2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
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.
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.
16 #define RENDER_INTERLACE_ENABLED 0x1
19 #include "psx_gpu_offsets.h"
21 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
23 #define edge_data_left_x_offset 0
24 #define edge_data_num_blocks_offset 2
25 #define edge_data_right_mask_offset 4
26 #define edge_data_y_offset 6
32 #define save_abi_regs() \
34 #define restore_abi_regs() \
37 #define save_abi_regs()
38 #define restore_abi_regs()
82 #define gw_by_l gw_bx_l
102 #define uvrg_xxxx0 q3
107 #define uvrg_xxxx1 q4
112 #define uvrg_xxxx2 q5
117 #define yyyy_uvrg0 q6
122 #define yyyy_uvrg1 q7
127 #define yyyy_uvrg2 q8
154 #define uvrg_base q11
197 #define uvrgb_phase q13
201 #include "arm_features.h"
203 #define function(name) FUNCTION(name):
205 #ifndef TEXRELS_FORBIDDEN
207 #define JT_OP_REL(table_label, index_reg, temp)
208 #define JT_OP(x...) x
209 #define JTE(start, target) target
213 #define JT_OP_REL(table_label, index_reg, temp) \
214 adr temp, table_label; \
215 ldr temp, [temp, index_reg, lsl #2]; \
219 #define JTE(start, target) (target - start)
224 #define flush_render_block_buffer _flush_render_block_buffer
225 #define update_texture_8bpp_cache _update_texture_8bpp_cache
226 #define setup_blocks_uv_adj_hack _setup_blocks_uv_adj_hack
234 function(compute_all_gradients)
235 // First compute the triangle area reciprocal and shift. The division will
236 // happen concurrently with much of the work which follows.
237 @ r12 = psx_gpu->triangle_area
238 ldr r12, [psx_gpu, #psx_gpu_triangle_area_offset]
239 stmdb sp!, { r4 - r11, lr }
242 @ load exponent of 62 into upper half of double
244 clz r14, r12 @ r14 = shift
246 movt r4, #((62 + 1023) << 4)
247 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
249 @ load area normalized into lower half of double
251 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
253 movt r4, #((1022 + 31) << 4)
256 add r4, r4, r12, lsr #11
259 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
261 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
262 // ( d0 * d1 ) - ( d2 * d3 ) =
263 // ( m0 ) - ( m1 ) = gradient
265 // This is split to do 12 elements at a time over three sets: a, b, and c.
266 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
267 // two of the slots are unused.
269 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
272 // First type is: uvrg bxxx xxxx
273 // Second type is: yyyy ybyy uvrg
274 // Since x_a and y_c are the same the same variable is used for both.
276 vld1.u32 { v0 }, [v_a, :128] @ v0 = { uvrg0, b0, x0, y0 }
277 ldrsh x0, [v_a, #8] @ load x0
279 vld1.u32 { v1 }, [v_b, :128] @ v1 = { uvrg1, b1, x1, y1}
280 ldrh x1, [v_b, #8] @ load x1
282 vld1.u32 { v2 }, [v_c, :128] @ v2 = { uvrg2, b2, x2, y2 }
283 ldrh x2, [v_c, #8] @ load x2
285 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
286 ldrh y0, [v_a, #10] @ load y0
288 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
289 ldrh y1, [v_b, #10] @ load y1
291 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
292 ldrh y2, [v_c, #10] @ load y2
294 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
295 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
297 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
298 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
300 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
301 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
303 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
304 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
306 ldrb b2, [v_c, #4] @ load b2
307 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
309 ldrb b1, [v_b, #4] @ load b1
310 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
312 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
313 vsub.s16 d0_ab, x1_ab, x0_ab
315 ldrb b0, [v_a, #4] @ load b0
316 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
318 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
319 vsub.s16 d2_ab, x2_ab, x1_ab
321 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
322 vsub.s16 d1_ab, y2_ab, y1_ab
324 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
325 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
327 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
328 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
330 vsub.s16 d3_ab, y1_ab, y0_ab
331 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
332 @ ((x2 - X1) * (b1 - b0))
333 vmull.s16 ga_uvrg_x, d0_a, d1_a
334 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
335 @ ((b2 - b1) * (y1 - y0))
336 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
337 movs gs_bx, ga_bx, asr #31
339 vmull.s16 ga_uvrg_y, d0_b, d1_b
340 rsbmi ga_bx, ga_bx, #0
342 @ r12 = psx_gpu->uvrgb_phase
343 ldr r12, [psx_gpu, #psx_gpu_uvrgb_phase_offset]
345 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
346 movs gs_by, ga_by, asr #31
348 vshr.u64 d0, d30, #22
349 add b_base, r12, b0, lsl #16
351 vdup.u32 uvrgb_phase, r12
353 rsbmi ga_by, ga_by, #0
354 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
356 @ r12 = psx_gpu->triangle_winding_offset
357 ldrb r12, [psx_gpu, #psx_gpu_triangle_winding_offset]
358 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
360 rsb r12, r12, #0 @ r12 = -(triangle->winding)
362 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
363 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
365 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
366 vdup.u32 r_shift, r14 @ r_shift = { shift, shift*, shift, shift* }
367 @ * - vshl.u64: ignored by hw
368 vadd.u32 uvrg_base, uvrgb_phase
369 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
371 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
372 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
374 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
375 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
376 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
377 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
379 vshl.u64 gw_rg_x, gw_rg_x, r_shift
380 vshl.u64 gw_uv_x, gw_uv_x, r_shift
381 vshl.u64 gw_rg_y, gw_rg_y, r_shift
382 vshl.u64 gw_uv_y, gw_uv_y, r_shift
384 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
385 vmovn.u64 g_uv_x, gw_uv_x
387 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
388 vmovn.u64 g_rg_x, gw_rg_x
390 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
391 vmovn.u64 g_uv_y, gw_uv_y
393 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
394 vmovn.u64 g_rg_y, gw_rg_y
396 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
397 mov ga_bx, ga_bx, lsl #13
399 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
400 mov ga_by, ga_by, lsl #13
403 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
405 vshl.u32 g_uvrg_x, g_uvrg_x, #4
406 vshl.u32 g_uvrg_y, g_uvrg_y, #4
408 umull gw_by_l, gw_by_h, ga_by, area_r_s
409 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
411 eor gs_bx, gs_bx, r12
412 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
414 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
415 eor gs_by, gs_by, r12
417 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
418 add store_a, psx_gpu, #psx_gpu_uvrg_offset
420 sub r11, r11, #(32 - 13)
422 add store_b, store_a, #16
425 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
426 vst1.u32 { uvrg_base }, [store_a, :128], store_inc
428 vst1.u32 { uvrg_dx1 }, [store_b, :128], store_inc
429 mov g_bx, gw_bx_h, lsr r11
431 vst1.u32 { g_uvrg_y }, [store_a, :128], store_inc
432 mov g_by, gw_by_h, lsr r11
434 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
435 [store_b, :128], store_inc
436 eor g_bx, g_bx, gs_bx
438 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
439 [store_b, :128], store_inc
440 sub g_bx, g_bx, gs_bx
443 eor g_by, g_by, gs_by
445 mls b_base, g_bx, x0, b_base
446 sub g_by, g_by, gs_by
451 add g_bx2, g_bx, g_bx
452 add g_bx3, g_bx, g_bx2
454 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
457 ldmia sp!, { r4 - r11, pc }
474 #define height_minor_a r7
475 #define height_minor_b r8
476 #define height_major r9
479 #define reciprocal_table_ptr r10
481 #define edge_alt_low r4
482 #define edge_alt_high r5
483 #define edge_dx_dy_alt r6
484 #define edge_shift_alt r10
486 #define edge_dx_dy_alt_low r4
487 #define edge_dx_dy_alt_high r5
489 #define span_edge_data r4
490 #define span_uvrg_offset r5
491 #define span_b_offset r6
499 #define alternate_x q0
500 #define alternate_dx_dy q1
501 #define alternate_x_32 q2
503 #define alternate_x_low d0
504 #define alternate_x_high d1
505 #define alternate_dx_dy_low d2
506 #define alternate_dx_dy_high d3
507 #define alternate_x_32_low d4
508 #define alternate_x_32_high d5
512 #define left_dx_dy q5
513 #define right_dx_dy q6
515 #define right_edge q8
517 #define left_x_low d6
518 #define left_x_high d7
519 #define right_x_low d8
520 #define right_x_high d9
521 #define left_dx_dy_low d10
522 #define left_dx_dy_high d11
523 #define right_dx_dy_low d12
524 #define right_dx_dy_high d13
525 #define left_edge_low d14
526 #define left_edge_high d15
527 #define right_edge_low d16
528 #define right_edge_high d17
530 #define y_mid_point d18
533 #define left_right_x_16 q11
534 #define span_shifts_y q12
537 #define span_shifts d24
542 #define left_right_x_16_low d22
543 #define left_right_x_16_high d23
549 #define alternate_x_16 d4
552 #define v_clip_low d6
554 #define right_x_32 q10
555 #define left_x_32 q11
556 #define alternate_select d24
558 #define right_x_32_low d20
559 #define right_x_32_high d21
560 #define left_x_32_low d22
561 #define left_x_32_high d23
563 #define tmp_max_blocks d20
566 #define edges_dx_dy d2
567 #define edge_shifts d3
568 #define edge_shifts_64 q2
570 #define edges_xy_left d0
571 #define edges_xy_right d1
573 #define height_reciprocals d6
581 #define heights_b d12
582 #define edges_dx_dy_64 q10
584 #define edges_dx_dy_64_left d20
585 #define edges_dx_dy_64_right d21
588 #define setup_spans_prologue() \
589 stmdb sp!, { r4 - r11, lr }; \
592 ldrsh x_a, [v_a, #8]; \
593 ldrsh x_b, [v_b, #8]; \
594 ldrsh x_c, [v_c, #8]; \
595 ldrsh y_a, [v_a, #10]; \
596 ldrsh y_b, [v_b, #10]; \
597 ldrsh y_c, [v_c, #10]; \
599 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
600 vld1.32 { uvrg }, [temp]; \
601 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
602 vld1.32 { uvrg_dy }, [temp]; \
603 ldr reciprocal_table_ptr, [psx_gpu, #psx_gpu_reciprocal_table_ptr_offset]; \
605 vmov.u32 c_0x01, #0x01 \
607 #define setup_spans_load_b() \
608 ldr b, [psx_gpu, #psx_gpu_b_offset]; \
609 ldr b_dy, [psx_gpu, #psx_gpu_b_dy_offset] \
611 #define setup_spans_prologue_b() \
612 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
613 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
615 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
616 vmov.u16 c_0x0004, #0x0004; \
618 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
619 vmov.u16 c_0x0001, #0x0001; \
621 vld1.u16 { left_edge_low[], left_edge_high[] }, [temp]; \
622 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
624 vld1.u16 { right_edge_low[], right_edge_high[] }, [temp]; \
625 vadd.u16 right_edge, right_edge, c_0x0001; \
627 vmov.u16 c_0x0007, #0x0007; \
628 vmvn.u16 c_0xFFFE, #0x0001 \
631 #define compute_edge_delta_x2() \
632 ldr temp, [reciprocal_table_ptr, height, lsl #2]; \
634 vdup.u32 heights, height; \
635 vsub.u32 widths, x_ends, x_starts; \
637 vdup.u32 edge_shifts, temp; \
638 vsub.u32 heights_b, heights, c_0x01; \
639 vshr.u32 height_reciprocals, edge_shifts, #10; \
641 vmla.s32 heights_b, x_starts, heights; \
642 vbic.u16 edge_shifts, #0xE0; \
643 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
644 vmull.s32 edges_xy, heights_b, height_reciprocals \
647 #define height_reciprocal_alt r11
648 #define height_b_alt r12
650 #define compute_edge_delta_x3(start_c, height_a, height_b) \
651 vmov heights, height_a, height_b; \
652 ldr temp, [reciprocal_table_ptr, height_a, lsl #2]; \
653 vmov.u32 edge_shifts[0], temp; \
654 ldr temp, [reciprocal_table_ptr, height_b, lsl #2]; \
655 vmov.u32 edge_shifts[1], temp; \
656 ldr edge_shift_alt, [reciprocal_table_ptr, height_minor_b, lsl #2]; \
658 vsub.u32 widths, x_ends, x_starts; \
659 sub width_alt, x_c, start_c; \
661 vsub.u32 heights_b, heights, c_0x01; \
662 sub height_b_alt, height_minor_b, #1; \
664 vshr.u32 height_reciprocals, edge_shifts, #10; \
665 lsr height_reciprocal_alt, edge_shift_alt, #10; \
667 vmla.s32 heights_b, x_starts, heights; \
668 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
670 vbic.u16 edge_shifts, #0xE0; \
671 and edge_shift_alt, edge_shift_alt, #0x1F; \
673 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
674 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
676 vmull.s32 edges_xy, heights_b, height_reciprocals; \
677 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
680 #define setup_spans_adjust_y_up() \
681 vsub.u32 y_x4, y_x4, c_0x0004 \
683 #define setup_spans_adjust_y_down() \
684 vadd.u32 y_x4, y_x4, c_0x0004 \
686 #define setup_spans_adjust_interpolants_up() \
687 vsub.u32 uvrg, uvrg, uvrg_dy; \
690 #define setup_spans_adjust_interpolants_down() \
691 vadd.u32 uvrg, uvrg, uvrg_dy; \
695 #define setup_spans_clip_interpolants_increment() \
696 mla b, b_dy, clip, b; \
697 vmla.s32 uvrg, uvrg_dy, v_clip \
699 #define setup_spans_clip_interpolants_decrement() \
700 mls b, b_dy, clip, b; \
701 vmls.s32 uvrg, uvrg_dy, v_clip \
703 #define setup_spans_clip_alternate_yes() \
704 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
706 #define setup_spans_clip_alternate_no() \
708 #define setup_spans_clip(direction, alternate_active) \
709 vdup.u32 v_clip, clip; \
710 setup_spans_clip_alternate_##alternate_active(); \
711 setup_spans_clip_interpolants_##direction(); \
712 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
715 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
716 vmovl.s32 edge_shifts_64, edge_shifts; \
717 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
719 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
720 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
722 vmov left_x_low, edges_xy_##left_index; \
723 vmov right_x_low, edges_xy_##right_index; \
725 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
726 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
727 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
728 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
730 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
731 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
733 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
734 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
737 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
738 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
740 vdup.u16 y_mid_point, y_b; \
741 rsb temp, edge_shift_alt, #32; \
743 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
744 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
745 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
746 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
748 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
749 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
750 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
751 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
753 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
754 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
757 #define setup_spans_y_select_up() \
758 vclt.s16 alternate_select, y_x4, y_mid_point \
760 #define setup_spans_y_select_down() \
761 vcgt.s16 alternate_select, y_x4, y_mid_point \
764 #define setup_spans_alternate_select_left() \
765 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
767 #define setup_spans_alternate_select_right() \
768 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
771 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
772 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
773 vshrn.s64 left_x_32_low, left_x, #32; \
774 vshrn.s64 right_x_32_low, right_x, #32; \
776 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
777 vadd.u64 left_x, left_x, left_dx_dy; \
778 vadd.u64 right_x, right_x, right_dx_dy; \
780 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
781 vshrn.s64 left_x_32_high, left_x, #32; \
782 vshrn.s64 right_x_32_high, right_x, #32; \
784 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
785 vadd.u64 left_x, left_x, left_dx_dy; \
786 vadd.u64 right_x, right_x, right_dx_dy; \
788 vmovn.u32 alternate_x_16, alternate_x_32; \
789 setup_spans_y_select_##direction(); \
790 vmovn.u32 left_right_x_16_low, left_x_32; \
792 vmovn.u32 left_right_x_16_high, right_x_32; \
793 setup_spans_alternate_select_##alternate(); \
795 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
796 str b, [span_b_offset], #4; \
797 setup_spans_adjust_interpolants_##direction(); \
799 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
801 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
802 str b, [span_b_offset], #4; \
803 setup_spans_adjust_interpolants_##direction(); \
805 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
807 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
808 str b, [span_b_offset], #4; \
809 setup_spans_adjust_interpolants_##direction(); \
811 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
812 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
813 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
815 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
816 str b, [span_b_offset], #4; \
817 setup_spans_adjust_interpolants_##direction(); \
819 vmov.u16 tmp_max_blocks, #MAX_BLOCKS_PER_ROW; \
820 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
821 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
822 vmin.u16 left_right_x_16_high, left_right_x_16_high, tmp_max_blocks; \
824 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
826 setup_spans_adjust_y_##direction() \
829 #define setup_spans_set_x4_alternate_no(alternate, direction) \
830 vshrn.s64 left_x_32_low, left_x, #32; \
831 vshrn.s64 right_x_32_low, right_x, #32; \
833 vadd.u64 left_x, left_x, left_dx_dy; \
834 vadd.u64 right_x, right_x, right_dx_dy; \
836 vshrn.s64 left_x_32_high, left_x, #32; \
837 vshrn.s64 right_x_32_high, right_x, #32; \
839 vadd.u64 left_x, left_x, left_dx_dy; \
840 vadd.u64 right_x, right_x, right_dx_dy; \
842 vmovn.u32 left_right_x_16_low, left_x_32; \
843 vmovn.u32 left_right_x_16_high, right_x_32; \
845 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
846 str b, [span_b_offset], #4; \
847 setup_spans_adjust_interpolants_##direction(); \
849 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
851 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
852 str b, [span_b_offset], #4; \
853 setup_spans_adjust_interpolants_##direction(); \
855 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
857 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
858 str b, [span_b_offset], #4; \
859 setup_spans_adjust_interpolants_##direction(); \
861 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
862 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
863 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
