2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
3 * Copyright (C) 2012 GraÅžvydas Ignotas "notaz" <notasas@gmail.com>
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of
8 * the License, or (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
16 #define RENDER_INTERLACE_ENABLED 0x1
19 #include "psx_gpu_offsets.h"
21 #define psx_gpu_b_dx_offset (psx_gpu_b_block_span_offset + 4)
23 #define edge_data_left_x_offset 0
24 #define edge_data_num_blocks_offset 2
25 #define edge_data_right_mask_offset 4
26 #define edge_data_y_offset 6
32 #define save_abi_regs() \
34 #define restore_abi_regs() \
37 #define save_abi_regs()
38 #define restore_abi_regs()
82 #define gw_by_l gw_bx_l
102 #define uvrg_xxxx0 q3
107 #define uvrg_xxxx1 q4
112 #define uvrg_xxxx2 q5
117 #define yyyy_uvrg0 q6
122 #define yyyy_uvrg1 q7
127 #define yyyy_uvrg2 q8
154 #define uvrg_base q11
197 #define uvrgb_phase q13
201 #include "arm_features.h"
203 #define function(name) FUNCTION(name):
205 #ifndef TEXRELS_FORBIDDEN
207 #define JT_OP_REL(table_label, index_reg, temp)
208 #define JT_OP(x...) x
209 #define JTE(start, target) target
213 #define JT_OP_REL(table_label, index_reg, temp) \
214 adr temp, table_label; \
215 ldr temp, [temp, index_reg, lsl #2]; \
219 #define JTE(start, target) (target - start)
224 #define flush_render_block_buffer _flush_render_block_buffer
225 #define update_texture_8bpp_cache _update_texture_8bpp_cache
233 function(compute_all_gradients)
234 // First compute the triangle area reciprocal and shift. The division will
235 // happen concurrently with much of the work which follows.
236 @ r12 = psx_gpu->triangle_area
237 ldr r12, [psx_gpu, #psx_gpu_triangle_area_offset]
238 stmdb sp!, { r4 - r11, lr }
241 @ load exponent of 62 into upper half of double
243 clz r14, r12 @ r14 = shift
245 movt r4, #((62 + 1023) << 4)
246 mov r12, r12, lsl r14 @ r12 = triangle_area_normalized
248 @ load area normalized into lower half of double
250 vmov.f64 d30, r5, r4 @ d30 = (1 << 62) + ta_n
252 movt r4, #((1022 + 31) << 4)
255 add r4, r4, r12, lsr #11
258 vdiv.f64 d30, d30, d31 @ d30 = ((1 << 62) + ta_n) / ta_n
260 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
261 // ( d0 * d1 ) - ( d2 * d3 ) =
262 // ( m0 ) - ( m1 ) = gradient
264 // This is split to do 12 elements at a time over three sets: a, b, and c.
265 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
266 // two of the slots are unused.
268 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
271 // First type is: uvrg bxxx xxxx
272 // Second type is: yyyy ybyy uvrg
273 // Since x_a and y_c are the same the same variable is used for both.
275 vld1.u32 { v0 }, [v_a, :128] @ v0 = { uvrg0, b0, x0, y0 }
276 ldrsh x0, [v_a, #8] @ load x0
278 vld1.u32 { v1 }, [v_b, :128] @ v1 = { uvrg1, b1, x1, y1}
279 ldrh x1, [v_b, #8] @ load x1
281 vld1.u32 { v2 }, [v_c, :128] @ v2 = { uvrg2, b2, x2, y2 }
282 ldrh x2, [v_c, #8] @ load x2
284 vmovl.u8 uvrg_xxxx0, uvrgb0 @ uvrg_xxxx0 = { uv0, rg0, b0-, -- }
285 ldrh y0, [v_a, #10] @ load y0
287 vmovl.u8 uvrg_xxxx1, uvrgb1 @ uvrg_xxxx1 = { uv1, rg1, b1-, -- }
288 ldrh y1, [v_b, #10] @ load y1
290 vmovl.u8 uvrg_xxxx2, uvrgb2 @ uvrg_xxxx2 = { uv2, rg2, b2-, -- }
291 ldrh y2, [v_c, #10] @ load y2
293 vmov.u8 uvrg0b, uvrg0 @ uvrg0b = { uv0, rg0 }
294 vdup.u16 xxxx0, x0_y0[0] @ xxxx0 = { xx0, xx0 }
296 orr x1_x2, x1, x2, lsl #16 @ x1_x2 = { x1, x2 }
297 pkhbt x0_x1, x0, x1, lsl #16 @ x0_x1 = { x0, x1 }
299 vmov.u8 uvrg1b, uvrg1 @ uvrg1b = { uv1, rg1 }
300 vdup.u16 xxxx1, x1_y1[0] @ xxxx1 = { xx1, xx1 }
302 vmov.u8 uvrg2b, uvrg2 @ uvrg2b = { uv2, rg2 }
303 vdup.u16 xxxx2, x2_y2[0] @ xxxx2 = { xx2, xx2 }
305 ldrb b2, [v_c, #4] @ load b2
306 orr y0_y1, y0, y1, lsl #16 @ y0_y1 = { y0, y1 }
308 ldrb b1, [v_b, #4] @ load b1
309 orr y1_y2, y1, y2, lsl #16 @ y1_y2 = { y1, y2 }
311 vdup.u16 yyyy0, x0_y0[1] @ yyyy0 = { yy0, yy0 }
312 vsub.s16 d0_ab, x1_ab, x0_ab
314 ldrb b0, [v_a, #4] @ load b0
315 orr b1_b2, b1, b2, lsl #16 @ b1_b2 = { b1, b2 }
317 vdup.u16 yyyy1, x1_y1[1] @ yyyy1 = { yy1, yy1 }
318 vsub.s16 d2_ab, x2_ab, x1_ab
320 vdup.u16 yyyy2, x2_y2[1] @ yyyy2 = { yy2, yy2 }
321 vsub.s16 d1_ab, y2_ab, y1_ab
323 orr b0_b1, b0, b1, lsl #16 @ b1_b2 = { b1, b2 }
324 ssub16 dx, x1_x2, x0_x1 @ dx = { x1 - x0, x2 - x1 }
326 ssub16 dy, y1_y2, y0_y1 @ dy = { y1 - y0, y2 - y1 }
327 ssub16 db, b1_b2, b0_b1 @ db = { b1 - b0, b2 - b1 }
329 vsub.s16 d3_ab, y1_ab, y0_ab
330 smusdx ga_by, dx, db @ ga_by = ((x1 - x0) * (b2 - b1)) -
331 @ ((x2 - X1) * (b1 - b0))
332 vmull.s16 ga_uvrg_x, d0_a, d1_a
333 smusdx ga_bx, db, dy @ ga_bx = ((b1 - b0) * (y2 - y1)) -
334 @ ((b2 - b1) * (y1 - y0))
335 vmlsl.s16 ga_uvrg_x, d2_a, d3_a
336 movs gs_bx, ga_bx, asr #31
338 vmull.s16 ga_uvrg_y, d0_b, d1_b
339 rsbmi ga_bx, ga_bx, #0
341 @ r12 = psx_gpu->uvrgb_phase
342 ldr r12, [psx_gpu, #psx_gpu_uvrgb_phase_offset]
344 vmlsl.s16 ga_uvrg_y, d2_b, d3_b
345 movs gs_by, ga_by, asr #31
347 vshr.u64 d0, d30, #22
348 add b_base, r12, b0, lsl #16
350 vdup.u32 uvrgb_phase, r12
352 rsbmi ga_by, ga_by, #0
353 vclt.s32 gs_uvrg_x, ga_uvrg_x, #0 @ gs_uvrg_x = ga_uvrg_x < 0
355 @ r12 = psx_gpu->triangle_winding_offset
356 ldrb r12, [psx_gpu, #psx_gpu_triangle_winding_offset]
357 vclt.s32 gs_uvrg_y, ga_uvrg_y, #0 @ gs_uvrg_y = ga_uvrg_y < 0
359 rsb r12, r12, #0 @ r12 = -(triangle->winding)
361 vdup.u32 w_mask, r12 @ w_mask = { -w, -w, -w, -w }
362 sub r14, r14, #(62 - 12) @ r14 = shift - (62 - FIXED_BITS)
364 vshll.u16 uvrg_base, uvrg0, #16 @ uvrg_base = uvrg0 << 16
365 vdup.u32 r_shift, r14 @ r_shift = { shift, shift*, shift, shift* }
366 @ * - vshl.u64: ignored by hw
367 vadd.u32 uvrg_base, uvrgb_phase
368 vabs.s32 ga_uvrg_x, ga_uvrg_x @ ga_uvrg_x = abs(ga_uvrg_x)
370 vmov area_r_s, s0 @ area_r_s = triangle_reciprocal
371 vabs.s32 ga_uvrg_y, ga_uvrg_y @ ga_uvrg_y = abs(ga_uvrg_y)
373 vmull.u32 gw_rg_x, ga_rg_x, d0[0]
374 vmull.u32 gw_uv_x, ga_uv_x, d0[0]
375 vmull.u32 gw_rg_y, ga_rg_y, d0[0]
376 vmull.u32 gw_uv_y, ga_uv_y, d0[0]
378 vshl.u64 gw_rg_x, gw_rg_x, r_shift
379 vshl.u64 gw_uv_x, gw_uv_x, r_shift
380 vshl.u64 gw_rg_y, gw_rg_y, r_shift
381 vshl.u64 gw_uv_y, gw_uv_y, r_shift
383 veor.u32 gs_uvrg_x, gs_uvrg_x, w_mask
384 vmovn.u64 g_uv_x, gw_uv_x
386 veor.u32 gs_uvrg_y, gs_uvrg_y, w_mask
387 vmovn.u64 g_rg_x, gw_rg_x
389 veor.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
390 vmovn.u64 g_uv_y, gw_uv_y
392 vsub.u32 g_uvrg_x, g_uvrg_x, gs_uvrg_x
393 vmovn.u64 g_rg_y, gw_rg_y
395 veor.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
396 mov ga_bx, ga_bx, lsl #13
398 vsub.u32 g_uvrg_y, g_uvrg_y, gs_uvrg_y
