730 | #define setup_sprite_column_data_single() \ |
| 4731 | mov column_data, height; \ |
| 4732 | ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] \ |
| 4733 | |
| 4734 | #define setup_sprite_column_data_multi() \ |
| 4735 | and height_rounded, height_rounded, #0xF; \ |
| 4736 | rsb column_data, offset_v, #16; \ |
| 4737 | \ |
| 4738 | add height_rounded, height_rounded, #1; \ |
| 4739 | sub tile_height, tile_height, #1; \ |
| 4740 | \ |
| 4741 | orr column_data, column_data, tile_height, lsl #16; \ |
| 4742 | ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ]; \ |
| 4743 | \ |
| 4744 | orr column_data, column_data, height_rounded, lsl #8 \ |
| 4745 | |
| 4746 | #define setup_sprite_setup_left_draw_mask_fb_ptr() \ |
| 4747 | vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \ |
| 4748 | vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \ |
| 4749 | |
| 4750 | #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \ |
| 4751 | mov fb_ptr_advance_column, #32; \ |
| 4752 | vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \ |
| 4753 | \ |
| 4754 | sub fb_ptr_advance_column, height, lsl #11; \ |
| 4755 | vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \ |
| 4756 | |
| 4757 | #define setup_sprite_setup_right_draw_mask_fb_ptr() \ |
| 4758 | vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \ |
| 4759 | vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \ |
| 4760 | |
| 4761 | #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \ |
| 4762 | edge, x4mode) \ |
| 4763 | setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \ |
| 4764 | setup_sprite_column_data_##multi_height(); \ |
| 4765 | vext.32 block_masks_shifted, block_masks, block_masks, #1; \ |
| 4766 | vorr.u32 block_masks, block_masks, block_masks_shifted; \ |
| 4767 | setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \ |
| 4768 | \ |
| 4769 | setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \ |
| 4770 | ldmia sp!, { r4 - r11, pc } \ |
| 4771 | |
| 4772 | #define setup_sprite_tiled_advance_column() \ |
| 4773 | add texture_offset_base, texture_offset_base, #0x100; \ |
| 4774 | tst texture_offset_base, #0xF00; \ |
| 4775 | subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \ |
| 4776 | |
| 4777 | #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \ |
| 4778 | right_mode, x4mode) \ |
| 4779 | setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\ |
| 4780 | setup_sprite_column_data_##multi_height(); \ |
| 4781 | \ |
| 4782 | setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \ |
| 4783 | \ |
| 4784 | setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\ |
| 4785 | \ |
| 4786 | subs tile_width, tile_width, #2; \ |
| 4787 | add fb_ptr, fb_ptr, fb_ptr_advance_column; \ |
| 4788 | \ |
| 4789 | beq 1f; \ |
| 4790 | \ |
| 4791 | vmov.u8 draw_masks_fb_ptrs, #0; \ |
| 4792 | vmov.u8 draw_masks_fb_ptrs2, #0; \ |
| 4793 | \ |
| 4794 | 0: \ |
| 4795 | setup_sprite_tiled_advance_column(); \ |
| 4796 | setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \ |
| 4797 | add fb_ptr, fb_ptr, fb_ptr_advance_column; \ |
| 4798 | subs tile_width, tile_width, #1; \ |
| 4799 | bne 0b; \ |
| 4800 | \ |
| 4801 | 1: \ |
| 4802 | setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \ |
| 4803 | \ |
| 4804 | setup_sprite_tiled_advance_column(); \ |
| 4805 | setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\ |
| 4806 | ldmia sp!, { r4 - r11, pc } \ |
| 4807 | |
| 4808 | |
| 4809 | #define setup_sprite_offset_u_adjust() \ |
| 4810 | |
| 4811 | #define setup_sprite_get_left_block_mask() \ |
| 4812 | and left_block_mask, left_block_mask, #0xFF \ |
| 4813 | |
| 4814 | #define setup_sprite_compare_left_block_mask() \ |
| 4815 | cmp left_block_mask, #0xFF \ |
| 4816 | |
| 4817 | #define setup_sprite_get_right_block_mask() \ |
| 4818 | uxtb right_block_mask, right_block_mask, ror #8 \ |
| 4819 | |
| 4820 | #define setup_sprite_compare_right_block_mask() \ |
| 4821 | cmp right_block_mask, #0xFF \ |
| 4822 | |
| 4823 | |
| 4824 | |
| 4825 | /* 4x stuff */ |
| 4826 | #define fb_ptr2 column_data |
| 4827 | |
| 4828 | #define setup_sprite_offset_u_adjust_4x() \ |
| 4829 | sub fb_ptr, fb_ptr, offset_u, lsl #1; \ |
| 4830 | lsl offset_u_right, #1; \ |
| 4831 | lsl offset_u, #1; \ |
| 4832 | add offset_u_right, #1 \ |
| 4833 | |
| 4834 | #define setup_sprite_get_left_block_mask_4x() \ |
| 4835 | sxth left_block_mask, left_block_mask \ |
| 4836 | |
| 4837 | #define setup_sprite_compare_left_block_mask_4x() \ |
| 4838 | cmp left_block_mask, #0xFFFFFFFF \ |
| 4839 | |
| 4840 | #define setup_sprite_get_right_block_mask_4x() \ |
| 4841 | sxth right_block_mask, right_block_mask, ror #16 \ |
| 4842 | |
| 4843 | #define setup_sprite_compare_right_block_mask_4x() \ |
| 4844 | cmp right_block_mask, #0xFFFFFFFF \ |
| 4845 | |
| 4846 | |
| 4847 | #define widen_texels_16bpp(texels_) \ |
| 4848 | vmov texels_wide_low, texels_; \ |
| 4849 | vmov texels_wide_high, texels_; \ |
| 4850 | vzip.16 texels_wide_low, texels_wide_high \ |
| 4851 | |
| 4852 | #define widen_texels_8bpp(texels_) \ |
| 4853 | vmov texels_wide_low, texels_; \ |
| 4854 | vmov texels_wide_high, texels_; \ |
| 4855 | vzip.8 texels_wide_low, texels_wide_high \ |
| 4856 | |
| 4857 | #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \ |
| 4858 | vst1.u32 { texels_ }, [ block_, :128 ]; \ |
| 4859 | add block_, block_, #40; \ |
| 4860 | \ |
| 4861 | vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \ |
| 4862 | vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \ |
| 4863 | add block_, block_, #24 \ |
| 4864 | |
| 4865 | /* assumes 16-byte offset already added to block_ */ |
| 4866 | #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \ |
| 4867 | vst1.u32 { texels_ }, [ block_, :64 ]; \ |
| 4868 | add block_, block_, #24; \ |
| 4869 | \ |
| 4870 | vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \ |
| 4871 | vst1.u32 { draw_mask_fb_ptr_ }, [ block_, :64 ]; \ |
| 4872 | add block_, block_, #40 \ |
| 4873 | |
| 4874 | #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \ |
| 4875 | draw_mask_fb_ptr_b_) \ |
| 4876 | widen_texels_16bpp(texels_low); \ |
| 4877 | add fb_ptr_tmp, fb_ptr, #1024*2; \ |
| 4878 | \ |
| 4879 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \ |
| 4880 | \ |
