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.
18 #define MAX_BLOCKS_PER_ROW 128
20 #define RENDER_STATE_MASK_EVALUATE 0x20
21 #define RENDER_FLAGS_MODULATE_TEXELS 0x1
22 #define RENDER_FLAGS_BLEND 0x2
23 #define RENDER_INTERLACE_ENABLED 0x1
25 #include "psx_gpu_offsets.h"
27 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
29 #define edge_data_left_x_offset 0
30 #define edge_data_num_blocks_offset 2
31 #define edge_data_right_mask_offset 4
32 #define edge_data_y_offset 6
38 #define save_abi_regs() \
40 #define restore_abi_regs() \
43 #define save_abi_regs()
44 #define restore_abi_regs()
88 #define gw_by_l gw_bx_l
108 #define uvrg_xxxx0 q3
113 #define uvrg_xxxx1 q4
118 #define uvrg_xxxx2 q5
123 #define yyyy_uvrg0 q6
128 #define yyyy_uvrg1 q7
133 #define yyyy_uvrg2 q8
160 #define uvrg_base q11
203 #define uvrgb_phase q13
207 #include "arm_features.h"
209 #define function(name) FUNCTION(name):
211 #ifndef TEXRELS_FORBIDDEN
213 #define JT_OP_REL(table_label, index_reg, temp)
214 #define JT_OP(x...) x
215 #define JTE(start, target) target
219 #define JT_OP_REL(table_label, index_reg, temp) \
220 adr temp, table_label; \
221 ldr temp, [temp, index_reg, lsl #2]; \
225 #define JTE(start, target) (target - start)
230 #define flush_render_block_buffer _flush_render_block_buffer
231 #define setup_sprite_untextured_simple _setup_sprite_untextured_simple
232 #define update_texture_8bpp_cache _update_texture_8bpp_cache
240 function(compute_all_gradients)
241 // First compute the triangle area reciprocal and shift. The division will
242 // happen concurrently with much of the work which follows.
243 @ r12 = psx_gpu->triangle_area
244 ldr r12, [psx_gpu, #psx_gpu_triangle_area_offset]
245 stmdb sp!, { r4 - r11, lr }
248 @ load exponent of 62 into upper half of double
250 clz r14, r12 @ r14 = shift
252 movt r4, #((62 + 1023) << 4)
253 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
255 @ load area normalized into lower half of double
257 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
259 movt r4, #((1022 + 31) << 4)
262 add r4, r4, r12, lsr #11
265 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
267 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
268 // ( d0 * d1 ) - ( d2 * d3 ) =
269 // ( m0 ) - ( m1 ) = gradient
271 // This is split to do 12 elements at a time over three sets: a, b, and c.
272 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
273 // two of the slots are unused.
275 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
278 // First type is: uvrg bxxx xxxx
279 // Second type is: yyyy ybyy uvrg
280 // Since x_a and y_c are the same the same variable is used for both.
282 vld1.u32 { v0 }, [v_a, :128] @ v0 = { uvrg0, b0, x0, y0 }
283 ldrsh x0, [v_a, #8] @ load x0
285 vld1.u32 { v1 }, [v_b, :128] @ v1 = { uvrg1, b1, x1, y1}
286 ldrh x1, [v_b, #8] @ load x1
288 vld1.u32 { v2 }, [v_c, :128] @ v2 = { uvrg2, b2, x2, y2 }
289 ldrh x2, [v_c, #8] @ load x2
291 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
292 ldrh y0, [v_a, #10] @ load y0
294 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
295 ldrh y1, [v_b, #10] @ load y1
297 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
298 ldrh y2, [v_c, #10] @ load y2
300 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
301 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
303 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
304 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
306 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
307 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
309 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
310 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
312 ldrb b2, [v_c, #4] @ load b2
313 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
315 ldrb b1, [v_b, #4] @ load b1
316 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
318 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
319 vsub.s16 d0_ab, x1_ab, x0_ab
321 ldrb b0, [v_a, #4] @ load b0
322 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
324 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
325 vsub.s16 d2_ab, x2_ab, x1_ab
327 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
328 vsub.s16 d1_ab, y2_ab, y1_ab
330 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
331 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
333 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
334 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
336 vsub.s16 d3_ab, y1_ab, y0_ab
337 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
338 @ ((x2 - X1) * (b1 - b0))
339 vmull.s16 ga_uvrg_x, d0_a, d1_a
340 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
341 @ ((b2 - b1) * (y1 - y0))
342 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
343 movs gs_bx, ga_bx, asr #31
345 vmull.s16 ga_uvrg_y, d0_b, d1_b
346 rsbmi ga_bx, ga_bx, #0
348 @ r12 = psx_gpu->uvrgb_phase
349 ldr r12, [psx_gpu, #psx_gpu_uvrgb_phase_offset]
351 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
352 movs gs_by, ga_by, asr #31
354 vshr.u64 d0, d30, #22
355 add b_base, r12, b0, lsl #16
357 vdup.u32 uvrgb_phase, r12
359 rsbmi ga_by, ga_by, #0
360 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
362 @ r12 = psx_gpu->triangle_winding_offset
363 ldrb r12, [psx_gpu, #psx_gpu_triangle_winding_offset]
364 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
366 rsb r12, r12, #0 @ r12 = -(triangle->winding)
368 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
369 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
371 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
372 vdup.u32 r_shift, r14 @ r_shift = { shift, shift*, shift, shift* }
373 @ * - vshl.u64: ignored by hw
374 vadd.u32 uvrg_base, uvrgb_phase
375 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
377 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
378 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
380 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
381 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
382 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
383 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
385 vshl.u64 gw_rg_x, gw_rg_x, r_shift
386 vshl.u64 gw_uv_x, gw_uv_x, r_shift
387 vshl.u64 gw_rg_y, gw_rg_y, r_shift
388 vshl.u64 gw_uv_y, gw_uv_y, r_shift
390 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
391 vmovn.u64 g_uv_x, gw_uv_x
393 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
394 vmovn.u64 g_rg_x, gw_rg_x
396 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
397 vmovn.u64 g_uv_y, gw_uv_y
399 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
400 vmovn.u64 g_rg_y, gw_rg_y
402 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
403 mov ga_bx, ga_bx, lsl #13
405 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
406 mov ga_by, ga_by, lsl #13
409 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
411 vshl.u32 g_uvrg_x, g_uvrg_x, #4
412 vshl.u32 g_uvrg_y, g_uvrg_y, #4
414 umull gw_by_l, gw_by_h, ga_by, area_r_s
415 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
417 eor gs_bx, gs_bx, r12
418 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
420 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
421 eor gs_by, gs_by, r12
423 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
424 add store_a, psx_gpu, #psx_gpu_uvrg_offset
426 sub r11, r11, #(32 - 13)
428 add store_b, store_a, #16
431 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
432 vst1.u32 { uvrg_base }, [store_a, :128], store_inc
434 vst1.u32 { uvrg_dx1 }, [store_b, :128], store_inc
435 mov g_bx, gw_bx_h, lsr r11
437 vst1.u32 { g_uvrg_y }, [store_a, :128], store_inc
438 mov g_by, gw_by_h, lsr r11
440 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
441 [store_b, :128], store_inc
442 eor g_bx, g_bx, gs_bx
444 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
445 [store_b, :128], store_inc
446 sub g_bx, g_bx, gs_bx
449 eor g_by, g_by, gs_by
451 mls b_base, g_bx, x0, b_base
452 sub g_by, g_by, gs_by
457 add g_bx2, g_bx, g_bx
458 add g_bx3, g_bx, g_bx2
460 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
463 ldmia sp!, { r4 - r11, pc }
480 #define height_minor_a r7
481 #define height_minor_b r8
482 #define height_major r9
485 #define reciprocal_table_ptr r10
487 #define edge_alt_low r4
488 #define edge_alt_high r5
489 #define edge_dx_dy_alt r6
490 #define edge_shift_alt r10
492 #define edge_dx_dy_alt_low r4
493 #define edge_dx_dy_alt_high r5
495 #define span_edge_data r4
496 #define span_uvrg_offset r5
497 #define span_b_offset r6
505 #define alternate_x q0
506 #define alternate_dx_dy q1
507 #define alternate_x_32 q2
509 #define alternate_x_low d0
510 #define alternate_x_high d1
511 #define alternate_dx_dy_low d2
512 #define alternate_dx_dy_high d3
513 #define alternate_x_32_low d4
514 #define alternate_x_32_high d5
518 #define left_dx_dy q5
519 #define right_dx_dy q6
521 #define right_edge q8
523 #define left_x_low d6
524 #define left_x_high d7
525 #define right_x_low d8
526 #define right_x_high d9
527 #define left_dx_dy_low d10
528 #define left_dx_dy_high d11
529 #define right_dx_dy_low d12
530 #define right_dx_dy_high d13
531 #define left_edge_low d14
532 #define left_edge_high d15
533 #define right_edge_low d16
534 #define right_edge_high d17
536 #define y_mid_point d18
539 #define left_right_x_16 q11
540 #define span_shifts_y q12
543 #define span_shifts d24
548 #define left_right_x_16_low d22
549 #define left_right_x_16_high d23
554 #define alternate_x_16 d4
557 #define v_clip_low d6
559 #define right_x_32 q10
560 #define left_x_32 q11
561 #define alternate_select d24
563 #define right_x_32_low d20
564 #define right_x_32_high d21
565 #define left_x_32_low d22
566 #define left_x_32_high d23
569 #define edges_dx_dy d2
570 #define edge_shifts d3
571 #define edge_shifts_64 q2
573 #define edges_xy_left d0
574 #define edges_xy_right d1
576 #define height_reciprocals d6
584 #define heights_b d12
585 #define edges_dx_dy_64 q10
587 #define edges_dx_dy_64_left d20
588 #define edges_dx_dy_64_right d21
591 #define setup_spans_prologue() \
592 stmdb sp!, { r4 - r11, lr }; \
595 ldrsh x_a, [v_a, #8]; \
596 ldrsh x_b, [v_b, #8]; \
597 ldrsh x_c, [v_c, #8]; \
598 ldrsh y_a, [v_a, #10]; \
599 ldrsh y_b, [v_b, #10]; \
600 ldrsh y_c, [v_c, #10]; \
602 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
603 vld1.32 { uvrg }, [temp]; \
604 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
605 vld1.32 { uvrg_dy }, [temp]; \
606 ldr reciprocal_table_ptr, [psx_gpu, #psx_gpu_reciprocal_table_ptr_offset]; \
608 vmov.u32 c_0x01, #0x01 \
610 #define setup_spans_load_b() \
611 ldr b, [psx_gpu, #psx_gpu_b_offset]; \
612 ldr b_dy, [psx_gpu, #psx_gpu_b_dy_offset] \
614 #define setup_spans_prologue_b() \
615 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
616 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
618 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
619 vmov.u16 c_0x0004, #0x0004; \
621 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
622 vmov.u16 c_0x0001, #0x0001; \
624 vld1.u16 { left_edge_low[], left_edge_high[] }, [temp]; \
625 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
627 vld1.u16 { right_edge_low[], right_edge_high[] }, [temp]; \
628 vadd.u16 right_edge, right_edge, c_0x0001; \
630 vmov.u16 c_0x0007, #0x0007; \
631 vmvn.u16 c_0xFFFE, #0x0001 \
634 #define compute_edge_delta_x2() \
635 ldr temp, [reciprocal_table_ptr, height, lsl #2]; \
637 vdup.u32 heights, height; \
638 vsub.u32 widths, x_ends, x_starts; \
640 vdup.u32 edge_shifts, temp; \
641 vsub.u32 heights_b, heights, c_0x01; \
642 vshr.u32 height_reciprocals, edge_shifts, #10; \
644 vmla.s32 heights_b, x_starts, heights; \
645 vbic.u16 edge_shifts, #0xE0; \
646 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
647 vmull.s32 edges_xy, heights_b, height_reciprocals \
650 #define height_reciprocal_alt r11
651 #define height_b_alt r12
653 #define compute_edge_delta_x3(start_c, height_a, height_b) \
654 vmov heights, height_a, height_b; \
655 ldr temp, [reciprocal_table_ptr, height_a, lsl #2]; \
656 vmov.u32 edge_shifts[0], temp; \
657 ldr temp, [reciprocal_table_ptr, height_b, lsl #2]; \
658 vmov.u32 edge_shifts[1], temp; \
659 ldr edge_shift_alt, [reciprocal_table_ptr, height_minor_b, lsl #2]; \
661 vsub.u32 widths, x_ends, x_starts; \
662 sub width_alt, x_c, start_c; \
664 vsub.u32 heights_b, heights, c_0x01; \
665 sub height_b_alt, height_minor_b, #1; \
667 vshr.u32 height_reciprocals, edge_shifts, #10; \
668 lsr height_reciprocal_alt, edge_shift_alt, #10; \
670 vmla.s32 heights_b, x_starts, heights; \
671 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
673 vbic.u16 edge_shifts, #0xE0; \
674 and edge_shift_alt, edge_shift_alt, #0x1F; \
676 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
677 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
679 vmull.s32 edges_xy, heights_b, height_reciprocals; \
680 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
683 #define setup_spans_adjust_y_up() \
684 vsub.u32 y_x4, y_x4, c_0x0004 \
686 #define setup_spans_adjust_y_down() \
687 vadd.u32 y_x4, y_x4, c_0x0004 \
689 #define setup_spans_adjust_interpolants_up() \
690 vsub.u32 uvrg, uvrg, uvrg_dy; \
693 #define setup_spans_adjust_interpolants_down() \
694 vadd.u32 uvrg, uvrg, uvrg_dy; \
698 #define setup_spans_clip_interpolants_increment() \
699 mla b, b_dy, clip, b; \
700 vmla.s32 uvrg, uvrg_dy, v_clip \
702 #define setup_spans_clip_interpolants_decrement() \
703 mls b, b_dy, clip, b; \
704 vmls.s32 uvrg, uvrg_dy, v_clip \
706 #define setup_spans_clip_alternate_yes() \
707 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
709 #define setup_spans_clip_alternate_no() \
711 #define setup_spans_clip(direction, alternate_active) \
712 vdup.u32 v_clip, clip; \
713 setup_spans_clip_alternate_##alternate_active(); \
714 setup_spans_clip_interpolants_##direction(); \
715 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
718 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
719 vmovl.s32 edge_shifts_64, edge_shifts; \
720 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
722 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
723 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
725 vmov left_x_low, edges_xy_##left_index; \
726 vmov right_x_low, edges_xy_##right_index; \
728 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
729 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
730 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
731 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
733 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
734 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
736 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
737 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
740 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
741 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
743 vdup.u16 y_mid_point, y_b; \
744 rsb temp, edge_shift_alt, #32; \
746 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
747 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
748 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
749 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
751 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
752 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
753 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
754 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
756 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
757 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
760 #define setup_spans_y_select_up() \
761 vclt.s16 alternate_select, y_x4, y_mid_point \
763 #define setup_spans_y_select_down() \
764 vcgt.s16 alternate_select, y_x4, y_mid_point \
767 #define setup_spans_alternate_select_left() \
768 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
770 #define setup_spans_alternate_select_right() \
771 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
774 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
775 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
776 vshrn.s64 left_x_32_low, left_x, #32; \
777 vshrn.s64 right_x_32_low, right_x, #32; \
779 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
780 vadd.u64 left_x, left_x, left_dx_dy; \
781 vadd.u64 right_x, right_x, right_dx_dy; \
783 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
784 vshrn.s64 left_x_32_high, left_x, #32; \
785 vshrn.s64 right_x_32_high, right_x, #32; \
787 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
788 vadd.u64 left_x, left_x, left_dx_dy; \
789 vadd.u64 right_x, right_x, right_dx_dy; \
791 vmovn.u32 alternate_x_16, alternate_x_32; \
792 setup_spans_y_select_##direction(); \
793 vmovn.u32 left_right_x_16_low, left_x_32; \
795 vmovn.u32 left_right_x_16_high, right_x_32; \
796 setup_spans_alternate_select_##alternate(); \
798 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
799 str b, [span_b_offset], #4; \
800 setup_spans_adjust_interpolants_##direction(); \
802 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
804 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
805 str b, [span_b_offset], #4; \
806 setup_spans_adjust_interpolants_##direction(); \
808 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
810 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
811 str b, [span_b_offset], #4; \
812 setup_spans_adjust_interpolants_##direction(); \
814 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
815 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
816 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
818 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
819 str b, [span_b_offset], #4; \
820 setup_spans_adjust_interpolants_##direction(); \
822 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
823 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