399 mov ga_by, ga_by, lsl #13
402 umull gw_bx_l, gw_bx_h, ga_bx, area_r_s
404 vshl.u32 g_uvrg_x, g_uvrg_x, #4
405 vshl.u32 g_uvrg_y, g_uvrg_y, #4
407 umull gw_by_l, gw_by_h, ga_by, area_r_s
408 vmls.s32 uvrg_base, ga_uvrg_x, x0_y0[0]
410 eor gs_bx, gs_bx, r12
411 vadd.u32 uvrg_dx2, uvrg_dx1, uvrg_dx1
413 veor.u32 uvrg_dx0, uvrg_dx0, uvrg_dx0
414 eor gs_by, gs_by, r12
416 rsb r11, r14, #0 @ r11 = negative shift for scalar lsr
417 add store_a, psx_gpu, #psx_gpu_uvrg_offset
419 sub r11, r11, #(32 - 13)
421 add store_b, store_a, #16
424 vadd.u32 uvrg_dx3, uvrg_dx2, uvrg_dx1
425 vst1.u32 { uvrg_base }, [store_a, :128], store_inc
427 vst1.u32 { uvrg_dx1 }, [store_b, :128], store_inc
428 mov g_bx, gw_bx_h, lsr r11
430 vst1.u32 { g_uvrg_y }, [store_a, :128], store_inc
431 mov g_by, gw_by_h, lsr r11
433 vst4.u32 { uvrg_dx0l, uvrg_dx1l, uvrg_dx2l, uvrg_dx3l }, \
434 [store_b, :128], store_inc
435 eor g_bx, g_bx, gs_bx
437 vst4.u32 { uvrg_dx0h, uvrg_dx1h, uvrg_dx2h, uvrg_dx3h }, \
438 [store_b, :128], store_inc
439 sub g_bx, g_bx, gs_bx
442 eor g_by, g_by, gs_by
444 mls b_base, g_bx, x0, b_base
445 sub g_by, g_by, gs_by
450 add g_bx2, g_bx, g_bx
451 add g_bx3, g_bx, g_bx2
453 stmia store_b, { g_bx0, g_bx, g_bx2, g_bx3, b_base, g_by }
456 ldmia sp!, { r4 - r11, pc }
473 #define height_minor_a r7
474 #define height_minor_b r8
475 #define height_major r9
478 #define reciprocal_table_ptr r10
480 #define edge_alt_low r4
481 #define edge_alt_high r5
482 #define edge_dx_dy_alt r6
483 #define edge_shift_alt r10
485 #define edge_dx_dy_alt_low r4
486 #define edge_dx_dy_alt_high r5
488 #define span_edge_data r4
489 #define span_uvrg_offset r5
490 #define span_b_offset r6
498 #define alternate_x q0
499 #define alternate_dx_dy q1
500 #define alternate_x_32 q2
502 #define alternate_x_low d0
503 #define alternate_x_high d1
504 #define alternate_dx_dy_low d2
505 #define alternate_dx_dy_high d3
506 #define alternate_x_32_low d4
507 #define alternate_x_32_high d5
511 #define left_dx_dy q5
512 #define right_dx_dy q6
514 #define right_edge q8
516 #define left_x_low d6
517 #define left_x_high d7
518 #define right_x_low d8
519 #define right_x_high d9
520 #define left_dx_dy_low d10
521 #define left_dx_dy_high d11
522 #define right_dx_dy_low d12
523 #define right_dx_dy_high d13
524 #define left_edge_low d14
525 #define left_edge_high d15
526 #define right_edge_low d16
527 #define right_edge_high d17
529 #define y_mid_point d18
532 #define left_right_x_16 q11
533 #define span_shifts_y q12
536 #define span_shifts d24
541 #define left_right_x_16_low d22
542 #define left_right_x_16_high d23
547 #define alternate_x_16 d4
550 #define v_clip_low d6
552 #define right_x_32 q10
553 #define left_x_32 q11
554 #define alternate_select d24
556 #define right_x_32_low d20
557 #define right_x_32_high d21
558 #define left_x_32_low d22
559 #define left_x_32_high d23
561 #define tmp_max_blocks d20
564 #define edges_dx_dy d2
565 #define edge_shifts d3
566 #define edge_shifts_64 q2
568 #define edges_xy_left d0
569 #define edges_xy_right d1
571 #define height_reciprocals d6
579 #define heights_b d12
580 #define edges_dx_dy_64 q10
582 #define edges_dx_dy_64_left d20
583 #define edges_dx_dy_64_right d21
586 #define setup_spans_prologue() \
587 stmdb sp!, { r4 - r11, lr }; \
590 ldrsh x_a, [v_a, #8]; \
591 ldrsh x_b, [v_b, #8]; \
592 ldrsh x_c, [v_c, #8]; \
593 ldrsh y_a, [v_a, #10]; \
594 ldrsh y_b, [v_b, #10]; \
595 ldrsh y_c, [v_c, #10]; \
597 add temp, psx_gpu, #psx_gpu_uvrg_offset; \
598 vld1.32 { uvrg }, [temp]; \
599 add temp, psx_gpu, #psx_gpu_uvrg_dy_offset; \
600 vld1.32 { uvrg_dy }, [temp]; \
601 ldr reciprocal_table_ptr, [psx_gpu, #psx_gpu_reciprocal_table_ptr_offset]; \
603 vmov.u32 c_0x01, #0x01 \
605 #define setup_spans_load_b() \
606 ldr b, [psx_gpu, #psx_gpu_b_offset]; \
607 ldr b_dy, [psx_gpu, #psx_gpu_b_dy_offset] \
609 #define setup_spans_prologue_b() \
610 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
611 add temp, psx_gpu, #psx_gpu_viewport_start_x_offset; \
613 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
614 vmov.u16 c_0x0004, #0x0004; \
616 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
617 vmov.u16 c_0x0001, #0x0001; \
619 vld1.u16 { left_edge_low[], left_edge_high[] }, [temp]; \
620 add temp, psx_gpu, #psx_gpu_viewport_end_x_offset; \
622 vld1.u16 { right_edge_low[], right_edge_high[] }, [temp]; \
623 vadd.u16 right_edge, right_edge, c_0x0001; \
625 vmov.u16 c_0x0007, #0x0007; \
626 vmvn.u16 c_0xFFFE, #0x0001 \
629 #define compute_edge_delta_x2() \
630 ldr temp, [reciprocal_table_ptr, height, lsl #2]; \
632 vdup.u32 heights, height; \
633 vsub.u32 widths, x_ends, x_starts; \
635 vdup.u32 edge_shifts, temp; \
636 vsub.u32 heights_b, heights, c_0x01; \
637 vshr.u32 height_reciprocals, edge_shifts, #10; \
639 vmla.s32 heights_b, x_starts, heights; \
640 vbic.u16 edge_shifts, #0xE0; \
641 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
642 vmull.s32 edges_xy, heights_b, height_reciprocals \
645 #define height_reciprocal_alt r11
646 #define height_b_alt r12
648 #define compute_edge_delta_x3(start_c, height_a, height_b) \
649 vmov heights, height_a, height_b; \
650 ldr temp, [reciprocal_table_ptr, height_a, lsl #2]; \
651 vmov.u32 edge_shifts[0], temp; \
652 ldr temp, [reciprocal_table_ptr, height_b, lsl #2]; \
653 vmov.u32 edge_shifts[1], temp; \
654 ldr edge_shift_alt, [reciprocal_table_ptr, height_minor_b, lsl #2]; \
656 vsub.u32 widths, x_ends, x_starts; \
657 sub width_alt, x_c, start_c; \
659 vsub.u32 heights_b, heights, c_0x01; \
660 sub height_b_alt, height_minor_b, #1; \
662 vshr.u32 height_reciprocals, edge_shifts, #10; \
663 lsr height_reciprocal_alt, edge_shift_alt, #10; \
665 vmla.s32 heights_b, x_starts, heights; \
666 mla height_b_alt, height_minor_b, start_c, height_b_alt; \
668 vbic.u16 edge_shifts, #0xE0; \
669 and edge_shift_alt, edge_shift_alt, #0x1F; \
671 vmul.s32 edges_dx_dy, widths, height_reciprocals; \
672 mul edge_dx_dy_alt, width_alt, height_reciprocal_alt; \
674 vmull.s32 edges_xy, heights_b, height_reciprocals; \
675 smull edge_alt_low, edge_alt_high, height_b_alt, height_reciprocal_alt \
678 #define setup_spans_adjust_y_up() \
679 vsub.u32 y_x4, y_x4, c_0x0004 \
681 #define setup_spans_adjust_y_down() \
682 vadd.u32 y_x4, y_x4, c_0x0004 \
684 #define setup_spans_adjust_interpolants_up() \
685 vsub.u32 uvrg, uvrg, uvrg_dy; \
688 #define setup_spans_adjust_interpolants_down() \
689 vadd.u32 uvrg, uvrg, uvrg_dy; \
693 #define setup_spans_clip_interpolants_increment() \