865 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
866 str b, [span_b_offset], #4; \
867 setup_spans_adjust_interpolants_##direction(); \
869 vmov.u16 tmp_max_blocks, #MAX_BLOCKS_PER_ROW; \
870 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
871 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
872 vmin.u16 left_right_x_16_high, left_right_x_16_high, tmp_max_blocks; \
874 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
876 setup_spans_adjust_y_##direction() \
879 #define edge_adjust_low r11
880 #define edge_adjust_high r12
882 #define setup_spans_alternate_adjust_yes() \
883 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
884 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
885 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
887 #define setup_spans_alternate_adjust_no() \
890 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
891 setup_spans_alternate_adjust_##alternate_active(); \
892 setup_spans_load_b(); \
894 ldrsh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
895 subs y_c, y_c, temp; \
896 subgt height, height, y_c; \
897 addgt height, height, #1; \
899 ldrsh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
900 subs clip, temp, y_a; \
903 sub height, height, clip; \
904 add y_a, y_a, clip; \
905 setup_spans_clip(increment, alternate_active); \
911 orr temp, y_a, y_a, lsl #16; \
913 add temp, temp, #(1 << 16); \
914 movgt height, #512; \
916 add y_a, y_a, #(2 << 16); \
917 vmov y_x4, temp, y_a; \
919 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
921 setup_spans_prologue_b(); \
923 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
926 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
927 subs height, height, #4; \
931 ldr temp, [psx_gpu, #psx_gpu_hacks_active_offset]; \
932 tst temp, #(AHACK_TEXTURE_ADJ_U | AHACK_TEXTURE_ADJ_V); \
934 add temp, span_uvrg_offset, height, lsl #4; \
935 vldr uv, [temp, #(-16*2)]; \
936 vstr uv, [temp, #(-16)]; \
941 #define setup_spans_alternate_pre_increment_yes() \
942 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
943 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
945 #define setup_spans_alternate_pre_increment_no() \
948 #define setup_spans_up_decrement_yes() \
949 suble height, height, #1 \
951 #define setup_spans_up_decrement_no() \
954 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
955 setup_spans_alternate_adjust_##alternate_active(); \
956 setup_spans_load_b(); \
959 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
960 subs temp, temp, y_c; \
961 subgt height, height, temp; \
962 setup_spans_up_decrement_##alternate_active(); \
964 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
965 subs clip, y_a, temp; \
968 sub height, height, clip; \
969 sub y_a, y_a, clip; \
970 setup_spans_clip(decrement, alternate_active); \
976 orr temp, y_a, y_a, lsl #16; \
978 sub temp, temp, #(1 << 16); \
979 movgt height, #512; \
981 sub y_a, y_a, #(2 << 16); \
982 vmov y_x4, temp, y_a; \
984 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
986 setup_spans_alternate_pre_increment_##alternate_active(); \
987 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
989 setup_spans_adjust_interpolants_up(); \
990 setup_spans_prologue_b(); \
992 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
995 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
996 subs height, height, #4; \
1000 ldr temp, [psx_gpu, #psx_gpu_hacks_active_offset]; \
1001 tst temp, #AHACK_TEXTURE_ADJ_V; \
1003 add temp, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1004 vldr uv, [temp, #16]; \
1005 vstr uv, [temp, #0]; \
1010 #define setup_spans_epilogue() \
1011 restore_abi_regs(); \
1012 ldmia sp!, { r4 - r11, pc } \
1015 #define setup_spans_up_up(minor, major) \
1016 setup_spans_prologue(); \
1017 sub height_minor_a, y_a, y_b; \
1018 sub height_minor_b, y_b, y_c; \
1019 sub height, y_a, y_c; \
1021 vdup.u32 x_starts, x_a; \
1022 vmov x_ends, x_c, x_b; \
1024 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1025 setup_spans_up(major, minor, minor, yes); \
1026 setup_spans_epilogue() \
1028 function(setup_spans_up_left)
1029 setup_spans_up_up(left, right)
1031 function(setup_spans_up_right)
1032 setup_spans_up_up(right, left)
1034 #define setup_spans_down_down(minor, major) \
1035 setup_spans_prologue(); \
1036 sub height_minor_a, y_b, y_a; \
1037 sub height_minor_b, y_c, y_b; \
1038 sub height, y_c, y_a; \
1040 vdup.u32 x_starts, x_a; \
1041 vmov x_ends, x_c, x_b; \
1043 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1044 setup_spans_down(major, minor, minor, yes); \
1045 setup_spans_epilogue() \
1047 function(setup_spans_down_left)
1048 setup_spans_down_down(left, right)
1050 function(setup_spans_down_right)
1051 setup_spans_down_down(right, left)
1054 #define setup_spans_up_flat() \
1055 sub height, y_a, y_c; \
1057 compute_edge_delta_x2(); \
1058 setup_spans_up(left, right, none, no); \
1059 setup_spans_epilogue() \
1061 function(setup_spans_up_a)
1062 setup_spans_prologue()
1064 vmov x_starts, x_a, x_b
1065 vdup.u32 x_ends, x_c
1067 setup_spans_up_flat()
1069 function(setup_spans_up_b)
1070 setup_spans_prologue()
1072 vdup.u32 x_starts, x_a
1073 vmov x_ends, x_b, x_c
1075 setup_spans_up_flat()
1077 #define setup_spans_down_flat() \
1078 sub height, y_c, y_a; \
1080 compute_edge_delta_x2(); \
1081 setup_spans_down(left, right, none, no); \
1082 setup_spans_epilogue() \
1084 function(setup_spans_down_a)
1085 setup_spans_prologue()
1087 vmov x_starts, x_a, x_b
1088 vdup.u32 x_ends, x_c
1090 setup_spans_down_flat()
1092 function(setup_spans_down_b)
1093 setup_spans_prologue()
1095 vdup.u32 x_starts, x_a
1096 vmov x_ends, x_b, x_c
1098 setup_spans_down_flat()
1103 #define edges_xy_b q11
1104 #define edges_dx_dy_b d26
1105 #define edge_shifts_b d27
1106 #define edges_dx_dy_and_shifts_b q13
1107 #define height_increment d20
1109 #define edges_dx_dy_and_shifts q1
1111 #define edges_xy_b_left d22
1112 #define edges_xy_b_right d23
1114 #define setup_spans_up_down_load_edge_set_b() \
1115 vmov edges_xy, edges_xy_b; \
1116 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1119 function(setup_spans_up_down)
1120 setup_spans_prologue()
1122 // s32 middle_y = y_a;
1123 sub height_minor_a, y_a, y_b
1124 sub height_minor_b, y_c, y_a
1125 sub height_major, y_c, y_b
1127 vmov x_starts, x_a, x_c
1128 vdup.u32 x_ends, x_b
1130 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1133 vmov height_increment, temp, height_minor_b
1134 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1136 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1137 vmov edges_xy_b_right, edges_xy_right
1139 vmov edge_shifts_b, edge_shifts
1140 vmov.u32 edge_shifts_b[0], edge_shift_alt
1142 vneg.s32 edges_dx_dy_b, edges_dx_dy
1143 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1147 setup_spans_load_b()
1150 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1151 subs temp, temp, y_b
1152 subgt height_minor_a, height_minor_a, temp
1154 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1155 subs clip, y_a, temp
1158 sub height_minor_a, height_minor_a, clip
1160 setup_spans_clip(decrement, no)
1163 cmp height_minor_a, #0
1166 orr temp, y_a, y_a, lsl #16
1167 sub temp, temp, #(1 << 16)
1169 sub y_a, y_a, #(2 << 16)
1170 vmov y_x4, temp, y_a
1172 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1174 strh height_minor_a, [psx_gpu, #psx_gpu_num_spans_offset]
1176 setup_spans_adjust_edges_alternate_no(left, right);
1177 setup_spans_adjust_interpolants_up()
1178 setup_spans_up_down_load_edge_set_b()
1180 setup_spans_prologue_b()
1184 setup_spans_set_x4_alternate_no(none, up)
1185 subs height_minor_a, height_minor_a, #4
1188 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1189 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1190 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1193 add temp, psx_gpu, #psx_gpu_uvrg_offset
1194 vld1.32 { uvrg }, [temp]
1197 setup_spans_load_b()
1199 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1201 subgt height_minor_b, height_minor_b, y_c
1202 addgt height_minor_b, height_minor_b, #1
1204 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1205 subs clip, temp, y_a
1208 sub height_minor_b, height_minor_b, clip
1210 setup_spans_clip(increment, no)
1213 cmp height_minor_b, #0
1216 orr temp, y_a, y_a, lsl #16
1217 add temp, temp, #(1 << 16)
1219 add y_a, y_a, #(2 << 16)
1220 vmov y_x4, temp, y_a
1222 setup_spans_adjust_edges_alternate_no(left, right)
1224 ldrh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1225 add temp, temp, height_minor_b
1227 cmp temp, #MAX_SPANS
1230 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1233 setup_spans_set_x4_alternate_no(none, down)
1234 subs height_minor_b, height_minor_b, #4
1238 ldr temp, [psx_gpu, #psx_gpu_hacks_active_offset]
1239 tst temp, #(AHACK_TEXTURE_ADJ_U | AHACK_TEXTURE_ADJ_V)
1241 add temp, span_uvrg_offset, height, lsl #4
1242 vldr uv, [temp, #(-16*2)]
1243 vstr uv, [temp, #(-16)]
1246 setup_spans_epilogue()
1249 setup_spans_up_down_load_edge_set_b()
1250 setup_spans_prologue_b()
1254 // FIXME: overflow corner case
1255 sub temp, temp, height_minor_b
1256 bics height_minor_b, #3
1257 add temp, temp, height_minor_b
1258 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1262 #undef span_uvrg_offset
1263 #undef span_edge_data
1264 #undef span_b_offset
1269 #define num_spans r1
1270 #define span_uvrg_offset r2
1271 #define span_edge_data r3
1272 #define span_b_offset r4
1274 #define span_num_blocks r6
1278 #define dither_offset_ptr r10
1279 #define block_ptr_a r11
1281 #define num_blocks r14
1283 #define uvrg_dx_ptr r2
1284 #define texture_mask_ptr r3
1285 #define hacks_active r6
1286 #define dither_shift r8
1287 #define dither_row r10
1292 #define block_ptr_b r10
1294 #define block_span_ptr r10
1295 #define right_mask r8
1315 #define b_whole_8 d14
1316 #define fb_mask_ptrs d15
1329 #define u_whole_low d16
1330 #define u_whole_high d17
1331 #define v_whole_low d18
1332 #define v_whole_high d19
1333 #define r_whole_low d20
1334 #define r_whole_high d21
1335 #define g_whole_low d22
1336 #define g_whole_high d23
1337 #define b_whole_low d24
1338 #define b_whole_high d25
1343 #define u_whole_8 d26
1344 #define v_whole_8 d27
1345 #define u_whole_8b d24
1346 #define r_whole_8 d24
1347 #define g_whole_8 d25
1349 #define uv_whole_8 q13
1350 #define uv_whole_8b q14
1352 #define dither_offsets q14
1353 #define texture_mask q15
1354 #define texture_mask_u d30
1355 #define texture_mask_v d31
1357 #define dither_offsets_short d28
1361 #define block_span q10
1366 #define draw_mask q1
1367 #define draw_mask_edge q13
1368 #define test_mask q0
1374 #define setup_blocks_texture_swizzled() \
1375 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1376 vsli.u8 u_whole_8, v_whole_8, #4; \
1377 vsri.u8 v_whole_8, u_whole_8b, #4 \
1379 #define setup_blocks_texture_unswizzled() \
1381 #define setup_blocks_uv_adj_hack_textured(hacks_active) \
1382 tst hacks_active, #(AHACK_TEXTURE_ADJ_U | AHACK_TEXTURE_ADJ_V); \
1385 /* pushing odd num of regs here realigns our unaligned stack */ \
1386 vstr texture_mask_u, [r0, #psx_gpu_saved_tmp_offset]; \
1387 vstr texture_mask_v, [r0, #psx_gpu_saved_tmp_offset + 8]; \
1388 push { r0 - r4, EXTRA_UNSAVED_REGS r12, r14 }; \
1389 mov r12, span_uvrg_offset; \
1390 sub r1, block_ptr_a, #64; \
1391 mov r2, span_edge_data; \
1393 bl setup_blocks_uv_adj_hack; /* psx_gpu=r0 */ \
1394 pop { r0 - r4, EXTRA_UNSAVED_REGS r12, r14 }; \
1395 vldr texture_mask_u, [r0, #psx_gpu_saved_tmp_offset]; \
1396 vldr texture_mask_v, [r0, #psx_gpu_saved_tmp_offset + 8]; \
1398 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1402 #define setup_blocks_shaded_textured_builder(swizzling) \
1405 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1406 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1407 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1409 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1410 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1412 cmp num_spans, #0; \
1415 stmdb sp!, { r4 - r11, r14 }; \
1417 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1419 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
1420 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1422 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1423 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1425 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1426 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1428 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1429 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1431 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1434 vmov.u8 fb_mask_ptrs, #0; \
1436 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1437 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1439 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1440 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1442 cmp span_num_blocks, #0; \
1445 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1446 add num_blocks, span_num_blocks, num_blocks; \
1448 cmp num_blocks, #MAX_BLOCKS; \
1452 ldr b, [span_b_offset]; \
1453 add fb_ptr, fb_ptr, y, lsl #11; \
1455 vdup.u32 v_left_x, left_x; \
1458 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1459 add fb_ptr, fb_ptr, left_x, lsl #1; \
1461 mla b, b_dx, left_x, b; \
1462 and dither_shift, left_x, #0x03; \
1464 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1465 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1467 mov dither_shift, dither_shift, lsl #3; \
1468 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1471 subs span_num_blocks, span_num_blocks, #1; \
1473 mov dither_row, dither_row, ror dither_shift; \
1474 mov b_dx4, b_dx, lsl #2; \
1476 vdup.u32 dither_offsets_short, dither_row; \
1477 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1479 vdup.u32 b_block, b; \
1480 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1482 vdup.u32 u_block, uv[0]; \
1483 mov b_dx8, b_dx, lsl #3; \
1485 vdup.u32 v_block, uv[1]; \
1486 vdup.u32 r_block, rg[0]; \
1487 vdup.u32 g_block, rg[1]; \
1489 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1491 vadd.u32 u_block, u_block, block_span; \
1492 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1494 vadd.u32 v_block, v_block, block_span; \
1495 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1497 vadd.u32 r_block, r_block, block_span; \
1498 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1500 vadd.u32 g_block, g_block, block_span; \
1501 vld1.u32 { block_span }, [block_span_ptr, :128]; \
1503 vadd.u32 b_block, b_block, block_span; \
1504 add block_ptr_b, block_ptr_a, #16; \
1506 vshrn.u32 u_whole_low, u_block, #16; \
1507 vshrn.u32 v_whole_low, v_block, #16; \
1508 vshrn.u32 r_whole_low, r_block, #16; \
1509 vshrn.u32 g_whole_low, g_block, #16; \
1511 vdup.u32 dx4, uv_dx4[0]; \
1512 vshrn.u32 b_whole_low, b_block, #16; \
1514 vaddhn.u32 u_whole_high, u_block, dx4; \
1515 vdup.u32 dx4, uv_dx4[1]; \
1517 vaddhn.u32 v_whole_high, v_block, dx4; \
1518 vdup.u32 dx4, rg_dx4[0]; \
1520 vaddhn.u32 r_whole_high, r_block, dx4; \
1521 vdup.u32 dx4, rg_dx4[1]; \
1523 vaddhn.u32 g_whole_high, g_block, dx4; \
1524 vdup.u32 dx4, b_dx4; \
1526 vaddhn.u32 b_whole_high, b_block, dx4; \
1527 vdup.u32 dx8, uv_dx8[0]; \
1529 vadd.u32 u_block, u_block, dx8; \
1530 vdup.u32 dx8, uv_dx8[1]; \
1532 vadd.u32 v_block, v_block, dx8; \
1533 vdup.u32 dx8, rg_dx8[0]; \
1535 vadd.u32 r_block, r_block, dx8; \
1536 vdup.u32 dx8, rg_dx8[1]; \
1538 vadd.u32 g_block, g_block, dx8; \
1539 vdup.u32 dx8, b_dx8; \
1541 vadd.u32 b_block, b_block, dx8; \
1542 vmovn.u16 u_whole_8, u_whole; \
1544 vmovn.u16 v_whole_8, v_whole; \
1546 vmovn.u16 b_whole_8, b_whole; \
1548 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1550 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1551 setup_blocks_texture_##swizzling(); \
1553 vmovn.u16 r_whole_8, r_whole; \
1557 vmovn.u16 g_whole_8, g_whole; \
1558 vshrn.u32 u_whole_low, u_block, #16; \
1560 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1561 vshrn.u32 v_whole_low, v_block, #16; \
1563 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1564 vshrn.u32 r_whole_low, r_block, #16; \
1566 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1567 vshrn.u32 g_whole_low, g_block, #16; \
1569 vdup.u32 dx4, uv_dx4[0]; \
1570 vshrn.u32 b_whole_low, b_block, #16; \
1572 vaddhn.u32 u_whole_high, u_block, dx4; \
1573 vdup.u32 dx4, uv_dx4[1]; \
1575 vaddhn.u32 v_whole_high, v_block, dx4; \
1576 vdup.u32 dx4, rg_dx4[0]; \
1578 vaddhn.u32 r_whole_high, r_block, dx4; \
1579 vdup.u32 dx4, rg_dx4[1]; \
1581 vaddhn.u32 g_whole_high, g_block, dx4; \
1582 vdup.u32 dx4, b_dx4; \
1584 vaddhn.u32 b_whole_high, b_block, dx4; \
1585 vdup.u32 dx8, uv_dx8[0]; \
1587 vadd.u32 u_block, u_block, dx8; \
1588 vdup.u32 dx8, uv_dx8[1]; \
1590 vadd.u32 v_block, v_block, dx8; \
1591 vdup.u32 dx8, rg_dx8[0]; \
1593 vadd.u32 r_block, r_block, dx8; \
1594 vdup.u32 dx8, rg_dx8[1]; \
1596 vadd.u32 g_block, g_block, dx8; \
1597 vdup.u32 dx8, b_dx8; \
1599 vadd.u32 b_block, b_block, dx8; \
1600 vmovn.u16 u_whole_8, u_whole; \
1602 add fb_ptr, fb_ptr, #16; \
1603 vmovn.u16 v_whole_8, v_whole; \
1605 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1606 vmovn.u16 b_whole_8, b_whole; \
1610 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1611 subs span_num_blocks, span_num_blocks, #1; \
1613 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1614 setup_blocks_texture_##swizzling(); \
1616 vmovn.u16 r_whole_8, r_whole; \
1620 vmovn.u16 g_whole_8, g_whole; \
1621 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1623 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1624 vdup.u8 draw_mask, right_mask; \
1626 ldr hacks_active, [psx_gpu, #psx_gpu_hacks_active_offset]; \
1627 vmov.u32 fb_mask_ptrs[0], right_mask; \
1628 vtst.u16 draw_mask, draw_mask, test_mask; \
1629 vzip.u8 u_whole_8, v_whole_8; \
1631 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1632 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1633 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1634 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1635 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1637 setup_blocks_uv_adj_hack_textured(hacks_active); \
1640 add span_uvrg_offset, span_uvrg_offset, #16; \
1641 add span_b_offset, span_b_offset, #4; \
1643 add span_edge_data, span_edge_data, #8; \
1644 subs num_spans, num_spans, #1; \
1646 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1649 restore_abi_regs(); \
1650 ldmia sp!, { r4 - r11, pc }; \
1653 vstr texture_mask_u, [r0, #psx_gpu_saved_tmp_offset]; \
1654 vstr texture_mask_v, [r0, #psx_gpu_saved_tmp_offset + 8]; \
1655 /* pushing odd num of regs here realigns our unaligned stack */ \
1656 push { r0 - r3, EXTRA_UNSAVED_REGS r12 }; /* r14=num_blocks */ \
1657 bl flush_render_block_buffer; \
1658 pop { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
1659 vldr texture_mask_u, [r0, #psx_gpu_saved_tmp_offset]; \
1660 vldr texture_mask_v, [r0, #psx_gpu_saved_tmp_offset + 8]; \
1662 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1663 vmov.u8 fb_mask_ptrs, #0; \
1665 mov num_blocks, span_num_blocks; \