825 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
827 setup_spans_adjust_y_##direction() \
830 #define setup_spans_set_x4_alternate_no(alternate, direction) \
831 vshrn.s64 left_x_32_low, left_x, #32; \
832 vshrn.s64 right_x_32_low, right_x, #32; \
834 vadd.u64 left_x, left_x, left_dx_dy; \
835 vadd.u64 right_x, right_x, right_dx_dy; \
837 vshrn.s64 left_x_32_high, left_x, #32; \
838 vshrn.s64 right_x_32_high, right_x, #32; \
840 vadd.u64 left_x, left_x, left_dx_dy; \
841 vadd.u64 right_x, right_x, right_dx_dy; \
843 vmovn.u32 left_right_x_16_low, left_x_32; \
844 vmovn.u32 left_right_x_16_high, right_x_32; \
846 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
847 str b, [span_b_offset], #4; \
848 setup_spans_adjust_interpolants_##direction(); \
850 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
852 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
853 str b, [span_b_offset], #4; \
854 setup_spans_adjust_interpolants_##direction(); \
856 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
858 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
859 str b, [span_b_offset], #4; \
860 setup_spans_adjust_interpolants_##direction(); \
862 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
863 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
864 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
866 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
867 str b, [span_b_offset], #4; \
868 setup_spans_adjust_interpolants_##direction(); \
870 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
871 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
873 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
875 setup_spans_adjust_y_##direction() \
878 #define edge_adjust_low r11
879 #define edge_adjust_high r12
881 #define setup_spans_alternate_adjust_yes() \
882 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
883 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
884 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
886 #define setup_spans_alternate_adjust_no() \
889 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
890 setup_spans_alternate_adjust_##alternate_active(); \
891 setup_spans_load_b(); \
893 ldrsh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
894 subs y_c, y_c, temp; \
895 subgt height, height, y_c; \
896 addgt height, height, #1; \
898 ldrsh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
899 subs clip, temp, y_a; \
902 sub height, height, clip; \
903 add y_a, y_a, clip; \
904 setup_spans_clip(increment, alternate_active); \
910 orr temp, y_a, y_a, lsl #16; \
911 add temp, temp, #(1 << 16); \
913 add y_a, y_a, #(2 << 16); \
914 vmov y_x4, temp, y_a; \
916 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
918 setup_spans_prologue_b(); \
920 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
923 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
924 subs height, height, #4; \
930 #define setup_spans_alternate_pre_increment_yes() \
931 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
932 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
934 #define setup_spans_alternate_pre_increment_no() \
937 #define setup_spans_up_decrement_yes() \
938 suble height, height, #1 \
940 #define setup_spans_up_decrement_no() \
943 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
944 setup_spans_alternate_adjust_##alternate_active(); \
945 setup_spans_load_b(); \
948 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
949 subs temp, temp, y_c; \
950 subgt height, height, temp; \
951 setup_spans_up_decrement_##alternate_active(); \
953 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
954 subs clip, y_a, temp; \
957 sub height, height, clip; \
958 sub y_a, y_a, clip; \
959 setup_spans_clip(decrement, alternate_active); \
965 orr temp, y_a, y_a, lsl #16; \
966 sub temp, temp, #(1 << 16); \
968 sub y_a, y_a, #(2 << 16); \
969 vmov y_x4, temp, y_a; \
971 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
973 setup_spans_alternate_pre_increment_##alternate_active(); \
974 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
976 setup_spans_adjust_interpolants_up(); \
977 setup_spans_prologue_b(); \
979 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
982 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
983 subs height, height, #4; \
989 #define setup_spans_epilogue() \
990 restore_abi_regs(); \
991 ldmia sp!, { r4 - r11, pc } \
994 #define setup_spans_up_up(minor, major) \
995 setup_spans_prologue(); \
996 sub height_minor_a, y_a, y_b; \
997 sub height_minor_b, y_b, y_c; \
998 sub height, y_a, y_c; \
1000 vdup.u32 x_starts, x_a; \
1001 vmov x_ends, x_c, x_b; \
1003 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1004 setup_spans_up(major, minor, minor, yes); \
1005 setup_spans_epilogue() \
1007 function(setup_spans_up_left)
1008 setup_spans_up_up(left, right)
1010 function(setup_spans_up_right)
1011 setup_spans_up_up(right, left)
1013 #define setup_spans_down_down(minor, major) \
1014 setup_spans_prologue(); \
1015 sub height_minor_a, y_b, y_a; \
1016 sub height_minor_b, y_c, y_b; \
1017 sub height, y_c, y_a; \
1019 vdup.u32 x_starts, x_a; \
1020 vmov x_ends, x_c, x_b; \
1022 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1023 setup_spans_down(major, minor, minor, yes); \
1024 setup_spans_epilogue() \
1026 function(setup_spans_down_left)
1027 setup_spans_down_down(left, right)
1029 function(setup_spans_down_right)
1030 setup_spans_down_down(right, left)
1033 #define setup_spans_up_flat() \
1034 sub height, y_a, y_c; \
1036 compute_edge_delta_x2(); \
1037 setup_spans_up(left, right, none, no); \
1038 setup_spans_epilogue() \
1040 function(setup_spans_up_a)
1041 setup_spans_prologue()
1043 vmov x_starts, x_a, x_b
1044 vdup.u32 x_ends, x_c
1046 setup_spans_up_flat()
1048 function(setup_spans_up_b)
1049 setup_spans_prologue()
1051 vdup.u32 x_starts, x_a
1052 vmov x_ends, x_b, x_c
1054 setup_spans_up_flat()
1056 #define setup_spans_down_flat() \
1057 sub height, y_c, y_a; \
1059 compute_edge_delta_x2(); \
1060 setup_spans_down(left, right, none, no); \
1061 setup_spans_epilogue() \
1063 function(setup_spans_down_a)
1064 setup_spans_prologue()
1066 vmov x_starts, x_a, x_b
1067 vdup.u32 x_ends, x_c
1069 setup_spans_down_flat()
1071 function(setup_spans_down_b)
1072 setup_spans_prologue()
1074 vdup.u32 x_starts, x_a
1075 vmov x_ends, x_b, x_c
1077 setup_spans_down_flat()
1082 #define edges_xy_b q11
1083 #define edges_dx_dy_b d26
1084 #define edge_shifts_b d27
1085 #define edges_dx_dy_and_shifts_b q13
1086 #define height_increment d20
1088 #define edges_dx_dy_and_shifts q1
1090 #define edges_xy_b_left d22
1091 #define edges_xy_b_right d23
1093 #define setup_spans_up_down_load_edge_set_b() \
1094 vmov edges_xy, edges_xy_b; \
1095 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1098 function(setup_spans_up_down)
1099 setup_spans_prologue()
1101 // s32 middle_y = y_a;
1102 sub height_minor_a, y_a, y_b
1103 sub height_minor_b, y_c, y_a
1104 sub height_major, y_c, y_b
1106 vmov x_starts, x_a, x_c
1107 vdup.u32 x_ends, x_b
1109 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1112 vmov height_increment, temp, height_minor_b
1113 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1115 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1116 vmov edges_xy_b_right, edges_xy_right
1118 vmov edge_shifts_b, edge_shifts
1119 vmov.u32 edge_shifts_b[0], edge_shift_alt
1121 vneg.s32 edges_dx_dy_b, edges_dx_dy
1122 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1126 setup_spans_load_b()
1129 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1130 subs temp, temp, y_b
1131 subgt height_minor_a, height_minor_a, temp
1133 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1134 subs clip, y_a, temp
1137 sub height_minor_a, height_minor_a, clip
1139 setup_spans_clip(decrement, no)
1142 cmp height_minor_a, #0
1145 orr temp, y_a, y_a, lsl #16
1146 sub temp, temp, #(1 << 16)
1148 sub y_a, y_a, #(2 << 16)
1149 vmov y_x4, temp, y_a
1151 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1153 strh height_minor_a, [psx_gpu, #psx_gpu_num_spans_offset]
1155 setup_spans_adjust_edges_alternate_no(left, right);
1156 setup_spans_adjust_interpolants_up()
1157 setup_spans_up_down_load_edge_set_b()
1159 setup_spans_prologue_b()
1163 setup_spans_set_x4_alternate_no(none, up)
1164 subs height_minor_a, height_minor_a, #4
1167 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1168 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1169 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1172 add temp, psx_gpu, #psx_gpu_uvrg_offset
1173 vld1.32 { uvrg }, [temp]
1176 setup_spans_load_b()
1178 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1180 subgt height_minor_b, height_minor_b, y_c
1181 addgt height_minor_b, height_minor_b, #1
1183 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1184 subs clip, temp, y_a
1187 sub height_minor_b, height_minor_b, clip
1189 setup_spans_clip(increment, no)
1192 cmp height_minor_b, #0
1195 orr temp, y_a, y_a, lsl #16
1196 add temp, temp, #(1 << 16)
1198 add y_a, y_a, #(2 << 16)
1199 vmov y_x4, temp, y_a
1201 setup_spans_adjust_edges_alternate_no(left, right)
1203 ldrh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1204 add temp, temp, height_minor_b
1206 cmp temp, #MAX_SPANS
1209 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1212 setup_spans_set_x4_alternate_no(none, down)
1213 subs height_minor_b, height_minor_b, #4
1217 setup_spans_epilogue()
1220 setup_spans_up_down_load_edge_set_b()
1221 setup_spans_prologue_b()
1225 // FIXME: overflow corner case
1226 sub temp, temp, height_minor_b
1227 bics height_minor_b, #3
1228 add temp, temp, height_minor_b
1229 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1233 #undef span_uvrg_offset
1234 #undef span_edge_data
1235 #undef span_b_offset
1240 #define num_spans r1
1241 #define span_uvrg_offset r2
1242 #define span_edge_data r3
1243 #define span_b_offset r4
1245 #define span_num_blocks r6
1249 #define dither_offset_ptr r10
1250 #define block_ptr_a r11
1252 #define num_blocks r14
1254 #define uvrg_dx_ptr r2
1255 #define texture_mask_ptr r3
1256 #define dither_shift r8
1257 #define dither_row r10
1262 #define block_ptr_b r10
1264 #define block_span_ptr r10
1265 #define right_mask r8
1284 #define b_whole_8 d14
1285 #define fb_mask_ptrs d15
1298 #define u_whole_low d16
1299 #define u_whole_high d17
1300 #define v_whole_low d18
1301 #define v_whole_high d19
1302 #define r_whole_low d20
1303 #define r_whole_high d21
1304 #define g_whole_low d22
1305 #define g_whole_high d23
1306 #define b_whole_low d24
1307 #define b_whole_high d25
1312 #define u_whole_8 d26
1313 #define v_whole_8 d27
1314 #define u_whole_8b d24
1315 #define r_whole_8 d24
1316 #define g_whole_8 d25
1318 #define uv_whole_8 q13
1319 #define uv_whole_8b q14
1321 #define dither_offsets q14
1322 #define texture_mask q15
1323 #define texture_mask_u d30
1324 #define texture_mask_v d31
1326 #define dither_offsets_short d28
1330 #define block_span q10
1335 #define draw_mask q1
1336 #define draw_mask_edge q13
1337 #define test_mask q0
1343 #define setup_blocks_texture_swizzled() \
1344 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1345 vsli.u8 u_whole_8, v_whole_8, #4; \
1346 vsri.u8 v_whole_8, u_whole_8b, #4 \
1348 #define setup_blocks_texture_unswizzled() \
1351 #define setup_blocks_shaded_textured_builder(swizzling) \
1354 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1355 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1356 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1358 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1359 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1361 cmp num_spans, #0; \
1364 stmdb sp!, { r4 - r11, r14 }; \
1366 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1368 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
1369 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1371 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1372 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1374 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1375 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1377 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1378 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1380 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1383 vmov.u8 fb_mask_ptrs, #0; \
1385 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1386 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1388 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1389 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1391 cmp span_num_blocks, #0; \
1394 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1395 add num_blocks, span_num_blocks, num_blocks; \
1397 cmp num_blocks, #MAX_BLOCKS; \
1401 ldr b, [span_b_offset]; \
1402 add fb_ptr, fb_ptr, y, lsl #11; \
1404 vdup.u32 v_left_x, left_x; \
1407 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1408 add fb_ptr, fb_ptr, left_x, lsl #1; \
1410 mla b, b_dx, left_x, b; \
1411 and dither_shift, left_x, #0x03; \
1413 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1414 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1416 mov dither_shift, dither_shift, lsl #3; \
1417 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1420 subs span_num_blocks, span_num_blocks, #1; \
1422 mov dither_row, dither_row, ror dither_shift; \
1423 mov b_dx4, b_dx, lsl #2; \
1425 vdup.u32 dither_offsets_short, dither_row; \
1426 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1428 vdup.u32 b_block, b; \
1429 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1431 vdup.u32 u_block, uv[0]; \
1432 mov b_dx8, b_dx, lsl #3; \
1434 vdup.u32 v_block, uv[1]; \
1435 vdup.u32 r_block, rg[0]; \
1436 vdup.u32 g_block, rg[1]; \
1438 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1440 vadd.u32 u_block, u_block, block_span; \
1441 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1443 vadd.u32 v_block, v_block, block_span; \
1444 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1446 vadd.u32 r_block, r_block, block_span; \
1447 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1449 vadd.u32 g_block, g_block, block_span; \
1450 vld1.u32 { block_span }, [block_span_ptr, :128]; \
1452 vadd.u32 b_block, b_block, block_span; \
1453 add block_ptr_b, block_ptr_a, #16; \
1455 vshrn.u32 u_whole_low, u_block, #16; \
1456 vshrn.u32 v_whole_low, v_block, #16; \
1457 vshrn.u32 r_whole_low, r_block, #16; \
1458 vshrn.u32 g_whole_low, g_block, #16; \
1460 vdup.u32 dx4, uv_dx4[0]; \
1461 vshrn.u32 b_whole_low, b_block, #16; \
1463 vaddhn.u32 u_whole_high, u_block, dx4; \
1464 vdup.u32 dx4, uv_dx4[1]; \
1466 vaddhn.u32 v_whole_high, v_block, dx4; \
1467 vdup.u32 dx4, rg_dx4[0]; \
1469 vaddhn.u32 r_whole_high, r_block, dx4; \
1470 vdup.u32 dx4, rg_dx4[1]; \
1472 vaddhn.u32 g_whole_high, g_block, dx4; \
1473 vdup.u32 dx4, b_dx4; \
1475 vaddhn.u32 b_whole_high, b_block, dx4; \
1476 vdup.u32 dx8, uv_dx8[0]; \
1478 vadd.u32 u_block, u_block, dx8; \
1479 vdup.u32 dx8, uv_dx8[1]; \
1481 vadd.u32 v_block, v_block, dx8; \
1482 vdup.u32 dx8, rg_dx8[0]; \
1484 vadd.u32 r_block, r_block, dx8; \
1485 vdup.u32 dx8, rg_dx8[1]; \
1487 vadd.u32 g_block, g_block, dx8; \
1488 vdup.u32 dx8, b_dx8; \
1490 vadd.u32 b_block, b_block, dx8; \
1491 vmovn.u16 u_whole_8, u_whole; \
1493 vmovn.u16 v_whole_8, v_whole; \
1495 vmovn.u16 b_whole_8, b_whole; \
1497 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1499 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1500 setup_blocks_texture_##swizzling(); \
1502 vmovn.u16 r_whole_8, r_whole; \
1506 vmovn.u16 g_whole_8, g_whole; \
1507 vshrn.u32 u_whole_low, u_block, #16; \
1509 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1510 vshrn.u32 v_whole_low, v_block, #16; \
1512 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1513 vshrn.u32 r_whole_low, r_block, #16; \
1515 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1516 vshrn.u32 g_whole_low, g_block, #16; \
1518 vdup.u32 dx4, uv_dx4[0]; \
1519 vshrn.u32 b_whole_low, b_block, #16; \
1521 vaddhn.u32 u_whole_high, u_block, dx4; \
1522 vdup.u32 dx4, uv_dx4[1]; \
1524 vaddhn.u32 v_whole_high, v_block, dx4; \
1525 vdup.u32 dx4, rg_dx4[0]; \
1527 vaddhn.u32 r_whole_high, r_block, dx4; \
1528 vdup.u32 dx4, rg_dx4[1]; \
1530 vaddhn.u32 g_whole_high, g_block, dx4; \
1531 vdup.u32 dx4, b_dx4; \
1533 vaddhn.u32 b_whole_high, b_block, dx4; \
1534 vdup.u32 dx8, uv_dx8[0]; \
1536 vadd.u32 u_block, u_block, dx8; \
1537 vdup.u32 dx8, uv_dx8[1]; \
1539 vadd.u32 v_block, v_block, dx8; \
1540 vdup.u32 dx8, rg_dx8[0]; \
1542 vadd.u32 r_block, r_block, dx8; \
1543 vdup.u32 dx8, rg_dx8[1]; \
1545 vadd.u32 g_block, g_block, dx8; \
1546 vdup.u32 dx8, b_dx8; \
1548 vadd.u32 b_block, b_block, dx8; \
1549 vmovn.u16 u_whole_8, u_whole; \
1551 add fb_ptr, fb_ptr, #16; \
1552 vmovn.u16 v_whole_8, v_whole; \
1554 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1555 vmovn.u16 b_whole_8, b_whole; \
1559 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1560 subs span_num_blocks, span_num_blocks, #1; \
1562 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1563 setup_blocks_texture_##swizzling(); \
1565 vmovn.u16 r_whole_8, r_whole; \
1569 vmovn.u16 g_whole_8, g_whole; \
1570 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1572 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1573 vdup.u8 draw_mask, right_mask; \
1575 vmov.u32 fb_mask_ptrs[0], right_mask; \
1576 vtst.u16 draw_mask, draw_mask, test_mask; \
1577 vzip.u8 u_whole_8, v_whole_8; \
1579 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1580 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1581 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1582 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1583 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1586 add span_uvrg_offset, span_uvrg_offset, #16; \