694 mla b, b_dy, clip, b; \
695 vmla.s32 uvrg, uvrg_dy, v_clip \
697 #define setup_spans_clip_interpolants_decrement() \
698 mls b, b_dy, clip, b; \
699 vmls.s32 uvrg, uvrg_dy, v_clip \
701 #define setup_spans_clip_alternate_yes() \
702 smlal edge_alt_low, edge_alt_high, edge_dx_dy_alt, clip \
704 #define setup_spans_clip_alternate_no() \
706 #define setup_spans_clip(direction, alternate_active) \
707 vdup.u32 v_clip, clip; \
708 setup_spans_clip_alternate_##alternate_active(); \
709 setup_spans_clip_interpolants_##direction(); \
710 vmlal.s32 edges_xy, edges_dx_dy, v_clip_low \
713 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
714 vmovl.s32 edge_shifts_64, edge_shifts; \
715 vmovl.s32 edges_dx_dy_64, edges_dx_dy; \
717 vshl.s64 edges_xy, edges_xy, edge_shifts_64; \
718 vshl.s64 edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64; \
720 vmov left_x_low, edges_xy_##left_index; \
721 vmov right_x_low, edges_xy_##right_index; \
723 vmov left_dx_dy_low, edges_dx_dy_64_##left_index; \
724 vmov left_dx_dy_high, edges_dx_dy_64_##left_index; \
725 vmov right_dx_dy_low, edges_dx_dy_64_##right_index; \
726 vmov right_dx_dy_high, edges_dx_dy_64_##right_index; \
728 vadd.u64 left_x_high, left_x_low, left_dx_dy_low; \
729 vadd.u64 right_x_high, right_x_low, right_dx_dy_low; \
731 vadd.u64 left_dx_dy, left_dx_dy, left_dx_dy; \
732 vadd.u64 right_dx_dy, right_dx_dy, right_dx_dy \
735 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
736 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
738 vdup.u16 y_mid_point, y_b; \
739 rsb temp, edge_shift_alt, #32; \
741 lsl edge_alt_high, edge_alt_high, edge_shift_alt; \
742 orr edge_alt_high, edge_alt_high, edge_alt_low, lsr temp; \
743 lsl edge_alt_low, edge_alt_low, edge_shift_alt; \
744 vmov alternate_x_low, edge_alt_low, edge_alt_high; \
746 asr edge_dx_dy_alt_high, edge_dx_dy_alt, temp; \
747 lsl edge_dx_dy_alt_low, edge_dx_dy_alt, edge_shift_alt; \
748 vmov alternate_dx_dy_low, edge_dx_dy_alt_low, edge_dx_dy_alt_high; \
749 vmov alternate_dx_dy_high, alternate_dx_dy_low; \
751 vadd.u64 alternate_x_high, alternate_x_low, alternate_dx_dy_low; \
752 vadd.u64 alternate_dx_dy, alternate_dx_dy, alternate_dx_dy \
755 #define setup_spans_y_select_up() \
756 vclt.s16 alternate_select, y_x4, y_mid_point \
758 #define setup_spans_y_select_down() \
759 vcgt.s16 alternate_select, y_x4, y_mid_point \
762 #define setup_spans_alternate_select_left() \
763 vbit.u16 left_right_x_16_low, alternate_x_16, alternate_select \
765 #define setup_spans_alternate_select_right() \
766 vbit.u16 left_right_x_16_high, alternate_x_16, alternate_select \
769 #define setup_spans_set_x4_alternate_yes(alternate, direction) \
770 vshrn.s64 alternate_x_32_low, alternate_x, #32; \
771 vshrn.s64 left_x_32_low, left_x, #32; \
772 vshrn.s64 right_x_32_low, right_x, #32; \
774 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
775 vadd.u64 left_x, left_x, left_dx_dy; \
776 vadd.u64 right_x, right_x, right_dx_dy; \
778 vshrn.s64 alternate_x_32_high, alternate_x, #32; \
779 vshrn.s64 left_x_32_high, left_x, #32; \
780 vshrn.s64 right_x_32_high, right_x, #32; \
782 vadd.u64 alternate_x, alternate_x, alternate_dx_dy; \
783 vadd.u64 left_x, left_x, left_dx_dy; \
784 vadd.u64 right_x, right_x, right_dx_dy; \
786 vmovn.u32 alternate_x_16, alternate_x_32; \
787 setup_spans_y_select_##direction(); \
788 vmovn.u32 left_right_x_16_low, left_x_32; \
790 vmovn.u32 left_right_x_16_high, right_x_32; \
791 setup_spans_alternate_select_##alternate(); \
793 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
794 str b, [span_b_offset], #4; \
795 setup_spans_adjust_interpolants_##direction(); \
797 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
799 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
800 str b, [span_b_offset], #4; \
801 setup_spans_adjust_interpolants_##direction(); \
803 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
805 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
806 str b, [span_b_offset], #4; \
807 setup_spans_adjust_interpolants_##direction(); \
809 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
810 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
811 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
813 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
814 str b, [span_b_offset], #4; \
815 setup_spans_adjust_interpolants_##direction(); \
817 vmov.u16 tmp_max_blocks, #MAX_BLOCKS_PER_ROW; \
818 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
819 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
820 vmin.u16 left_right_x_16_high, left_right_x_16_high, tmp_max_blocks; \
822 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
824 setup_spans_adjust_y_##direction() \
827 #define setup_spans_set_x4_alternate_no(alternate, direction) \
828 vshrn.s64 left_x_32_low, left_x, #32; \
829 vshrn.s64 right_x_32_low, right_x, #32; \
831 vadd.u64 left_x, left_x, left_dx_dy; \
832 vadd.u64 right_x, right_x, right_dx_dy; \
834 vshrn.s64 left_x_32_high, left_x, #32; \
835 vshrn.s64 right_x_32_high, right_x, #32; \
837 vadd.u64 left_x, left_x, left_dx_dy; \
838 vadd.u64 right_x, right_x, right_dx_dy; \
840 vmovn.u32 left_right_x_16_low, left_x_32; \
841 vmovn.u32 left_right_x_16_high, right_x_32; \
843 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
844 str b, [span_b_offset], #4; \
845 setup_spans_adjust_interpolants_##direction(); \
847 vmax.s16 left_right_x_16, left_right_x_16, left_edge; \
849 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
850 str b, [span_b_offset], #4; \
851 setup_spans_adjust_interpolants_##direction(); \
853 vmin.s16 left_right_x_16, left_right_x_16, right_edge; \
855 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
856 str b, [span_b_offset], #4; \
857 setup_spans_adjust_interpolants_##direction(); \
859 vsub.u16 left_right_x_16_high, left_right_x_16_high, left_right_x_16_low; \
860 vadd.u16 left_right_x_16_high, left_right_x_16_high, c_0x0007; \
861 vand.u16 span_shifts, left_right_x_16_high, c_0x0007; \
863 vst1.u32 { uvrg }, [span_uvrg_offset, :128]!; \
864 str b, [span_b_offset], #4; \
865 setup_spans_adjust_interpolants_##direction(); \
867 vmov.u16 tmp_max_blocks, #MAX_BLOCKS_PER_ROW; \
868 vshr.u16 left_right_x_16_high, left_right_x_16_high, #3; \
869 vshl.u16 span_shifts, c_0xFFFE, span_shifts; \
870 vmin.u16 left_right_x_16_high, left_right_x_16_high, tmp_max_blocks; \
872 vst4.u16 { left_right_x_16, span_shifts_y }, [span_edge_data]!; \
874 setup_spans_adjust_y_##direction() \
877 #define edge_adjust_low r11