1666 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1670 setup_blocks_shaded_textured_builder(swizzled)
1671 setup_blocks_shaded_textured_builder(unswizzled)
1674 #define setup_blocks_unshaded_textured_builder(swizzling) \
1677 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1678 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1679 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1681 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1682 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1684 cmp num_spans, #0; \
1687 stmdb sp!, { r4 - r11, r14 }; \
1689 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1691 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1693 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1694 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1696 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1697 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1699 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1701 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1704 vmov.u8 fb_mask_ptrs, #0; \
1706 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1707 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1709 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1710 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1712 cmp span_num_blocks, #0; \
1715 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1716 add num_blocks, span_num_blocks, num_blocks; \
1718 cmp num_blocks, #MAX_BLOCKS; \
1722 add fb_ptr, fb_ptr, y, lsl #11; \
1724 vdup.u32 v_left_x, left_x; \
1727 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1728 add fb_ptr, fb_ptr, left_x, lsl #1; \
1730 and dither_shift, left_x, #0x03; \
1732 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1733 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1735 mov dither_shift, dither_shift, lsl #3; \
1736 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1739 subs span_num_blocks, span_num_blocks, #1; \
1741 mov dither_row, dither_row, ror dither_shift; \
1743 vdup.u32 dither_offsets_short, dither_row; \
1744 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1746 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1748 vdup.u32 u_block, uv[0]; \
1750 vdup.u32 v_block, uv[1]; \
1751 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1753 vadd.u32 u_block, u_block, block_span; \
1754 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1756 vadd.u32 v_block, v_block, block_span; \
1757 add block_ptr_b, block_ptr_a, #16; \
1759 vshrn.u32 u_whole_low, u_block, #16; \
1760 vshrn.u32 v_whole_low, v_block, #16; \
1762 vdup.u32 dx4, uv_dx4[0]; \
1764 vaddhn.u32 u_whole_high, u_block, dx4; \
1765 vdup.u32 dx4, uv_dx4[1]; \
1767 vaddhn.u32 v_whole_high, v_block, dx4; \
1768 vdup.u32 dx8, uv_dx8[0]; \
1770 vadd.u32 u_block, u_block, dx8; \
1771 vdup.u32 dx8, uv_dx8[1]; \
1773 vadd.u32 v_block, v_block, dx8; \
1774 vmovn.u16 u_whole_8, u_whole; \
1776 vmovn.u16 v_whole_8, v_whole; \
1779 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1781 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1782 setup_blocks_texture_##swizzling(); \
1787 vshrn.u32 u_whole_low, u_block, #16; \
1789 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1790 vshrn.u32 v_whole_low, v_block, #16; \
1792 add block_ptr_b, block_ptr_b, #32; \
1793 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1795 vdup.u32 dx4, uv_dx4[0]; \
1796 vaddhn.u32 u_whole_high, u_block, dx4; \
1797 vdup.u32 dx4, uv_dx4[1]; \
1799 vaddhn.u32 v_whole_high, v_block, dx4; \
1800 vdup.u32 dx8, uv_dx8[0]; \
1802 vadd.u32 u_block, u_block, dx8; \
1803 vdup.u32 dx8, uv_dx8[1]; \
1805 vadd.u32 v_block, v_block, dx8; \
1806 vmovn.u16 u_whole_8, u_whole; \
1808 add fb_ptr, fb_ptr, #16; \
1809 vmovn.u16 v_whole_8, v_whole; \
1811 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1814 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1815 subs span_num_blocks, span_num_blocks, #1; \
1817 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1818 setup_blocks_texture_##swizzling(); \
1823 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1825 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1826 vdup.u8 draw_mask, right_mask; \
1828 ldr hacks_active, [psx_gpu, #psx_gpu_hacks_active_offset]; \
1829 vmov.u32 fb_mask_ptrs[0], right_mask; \
1830 vtst.u16 draw_mask, draw_mask, test_mask; \
1831 vzip.u8 u_whole_8, v_whole_8; \
1833 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1834 add block_ptr_b, block_ptr_b, #32; \
1835 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1836 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1837 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1839 setup_blocks_uv_adj_hack_textured(hacks_active); \
1842 add span_uvrg_offset, span_uvrg_offset, #16; \
1843 add span_edge_data, span_edge_data, #8; \
1844 subs num_spans, num_spans, #1; \
1846 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1849 restore_abi_regs(); \
1850 ldmia sp!, { r4 - r11, pc }; \
1853 vstr texture_mask_u, [r0, #psx_gpu_saved_tmp_offset]; \
1854 vstr texture_mask_v, [r0, #psx_gpu_saved_tmp_offset + 8]; \
1855 push { r0 - r3, EXTRA_UNSAVED_REGS r12 }; /* r14=num_blocks */ \
1856 bl flush_render_block_buffer; \
1857 pop { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
1858 vldr texture_mask_u, [r0, #psx_gpu_saved_tmp_offset]; \
1859 vldr texture_mask_v, [r0, #psx_gpu_saved_tmp_offset + 8]; \
1861 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1862 vmov.u8 fb_mask_ptrs, #0; \
1864 mov num_blocks, span_num_blocks; \
1865 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1869 setup_blocks_unshaded_textured_builder(swizzled)
1870 setup_blocks_unshaded_textured_builder(unswizzled)
1875 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1876 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1877 veor.u32 draw_mask, draw_mask, draw_mask
1882 stmdb sp!, { r4 - r11, r14 }
1884 vld1.u32 { test_mask }, [psx_gpu, :128]
1886 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1888 ubfx color_r, color, #3, #5
1889 ubfx color_g, color, #11, #5
1890 ubfx color_b, color, #19, #5
1892 orr color, color_r, color_b, lsl #10
1893 orr color, color, color_g, lsl #5
1895 vdup.u16 colors, color
1897 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1898 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1900 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1901 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1904 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1905 ldrh y, [span_edge_data, #edge_data_y_offset]
1907 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1909 cmp span_num_blocks, #0
1912 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1913 add num_blocks, span_num_blocks, num_blocks
1915 cmp num_blocks, #MAX_BLOCKS
1919 add fb_ptr, fb_ptr, y, lsl #11
1922 add fb_ptr, fb_ptr, left_x, lsl #1
1925 subs span_num_blocks, span_num_blocks, #1
1927 add block_ptr_b, block_ptr_a, #16
1930 vmov.u32 fb_mask_ptrs[1], fb_ptr
1934 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_32
1935 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1936 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1938 add fb_ptr, fb_ptr, #16
1939 add block_ptr_b, block_ptr_b, #32
1943 vmov.u32 fb_mask_ptrs[1], fb_ptr
1944 subs span_num_blocks, span_num_blocks, #1
1949 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]
1951 vdup.u8 draw_mask_edge, right_mask
1952 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1954 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1955 vst1.u32 { draw_mask_edge }, [block_ptr_a, :128], c_32
1956 add block_ptr_b, block_ptr_b, #32
1957 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1960 add span_edge_data, span_edge_data, #8
1961 subs num_spans, num_spans, #1
1963 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1967 ldmia sp!, { r4 - r11, pc }
1970 vstr d4, [r0, #psx_gpu_saved_tmp_offset] /* colors */
1971 vstr d5, [r0, #psx_gpu_saved_tmp_offset + 8]
1972 push { r0 - r3, EXTRA_UNSAVED_REGS r12 }
1973 bl flush_render_block_buffer
1974 pop { r0 - r3, EXTRA_UNSAVED_REGS r12 }
1975 vldr d4, [r0, #psx_gpu_saved_tmp_offset]
1976 vldr d5, [r0, #psx_gpu_saved_tmp_offset + 8]
1978 vld1.u32 { test_mask }, [psx_gpu, :128]
1979 veor.u32 draw_mask, draw_mask, draw_mask
1981 mov num_blocks, span_num_blocks
1982 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1986 #define mask_msb_scalar r14
1988 #define msb_mask q15
1990 #define pixels_low d16
1992 #define msb_mask_low d30
1993 #define msb_mask_high d31
1998 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1999 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
2004 stmdb sp!, { r4 - r11, r14 }
2006 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
2008 ubfx color_r, color, #3, #5
2009 ubfx color_g, color, #11, #5
2011 ldrh mask_msb_scalar, [psx_gpu, #psx_gpu_mask_msb_offset]
2012 ubfx color_b, color, #19, #5
2014 orr color, color_r, color_b, lsl #10
2015 orr color, color, color_g, lsl #5
2016 orr color, color, mask_msb_scalar
2018 vdup.u16 colors, color
2020 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
2021 orr color, color, color, lsl #16
2025 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
2026 ldrh y, [span_edge_data, #edge_data_y_offset]
2028 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
2030 cmp span_num_blocks, #0
2033 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
2035 add fb_ptr, fb_ptr, y, lsl #11
2036 subs span_num_blocks, span_num_blocks, #1
2038 add fb_ptr, fb_ptr, left_x, lsl #1
2042 vst1.u32 { colors }, [fb_ptr]!
2043 subs span_num_blocks, span_num_blocks, #1
2048 ldrb right_mask, [span_edge_data, #edge_data_right_mask_offset]
2050 cmp right_mask, #0x0
2053 tst right_mask, #0xF
2054 streq color, [fb_ptr], #4
2055 moveq right_mask, right_mask, lsr #4
2056 streq color, [fb_ptr], #4
2058 tst right_mask, #0x3
2059 streq color, [fb_ptr], #4
2060 moveq right_mask, right_mask, lsr #2
2062 tst right_mask, #0x1
2063 strheq color, [fb_ptr]
2066 add span_edge_data, span_edge_data, #8
2067 subs num_spans, num_spans, #1
2070 ldmia sp!, { r4 - r11, pc }
2073 vst1.u32 { colors }, [fb_ptr]
2080 #define rg_dx_ptr r2
2098 #undef dither_offsets
2118 #define r_whole_low d6
2119 #define r_whole_high d7
2120 #define g_whole_low d8
2121 #define g_whole_high d9
2122 #define b_whole_low d10
2123 #define b_whole_high d11
2125 #define gb_whole_8 q6
2127 #define g_whole_8 d12
2128 #define b_whole_8 d13
2130 #define r_whole_8 d14
2142 #define block_span q5
2150 #define d128_0x7 q13
2154 #define dither_offsets q14
2155 #define draw_mask q15
2157 #define dither_offsets_low d28
2160 #define test_mask q10
2163 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2164 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2165 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2167 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2168 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2169 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2171 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2173 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2176 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2179 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2180 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2181 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2183 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2185 cmp num_spans, #0; \
2188 stmdb sp!, { r4 - r11, r14 }; \
2190 vshl.u32 rg_dx4, rg_dx, #2; \
2192 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2193 vshl.u32 rg_dx8, rg_dx, #3; \
2195 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2197 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2198 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2200 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2201 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2203 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2204 vmov.u8 d64_1, #1; \
2206 vmov.u8 d128_4, #4; \
2207 vmov.u8 d64_128, #128; \
2209 vmov.u8 d128_0x7, #0x7; \
2212 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2213 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2215 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2216 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2218 cmp span_num_blocks, #0; \
2221 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2222 add num_blocks, span_num_blocks, num_blocks; \
2224 cmp num_blocks, #MAX_BLOCKS; \
2228 ldr b, [span_b_offset]; \
2229 add fb_ptr, fb_ptr, y, lsl #11; \
2231 vdup.u32 v_left_x, left_x; \
2234 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2235 add fb_ptr, fb_ptr, left_x, lsl #1; \
2237 mla b, b_dx, left_x, b; \
2238 and dither_shift, left_x, #0x03; \
2240 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2241 vshr.u32 rg_dx, rg_dx4, #2; \
2243 mov dither_shift, dither_shift, lsl #3; \
2244 vmla.u32 rg, rg_dx, v_left_x; \
2247 subs span_num_blocks, span_num_blocks, #1; \
2249 mov dither_row, dither_row, ror dither_shift; \
2250 mov b_dx4, b_dx, lsl #2; \
2252 vdup.u32 dither_offsets, dither_row; \
2253 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2255 vdup.u32 b_block, b; \
2256 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2258 mov b_dx8, b_dx, lsl #3; \
2259 vdup.u32 r_block, rg[0]; \
2260 vdup.u32 g_block, rg[1]; \
2262 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2264 vadd.u32 r_block, r_block, block_span; \
2265 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2267 vadd.u32 g_block, g_block, block_span; \
2268 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2270 vadd.u32 b_block, b_block, block_span; \
2271 add block_ptr_b, block_ptr_a, #16; \
2273 vshrn.u32 r_whole_low, r_block, #16; \
2274 vshrn.u32 g_whole_low, g_block, #16; \
2275 vshrn.u32 b_whole_low, b_block, #16; \
2276 vdup.u32 dx4, rg_dx4[0]; \
2278 vaddhn.u32 r_whole_high, r_block, dx4; \
2279 vdup.u32 dx4, rg_dx4[1]; \
2281 vaddhn.u32 g_whole_high, g_block, dx4; \
2282 vdup.u32 dx4, b_dx4; \
2284 vaddhn.u32 b_whole_high, b_block, dx4; \
2285 vdup.u32 dx8, rg_dx8[0]; \
2287 vadd.u32 r_block, r_block, dx8; \
2288 vdup.u32 dx8, rg_dx8[1]; \
2290 vadd.u32 g_block, g_block, dx8; \
2291 vdup.u32 dx8, b_dx8; \
2293 vadd.u32 b_block, b_block, dx8; \
2295 vmovn.u16 r_whole_8, r_whole; \
2296 vmovn.u16 g_whole_8, g_whole; \
2297 vmovn.u16 b_whole_8, b_whole; \
2300 veor.u32 draw_mask, draw_mask, draw_mask; \
2303 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2304 vshrn.u32 r_whole_low, r_block, #16; \
2306 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2307 vshrn.u32 g_whole_low, g_block, #16; \
2309 vshrn.u32 b_whole_low, b_block, #16; \
2310 str fb_ptr, [block_ptr_a, #44]; \
2312 vdup.u32 dx4, rg_dx4[0]; \
2313 vshr.u8 r_whole_8, r_whole_8, #3; \
2314 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2316 vaddhn.u32 r_whole_high, r_block, dx4; \
2317 vdup.u32 dx4, rg_dx4[1]; \
2319 vaddhn.u32 g_whole_high, g_block, dx4; \
2320 vdup.u32 dx4, b_dx4; \
2322 vaddhn.u32 b_whole_high, b_block, dx4; \
2323 vdup.u32 dx8, rg_dx8[0]; \
2325 vmull.u8 pixels, r_whole_8, d64_1; \
2326 vmlal.u8 pixels, g_whole_8, d64_4; \
2327 vmlal.u8 pixels, b_whole_8, d64_128; \
2329 vadd.u32 r_block, r_block, dx8; \
2330 vdup.u32 dx8, rg_dx8[1]; \
2332 vadd.u32 g_block, g_block, dx8; \
2333 vdup.u32 dx8, b_dx8; \
2335 vadd.u32 b_block, b_block, dx8; \
2336 add fb_ptr, fb_ptr, #16; \
2338 vmovn.u16 r_whole_8, r_whole; \
2339 vmovn.u16 g_whole_8, g_whole; \
2340 vmovn.u16 b_whole_8, b_whole; \
2342 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2343 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2347 subs span_num_blocks, span_num_blocks, #1; \
2351 str fb_ptr, [block_ptr_a, #44]; \
2352 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2354 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2355 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2357 vshr.u8 r_whole_8, r_whole_8, #3; \
2358 vdup.u8 draw_mask, right_mask; \
2360 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2361 vld1.u32 { test_mask }, [psx_gpu, :128]; \
2363 vtst.u16 draw_mask, draw_mask, test_mask; \
2365 vmull.u8 pixels, r_whole_8, d64_1; \
2366 vmlal.u8 pixels, g_whole_8, d64_4; \
2367 vmlal.u8 pixels, b_whole_8, d64_128; \
2369 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2370 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2373 add span_uvrg_offset, span_uvrg_offset, #16; \
2374 add span_b_offset, span_b_offset, #4; \
2376 add span_edge_data, span_edge_data, #8; \
2377 subs num_spans, num_spans, #1; \
2379 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2382 restore_abi_regs(); \
2383 pop { r4 - r11, pc }; \
2386 vstr rg_dx4, [r0, #psx_gpu_saved_tmp_offset]; \
2387 push { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
2388 bl flush_render_block_buffer; \
2389 pop { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
2390 vldr rg_dx4, [r0, #psx_gpu_saved_tmp_offset]; \
2392 vmov.u8 d64_1, #1; \
2393 vmov.u8 d128_4, #4; \
2394 vmov.u8 d64_128, #128; \
2395 vmov.u8 d128_0x7, #0x7; \
2397 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2399 mov num_blocks, span_num_blocks; \
2400 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2404 setup_blocks_shaded_untextured_indirect_builder(undithered)
2405 setup_blocks_shaded_untextured_indirect_builder(dithered)
2410 #define mask_msb_ptr r14
2412 #define draw_mask q0
2413 #define pixels_low d16
2414 #define pixels_high d17
2418 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2421 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2422 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2423 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2425 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2427 cmp num_spans, #0; \
2430 stmdb sp!, { r4 - r11, r14 }; \
2432 vshl.u32 rg_dx4, rg_dx, #2; \
2434 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2435 vshl.u32 rg_dx8, rg_dx, #3; \
2437 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2438 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2440 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2441 vmov.u8 d64_1, #1; \
2443 vmov.u8 d128_4, #4; \
2444 vmov.u8 d64_128, #128; \
2446 vmov.u8 d128_0x7, #0x7; \
2447 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2448 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
2451 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2452 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2454 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2455 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2457 cmp span_num_blocks, #0; \
2460 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2461 add fb_ptr, fb_ptr, y, lsl #11; \
2463 ldr b, [span_b_offset]; \
2464 vdup.u32 v_left_x, left_x; \
2467 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2468 add fb_ptr, fb_ptr, left_x, lsl #1; \
2470 mla b, b_dx, left_x, b; \
2471 and dither_shift, left_x, #0x03; \
2473 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2474 vshr.u32 rg_dx, rg_dx4, #2; \
2476 mov dither_shift, dither_shift, lsl #3; \
2477 vmla.u32 rg, rg_dx, v_left_x; \
2479 subs span_num_blocks, span_num_blocks, #1; \
2481 mov dither_row, dither_row, ror dither_shift; \
2482 mov b_dx4, b_dx, lsl #2; \