1587 add span_b_offset, span_b_offset, #4; \
1589 add span_edge_data, span_edge_data, #8; \
1590 subs num_spans, num_spans, #1; \
1592 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1595 restore_abi_regs(); \
1596 ldmia sp!, { r4 - r11, pc }; \
1599 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1600 vpush { texture_mask }; \
1601 vpush { uvrg_dx4 }; \
1603 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; /* r14=num_blocks */ \
1604 bl flush_render_block_buffer; \
1605 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
1607 vpop { uvrg_dx4 }; \
1608 vpop { texture_mask }; \
1610 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1611 vmov.u8 fb_mask_ptrs, #0; \
1613 mov num_blocks, span_num_blocks; \
1614 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1618 setup_blocks_shaded_textured_builder(swizzled)
1619 setup_blocks_shaded_textured_builder(unswizzled)
1622 #define setup_blocks_unshaded_textured_builder(swizzling) \
1625 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1626 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1627 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1629 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1630 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1632 cmp num_spans, #0; \
1635 stmdb sp!, { r4 - r11, r14 }; \
1637 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1639 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1641 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1642 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1644 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1645 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1647 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1649 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1652 vmov.u8 fb_mask_ptrs, #0; \
1654 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1655 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1657 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1658 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1660 cmp span_num_blocks, #0; \
1663 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1664 add num_blocks, span_num_blocks, num_blocks; \
1666 cmp num_blocks, #MAX_BLOCKS; \
1670 add fb_ptr, fb_ptr, y, lsl #11; \
1672 vdup.u32 v_left_x, left_x; \
1675 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1676 add fb_ptr, fb_ptr, left_x, lsl #1; \
1678 and dither_shift, left_x, #0x03; \
1680 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1681 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1683 mov dither_shift, dither_shift, lsl #3; \
1684 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1687 subs span_num_blocks, span_num_blocks, #1; \
1689 mov dither_row, dither_row, ror dither_shift; \
1691 vdup.u32 dither_offsets_short, dither_row; \
1692 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1694 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1696 vdup.u32 u_block, uv[0]; \
1698 vdup.u32 v_block, uv[1]; \
1699 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1701 vadd.u32 u_block, u_block, block_span; \
1702 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1704 vadd.u32 v_block, v_block, block_span; \
1705 add block_ptr_b, block_ptr_a, #16; \
1707 vshrn.u32 u_whole_low, u_block, #16; \
1708 vshrn.u32 v_whole_low, v_block, #16; \
1710 vdup.u32 dx4, uv_dx4[0]; \
1712 vaddhn.u32 u_whole_high, u_block, dx4; \
1713 vdup.u32 dx4, uv_dx4[1]; \
1715 vaddhn.u32 v_whole_high, v_block, dx4; \
1716 vdup.u32 dx8, uv_dx8[0]; \
1718 vadd.u32 u_block, u_block, dx8; \
1719 vdup.u32 dx8, uv_dx8[1]; \
1721 vadd.u32 v_block, v_block, dx8; \
1722 vmovn.u16 u_whole_8, u_whole; \
1724 vmovn.u16 v_whole_8, v_whole; \
1727 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1729 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1730 setup_blocks_texture_##swizzling(); \
1735 vshrn.u32 u_whole_low, u_block, #16; \
1737 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1738 vshrn.u32 v_whole_low, v_block, #16; \
1740 add block_ptr_b, block_ptr_b, #32; \
1741 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1743 vdup.u32 dx4, uv_dx4[0]; \
1744 vaddhn.u32 u_whole_high, u_block, dx4; \
1745 vdup.u32 dx4, uv_dx4[1]; \
1747 vaddhn.u32 v_whole_high, v_block, dx4; \
1748 vdup.u32 dx8, uv_dx8[0]; \
1750 vadd.u32 u_block, u_block, dx8; \
1751 vdup.u32 dx8, uv_dx8[1]; \
1753 vadd.u32 v_block, v_block, dx8; \
1754 vmovn.u16 u_whole_8, u_whole; \
1756 add fb_ptr, fb_ptr, #16; \
1757 vmovn.u16 v_whole_8, v_whole; \
1759 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1762 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1763 subs span_num_blocks, span_num_blocks, #1; \
1765 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1766 setup_blocks_texture_##swizzling(); \
1771 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1773 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1774 vdup.u8 draw_mask, right_mask; \
1776 vmov.u32 fb_mask_ptrs[0], right_mask; \
1777 vtst.u16 draw_mask, draw_mask, test_mask; \
1778 vzip.u8 u_whole_8, v_whole_8; \
1780 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1781 add block_ptr_b, block_ptr_b, #32; \
1782 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1783 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1784 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1787 add span_uvrg_offset, span_uvrg_offset, #16; \
1788 add span_edge_data, span_edge_data, #8; \
1789 subs num_spans, num_spans, #1; \
1791 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1794 restore_abi_regs(); \
1795 ldmia sp!, { r4 - r11, pc }; \
1798 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1799 vpush { texture_mask }; \
1800 vpush { uvrg_dx4 }; \
1802 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; /* r14=num_blocks */ \
1803 bl flush_render_block_buffer; \
1804 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
1806 vpop { uvrg_dx4 }; \
1807 vpop { texture_mask }; \
1809 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1810 vmov.u8 fb_mask_ptrs, #0; \
1812 mov num_blocks, span_num_blocks; \
1813 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1817 setup_blocks_unshaded_textured_builder(swizzled)
1818 setup_blocks_unshaded_textured_builder(unswizzled)
1823 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1824 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1825 veor.u32 draw_mask, draw_mask, draw_mask
1830 stmdb sp!, { r4 - r11, r14 }
1832 vld1.u32 { test_mask }, [psx_gpu, :128]
1834 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1836 ubfx color_r, color, #3, #5
1837 ubfx color_g, color, #11, #5
1838 ubfx color_b, color, #19, #5
1840 orr color, color_r, color_b, lsl #10
1841 orr color, color, color_g, lsl #5
1843 vdup.u16 colors, color
1845 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1846 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1848 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1849 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1852 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1853 ldrh y, [span_edge_data, #edge_data_y_offset]
1855 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1857 cmp span_num_blocks, #0
1860 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1861 add num_blocks, span_num_blocks, num_blocks
1863 cmp num_blocks, #MAX_BLOCKS
1867 add fb_ptr, fb_ptr, y, lsl #11
1870 add fb_ptr, fb_ptr, left_x, lsl #1
1873 subs span_num_blocks, span_num_blocks, #1
1875 add block_ptr_b, block_ptr_a, #16
1878 vmov.u32 fb_mask_ptrs[1], fb_ptr
1882 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_32
1883 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1884 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1886 add fb_ptr, fb_ptr, #16
1887 add block_ptr_b, block_ptr_b, #32
1891 vmov.u32 fb_mask_ptrs[1], fb_ptr
1892 subs span_num_blocks, span_num_blocks, #1
1897 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]
1899 vdup.u8 draw_mask_edge, right_mask
1900 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1902 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1903 vst1.u32 { draw_mask_edge }, [block_ptr_a, :128], c_32
1904 add block_ptr_b, block_ptr_b, #32
1905 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1908 add span_edge_data, span_edge_data, #8
1909 subs num_spans, num_spans, #1
1911 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1915 ldmia sp!, { r4 - r11, pc }
1920 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1921 bl flush_render_block_buffer
1922 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1926 vld1.u32 { test_mask }, [psx_gpu, :128]
1927 veor.u32 draw_mask, draw_mask, draw_mask
1929 mov num_blocks, span_num_blocks
1930 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1934 #define mask_msb_scalar r14
1936 #define msb_mask q15
1938 #define pixels_low d16
1940 #define msb_mask_low d30
1941 #define msb_mask_high d31
1946 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1947 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1952 stmdb sp!, { r4 - r11, r14 }
1954 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1956 ubfx color_r, color, #3, #5
1957 ubfx color_g, color, #11, #5
1959 ldrh mask_msb_scalar, [psx_gpu, #psx_gpu_mask_msb_offset]
1960 ubfx color_b, color, #19, #5
1962 orr color, color_r, color_b, lsl #10
1963 orr color, color, color_g, lsl #5
1964 orr color, color, mask_msb_scalar
1966 vdup.u16 colors, color
1968 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1969 orr color, color, color, lsl #16
1973 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1974 ldrh y, [span_edge_data, #edge_data_y_offset]
1976 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1978 cmp span_num_blocks, #0
1981 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1983 add fb_ptr, fb_ptr, y, lsl #11
1984 subs span_num_blocks, span_num_blocks, #1
1986 add fb_ptr, fb_ptr, left_x, lsl #1
1990 vst1.u32 { colors }, [fb_ptr]!
1991 subs span_num_blocks, span_num_blocks, #1
1996 ldrb right_mask, [span_edge_data, #edge_data_right_mask_offset]
1998 cmp right_mask, #0x0
2001 tst right_mask, #0xF
2002 streq color, [fb_ptr], #4
2003 moveq right_mask, right_mask, lsr #4
2004 streq color, [fb_ptr], #4
2006 tst right_mask, #0x3
2007 streq color, [fb_ptr], #4
2008 moveq right_mask, right_mask, lsr #2
2010 tst right_mask, #0x1
2011 strheq color, [fb_ptr]
2014 add span_edge_data, span_edge_data, #8
2015 subs num_spans, num_spans, #1
2018 ldmia sp!, { r4 - r11, pc }
2021 vst1.u32 { colors }, [fb_ptr]
2028 #define rg_dx_ptr r2
2046 #undef dither_offsets
2066 #define r_whole_low d6
2067 #define r_whole_high d7
2068 #define g_whole_low d8
2069 #define g_whole_high d9
2070 #define b_whole_low d10
2071 #define b_whole_high d11
2073 #define gb_whole_8 q6
2075 #define g_whole_8 d12
2076 #define b_whole_8 d13
2078 #define r_whole_8 d14
2090 #define block_span q5
2098 #define d128_0x7 q13
2102 #define dither_offsets q14
2103 #define draw_mask q15
2105 #define dither_offsets_low d28
2108 #define test_mask q10
2111 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2112 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2113 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2115 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2116 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2117 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2119 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2121 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2124 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2127 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2128 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2129 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2131 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2133 cmp num_spans, #0; \
2136 stmdb sp!, { r4 - r11, r14 }; \
2138 vshl.u32 rg_dx4, rg_dx, #2; \
2140 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2141 vshl.u32 rg_dx8, rg_dx, #3; \
2143 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2145 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2146 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2148 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2149 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2151 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2152 vmov.u8 d64_1, #1; \
2154 vmov.u8 d128_4, #4; \
2155 vmov.u8 d64_128, #128; \
2157 vmov.u8 d128_0x7, #0x7; \
2160 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2161 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2163 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2164 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2166 cmp span_num_blocks, #0; \
2169 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2170 add num_blocks, span_num_blocks, num_blocks; \
2172 cmp num_blocks, #MAX_BLOCKS; \
2176 ldr b, [span_b_offset]; \
2177 add fb_ptr, fb_ptr, y, lsl #11; \
2179 vdup.u32 v_left_x, left_x; \
2182 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2183 add fb_ptr, fb_ptr, left_x, lsl #1; \
2185 mla b, b_dx, left_x, b; \
2186 and dither_shift, left_x, #0x03; \
2188 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2189 vshr.u32 rg_dx, rg_dx4, #2; \
2191 mov dither_shift, dither_shift, lsl #3; \
2192 vmla.u32 rg, rg_dx, v_left_x; \
2195 subs span_num_blocks, span_num_blocks, #1; \
2197 mov dither_row, dither_row, ror dither_shift; \
2198 mov b_dx4, b_dx, lsl #2; \
2200 vdup.u32 dither_offsets, dither_row; \
2201 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2203 vdup.u32 b_block, b; \
2204 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2206 mov b_dx8, b_dx, lsl #3; \
2207 vdup.u32 r_block, rg[0]; \
2208 vdup.u32 g_block, rg[1]; \
2210 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2212 vadd.u32 r_block, r_block, block_span; \
2213 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2215 vadd.u32 g_block, g_block, block_span; \
2216 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2218 vadd.u32 b_block, b_block, block_span; \
2219 add block_ptr_b, block_ptr_a, #16; \
2221 vshrn.u32 r_whole_low, r_block, #16; \
2222 vshrn.u32 g_whole_low, g_block, #16; \
2223 vshrn.u32 b_whole_low, b_block, #16; \
2224 vdup.u32 dx4, rg_dx4[0]; \
2226 vaddhn.u32 r_whole_high, r_block, dx4; \
2227 vdup.u32 dx4, rg_dx4[1]; \
2229 vaddhn.u32 g_whole_high, g_block, dx4; \
2230 vdup.u32 dx4, b_dx4; \
2232 vaddhn.u32 b_whole_high, b_block, dx4; \
2233 vdup.u32 dx8, rg_dx8[0]; \
2235 vadd.u32 r_block, r_block, dx8; \
2236 vdup.u32 dx8, rg_dx8[1]; \
2238 vadd.u32 g_block, g_block, dx8; \
2239 vdup.u32 dx8, b_dx8; \
2241 vadd.u32 b_block, b_block, dx8; \
2243 vmovn.u16 r_whole_8, r_whole; \
2244 vmovn.u16 g_whole_8, g_whole; \
2245 vmovn.u16 b_whole_8, b_whole; \
2248 veor.u32 draw_mask, draw_mask, draw_mask; \
2251 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2252 vshrn.u32 r_whole_low, r_block, #16; \
2254 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2255 vshrn.u32 g_whole_low, g_block, #16; \
2257 vshrn.u32 b_whole_low, b_block, #16; \
2258 str fb_ptr, [block_ptr_a, #44]; \
2260 vdup.u32 dx4, rg_dx4[0]; \
2261 vshr.u8 r_whole_8, r_whole_8, #3; \
2262 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2264 vaddhn.u32 r_whole_high, r_block, dx4; \
2265 vdup.u32 dx4, rg_dx4[1]; \
2267 vaddhn.u32 g_whole_high, g_block, dx4; \
2268 vdup.u32 dx4, b_dx4; \
2270 vaddhn.u32 b_whole_high, b_block, dx4; \
2271 vdup.u32 dx8, rg_dx8[0]; \
2273 vmull.u8 pixels, r_whole_8, d64_1; \
2274 vmlal.u8 pixels, g_whole_8, d64_4; \
2275 vmlal.u8 pixels, b_whole_8, d64_128; \
2277 vadd.u32 r_block, r_block, dx8; \
2278 vdup.u32 dx8, rg_dx8[1]; \
2280 vadd.u32 g_block, g_block, dx8; \
2281 vdup.u32 dx8, b_dx8; \
2283 vadd.u32 b_block, b_block, dx8; \
2284 add fb_ptr, fb_ptr, #16; \
2286 vmovn.u16 r_whole_8, r_whole; \
2287 vmovn.u16 g_whole_8, g_whole; \
2288 vmovn.u16 b_whole_8, b_whole; \
2290 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2291 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2295 subs span_num_blocks, span_num_blocks, #1; \
2299 str fb_ptr, [block_ptr_a, #44]; \
2300 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2302 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2303 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2305 vshr.u8 r_whole_8, r_whole_8, #3; \
2306 vdup.u8 draw_mask, right_mask; \
2308 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2309 vld1.u32 { test_mask }, [psx_gpu, :128]; \
2311 vtst.u16 draw_mask, draw_mask, test_mask; \
2313 vmull.u8 pixels, r_whole_8, d64_1; \
2314 vmlal.u8 pixels, g_whole_8, d64_4; \
2315 vmlal.u8 pixels, b_whole_8, d64_128; \
2317 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2318 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2321 add span_uvrg_offset, span_uvrg_offset, #16; \
2322 add span_b_offset, span_b_offset, #4; \
2324 add span_edge_data, span_edge_data, #8; \
2325 subs num_spans, num_spans, #1; \
2327 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2330 restore_abi_regs(); \
2331 ldmia sp!, { r4 - r11, pc }; \
2334 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2337 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2338 bl flush_render_block_buffer; \
2339 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2343 vmov.u8 d64_1, #1; \
2344 vmov.u8 d128_4, #4; \
2345 vmov.u8 d64_128, #128; \
2346 vmov.u8 d128_0x7, #0x7; \
2348 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2350 mov num_blocks, span_num_blocks; \
2351 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2355 setup_blocks_shaded_untextured_indirect_builder(undithered)
2356 setup_blocks_shaded_untextured_indirect_builder(dithered)
2361 #define mask_msb_ptr r14
2363 #define draw_mask q0
2364 #define pixels_low d16
2365 #define pixels_high d17