| 4881 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \ |
| 4882 | widen_texels_16bpp(texels_high); \ |
| 4883 | \ |
| 4884 | add fb_ptr_tmp, fb_ptr, #8*2; \ |
| 4885 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \ |
| 4886 | \ |
| 4887 | add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \ |
| 4888 | write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \ |
| 4889 | |
| 4890 | #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \ |
| 4891 | draw_mask_fb_ptr_b_) \ |
| 4892 | widen_texels_8bpp(texels); \ |
| 4893 | add fb_ptr_tmp, fb_ptr, #1024*2; \ |
| 4894 | \ |
| 4895 | write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \ |
| 4896 | write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \ |
| 4897 | \ |
| 4898 | add fb_ptr_tmp, fb_ptr, #8*2; \ |
| 4899 | write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \ |
| 4900 | \ |
| 4901 | add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \ |
| 4902 | write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \ |
| 4903 | |
| 4904 | |
| 4905 | #define setup_sprite_tiled_initialize_4bpp_4x() \ |
| 4906 | ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ]; \ |
| 4907 | vld1.u32 { clut_a, clut_b }, [ clut_ptr, :128 ]; \ |
| 4908 | \ |
| 4909 | vuzp.u8 clut_a, clut_b \ |
| 4910 | |
| 4911 | #define setup_sprite_tiled_initialize_8bpp_4x() \ |
| 4912 | |
| 4913 | |
| 4914 | #define setup_sprite_block_count_single_4x() \ |
| 4915 | sub_tile_height, lsl #2 \ |
| 4916 | |
| 4917 | #define setup_sprite_block_count_double_4x() \ |
| 4918 | sub_tile_height, lsl #(1+2) \ |
| 4919 | |
| 4920 | #define setup_sprite_tile_full_4bpp_4x(edge) \ |
| 4921 | setup_sprite_tile_add_blocks(double_4x); \ |
| 4922 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 4923 | \ |
| 4924 | 4: \ |
| 4925 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4926 | pld [ fb_ptr ]; \ |
| 4927 | \ |
| 4928 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4929 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4930 | \ |
| 4931 | add texture_block_ptr, texture_offset, #8; \ |
| 4932 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4933 | \ |
| 4934 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 4935 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4936 | \ |
| 4937 | vzip.8 texels_low, texels_high; \ |
| 4938 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \ |
| 4939 | draw_mask_fb_ptr_left_b); \ |
| 4940 | \ |
| 4941 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4942 | pld [ fb_ptr, #2048 ]; \ |
| 4943 | \ |
| 4944 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4945 | add fb_ptr, fb_ptr, #16*2; \ |
| 4946 | \ |
| 4947 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4948 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4949 | \ |
| 4950 | vzip.8 texels_low, texels_high; \ |
| 4951 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \ |
| 4952 | draw_mask_fb_ptr_right_b); \ |
| 4953 | \ |
| 4954 | add texture_offset, texture_offset, #0x10; \ |
| 4955 | add fb_ptr, fb_ptr, #(2048 - 16) * 2; \ |
| 4956 | \ |
| 4957 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4958 | bne 4b; \ |
| 4959 | \ |
| 4960 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 4961 | add texture_offset, texture_offset, #0xF00; \ |
| 4962 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 4963 | |
| 4964 | |
| 4965 | #define setup_sprite_tile_half_4bpp_4x(edge) \ |
| 4966 | setup_sprite_tile_add_blocks(single_4x); \ |
| 4967 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 4968 | \ |
| 4969 | 4: \ |
| 4970 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 4971 | pld [ fb_ptr ]; \ |
| 4972 | \ |
| 4973 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4974 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 4975 | \ |
| 4976 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 4977 | vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \ |
| 4978 | \ |
| 4979 | vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \ |
| 4980 | add texture_offset, texture_offset, #0x10; \ |
| 4981 | \ |
| 4982 | vzip.8 texels_low, texels_high; \ |
| 4983 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \ |
| 4984 | draw_mask_fb_ptr_##edge##_b); \ |
| 4985 | \ |
| 4986 | pld [ fb_ptr, #2048 ]; \ |
| 4987 | add fb_ptr, fb_ptr, #2048 * 2; \ |
| 4988 | \ |
| 4989 | subs sub_tile_height, sub_tile_height, #1; \ |
| 4990 | bne 4b; \ |
| 4991 | \ |
| 4992 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 4993 | add texture_offset, texture_offset, #0xF00; \ |
| 4994 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 4995 | |
| 4996 | |
| 4997 | #define setup_sprite_tile_full_8bpp_4x(edge) \ |
| 4998 | setup_sprite_tile_add_blocks(double_4x); \ |
| 4999 | add block, block, #16; \ |
| 5000 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 5001 | \ |
| 5002 | 4: \ |
| 5003 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 5004 | pld [ fb_ptr ]; \ |
| 5005 | \ |
| 5006 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 5007 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 5008 | \ |
| 5009 | add texture_block_ptr, texture_offset, #8; \ |
| 5010 | do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \ |
| 5011 | draw_mask_fb_ptr_left_b); \ |
| 5012 | \ |
| 5013 | pld [ fb_ptr, #2048 ]; \ |
| 5014 | and texture_block_ptr, texture_block_ptr, texture_mask; \ |
| 5015 | \ |
| 5016 | add fb_ptr, fb_ptr, #16*2; \ |
| 5017 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 5018 | \ |
| 5019 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 5020 | \ |
| 5021 | do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \ |
| 5022 | draw_mask_fb_ptr_right_b); \ |
| 5023 | \ |
| 5024 | add texture_offset, texture_offset, #0x10; \ |
| 5025 | add fb_ptr, fb_ptr, #(2048 - 16) * 2; \ |
| 5026 | \ |
| 5027 | subs sub_tile_height, sub_tile_height, #1; \ |
| 5028 | bne 4b; \ |
| 5029 | \ |
| 5030 | sub block, block, #16; \ |
| 5031 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 5032 | add texture_offset, texture_offset, #0xF00; \ |
| 5033 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 5034 | |
| 5035 | |
| 5036 | #define setup_sprite_tile_half_8bpp_4x(edge) \ |
| 5037 | setup_sprite_tile_add_blocks(single_4x); \ |
| 5038 | add block, block, #16; \ |