2484 vdup.u32 dither_offsets, dither_row; \
2485 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2487 vdup.u32 b_block, b; \
2488 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2490 mov b_dx8, b_dx, lsl #3; \
2491 vdup.u32 r_block, rg[0]; \
2492 vdup.u32 g_block, rg[1]; \
2494 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2496 vadd.u32 r_block, r_block, block_span; \
2497 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2499 vadd.u32 g_block, g_block, block_span; \
2500 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2502 vadd.u32 b_block, b_block, block_span; \
2503 add block_ptr_b, block_ptr_a, #16; \
2505 vshrn.u32 r_whole_low, r_block, #16; \
2506 vshrn.u32 g_whole_low, g_block, #16; \
2507 vshrn.u32 b_whole_low, b_block, #16; \
2508 vdup.u32 dx4, rg_dx4[0]; \
2510 vaddhn.u32 r_whole_high, r_block, dx4; \
2511 vdup.u32 dx4, rg_dx4[1]; \
2513 vaddhn.u32 g_whole_high, g_block, dx4; \
2514 vdup.u32 dx4, b_dx4; \
2516 vaddhn.u32 b_whole_high, b_block, dx4; \
2517 vdup.u32 dx8, rg_dx8[0]; \
2519 vadd.u32 r_block, r_block, dx8; \
2520 vdup.u32 dx8, rg_dx8[1]; \
2522 vadd.u32 g_block, g_block, dx8; \
2523 vdup.u32 dx8, b_dx8; \
2525 vadd.u32 b_block, b_block, dx8; \
2527 vmovn.u16 r_whole_8, r_whole; \
2528 vmovn.u16 g_whole_8, g_whole; \
2529 vmovn.u16 b_whole_8, b_whole; \
2534 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2535 vshrn.u32 r_whole_low, r_block, #16; \
2537 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2538 vshrn.u32 g_whole_low, g_block, #16; \
2540 vshrn.u32 b_whole_low, b_block, #16; \
2542 vdup.u32 dx4, rg_dx4[0]; \
2543 vshr.u8 r_whole_8, r_whole_8, #3; \
2544 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2546 vaddhn.u32 r_whole_high, r_block, dx4; \
2547 vdup.u32 dx4, rg_dx4[1]; \
2549 vmov pixels, msb_mask; \
2550 vaddhn.u32 g_whole_high, g_block, dx4; \
2551 vdup.u32 dx4, b_dx4; \
2553 vaddhn.u32 b_whole_high, b_block, dx4; \
2554 vdup.u32 dx8, rg_dx8[0]; \
2556 vmlal.u8 pixels, r_whole_8, d64_1; \
2557 vmlal.u8 pixels, g_whole_8, d64_4; \
2558 vmlal.u8 pixels, b_whole_8, d64_128; \
2560 vadd.u32 r_block, r_block, dx8; \
2561 vdup.u32 dx8, rg_dx8[1]; \
2563 vadd.u32 g_block, g_block, dx8; \
2564 vdup.u32 dx8, b_dx8; \
2566 vadd.u32 b_block, b_block, dx8; \
2568 vmovn.u16 r_whole_8, r_whole; \
2569 vmovn.u16 g_whole_8, g_whole; \
2570 vmovn.u16 b_whole_8, b_whole; \
2572 vst1.u32 { pixels }, [fb_ptr]!; \
2573 subs span_num_blocks, span_num_blocks, #1; \
2577 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2579 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2580 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2582 vshr.u8 r_whole_8, r_whole_8, #3; \
2583 rbit right_mask, right_mask; \
2584 vmov pixels, msb_mask; \
2585 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2586 clz right_mask, right_mask; \
2588 vmlal.u8 pixels, r_whole_8, d64_1; \
2589 vmlal.u8 pixels, g_whole_8, d64_4; \
2590 vmlal.u8 pixels, b_whole_8, d64_128; \
2592 JT_OP_REL(100f, right_mask, temp); \
2593 JT_OP(ldr pc, [pc, right_mask, lsl #2]); \
2597 .word JTE(100b, 4f); \
2598 .word JTE(100b, 5f); \
2599 .word JTE(100b, 6f); \
2600 .word JTE(100b, 7f); \
2601 .word JTE(100b, 8f); \
2602 .word JTE(100b, 9f); \
2603 .word JTE(100b, 10f); \
2604 .word JTE(100b, 11f); \
2607 vst1.u16 { pixels_low[0] }, [fb_ptr]; \
2611 vst1.u32 { pixels_low[0] }, [fb_ptr]; \
2615 vst1.u32 { pixels_low[0] }, [fb_ptr]!; \
2616 vst1.u16 { pixels_low[2] }, [fb_ptr]; \
2620 vst1.u32 { pixels_low }, [fb_ptr]; \
2624 vst1.u32 { pixels_low }, [fb_ptr]!; \
2625 vst1.u16 { pixels_high[0] }, [fb_ptr]; \
2629 vst1.u32 { pixels_low }, [fb_ptr]!; \
2630 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2634 vst1.u32 { pixels_low }, [fb_ptr]!; \
2635 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2636 vst1.u16 { pixels_high[2] }, [fb_ptr]; \
2640 vst1.u32 { pixels }, [fb_ptr]; \
2644 add span_uvrg_offset, span_uvrg_offset, #16; \
2645 add span_b_offset, span_b_offset, #4; \
2647 add span_edge_data, span_edge_data, #8; \
2648 subs num_spans, num_spans, #1; \
2652 restore_abi_regs(); \
2653 ldmia sp!, { r4 - r11, pc } \
2655 setup_blocks_shaded_untextured_direct_builder(undithered)
2656 setup_blocks_shaded_untextured_direct_builder(dithered)
2665 #define block_ptr r1
2666 #define num_blocks r2
2679 #define texture_ptr r11
2693 #define pixels_d r10
2697 #define clut_ptr r12
2698 #define current_texture_mask r5
2699 #define dirty_textures_mask r6
2703 #define clut_low_a d2
2704 #define clut_low_b d3
2705 #define clut_high_a d4
2706 #define clut_high_b d5
2711 #define texels_low d6
2712 #define texels_high d7
2716 function(texture_blocks_untextured)
2722 function(texture_blocks_4bpp)
2723 stmdb sp!, { r3 - r11, r14 }
2724 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2726 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2727 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2729 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2730 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]
2732 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2733 vuzp.u8 clut_a, clut_b
2735 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
2736 tst dirty_textures_mask, current_texture_mask
2742 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2744 uxtah uv_0, texture_ptr, uv_01
2745 uxtah uv_1, texture_ptr, uv_01, ror #16
2747 uxtah uv_2, texture_ptr, uv_23
2748 uxtah uv_3, texture_ptr, uv_23, ror #16
2750 uxtah uv_4, texture_ptr, uv_45
2751 ldrb pixel_0, [uv_0]
2753 uxtah uv_5, texture_ptr, uv_45, ror #16
2754 ldrb pixel_1, [uv_1]
2756 uxtah uv_6, texture_ptr, uv_67
2757 ldrb pixel_2, [uv_2]
2759 uxtah uv_7, texture_ptr, uv_67, ror #16
2760 ldrb pixel_3, [uv_3]
2762 ldrb pixel_4, [uv_4]
2763 subs num_blocks, num_blocks, #1
2765 ldrb pixel_5, [uv_5]
2766 orr pixels_a, pixel_0, pixel_1, lsl #8
2768 ldrb pixel_6, [uv_6]
2769 orr pixels_b, pixel_4, pixel_5, lsl #8
2771 ldrb pixel_7, [uv_7]
2772 orr pixels_a, pixels_a, pixel_2, lsl #16
2774 orr pixels_b, pixels_b, pixel_6, lsl #16
2775 orr pixels_a, pixels_a, pixel_3, lsl #24
2777 orr pixels_b, pixels_b, pixel_7, lsl #24
2778 vmov texels, pixels_a, pixels_b
2780 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2781 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2783 vst2.u8 { texels_low, texels_high }, [block_ptr, :128], c_64
2786 ldmia sp!, { r3 - r11, pc }
2789 stmdb sp!, { r1 - r2 }
2790 bl update_texture_4bpp_cache
2793 ldmia sp!, { r1 - r2 }
2799 function(texture_blocks_8bpp)
2800 push { r4 - r11, lr }
2801 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2803 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2804 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2806 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2807 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2809 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]
2810 tst dirty_textures_mask, current_texture_mask
2816 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2818 uxtah uv_0, texture_ptr, uv_01
2819 uxtah uv_1, texture_ptr, uv_01, ror #16
2821 uxtah uv_2, texture_ptr, uv_23
2822 uxtah uv_3, texture_ptr, uv_23, ror #16
2824 uxtah uv_4, texture_ptr, uv_45
2825 ldrb pixel_0, [uv_0]
2827 uxtah uv_5, texture_ptr, uv_45, ror #16
2828 ldrb pixel_1, [uv_1]
2830 uxtah uv_6, texture_ptr, uv_67
2831 ldrb pixel_2, [uv_2]
2833 uxtah uv_7, texture_ptr, uv_67, ror #16
2834 ldrb pixel_3, [uv_3]
2836 ldrb pixel_4, [uv_4]
2837 add pixel_0, pixel_0, pixel_0
2839 ldrb pixel_5, [uv_5]
2840 add pixel_1, pixel_1, pixel_1
2842 ldrb pixel_6, [uv_6]
2843 add pixel_2, pixel_2, pixel_2
2845 ldrb pixel_7, [uv_7]
2846 add pixel_3, pixel_3, pixel_3
2848 ldrh pixel_0, [clut_ptr, pixel_0]
2849 add pixel_4, pixel_4, pixel_4
2851 ldrh pixel_1, [clut_ptr, pixel_1]
2852 add pixel_5, pixel_5, pixel_5
2854 ldrh pixel_2, [clut_ptr, pixel_2]
2855 add pixel_6, pixel_6, pixel_6
2857 ldrh pixel_3, [clut_ptr, pixel_3]
2858 add pixel_7, pixel_7, pixel_7
2860 ldrh pixel_4, [clut_ptr, pixel_4]
2861 orr pixels_a, pixel_0, pixel_1, lsl #16
2863 ldrh pixel_5, [clut_ptr, pixel_5]
2864 orr pixels_c, pixel_2, pixel_3, lsl #16
2866 ldrh pixel_6, [clut_ptr, pixel_6]
2867 subs num_blocks, num_blocks, #1
2869 ldrh pixel_7, [clut_ptr, pixel_7]
2870 orr pixels_b, pixel_4, pixel_5, lsl #16
2872 orr pixels_d, pixel_6, pixel_7, lsl #16
2873 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2875 add block_ptr, block_ptr, #64
2878 pop { r4 - r11, pc }
2881 /* pushing odd num of regs here realigns our unaligned stack */
2882 push { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2883 bl update_texture_8bpp_cache
2884 pop { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2914 #define block_ptr r1
2915 #define num_blocks r2
2959 #define texture_ptr r12
2964 function(texture_blocks_16bpp)
2965 stmdb sp!, { r3 - r11, r14 }
2966 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2968 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2969 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2972 ldrh uv_0, [block_ptr]
2973 subs num_blocks, num_blocks, #1
2975 ldrh uv_1, [block_ptr, #2]
2977 and v_0, uv_0, #0xFF00
2978 and v_1, uv_1, #0xFF00
2980 and u_0, uv_0, #0xFF
2981 and u_1, uv_1, #0xFF
2983 add uv_0, u_0, v_0, lsl #2
2984 ldrh uv_2, [block_ptr, #4]
2986 add uv_1, u_1, v_1, lsl #2
2987 ldrh uv_3, [block_ptr, #6]
2989 add uv_0, uv_0, uv_0
2990 add uv_1, uv_1, uv_1
2992 and v_2, uv_2, #0xFF00
2993 and v_3, uv_3, #0xFF00
2995 and u_2, uv_2, #0xFF
2996 and u_3, uv_3, #0xFF
2998 add uv_2, u_2, v_2, lsl #2
2999 ldrh uv_4, [block_ptr, #8]
3001 add uv_3, u_3, v_3, lsl #2
3002 ldrh uv_5, [block_ptr, #10]
3004 add uv_2, uv_2, uv_2
3005 add uv_3, uv_3, uv_3
3007 and v_4, uv_4, #0xFF00
3008 and v_5, uv_5, #0xFF00
3010 and u_4, uv_4, #0xFF
3011 and u_5, uv_5, #0xFF
3013 add uv_4, u_4, v_4, lsl #2
3014 ldrh uv_6, [block_ptr, #12]
3016 add uv_5, u_5, v_5, lsl #2
3017 ldrh uv_7, [block_ptr, #14]
3019 add uv_4, uv_4, uv_4
3020 ldrh pixel_0, [texture_ptr, uv_0]
3022 add uv_5, uv_5, uv_5
3023 ldrh pixel_1, [texture_ptr, uv_1]
3025 and v_6, uv_6, #0xFF00
3026 ldrh pixel_2, [texture_ptr, uv_2]
3028 and v_7, uv_7, #0xFF00
3029 ldrh pixel_3, [texture_ptr, uv_3]
3031 and u_6, uv_6, #0xFF
3032 ldrh pixel_4, [texture_ptr, uv_4]
3034 and u_7, uv_7, #0xFF
3035 ldrh pixel_5, [texture_ptr, uv_5]
3037 add uv_6, u_6, v_6, lsl #2
3038 add uv_7, u_7, v_7, lsl #2
3040 add uv_6, uv_6, uv_6
3041 add uv_7, uv_7, uv_7
3043 orr pixels_a, pixel_0, pixel_1, lsl #16
3044 orr pixels_b, pixel_2, pixel_3, lsl #16
3046 ldrh pixel_6, [texture_ptr, uv_6]
3047 orr pixels_c, pixel_4, pixel_5, lsl #16
3049 ldrh pixel_7, [texture_ptr, uv_7]
3050 orr pixels_d, pixel_6, pixel_7, lsl #16
3052 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
3053 add block_ptr, block_ptr, #64
3057 ldmia sp!, { r3 - r11, pc }
3072 #undef msb_mask_high
3080 #define num_blocks r1
3081 #define color_ptr r2
3082 #define colors_scalar r2
3083 #define colors_scalar_compare r3
3084 #define mask_msb_ptr r2
3086 #define block_ptr_load_a r0
3087 #define block_ptr_store r3
3088 #define block_ptr_load_b r12
3093 #define draw_mask_bits_scalar r5
3095 #define d128_0x07 q0
3096 #define d128_0x1F q1
3097 #define d128_0x8000 q2
3098 #define test_mask q3
3100 #define colors_rg q5
3101 #define colors_b_dm_bits q6
3102 #define texels_rg q7
3105 #define pixels_b q10
3107 #define zero_mask q4
3108 #define draw_mask q12
3109 #define msb_mask q13
3111 #define fb_pixels q8
3113 #define pixels_gb_low q9
3115 #define colors_r d10
3116 #define colors_g d11
3117 #define colors_b d12
3118 #define draw_mask_bits d13
3119 #define texels_r d14
3120 #define texels_g d15
3121 #define pixels_r_low d16
3122 #define pixels_g_low d18
3123 #define pixels_b_low d19
3124 #define msb_mask_low d26
3125 #define msb_mask_high d27
3130 #define texels_b d31
3132 #define shade_blocks_textured_modulated_prologue_indirect() \
3134 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3136 #define shade_blocks_textured_modulated_prologue_direct() \
3137 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3138 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] \
3141 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3143 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3144 ldr colors_scalar, [psx_gpu, #psx_gpu_triangle_color_offset]; \
3145 movw colors_scalar_compare, #0x8080; \
3147 movt colors_scalar_compare, #0x80; \
3148 cmp colors_scalar, colors_scalar_compare; \
3149 beq shade_blocks_textured_unmodulated_##target \
3151 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3153 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3154 shade_blocks_textured_false_modulation_check_##dithering(target); \
3155 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3156 vld1.u32 { colors_r[] }, [color_ptr, :32]; \
3157 vdup.u8 colors_g, colors_r[1]; \
3158 vdup.u8 colors_b, colors_r[2]; \
3159 vdup.u8 colors_r, colors_r[0] \
3162 #define shade_blocks_textured_modulated_load_dithered(target) \
3163 vld1.u32 { target }, [block_ptr_load_b, :128] \
3165 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3166 vld1.u32 { target }, [block_ptr_load_b, :128], c_32 \
3168 #define shade_blocks_textured_modulated_load_undithered(target) \
3170 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3171 add block_ptr_load_b, block_ptr_load_b, #32 \
3173 #define shade_blocks_textured_modulate_dithered(channel) \
3174 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3176 #define shade_blocks_textured_modulate_undithered(channel) \
3177 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3180 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3181 vst1.u32 { draw_mask }, [block_ptr_store, :128]! \
3183 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3184 ldr fb_ptr, [block_ptr_load_b, #(offset - 64)]; \
3185 vld1.u32 { fb_pixels }, [fb_ptr]; \
3186 vbit.u16 pixels, fb_pixels, draw_mask \
3188 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3189 vst1.u32 { pixels }, [block_ptr_store, :128], c_48 \
3191 #define shade_blocks_textured_modulated_store_pixels_direct() \
3192 vst1.u32 { pixels }, [fb_ptr] \
3195 #define shade_blocks_textured_modulated_load_rg_shaded() \
3196 vld1.u32 { colors_r, colors_g }, [block_ptr_load_b, :128], c_32 \
3198 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3199 add block_ptr_load_b, block_ptr_load_b, #32 \
3201 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3202 vld1.u32 { colors_b, draw_mask_bits }, [block_ptr_load_a, :128], c_32 \
3204 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3205 ldr draw_mask_bits_scalar, [block_ptr_load_a, #8]; \
3206 add block_ptr_load_a, block_ptr_load_a, #32 \
3208 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3209 vdup.u16 draw_mask, draw_mask_bits[0] \
3211 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3212 vdup.u16 draw_mask, draw_mask_bits_scalar \
3215 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3217 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3218 vorr.u16 pixels, pixels, msb_mask \
3221 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3224 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3226 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3227 stmdb sp!, { r4 - r5, lr }; \
3228 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3230 vld1.u32 { test_mask }, [psx_gpu, :128]; \
3232 shade_blocks_textured_modulated_prologue_##target(); \
3234 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3237 add block_ptr_load_b, block_ptr_load_a, #16; \
3238 vmov.u8 d64_1, #1; \
3239 vmov.u8 d64_4, #4; \
3240 vmov.u8 d64_128, #128; \
3242 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3243 vmov.u8 d128_0x07, #0x07; \
3245 shade_blocks_textured_modulated_load_rg_##shading(); \
3246 vmov.u8 d128_0x1F, #0x1F; \
3248 shade_blocks_textured_modulated_load_bdm_##shading(); \
3249 vmov.u16 d128_0x8000, #0x8000; \
3251 vmovn.u16 texels_r, texels; \
3252 vshrn.u16 texels_g, texels, #5; \
3254 vshrn.u16 texels_b, texels, #7; \
3255 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3257 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3258 vtst.u16 draw_mask, draw_mask, test_mask; \
3260 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3261 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3263 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3264 vshr.u8 texels_b, texels_b, #3; \
3266 shade_blocks_textured_modulate_##dithering(r); \
3267 shade_blocks_textured_modulate_##dithering(g); \
3268 shade_blocks_textured_modulate_##dithering(b); \
3270 vand.u16 pixels, texels, d128_0x8000; \
3271 vceq.u16 zero_mask, texels, #0; \
3273 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3274 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3275 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3277 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3278 vorr.u16 draw_mask, draw_mask, zero_mask; \
3279 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3280 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3282 subs num_blocks, num_blocks, #1; \
3288 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3289 shade_blocks_textured_modulated_load_rg_##shading(); \
3290 vshrn.u16 texels_g, texels, #5; \
3292 shade_blocks_textured_modulated_load_bdm_##shading(); \
3293 vshrn.u16 texels_b, texels, #7; \
3295 pld [block_ptr_load_a]; \
3296 vmovn.u16 texels_r, texels; \
3297 vmlal.u8 pixels, pixels_r_low, d64_1; \
3299 vmlal.u8 pixels, pixels_g_low, d64_4; \
3300 vmlal.u8 pixels, pixels_b_low, d64_128; \
3301 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3303 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3304 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3306 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3307 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3309 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3310 vtst.u16 draw_mask, draw_mask, test_mask; \
3312 shade_blocks_textured_modulated_store_pixels_##target(); \
3313 vshr.u8 texels_b, texels_b, #3; \
3315 shade_blocks_textured_modulate_##dithering(r); \
3316 shade_blocks_textured_modulate_##dithering(g); \
3317 shade_blocks_textured_modulate_##dithering(b); \
3319 vand.u16 pixels, texels, d128_0x8000; \
3320 vceq.u16 zero_mask, texels, #0; \
3322 subs num_blocks, num_blocks, #1; \
3324 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3325 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3326 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3328 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3329 vorr.u16 draw_mask, draw_mask, zero_mask; \
3330 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3331 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3336 vmlal.u8 pixels, pixels_r_low, d64_1; \
3337 vmlal.u8 pixels, pixels_g_low, d64_4; \
3338 vmlal.u8 pixels, pixels_b_low, d64_128; \
3340 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3341 shade_blocks_textured_modulated_store_pixels_##target(); \
3343 ldmia sp!, { r4 - r5, lr }; \
3344 restore_abi_regs(); \