878 #define edge_adjust_high r12
880 #define setup_spans_alternate_adjust_yes() \
881 smull edge_adjust_low, edge_adjust_high, edge_dx_dy_alt, height_minor_a; \
882 subs edge_alt_low, edge_alt_low, edge_adjust_low; \
883 sbc edge_alt_high, edge_alt_high, edge_adjust_high \
885 #define setup_spans_alternate_adjust_no() \
888 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
889 setup_spans_alternate_adjust_##alternate_active(); \
890 setup_spans_load_b(); \
892 ldrsh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
893 subs y_c, y_c, temp; \
894 subgt height, height, y_c; \
895 addgt height, height, #1; \
897 ldrsh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
898 subs clip, temp, y_a; \
901 sub height, height, clip; \
902 add y_a, y_a, clip; \
903 setup_spans_clip(increment, alternate_active); \
909 orr temp, y_a, y_a, lsl #16; \
911 add temp, temp, #(1 << 16); \
912 movgt height, #512; \
914 add y_a, y_a, #(2 << 16); \
915 vmov y_x4, temp, y_a; \
917 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
919 setup_spans_prologue_b(); \
921 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
924 setup_spans_set_x4_alternate_##alternate_active(alternate, down); \
925 subs height, height, #4; \
931 #define setup_spans_alternate_pre_increment_yes() \
932 adds edge_alt_low, edge_alt_low, edge_dx_dy_alt; \
933 adc edge_alt_high, edge_alt_high, edge_dx_dy_alt, asr #31 \
935 #define setup_spans_alternate_pre_increment_no() \
938 #define setup_spans_up_decrement_yes() \
939 suble height, height, #1 \
941 #define setup_spans_up_decrement_no() \
944 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
945 setup_spans_alternate_adjust_##alternate_active(); \
946 setup_spans_load_b(); \
949 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]; \
950 subs temp, temp, y_c; \
951 subgt height, height, temp; \
952 setup_spans_up_decrement_##alternate_active(); \
954 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]; \
955 subs clip, y_a, temp; \
958 sub height, height, clip; \
959 sub y_a, y_a, clip; \
960 setup_spans_clip(decrement, alternate_active); \
966 orr temp, y_a, y_a, lsl #16; \
968 sub temp, temp, #(1 << 16); \
969 movgt height, #512; \
971 sub y_a, y_a, #(2 << 16); \
972 vmov y_x4, temp, y_a; \
974 vaddw.s32 edges_xy, edges_xy, edges_dx_dy; \
976 setup_spans_alternate_pre_increment_##alternate_active(); \
977 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
979 setup_spans_adjust_interpolants_up(); \
980 setup_spans_prologue_b(); \
982 strh height, [psx_gpu, #psx_gpu_num_spans_offset]; \
985 setup_spans_set_x4_alternate_##alternate_active(alternate, up); \
986 subs height, height, #4; \
992 #define setup_spans_epilogue() \
993 restore_abi_regs(); \
994 ldmia sp!, { r4 - r11, pc } \
997 #define setup_spans_up_up(minor, major) \
998 setup_spans_prologue(); \
999 sub height_minor_a, y_a, y_b; \
1000 sub height_minor_b, y_b, y_c; \
1001 sub height, y_a, y_c; \
1003 vdup.u32 x_starts, x_a; \
1004 vmov x_ends, x_c, x_b; \
1006 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1007 setup_spans_up(major, minor, minor, yes); \
1008 setup_spans_epilogue() \
1010 function(setup_spans_up_left)
1011 setup_spans_up_up(left, right)
1013 function(setup_spans_up_right)
1014 setup_spans_up_up(right, left)
1016 #define setup_spans_down_down(minor, major) \
1017 setup_spans_prologue(); \
1018 sub height_minor_a, y_b, y_a; \
1019 sub height_minor_b, y_c, y_b; \
1020 sub height, y_c, y_a; \
1022 vdup.u32 x_starts, x_a; \
1023 vmov x_ends, x_c, x_b; \
1025 compute_edge_delta_x3(x_b, height_major, height_minor_a); \
1026 setup_spans_down(major, minor, minor, yes); \
1027 setup_spans_epilogue() \
1029 function(setup_spans_down_left)
1030 setup_spans_down_down(left, right)
1032 function(setup_spans_down_right)
1033 setup_spans_down_down(right, left)
1036 #define setup_spans_up_flat() \
1037 sub height, y_a, y_c; \
1039 compute_edge_delta_x2(); \
1040 setup_spans_up(left, right, none, no); \
1041 setup_spans_epilogue() \
1043 function(setup_spans_up_a)
1044 setup_spans_prologue()
1046 vmov x_starts, x_a, x_b
1047 vdup.u32 x_ends, x_c
1049 setup_spans_up_flat()
1051 function(setup_spans_up_b)
1052 setup_spans_prologue()
1054 vdup.u32 x_starts, x_a
1055 vmov x_ends, x_b, x_c
1057 setup_spans_up_flat()
1059 #define setup_spans_down_flat() \
1060 sub height, y_c, y_a; \
1062 compute_edge_delta_x2(); \
1063 setup_spans_down(left, right, none, no); \
1064 setup_spans_epilogue() \
1066 function(setup_spans_down_a)
1067 setup_spans_prologue()
1069 vmov x_starts, x_a, x_b
1070 vdup.u32 x_ends, x_c
1072 setup_spans_down_flat()
1074 function(setup_spans_down_b)
1075 setup_spans_prologue()
1077 vdup.u32 x_starts, x_a
1078 vmov x_ends, x_b, x_c
1080 setup_spans_down_flat()
1085 #define edges_xy_b q11
1086 #define edges_dx_dy_b d26
1087 #define edge_shifts_b d27
1088 #define edges_dx_dy_and_shifts_b q13
1089 #define height_increment d20
1091 #define edges_dx_dy_and_shifts q1
1093 #define edges_xy_b_left d22
1094 #define edges_xy_b_right d23
1096 #define setup_spans_up_down_load_edge_set_b() \
1097 vmov edges_xy, edges_xy_b; \
1098 vmov edges_dx_dy_and_shifts, edges_dx_dy_and_shifts_b \
1101 function(setup_spans_up_down)
1102 setup_spans_prologue()
1104 // s32 middle_y = y_a;
1105 sub height_minor_a, y_a, y_b
1106 sub height_minor_b, y_c, y_a
1107 sub height_major, y_c, y_b
1109 vmov x_starts, x_a, x_c
1110 vdup.u32 x_ends, x_b
1112 compute_edge_delta_x3(x_a, height_minor_a, height_major)
1115 vmov height_increment, temp, height_minor_b
1116 vmlal.s32 edges_xy, edges_dx_dy, height_increment
1118 vmov edges_xy_b_left, edge_alt_low, edge_alt_high
1119 vmov edges_xy_b_right, edges_xy_right
1121 vmov edge_shifts_b, edge_shifts
1122 vmov.u32 edge_shifts_b[0], edge_shift_alt
1124 vneg.s32 edges_dx_dy_b, edges_dx_dy
1125 vmov.u32 edges_dx_dy_b[0], edge_dx_dy_alt
1129 setup_spans_load_b()
1132 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1133 subs temp, temp, y_b
1134 subgt height_minor_a, height_minor_a, temp
1136 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1137 subs clip, y_a, temp
1140 sub height_minor_a, height_minor_a, clip
1142 setup_spans_clip(decrement, no)
1145 cmp height_minor_a, #0
1148 orr temp, y_a, y_a, lsl #16
1149 sub temp, temp, #(1 << 16)
1151 sub y_a, y_a, #(2 << 16)
1152 vmov y_x4, temp, y_a
1154 vaddw.s32 edges_xy, edges_xy, edges_dx_dy
1156 strh height_minor_a, [psx_gpu, #psx_gpu_num_spans_offset]
1158 setup_spans_adjust_edges_alternate_no(left, right);
1159 setup_spans_adjust_interpolants_up()