| 5039 | str column_data, [sp, #-8]!; /* fb_ptr2 */ \ |
| 5040 | \ |
| 5041 | 4: \ |
| 5042 | and texture_block_ptr, texture_offset, texture_mask; \ |
| 5043 | pld [ fb_ptr ]; \ |
| 5044 | \ |
| 5045 | add texture_block_ptr, texture_page_ptr, texture_block_ptr; \ |
| 5046 | vld1.u32 { texels }, [ texture_block_ptr, :64 ]; \ |
| 5047 | \ |
| 5048 | pld [ fb_ptr, #2048 ]; \ |
| 5049 | do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \ |
| 5050 | draw_mask_fb_ptr_##edge##_b); \ |
| 5051 | \ |
| 5052 | add texture_offset, texture_offset, #0x10; \ |
| 5053 | add fb_ptr, fb_ptr, #2048 * 2; \ |
| 5054 | \ |
| 5055 | subs sub_tile_height, sub_tile_height, #1; \ |
| 5056 | bne 4b; \ |
| 5057 | \ |
| 5058 | sub block, block, #16; \ |
| 5059 | ldr column_data, [sp], #8; /* fb_ptr2 */ \ |
| 5060 | add texture_offset, texture_offset, #0xF00; \ |
| 5061 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] \ |
| 5062 | |
| 5063 | |
| 5064 | #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \ |
| 5065 | add texture_offset, texture_offset_base, #8; \ |
| 5066 | add fb_ptr, fb_ptr, #16 * 2 \ |
| 5067 | |
| 5068 | #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \ |
| 5069 | mov texture_offset, texture_offset_base \ |
| 5070 | |
| 5071 | #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \ |
| 5072 | setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \ |
| 5073 | |
| 5074 | #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \ |
| 5075 | mov texture_offset, texture_offset_base \ |
| 5076 | |
| 5077 | #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \ |
| 5078 | sub fb_ptr, fb_ptr, #16 * 2 \ |
| 5079 | |
| 5080 | #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \ |
| 5081 | |
| 5082 | #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \ |
| 5083 | setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \ |
| 5084 | |
| 5085 | #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \ |
| 5086 | |
| 5087 | |
| 5088 | #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \ |
| 5089 | vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \ |
| 5090 | vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \ |
| 5091 | vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \ |
| 5092 | vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \ |
| 5093 | |
| 5094 | #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \ |
| 5095 | mov fb_ptr_advance_column, #32 * 2; \ |
| 5096 | vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \ |
| 5097 | vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \ |
| 5098 | sub fb_ptr_advance_column, height, lsl #11 + 1; \ |
| 5099 | vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \ |
| 5100 | vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \ |
| 5101 | |
| 5102 | #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \ |
| 5103 | vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \ |
| 5104 | vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \ |
| 5105 | vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \ |
| 5106 | vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \ |
| 5107 | |
| 5108 | |
| 5109 | // r0: psx_gpu |
| 5110 | // r1: x |
| 5111 | // r2: y |
| 5112 | // r3: u |
| 5113 | // [ sp ]: v |
| 5114 | // [ sp + 4 ]: width |
| 5115 | // [ sp + 8 ]: height |
| 5116 | // [ sp + 12 ]: color (unused) |
| 5117 | |
| 5118 | #define setup_sprite_tiled_builder(texture_mode, x4mode) \ |
| 5119 | \ |
| 5120 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \ |
| 5121 | x4mode); \ |
| 5122 | setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \ |
| 5123 | x4mode); \ |
| 5124 | setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \ |
| 5125 | x4mode); \ |
| 5126 | setup_sprite_tile_column_width_single(texture_mode, single, full, none, \ |
| 5127 | x4mode); \ |
| 5128 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \ |
| 5129 | x4mode); \ |
| 5130 | setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \ |
| 5131 | x4mode); \ |
| 5132 | setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \ |
| 5133 | x4mode); \ |
| 5134 | setup_sprite_tile_column_width_single(texture_mode, single, half, right, \ |
| 5135 | x4mode); \ |
| 5136 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \ |
| 5137 | x4mode); \ |
| 5138 | setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \ |
| 5139 | x4mode); \ |
| 5140 | setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \ |
| 5141 | x4mode); \ |
| 5142 | setup_sprite_tile_column_width_single(texture_mode, single, half, left, \ |
| 5143 | x4mode); \ |
| 5144 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \ |
| 5145 | x4mode); \ |
| 5146 | setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \ |
| 5147 | x4mode); \ |
| 5148 | \ |
| 5149 | .align 4; \ |
| 5150 | \ |
| 5151 | function(setup_sprite_##texture_mode##x4mode) \ |
| 5152 | stmdb sp!, { r4 - r11, r14 }; \ |
| 5153 | setup_sprite_tiled_initialize_##texture_mode##x4mode(); \ |
| 5154 | \ |
| 5155 | ldr v, [ sp, #36 ]; \ |
| 5156 | and offset_u, u, #0xF; \ |
| 5157 | \ |
| 5158 | ldr width, [ sp, #40 ]; \ |
| 5159 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ]; \ |
| 5160 | \ |
| 5161 | ldr height, [ sp, #44 ]; \ |
| 5162 | add fb_ptr, fb_ptr, y, lsl #11; \ |
| 5163 | \ |
| 5164 | add fb_ptr, fb_ptr, x, lsl #1; \ |
| 5165 | and offset_v, v, #0xF; \ |
| 5166 | \ |
| 5167 | sub fb_ptr, fb_ptr, offset_u, lsl #1; \ |
| 5168 | add width_rounded, offset_u, width; \ |
| 5169 | \ |
| 5170 | add height_rounded, offset_v, height; \ |
| 5171 | add width_rounded, width_rounded, #15; \ |
| 5172 | \ |
| 5173 | add height_rounded, height_rounded, #15; \ |
| 5174 | mov tile_width, width_rounded, lsr #4; \ |
| 5175 | \ |
| 5176 | /* texture_offset_base = VH-VL-00-00 */\ |
| 5177 | mov texture_offset_base, v, lsl #8; \ |
| 5178 | and offset_u_right, width_rounded, #0xF; \ |
| 5179 | \ |
| 5180 | /* texture_offset_base = VH-UH-UL-00 */\ |
| 5181 | bfi texture_offset_base, u, #4, #8; \ |
| 5182 | mov right_block_mask, #0xFFFFFFFE; \ |
| 5183 | \ |
| 5184 | setup_sprite_offset_u_adjust##x4mode(); \ |
| 5185 | \ |
| 5186 | /* texture_offset_base = VH-UH-VL-00 */\ |
| 5187 | bfi texture_offset_base, v, #4, #4; \ |
| 5188 | mov left_block_mask, #0xFFFFFFFF; \ |
| 5189 | \ |