3348 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3349 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3350 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3351 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3353 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3354 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3355 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3356 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3374 #undef msb_mask_high
3377 #define num_blocks r1
3378 #define mask_msb_ptr r2
3379 #define color_ptr r3
3381 #define block_ptr_load r0
3382 #define draw_mask_store_ptr r3
3383 #define draw_mask_bits_ptr r12
3384 #define draw_mask_ptr r12
3385 #define pixel_store_ptr r14
3387 #define fb_ptr_cmp r4
3390 #define fb_ptr_next r14
3394 #define test_mask q0
3396 #define draw_mask q2
3397 #define zero_mask q3
3398 #define draw_mask_combined q4
3399 #define fb_pixels q5
3400 #define fb_pixels_next q6
3403 #define draw_mask_low d4
3404 #define draw_mask_high d5
3405 #define msb_mask_low d14
3406 #define msb_mask_high d15
3409 function(shade_blocks_textured_unmodulated_indirect)
3410 stmdb sp!, { r4, r14 }
3412 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3414 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3415 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3417 vld1.u32 { test_mask }, [psx_gpu, :128]
3418 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3421 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3423 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3424 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3425 [draw_mask_bits_ptr, :16], c_64
3426 vceq.u16 zero_mask, pixels, #0
3428 vtst.u16 draw_mask, draw_mask, test_mask
3429 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3431 subs num_blocks, num_blocks, #1
3435 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3436 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3438 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3439 [draw_mask_bits_ptr, :16], c_64
3440 vceq.u16 zero_mask, pixels, #0
3442 vtst.u16 draw_mask, draw_mask, test_mask
3443 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3445 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3446 subs num_blocks, num_blocks, #1
3451 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3452 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3455 ldmia sp!, { r4, pc }
3460 function(shade_blocks_textured_unmodulated_direct)
3461 stmdb sp!, { r4, r14 }
3463 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3465 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3466 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3468 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3471 vld1.u32 { test_mask }, [psx_gpu, :128]
3472 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3474 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3475 [draw_mask_bits_ptr, :16], c_64
3476 ldr fb_ptr_next, [block_ptr_load, #44]
3478 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3479 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3480 vceq.u16 zero_mask, pixels, #0
3481 vtst.u16 draw_mask, draw_mask, test_mask
3483 subs num_blocks, num_blocks, #1
3487 mov fb_ptr, fb_ptr_next
3488 ldr fb_ptr_next, [block_ptr_load, #44]
3490 vorr.u16 pixels, pixels, msb_mask
3492 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3493 vmov fb_pixels, fb_pixels_next
3495 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3496 [draw_mask_bits_ptr, :16], c_64
3497 vbif.u16 fb_pixels, pixels, draw_mask_combined
3499 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3500 pld [fb_ptr_next, #64]
3502 add fb_ptr_cmp, fb_ptr_cmp, #14
3503 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3508 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3509 vceq.u16 zero_mask, pixels, #0
3511 vst1.u16 { fb_pixels }, [fb_ptr]
3512 vtst.u16 draw_mask, draw_mask, test_mask
3515 subs num_blocks, num_blocks, #1
3519 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3520 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3522 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3525 ldmia sp!, { r4, pc }
3528 vst1.u16 { fb_pixels }, [fb_ptr]
3529 vceq.u16 zero_mask, pixels, #0
3531 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3532 vtst.u16 draw_mask, draw_mask, test_mask
3537 function(shade_blocks_unshaded_untextured_indirect)
3542 function(shade_blocks_unshaded_untextured_direct)
3543 stmdb sp!, { r4, r14 }
3545 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3547 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3548 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3550 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3551 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3553 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3554 vld1.u16 { pixels }, [color_ptr, :128]
3557 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3559 vorr.u16 pixels, pixels, msb_mask
3560 subs num_blocks, num_blocks, #1
3562 ldr fb_ptr_next, [block_ptr_load], #64
3564 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3568 vmov fb_pixels, fb_pixels_next
3569 mov fb_ptr, fb_ptr_next
3570 ldr fb_ptr_next, [block_ptr_load], #64
3572 vbif.u16 fb_pixels, pixels, draw_mask
3573 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3575 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3576 add fb_ptr_cmp, fb_ptr_cmp, #14
3580 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3581 vst1.u16 { fb_pixels }, [fb_ptr]
3584 subs num_blocks, num_blocks, #1
3588 vbif.u16 fb_pixels_next, pixels, draw_mask
3589 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3592 ldmia sp!, { r4, pc }
3595 vst1.u16 { fb_pixels }, [fb_ptr]
3596 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3600 #undef draw_mask_ptr
3607 #define num_blocks r1
3608 #define msb_mask_ptr r2
3609 #define pixel_ptr r3
3610 #define draw_mask_ptr r0
3613 #define fb_ptr_next r14
3614 #define fb_ptr_cmp r4
3622 #undef msb_mask_high
3623 #undef draw_mask_next
3626 #undef fb_pixels_next
3629 #define draw_mask q1
3631 #define fb_pixels q3
3632 #define blend_pixels q4
3633 #define pixels_no_msb q5
3634 #define blend_mask q6
3635 #define fb_pixels_no_msb q7
3636 #define d128_0x8000 q8
3637 #define d128_0x0421 q9
3638 #define fb_pixels_next q10
3639 #define blend_pixels_next q11
3640 #define pixels_next q12
3641 #define draw_mask_next q13
3642 #define write_mask q14
3644 #define pixels_rb q5
3645 #define pixels_mg q7
3647 #define d128_0x7C1F q8
3648 #define d128_0x03E0 q9
3649 #define fb_pixels_rb q10
3650 #define fb_pixels_g q11
3651 #define fb_pixels_masked q11
3652 #define d128_0x83E0 q15
3653 #define pixels_fourth q7
3654 #define d128_0x1C07 q12
3655 #define d128_0x00E0 q13
3656 #define d128_0x80E0 q13
3658 #define msb_mask_low d0
3659 #define msb_mask_high d1
3661 #define blend_blocks_average_set_blend_mask_textured(source) \
3662 vclt.s16 blend_mask, source, #0 \
3664 #define blend_blocks_average_set_stp_bit_textured() \
3665 vorr.u16 blend_pixels, #0x8000 \
3667 #define blend_blocks_average_combine_textured(source) \
3668 vbif.u16 blend_pixels, source, blend_mask \
3670 #define blend_blocks_average_set_blend_mask_untextured(source) \
3672 #define blend_blocks_average_set_stp_bit_untextured() \
3674 #define blend_blocks_average_combine_untextured(source) \
3676 #define blend_blocks_average_mask_set_on() \
3677 vclt.s16 write_mask, fb_pixels_next, #0 \
3679 #define blend_blocks_average_mask_copy_on() \
3680 vorr.u16 draw_mask, draw_mask_next, write_mask \
3682 #define blend_blocks_average_mask_copy_b_on() \
3683 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3685 #define blend_blocks_average_mask_set_off() \
3687 #define blend_blocks_average_mask_copy_off() \
3688 vmov draw_mask, draw_mask_next \
3690 #define blend_blocks_average_mask_copy_b_off() \
3692 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3695 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3696 stmdb sp!, { r4, r14 }; \
3698 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3699 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3701 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3702 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3704 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3707 vmov.u16 d128_0x8000, #0x8000; \
3708 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3709 ldr fb_ptr_next, [pixel_ptr, #28]; \
3711 vmov.u16 d128_0x0421, #0x0400; \
3712 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3714 vorr.u16 d128_0x0421, #0x0021; \
3715 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3717 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3718 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3719 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3720 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3721 blend_blocks_average_mask_set_##mask_evaluate(); \
3722 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3724 subs num_blocks, num_blocks, #1; \
3728 mov fb_ptr, fb_ptr_next; \
3729 ldr fb_ptr_next, [pixel_ptr, #28]; \
3731 vmov pixels, pixels_next; \
3732 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3734 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3736 blend_blocks_average_mask_copy_##mask_evaluate(); \
3737 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3739 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3740 blend_blocks_average_set_stp_bit_##texturing(); \
3741 vmov fb_pixels, fb_pixels_next; \
3742 blend_blocks_average_combine_##texturing(pixels); \
3744 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3745 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3746 cmp fb_ptr_cmp, #28; \
3749 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3750 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3752 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3753 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3755 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3756 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3758 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3759 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3760 blend_blocks_average_mask_set_##mask_evaluate(); \
3761 vst1.u16 { fb_pixels }, [fb_ptr]; \
3764 subs num_blocks, num_blocks, #1; \
3768 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3769 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3771 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3772 blend_blocks_average_set_stp_bit_##texturing(); \
3773 blend_blocks_average_combine_##texturing(pixels_next); \
3775 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3776 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3777 vst1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3779 restore_abi_regs(); \
3780 ldmia sp!, { r4, pc }; \
3783 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3784 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3785 vst1.u16 { fb_pixels }, [fb_ptr]; \
3787 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3788 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3789 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3790 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3791 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3792 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3796 blend_blocks_average_builder(textured, off)
3797 blend_blocks_average_builder(untextured, off)
3798 blend_blocks_average_builder(textured, on)
3799 blend_blocks_average_builder(untextured, on)
3802 #define blend_blocks_add_mask_set_on() \
3803 vclt.s16 write_mask, fb_pixels, #0 \
3805 #define blend_blocks_add_mask_copy_on() \
3806 vorr.u16 draw_mask, draw_mask, write_mask \
3808 #define blend_blocks_add_mask_set_off() \
3810 #define blend_blocks_add_mask_copy_off() \
3813 #define blend_blocks_add_textured_builder(mask_evaluate) \
3816 function(blend_blocks_textured_add_##mask_evaluate) \
3817 stmdb sp!, { r4, r14 }; \
3819 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3820 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3822 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3823 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3825 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3828 vmov.u16 d128_0x7C1F, #0x7C00; \
3829 vmov.u16 d128_0x03E0, #0x0300; \
3830 vmov.u16 d128_0x83E0, #0x8000; \
3831 vorr.u16 d128_0x03E0, #0x00E0; \
3832 vorr.u16 d128_0x7C1F, #0x001F; \
3833 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3835 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3836 ldr fb_ptr_next, [pixel_ptr, #28]; \
3837 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3838 vclt.s16 blend_mask, pixels, #0; \
3839 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3840 blend_blocks_add_mask_set_##mask_evaluate(); \
3841 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3843 blend_blocks_add_mask_copy_##mask_evaluate(); \
3844 vorr.u16 pixels, pixels, msb_mask; \
3845 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3846 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3847 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3848 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3849 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3850 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3851 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3852 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3854 subs num_blocks, num_blocks, #1; \
3858 mov fb_ptr, fb_ptr_next; \
3860 ldr fb_ptr_next, [pixel_ptr, #28]; \
3862 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3863 vclt.s16 blend_mask, pixels, #0; \
3865 vorr.u16 pixels, pixels, msb_mask; \
3866 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3867 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3869 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3870 pld [fb_ptr_next, #64]; \
3872 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3873 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3875 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3876 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3878 cmp fb_ptr_cmp, #28; \
3881 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3882 blend_blocks_add_mask_set_##mask_evaluate(); \
3883 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3884 blend_blocks_add_mask_copy_##mask_evaluate(); \
3885 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3886 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3887 vst1.u16 { blend_pixels }, [fb_ptr]; \
3890 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3891 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3892 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3893 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3894 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3896 subs num_blocks, num_blocks, #1; \
3900 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3901 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3902 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3904 restore_abi_regs(); \
3905 ldmia sp!, { r4, pc }; \
3908 vst1.u16 { blend_pixels }, [fb_ptr]; \
3909 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3911 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3912 blend_blocks_add_mask_set_##mask_evaluate(); \
3913 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3914 blend_blocks_add_mask_copy_##mask_evaluate(); \
3915 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3919 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3922 function(blend_blocks_untextured_add_##mask_evaluate) \
3923 stmdb sp!, { r4, r14 }; \
3925 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3926 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3928 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3929 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3931 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3934 vmov.u16 d128_0x7C1F, #0x7C00; \
3935 vmov.u16 d128_0x03E0, #0x0300; \
3936 vorr.u16 d128_0x7C1F, #0x001F; \
3937 vorr.u16 d128_0x03E0, #0x00E0; \
3939 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3940 ldr fb_ptr_next, [pixel_ptr, #28]; \
3941 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3942 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3943 blend_blocks_add_mask_set_##mask_evaluate(); \
3944 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3946 blend_blocks_add_mask_copy_##mask_evaluate(); \
3947 vand.u16 pixels_g, pixels, d128_0x03E0; \
3948 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3949 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3950 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3951 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3952 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3953 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3955 subs num_blocks, num_blocks, #1; \
3959 mov fb_ptr, fb_ptr_next; \
3961 ldr fb_ptr_next, [pixel_ptr, #28]; \
3963 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3965 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3966 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3967 vand.u16 pixels_g, pixels, d128_0x03E0; \
3969 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3970 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3972 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3973 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3974 cmp fb_ptr_cmp, #28; \
3977 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3978 blend_blocks_add_mask_set_##mask_evaluate(); \
3979 blend_blocks_add_mask_copy_##mask_evaluate(); \
3980 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3981 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3982 vst1.u16 { blend_pixels }, [fb_ptr]; \
3985 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3986 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3987 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3988 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3989 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3991 subs num_blocks, num_blocks, #1; \
3995 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3996 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3997 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3998 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4000 restore_abi_regs(); \
4001 ldmia sp!, { r4, pc }; \
4004 vst1.u16 { blend_pixels }, [fb_ptr]; \
4005 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
4007 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4008 blend_blocks_add_mask_set_##mask_evaluate(); \
4009 blend_blocks_add_mask_copy_##mask_evaluate(); \
4010 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4014 blend_blocks_add_textured_builder(off)
4015 blend_blocks_add_textured_builder(on)
4016 blend_blocks_add_untextured_builder(off)
4017 blend_blocks_add_untextured_builder(on)
4019 #define blend_blocks_subtract_set_blend_mask_textured() \
4020 vclt.s16 blend_mask, pixels_next, #0 \
4022 #define blend_blocks_subtract_combine_textured() \
4023 vbif.u16 blend_pixels, pixels, blend_mask \
4025 #define blend_blocks_subtract_set_stp_textured() \
4026 vorr.u16 blend_pixels, #0x8000 \
4028 #define blend_blocks_subtract_msb_mask_textured() \
4029 vorr.u16 pixels, pixels_next, msb_mask \
4031 #define blend_blocks_subtract_set_blend_mask_untextured() \
4033 #define blend_blocks_subtract_combine_untextured() \