1160 setup_spans_up_down_load_edge_set_b()
1162 setup_spans_prologue_b()
1166 setup_spans_set_x4_alternate_no(none, up)
1167 subs height_minor_a, height_minor_a, #4
1170 add span_edge_data, span_edge_data, height_minor_a, lsl #3
1171 add span_uvrg_offset, span_uvrg_offset, height_minor_a, lsl #4
1172 add span_b_offset, span_b_offset, height_minor_a, lsl #2
1175 add temp, psx_gpu, #psx_gpu_uvrg_offset
1176 vld1.32 { uvrg }, [temp]
1179 setup_spans_load_b()
1181 ldrh temp, [psx_gpu, #psx_gpu_viewport_end_y_offset]
1183 subgt height_minor_b, height_minor_b, y_c
1184 addgt height_minor_b, height_minor_b, #1
1186 ldrh temp, [psx_gpu, #psx_gpu_viewport_start_y_offset]
1187 subs clip, temp, y_a
1190 sub height_minor_b, height_minor_b, clip
1192 setup_spans_clip(increment, no)
1195 cmp height_minor_b, #0
1198 orr temp, y_a, y_a, lsl #16
1199 add temp, temp, #(1 << 16)
1201 add y_a, y_a, #(2 << 16)
1202 vmov y_x4, temp, y_a
1204 setup_spans_adjust_edges_alternate_no(left, right)
1206 ldrh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1207 add temp, temp, height_minor_b
1209 cmp temp, #MAX_SPANS
1212 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1215 setup_spans_set_x4_alternate_no(none, down)
1216 subs height_minor_b, height_minor_b, #4
1220 setup_spans_epilogue()
1223 setup_spans_up_down_load_edge_set_b()
1224 setup_spans_prologue_b()
1228 // FIXME: overflow corner case
1229 sub temp, temp, height_minor_b
1230 bics height_minor_b, #3
1231 add temp, temp, height_minor_b
1232 strh temp, [psx_gpu, #psx_gpu_num_spans_offset]
1236 #undef span_uvrg_offset
1237 #undef span_edge_data
1238 #undef span_b_offset
1243 #define num_spans r1
1244 #define span_uvrg_offset r2
1245 #define span_edge_data r3
1246 #define span_b_offset r4
1248 #define span_num_blocks r6
1252 #define dither_offset_ptr r10
1253 #define block_ptr_a r11
1255 #define num_blocks r14
1257 #define uvrg_dx_ptr r2
1258 #define texture_mask_ptr r3
1259 #define dither_shift r8
1260 #define dither_row r10
1265 #define block_ptr_b r10
1267 #define block_span_ptr r10
1268 #define right_mask r8
1287 #define b_whole_8 d14
1288 #define fb_mask_ptrs d15
1301 #define u_whole_low d16
1302 #define u_whole_high d17
1303 #define v_whole_low d18
1304 #define v_whole_high d19
1305 #define r_whole_low d20
1306 #define r_whole_high d21
1307 #define g_whole_low d22
1308 #define g_whole_high d23
1309 #define b_whole_low d24
1310 #define b_whole_high d25
1315 #define u_whole_8 d26
1316 #define v_whole_8 d27
1317 #define u_whole_8b d24
1318 #define r_whole_8 d24
1319 #define g_whole_8 d25
1321 #define uv_whole_8 q13
1322 #define uv_whole_8b q14
1324 #define dither_offsets q14
1325 #define texture_mask q15
1326 #define texture_mask_u d30
1327 #define texture_mask_v d31
1329 #define dither_offsets_short d28
1333 #define block_span q10
1338 #define draw_mask q1
1339 #define draw_mask_edge q13
1340 #define test_mask q0
1346 #define setup_blocks_texture_swizzled() \
1347 vand.u8 u_whole_8b, u_whole_8, texture_mask_u; \
1348 vsli.u8 u_whole_8, v_whole_8, #4; \
1349 vsri.u8 v_whole_8, u_whole_8b, #4 \
1351 #define setup_blocks_texture_unswizzled() \
1354 #define setup_blocks_shaded_textured_builder(swizzling) \
1357 function(setup_blocks_shaded_textured_dithered_##swizzling##_indirect) \
1358 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1359 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1361 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1362 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1364 cmp num_spans, #0; \
1367 stmdb sp!, { r4 - r11, r14 }; \
1369 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1371 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
1372 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1374 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1375 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1377 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1378 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1380 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
1381 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1383 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1386 vmov.u8 fb_mask_ptrs, #0; \
1388 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1389 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1391 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1392 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1394 cmp span_num_blocks, #0; \
1397 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1398 add num_blocks, span_num_blocks, num_blocks; \
1400 cmp num_blocks, #MAX_BLOCKS; \
1404 ldr b, [span_b_offset]; \
1405 add fb_ptr, fb_ptr, y, lsl #11; \
1407 vdup.u32 v_left_x, left_x; \
1410 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1411 add fb_ptr, fb_ptr, left_x, lsl #1; \
1413 mla b, b_dx, left_x, b; \
1414 and dither_shift, left_x, #0x03; \
1416 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1417 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1419 mov dither_shift, dither_shift, lsl #3; \
1420 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1423 subs span_num_blocks, span_num_blocks, #1; \
1425 mov dither_row, dither_row, ror dither_shift; \
1426 mov b_dx4, b_dx, lsl #2; \
1428 vdup.u32 dither_offsets_short, dither_row; \
1429 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1431 vdup.u32 b_block, b; \
1432 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1434 vdup.u32 u_block, uv[0]; \
1435 mov b_dx8, b_dx, lsl #3; \
1437 vdup.u32 v_block, uv[1]; \
1438 vdup.u32 r_block, rg[0]; \
1439 vdup.u32 g_block, rg[1]; \
1441 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1443 vadd.u32 u_block, u_block, block_span; \
1444 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1446 vadd.u32 v_block, v_block, block_span; \
1447 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1449 vadd.u32 r_block, r_block, block_span; \
1450 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1452 vadd.u32 g_block, g_block, block_span; \
1453 vld1.u32 { block_span }, [block_span_ptr, :128]; \
1455 vadd.u32 b_block, b_block, block_span; \
1456 add block_ptr_b, block_ptr_a, #16; \
1458 vshrn.u32 u_whole_low, u_block, #16; \
1459 vshrn.u32 v_whole_low, v_block, #16; \
1460 vshrn.u32 r_whole_low, r_block, #16; \
1461 vshrn.u32 g_whole_low, g_block, #16; \
1463 vdup.u32 dx4, uv_dx4[0]; \
1464 vshrn.u32 b_whole_low, b_block, #16; \
1466 vaddhn.u32 u_whole_high, u_block, dx4; \
1467 vdup.u32 dx4, uv_dx4[1]; \
1469 vaddhn.u32 v_whole_high, v_block, dx4; \
1470 vdup.u32 dx4, rg_dx4[0]; \