| 5190 | mov tile_height, height_rounded, lsr #4; \ |
| 5191 | mvn left_block_mask, left_block_mask, lsl offset_u; \ |
| 5192 | \ |
| 5193 | /* texture_mask = HH-HL-WH-WL */\ |
| 5194 | ldrh texture_mask, [ psx_gpu, #psx_gpu_texture_mask_width_offset ]; \ |
| 5195 | mov right_block_mask, right_block_mask, lsl offset_u_right; \ |
| 5196 | \ |
| 5197 | /* texture_mask_rev = WH-WL-HH-HL */\ |
| 5198 | rev16 texture_mask_rev, texture_mask; \ |
| 5199 | vmov block_masks, left_block_mask, right_block_mask; \ |
| 5200 | \ |
| 5201 | /* texture_mask = HH-HL-HL-WL */\ |
| 5202 | bfi texture_mask, texture_mask_rev, #4, #4; \ |
| 5203 | /* texture_mask_rev = 00-00-00-WH */\ |
| 5204 | mov texture_mask_rev, texture_mask_rev, lsr #12; \ |
| 5205 | \ |
| 5206 | /* texture_mask = HH-WH-HL-WL */\ |
| 5207 | bfi texture_mask, texture_mask_rev, #8, #4; \ |
| 5208 | setup_sprite_get_left_block_mask##x4mode(); \ |
| 5209 | \ |
| 5210 | mov control_mask, #0; \ |
| 5211 | setup_sprite_compare_left_block_mask##x4mode(); \ |
| 5212 | \ |
| 5213 | setup_sprite_get_right_block_mask##x4mode(); \ |
| 5214 | orreq control_mask, control_mask, #0x4; \ |
| 5215 | \ |
| 5216 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]; \ |
| 5217 | setup_sprite_compare_right_block_mask##x4mode(); \ |
| 5218 | \ |
| 5219 | orreq control_mask, control_mask, #0x8; \ |
| 5220 | cmp tile_width, #1; \ |
| 5221 | \ |
| 5222 | add block, psx_gpu, #psx_gpu_blocks_offset; \ |
| 5223 | orreq control_mask, control_mask, #0x1; \ |
| 5224 | \ |
| 5225 | cmp tile_height, #1; \ |
| 5226 | add block, block, num_blocks, lsl #6; \ |
| 5227 | \ |
| 5228 | orreq control_mask, control_mask, #0x2; \ |
| 5229 | ldr pc, [ pc, control_mask, lsl #2 ]; \ |
| 5230 | nop; \ |
| 5231 | \ |
| 5232 | .word setup_sprite_##texture_mode##_multi_multi_full_full##x4mode; \ |
| 5233 | .word setup_sprite_##texture_mode##_single_multi_full_none##x4mode; \ |
| 5234 | .word setup_sprite_##texture_mode##_multi_single_full_full##x4mode; \ |
| 5235 | .word setup_sprite_##texture_mode##_single_single_full_none##x4mode; \ |
| 5236 | .word setup_sprite_##texture_mode##_multi_multi_half_full##x4mode; \ |
| 5237 | .word setup_sprite_##texture_mode##_single_multi_half_right##x4mode; \ |
| 5238 | .word setup_sprite_##texture_mode##_multi_single_half_full##x4mode; \ |
| 5239 | .word setup_sprite_##texture_mode##_single_single_half_right##x4mode; \ |
| 5240 | .word setup_sprite_##texture_mode##_multi_multi_full_half##x4mode; \ |
| 5241 | .word setup_sprite_##texture_mode##_single_multi_half_left##x4mode; \ |
| 5242 | .word setup_sprite_##texture_mode##_multi_single_full_half##x4mode; \ |
| 5243 | .word setup_sprite_##texture_mode##_single_single_half_left##x4mode; \ |
| 5244 | .word setup_sprite_##texture_mode##_multi_multi_half_half##x4mode; \ |
| 5245 | .word 0x00000000; \ |
| 5246 | .word setup_sprite_##texture_mode##_multi_single_half_half##x4mode; \ |
| 5247 | |
| 5248 | |
| 5249 | setup_sprite_tiled_builder(4bpp,); |
| 5250 | setup_sprite_tiled_builder(8bpp,); |
| 5251 | |
| 5252 | #undef draw_mask_fb_ptr_left |
| 5253 | #undef draw_mask_fb_ptr_right |
| 5254 | |
| 5255 | setup_sprite_tiled_builder(4bpp, _4x); |
| 5256 | setup_sprite_tiled_builder(8bpp, _4x); |
| 5257 | |
| 5258 | |
| 5259 | #undef block_ptr |
| 5260 | #undef num_blocks |
| 5261 | #undef clut_ptr |
| 5262 | |
| 5263 | #define psx_gpu r0 |
| 5264 | #define block_ptr r0 |
| 5265 | #define num_blocks r1 |
| 5266 | #define clut_ptr r2 |
| 5267 | #define texel_shift_mask r3 |
| 5268 | #define block_pixels_a r4 |
| 5269 | #define block_pixels_b r5 |
| 5270 | #define texel_0 r6 |
| 5271 | #define texel_2 r7 |
| 5272 | #define texel_4 r8 |
| 5273 | #define texel_6 r9 |
| 5274 | #define texel_1 r10 |
| 5275 | #define texel_3 r11 |
| 5276 | #define texel_5 r12 |
| 5277 | #define texel_7 r14 |
| 5278 | #define texels_01 r6 |
| 5279 | #define texels_23 r7 |
| 5280 | #define texels_45 r8 |
| 5281 | #define texels_67 r9 |
| 5282 | |
| 5283 | function(texture_sprite_blocks_8bpp) |
| 5284 | stmdb sp!, { r4 - r11, r14 } |
| 5285 | movw texel_shift_mask, #(0xFF << 1) |
| 5286 | |
| 5287 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5288 | ldr clut_ptr, [ psx_gpu, #psx_gpu_clut_ptr_offset ] |
| 5289 | |
| 5290 | add block_ptr, psx_gpu, #psx_gpu_blocks_offset |
| 5291 | ldr block_pixels_a, [ block_ptr, #16 ] |
| 5292 | |
| 5293 | 0: |
| 5294 | and texel_0, texel_shift_mask, block_pixels_a, lsl #1 |
| 5295 | ldr block_pixels_b, [ block_ptr, #20 ] |
| 5296 | |
| 5297 | and texel_1, texel_shift_mask, block_pixels_a, lsr #7 |
| 5298 | ldrh texel_0, [ clut_ptr, texel_0 ] |
| 5299 | |
| 5300 | and texel_2, texel_shift_mask, block_pixels_a, lsr #15 |
| 5301 | ldrh texel_1, [ clut_ptr, texel_1 ] |
| 5302 | |
| 5303 | and texel_3, texel_shift_mask, block_pixels_a, lsr #23 |
| 5304 | ldr block_pixels_a, [ block_ptr, #(64 + 16) ] |
| 5305 | |
| 5306 | ldrh texel_2, [ clut_ptr, texel_2 ] |
| 5307 | and texel_4, texel_shift_mask, block_pixels_b, lsl #1 |
| 5308 | |
| 5309 | ldrh texel_3, [ clut_ptr, texel_3 ] |
| 5310 | and texel_5, texel_shift_mask, block_pixels_b, lsr #7 |
| 5311 | |
| 5312 | ldrh texel_4, [ clut_ptr, texel_4 ] |
| 5313 | and texel_6, texel_shift_mask, block_pixels_b, lsr #15 |
| 5314 | |
| 5315 | ldrh texel_5, [ clut_ptr, texel_5 ] |
| 5316 | and texel_7, texel_shift_mask, block_pixels_b, lsr #23 |
| 5317 | |
| 5318 | ldrh texel_6, [ clut_ptr, texel_6 ] |
| 5319 | orr texels_01, texel_0, texel_1, lsl #16 |
| 5320 | |
| 5321 | ldrh texel_7, [ clut_ptr, texel_7 ] |
| 5322 | orr texels_23, texel_2, texel_3, lsl #16 |
| 5323 | |
| 5324 | orr texels_45, texel_4, texel_5, lsl #16 |
| 5325 | str texels_01, [ block_ptr, #0 ] |
| 5326 | |
| 5327 | orr texels_67, texel_6, texel_7, lsl #16 |
| 5328 | str texels_23, [ block_ptr, #4 ] |
| 5329 | |
| 5330 | subs num_blocks, num_blocks, #1 |
| 5331 | str texels_45, [ block_ptr, #8 ] |
| 5332 | |
| 5333 | str texels_67, [ block_ptr, #12 ] |
| 5334 | add block_ptr, block_ptr, #64 |
| 5335 | |
| 5336 | bne 0b |
| 5337 | |
| 5338 | ldmia sp!, { r4 - r11, pc } |
| 5339 | |
| 5340 | |
| 5341 | #undef width_rounded |
| 5342 | #undef texture_mask |
| 5343 | #undef num_blocks |
| 5344 | #undef texture_offset |
| 5345 | #undef texels_low |
| 5346 | #undef texels_high |