2369 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2372 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2373 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2374 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2376 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2378 cmp num_spans, #0; \
2381 stmdb sp!, { r4 - r11, r14 }; \
2383 vshl.u32 rg_dx4, rg_dx, #2; \
2385 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2386 vshl.u32 rg_dx8, rg_dx, #3; \
2388 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2389 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2391 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2392 vmov.u8 d64_1, #1; \
2394 vmov.u8 d128_4, #4; \
2395 vmov.u8 d64_128, #128; \
2397 vmov.u8 d128_0x7, #0x7; \
2398 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2399 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
2402 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2403 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2405 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2406 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2408 cmp span_num_blocks, #0; \
2411 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2412 add fb_ptr, fb_ptr, y, lsl #11; \
2414 ldr b, [span_b_offset]; \
2415 vdup.u32 v_left_x, left_x; \
2418 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2419 add fb_ptr, fb_ptr, left_x, lsl #1; \
2421 mla b, b_dx, left_x, b; \
2422 and dither_shift, left_x, #0x03; \
2424 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2425 vshr.u32 rg_dx, rg_dx4, #2; \
2427 mov dither_shift, dither_shift, lsl #3; \
2428 vmla.u32 rg, rg_dx, v_left_x; \
2430 subs span_num_blocks, span_num_blocks, #1; \
2432 mov dither_row, dither_row, ror dither_shift; \
2433 mov b_dx4, b_dx, lsl #2; \
2435 vdup.u32 dither_offsets, dither_row; \
2436 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2438 vdup.u32 b_block, b; \
2439 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2441 mov b_dx8, b_dx, lsl #3; \
2442 vdup.u32 r_block, rg[0]; \
2443 vdup.u32 g_block, rg[1]; \
2445 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2447 vadd.u32 r_block, r_block, block_span; \
2448 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2450 vadd.u32 g_block, g_block, block_span; \
2451 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2453 vadd.u32 b_block, b_block, block_span; \
2454 add block_ptr_b, block_ptr_a, #16; \
2456 vshrn.u32 r_whole_low, r_block, #16; \
2457 vshrn.u32 g_whole_low, g_block, #16; \
2458 vshrn.u32 b_whole_low, b_block, #16; \
2459 vdup.u32 dx4, rg_dx4[0]; \
2461 vaddhn.u32 r_whole_high, r_block, dx4; \
2462 vdup.u32 dx4, rg_dx4[1]; \
2464 vaddhn.u32 g_whole_high, g_block, dx4; \
2465 vdup.u32 dx4, b_dx4; \
2467 vaddhn.u32 b_whole_high, b_block, dx4; \
2468 vdup.u32 dx8, rg_dx8[0]; \
2470 vadd.u32 r_block, r_block, dx8; \
2471 vdup.u32 dx8, rg_dx8[1]; \
2473 vadd.u32 g_block, g_block, dx8; \
2474 vdup.u32 dx8, b_dx8; \
2476 vadd.u32 b_block, b_block, dx8; \
2478 vmovn.u16 r_whole_8, r_whole; \
2479 vmovn.u16 g_whole_8, g_whole; \
2480 vmovn.u16 b_whole_8, b_whole; \
2485 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2486 vshrn.u32 r_whole_low, r_block, #16; \
2488 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2489 vshrn.u32 g_whole_low, g_block, #16; \
2491 vshrn.u32 b_whole_low, b_block, #16; \
2493 vdup.u32 dx4, rg_dx4[0]; \
2494 vshr.u8 r_whole_8, r_whole_8, #3; \
2495 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2497 vaddhn.u32 r_whole_high, r_block, dx4; \
2498 vdup.u32 dx4, rg_dx4[1]; \
2500 vmov pixels, msb_mask; \
2501 vaddhn.u32 g_whole_high, g_block, dx4; \
2502 vdup.u32 dx4, b_dx4; \
2504 vaddhn.u32 b_whole_high, b_block, dx4; \
2505 vdup.u32 dx8, rg_dx8[0]; \
2507 vmlal.u8 pixels, r_whole_8, d64_1; \
2508 vmlal.u8 pixels, g_whole_8, d64_4; \
2509 vmlal.u8 pixels, b_whole_8, d64_128; \
2511 vadd.u32 r_block, r_block, dx8; \
2512 vdup.u32 dx8, rg_dx8[1]; \
2514 vadd.u32 g_block, g_block, dx8; \
2515 vdup.u32 dx8, b_dx8; \
2517 vadd.u32 b_block, b_block, dx8; \
2519 vmovn.u16 r_whole_8, r_whole; \
2520 vmovn.u16 g_whole_8, g_whole; \
2521 vmovn.u16 b_whole_8, b_whole; \
2523 vst1.u32 { pixels }, [fb_ptr]!; \
2524 subs span_num_blocks, span_num_blocks, #1; \
2528 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2530 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2531 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2533 vshr.u8 r_whole_8, r_whole_8, #3; \
2534 rbit right_mask, right_mask; \
2535 vmov pixels, msb_mask; \
2536 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2537 clz right_mask, right_mask; \
2539 vmlal.u8 pixels, r_whole_8, d64_1; \
2540 vmlal.u8 pixels, g_whole_8, d64_4; \
2541 vmlal.u8 pixels, b_whole_8, d64_128; \
2543 JT_OP_REL(100f, right_mask, temp); \
2544 JT_OP(ldr pc, [pc, right_mask, lsl #2]); \
2548 .word JTE(100b, 4f); \
2549 .word JTE(100b, 5f); \
2550 .word JTE(100b, 6f); \
2551 .word JTE(100b, 7f); \
2552 .word JTE(100b, 8f); \
2553 .word JTE(100b, 9f); \
2554 .word JTE(100b, 10f); \
2555 .word JTE(100b, 11f); \
2558 vst1.u16 { pixels_low[0] }, [fb_ptr]; \
2562 vst1.u32 { pixels_low[0] }, [fb_ptr]; \
2566 vst1.u32 { pixels_low[0] }, [fb_ptr]!; \
2567 vst1.u16 { pixels_low[2] }, [fb_ptr]; \
2571 vst1.u32 { pixels_low }, [fb_ptr]; \
2575 vst1.u32 { pixels_low }, [fb_ptr]!; \
2576 vst1.u16 { pixels_high[0] }, [fb_ptr]; \
2580 vst1.u32 { pixels_low }, [fb_ptr]!; \
2581 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2585 vst1.u32 { pixels_low }, [fb_ptr]!; \
2586 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2587 vst1.u16 { pixels_high[2] }, [fb_ptr]; \
2591 vst1.u32 { pixels }, [fb_ptr]; \
2595 add span_uvrg_offset, span_uvrg_offset, #16; \
2596 add span_b_offset, span_b_offset, #4; \
2598 add span_edge_data, span_edge_data, #8; \
2599 subs num_spans, num_spans, #1; \
2603 restore_abi_regs(); \
2604 ldmia sp!, { r4 - r11, pc } \
2606 setup_blocks_shaded_untextured_direct_builder(undithered)
2607 setup_blocks_shaded_untextured_direct_builder(dithered)
2616 #define block_ptr r1
2617 #define num_blocks r2
2630 #define texture_ptr r11
2644 #define pixels_d r10
2648 #define clut_ptr r12
2649 #define current_texture_mask r5
2650 #define dirty_textures_mask r6
2654 #define clut_low_a d2
2655 #define clut_low_b d3
2656 #define clut_high_a d4
2657 #define clut_high_b d5
2662 #define texels_low d6
2663 #define texels_high d7
2667 function(texture_blocks_untextured)
2673 function(texture_blocks_4bpp)
2674 stmdb sp!, { r3 - r11, r14 }
2675 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2677 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2678 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2680 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2681 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]
2683 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2684 vuzp.u8 clut_a, clut_b
2686 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
2687 tst dirty_textures_mask, current_texture_mask
2693 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2695 uxtah uv_0, texture_ptr, uv_01
2696 uxtah uv_1, texture_ptr, uv_01, ror #16
2698 uxtah uv_2, texture_ptr, uv_23
2699 uxtah uv_3, texture_ptr, uv_23, ror #16
2701 uxtah uv_4, texture_ptr, uv_45
2702 ldrb pixel_0, [uv_0]
2704 uxtah uv_5, texture_ptr, uv_45, ror #16
2705 ldrb pixel_1, [uv_1]
2707 uxtah uv_6, texture_ptr, uv_67
2708 ldrb pixel_2, [uv_2]
2710 uxtah uv_7, texture_ptr, uv_67, ror #16
2711 ldrb pixel_3, [uv_3]
2713 ldrb pixel_4, [uv_4]
2714 subs num_blocks, num_blocks, #1
2716 ldrb pixel_5, [uv_5]
2717 orr pixels_a, pixel_0, pixel_1, lsl #8
2719 ldrb pixel_6, [uv_6]
2720 orr pixels_b, pixel_4, pixel_5, lsl #8
2722 ldrb pixel_7, [uv_7]
2723 orr pixels_a, pixels_a, pixel_2, lsl #16
2725 orr pixels_b, pixels_b, pixel_6, lsl #16
2726 orr pixels_a, pixels_a, pixel_3, lsl #24
2728 orr pixels_b, pixels_b, pixel_7, lsl #24
2729 vmov texels, pixels_a, pixels_b
2731 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2732 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2734 vst2.u8 { texels_low, texels_high }, [block_ptr, :128], c_64
2737 ldmia sp!, { r3 - r11, pc }
2740 stmdb sp!, { r1 - r2 }
2741 bl update_texture_4bpp_cache
2744 ldmia sp!, { r1 - r2 }
2750 function(texture_blocks_8bpp)
2751 stmdb sp!, { r3 - r11, r14 }
2752 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2754 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2755 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2757 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2758 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2760 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]
2761 tst dirty_textures_mask, current_texture_mask
2767 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2769 uxtah uv_0, texture_ptr, uv_01
2770 uxtah uv_1, texture_ptr, uv_01, ror #16
2772 uxtah uv_2, texture_ptr, uv_23
2773 uxtah uv_3, texture_ptr, uv_23, ror #16
2775 uxtah uv_4, texture_ptr, uv_45
2776 ldrb pixel_0, [uv_0]
2778 uxtah uv_5, texture_ptr, uv_45, ror #16
2779 ldrb pixel_1, [uv_1]
2781 uxtah uv_6, texture_ptr, uv_67
2782 ldrb pixel_2, [uv_2]
2784 uxtah uv_7, texture_ptr, uv_67, ror #16
2785 ldrb pixel_3, [uv_3]
2787 ldrb pixel_4, [uv_4]
2788 add pixel_0, pixel_0, pixel_0
2790 ldrb pixel_5, [uv_5]
2791 add pixel_1, pixel_1, pixel_1
2793 ldrb pixel_6, [uv_6]
2794 add pixel_2, pixel_2, pixel_2
2796 ldrb pixel_7, [uv_7]
2797 add pixel_3, pixel_3, pixel_3
2799 ldrh pixel_0, [clut_ptr, pixel_0]
2800 add pixel_4, pixel_4, pixel_4
2802 ldrh pixel_1, [clut_ptr, pixel_1]
2803 add pixel_5, pixel_5, pixel_5
2805 ldrh pixel_2, [clut_ptr, pixel_2]
2806 add pixel_6, pixel_6, pixel_6
2808 ldrh pixel_3, [clut_ptr, pixel_3]
2809 add pixel_7, pixel_7, pixel_7
2811 ldrh pixel_4, [clut_ptr, pixel_4]
2812 orr pixels_a, pixel_0, pixel_1, lsl #16
2814 ldrh pixel_5, [clut_ptr, pixel_5]
2815 orr pixels_c, pixel_2, pixel_3, lsl #16
2817 ldrh pixel_6, [clut_ptr, pixel_6]
2818 subs num_blocks, num_blocks, #1
2820 ldrh pixel_7, [clut_ptr, pixel_7]
2821 orr pixels_b, pixel_4, pixel_5, lsl #16
2823 orr pixels_d, pixel_6, pixel_7, lsl #16
2824 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2826 add block_ptr, block_ptr, #64
2829 ldmia sp!, { r3 - r11, pc }
2832 stmdb sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2834 bl update_texture_8bpp_cache
2836 ldmia sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2866 #define block_ptr r1
2867 #define num_blocks r2
2911 #define texture_ptr r12
2916 function(texture_blocks_16bpp)
2917 stmdb sp!, { r3 - r11, r14 }
2918 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2920 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2921 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2924 ldrh uv_0, [block_ptr]
2925 subs num_blocks, num_blocks, #1
2927 ldrh uv_1, [block_ptr, #2]
2929 and v_0, uv_0, #0xFF00
2930 and v_1, uv_1, #0xFF00
2932 and u_0, uv_0, #0xFF
2933 and u_1, uv_1, #0xFF
2935 add uv_0, u_0, v_0, lsl #2
2936 ldrh uv_2, [block_ptr, #4]
2938 add uv_1, u_1, v_1, lsl #2
2939 ldrh uv_3, [block_ptr, #6]
2941 add uv_0, uv_0, uv_0
2942 add uv_1, uv_1, uv_1
2944 and v_2, uv_2, #0xFF00
2945 and v_3, uv_3, #0xFF00
2947 and u_2, uv_2, #0xFF
2948 and u_3, uv_3, #0xFF
2950 add uv_2, u_2, v_2, lsl #2
2951 ldrh uv_4, [block_ptr, #8]
2953 add uv_3, u_3, v_3, lsl #2
2954 ldrh uv_5, [block_ptr, #10]
2956 add uv_2, uv_2, uv_2
2957 add uv_3, uv_3, uv_3
2959 and v_4, uv_4, #0xFF00
2960 and v_5, uv_5, #0xFF00
2962 and u_4, uv_4, #0xFF
2963 and u_5, uv_5, #0xFF
2965 add uv_4, u_4, v_4, lsl #2
2966 ldrh uv_6, [block_ptr, #12]
2968 add uv_5, u_5, v_5, lsl #2
2969 ldrh uv_7, [block_ptr, #14]
2971 add uv_4, uv_4, uv_4
2972 ldrh pixel_0, [texture_ptr, uv_0]
2974 add uv_5, uv_5, uv_5
2975 ldrh pixel_1, [texture_ptr, uv_1]
2977 and v_6, uv_6, #0xFF00
2978 ldrh pixel_2, [texture_ptr, uv_2]
2980 and v_7, uv_7, #0xFF00
2981 ldrh pixel_3, [texture_ptr, uv_3]
2983 and u_6, uv_6, #0xFF
2984 ldrh pixel_4, [texture_ptr, uv_4]
2986 and u_7, uv_7, #0xFF
2987 ldrh pixel_5, [texture_ptr, uv_5]
2989 add uv_6, u_6, v_6, lsl #2
2990 add uv_7, u_7, v_7, lsl #2
2992 add uv_6, uv_6, uv_6
2993 add uv_7, uv_7, uv_7
2995 orr pixels_a, pixel_0, pixel_1, lsl #16
2996 orr pixels_b, pixel_2, pixel_3, lsl #16
2998 ldrh pixel_6, [texture_ptr, uv_6]
2999 orr pixels_c, pixel_4, pixel_5, lsl #16
3001 ldrh pixel_7, [texture_ptr, uv_7]
3002 orr pixels_d, pixel_6, pixel_7, lsl #16
3004 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
3005 add block_ptr, block_ptr, #64
3009 ldmia sp!, { r3 - r11, pc }
3024 #undef msb_mask_high
3032 #define num_blocks r1
3033 #define color_ptr r2
3034 #define colors_scalar r2
3035 #define colors_scalar_compare r3
3036 #define mask_msb_ptr r2
3038 #define block_ptr_load_a r0
3039 #define block_ptr_store r3
3040 #define block_ptr_load_b r12
3045 #define draw_mask_bits_scalar r5
3047 #define d128_0x07 q0
3048 #define d128_0x1F q1
3049 #define d128_0x8000 q2
3050 #define test_mask q3
3052 #define colors_rg q5
3053 #define colors_b_dm_bits q6
3054 #define texels_rg q7
3057 #define pixels_b q10
3059 #define zero_mask q4
3060 #define draw_mask q12
3061 #define msb_mask q13
3063 #define fb_pixels q8
3065 #define pixels_gb_low q9
3067 #define colors_r d10
3068 #define colors_g d11
3069 #define colors_b d12
3070 #define draw_mask_bits d13
3071 #define texels_r d14
3072 #define texels_g d15
3073 #define pixels_r_low d16
3074 #define pixels_g_low d18
3075 #define pixels_b_low d19
3076 #define msb_mask_low d26
3077 #define msb_mask_high d27
3082 #define texels_b d31
3084 #define shade_blocks_textured_modulated_prologue_indirect() \
3086 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3088 #define shade_blocks_textured_modulated_prologue_direct() \
3089 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3090 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] \
3093 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3095 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3096 ldr colors_scalar, [psx_gpu, #psx_gpu_triangle_color_offset]; \
3097 movw colors_scalar_compare, #0x8080; \
3099 movt colors_scalar_compare, #0x80; \
3100 cmp colors_scalar, colors_scalar_compare; \
3101 beq shade_blocks_textured_unmodulated_##target \
3103 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3105 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3106 shade_blocks_textured_false_modulation_check_##dithering(target); \
3107 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3108 vld1.u32 { colors_r[] }, [color_ptr, :32]; \
3109 vdup.u8 colors_g, colors_r[1]; \
3110 vdup.u8 colors_b, colors_r[2]; \
3111 vdup.u8 colors_r, colors_r[0] \
3114 #define shade_blocks_textured_modulated_load_dithered(target) \
3115 vld1.u32 { target }, [block_ptr_load_b, :128] \
3117 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3118 vld1.u32 { target }, [block_ptr_load_b, :128], c_32 \
3120 #define shade_blocks_textured_modulated_load_undithered(target) \
3122 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3123 add block_ptr_load_b, block_ptr_load_b, #32 \
3125 #define shade_blocks_textured_modulate_dithered(channel) \
3126 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3128 #define shade_blocks_textured_modulate_undithered(channel) \
3129 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3132 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3133 vst1.u32 { draw_mask }, [block_ptr_store, :128]! \
3135 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3136 ldr fb_ptr, [block_ptr_load_b, #(offset - 64)]; \
3137 vld1.u32 { fb_pixels }, [fb_ptr]; \
3138 vbit.u16 pixels, fb_pixels, draw_mask \
3140 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3141 vst1.u32 { pixels }, [block_ptr_store, :128], c_48 \
3143 #define shade_blocks_textured_modulated_store_pixels_direct() \
3144 vst1.u32 { pixels }, [fb_ptr] \
3147 #define shade_blocks_textured_modulated_load_rg_shaded() \
3148 vld1.u32 { colors_r, colors_g }, [block_ptr_load_b, :128], c_32 \
3150 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3151 add block_ptr_load_b, block_ptr_load_b, #32 \
3153 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3154 vld1.u32 { colors_b, draw_mask_bits }, [block_ptr_load_a, :128], c_32 \
3156 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3157 ldr draw_mask_bits_scalar, [block_ptr_load_a, #8]; \
3158 add block_ptr_load_a, block_ptr_load_a, #32 \
3160 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3161 vdup.u16 draw_mask, draw_mask_bits[0] \
3163 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3164 vdup.u16 draw_mask, draw_mask_bits_scalar \
3167 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3169 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3170 vorr.u16 pixels, pixels, msb_mask \
3173 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3176 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3178 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3179 stmdb sp!, { r4 - r5, lr }; \
3180 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3182 vld1.u32 { test_mask }, [psx_gpu, :128]; \
3184 shade_blocks_textured_modulated_prologue_##target(); \
3186 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3189 add block_ptr_load_b, block_ptr_load_a, #16; \
3190 vmov.u8 d64_1, #1; \
3191 vmov.u8 d64_4, #4; \
3192 vmov.u8 d64_128, #128; \
3194 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3195 vmov.u8 d128_0x07, #0x07; \
3197 shade_blocks_textured_modulated_load_rg_##shading(); \
3198 vmov.u8 d128_0x1F, #0x1F; \
3200 shade_blocks_textured_modulated_load_bdm_##shading(); \
3201 vmov.u16 d128_0x8000, #0x8000; \
3203 vmovn.u16 texels_r, texels; \
3204 vshrn.u16 texels_g, texels, #5; \
3206 vshrn.u16 texels_b, texels, #7; \