1472 vaddhn.u32 r_whole_high, r_block, dx4; \
1473 vdup.u32 dx4, rg_dx4[1]; \
1475 vaddhn.u32 g_whole_high, g_block, dx4; \
1476 vdup.u32 dx4, b_dx4; \
1478 vaddhn.u32 b_whole_high, b_block, dx4; \
1479 vdup.u32 dx8, uv_dx8[0]; \
1481 vadd.u32 u_block, u_block, dx8; \
1482 vdup.u32 dx8, uv_dx8[1]; \
1484 vadd.u32 v_block, v_block, dx8; \
1485 vdup.u32 dx8, rg_dx8[0]; \
1487 vadd.u32 r_block, r_block, dx8; \
1488 vdup.u32 dx8, rg_dx8[1]; \
1490 vadd.u32 g_block, g_block, dx8; \
1491 vdup.u32 dx8, b_dx8; \
1493 vadd.u32 b_block, b_block, dx8; \
1494 vmovn.u16 u_whole_8, u_whole; \
1496 vmovn.u16 v_whole_8, v_whole; \
1498 vmovn.u16 b_whole_8, b_whole; \
1500 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1502 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1503 setup_blocks_texture_##swizzling(); \
1505 vmovn.u16 r_whole_8, r_whole; \
1509 vmovn.u16 g_whole_8, g_whole; \
1510 vshrn.u32 u_whole_low, u_block, #16; \
1512 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1513 vshrn.u32 v_whole_low, v_block, #16; \
1515 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1516 vshrn.u32 r_whole_low, r_block, #16; \
1518 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1519 vshrn.u32 g_whole_low, g_block, #16; \
1521 vdup.u32 dx4, uv_dx4[0]; \
1522 vshrn.u32 b_whole_low, b_block, #16; \
1524 vaddhn.u32 u_whole_high, u_block, dx4; \
1525 vdup.u32 dx4, uv_dx4[1]; \
1527 vaddhn.u32 v_whole_high, v_block, dx4; \
1528 vdup.u32 dx4, rg_dx4[0]; \
1530 vaddhn.u32 r_whole_high, r_block, dx4; \
1531 vdup.u32 dx4, rg_dx4[1]; \
1533 vaddhn.u32 g_whole_high, g_block, dx4; \
1534 vdup.u32 dx4, b_dx4; \
1536 vaddhn.u32 b_whole_high, b_block, dx4; \
1537 vdup.u32 dx8, uv_dx8[0]; \
1539 vadd.u32 u_block, u_block, dx8; \
1540 vdup.u32 dx8, uv_dx8[1]; \
1542 vadd.u32 v_block, v_block, dx8; \
1543 vdup.u32 dx8, rg_dx8[0]; \
1545 vadd.u32 r_block, r_block, dx8; \
1546 vdup.u32 dx8, rg_dx8[1]; \
1548 vadd.u32 g_block, g_block, dx8; \
1549 vdup.u32 dx8, b_dx8; \
1551 vadd.u32 b_block, b_block, dx8; \
1552 vmovn.u16 u_whole_8, u_whole; \
1554 add fb_ptr, fb_ptr, #16; \
1555 vmovn.u16 v_whole_8, v_whole; \
1557 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1558 vmovn.u16 b_whole_8, b_whole; \
1562 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1563 subs span_num_blocks, span_num_blocks, #1; \
1565 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1566 setup_blocks_texture_##swizzling(); \
1568 vmovn.u16 r_whole_8, r_whole; \
1572 vmovn.u16 g_whole_8, g_whole; \
1573 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1575 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1576 vdup.u8 draw_mask, right_mask; \
1578 vmov.u32 fb_mask_ptrs[0], right_mask; \
1579 vtst.u16 draw_mask, draw_mask, test_mask; \
1580 vzip.u8 u_whole_8, v_whole_8; \
1582 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1583 vst1.u32 { r_whole_8, g_whole_8 }, [block_ptr_b, :128], c_32; \
1584 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1585 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1586 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1589 add span_uvrg_offset, span_uvrg_offset, #16; \
1590 add span_b_offset, span_b_offset, #4; \
1592 add span_edge_data, span_edge_data, #8; \
1593 subs num_spans, num_spans, #1; \
1595 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1598 restore_abi_regs(); \
1599 ldmia sp!, { r4 - r11, pc }; \
1602 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1603 vpush { texture_mask }; \
1604 vpush { uvrg_dx4 }; \
1606 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; /* r14=num_blocks */ \
1607 bl flush_render_block_buffer; \
1608 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
1610 vpop { uvrg_dx4 }; \
1611 vpop { texture_mask }; \
1613 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1614 vmov.u8 fb_mask_ptrs, #0; \
1616 mov num_blocks, span_num_blocks; \
1617 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1621 setup_blocks_shaded_textured_builder(swizzled)
1622 setup_blocks_shaded_textured_builder(unswizzled)
1625 #define setup_blocks_unshaded_textured_builder(swizzling) \
1628 function(setup_blocks_unshaded_textured_dithered_##swizzling##_indirect) \
1629 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
1630 add uvrg_dx_ptr, psx_gpu, #psx_gpu_uvrg_dx_offset; \
1632 vld1.u32 { uvrg_dx }, [uvrg_dx_ptr, :128]; \
1633 add texture_mask_ptr, psx_gpu, #psx_gpu_texture_mask_width_offset; \
1635 cmp num_spans, #0; \
1638 stmdb sp!, { r4 - r11, r14 }; \
1640 vshl.u32 uvrg_dx4, uvrg_dx, #2; \
1642 vshl.u32 uvrg_dx8, uvrg_dx, #3; \
1644 vld2.u8 { texture_mask_u[], texture_mask_v[] }, [texture_mask_ptr, :16]; \
1645 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
1647 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1648 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
1650 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1652 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
1655 vmov.u8 fb_mask_ptrs, #0; \
1657 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
1658 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
1660 ldrh y, [span_edge_data, #edge_data_y_offset]; \
1661 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
1663 cmp span_num_blocks, #0; \
1666 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
1667 add num_blocks, span_num_blocks, num_blocks; \
1669 cmp num_blocks, #MAX_BLOCKS; \
1673 add fb_ptr, fb_ptr, y, lsl #11; \
1675 vdup.u32 v_left_x, left_x; \
1678 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
1679 add fb_ptr, fb_ptr, left_x, lsl #1; \
1681 and dither_shift, left_x, #0x03; \
1683 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
1684 vshr.u32 uvrg_dx, uvrg_dx4, #2; \
1686 mov dither_shift, dither_shift, lsl #3; \
1687 vmla.u32 uvrg, uvrg_dx, v_left_x; \
1690 subs span_num_blocks, span_num_blocks, #1; \
1692 mov dither_row, dither_row, ror dither_shift; \
1694 vdup.u32 dither_offsets_short, dither_row; \
1695 add block_span_ptr, psx_gpu, #psx_gpu_u_block_span_offset; \
1697 vshll.s8 dither_offsets, dither_offsets_short, #4; \
1699 vdup.u32 u_block, uv[0]; \
1701 vdup.u32 v_block, uv[1]; \
1702 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1704 vadd.u32 u_block, u_block, block_span; \
1705 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
1707 vadd.u32 v_block, v_block, block_span; \
1708 add block_ptr_b, block_ptr_a, #16; \
1710 vshrn.u32 u_whole_low, u_block, #16; \
1711 vshrn.u32 v_whole_low, v_block, #16; \
1713 vdup.u32 dx4, uv_dx4[0]; \
1715 vaddhn.u32 u_whole_high, u_block, dx4; \