| 5347 | #undef texels_wide_low |
| 5348 | #undef texels_wide_high |
| 5349 | #undef texels_wide |
| 5350 | #undef fb_ptr2 |
| 5351 | |
| 5352 | #define psx_gpu r0 |
| 5353 | #define x r1 |
| 5354 | #define y r2 |
| 5355 | #define u r3 |
| 5356 | #define v r4 |
| 5357 | #define width r5 |
| 5358 | #define height r6 |
| 5359 | #define left_offset r8 |
| 5360 | #define width_rounded r9 |
| 5361 | #define right_width r10 |
| 5362 | |
| 5363 | #define block_width r11 |
| 5364 | |
| 5365 | #define texture_offset_base r1 |
| 5366 | #define texture_mask r2 |
| 5367 | #define texture_page_ptr r3 |
| 5368 | #define num_blocks r4 |
| 5369 | #define block r5 |
| 5370 | #define fb_ptr r7 |
| 5371 | #define texture_offset r8 |
| 5372 | #define blocks_remaining r9 |
| 5373 | #define fb_ptr2 r10 |
| 5374 | #define fb_ptr_pitch r12 |
| 5375 | #define texture_block_ptr r14 |
| 5376 | |
| 5377 | #define texture_mask_width r2 |
| 5378 | #define texture_mask_height r3 |
| 5379 | #define left_mask_bits r4 |
| 5380 | #define right_mask_bits r5 |
| 5381 | |
| 5382 | |
| 5383 | #undef block_masks |
| 5384 | #undef block_masks_shifted |
| 5385 | #undef texels |
| 5386 | |
| 5387 | #define block_masks d0 |
| 5388 | #define block_masks_shifted d1 |
| 5389 | #define draw_mask_fb_ptr d2 |
| 5390 | #define texels q2 |
| 5391 | |
| 5392 | #define draw_mask_fb_ptr_a d2 |
| 5393 | #define draw_mask_fb_ptr_b d3 |
| 5394 | #define texels_low d4 |
| 5395 | #define texels_high d5 |
| 5396 | #define texels_wide_low d6 |
| 5397 | #define texels_wide_high d7 |
| 5398 | #define texels_wide q3 |
| 5399 | |
| 5400 | |
| 5401 | setup_sprites_16bpp_flush: |
| 5402 | vpush { d0 - d3 } |
| 5403 | |
| 5404 | stmdb sp!, { r0 - r3, r12, r14 } |
| 5405 | bl flush_render_block_buffer |
| 5406 | ldmia sp!, { r0 - r3, r12, r14 } |
| 5407 | |
| 5408 | vpop { d0 - d3 } |
| 5409 | |
| 5410 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5411 | mov num_blocks, block_width |
| 5412 | |
| 5413 | bx lr |
| 5414 | |
| 5415 | function(setup_sprite_16bpp) |
| 5416 | stmdb sp!, { r4 - r11, r14 } |
| 5417 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ] |
| 5418 | |
| 5419 | ldr v, [ sp, #36 ] |
| 5420 | add fb_ptr, fb_ptr, y, lsl #11 |
| 5421 | |
| 5422 | ldr width, [ sp, #40 ] |
| 5423 | add fb_ptr, fb_ptr, x, lsl #1 |
| 5424 | |
| 5425 | ldr height, [ sp, #44 ] |
| 5426 | and left_offset, u, #0x7 |
| 5427 | |
| 5428 | add texture_offset_base, u, u |
| 5429 | add width_rounded, width, #7 |
| 5430 | |
| 5431 | add texture_offset_base, v, lsl #11 |
| 5432 | mov left_mask_bits, #0xFF |
| 5433 | |
| 5434 | ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ] |
| 5435 | add width_rounded, width_rounded, left_offset |
| 5436 | |
| 5437 | ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ] |
| 5438 | sub fb_ptr, fb_ptr, left_offset, lsl #1 |
| 5439 | |
| 5440 | add texture_mask, texture_mask_width, texture_mask_width |
| 5441 | mov right_mask_bits, #0xFE |
| 5442 | |
| 5443 | and right_width, width_rounded, #0x7 |
| 5444 | mvn left_mask_bits, left_mask_bits, lsl left_offset |
| 5445 | |
| 5446 | add texture_mask, texture_mask_height, lsl #11 |
| 5447 | mov block_width, width_rounded, lsr #3 |
| 5448 | |
| 5449 | mov right_mask_bits, right_mask_bits, lsl right_width |
| 5450 | movw fb_ptr_pitch, #(2048 + 16) |
| 5451 | |
| 5452 | sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4 |
| 5453 | vmov block_masks, left_mask_bits, right_mask_bits |
| 5454 | |
| 5455 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5456 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5457 | |
| 5458 | bic texture_offset_base, texture_offset_base, #0xF |
| 5459 | cmp block_width, #1 |
| 5460 | |
| 5461 | ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] |
| 5462 | add block, block, num_blocks, lsl #6 |
| 5463 | |
| 5464 | bne 0f |
| 5465 | |
| 5466 | vext.32 block_masks_shifted, block_masks, block_masks, #1 |
| 5467 | vorr.u32 block_masks, block_masks, block_masks_shifted |
| 5468 | vdup.u8 draw_mask_fb_ptr, block_masks[0] |
| 5469 | |
| 5470 | 1: |
| 5471 | add num_blocks, num_blocks, #1 |
| 5472 | cmp num_blocks, #MAX_BLOCKS |
| 5473 | blgt setup_sprites_16bpp_flush |
| 5474 | |
| 5475 | and texture_block_ptr, texture_offset_base, texture_mask |
| 5476 | subs height, height, #1 |
| 5477 | |
| 5478 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5479 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5480 | |
| 5481 | vst1.u32 { texels }, [ block, :128 ] |
| 5482 | add block, block, #40 |
| 5483 | |
| 5484 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5485 | pld [ fb_ptr ] |
| 5486 | |
| 5487 | vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ] |
| 5488 | |
| 5489 | add block, block, #24 |
| 5490 | add texture_offset_base, texture_offset_base, #2048 |
| 5491 | add fb_ptr, fb_ptr, #2048 |
| 5492 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5493 | bne 1b |
| 5494 | |
| 5495 | ldmia sp!, { r4 - r11, pc } |
| 5496 | |
| 5497 | 0: |
| 5498 | add num_blocks, num_blocks, block_width |
| 5499 | mov texture_offset, texture_offset_base |
| 5500 | |
| 5501 | cmp num_blocks, #MAX_BLOCKS |
| 5502 | blgt setup_sprites_16bpp_flush |
| 5503 | |
| 5504 | add texture_offset_base, texture_offset_base, #2048 |
| 5505 | and texture_block_ptr, texture_offset, texture_mask |
| 5506 | |
| 5507 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5508 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5509 | |
| 5510 | vst1.u32 { texels }, [ block, :128 ] |
| 5511 | add block, block, #40 |
| 5512 | |
| 5513 | vdup.u8 draw_mask_fb_ptr, block_masks[0] |
| 5514 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5515 | pld [ fb_ptr ] |
| 5516 | |
| 5517 | vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ] |
| 5518 | subs blocks_remaining, block_width, #2 |
| 5519 | |
| 5520 | add texture_offset, texture_offset, #16 |
| 5521 | add fb_ptr, fb_ptr, #16 |
| 5522 | |
| 5523 | vmov.u8 draw_mask_fb_ptr, #0 |
| 5524 | |
| 5525 | add block, block, #24 |
| 5526 | beq 2f |
| 5527 | |
| 5528 | 1: |
| 5529 | and texture_block_ptr, texture_offset, texture_mask |
| 5530 | subs blocks_remaining, blocks_remaining, #1 |
| 5531 | |
| 5532 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5533 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5534 | |