3207 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3209 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3210 vtst.u16 draw_mask, draw_mask, test_mask; \
3212 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3213 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3215 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3216 vshr.u8 texels_b, texels_b, #3; \
3218 shade_blocks_textured_modulate_##dithering(r); \
3219 shade_blocks_textured_modulate_##dithering(g); \
3220 shade_blocks_textured_modulate_##dithering(b); \
3222 vand.u16 pixels, texels, d128_0x8000; \
3223 vceq.u16 zero_mask, texels, #0; \
3225 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3226 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3227 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3229 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3230 vorr.u16 draw_mask, draw_mask, zero_mask; \
3231 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3232 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3234 subs num_blocks, num_blocks, #1; \
3240 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3241 shade_blocks_textured_modulated_load_rg_##shading(); \
3242 vshrn.u16 texels_g, texels, #5; \
3244 shade_blocks_textured_modulated_load_bdm_##shading(); \
3245 vshrn.u16 texels_b, texels, #7; \
3247 pld [block_ptr_load_a]; \
3248 vmovn.u16 texels_r, texels; \
3249 vmlal.u8 pixels, pixels_r_low, d64_1; \
3251 vmlal.u8 pixels, pixels_g_low, d64_4; \
3252 vmlal.u8 pixels, pixels_b_low, d64_128; \
3253 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3255 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3256 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3258 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3259 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3261 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3262 vtst.u16 draw_mask, draw_mask, test_mask; \
3264 shade_blocks_textured_modulated_store_pixels_##target(); \
3265 vshr.u8 texels_b, texels_b, #3; \
3267 shade_blocks_textured_modulate_##dithering(r); \
3268 shade_blocks_textured_modulate_##dithering(g); \
3269 shade_blocks_textured_modulate_##dithering(b); \
3271 vand.u16 pixels, texels, d128_0x8000; \
3272 vceq.u16 zero_mask, texels, #0; \
3274 subs num_blocks, num_blocks, #1; \
3276 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3277 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3278 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3280 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3281 vorr.u16 draw_mask, draw_mask, zero_mask; \
3282 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3283 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3288 vmlal.u8 pixels, pixels_r_low, d64_1; \
3289 vmlal.u8 pixels, pixels_g_low, d64_4; \
3290 vmlal.u8 pixels, pixels_b_low, d64_128; \
3292 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3293 shade_blocks_textured_modulated_store_pixels_##target(); \
3295 ldmia sp!, { r4 - r5, lr }; \
3296 restore_abi_regs(); \
3300 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3301 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3302 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3303 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3305 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3306 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3307 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3308 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3326 #undef msb_mask_high
3329 #define num_blocks r1
3330 #define mask_msb_ptr r2
3331 #define color_ptr r3
3333 #define block_ptr_load r0
3334 #define draw_mask_store_ptr r3
3335 #define draw_mask_bits_ptr r12
3336 #define draw_mask_ptr r12
3337 #define pixel_store_ptr r14
3339 #define fb_ptr_cmp r4
3342 #define fb_ptr_next r14
3346 #define test_mask q0
3348 #define draw_mask q2
3349 #define zero_mask q3
3350 #define draw_mask_combined q4
3351 #define fb_pixels q5
3352 #define fb_pixels_next q6
3355 #define draw_mask_low d4
3356 #define draw_mask_high d5
3357 #define msb_mask_low d14
3358 #define msb_mask_high d15
3361 function(shade_blocks_textured_unmodulated_indirect)
3362 stmdb sp!, { r4, r14 }
3364 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3366 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3367 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3369 vld1.u32 { test_mask }, [psx_gpu, :128]
3370 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3373 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3375 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3376 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3377 [draw_mask_bits_ptr, :16], c_64
3378 vceq.u16 zero_mask, pixels, #0
3380 vtst.u16 draw_mask, draw_mask, test_mask
3381 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3383 subs num_blocks, num_blocks, #1
3387 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3388 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3390 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3391 [draw_mask_bits_ptr, :16], c_64
3392 vceq.u16 zero_mask, pixels, #0
3394 vtst.u16 draw_mask, draw_mask, test_mask
3395 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3397 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3398 subs num_blocks, num_blocks, #1
3403 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3404 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3407 ldmia sp!, { r4, pc }
3412 function(shade_blocks_textured_unmodulated_direct)
3413 stmdb sp!, { r4, r14 }
3415 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3417 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3418 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3420 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3423 vld1.u32 { test_mask }, [psx_gpu, :128]
3424 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3426 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3427 [draw_mask_bits_ptr, :16], c_64
3428 ldr fb_ptr_next, [block_ptr_load, #44]
3430 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3431 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3432 vceq.u16 zero_mask, pixels, #0
3433 vtst.u16 draw_mask, draw_mask, test_mask
3435 subs num_blocks, num_blocks, #1
3439 mov fb_ptr, fb_ptr_next
3440 ldr fb_ptr_next, [block_ptr_load, #44]
3442 vorr.u16 pixels, pixels, msb_mask
3444 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3445 vmov fb_pixels, fb_pixels_next
3447 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3448 [draw_mask_bits_ptr, :16], c_64
3449 vbif.u16 fb_pixels, pixels, draw_mask_combined
3451 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3452 pld [fb_ptr_next, #64]
3454 add fb_ptr_cmp, fb_ptr_cmp, #14
3455 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3460 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3461 vceq.u16 zero_mask, pixels, #0
3463 vst1.u16 { fb_pixels }, [fb_ptr]
3464 vtst.u16 draw_mask, draw_mask, test_mask
3467 subs num_blocks, num_blocks, #1
3471 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3472 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3474 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3477 ldmia sp!, { r4, pc }
3480 vst1.u16 { fb_pixels }, [fb_ptr]
3481 vceq.u16 zero_mask, pixels, #0
3483 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3484 vtst.u16 draw_mask, draw_mask, test_mask
3489 function(shade_blocks_unshaded_untextured_indirect)
3494 function(shade_blocks_unshaded_untextured_direct)
3495 stmdb sp!, { r4, r14 }
3497 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3499 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3500 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3502 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3503 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3505 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3506 vld1.u16 { pixels }, [color_ptr, :128]
3509 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3511 vorr.u16 pixels, pixels, msb_mask
3512 subs num_blocks, num_blocks, #1
3514 ldr fb_ptr_next, [block_ptr_load], #64
3516 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3520 vmov fb_pixels, fb_pixels_next
3521 mov fb_ptr, fb_ptr_next
3522 ldr fb_ptr_next, [block_ptr_load], #64
3524 vbif.u16 fb_pixels, pixels, draw_mask
3525 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3527 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3528 add fb_ptr_cmp, fb_ptr_cmp, #14
3532 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3533 vst1.u16 { fb_pixels }, [fb_ptr]
3536 subs num_blocks, num_blocks, #1
3540 vbif.u16 fb_pixels_next, pixels, draw_mask
3541 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3544 ldmia sp!, { r4, pc }
3547 vst1.u16 { fb_pixels }, [fb_ptr]
3548 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3552 #undef draw_mask_ptr
3559 #define num_blocks r1
3560 #define msb_mask_ptr r2
3561 #define pixel_ptr r3
3562 #define draw_mask_ptr r0
3565 #define fb_ptr_next r14
3566 #define fb_ptr_cmp r4
3574 #undef msb_mask_high
3575 #undef draw_mask_next
3578 #undef fb_pixels_next
3581 #define draw_mask q1
3583 #define fb_pixels q3
3584 #define blend_pixels q4
3585 #define pixels_no_msb q5
3586 #define blend_mask q6
3587 #define fb_pixels_no_msb q7
3588 #define d128_0x8000 q8
3589 #define d128_0x0421 q9
3590 #define fb_pixels_next q10
3591 #define blend_pixels_next q11
3592 #define pixels_next q12
3593 #define draw_mask_next q13
3594 #define write_mask q14
3596 #define pixels_rb q5
3597 #define pixels_mg q7
3599 #define d128_0x7C1F q8
3600 #define d128_0x03E0 q9
3601 #define fb_pixels_rb q10
3602 #define fb_pixels_g q11
3603 #define fb_pixels_masked q11
3604 #define d128_0x83E0 q15
3605 #define pixels_fourth q7
3606 #define d128_0x1C07 q12
3607 #define d128_0x00E0 q13
3608 #define d128_0x80E0 q13
3610 #define msb_mask_low d0
3611 #define msb_mask_high d1
3613 #define blend_blocks_average_set_blend_mask_textured(source) \
3614 vclt.s16 blend_mask, source, #0 \
3616 #define blend_blocks_average_set_stp_bit_textured() \
3617 vorr.u16 blend_pixels, #0x8000 \
3619 #define blend_blocks_average_combine_textured(source) \
3620 vbif.u16 blend_pixels, source, blend_mask \
3622 #define blend_blocks_average_set_blend_mask_untextured(source) \
3624 #define blend_blocks_average_set_stp_bit_untextured() \
3626 #define blend_blocks_average_combine_untextured(source) \
3628 #define blend_blocks_average_mask_set_on() \
3629 vclt.s16 write_mask, fb_pixels_next, #0 \
3631 #define blend_blocks_average_mask_copy_on() \
3632 vorr.u16 draw_mask, draw_mask_next, write_mask \
3634 #define blend_blocks_average_mask_copy_b_on() \
3635 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3637 #define blend_blocks_average_mask_set_off() \
3639 #define blend_blocks_average_mask_copy_off() \
3640 vmov draw_mask, draw_mask_next \
3642 #define blend_blocks_average_mask_copy_b_off() \
3644 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3647 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3648 stmdb sp!, { r4, r14 }; \
3650 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3651 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3653 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3654 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3656 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3659 vmov.u16 d128_0x8000, #0x8000; \
3660 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3661 ldr fb_ptr_next, [pixel_ptr, #28]; \
3663 vmov.u16 d128_0x0421, #0x0400; \
3664 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3666 vorr.u16 d128_0x0421, #0x0021; \
3667 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3669 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3670 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3671 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3672 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3673 blend_blocks_average_mask_set_##mask_evaluate(); \
3674 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3676 subs num_blocks, num_blocks, #1; \
3680 mov fb_ptr, fb_ptr_next; \
3681 ldr fb_ptr_next, [pixel_ptr, #28]; \
3683 vmov pixels, pixels_next; \
3684 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3686 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3688 blend_blocks_average_mask_copy_##mask_evaluate(); \
3689 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3691 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3692 blend_blocks_average_set_stp_bit_##texturing(); \
3693 vmov fb_pixels, fb_pixels_next; \
3694 blend_blocks_average_combine_##texturing(pixels); \
3696 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3697 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3698 cmp fb_ptr_cmp, #28; \
3701 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3702 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3704 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3705 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3707 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3708 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3710 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3711 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3712 blend_blocks_average_mask_set_##mask_evaluate(); \
3713 vst1.u16 { fb_pixels }, [fb_ptr]; \
3716 subs num_blocks, num_blocks, #1; \
3720 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3721 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3723 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3724 blend_blocks_average_set_stp_bit_##texturing(); \
3725 blend_blocks_average_combine_##texturing(pixels_next); \
3727 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3728 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3729 vst1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3731 restore_abi_regs(); \
3732 ldmia sp!, { r4, pc }; \
3735 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3736 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3737 vst1.u16 { fb_pixels }, [fb_ptr]; \
3739 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3740 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3741 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3742 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3743 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3744 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3748 blend_blocks_average_builder(textured, off)
3749 blend_blocks_average_builder(untextured, off)
3750 blend_blocks_average_builder(textured, on)
3751 blend_blocks_average_builder(untextured, on)
3754 #define blend_blocks_add_mask_set_on() \
3755 vclt.s16 write_mask, fb_pixels, #0 \
3757 #define blend_blocks_add_mask_copy_on() \
3758 vorr.u16 draw_mask, draw_mask, write_mask \
3760 #define blend_blocks_add_mask_set_off() \
3762 #define blend_blocks_add_mask_copy_off() \
3765 #define blend_blocks_add_textured_builder(mask_evaluate) \
3768 function(blend_blocks_textured_add_##mask_evaluate) \
3769 stmdb sp!, { r4, r14 }; \
3771 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3772 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3774 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3775 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3777 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3780 vmov.u16 d128_0x7C1F, #0x7C00; \
3781 vmov.u16 d128_0x03E0, #0x0300; \
3782 vmov.u16 d128_0x83E0, #0x8000; \
3783 vorr.u16 d128_0x03E0, #0x00E0; \
3784 vorr.u16 d128_0x7C1F, #0x001F; \
3785 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3787 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3788 ldr fb_ptr_next, [pixel_ptr, #28]; \
3789 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3790 vclt.s16 blend_mask, pixels, #0; \
3791 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3792 blend_blocks_add_mask_set_##mask_evaluate(); \
3793 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3795 blend_blocks_add_mask_copy_##mask_evaluate(); \
3796 vorr.u16 pixels, pixels, msb_mask; \
3797 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3798 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3799 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3800 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3801 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3802 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3803 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3804 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3806 subs num_blocks, num_blocks, #1; \
3810 mov fb_ptr, fb_ptr_next; \
3812 ldr fb_ptr_next, [pixel_ptr, #28]; \
3814 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3815 vclt.s16 blend_mask, pixels, #0; \
3817 vorr.u16 pixels, pixels, msb_mask; \
3818 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3819 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3821 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3822 pld [fb_ptr_next, #64]; \
3824 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3825 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3827 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3828 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3830 cmp fb_ptr_cmp, #28; \
3833 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3834 blend_blocks_add_mask_set_##mask_evaluate(); \
3835 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3836 blend_blocks_add_mask_copy_##mask_evaluate(); \
3837 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3838 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3839 vst1.u16 { blend_pixels }, [fb_ptr]; \
3842 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3843 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3844 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3845 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3846 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3848 subs num_blocks, num_blocks, #1; \