1716 vdup.u32 dx4, uv_dx4[1]; \
1718 vaddhn.u32 v_whole_high, v_block, dx4; \
1719 vdup.u32 dx8, uv_dx8[0]; \
1721 vadd.u32 u_block, u_block, dx8; \
1722 vdup.u32 dx8, uv_dx8[1]; \
1724 vadd.u32 v_block, v_block, dx8; \
1725 vmovn.u16 u_whole_8, u_whole; \
1727 vmovn.u16 v_whole_8, v_whole; \
1730 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1732 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1733 setup_blocks_texture_##swizzling(); \
1738 vshrn.u32 u_whole_low, u_block, #16; \
1740 vst2.u8 { u_whole_8, v_whole_8 }, [block_ptr_a, :128], c_32; \
1741 vshrn.u32 v_whole_low, v_block, #16; \
1743 add block_ptr_b, block_ptr_b, #32; \
1744 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1746 vdup.u32 dx4, uv_dx4[0]; \
1747 vaddhn.u32 u_whole_high, u_block, dx4; \
1748 vdup.u32 dx4, uv_dx4[1]; \
1750 vaddhn.u32 v_whole_high, v_block, dx4; \
1751 vdup.u32 dx8, uv_dx8[0]; \
1753 vadd.u32 u_block, u_block, dx8; \
1754 vdup.u32 dx8, uv_dx8[1]; \
1756 vadd.u32 v_block, v_block, dx8; \
1757 vmovn.u16 u_whole_8, u_whole; \
1759 add fb_ptr, fb_ptr, #16; \
1760 vmovn.u16 v_whole_8, v_whole; \
1762 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1765 vmov.u32 fb_mask_ptrs[1], fb_ptr; \
1766 subs span_num_blocks, span_num_blocks, #1; \
1768 vand.u8 uv_whole_8, uv_whole_8, texture_mask; \
1769 setup_blocks_texture_##swizzling(); \
1774 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
1776 vld1.u32 { test_mask }, [psx_gpu, :128]; \
1777 vdup.u8 draw_mask, right_mask; \
1779 vmov.u32 fb_mask_ptrs[0], right_mask; \
1780 vtst.u16 draw_mask, draw_mask, test_mask; \
1781 vzip.u8 u_whole_8, v_whole_8; \
1783 vbic.u16 uv_whole_8, uv_whole_8, draw_mask; \
1784 add block_ptr_b, block_ptr_b, #32; \
1785 vst1.u32 { uv_whole_8 }, [block_ptr_a, :128], c_32; \
1786 vst1.u32 { dither_offsets }, [block_ptr_b, :128], c_32; \
1787 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32; \
1790 add span_uvrg_offset, span_uvrg_offset, #16; \
1791 add span_edge_data, span_edge_data, #8; \
1792 subs num_spans, num_spans, #1; \
1794 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
1797 restore_abi_regs(); \
1798 ldmia sp!, { r4 - r11, pc }; \
1801 /* TODO: Load from psx_gpu instead of saving/restoring these */\
1802 vpush { texture_mask }; \
1803 vpush { uvrg_dx4 }; \
1805 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; /* r14=num_blocks */ \
1806 bl flush_render_block_buffer; \
1807 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12 }; \
1809 vpop { uvrg_dx4 }; \
1810 vpop { texture_mask }; \
1812 vadd.u32 uvrg_dx8, uvrg_dx4, uvrg_dx4; \
1813 vmov.u8 fb_mask_ptrs, #0; \
1815 mov num_blocks, span_num_blocks; \
1816 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
1820 setup_blocks_unshaded_textured_builder(swizzled)
1821 setup_blocks_unshaded_textured_builder(unswizzled)
1826 function(setup_blocks_unshaded_untextured_undithered_unswizzled_indirect)
1827 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1828 veor.u32 draw_mask, draw_mask, draw_mask
1833 stmdb sp!, { r4 - r11, r14 }
1835 vld1.u32 { test_mask }, [psx_gpu, :128]
1837 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1839 ubfx color_r, color, #3, #5
1840 ubfx color_g, color, #11, #5
1841 ubfx color_b, color, #19, #5
1843 orr color, color_r, color_b, lsl #10
1844 orr color, color, color_g, lsl #5
1846 vdup.u16 colors, color
1848 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1849 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1851 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1852 add block_ptr_a, block_ptr_a, num_blocks, lsl #6
1855 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1856 ldrh y, [span_edge_data, #edge_data_y_offset]
1858 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1860 cmp span_num_blocks, #0
1863 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1864 add num_blocks, span_num_blocks, num_blocks
1866 cmp num_blocks, #MAX_BLOCKS
1870 add fb_ptr, fb_ptr, y, lsl #11
1873 add fb_ptr, fb_ptr, left_x, lsl #1
1876 subs span_num_blocks, span_num_blocks, #1
1878 add block_ptr_b, block_ptr_a, #16
1881 vmov.u32 fb_mask_ptrs[1], fb_ptr
1885 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_32
1886 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1887 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1889 add fb_ptr, fb_ptr, #16
1890 add block_ptr_b, block_ptr_b, #32
1894 vmov.u32 fb_mask_ptrs[1], fb_ptr
1895 subs span_num_blocks, span_num_blocks, #1
1900 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]
1902 vdup.u8 draw_mask_edge, right_mask
1903 vtst.u16 draw_mask_edge, draw_mask_edge, test_mask
1905 vst1.u32 { colors }, [block_ptr_b, :128], c_32
1906 vst1.u32 { draw_mask_edge }, [block_ptr_a, :128], c_32
1907 add block_ptr_b, block_ptr_b, #32
1908 vst1.u32 { b_whole_8, fb_mask_ptrs }, [block_ptr_a, :128], c_32
1911 add span_edge_data, span_edge_data, #8
1912 subs num_spans, num_spans, #1
1914 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
1918 ldmia sp!, { r4 - r11, pc }
1923 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1924 bl flush_render_block_buffer
1925 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
1929 vld1.u32 { test_mask }, [psx_gpu, :128]
1930 veor.u32 draw_mask, draw_mask, draw_mask
1932 mov num_blocks, span_num_blocks
1933 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset
1937 #define mask_msb_scalar r14
1939 #define msb_mask q15
1941 #define pixels_low d16
1943 #define msb_mask_low d30
1944 #define msb_mask_high d31
1949 function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
1950 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]
1955 stmdb sp!, { r4 - r11, r14 }
1957 ldr color, [psx_gpu, #psx_gpu_triangle_color_offset]
1959 ubfx color_r, color, #3, #5
1960 ubfx color_g, color, #11, #5
1962 ldrh mask_msb_scalar, [psx_gpu, #psx_gpu_mask_msb_offset]
1963 ubfx color_b, color, #19, #5
1965 orr color, color_r, color_b, lsl #10
1966 orr color, color, color_g, lsl #5
1967 orr color, color, mask_msb_scalar
1969 vdup.u16 colors, color
1971 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset
1972 orr color, color, color, lsl #16
1976 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]
1977 ldrh y, [span_edge_data, #edge_data_y_offset]
1979 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
1981 cmp span_num_blocks, #0
1984 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]
1986 add fb_ptr, fb_ptr, y, lsl #11
1987 subs span_num_blocks, span_num_blocks, #1
1989 add fb_ptr, fb_ptr, left_x, lsl #1
1993 vst1.u32 { colors }, [fb_ptr]!