| 5535 | vst1.u32 { texels }, [ block, :128 ] |
| 5536 | add block, block, #40 |
| 5537 | |
| 5538 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5539 | pld [ fb_ptr ] |
| 5540 | |
| 5541 | vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ] |
| 5542 | |
| 5543 | add texture_offset, texture_offset, #16 |
| 5544 | add fb_ptr, fb_ptr, #16 |
| 5545 | |
| 5546 | add block, block, #24 |
| 5547 | bne 1b |
| 5548 | |
| 5549 | 2: |
| 5550 | and texture_block_ptr, texture_offset, texture_mask |
| 5551 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5552 | |
| 5553 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5554 | vdup.u8 draw_mask_fb_ptr, block_masks[4] |
| 5555 | |
| 5556 | vst1.u32 { texels }, [ block, :128 ] |
| 5557 | add block, block, #40 |
| 5558 | |
| 5559 | vmov.u32 draw_mask_fb_ptr[1], fb_ptr |
| 5560 | vst1.u32 { draw_mask_fb_ptr }, [ block, :64 ] |
| 5561 | |
| 5562 | add block, block, #24 |
| 5563 | subs height, height, #1 |
| 5564 | |
| 5565 | add fb_ptr, fb_ptr, fb_ptr_pitch |
| 5566 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5567 | |
| 5568 | bne 0b |
| 5569 | |
| 5570 | ldmia sp!, { r4 - r11, pc } |
| 5571 | |
| 5572 | |
| 5573 | // 4x version |
| 5574 | // FIXME: duplicate code with normal version :( |
| 5575 | #undef draw_mask_fb_ptr |
| 5576 | |
| 5577 | function(setup_sprite_16bpp_4x) |
| 5578 | stmdb sp!, { r4 - r11, r14 } |
| 5579 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ] |
| 5580 | |
| 5581 | ldr v, [ sp, #36 ] |
| 5582 | add fb_ptr, fb_ptr, y, lsl #11 |
| 5583 | |
| 5584 | ldr width, [ sp, #40 ] |
| 5585 | add fb_ptr, fb_ptr, x, lsl #1 |
| 5586 | |
| 5587 | ldr height, [ sp, #44 ] |
| 5588 | and left_offset, u, #0x7 |
| 5589 | |
| 5590 | add texture_offset_base, u, u |
| 5591 | add width_rounded, width, #7 |
| 5592 | |
| 5593 | add texture_offset_base, v, lsl #11 |
| 5594 | movw left_mask_bits, #0xFFFF |
| 5595 | |
| 5596 | ldrb texture_mask_width, [ psx_gpu, #psx_gpu_texture_mask_width_offset ] |
| 5597 | add width_rounded, width_rounded, left_offset |
| 5598 | |
| 5599 | lsl left_offset, #1 |
| 5600 | |
| 5601 | ldrb texture_mask_height, [ psx_gpu, #psx_gpu_texture_mask_height_offset ] |
| 5602 | sub fb_ptr, fb_ptr, left_offset, lsl #1 |
| 5603 | |
| 5604 | add texture_mask, texture_mask_width, texture_mask_width |
| 5605 | movw right_mask_bits, #0xFFFC |
| 5606 | |
| 5607 | and right_width, width_rounded, #0x7 |
| 5608 | mvn left_mask_bits, left_mask_bits, lsl left_offset |
| 5609 | |
| 5610 | lsl right_width, #1 |
| 5611 | |
| 5612 | add texture_mask, texture_mask_height, lsl #11 |
| 5613 | mov block_width, width_rounded, lsr #3 |
| 5614 | |
| 5615 | mov right_mask_bits, right_mask_bits, lsl right_width |
| 5616 | movw fb_ptr_pitch, #(2048 + 16) * 2 |
| 5617 | |
| 5618 | sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1 |
| 5619 | vmov block_masks, left_mask_bits, right_mask_bits |
| 5620 | |
| 5621 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5622 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5623 | |
| 5624 | bic texture_offset_base, texture_offset_base, #0xF |
| 5625 | cmp block_width, #1 |
| 5626 | |
| 5627 | ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_ptr_offset ] |
| 5628 | add block, block, num_blocks, lsl #6 |
| 5629 | |
| 5630 | lsl block_width, #2 |
| 5631 | bne 0f |
| 5632 | |
| 5633 | vext.32 block_masks_shifted, block_masks, block_masks, #1 |
| 5634 | vorr.u32 block_masks, block_masks, block_masks_shifted |
| 5635 | vdup.u8 draw_mask_fb_ptr_a, block_masks[0] |
| 5636 | vdup.u8 draw_mask_fb_ptr_b, block_masks[1] |
| 5637 | |
| 5638 | 1: |
| 5639 | add num_blocks, num_blocks, block_width |
| 5640 | cmp num_blocks, #MAX_BLOCKS |
| 5641 | blgt setup_sprites_16bpp_flush |
| 5642 | |
| 5643 | and texture_block_ptr, texture_offset_base, texture_mask |
| 5644 | subs height, height, #1 |
| 5645 | |
| 5646 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5647 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5648 | |
| 5649 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5650 | |
| 5651 | add texture_offset_base, texture_offset_base, #2048 |
| 5652 | add fb_ptr, fb_ptr, #2048*2 |
| 5653 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5654 | bne 1b |
| 5655 | |
| 5656 | ldmia sp!, { r4 - r11, pc } |
| 5657 | |
| 5658 | 0: |
| 5659 | add num_blocks, num_blocks, block_width |
| 5660 | mov texture_offset, texture_offset_base |
| 5661 | |
| 5662 | vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits |
| 5663 | vdup.u8 draw_mask_fb_ptr_b, block_masks[1] |
| 5664 | |
| 5665 | cmp num_blocks, #MAX_BLOCKS |
| 5666 | blgt setup_sprites_16bpp_flush |
| 5667 | |
| 5668 | add texture_offset_base, texture_offset_base, #2048 |
| 5669 | and texture_block_ptr, texture_offset, texture_mask |
| 5670 | |
| 5671 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5672 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5673 | |
| 5674 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5675 | |
| 5676 | subs blocks_remaining, block_width, #2*4 |
| 5677 | add texture_offset, texture_offset, #16 |
| 5678 | |
| 5679 | vmov.u8 draw_mask_fb_ptr_a, #0 |
| 5680 | vmov.u8 draw_mask_fb_ptr_b, #0 |
| 5681 | |
| 5682 | add fb_ptr, fb_ptr, #16*2 |
| 5683 | beq 2f |
| 5684 | |
| 5685 | 1: |
| 5686 | and texture_block_ptr, texture_offset, texture_mask |
| 5687 | subs blocks_remaining, blocks_remaining, #4 |
| 5688 | |
| 5689 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5690 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5691 | |
| 5692 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5693 | add texture_offset, texture_offset, #16 |
| 5694 | |
| 5695 | add fb_ptr, fb_ptr, #16*2 |
| 5696 | bgt 1b |
| 5697 | |
| 5698 | 2: |
| 5699 | vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits |
| 5700 | vdup.u8 draw_mask_fb_ptr_b, block_masks[5] |
| 5701 | |
| 5702 | and texture_block_ptr, texture_offset, texture_mask |
| 5703 | add texture_block_ptr, texture_page_ptr, texture_block_ptr |
| 5704 | |
| 5705 | vld1.u32 { texels }, [ texture_block_ptr, :128 ] |
| 5706 | |
| 5707 | do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b) |
| 5708 | subs height, height, #1 |
| 5709 | |
| 5710 | add fb_ptr, fb_ptr, fb_ptr_pitch |