3852 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3853 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3854 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3856 restore_abi_regs(); \
3857 ldmia sp!, { r4, pc }; \
3860 vst1.u16 { blend_pixels }, [fb_ptr]; \
3861 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3863 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3864 blend_blocks_add_mask_set_##mask_evaluate(); \
3865 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3866 blend_blocks_add_mask_copy_##mask_evaluate(); \
3867 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3871 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3874 function(blend_blocks_untextured_add_##mask_evaluate) \
3875 stmdb sp!, { r4, r14 }; \
3877 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3878 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3880 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3881 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3883 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3886 vmov.u16 d128_0x7C1F, #0x7C00; \
3887 vmov.u16 d128_0x03E0, #0x0300; \
3888 vorr.u16 d128_0x7C1F, #0x001F; \
3889 vorr.u16 d128_0x03E0, #0x00E0; \
3891 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3892 ldr fb_ptr_next, [pixel_ptr, #28]; \
3893 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3894 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3895 blend_blocks_add_mask_set_##mask_evaluate(); \
3896 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3898 blend_blocks_add_mask_copy_##mask_evaluate(); \
3899 vand.u16 pixels_g, pixels, d128_0x03E0; \
3900 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3901 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3902 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3903 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3904 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3905 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3907 subs num_blocks, num_blocks, #1; \
3911 mov fb_ptr, fb_ptr_next; \
3913 ldr fb_ptr_next, [pixel_ptr, #28]; \
3915 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3917 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3918 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3919 vand.u16 pixels_g, pixels, d128_0x03E0; \
3921 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3922 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3924 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3925 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3926 cmp fb_ptr_cmp, #28; \
3929 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3930 blend_blocks_add_mask_set_##mask_evaluate(); \
3931 blend_blocks_add_mask_copy_##mask_evaluate(); \
3932 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3933 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3934 vst1.u16 { blend_pixels }, [fb_ptr]; \
3937 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3938 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3939 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3940 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3941 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3943 subs num_blocks, num_blocks, #1; \
3947 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3948 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3949 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3950 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3952 restore_abi_regs(); \
3953 ldmia sp!, { r4, pc }; \
3956 vst1.u16 { blend_pixels }, [fb_ptr]; \
3957 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3959 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3960 blend_blocks_add_mask_set_##mask_evaluate(); \
3961 blend_blocks_add_mask_copy_##mask_evaluate(); \
3962 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3966 blend_blocks_add_textured_builder(off)
3967 blend_blocks_add_textured_builder(on)
3968 blend_blocks_add_untextured_builder(off)
3969 blend_blocks_add_untextured_builder(on)
3971 #define blend_blocks_subtract_set_blend_mask_textured() \
3972 vclt.s16 blend_mask, pixels_next, #0 \
3974 #define blend_blocks_subtract_combine_textured() \
3975 vbif.u16 blend_pixels, pixels, blend_mask \
3977 #define blend_blocks_subtract_set_stp_textured() \
3978 vorr.u16 blend_pixels, #0x8000 \
3980 #define blend_blocks_subtract_msb_mask_textured() \
3981 vorr.u16 pixels, pixels_next, msb_mask \
3983 #define blend_blocks_subtract_set_blend_mask_untextured() \
3985 #define blend_blocks_subtract_combine_untextured() \
3987 #define blend_blocks_subtract_set_stp_untextured() \
3988 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3990 #define blend_blocks_subtract_msb_mask_untextured() \
3993 #define blend_blocks_subtract_mask_set_on() \
3994 vclt.s16 write_mask, fb_pixels, #0 \
3996 #define blend_blocks_subtract_mask_copy_on() \
3997 vorr.u16 draw_mask, draw_mask_next, write_mask \
3999 #define blend_blocks_subtract_mask_set_off() \
4001 #define blend_blocks_subtract_mask_copy_off() \
4002 vmov draw_mask, draw_mask_next \
4005 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
4008 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
4009 stmdb sp!, { r4, r14 }; \
4011 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4012 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4014 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4015 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4017 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4020 vmov.u16 d128_0x7C1F, #0x7C00; \
4021 vmov.u16 d128_0x03E0, #0x0300; \
4022 vorr.u16 d128_0x7C1F, #0x001F; \
4023 vorr.u16 d128_0x03E0, #0x00E0; \
4025 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4026 ldr fb_ptr_next, [pixel_ptr, #28]; \
4027 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4028 blend_blocks_subtract_set_blend_mask_##texturing(); \
4029 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4030 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4031 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4033 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4034 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4035 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4036 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4037 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4039 subs num_blocks, num_blocks, #1; \
4043 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4044 mov fb_ptr, fb_ptr_next; \
4045 ldr fb_ptr_next, [pixel_ptr, #28]; \
4047 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4048 blend_blocks_subtract_msb_mask_##texturing(); \
4050 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4051 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4052 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4053 blend_blocks_subtract_set_stp_##texturing(); \
4054 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4055 blend_blocks_subtract_combine_##texturing(); \
4056 blend_blocks_subtract_set_blend_mask_##texturing(); \
4057 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4059 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4060 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4061 cmp fb_ptr_cmp, #28; \
4064 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4065 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4066 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4067 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4068 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4069 vst1.u16 { blend_pixels }, [fb_ptr]; \
4070 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4073 subs num_blocks, num_blocks, #1; \
4077 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4079 blend_blocks_subtract_msb_mask_##texturing(); \
4080 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4081 blend_blocks_subtract_set_stp_##texturing(); \
4082 blend_blocks_subtract_combine_##texturing(); \
4083 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4084 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4086 restore_abi_regs(); \
4087 ldmia sp!, { r4, pc }; \
4090 vst1.u16 { blend_pixels }, [fb_ptr]; \
4091 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4092 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4093 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4094 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4095 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4096 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4100 blend_blocks_subtract_builder(textured, off)
4101 blend_blocks_subtract_builder(textured, on)
4102 blend_blocks_subtract_builder(untextured, off)
4103 blend_blocks_subtract_builder(untextured, on)
4106 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4109 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4110 stmdb sp!, { r4, r14 }; \
4112 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4113 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4115 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4116 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4118 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4121 vmov.u16 d128_0x7C1F, #0x7C00; \
4122 vmov.u16 d128_0x03E0, #0x0300; \
4123 vmov.u16 d128_0x1C07, #0x1C00; \
4124 vmov.u16 d128_0x00E0, #0x00E0; \
4125 vorr.u16 d128_0x7C1F, #0x001F; \
4126 vorr.u16 d128_0x03E0, #0x00E0; \
4127 vorr.u16 d128_0x1C07, #0x0007; \
4129 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4130 ldr fb_ptr_next, [pixel_ptr, #28]; \
4131 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4132 vclt.s16 blend_mask, pixels, #0; \
4133 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4134 blend_blocks_add_mask_set_##mask_evaluate(); \
4135 vshr.s16 pixels_fourth, pixels, #2; \
4136 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4138 blend_blocks_add_mask_copy_##mask_evaluate(); \
4139 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4140 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4141 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4142 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4143 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4144 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4145 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4147 subs num_blocks, num_blocks, #1; \
4151 mov fb_ptr, fb_ptr_next; \
4152 ldr fb_ptr_next, [pixel_ptr, #28]; \
4154 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4155 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4156 vbif.u16 blend_pixels, pixels, blend_mask; \
4158 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4159 vclt.s16 blend_mask, pixels, #0; \
4160 vshr.s16 pixels_fourth, pixels, #2; \
4161 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4162 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4164 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4165 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4167 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4168 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4169 cmp fb_ptr_cmp, #28; \
4172 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4173 blend_blocks_add_mask_set_##mask_evaluate(); \
4174 blend_blocks_add_mask_copy_##mask_evaluate(); \
4175 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4176 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4177 vst1.u16 { blend_pixels }, [fb_ptr]; \
4180 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4181 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4182 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4183 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4184 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4186 subs num_blocks, num_blocks, #1; \
4190 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4191 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4192 vbif.u16 blend_pixels, pixels, blend_mask; \
4193 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4194 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4195 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4197 restore_abi_regs(); \
4198 ldmia sp!, { r4, pc }; \
4201 vst1.u16 { blend_pixels }, [fb_ptr]; \
4202 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4204 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4205 blend_blocks_add_mask_set_##mask_evaluate(); \
4206 blend_blocks_add_mask_copy_##mask_evaluate(); \
4207 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4212 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4215 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4216 stmdb sp!, { r4, r14 }; \
4218 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4219 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4221 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4222 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4224 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4227 vmov.u16 d128_0x7C1F, #0x7C00; \
4228 vmov.u16 d128_0x03E0, #0x0300; \
4229 vmov.u16 d128_0x1C07, #0x1C00; \
4230 vmov.u16 d128_0x00E0, #0x00E0; \
4231 vorr.u16 d128_0x7C1F, #0x001F; \
4232 vorr.u16 d128_0x03E0, #0x00E0; \
4233 vorr.u16 d128_0x1C07, #0x0007; \
4235 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4236 ldr fb_ptr_next, [pixel_ptr, #28]; \
4237 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4238 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4239 blend_blocks_add_mask_set_##mask_evaluate(); \
4240 vshr.s16 pixels_fourth, pixels, #2; \
4241 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4243 blend_blocks_add_mask_copy_##mask_evaluate(); \
4244 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4245 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4246 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4247 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4248 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4249 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4250 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4252 subs num_blocks, num_blocks, #1; \
4256 mov fb_ptr, fb_ptr_next; \
4257 ldr fb_ptr_next, [pixel_ptr, #28]; \
4259 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4261 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4262 vshr.s16 pixels_fourth, pixels, #2; \
4263 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4264 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4266 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4267 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4269 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4270 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4271 cmp fb_ptr_cmp, #28; \
4274 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4275 blend_blocks_add_mask_set_##mask_evaluate(); \
4276 blend_blocks_add_mask_copy_##mask_evaluate(); \
4277 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4278 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4279 vst1.u16 { blend_pixels }, [fb_ptr]; \
4282 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4283 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4284 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4285 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4286 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4288 subs num_blocks, num_blocks, #1; \
4292 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4293 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4294 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4295 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4297 restore_abi_regs(); \
4298 ldmia sp!, { r4, pc }; \
4301 vst1.u16 { blend_pixels }, [fb_ptr]; \
4302 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4304 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4305 blend_blocks_add_mask_set_##mask_evaluate(); \
4306 blend_blocks_add_mask_copy_##mask_evaluate(); \
4307 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4311 blend_blocks_add_fourth_textured_builder(off)
4312 blend_blocks_add_fourth_textured_builder(on)
4313 blend_blocks_add_fourth_untextured_builder(off)
4314 blend_blocks_add_fourth_untextured_builder(on)
4316 // TODO: Optimize this more. Need a scene that actually uses it for
4321 function(blend_blocks_textured_unblended_on)
4322 stmdb sp!, { r4, r14 }
4324 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4325 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
4327 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4328 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
4330 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4333 ldr fb_ptr, [pixel_ptr, #28]
4334 vld1.u16 { fb_pixels }, [fb_ptr]
4335 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4336 vclt.s16 write_mask, fb_pixels, #0
4337 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4339 subs num_blocks, num_blocks, #1
4343 vorr.u16 pixels, pixels, msb_mask
4344 vorr.u16 draw_mask, draw_mask, write_mask
4345 vbif.u16 fb_pixels, pixels, draw_mask
4346 vst1.u16 { fb_pixels }, [fb_ptr]
4348 ldr fb_ptr, [pixel_ptr, #28]
4349 vld1.u16 { fb_pixels }, [fb_ptr]
4350 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4351 vclt.s16 write_mask, fb_pixels, #0
4352 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4354 subs num_blocks, num_blocks, #1
4358 vorr.u16 pixels, pixels, msb_mask
4359 vorr.u16 draw_mask, draw_mask, write_mask
4360 vbif.u16 fb_pixels, pixels, draw_mask
4361 vst1.u16 { fb_pixels }, [fb_ptr]
4364 ldmia sp!, { r4, pc }
4367 function(blend_blocks_textured_unblended_off)
4377 vld1.u32 { u_whole_8, v_whole_8 }, [r1, :128], r3
4397 #define num_width r12
4407 function(render_block_fill_body)
4408 vdup.u16 colors_a, color
4411 vmov colors_b, colors_a
4412 sub pitch, pitch, width, lsl #1
4414 mov num_width, width
4417 vst1.u32 { colors_a, colors_b }, [vram_ptr, :256]!