4035 #define blend_blocks_subtract_set_stp_untextured() \
4036 vorr.u16 blend_pixels, blend_pixels, msb_mask \
4038 #define blend_blocks_subtract_msb_mask_untextured() \
4041 #define blend_blocks_subtract_mask_set_on() \
4042 vclt.s16 write_mask, fb_pixels, #0 \
4044 #define blend_blocks_subtract_mask_copy_on() \
4045 vorr.u16 draw_mask, draw_mask_next, write_mask \
4047 #define blend_blocks_subtract_mask_set_off() \
4049 #define blend_blocks_subtract_mask_copy_off() \
4050 vmov draw_mask, draw_mask_next \
4053 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
4056 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
4057 stmdb sp!, { r4, r14 }; \
4059 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4060 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4062 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4063 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4065 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4068 vmov.u16 d128_0x7C1F, #0x7C00; \
4069 vmov.u16 d128_0x03E0, #0x0300; \
4070 vorr.u16 d128_0x7C1F, #0x001F; \
4071 vorr.u16 d128_0x03E0, #0x00E0; \
4073 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4074 ldr fb_ptr_next, [pixel_ptr, #28]; \
4075 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4076 blend_blocks_subtract_set_blend_mask_##texturing(); \
4077 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4078 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4079 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4081 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4082 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4083 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4084 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4085 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4087 subs num_blocks, num_blocks, #1; \
4091 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4092 mov fb_ptr, fb_ptr_next; \
4093 ldr fb_ptr_next, [pixel_ptr, #28]; \
4095 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4096 blend_blocks_subtract_msb_mask_##texturing(); \
4098 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4099 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4100 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4101 blend_blocks_subtract_set_stp_##texturing(); \
4102 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4103 blend_blocks_subtract_combine_##texturing(); \
4104 blend_blocks_subtract_set_blend_mask_##texturing(); \
4105 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4107 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4108 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4109 cmp fb_ptr_cmp, #28; \
4112 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4113 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4114 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4115 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4116 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4117 vst1.u16 { blend_pixels }, [fb_ptr]; \
4118 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4121 subs num_blocks, num_blocks, #1; \
4125 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4127 blend_blocks_subtract_msb_mask_##texturing(); \
4128 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4129 blend_blocks_subtract_set_stp_##texturing(); \
4130 blend_blocks_subtract_combine_##texturing(); \
4131 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4132 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4134 restore_abi_regs(); \
4135 ldmia sp!, { r4, pc }; \
4138 vst1.u16 { blend_pixels }, [fb_ptr]; \
4139 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4140 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4141 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4142 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4143 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4144 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4148 blend_blocks_subtract_builder(textured, off)
4149 blend_blocks_subtract_builder(textured, on)
4150 blend_blocks_subtract_builder(untextured, off)
4151 blend_blocks_subtract_builder(untextured, on)
4154 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4157 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4158 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]; \
4163 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4164 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4166 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
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; \
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 vclt.s16 blend_mask, pixels, #0; \
4181 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4182 blend_blocks_add_mask_set_##mask_evaluate(); \
4183 vshr.s16 pixels_fourth, pixels, #2; \
4184 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4186 blend_blocks_add_mask_copy_##mask_evaluate(); \
4187 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4188 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4189 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4190 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4191 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4192 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4193 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4195 subs num_blocks, num_blocks, #1; \
4199 mov fb_ptr, fb_ptr_next; \
4200 ldr fb_ptr_next, [pixel_ptr, #28]; \
4202 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4203 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4204 vbif.u16 blend_pixels, pixels, blend_mask; \
4206 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4207 vclt.s16 blend_mask, pixels, #0; \
4208 vshr.s16 pixels_fourth, pixels, #2; \
4209 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4210 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4212 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4213 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4215 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4216 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4217 cmp fb_ptr_cmp, #28; \
4220 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4221 blend_blocks_add_mask_set_##mask_evaluate(); \
4222 blend_blocks_add_mask_copy_##mask_evaluate(); \
4223 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4224 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4225 vst1.u16 { blend_pixels }, [fb_ptr]; \
4228 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4229 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4230 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4231 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4232 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4234 subs num_blocks, num_blocks, #1; \
4238 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4239 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4240 vbif.u16 blend_pixels, pixels, blend_mask; \
4241 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4242 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4243 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4245 restore_abi_regs(); \
4246 ldmia sp!, { r4, pc }; \
4249 vst1.u16 { blend_pixels }, [fb_ptr]; \
4250 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4252 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4253 blend_blocks_add_mask_set_##mask_evaluate(); \
4254 blend_blocks_add_mask_copy_##mask_evaluate(); \
4255 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4260 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4263 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4264 stmdb sp!, { r4, r14 }; \
4266 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4267 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4269 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4270 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4272 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4275 vmov.u16 d128_0x7C1F, #0x7C00; \
4276 vmov.u16 d128_0x03E0, #0x0300; \
4277 vmov.u16 d128_0x1C07, #0x1C00; \
4278 vmov.u16 d128_0x00E0, #0x00E0; \
4279 vorr.u16 d128_0x7C1F, #0x001F; \
4280 vorr.u16 d128_0x03E0, #0x00E0; \
4281 vorr.u16 d128_0x1C07, #0x0007; \
4283 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4284 ldr fb_ptr_next, [pixel_ptr, #28]; \
4285 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4286 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4287 blend_blocks_add_mask_set_##mask_evaluate(); \
4288 vshr.s16 pixels_fourth, pixels, #2; \
4289 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4291 blend_blocks_add_mask_copy_##mask_evaluate(); \
4292 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4293 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4294 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4295 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4296 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4297 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4298 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4300 subs num_blocks, num_blocks, #1; \
4304 mov fb_ptr, fb_ptr_next; \
4305 ldr fb_ptr_next, [pixel_ptr, #28]; \
4307 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4309 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4310 vshr.s16 pixels_fourth, pixels, #2; \
4311 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4312 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4314 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4315 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4317 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4318 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4319 cmp fb_ptr_cmp, #28; \
4322 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4323 blend_blocks_add_mask_set_##mask_evaluate(); \
4324 blend_blocks_add_mask_copy_##mask_evaluate(); \
4325 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4326 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4327 vst1.u16 { blend_pixels }, [fb_ptr]; \
4330 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4331 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4332 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4333 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4334 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4336 subs num_blocks, num_blocks, #1; \
4340 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4341 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4342 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4343 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4345 restore_abi_regs(); \
4346 ldmia sp!, { r4, pc }; \
4349 vst1.u16 { blend_pixels }, [fb_ptr]; \
4350 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4352 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4353 blend_blocks_add_mask_set_##mask_evaluate(); \
4354 blend_blocks_add_mask_copy_##mask_evaluate(); \
4355 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4359 blend_blocks_add_fourth_textured_builder(off)
4360 blend_blocks_add_fourth_textured_builder(on)
4361 blend_blocks_add_fourth_untextured_builder(off)
4362 blend_blocks_add_fourth_untextured_builder(on)
4364 // TODO: Optimize this more. Need a scene that actually uses it for
4369 function(blend_blocks_textured_unblended_on)
4370 stmdb sp!, { r4, r14 }
4372 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4373 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
4375 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4376 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
4378 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4381 ldr fb_ptr, [pixel_ptr, #28]
4382 vld1.u16 { fb_pixels }, [fb_ptr]
4383 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4384 vclt.s16 write_mask, fb_pixels, #0
4385 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4387 subs num_blocks, num_blocks, #1
4391 vorr.u16 pixels, pixels, msb_mask
4392 vorr.u16 draw_mask, draw_mask, write_mask
4393 vbif.u16 fb_pixels, pixels, draw_mask
4394 vst1.u16 { fb_pixels }, [fb_ptr]
4396 ldr fb_ptr, [pixel_ptr, #28]
4397 vld1.u16 { fb_pixels }, [fb_ptr]
4398 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4399 vclt.s16 write_mask, fb_pixels, #0
4400 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4402 subs num_blocks, num_blocks, #1
4406 vorr.u16 pixels, pixels, msb_mask
4407 vorr.u16 draw_mask, draw_mask, write_mask
4408 vbif.u16 fb_pixels, pixels, draw_mask
4409 vst1.u16 { fb_pixels }, [fb_ptr]
4412 ldmia sp!, { r4, pc }
4415 function(blend_blocks_textured_unblended_off)
4425 vld1.u32 { u_whole_8, v_whole_8 }, [r1, :128], r3
4442 #undef dirty_textures_mask
4444 #undef current_texture_mask
4455 #define offset_u_right r10
4456 #define width_rounded r11
4457 #define height_rounded r12
4459 #define texture_offset_base r1
4460 #define tile_width r2
4461 #define tile_height r3
4462 #define num_blocks r4
4464 #define sub_tile_height r6
4466 #define texture_mask r8
4467 #define column_data r9
4468 #define texture_offset r10
4469 #define tiles_remaining r11
4470 #define fb_ptr_advance_column r12
4471 #define texture_block_ptr r14
4475 #define texture_page_ptr r3
4476 #define left_block_mask r4
4477 #define right_block_mask r5
4478 #define texture_mask_rev r10
4479 #define control_mask r11
4481 #define dirty_textures_mask r4
4483 #define current_texture_mask r6
4497 #define draw_masks_fb_ptrs q1
4499 #define draw_mask_fb_ptr_left d2
4500 #define draw_mask_fb_ptr_right d3
4502 #define draw_mask_fb_ptr_left_a d2
4503 #define draw_mask_fb_ptr_left_b d3
4504 #define draw_mask_fb_ptr_right_a d10
4505 #define draw_mask_fb_ptr_right_b d11
4506 #define draw_masks_fb_ptrs2 q5
4508 #define clut_low_a d4
4509 #define clut_low_b d5
4510 #define clut_high_a d6
4511 #define clut_high_b d7
4513 #define block_masks d8
4514 #define block_masks_shifted d9
4519 #define texels_low d12
4520 #define texels_high d13
4522 #define texels_wide_low d14
4523 #define texels_wide_high d15
4524 #define texels_wide q7
4528 setup_sprite_flush_blocks:
4529 push { r0 - r3, EXTRA_UNSAVED_REGS r12, lr }
4530 add block, r0, #psx_gpu_saved_tmp_offset /* r5 */
4531 vstmia block, { q1 - q3 }
4532 bl flush_render_block_buffer
4533 vldmia block, { q1 - q3 }
4534 pop { r0 - r3, EXTRA_UNSAVED_REGS r12, lr }
4536 add block, psx_gpu, #psx_gpu_blocks_offset
4540 setup_sprite_update_texture_4bpp_cache:
4541 push { r0 - r4, lr }
4542 bl update_texture_4bpp_cache
4546 setup_sprite_update_texture_8bpp_cache:
4547 push { r0 - r4, EXTRA_UNSAVED_REGS lr }
4548 bl update_texture_8bpp_cache
4549 pop { r0 - r4, EXTRA_UNSAVED_REGS pc }
4552 #define setup_sprite_tiled_initialize_4bpp() \
4553 ldr dirty_textures_mask, \
4554 [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]; \
4555 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4557 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4558 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4560 tst current_texture_mask, dirty_textures_mask; \
4561 vuzp.u8 clut_a, clut_b; \
4563 blne setup_sprite_update_texture_4bpp_cache \
4565 #define setup_sprite_tiled_initialize_8bpp() \
4566 ldr dirty_textures_mask, \
4567 [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]; \
4568 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4570 tst current_texture_mask, dirty_textures_mask; \
4571 blne setup_sprite_update_texture_8bpp_cache \
4574 #define setup_sprite_block_count_single() \
4577 #define setup_sprite_block_count_double() \
4578 sub_tile_height, lsl #1 \
4580 #define setup_sprite_tile_add_blocks(type) \
4581 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4582 cmp num_blocks, #MAX_BLOCKS; \
4584 movgt num_blocks, setup_sprite_block_count_##type(); \
4585 blgt setup_sprite_flush_blocks \
4588 #define setup_sprite_tile_full_4bpp(edge) \
4589 setup_sprite_tile_add_blocks(double); \
4592 and texture_block_ptr, texture_offset, texture_mask; \
4593 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4596 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4597 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4599 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4600 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4602 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4603 add texture_block_ptr, texture_offset, #8; \
4605 and texture_block_ptr, texture_block_ptr, texture_mask; \
4606 add block, block, #40; \
4608 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4609 add fb_ptr, fb_ptr, #16; \
4611 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4612 add block, block, #24; \
4614 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4615 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4618 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4619 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4621 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4622 add block, block, #40; \
4624 add texture_offset, texture_offset, #0x10; \
4625 add fb_ptr, fb_ptr, #(2048 - 16); \
4627 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4628 add block, block, #24; \
4630 subs sub_tile_height, sub_tile_height, #1; \
4633 add texture_offset, texture_offset, #0xF00; \
4634 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4637 #define setup_sprite_tile_half_4bpp(edge) \
4638 setup_sprite_tile_add_blocks(single); \
4641 and texture_block_ptr, texture_offset, texture_mask; \
4642 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4645 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4646 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4648 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4649 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4651 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4652 add block, block, #40; \
4654 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4655 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4657 add block, block, #24; \
4658 add texture_offset, texture_offset, #0x10; \
4660 add fb_ptr, fb_ptr, #2048; \
4661 subs sub_tile_height, sub_tile_height, #1; \
4665 add texture_offset, texture_offset, #0xF00; \
4666 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4669 #define setup_sprite_tile_full_8bpp(edge) \
4670 setup_sprite_tile_add_blocks(double); \
4671 add block, block, #16; \
4674 and texture_block_ptr, texture_offset, texture_mask; \
4675 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4678 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4679 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4681 add texture_block_ptr, texture_offset, #8; \
4682 vst1.u32 { texels }, [block, :64]; \
4684 and texture_block_ptr, texture_block_ptr, texture_mask; \
4685 add block, block, #24; \
4687 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4689 add fb_ptr, fb_ptr, #16; \
4690 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4692 add block, block, #40; \
4693 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4696 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4697 vst1.u32 { texels }, [block, :64]; \
4698 add block, block, #24; \
4700 add texture_offset, texture_offset, #0x10; \
4701 add fb_ptr, fb_ptr, #(2048 - 16); \
4703 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4704 add block, block, #40; \
4706 subs sub_tile_height, sub_tile_height, #1; \