| 5711 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5712 | |
| 5713 | bne 0b |
| 5714 | |
| 5715 | ldmia sp!, { r4 - r11, pc } |
| 5716 | |
| 5717 | |
| 5718 | #undef width |
| 5719 | #undef right_width |
| 5720 | #undef right_mask_bits |
| 5721 | #undef color |
| 5722 | #undef height |
| 5723 | #undef blocks_remaining |
| 5724 | #undef colors |
| 5725 | #undef right_mask |
| 5726 | #undef test_mask |
| 5727 | #undef draw_mask |
| 5728 | |
| 5729 | #define psx_gpu r0 |
| 5730 | #define x r1 |
| 5731 | #define y r2 |
| 5732 | #define width r3 |
| 5733 | #define right_width r5 |
| 5734 | #define right_mask_bits r6 |
| 5735 | #define fb_ptr r7 |
| 5736 | #define color r8 |
| 5737 | #define height r9 |
| 5738 | #define fb_ptr_pitch r12 |
| 5739 | |
| 5740 | // referenced by setup_sprites_16bpp_flush |
| 5741 | #define num_blocks r4 |
| 5742 | #define block r5 |
| 5743 | #define block_width r11 |
| 5744 | |
| 5745 | #define color_r r1 |
| 5746 | #define color_g r2 |
| 5747 | #define color_b r8 |
| 5748 | #define blocks_remaining r6 |
| 5749 | |
| 5750 | #define colors q0 |
| 5751 | #define right_mask q1 |
| 5752 | #define test_mask q2 |
| 5753 | #define draw_mask q2 |
| 5754 | #define draw_mask_bits_fb_ptr d6 |
| 5755 | |
| 5756 | |
| 5757 | .align 3 |
| 5758 | |
| 5759 | function(setup_sprite_untextured) |
| 5760 | ldrh r12, [ psx_gpu, #psx_gpu_render_state_offset ] |
| 5761 | tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \ |
| 5762 | | RENDER_FLAGS_BLEND) |
| 5763 | ldreqb r12, [ psx_gpu, #psx_gpu_render_mode_offset ] |
| 5764 | tsteq r12, #RENDER_INTERLACE_ENABLED |
| 5765 | beq setup_sprite_untextured_simple |
| 5766 | |
| 5767 | stmdb sp!, { r4 - r11, r14 } |
| 5768 | |
| 5769 | ldr width, [ sp, #40 ] |
| 5770 | ldr fb_ptr, [ psx_gpu, #psx_gpu_vram_out_ptr_offset ] |
| 5771 | |
| 5772 | ldr height, [ sp, #44 ] |
| 5773 | add fb_ptr, fb_ptr, y, lsl #11 |
| 5774 | |
| 5775 | add fb_ptr, fb_ptr, x, lsl #1 |
| 5776 | sub right_width, width, #1 |
| 5777 | |
| 5778 | ldr color, [ sp, #48 ] |
| 5779 | and right_width, #7 |
| 5780 | |
| 5781 | add block_width, width, #7 |
| 5782 | add right_width, #1 |
| 5783 | |
| 5784 | lsr block_width, #3 |
| 5785 | mov right_mask_bits, #0xff |
| 5786 | |
| 5787 | sub fb_ptr_pitch, block_width, #1 |
| 5788 | lsl right_mask_bits, right_width |
| 5789 | |
| 5790 | lsl fb_ptr_pitch, #3+1 |
| 5791 | ubfx color_r, color, #3, #5 |
| 5792 | |
| 5793 | rsb fb_ptr_pitch, #1024*2 |
| 5794 | ubfx color_g, color, #11, #5 |
| 5795 | |
| 5796 | vld1.u32 { test_mask }, [ psx_gpu, :128 ] |
| 5797 | ubfx color_b, color, #19, #5 |
| 5798 | |
| 5799 | vdup.u16 right_mask, right_mask_bits |
| 5800 | orr color, color_r, color_b, lsl #10 |
| 5801 | |
| 5802 | ldrh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5803 | orr color, color, color_g, lsl #5 |
| 5804 | |
| 5805 | vtst.u16 right_mask, right_mask, test_mask |
| 5806 | add block, psx_gpu, #psx_gpu_blocks_offset |
| 5807 | |
| 5808 | vdup.u16 colors, color |
| 5809 | add block, block, num_blocks, lsl #6 |
| 5810 | |
| 5811 | |
| 5812 | setup_sprite_untextured_height_loop: |
| 5813 | add num_blocks, block_width |
| 5814 | sub blocks_remaining, block_width, #1 |
| 5815 | |
| 5816 | cmp num_blocks, #MAX_BLOCKS |
| 5817 | blgt setup_sprites_16bpp_flush |
| 5818 | |
| 5819 | cmp blocks_remaining, #0 |
| 5820 | ble 1f |
| 5821 | |
| 5822 | vmov.u8 draw_mask, #0 /* zero_mask */ |
| 5823 | vmov.u8 draw_mask_bits_fb_ptr, #0 |
| 5824 | |
| 5825 | 0: |
| 5826 | vst1.u32 { draw_mask }, [ block, :128 ]! |
| 5827 | subs blocks_remaining, #1 |
| 5828 | |
| 5829 | vst1.u32 { colors }, [ block, :128 ] |
| 5830 | add block, block, #24 |
| 5831 | |
| 5832 | vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr |
| 5833 | vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ] |
| 5834 | |
| 5835 | add block, block, #24 |
| 5836 | add fb_ptr, #8*2 |
| 5837 | bgt 0b |
| 5838 | |
| 5839 | 1: |
| 5840 | vst1.u32 { right_mask }, [ block, :128 ]! |
| 5841 | subs height, #1 |
| 5842 | |
| 5843 | vst1.u32 { colors }, [ block, :128 ] |
| 5844 | add block, block, #24 |
| 5845 | |
| 5846 | vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr |
| 5847 | vst1.u32 { draw_mask_bits_fb_ptr }, [ block, :64 ] |
| 5848 | |
| 5849 | add block, block, #24 |
| 5850 | add fb_ptr, fb_ptr_pitch |
| 5851 | |
| 5852 | strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ] |
| 5853 | bgt setup_sprite_untextured_height_loop |
| 5854 | |
| 5855 | ldmia sp!, { r4 - r11, pc } |
| 5856 | |
| 5857 | |
| 5858 | |
| 5859 | #undef texture_page_ptr |
| 5860 | #undef vram_ptr |
| 5861 | #undef dirty_textures_mask |
| 5862 | #undef current_texture_mask |
| 5863 | |
| 5864 | #define psx_gpu r0 |
| 5865 | #define current_texture_page r1 |
| 5866 | #define texture_page_ptr r2 |
| 5867 | #define vram_ptr_a r3 |
| 5868 | #define current_texture_page_x r12 |
| 5869 | #define current_texture_page_y r4 |
| 5870 | #define dirty_textures_mask r5 |
| 5871 | #define tile_y r6 |
| 5872 | #define tile_x r7 |
| 5873 | #define sub_y r8 |
| 5874 | #define current_texture_mask r9 |
| 5875 | #define c_4096 r10 |
| 5876 | #define vram_ptr_b r11 |
| 5877 | |
| 5878 | #define texel_block_a d0 |
| 5879 | #define texel_block_b d1 |
| 5880 | #define texel_block_expanded_a q1 |
| 5881 | #define texel_block_expanded_b q2 |
| 5882 | #define texel_block_expanded_ab q2 |
| 5883 | #define texel_block_expanded_c q3 |
| 5884 | #define texel_block_expanded_d q4 |
| 5885 | #define texel_block_expanded_cd q3 |
| 5886 | |
| 5887 | function(update_texture_4bpp_cache) |
| 5888 | stmdb sp!, { r4 - r11, r14 } |
| 5889 | vpush { q0 - q3 } |
| 5890 | |
| 5891 | ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ] |
| 5892 | |
| 5893 | ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ] |
| 5894 | ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ] |
| 5895 | |
| 5896 | and current_texture_page_x, current_texture_page, #0xF |
| 5897 | ldr current_texture_mask, [ psx_gpu, #psx_gpu_current_texture_mask_offset ] |
| 5898 | |
| 5899 | mov current_texture_page_y, current_texture_page, lsr #4 |
| 5900 | ldr dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ] |
| 5901 | |
| 5902 | add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19 |