4419 subs num_width, num_width, #16
4422 add vram_ptr, vram_ptr, pitch
4423 mov num_width, width
4425 subs height, height, #1
4439 #undef dirty_textures_mask
4441 #undef current_texture_mask
4452 #define offset_u_right r10
4453 #define width_rounded r11
4454 #define height_rounded r12
4456 #define texture_offset_base r1
4457 #define tile_width r2
4458 #define tile_height r3
4459 #define num_blocks r4
4461 #define sub_tile_height r6
4463 #define texture_mask r8
4464 #define column_data r9
4465 #define texture_offset r10
4466 #define tiles_remaining r11
4467 #define fb_ptr_advance_column r12
4468 #define texture_block_ptr r14
4472 #define texture_page_ptr r3
4473 #define left_block_mask r4
4474 #define right_block_mask r5
4475 #define texture_mask_rev r10
4476 #define control_mask r11
4478 #define dirty_textures_mask r4
4480 #define current_texture_mask r6
4494 #define draw_masks_fb_ptrs q1
4496 #define draw_mask_fb_ptr_left d2
4497 #define draw_mask_fb_ptr_right d3
4499 #define draw_mask_fb_ptr_left_a d2
4500 #define draw_mask_fb_ptr_left_b d3
4501 #define draw_mask_fb_ptr_right_a d10
4502 #define draw_mask_fb_ptr_right_b d11
4503 #define draw_masks_fb_ptrs2 q5
4505 #define clut_low_a d4
4506 #define clut_low_b d5
4507 #define clut_high_a d6
4508 #define clut_high_b d7
4510 #define block_masks d8
4511 #define block_masks_shifted d9
4516 #define texels_low d12
4517 #define texels_high d13
4519 #define texels_wide_low d14
4520 #define texels_wide_high d15
4521 #define texels_wide q7
4524 setup_sprite_flush_blocks:
4527 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4528 bl flush_render_block_buffer
4529 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4533 add block, psx_gpu, #psx_gpu_blocks_offset
4537 setup_sprite_update_texture_4bpp_cache:
4538 stmdb sp!, { r0 - r3, r14 }
4539 bl update_texture_4bpp_cache
4540 ldmia sp!, { r0 - r3, pc }
4543 setup_sprite_update_texture_8bpp_cache:
4544 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r14 }
4545 bl update_texture_8bpp_cache
4546 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS pc }
4549 #define setup_sprite_tiled_initialize_4bpp() \
4550 ldr dirty_textures_mask, \
4551 [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]; \
4552 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4554 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4555 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4557 tst current_texture_mask, dirty_textures_mask; \
4558 vuzp.u8 clut_a, clut_b; \
4560 blne setup_sprite_update_texture_4bpp_cache \
4562 #define setup_sprite_tiled_initialize_8bpp() \
4563 ldr dirty_textures_mask, \
4564 [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]; \
4565 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4567 tst current_texture_mask, dirty_textures_mask; \
4568 blne setup_sprite_update_texture_8bpp_cache \
4571 #define setup_sprite_block_count_single() \
4574 #define setup_sprite_block_count_double() \
4575 sub_tile_height, lsl #1 \
4577 #define setup_sprite_tile_add_blocks(type) \
4578 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4579 cmp num_blocks, #MAX_BLOCKS; \
4581 movgt num_blocks, setup_sprite_block_count_##type(); \
4582 blgt setup_sprite_flush_blocks \
4585 #define setup_sprite_tile_full_4bpp(edge) \
4586 setup_sprite_tile_add_blocks(double); \
4589 and texture_block_ptr, texture_offset, texture_mask; \
4590 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4593 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4594 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4596 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4597 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4599 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4600 add texture_block_ptr, texture_offset, #8; \
4602 and texture_block_ptr, texture_block_ptr, texture_mask; \
4603 add block, block, #40; \
4605 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4606 add fb_ptr, fb_ptr, #16; \
4608 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4609 add block, block, #24; \
4611 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4612 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4615 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4616 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4618 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4619 add block, block, #40; \
4621 add texture_offset, texture_offset, #0x10; \
4622 add fb_ptr, fb_ptr, #(2048 - 16); \
4624 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4625 add block, block, #24; \
4627 subs sub_tile_height, sub_tile_height, #1; \
4630 add texture_offset, texture_offset, #0xF00; \
4631 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4634 #define setup_sprite_tile_half_4bpp(edge) \
4635 setup_sprite_tile_add_blocks(single); \
4638 and texture_block_ptr, texture_offset, texture_mask; \
4639 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4642 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4643 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4645 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4646 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4648 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4649 add block, block, #40; \
4651 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4652 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4654 add block, block, #24; \
4655 add texture_offset, texture_offset, #0x10; \
4657 add fb_ptr, fb_ptr, #2048; \
4658 subs sub_tile_height, sub_tile_height, #1; \
4662 add texture_offset, texture_offset, #0xF00; \
4663 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4666 #define setup_sprite_tile_full_8bpp(edge) \
4667 setup_sprite_tile_add_blocks(double); \
4668 add block, block, #16; \
4671 and texture_block_ptr, texture_offset, texture_mask; \
4672 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4675 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4676 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4678 add texture_block_ptr, texture_offset, #8; \
4679 vst1.u32 { texels }, [block, :64]; \
4681 and texture_block_ptr, texture_block_ptr, texture_mask; \
4682 add block, block, #24; \
4684 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4686 add fb_ptr, fb_ptr, #16; \
4687 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4689 add block, block, #40; \
4690 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4693 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4694 vst1.u32 { texels }, [block, :64]; \
4695 add block, block, #24; \
4697 add texture_offset, texture_offset, #0x10; \
4698 add fb_ptr, fb_ptr, #(2048 - 16); \
4700 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4701 add block, block, #40; \
4703 subs sub_tile_height, sub_tile_height, #1; \
4706 sub block, block, #16; \
4707 add texture_offset, texture_offset, #0xF00; \
4708 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4711 #define setup_sprite_tile_half_8bpp(edge) \
4712 setup_sprite_tile_add_blocks(single); \
4713 add block, block, #16; \
4716 and texture_block_ptr, texture_offset, texture_mask; \
4717 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4720 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4721 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4723 vst1.u32 { texels }, [block, :64]; \
4724 add block, block, #24; \
4726 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4727 add block, block, #40; \
4729 add texture_offset, texture_offset, #0x10; \
4730 add fb_ptr, fb_ptr, #2048; \
4732 subs sub_tile_height, sub_tile_height, #1; \
4735 sub block, block, #16; \
4736 add texture_offset, texture_offset, #0xF00; \
4737 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4740 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4741 add texture_offset, texture_offset_base, #8; \
4742 add fb_ptr, fb_ptr, #16 \
4744 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4745 mov texture_offset, texture_offset_base \
4747 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4748 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4750 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4751 mov texture_offset, texture_offset_base \
4753 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4754 sub fb_ptr, fb_ptr, #16 \
4756 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4758 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4759 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4761 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4764 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4766 mov sub_tile_height, column_data; \
4767 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4768 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4769 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4771 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4773 and sub_tile_height, column_data, #0xFF; \
4774 mov tiles_remaining, column_data, lsr #16; \
4775 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4776 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4778 subs tiles_remaining, tiles_remaining, #1; \
4782 mov sub_tile_height, #16; \
4783 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4784 subs tiles_remaining, tiles_remaining, #1; \
4788 uxtb sub_tile_height, column_data, ror #8; \
4789 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4790 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4793 #define setup_sprite_column_data_single() \
4794 mov column_data, height; \
4795 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] \
4797 #define setup_sprite_column_data_multi() \
4798 and height_rounded, height_rounded, #0xF; \
4799 rsb column_data, offset_v, #16; \
4801 add height_rounded, height_rounded, #1; \
4802 sub tile_height, tile_height, #1; \
4804 orr column_data, column_data, tile_height, lsl #16; \
4805 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]; \
4807 orr column_data, column_data, height_rounded, lsl #8 \
4809 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4810 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4811 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4813 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4814 mov fb_ptr_advance_column, #32; \
4815 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4817 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4818 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4820 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4821 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4822 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4824 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4826 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4827 setup_sprite_column_data_##multi_height(); \
4828 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4829 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4830 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4832 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4833 restore_abi_regs(); \
4834 ldmia sp!, { r4 - r11, pc } \
4836 #define setup_sprite_tiled_advance_column() \
4837 add texture_offset_base, texture_offset_base, #0x100; \
4838 tst texture_offset_base, #0xF00; \
4839 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4841 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4842 right_mode, x4mode) \
4843 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4844 setup_sprite_column_data_##multi_height(); \
4846 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4848 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4850 subs tile_width, tile_width, #2; \
4851 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4855 vmov.u8 draw_masks_fb_ptrs, #0; \
4856 vmov.u8 draw_masks_fb_ptrs2, #0; \
4859 setup_sprite_tiled_advance_column(); \
4860 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4861 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4862 subs tile_width, tile_width, #1; \
4866 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4868 setup_sprite_tiled_advance_column(); \
4869 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4870 restore_abi_regs(); \
4871 ldmia sp!, { r4 - r11, pc } \
4874 #define setup_sprite_offset_u_adjust() \
4876 #define setup_sprite_get_left_block_mask() \
4877 and left_block_mask, left_block_mask, #0xFF \
4879 #define setup_sprite_compare_left_block_mask() \
4880 cmp left_block_mask, #0xFF \
4882 #define setup_sprite_get_right_block_mask() \
4883 uxtb right_block_mask, right_block_mask, ror #8 \
4885 #define setup_sprite_compare_right_block_mask() \
4886 cmp right_block_mask, #0xFF \
4891 #define fb_ptr2 column_data
4893 #define setup_sprite_offset_u_adjust_4x() \
4894 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4895 lsl offset_u_right, #1; \
4897 add offset_u_right, #1 \
4899 #define setup_sprite_get_left_block_mask_4x() \
4900 sxth left_block_mask, left_block_mask \
4902 #define setup_sprite_compare_left_block_mask_4x() \
4903 cmp left_block_mask, #0xFFFFFFFF \
4905 #define setup_sprite_get_right_block_mask_4x() \
4906 sxth right_block_mask, right_block_mask, ror #16 \
4908 #define setup_sprite_compare_right_block_mask_4x() \
4909 cmp right_block_mask, #0xFFFFFFFF \
4912 #define widen_texels_16bpp(texels_) \
4913 vmov texels_wide_low, texels_; \
4914 vmov texels_wide_high, texels_; \
4915 vzip.16 texels_wide_low, texels_wide_high \
4917 #define widen_texels_8bpp(texels_) \
4918 vmov texels_wide_low, texels_; \
4919 vmov texels_wide_high, texels_; \
4920 vzip.8 texels_wide_low, texels_wide_high \
4922 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4923 vst1.u32 { texels_ }, [block_, :128]; \
4924 add block_, block_, #40; \
4926 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4927 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4928 add block_, block_, #24 \
4930 /* assumes 16-byte offset already added to block_ */
4931 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4932 vst1.u32 { texels_ }, [block_, :64]; \
4933 add block_, block_, #24; \
4935 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4936 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4937 add block_, block_, #40 \
4939 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4940 draw_mask_fb_ptr_b_) \
4941 widen_texels_16bpp(texels_low); \
4942 add fb_ptr_tmp, fb_ptr, #1024*2; \
4944 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4946 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4947 widen_texels_16bpp(texels_high); \
4949 add fb_ptr_tmp, fb_ptr, #8*2; \
4950 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4952 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4953 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4955 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4956 draw_mask_fb_ptr_b_) \
4957 widen_texels_8bpp(texels); \
4958 add fb_ptr_tmp, fb_ptr, #1024*2; \
4960 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4961 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4963 add fb_ptr_tmp, fb_ptr, #8*2; \
4964 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4966 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4967 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4970 #define setup_sprite_tiled_initialize_4bpp_4x() \
4971 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4972 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4974 vuzp.u8 clut_a, clut_b \
4976 #define setup_sprite_tiled_initialize_8bpp_4x() \
4979 #define setup_sprite_block_count_single_4x() \
4980 sub_tile_height, lsl #2 \
4982 #define setup_sprite_block_count_double_4x() \
4983 sub_tile_height, lsl #(1+2) \
4985 #define setup_sprite_tile_full_4bpp_4x(edge) \
4986 setup_sprite_tile_add_blocks(double_4x); \
4987 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4990 and texture_block_ptr, texture_offset, texture_mask; \
4993 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4994 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4996 add texture_block_ptr, texture_offset, #8; \
4997 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4999 and texture_block_ptr, texture_block_ptr, texture_mask; \
5000 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5002 vzip.8 texels_low, texels_high; \
5003 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5004 draw_mask_fb_ptr_left_b); \
5006 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5007 pld [fb_ptr, #2048]; \
5009 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5010 add fb_ptr, fb_ptr, #16*2; \
5012 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5013 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5015 vzip.8 texels_low, texels_high; \
5016 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5017 draw_mask_fb_ptr_right_b); \
5019 add texture_offset, texture_offset, #0x10; \
5020 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5022 subs sub_tile_height, sub_tile_height, #1; \
5025 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5026 add texture_offset, texture_offset, #0xF00; \
5027 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5030 #define setup_sprite_tile_half_4bpp_4x(edge) \
5031 setup_sprite_tile_add_blocks(single_4x); \
5032 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5035 and texture_block_ptr, texture_offset, texture_mask; \
5038 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5039 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5041 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5042 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5044 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5045 add texture_offset, texture_offset, #0x10; \
5047 vzip.8 texels_low, texels_high; \
5048 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5049 draw_mask_fb_ptr_##edge##_b); \
5051 pld [fb_ptr, #2048]; \
5052 add fb_ptr, fb_ptr, #2048 * 2; \
5054 subs sub_tile_height, sub_tile_height, #1; \
5057 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5058 add texture_offset, texture_offset, #0xF00; \
5059 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5062 #define setup_sprite_tile_full_8bpp_4x(edge) \
5063 setup_sprite_tile_add_blocks(double_4x); \
5064 add block, block, #16; \
5065 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5068 and texture_block_ptr, texture_offset, texture_mask; \
5071 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5072 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5074 add texture_block_ptr, texture_offset, #8; \
5075 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5076 draw_mask_fb_ptr_left_b); \
5078 pld [fb_ptr, #2048]; \
5079 and texture_block_ptr, texture_block_ptr, texture_mask; \
5081 add fb_ptr, fb_ptr, #16*2; \
5082 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5084 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5086 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5087 draw_mask_fb_ptr_right_b); \
5089 add texture_offset, texture_offset, #0x10; \
5090 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5092 subs sub_tile_height, sub_tile_height, #1; \
5095 sub block, block, #16; \
5096 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5097 add texture_offset, texture_offset, #0xF00; \
5098 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5101 #define setup_sprite_tile_half_8bpp_4x(edge) \
5102 setup_sprite_tile_add_blocks(single_4x); \
5103 add block, block, #16; \
5104 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5107 and texture_block_ptr, texture_offset, texture_mask; \
5110 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5111 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5113 pld [fb_ptr, #2048]; \
5114 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5115 draw_mask_fb_ptr_##edge##_b); \
5117 add texture_offset, texture_offset, #0x10; \
5118 add fb_ptr, fb_ptr, #2048 * 2; \
5120 subs sub_tile_height, sub_tile_height, #1; \
5123 sub block, block, #16; \
5124 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5125 add texture_offset, texture_offset, #0xF00; \
5126 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5129 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5130 add texture_offset, texture_offset_base, #8; \
5131 add fb_ptr, fb_ptr, #16 * 2 \
5133 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5134 mov texture_offset, texture_offset_base \
5136 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5137 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5139 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5140 mov texture_offset, texture_offset_base \
5142 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5143 sub fb_ptr, fb_ptr, #16 * 2 \
5145 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5147 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5148 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5150 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5153 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5154 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5155 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5156 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5157 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5159 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5160 mov fb_ptr_advance_column, #32 * 2; \
5161 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5162 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5163 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5164 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5165 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5167 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5168 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5169 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5170 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5171 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5181 // [sp + 12]: color (unused)
5183 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5185 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5187 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5189 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5191 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5193 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5195 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5197 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5199 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5201 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5203 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5205 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5207 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5209 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5211 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5216 function(setup_sprite_##texture_mode##x4mode) \