4709 sub block, block, #16; \
4710 add texture_offset, texture_offset, #0xF00; \
4711 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4714 #define setup_sprite_tile_half_8bpp(edge) \
4715 setup_sprite_tile_add_blocks(single); \
4716 add block, block, #16; \
4719 and texture_block_ptr, texture_offset, texture_mask; \
4720 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4723 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4724 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4726 vst1.u32 { texels }, [block, :64]; \
4727 add block, block, #24; \
4729 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4730 add block, block, #40; \
4732 add texture_offset, texture_offset, #0x10; \
4733 add fb_ptr, fb_ptr, #2048; \
4735 subs sub_tile_height, sub_tile_height, #1; \
4738 sub block, block, #16; \
4739 add texture_offset, texture_offset, #0xF00; \
4740 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4743 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4744 add texture_offset, texture_offset_base, #8; \
4745 add fb_ptr, fb_ptr, #16 \
4747 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4748 mov texture_offset, texture_offset_base \
4750 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4751 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4753 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4754 mov texture_offset, texture_offset_base \
4756 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4757 sub fb_ptr, fb_ptr, #16 \
4759 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4761 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4762 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4764 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4767 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4769 mov sub_tile_height, column_data; \
4770 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4771 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4772 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4774 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4776 and sub_tile_height, column_data, #0xFF; \
4777 mov tiles_remaining, column_data, lsr #16; \
4778 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4779 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4781 subs tiles_remaining, tiles_remaining, #1; \
4785 mov sub_tile_height, #16; \
4786 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4787 subs tiles_remaining, tiles_remaining, #1; \
4791 uxtb sub_tile_height, column_data, ror #8; \
4792 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4793 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4796 #define setup_sprite_column_data_single() \
4797 mov column_data, height; \
4798 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] \
4800 #define setup_sprite_column_data_multi() \
4801 and height_rounded, height_rounded, #0xF; \
4802 rsb column_data, offset_v, #16; \
4804 add height_rounded, height_rounded, #1; \
4805 sub tile_height, tile_height, #1; \
4807 orr column_data, column_data, tile_height, lsl #16; \
4808 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]; \
4810 orr column_data, column_data, height_rounded, lsl #8 \
4812 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4813 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4814 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4816 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4817 mov fb_ptr_advance_column, #32; \
4818 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4820 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4821 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4823 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4824 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4825 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4827 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4829 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4830 setup_sprite_column_data_##multi_height(); \
4831 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4832 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4833 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4835 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4837 pop { r3 - r11, pc } \
4839 #define setup_sprite_tiled_advance_column() \
4840 add texture_offset_base, texture_offset_base, #0x100; \
4841 tst texture_offset_base, #0xF00; \
4842 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4844 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4845 right_mode, x4mode) \
4846 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4847 setup_sprite_column_data_##multi_height(); \
4849 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4851 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4853 subs tile_width, tile_width, #2; \
4854 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4858 vmov.u8 draw_masks_fb_ptrs, #0; \
4859 vmov.u8 draw_masks_fb_ptrs2, #0; \
4862 setup_sprite_tiled_advance_column(); \
4863 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4864 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4865 subs tile_width, tile_width, #1; \
4869 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4871 setup_sprite_tiled_advance_column(); \
4872 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4874 pop { r3 - r11, pc } \
4877 #define setup_sprite_offset_u_adjust() \
4879 #define setup_sprite_get_left_block_mask() \
4880 and left_block_mask, left_block_mask, #0xFF \
4882 #define setup_sprite_compare_left_block_mask() \
4883 cmp left_block_mask, #0xFF \
4885 #define setup_sprite_get_right_block_mask() \
4886 uxtb right_block_mask, right_block_mask, ror #8 \
4888 #define setup_sprite_compare_right_block_mask() \
4889 cmp right_block_mask, #0xFF \
4894 #define fb_ptr2 column_data
4896 #define setup_sprite_offset_u_adjust_4x() \
4897 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4898 lsl offset_u_right, #1; \
4900 add offset_u_right, #1 \
4902 #define setup_sprite_get_left_block_mask_4x() \
4903 sxth left_block_mask, left_block_mask \
4905 #define setup_sprite_compare_left_block_mask_4x() \
4906 cmp left_block_mask, #0xFFFFFFFF \
4908 #define setup_sprite_get_right_block_mask_4x() \
4909 sxth right_block_mask, right_block_mask, ror #16 \
4911 #define setup_sprite_compare_right_block_mask_4x() \
4912 cmp right_block_mask, #0xFFFFFFFF \
4915 #define widen_texels_16bpp(texels_) \
4916 vmov texels_wide_low, texels_; \
4917 vmov texels_wide_high, texels_; \
4918 vzip.16 texels_wide_low, texels_wide_high \
4920 #define widen_texels_8bpp(texels_) \
4921 vmov texels_wide_low, texels_; \
4922 vmov texels_wide_high, texels_; \
4923 vzip.8 texels_wide_low, texels_wide_high \
4925 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4926 vst1.u32 { texels_ }, [block_, :128]; \
4927 add block_, block_, #40; \
4929 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4930 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4931 add block_, block_, #24 \
4933 /* assumes 16-byte offset already added to block_ */
4934 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4935 vst1.u32 { texels_ }, [block_, :64]; \
4936 add block_, block_, #24; \
4938 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4939 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4940 add block_, block_, #40 \
4942 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4943 draw_mask_fb_ptr_b_) \
4944 widen_texels_16bpp(texels_low); \
4945 add fb_ptr_tmp, fb_ptr, #1024*2; \
4947 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4949 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4950 widen_texels_16bpp(texels_high); \
4952 add fb_ptr_tmp, fb_ptr, #8*2; \
4953 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4955 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4956 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4958 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4959 draw_mask_fb_ptr_b_) \
4960 widen_texels_8bpp(texels); \
4961 add fb_ptr_tmp, fb_ptr, #1024*2; \
4963 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4964 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4966 add fb_ptr_tmp, fb_ptr, #8*2; \
4967 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4969 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4970 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4973 #define setup_sprite_tiled_initialize_4bpp_4x() \
4974 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4975 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4977 vuzp.u8 clut_a, clut_b \
4979 #define setup_sprite_tiled_initialize_8bpp_4x() \
4982 #define setup_sprite_block_count_single_4x() \
4983 sub_tile_height, lsl #2 \
4985 #define setup_sprite_block_count_double_4x() \
4986 sub_tile_height, lsl #(1+2) \
4988 #define setup_sprite_tile_full_4bpp_4x(edge) \
4989 setup_sprite_tile_add_blocks(double_4x); \
4990 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4993 and texture_block_ptr, texture_offset, texture_mask; \
4996 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4997 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4999 add texture_block_ptr, texture_offset, #8; \
5000 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5002 and texture_block_ptr, texture_block_ptr, texture_mask; \
5003 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5005 vzip.8 texels_low, texels_high; \
5006 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5007 draw_mask_fb_ptr_left_b); \
5009 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5010 pld [fb_ptr, #2048]; \
5012 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5013 add fb_ptr, fb_ptr, #16*2; \
5015 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5016 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5018 vzip.8 texels_low, texels_high; \
5019 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5020 draw_mask_fb_ptr_right_b); \
5022 add texture_offset, texture_offset, #0x10; \
5023 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5025 subs sub_tile_height, sub_tile_height, #1; \
5028 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5029 add texture_offset, texture_offset, #0xF00; \
5030 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5033 #define setup_sprite_tile_half_4bpp_4x(edge) \
5034 setup_sprite_tile_add_blocks(single_4x); \
5035 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5038 and texture_block_ptr, texture_offset, texture_mask; \
5041 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5042 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5044 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5045 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5047 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5048 add texture_offset, texture_offset, #0x10; \
5050 vzip.8 texels_low, texels_high; \
5051 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5052 draw_mask_fb_ptr_##edge##_b); \
5054 pld [fb_ptr, #2048]; \
5055 add fb_ptr, fb_ptr, #2048 * 2; \
5057 subs sub_tile_height, sub_tile_height, #1; \
5060 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5061 add texture_offset, texture_offset, #0xF00; \
5062 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5065 #define setup_sprite_tile_full_8bpp_4x(edge) \
5066 setup_sprite_tile_add_blocks(double_4x); \
5067 add block, block, #16; \
5068 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5071 and texture_block_ptr, texture_offset, texture_mask; \
5074 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5075 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5077 add texture_block_ptr, texture_offset, #8; \
5078 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5079 draw_mask_fb_ptr_left_b); \
5081 pld [fb_ptr, #2048]; \
5082 and texture_block_ptr, texture_block_ptr, texture_mask; \
5084 add fb_ptr, fb_ptr, #16*2; \
5085 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5087 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5089 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5090 draw_mask_fb_ptr_right_b); \
5092 add texture_offset, texture_offset, #0x10; \
5093 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5095 subs sub_tile_height, sub_tile_height, #1; \
5098 sub block, block, #16; \
5099 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5100 add texture_offset, texture_offset, #0xF00; \
5101 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5104 #define setup_sprite_tile_half_8bpp_4x(edge) \
5105 setup_sprite_tile_add_blocks(single_4x); \
5106 add block, block, #16; \
5107 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5110 and texture_block_ptr, texture_offset, texture_mask; \
5113 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5114 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5116 pld [fb_ptr, #2048]; \
5117 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5118 draw_mask_fb_ptr_##edge##_b); \
5120 add texture_offset, texture_offset, #0x10; \
5121 add fb_ptr, fb_ptr, #2048 * 2; \
5123 subs sub_tile_height, sub_tile_height, #1; \
5126 sub block, block, #16; \
5127 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5128 add texture_offset, texture_offset, #0xF00; \
5129 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5132 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5133 add texture_offset, texture_offset_base, #8; \
5134 add fb_ptr, fb_ptr, #16 * 2 \
5136 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5137 mov texture_offset, texture_offset_base \
5139 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5140 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5142 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5143 mov texture_offset, texture_offset_base \
5145 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5146 sub fb_ptr, fb_ptr, #16 * 2 \
5148 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5150 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5151 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5153 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5156 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5157 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5158 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5159 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5160 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5162 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5163 mov fb_ptr_advance_column, #32 * 2; \
5164 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5165 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5166 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5167 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5168 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5170 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5171 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5172 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5173 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5174 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5184 // [sp + 12]: color (unused)
5186 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5188 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5190 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5192 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5194 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5196 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5198 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5200 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5202 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5204 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5206 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5208 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5210 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5212 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5214 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5219 function(setup_sprite_##texture_mode##x4mode) \
5220 push { r3 - r11, lr }; \
5221 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5223 ldr v, [sp, #4*(10+0)]; \
5224 and offset_u, u, #0xF; \
5226 ldr width, [sp, #4*(10+1)]; \
5227 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
5229 ldr height, [sp, #4*(10+2)]; \
5230 add fb_ptr, fb_ptr, y, lsl #11; \
5232 vpush { q4 - q7 }; \
5234 add fb_ptr, fb_ptr, x, lsl #1; \
5235 and offset_v, v, #0xF; \
5237 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5238 add width_rounded, offset_u, width; \
5240 add height_rounded, offset_v, height; \
5241 add width_rounded, width_rounded, #15; \
5243 add height_rounded, height_rounded, #15; \
5244 mov tile_width, width_rounded, lsr #4; \
5246 /* texture_offset_base = VH-VL-00-00 */\
5247 mov texture_offset_base, v, lsl #8; \
5248 and offset_u_right, width_rounded, #0xF; \
5250 /* texture_offset_base = VH-UH-UL-00 */\
5251 bfi texture_offset_base, u, #4, #8; \
5252 mov right_block_mask, #0xFFFFFFFE; \
5254 setup_sprite_offset_u_adjust##x4mode(); \
5256 /* texture_offset_base = VH-UH-VL-00 */\
5257 bfi texture_offset_base, v, #4, #4; \
5258 mov left_block_mask, #0xFFFFFFFF; \
5260 mov tile_height, height_rounded, lsr #4; \
5261 mvn left_block_mask, left_block_mask, lsl offset_u; \
5263 /* texture_mask = HH-HL-WH-WL */\
5264 ldrh texture_mask, [psx_gpu, #psx_gpu_texture_mask_width_offset]; \
5265 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5267 /* texture_mask_rev = WH-WL-HH-HL */\
5268 rev16 texture_mask_rev, texture_mask; \
5269 vmov block_masks, left_block_mask, right_block_mask; \
5271 /* texture_mask = HH-HL-HL-WL */\
5272 bfi texture_mask, texture_mask_rev, #4, #4; \
5273 /* texture_mask_rev = 00-00-00-WH */\
5274 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5276 /* texture_mask = HH-WH-HL-WL */\
5277 bfi texture_mask, texture_mask_rev, #8, #4; \
5278 setup_sprite_get_left_block_mask##x4mode(); \
5280 mov control_mask, #0; \
5281 setup_sprite_compare_left_block_mask##x4mode(); \
5283 setup_sprite_get_right_block_mask##x4mode(); \
5284 orreq control_mask, control_mask, #0x4; \
5286 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
5287 setup_sprite_compare_right_block_mask##x4mode(); \
5289 orreq control_mask, control_mask, #0x8; \
5290 cmp tile_width, #1; \
5292 add block, psx_gpu, #psx_gpu_blocks_offset; \
5293 orreq control_mask, control_mask, #0x1; \
5295 cmp tile_height, #1; \
5296 add block, block, num_blocks, lsl #6; \
5298 orreq control_mask, control_mask, #0x2; \
5299 JT_OP_REL(9f, control_mask, temp); \
5300 JT_OP(ldr pc, [pc, control_mask, lsl #2]); \
5304 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \
5305 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \
5306 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \
5307 .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \
5308 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \
5309 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \
5310 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \
5311 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\
5312 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \
5313 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \
5314 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \
5315 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \
5316 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \
5318 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \
5321 setup_sprite_tiled_builder(4bpp,);
5322 setup_sprite_tiled_builder(8bpp,);
5324 #undef draw_mask_fb_ptr_left
5325 #undef draw_mask_fb_ptr_right
5327 setup_sprite_tiled_builder(4bpp, _4x);
5328 setup_sprite_tiled_builder(8bpp, _4x);
5336 #define block_ptr r0
5337 #define num_blocks r1
5339 #define texel_shift_mask r3
5340 #define block_pixels_a r4
5341 #define block_pixels_b r5
5350 #define texels_01 r6
5351 #define texels_23 r7
5352 #define texels_45 r8
5353 #define texels_67 r9
5355 function(texture_sprite_blocks_8bpp)
5356 push { r4 - r11, r14 }
5357 movw texel_shift_mask, #(0xFF << 1)
5359 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5360 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
5362 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5363 ldr block_pixels_a, [block_ptr, #16]
5366 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5367 ldr block_pixels_b, [block_ptr, #20]
5369 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5370 ldrh texel_0, [clut_ptr, texel_0]
5372 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5373 ldrh texel_1, [clut_ptr, texel_1]
5375 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5376 ldr block_pixels_a, [block_ptr, #(64 + 16)]
5378 ldrh texel_2, [clut_ptr, texel_2]
5379 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5381 ldrh texel_3, [clut_ptr, texel_3]
5382 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5384 ldrh texel_4, [clut_ptr, texel_4]
5385 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5387 ldrh texel_5, [clut_ptr, texel_5]
5388 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5390 ldrh texel_6, [clut_ptr, texel_6]
5391 orr texels_01, texel_0, texel_1, lsl #16
5393 ldrh texel_7, [clut_ptr, texel_7]
5394 orr texels_23, texel_2, texel_3, lsl #16
5396 orr texels_45, texel_4, texel_5, lsl #16
5397 str texels_01, [block_ptr, #0]
5399 orr texels_67, texel_6, texel_7, lsl #16
5400 str texels_23, [block_ptr, #4]
5402 subs num_blocks, num_blocks, #1
5403 str texels_45, [block_ptr, #8]
5405 str texels_67, [block_ptr, #12]
5406 add block_ptr, block_ptr, #64
5411 pop { r4 - r11, pc }
5414 #undef width_rounded
5417 #undef texture_offset
5420 #undef texels_wide_low
5421 #undef texels_wide_high
5433 #define left_offset r8
5434 #define width_rounded r9
5435 #define right_width r10
5437 #define block_width r11
5439 #define texture_offset_base r1
5440 #define texture_mask r2
5441 #define texture_page_ptr r3
5442 #define num_blocks r4
5445 #define texture_offset r8
5446 #define blocks_remaining r9
5448 #define fb_ptr_pitch r12
5449 #define texture_block_ptr r14
5451 #define texture_mask_width r2
5452 #define texture_mask_height r3
5453 #define left_mask_bits r4
5454 #define right_mask_bits r5
5458 #undef block_masks_shifted
5461 #define block_masks d0
5462 #define block_masks_shifted d1
5463 #define draw_mask_fb_ptr d2
5466 #define draw_mask_fb_ptr_a d2
5467 #define draw_mask_fb_ptr_b d3
5468 #define texels_low d4
5469 #define texels_high d5
5470 #define texels_wide_low d6
5471 #define texels_wide_high d7
5472 #define texels_wide q3
5475 setup_sprites_16bpp_flush:
5476 push { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5477 add r1, r0, #psx_gpu_saved_tmp_offset
5478 vstmia r1, { d0 - d3 }
5479 bl flush_render_block_buffer
5480 pop { r0 - r3, EXTRA_UNSAVED_REGS r12 }
5481 add lr, r0, #psx_gpu_saved_tmp_offset
5482 vldmia lr, { d0 - d3 }
5484 add block, psx_gpu, #psx_gpu_blocks_offset
5485 mov num_blocks, block_width
5489 function(setup_sprite_16bpp)
5490 push { r3 - r11, lr }
5491 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5493 ldr v, [sp, #4*(10+0)]
5494 add fb_ptr, fb_ptr, y, lsl #11
5496 ldr width, [sp, #4*(10+1)]
5497 add fb_ptr, fb_ptr, x, lsl #1
5499 ldr height, [sp, #4*(10+2)]
5500 and left_offset, u, #0x7
5502 add texture_offset_base, u, u
5503 add width_rounded, width, #7
5505 add texture_offset_base, texture_offset_base, v, lsl #11
5506 mov left_mask_bits, #0xFF
5508 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5509 add width_rounded, width_rounded, left_offset
5511 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5512 sub fb_ptr, fb_ptr, left_offset, lsl #1
5514 add texture_mask, texture_mask_width, texture_mask_width
5515 mov right_mask_bits, #0xFE
5517 and right_width, width_rounded, #0x7
5518 mvn left_mask_bits, left_mask_bits, lsl left_offset
5520 add texture_mask, texture_mask, texture_mask_height, lsl #11
5521 mov block_width, width_rounded, lsr #3
5523 mov right_mask_bits, right_mask_bits, lsl right_width
5524 movw fb_ptr_pitch, #(2048 + 16)
5526 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5527 vmov block_masks, left_mask_bits, right_mask_bits
5529 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5530 add block, psx_gpu, #psx_gpu_blocks_offset
5532 bic texture_offset_base, texture_offset_base, #0xF
5535 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5536 add block, block, num_blocks, lsl #6
5540 vext.32 block_masks_shifted, block_masks, block_masks, #1
5541 vorr.u32 block_masks, block_masks, block_masks_shifted
5542 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5545 add num_blocks, num_blocks, #1
5546 cmp num_blocks, #MAX_BLOCKS
5547 blgt setup_sprites_16bpp_flush
5549 and texture_block_ptr, texture_offset_base, texture_mask
5550 subs height, height, #1
5552 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5553 vld1.u32 { texels }, [texture_block_ptr, :128]
5555 vst1.u32 { texels }, [block, :128]
5556 add block, block, #40
5558 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5561 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5563 add block, block, #24
5564 add texture_offset_base, texture_offset_base, #2048
5565 add fb_ptr, fb_ptr, #2048
5566 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5569 pop { r3 - r11, pc }
5572 add num_blocks, num_blocks, block_width
5573 mov texture_offset, texture_offset_base
5575 cmp num_blocks, #MAX_BLOCKS
5576 blgt setup_sprites_16bpp_flush
5578 add texture_offset_base, texture_offset_base, #2048
5579 and texture_block_ptr, texture_offset, texture_mask
5581 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5582 vld1.u32 { texels }, [texture_block_ptr, :128]
5584 vst1.u32 { texels }, [block, :128]
5585 add block, block, #40
5587 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5588 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5591 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5592 subs blocks_remaining, block_width, #2
5594 add texture_offset, texture_offset, #16
5595 add fb_ptr, fb_ptr, #16
5597 vmov.u8 draw_mask_fb_ptr, #0
5599 add block, block, #24
5603 and texture_block_ptr, texture_offset, texture_mask
5604 subs blocks_remaining, blocks_remaining, #1
5606 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5607 vld1.u32 { texels }, [texture_block_ptr, :128]
5609 vst1.u32 { texels }, [block, :128]
5610 add block, block, #40
5612 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5615 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5617 add texture_offset, texture_offset, #16
5618 add fb_ptr, fb_ptr, #16
5620 add block, block, #24
5624 and texture_block_ptr, texture_offset, texture_mask
5625 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5627 vld1.u32 { texels }, [texture_block_ptr, :128]
5628 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5630 vst1.u32 { texels }, [block, :128]
5631 add block, block, #40
5633 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5634 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5636 add block, block, #24
5637 subs height, height, #1
5639 add fb_ptr, fb_ptr, fb_ptr_pitch
5640 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5645 pop { r3 - r11, pc }
5649 // FIXME: duplicate code with normal version :(
5650 #undef draw_mask_fb_ptr
5652 function(setup_sprite_16bpp_4x)
5653 push { r3 - r11, lr }
5654 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5656 ldr v, [sp, #4*(10+0)]
5657 add fb_ptr, fb_ptr, y, lsl #11
5659 ldr width, [sp, #4*(10+1)]
5660 add fb_ptr, fb_ptr, x, lsl #1
5662 ldr height, [sp, #4*(10+2)]
5663 and left_offset, u, #0x7
5665 add texture_offset_base, u, u
5666 add width_rounded, width, #7
5668 add texture_offset_base, texture_offset_base, v, lsl #11
5669 movw left_mask_bits, #0xFFFF
5671 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5672 add width_rounded, width_rounded, left_offset
5676 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5677 sub fb_ptr, fb_ptr, left_offset, lsl #1
5679 add texture_mask, texture_mask_width, texture_mask_width
5680 movw right_mask_bits, #0xFFFC
5682 and right_width, width_rounded, #0x7
5683 mvn left_mask_bits, left_mask_bits, lsl left_offset
5687 add texture_mask, texture_mask, texture_mask_height, lsl #11
5688 mov block_width, width_rounded, lsr #3
5690 mov right_mask_bits, right_mask_bits, lsl right_width
5691 movw fb_ptr_pitch, #(2048 + 16) * 2
5693 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5694 vmov block_masks, left_mask_bits, right_mask_bits
5696 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5697 add block, psx_gpu, #psx_gpu_blocks_offset
5699 bic texture_offset_base, texture_offset_base, #0xF
5702 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5703 add block, block, num_blocks, lsl #6
5708 vext.32 block_masks_shifted, block_masks, block_masks, #1
5709 vorr.u32 block_masks, block_masks, block_masks_shifted
5710 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5711 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5714 add num_blocks, num_blocks, block_width
5715 cmp num_blocks, #MAX_BLOCKS
5716 blgt setup_sprites_16bpp_flush
5718 and texture_block_ptr, texture_offset_base, texture_mask
5719 subs height, height, #1
5721 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5722 vld1.u32 { texels }, [texture_block_ptr, :128]
5724 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5726 add texture_offset_base, texture_offset_base, #2048
5727 add fb_ptr, fb_ptr, #2048*2
5728 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5731 pop { r3 - r11, pc }
5734 add num_blocks, num_blocks, block_width
5735 mov texture_offset, texture_offset_base
5737 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5738 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5740 cmp num_blocks, #MAX_BLOCKS
5741 blgt setup_sprites_16bpp_flush
5743 add texture_offset_base, texture_offset_base, #2048
5744 and texture_block_ptr, texture_offset, texture_mask
5746 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5747 vld1.u32 { texels }, [texture_block_ptr, :128]
5749 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5751 subs blocks_remaining, block_width, #2*4
5752 add texture_offset, texture_offset, #16
5754 vmov.u8 draw_mask_fb_ptr_a, #0
5755 vmov.u8 draw_mask_fb_ptr_b, #0
5757 add fb_ptr, fb_ptr, #16*2
5761 and texture_block_ptr, texture_offset, texture_mask
5762 subs blocks_remaining, blocks_remaining, #4
5764 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5765 vld1.u32 { texels }, [texture_block_ptr, :128]
5767 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5768 add texture_offset, texture_offset, #16
5770 add fb_ptr, fb_ptr, #16*2
5774 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5775 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5777 and texture_block_ptr, texture_offset, texture_mask
5778 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5780 vld1.u32 { texels }, [texture_block_ptr, :128]
5782 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5783 subs height, height, #1
5785 add fb_ptr, fb_ptr, fb_ptr_pitch
5786 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5791 pop { r3 - r11, pc }
5796 #undef right_mask_bits
5799 #undef blocks_remaining
5809 #define right_width r5
5810 #define right_mask_bits r6
5814 #define fb_ptr_pitch r12
5816 // referenced by setup_sprites_16bpp_flush
5817 #define num_blocks r4
5819 #define block_width r11
5824 #define blocks_remaining r6
5827 #define right_mask q1
5828 #define test_mask q2
5829 #define draw_mask q2
5830 #define draw_mask_bits_fb_ptr d6
5835 function(setup_sprite_untextured_512)
5836 push { r4 - r11, r14 }
5838 ldr width, [sp, #4*(9+1)]
5839 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5841 ldr height, [sp, #4*(9+2)]
5842 add fb_ptr, fb_ptr, y, lsl #11
5844 add fb_ptr, fb_ptr, x, lsl #1
5845 sub right_width, width, #1
5847 ldr color, [sp, #4*(9+3)]
5850 add block_width, width, #7
5854 mov right_mask_bits, #0xff
5856 sub fb_ptr_pitch, block_width, #1
5857 lsl right_mask_bits, right_width
5859 lsl fb_ptr_pitch, #3+1
5860 ubfx color_r, color, #3, #5
5862 rsb fb_ptr_pitch, #1024*2
5863 ubfx color_g, color, #11, #5
5865 vld1.u32 { test_mask }, [psx_gpu, :128]
5866 ubfx color_b, color, #19, #5
5868 vdup.u16 right_mask, right_mask_bits
5869 orr color, color_r, color_b, lsl #10
5871 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5872 orr color, color, color_g, lsl #5
5874 vtst.u16 right_mask, right_mask, test_mask
5875 add block, psx_gpu, #psx_gpu_blocks_offset
5877 vdup.u16 colors, color
5878 add block, block, num_blocks, lsl #6
5881 setup_sprite_untextured_height_loop:
5882 add num_blocks, block_width
5883 sub blocks_remaining, block_width, #1
5885 cmp num_blocks, #MAX_BLOCKS
5886 blgt setup_sprites_16bpp_flush
5888 cmp blocks_remaining, #0
5891 vmov.u8 draw_mask, #0 /* zero_mask */
5892 vmov.u8 draw_mask_bits_fb_ptr, #0
5895 vst1.u32 { draw_mask }, [block, :128]!
5896 subs blocks_remaining, #1
5898 vst1.u32 { colors }, [block, :128]
5899 add block, block, #24
5901 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5902 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5904 add block, block, #24
5909 vst1.u32 { right_mask }, [block, :128]!
5912 vst1.u32 { colors }, [block, :128]
5913 add block, block, #24
5915 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5916 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5918 add block, block, #24
5919 add fb_ptr, fb_ptr_pitch
5921 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5922 bgt setup_sprite_untextured_height_loop
5924 pop { r4 - r11, pc }
5928 #undef texture_page_ptr
5930 #undef dirty_textures_mask
5931 #undef current_texture_mask
5934 #define current_texture_page r1
5935 #define texture_page_ptr r2
5936 #define vram_ptr_a r3
5937 #define current_texture_page_x r12
5938 #define current_texture_page_y r4
5939 #define dirty_textures_mask r5
5943 #define current_texture_mask r9
5945 #define vram_ptr_b r11
5947 #define texel_block_a d0
5948 #define texel_block_b d1
5949 #define texel_block_expanded_a q1
5950 #define texel_block_expanded_b q2
5951 #define texel_block_expanded_ab q2
5952 #define texel_block_expanded_c q3
5953 #define texel_block_expanded_d q0
5954 #define texel_block_expanded_cd q3
5956 function(update_texture_4bpp_cache)
5957 push { r3 - r11, r14 }
5960 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
5962 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
5963 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
5965 and current_texture_page_x, current_texture_page, #0xF
5966 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
5968 mov current_texture_page_y, current_texture_page, lsr #4
5969 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5971 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5974 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5975 bic dirty_textures_mask, current_texture_mask
5978 str dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5983 add vram_ptr_b, vram_ptr_a, #2048
5986 vld1.u32 { texel_block_a }, [vram_ptr_a, :64], c_4096
5987 vld1.u32 { texel_block_b }, [vram_ptr_b, :64], c_4096
5989 vmovl.u8 texel_block_expanded_a, texel_block_a
5990 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5991 vmovl.u8 texel_block_expanded_c, texel_block_b
5992 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5994 vbic.u16 texel_block_expanded_a, #0x00F0
5995 vbic.u16 texel_block_expanded_b, #0x00F0
5996 vbic.u16 texel_block_expanded_c, #0x00F0
5997 vbic.u16 texel_block_expanded_d, #0x00F0
5999 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
6000 texel_block_expanded_b
6001 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
6002 texel_block_expanded_d
6004 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
6005 [texture_page_ptr, :256]!
6007 subs sub_y, sub_y, #1
6011 add vram_ptr_a, vram_ptr_a, #8
6012 add vram_ptr_b, vram_ptr_b, #8
6014 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6015 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6017 subs tile_x, tile_x, #1
6021 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6022 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6024 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6025 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6027 subs tile_y, tile_y, #1
6031 pop { r3 - r11, pc }
6034 #undef current_texture_page
6037 #define texture_page r1
6038 #define texture_page_ptr r2
6039 #define vram_ptr_a r3
6040 #define texture_page_x r12
6041 #define texture_page_y r4
6042 #define current_texture_page r5
6047 #define vram_ptr_b r11
6059 function(update_texture_8bpp_cache_slice)
6060 stmdb sp!, { r4 - r11, r14 }
6062 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
6063 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
6065 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
6068 and texture_page_x, texture_page, #0xF
6069 mov texture_page_y, texture_page, lsr #4
6071 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6074 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6075 eor current_texture_page, current_texture_page, texture_page
6077 ands current_texture_page, current_texture_page, #0x1
6080 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6083 add vram_ptr_b, vram_ptr_a, #2048
6086 vld1.u32 { texels_a }, [vram_ptr_a, :128], c_4096
6087 vld1.u32 { texels_b }, [vram_ptr_b, :128], c_4096
6088 vld1.u32 { texels_c }, [vram_ptr_a, :128], c_4096
6089 vld1.u32 { texels_d }, [vram_ptr_b, :128], c_4096
6091 vst1.u32 { texels_a, texels_b }, [texture_page_ptr, :256]!
6092 vst1.u32 { texels_c, texels_d }, [texture_page_ptr, :256]!
6094 subs sub_y, sub_y, #1
6099 add vram_ptr_a, vram_ptr_a, #16
6100 add vram_ptr_b, vram_ptr_b, #16
6102 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6103 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6105 subs tile_x, tile_x, #1
6110 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6111 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6113 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6114 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6116 subs tile_y, tile_y, #1
6117 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6121 ldmia sp!, { r4 - r11, pc }
6124 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6125 function(scale2x_tiles8)
6129 add r12, r0, #1024*2
6134 vld1.u16 { q0 }, [r1, :128]!
6135 vld1.u16 { q2 }, [r1, :128]!
6141 vst1.u16 { q0, q1 }, [r0, :128]!
6142 vst1.u16 { q0, q1 }, [r12, :128]!
6144 vst1.u16 { q2, q3 }, [r0, :128]!
6145 vst1.u16 { q2, q3 }, [r12, :128]!
6152 sub r0, r0, r2, lsl #4+1
6154 add r12, r0, #1024*2
6160 // vim:filetype=armasm
notaz.gp2x.de / pcsx_rearmed.git / blob