| 5903 | mov tile_y, #16 |
| 5904 | |
| 5905 | add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7 |
| 5906 | bic dirty_textures_mask, current_texture_mask |
| 5907 | |
| 5908 | mov tile_x, #16 |
| 5909 | str dirty_textures_mask, [ psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset ] |
| 5910 | |
| 5911 | mov sub_y, #8 |
| 5912 | movw c_4096, #4096 |
| 5913 | |
| 5914 | add vram_ptr_b, vram_ptr_a, #2048 |
| 5915 | |
| 5916 | 0: |
| 5917 | vld1.u32 { texel_block_a }, [ vram_ptr_a, :64 ], c_4096 |
| 5918 | vld1.u32 { texel_block_b }, [ vram_ptr_b, :64 ], c_4096 |
| 5919 | |
| 5920 | vmovl.u8 texel_block_expanded_a, texel_block_a |
| 5921 | vshll.u8 texel_block_expanded_b, texel_block_a, #4 |
| 5922 | vmovl.u8 texel_block_expanded_c, texel_block_b |
| 5923 | vshll.u8 texel_block_expanded_d, texel_block_b, #4 |
| 5924 | |
| 5925 | vbic.u16 texel_block_expanded_a, #0x00F0 |
| 5926 | vbic.u16 texel_block_expanded_b, #0x00F0 |
| 5927 | vbic.u16 texel_block_expanded_c, #0x00F0 |
| 5928 | vbic.u16 texel_block_expanded_d, #0x00F0 |
| 5929 | |
| 5930 | vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \ |
| 5931 | texel_block_expanded_b |
| 5932 | vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \ |
| 5933 | texel_block_expanded_d |
| 5934 | |
| 5935 | vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \ |
| 5936 | [ texture_page_ptr, :256 ]! |
| 5937 | |
| 5938 | subs sub_y, sub_y, #1 |
| 5939 | bne 0b |
| 5940 | |
| 5941 | mov sub_y, #8 |
| 5942 | add vram_ptr_a, vram_ptr_a, #8 |
| 5943 | add vram_ptr_b, vram_ptr_b, #8 |
| 5944 | |
| 5945 | sub vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 5946 | sub vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 5947 | |
| 5948 | subs tile_x, tile_x, #1 |
| 5949 | bne 0b |
| 5950 | |
| 5951 | mov tile_x, #16 |
| 5952 | add vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 5953 | add vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 5954 | |
| 5955 | sub vram_ptr_a, vram_ptr_a, #(8 * 16) |
| 5956 | sub vram_ptr_b, vram_ptr_b, #(8 * 16) |
| 5957 | |
| 5958 | subs tile_y, tile_y, #1 |
| 5959 | bne 0b |
| 5960 | |
| 5961 | vpop { q0 - q3 } |
| 5962 | ldmia sp!, { r4 - r11, pc } |
| 5963 | |
| 5964 | |
| 5965 | #undef current_texture_page |
| 5966 | |
| 5967 | #define psx_gpu r0 |
| 5968 | #define texture_page r1 |
| 5969 | #define texture_page_ptr r2 |
| 5970 | #define vram_ptr_a r3 |
| 5971 | #define texture_page_x r12 |
| 5972 | #define texture_page_y r4 |
| 5973 | #define current_texture_page r5 |
| 5974 | #define tile_y r6 |
| 5975 | #define tile_x r7 |
| 5976 | #define sub_y r8 |
| 5977 | #define c_4096 r10 |
| 5978 | #define vram_ptr_b r11 |
| 5979 | |
| 5980 | |
| 5981 | #undef texels_a |
| 5982 | #undef texels_b |
| 5983 | |
| 5984 | #define texels_a q0 |
| 5985 | #define texels_b q1 |
| 5986 | #define texels_c q2 |
| 5987 | #define texels_d q3 |
| 5988 | |
| 5989 | |
| 5990 | function(update_texture_8bpp_cache_slice) |
| 5991 | stmdb sp!, { r4 - r11, r14 } |
| 5992 | vpush { q0 - q3 } |
| 5993 | |
| 5994 | ldrb current_texture_page, [ psx_gpu, #psx_gpu_current_texture_page_offset ] |
| 5995 | ldr vram_ptr_a, [ psx_gpu, #psx_gpu_vram_ptr_offset ] |
| 5996 | |
| 5997 | ldr texture_page_ptr, [ psx_gpu, #psx_gpu_texture_page_base_offset ] |
| 5998 | mov tile_y, #16 |
| 5999 | |
| 6000 | and texture_page_x, texture_page, #0xF |
| 6001 | mov texture_page_y, texture_page, lsr #4 |
| 6002 | |
| 6003 | add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7 |
| 6004 | mov tile_x, #8 |
| 6005 | |
| 6006 | add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19 |
| 6007 | eor current_texture_page, current_texture_page, texture_page |
| 6008 | |
| 6009 | ands current_texture_page, current_texture_page, #0x1 |
| 6010 | mov sub_y, #4 |
| 6011 | |
| 6012 | addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16) |
| 6013 | movw c_4096, #4096 |
| 6014 | |
| 6015 | add vram_ptr_b, vram_ptr_a, #2048 |
| 6016 | |
| 6017 | 0: |
| 6018 | vld1.u32 { texels_a }, [ vram_ptr_a, :128 ], c_4096 |
| 6019 | vld1.u32 { texels_b }, [ vram_ptr_b, :128 ], c_4096 |
| 6020 | vld1.u32 { texels_c }, [ vram_ptr_a, :128 ], c_4096 |
| 6021 | vld1.u32 { texels_d }, [ vram_ptr_b, :128 ], c_4096 |
| 6022 | |
| 6023 | vst1.u32 { texels_a, texels_b }, [ texture_page_ptr, :256 ]! |
| 6024 | vst1.u32 { texels_c, texels_d }, [ texture_page_ptr, :256 ]! |
| 6025 | |
| 6026 | subs sub_y, sub_y, #1 |
| 6027 | bne 0b |
| 6028 | |
| 6029 | mov sub_y, #4 |
| 6030 | |
| 6031 | add vram_ptr_a, vram_ptr_a, #16 |
| 6032 | add vram_ptr_b, vram_ptr_b, #16 |
| 6033 | |
| 6034 | sub vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 6035 | sub vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 6036 | |
| 6037 | subs tile_x, tile_x, #1 |
| 6038 | bne 0b |
| 6039 | |
| 6040 | mov tile_x, #8 |
| 6041 | |
| 6042 | add vram_ptr_a, vram_ptr_a, #(16 * 2048) |
| 6043 | add vram_ptr_b, vram_ptr_b, #(16 * 2048) |
| 6044 | |
| 6045 | sub vram_ptr_a, vram_ptr_a, #(8 * 16) |
| 6046 | sub vram_ptr_b, vram_ptr_b, #(8 * 16) |
| 6047 | |
| 6048 | subs tile_y, tile_y, #1 |
| 6049 | add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16) |
| 6050 | |
| 6051 | bne 0b |
| 6052 | |
| 6053 | vpop { q0 - q3 } |
| 6054 | ldmia sp!, { r4 - r11, pc } |
| 6055 | |
| 6056 | |
| 6057 | /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */ |
| 6058 | function(scale2x_tiles8) |
| 6059 | push { r4, r14 } |
| 6060 | |
| 6061 | mov r4, r1 |
| 6062 | add r12, r0, #1024*2 |
| 6063 | mov r14, r2 |
| 6064 | |
| 6065 | 0: |
| 6066 | vld1.u16 { q0 }, [ r1, :128 ]! |
| 6067 | vld1.u16 { q2 }, [ r1, :128 ]! |
| 6068 | vmov q1, q0 |
| 6069 | vmov q3, q2 |
| 6070 | vzip.16 q0, q1 |
| 6071 | vzip.16 q2, q3 |
| 6072 | subs r14, #2 |
| 6073 | vst1.u16 { q0, q1 }, [ r0, :128 ]! |
| 6074 | vst1.u16 { q0, q1 }, [ r12, :128 ]! |
| 6075 | blt 1f |
| 6076 | vst1.u16 { q2, q3 }, [ r0, :128 ]! |
| 6077 | vst1.u16 { q2, q3 }, [ r12, :128 ]! |
| 6078 | bgt 0b |
| 6079 | 1: |
| 6080 | subs r3, #1 |
| 6081 | mov r14, r2 |
| 6082 | add r0, #1024*2*2 |
| 6083 | add r4, #1024*2 |
| 6084 | sub r0, r2, lsl #4+1 |
| 6085 | mov r1, r4 |
| 6086 | add r12, r0, #1024*2 |
| 6087 | bgt 0b |
| 6088 | nop |
| 6089 | |
| 6090 | pop { r4, pc } |
| 6091 | |
| 6092 | // vim:filetype=armasm |