5217 stmdb sp!, { r4 - r11, r14 }; \
5218 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5221 and offset_u, u, #0xF; \
5223 ldr width, [sp, #40]; \
5224 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
5226 ldr height, [sp, #44]; \
5227 add fb_ptr, fb_ptr, y, lsl #11; \
5231 add fb_ptr, fb_ptr, x, lsl #1; \
5232 and offset_v, v, #0xF; \
5234 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5235 add width_rounded, offset_u, width; \
5237 add height_rounded, offset_v, height; \
5238 add width_rounded, width_rounded, #15; \
5240 add height_rounded, height_rounded, #15; \
5241 mov tile_width, width_rounded, lsr #4; \
5243 /* texture_offset_base = VH-VL-00-00 */\
5244 mov texture_offset_base, v, lsl #8; \
5245 and offset_u_right, width_rounded, #0xF; \
5247 /* texture_offset_base = VH-UH-UL-00 */\
5248 bfi texture_offset_base, u, #4, #8; \
5249 mov right_block_mask, #0xFFFFFFFE; \
5251 setup_sprite_offset_u_adjust##x4mode(); \
5253 /* texture_offset_base = VH-UH-VL-00 */\
5254 bfi texture_offset_base, v, #4, #4; \
5255 mov left_block_mask, #0xFFFFFFFF; \
5257 mov tile_height, height_rounded, lsr #4; \
5258 mvn left_block_mask, left_block_mask, lsl offset_u; \
5260 /* texture_mask = HH-HL-WH-WL */\
5261 ldrh texture_mask, [psx_gpu, #psx_gpu_texture_mask_width_offset]; \
5262 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5264 /* texture_mask_rev = WH-WL-HH-HL */\
5265 rev16 texture_mask_rev, texture_mask; \
5266 vmov block_masks, left_block_mask, right_block_mask; \
5268 /* texture_mask = HH-HL-HL-WL */\
5269 bfi texture_mask, texture_mask_rev, #4, #4; \
5270 /* texture_mask_rev = 00-00-00-WH */\
5271 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5273 /* texture_mask = HH-WH-HL-WL */\
5274 bfi texture_mask, texture_mask_rev, #8, #4; \
5275 setup_sprite_get_left_block_mask##x4mode(); \
5277 mov control_mask, #0; \
5278 setup_sprite_compare_left_block_mask##x4mode(); \
5280 setup_sprite_get_right_block_mask##x4mode(); \
5281 orreq control_mask, control_mask, #0x4; \
5283 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
5284 setup_sprite_compare_right_block_mask##x4mode(); \
5286 orreq control_mask, control_mask, #0x8; \
5287 cmp tile_width, #1; \
5289 add block, psx_gpu, #psx_gpu_blocks_offset; \
5290 orreq control_mask, control_mask, #0x1; \
5292 cmp tile_height, #1; \
5293 add block, block, num_blocks, lsl #6; \
5295 orreq control_mask, control_mask, #0x2; \
5296 JT_OP_REL(9f, control_mask, temp); \
5297 JT_OP(ldr pc, [pc, control_mask, lsl #2]); \
5301 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \
5302 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \
5303 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \
5304 .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \
5305 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \
5306 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \
5307 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \
5308 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\
5309 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \
5310 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \
5311 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \
5312 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \
5313 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \
5315 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \
5318 setup_sprite_tiled_builder(4bpp,);
5319 setup_sprite_tiled_builder(8bpp,);
5321 #undef draw_mask_fb_ptr_left
5322 #undef draw_mask_fb_ptr_right
5324 setup_sprite_tiled_builder(4bpp, _4x);
5325 setup_sprite_tiled_builder(8bpp, _4x);
5333 #define block_ptr r0
5334 #define num_blocks r1
5336 #define texel_shift_mask r3
5337 #define block_pixels_a r4
5338 #define block_pixels_b r5
5347 #define texels_01 r6
5348 #define texels_23 r7
5349 #define texels_45 r8
5350 #define texels_67 r9
5352 function(texture_sprite_blocks_8bpp)
5353 stmdb sp!, { r4 - r11, r14 }
5354 movw texel_shift_mask, #(0xFF << 1)
5356 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5357 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
5359 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5360 ldr block_pixels_a, [block_ptr, #16]
5363 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5364 ldr block_pixels_b, [block_ptr, #20]
5366 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5367 ldrh texel_0, [clut_ptr, texel_0]
5369 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5370 ldrh texel_1, [clut_ptr, texel_1]
5372 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5373 ldr block_pixels_a, [block_ptr, #(64 + 16)]
5375 ldrh texel_2, [clut_ptr, texel_2]
5376 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5378 ldrh texel_3, [clut_ptr, texel_3]
5379 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5381 ldrh texel_4, [clut_ptr, texel_4]
5382 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5384 ldrh texel_5, [clut_ptr, texel_5]
5385 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5387 ldrh texel_6, [clut_ptr, texel_6]
5388 orr texels_01, texel_0, texel_1, lsl #16
5390 ldrh texel_7, [clut_ptr, texel_7]
5391 orr texels_23, texel_2, texel_3, lsl #16
5393 orr texels_45, texel_4, texel_5, lsl #16
5394 str texels_01, [block_ptr, #0]
5396 orr texels_67, texel_6, texel_7, lsl #16
5397 str texels_23, [block_ptr, #4]
5399 subs num_blocks, num_blocks, #1
5400 str texels_45, [block_ptr, #8]
5402 str texels_67, [block_ptr, #12]
5403 add block_ptr, block_ptr, #64
5407 ldmia sp!, { r4 - r11, pc }
5410 #undef width_rounded
5413 #undef texture_offset
5416 #undef texels_wide_low
5417 #undef texels_wide_high
5429 #define left_offset r8
5430 #define width_rounded r9
5431 #define right_width r10
5433 #define block_width r11
5435 #define texture_offset_base r1
5436 #define texture_mask r2
5437 #define texture_page_ptr r3
5438 #define num_blocks r4
5441 #define texture_offset r8
5442 #define blocks_remaining r9
5444 #define fb_ptr_pitch r12
5445 #define texture_block_ptr r14
5447 #define texture_mask_width r2
5448 #define texture_mask_height r3
5449 #define left_mask_bits r4
5450 #define right_mask_bits r5
5454 #undef block_masks_shifted
5457 #define block_masks d0
5458 #define block_masks_shifted d1
5459 #define draw_mask_fb_ptr d2
5462 #define draw_mask_fb_ptr_a d2
5463 #define draw_mask_fb_ptr_b d3
5464 #define texels_low d4
5465 #define texels_high d5
5466 #define texels_wide_low d6
5467 #define texels_wide_high d7
5468 #define texels_wide q3
5471 setup_sprites_16bpp_flush:
5474 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5475 bl flush_render_block_buffer
5476 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5480 add block, psx_gpu, #psx_gpu_blocks_offset
5481 mov num_blocks, block_width
5485 function(setup_sprite_16bpp)
5486 stmdb sp!, { r4 - r11, r14 }
5487 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5490 add fb_ptr, fb_ptr, y, lsl #11
5492 ldr width, [sp, #40]
5493 add fb_ptr, fb_ptr, x, lsl #1
5495 ldr height, [sp, #44]
5496 and left_offset, u, #0x7
5498 add texture_offset_base, u, u
5499 add width_rounded, width, #7
5501 add texture_offset_base, texture_offset_base, v, lsl #11
5502 mov left_mask_bits, #0xFF
5504 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5505 add width_rounded, width_rounded, left_offset
5507 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5508 sub fb_ptr, fb_ptr, left_offset, lsl #1
5510 add texture_mask, texture_mask_width, texture_mask_width
5511 mov right_mask_bits, #0xFE
5513 and right_width, width_rounded, #0x7
5514 mvn left_mask_bits, left_mask_bits, lsl left_offset
5516 add texture_mask, texture_mask, texture_mask_height, lsl #11
5517 mov block_width, width_rounded, lsr #3
5519 mov right_mask_bits, right_mask_bits, lsl right_width
5520 movw fb_ptr_pitch, #(2048 + 16)
5522 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5523 vmov block_masks, left_mask_bits, right_mask_bits
5525 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5526 add block, psx_gpu, #psx_gpu_blocks_offset
5528 bic texture_offset_base, texture_offset_base, #0xF
5531 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5532 add block, block, num_blocks, lsl #6
5536 vext.32 block_masks_shifted, block_masks, block_masks, #1
5537 vorr.u32 block_masks, block_masks, block_masks_shifted
5538 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5541 add num_blocks, num_blocks, #1
5542 cmp num_blocks, #MAX_BLOCKS
5543 blgt setup_sprites_16bpp_flush
5545 and texture_block_ptr, texture_offset_base, texture_mask
5546 subs height, height, #1
5548 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5549 vld1.u32 { texels }, [texture_block_ptr, :128]
5551 vst1.u32 { texels }, [block, :128]
5552 add block, block, #40
5554 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5557 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5559 add block, block, #24
5560 add texture_offset_base, texture_offset_base, #2048
5561 add fb_ptr, fb_ptr, #2048
5562 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5565 ldmia sp!, { r4 - r11, pc }
5568 add num_blocks, num_blocks, block_width
5569 mov texture_offset, texture_offset_base
5571 cmp num_blocks, #MAX_BLOCKS
5572 blgt setup_sprites_16bpp_flush
5574 add texture_offset_base, texture_offset_base, #2048
5575 and texture_block_ptr, texture_offset, texture_mask
5577 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5578 vld1.u32 { texels }, [texture_block_ptr, :128]
5580 vst1.u32 { texels }, [block, :128]
5581 add block, block, #40
5583 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5584 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5587 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5588 subs blocks_remaining, block_width, #2
5590 add texture_offset, texture_offset, #16
5591 add fb_ptr, fb_ptr, #16
5593 vmov.u8 draw_mask_fb_ptr, #0
5595 add block, block, #24
5599 and texture_block_ptr, texture_offset, texture_mask
5600 subs blocks_remaining, blocks_remaining, #1
5602 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5603 vld1.u32 { texels }, [texture_block_ptr, :128]
5605 vst1.u32 { texels }, [block, :128]
5606 add block, block, #40
5608 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5611 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5613 add texture_offset, texture_offset, #16
5614 add fb_ptr, fb_ptr, #16
5616 add block, block, #24
5620 and texture_block_ptr, texture_offset, texture_mask
5621 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5623 vld1.u32 { texels }, [texture_block_ptr, :128]
5624 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5626 vst1.u32 { texels }, [block, :128]
5627 add block, block, #40
5629 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5630 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5632 add block, block, #24
5633 subs height, height, #1
5635 add fb_ptr, fb_ptr, fb_ptr_pitch
5636 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5640 ldmia sp!, { r4 - r11, pc }
5644 // FIXME: duplicate code with normal version :(
5645 #undef draw_mask_fb_ptr
5647 function(setup_sprite_16bpp_4x)
5648 stmdb sp!, { r4 - r11, r14 }
5649 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5652 add fb_ptr, fb_ptr, y, lsl #11
5654 ldr width, [sp, #40]
5655 add fb_ptr, fb_ptr, x, lsl #1
5657 ldr height, [sp, #44]
5658 and left_offset, u, #0x7
5660 add texture_offset_base, u, u
5661 add width_rounded, width, #7
5663 add texture_offset_base, texture_offset_base, v, lsl #11
5664 movw left_mask_bits, #0xFFFF
5666 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5667 add width_rounded, width_rounded, left_offset
5671 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5672 sub fb_ptr, fb_ptr, left_offset, lsl #1
5674 add texture_mask, texture_mask_width, texture_mask_width
5675 movw right_mask_bits, #0xFFFC
5677 and right_width, width_rounded, #0x7
5678 mvn left_mask_bits, left_mask_bits, lsl left_offset
5682 add texture_mask, texture_mask, texture_mask_height, lsl #11
5683 mov block_width, width_rounded, lsr #3
5685 mov right_mask_bits, right_mask_bits, lsl right_width
5686 movw fb_ptr_pitch, #(2048 + 16) * 2
5688 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5689 vmov block_masks, left_mask_bits, right_mask_bits
5691 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5692 add block, psx_gpu, #psx_gpu_blocks_offset
5694 bic texture_offset_base, texture_offset_base, #0xF
5697 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5698 add block, block, num_blocks, lsl #6
5703 vext.32 block_masks_shifted, block_masks, block_masks, #1
5704 vorr.u32 block_masks, block_masks, block_masks_shifted
5705 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5706 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5709 add num_blocks, num_blocks, block_width
5710 cmp num_blocks, #MAX_BLOCKS
5711 blgt setup_sprites_16bpp_flush
5713 and texture_block_ptr, texture_offset_base, texture_mask
5714 subs height, height, #1
5716 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5717 vld1.u32 { texels }, [texture_block_ptr, :128]
5719 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5721 add texture_offset_base, texture_offset_base, #2048
5722 add fb_ptr, fb_ptr, #2048*2
5723 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5726 ldmia sp!, { r4 - r11, pc }
5729 add num_blocks, num_blocks, block_width
5730 mov texture_offset, texture_offset_base
5732 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5733 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5735 cmp num_blocks, #MAX_BLOCKS
5736 blgt setup_sprites_16bpp_flush
5738 add texture_offset_base, texture_offset_base, #2048
5739 and texture_block_ptr, texture_offset, texture_mask
5741 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5742 vld1.u32 { texels }, [texture_block_ptr, :128]
5744 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5746 subs blocks_remaining, block_width, #2*4
5747 add texture_offset, texture_offset, #16
5749 vmov.u8 draw_mask_fb_ptr_a, #0
5750 vmov.u8 draw_mask_fb_ptr_b, #0
5752 add fb_ptr, fb_ptr, #16*2
5756 and texture_block_ptr, texture_offset, texture_mask
5757 subs blocks_remaining, blocks_remaining, #4
5759 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5760 vld1.u32 { texels }, [texture_block_ptr, :128]
5762 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5763 add texture_offset, texture_offset, #16
5765 add fb_ptr, fb_ptr, #16*2
5769 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5770 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5772 and texture_block_ptr, texture_offset, texture_mask
5773 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5775 vld1.u32 { texels }, [texture_block_ptr, :128]
5777 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5778 subs height, height, #1
5780 add fb_ptr, fb_ptr, fb_ptr_pitch
5781 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5785 ldmia sp!, { r4 - r11, pc }
5790 #undef right_mask_bits
5793 #undef blocks_remaining
5803 #define right_width r5
5804 #define right_mask_bits r6
5808 #define fb_ptr_pitch r12
5810 // referenced by setup_sprites_16bpp_flush
5811 #define num_blocks r4
5813 #define block_width r11
5818 #define blocks_remaining r6
5821 #define right_mask q1
5822 #define test_mask q2
5823 #define draw_mask q2
5824 #define draw_mask_bits_fb_ptr d6
5829 function(setup_sprite_untextured)
5830 ldrh r12, [psx_gpu, #psx_gpu_render_state_offset]
5831 tst r12, #(RENDER_STATE_MASK_EVALUATE | RENDER_FLAGS_MODULATE_TEXELS \
5832 | RENDER_FLAGS_BLEND)
5833 ldrbeq r12, [psx_gpu, #psx_gpu_render_mode_offset]
5834 tsteq r12, #RENDER_INTERLACE_ENABLED
5835 beq setup_sprite_untextured_simple
5837 stmdb sp!, { r4 - r11, r14 }
5839 ldr width, [sp, #40]
5840 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5842 ldr height, [sp, #44]
5843 add fb_ptr, fb_ptr, y, lsl #11
5845 add fb_ptr, fb_ptr, x, lsl #1
5846 sub right_width, width, #1
5848 ldr color, [sp, #48]
5851 add block_width, width, #7
5855 mov right_mask_bits, #0xff
5857 sub fb_ptr_pitch, block_width, #1
5858 lsl right_mask_bits, right_width
5860 lsl fb_ptr_pitch, #3+1
5861 ubfx color_r, color, #3, #5
5863 rsb fb_ptr_pitch, #1024*2
5864 ubfx color_g, color, #11, #5
5866 vld1.u32 { test_mask }, [psx_gpu, :128]
5867 ubfx color_b, color, #19, #5
5869 vdup.u16 right_mask, right_mask_bits
5870 orr color, color_r, color_b, lsl #10
5872 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5873 orr color, color, color_g, lsl #5
5875 vtst.u16 right_mask, right_mask, test_mask
5876 add block, psx_gpu, #psx_gpu_blocks_offset
5878 vdup.u16 colors, color
5879 add block, block, num_blocks, lsl #6
5882 setup_sprite_untextured_height_loop:
5883 add num_blocks, block_width
5884 sub blocks_remaining, block_width, #1
5886 cmp num_blocks, #MAX_BLOCKS
5887 blgt setup_sprites_16bpp_flush
5889 cmp blocks_remaining, #0
5892 vmov.u8 draw_mask, #0 /* zero_mask */
5893 vmov.u8 draw_mask_bits_fb_ptr, #0
5896 vst1.u32 { draw_mask }, [block, :128]!
5897 subs blocks_remaining, #1
5899 vst1.u32 { colors }, [block, :128]
5900 add block, block, #24
5902 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5903 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5905 add block, block, #24
5910 vst1.u32 { right_mask }, [block, :128]!
5913 vst1.u32 { colors }, [block, :128]
5914 add block, block, #24
5916 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5917 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5919 add block, block, #24
5920 add fb_ptr, fb_ptr_pitch
5922 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5923 bgt setup_sprite_untextured_height_loop
5925 ldmia sp!, { r4 - r11, pc }
5929 #undef texture_page_ptr
5931 #undef dirty_textures_mask
5932 #undef current_texture_mask
5935 #define current_texture_page r1
5936 #define texture_page_ptr r2
5937 #define vram_ptr_a r3
5938 #define current_texture_page_x r12
5939 #define current_texture_page_y r4
5940 #define dirty_textures_mask r5
5944 #define current_texture_mask r9
5946 #define vram_ptr_b r11
5948 #define texel_block_a d0
5949 #define texel_block_b d1
5950 #define texel_block_expanded_a q1
5951 #define texel_block_expanded_b q2
5952 #define texel_block_expanded_ab q2
5953 #define texel_block_expanded_c q3
5954 #define texel_block_expanded_d q0
5955 #define texel_block_expanded_cd q3
5957 function(update_texture_4bpp_cache)
5958 stmdb sp!, { r4 - r11, r14 }
5961 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
5963 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
5964 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
5966 and current_texture_page_x, current_texture_page, #0xF
5967 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
5969 mov current_texture_page_y, current_texture_page, lsr #4
5970 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5972 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5975 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5976 bic dirty_textures_mask, current_texture_mask
5979 str dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5984 add vram_ptr_b, vram_ptr_a, #2048
5987 vld1.u32 { texel_block_a }, [vram_ptr_a, :64], c_4096
5988 vld1.u32 { texel_block_b }, [vram_ptr_b, :64], c_4096
5990 vmovl.u8 texel_block_expanded_a, texel_block_a
5991 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5992 vmovl.u8 texel_block_expanded_c, texel_block_b
5993 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5995 vbic.u16 texel_block_expanded_a, #0x00F0
5996 vbic.u16 texel_block_expanded_b, #0x00F0
5997 vbic.u16 texel_block_expanded_c, #0x00F0
5998 vbic.u16 texel_block_expanded_d, #0x00F0
6000 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
6001 texel_block_expanded_b
6002 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
6003 texel_block_expanded_d
6005 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
6006 [texture_page_ptr, :256]!
6008 subs sub_y, sub_y, #1
6012 add vram_ptr_a, vram_ptr_a, #8
6013 add vram_ptr_b, vram_ptr_b, #8
6015 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6016 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6018 subs tile_x, tile_x, #1
6022 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6023 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6025 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6026 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6028 subs tile_y, tile_y, #1
6032 ldmia sp!, { r4 - r11, pc }
6035 #undef current_texture_page
6038 #define texture_page r1
6039 #define texture_page_ptr r2
6040 #define vram_ptr_a r3
6041 #define texture_page_x r12
6042 #define texture_page_y r4
6043 #define current_texture_page r5
6048 #define vram_ptr_b r11
6060 function(update_texture_8bpp_cache_slice)
6061 stmdb sp!, { r4 - r11, r14 }
6064 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
6065 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
6067 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
6070 and texture_page_x, texture_page, #0xF
6071 mov texture_page_y, texture_page, lsr #4
6073 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6076 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6077 eor current_texture_page, current_texture_page, texture_page
6079 ands current_texture_page, current_texture_page, #0x1
6082 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6085 add vram_ptr_b, vram_ptr_a, #2048
6088 vld1.u32 { texels_a }, [vram_ptr_a, :128], c_4096
6089 vld1.u32 { texels_b }, [vram_ptr_b, :128], c_4096
6090 vld1.u32 { texels_c }, [vram_ptr_a, :128], c_4096
6091 vld1.u32 { texels_d }, [vram_ptr_b, :128], c_4096
6093 vst1.u32 { texels_a, texels_b }, [texture_page_ptr, :256]!
6094 vst1.u32 { texels_c, texels_d }, [texture_page_ptr, :256]!
6096 subs sub_y, sub_y, #1
6101 add vram_ptr_a, vram_ptr_a, #16
6102 add vram_ptr_b, vram_ptr_b, #16
6104 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6105 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6107 subs tile_x, tile_x, #1
6112 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6113 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6115 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6116 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6118 subs tile_y, tile_y, #1
6119 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6124 ldmia sp!, { r4 - r11, pc }
6127 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6128 function(scale2x_tiles8)
6132 add r12, r0, #1024*2
6136 vld1.u16 { q0 }, [r1, :128]!
6137 vld1.u16 { q2 }, [r1, :128]!
6143 vst1.u16 { q0, q1 }, [r0, :128]!
6144 vst1.u16 { q0, q1 }, [r12, :128]!
6146 vst1.u16 { q2, q3 }, [r0, :128]!
6147 vst1.u16 { q2, q3 }, [r12, :128]!
6154 sub r0, r0, r2, lsl #4+1
6156 add r12, r0, #1024*2
6162 // vim:filetype=armasm