1994 subs span_num_blocks, span_num_blocks, #1
1999 ldrb right_mask, [span_edge_data, #edge_data_right_mask_offset]
2001 cmp right_mask, #0x0
2004 tst right_mask, #0xF
2005 streq color, [fb_ptr], #4
2006 moveq right_mask, right_mask, lsr #4
2007 streq color, [fb_ptr], #4
2009 tst right_mask, #0x3
2010 streq color, [fb_ptr], #4
2011 moveq right_mask, right_mask, lsr #2
2013 tst right_mask, #0x1
2014 strheq color, [fb_ptr]
2017 add span_edge_data, span_edge_data, #8
2018 subs num_spans, num_spans, #1
2021 ldmia sp!, { r4 - r11, pc }
2024 vst1.u32 { colors }, [fb_ptr]
2031 #define rg_dx_ptr r2
2049 #undef dither_offsets
2069 #define r_whole_low d6
2070 #define r_whole_high d7
2071 #define g_whole_low d8
2072 #define g_whole_high d9
2073 #define b_whole_low d10
2074 #define b_whole_high d11
2076 #define gb_whole_8 q6
2078 #define g_whole_8 d12
2079 #define b_whole_8 d13
2081 #define r_whole_8 d14
2093 #define block_span q5
2101 #define d128_0x7 q13
2105 #define dither_offsets q14
2106 #define draw_mask q15
2108 #define dither_offsets_low d28
2111 #define test_mask q10
2114 #define setup_blocks_shaded_untextured_dither_a_dithered() \
2115 vqadd.u8 r_whole_8, r_whole_8, dither_offsets_low; \
2116 vqadd.u8 gb_whole_8, gb_whole_8, dither_offsets; \
2118 #define setup_blocks_shaded_untextured_dither_b_dithered() \
2119 vqsub.u8 r_whole_8, r_whole_8, d64_4; \
2120 vqsub.u8 gb_whole_8, gb_whole_8, d128_4 \
2122 #define setup_blocks_shaded_untextured_dither_a_undithered() \
2124 #define setup_blocks_shaded_untextured_dither_b_undithered() \
2127 #define setup_blocks_shaded_untextured_indirect_builder(dithering) \
2130 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_indirect) \
2131 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2132 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2134 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2136 cmp num_spans, #0; \
2139 stmdb sp!, { r4 - r11, r14 }; \
2141 vshl.u32 rg_dx4, rg_dx, #2; \
2143 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2144 vshl.u32 rg_dx8, rg_dx, #3; \
2146 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2148 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2149 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2151 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2152 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2154 add block_ptr_a, block_ptr_a, num_blocks, lsl #6; \
2155 vmov.u8 d64_1, #1; \
2157 vmov.u8 d128_4, #4; \
2158 vmov.u8 d64_128, #128; \
2160 vmov.u8 d128_0x7, #0x7; \
2163 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2164 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2166 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2167 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2169 cmp span_num_blocks, #0; \
2172 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2173 add num_blocks, span_num_blocks, num_blocks; \
2175 cmp num_blocks, #MAX_BLOCKS; \
2179 ldr b, [span_b_offset]; \
2180 add fb_ptr, fb_ptr, y, lsl #11; \
2182 vdup.u32 v_left_x, left_x; \
2185 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2186 add fb_ptr, fb_ptr, left_x, lsl #1; \
2188 mla b, b_dx, left_x, b; \
2189 and dither_shift, left_x, #0x03; \
2191 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2192 vshr.u32 rg_dx, rg_dx4, #2; \
2194 mov dither_shift, dither_shift, lsl #3; \
2195 vmla.u32 rg, rg_dx, v_left_x; \
2198 subs span_num_blocks, span_num_blocks, #1; \
2200 mov dither_row, dither_row, ror dither_shift; \
2201 mov b_dx4, b_dx, lsl #2; \
2203 vdup.u32 dither_offsets, dither_row; \
2204 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2206 vdup.u32 b_block, b; \
2207 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2209 mov b_dx8, b_dx, lsl #3; \
2210 vdup.u32 r_block, rg[0]; \
2211 vdup.u32 g_block, rg[1]; \
2213 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2215 vadd.u32 r_block, r_block, block_span; \
2216 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2218 vadd.u32 g_block, g_block, block_span; \
2219 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2221 vadd.u32 b_block, b_block, block_span; \
2222 add block_ptr_b, block_ptr_a, #16; \
2224 vshrn.u32 r_whole_low, r_block, #16; \
2225 vshrn.u32 g_whole_low, g_block, #16; \
2226 vshrn.u32 b_whole_low, b_block, #16; \
2227 vdup.u32 dx4, rg_dx4[0]; \
2229 vaddhn.u32 r_whole_high, r_block, dx4; \
2230 vdup.u32 dx4, rg_dx4[1]; \
2232 vaddhn.u32 g_whole_high, g_block, dx4; \
2233 vdup.u32 dx4, b_dx4; \
2235 vaddhn.u32 b_whole_high, b_block, dx4; \
2236 vdup.u32 dx8, rg_dx8[0]; \
2238 vadd.u32 r_block, r_block, dx8; \
2239 vdup.u32 dx8, rg_dx8[1]; \
2241 vadd.u32 g_block, g_block, dx8; \
2242 vdup.u32 dx8, b_dx8; \
2244 vadd.u32 b_block, b_block, dx8; \
2246 vmovn.u16 r_whole_8, r_whole; \
2247 vmovn.u16 g_whole_8, g_whole; \
2248 vmovn.u16 b_whole_8, b_whole; \
2251 veor.u32 draw_mask, draw_mask, draw_mask; \
2254 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2255 vshrn.u32 r_whole_low, r_block, #16; \
2257 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2258 vshrn.u32 g_whole_low, g_block, #16; \
2260 vshrn.u32 b_whole_low, b_block, #16; \
2261 str fb_ptr, [block_ptr_a, #44]; \
2263 vdup.u32 dx4, rg_dx4[0]; \
2264 vshr.u8 r_whole_8, r_whole_8, #3; \
2265 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2267 vaddhn.u32 r_whole_high, r_block, dx4; \
2268 vdup.u32 dx4, rg_dx4[1]; \
2270 vaddhn.u32 g_whole_high, g_block, dx4; \
2271 vdup.u32 dx4, b_dx4; \
2273 vaddhn.u32 b_whole_high, b_block, dx4; \
2274 vdup.u32 dx8, rg_dx8[0]; \
2276 vmull.u8 pixels, r_whole_8, d64_1; \
2277 vmlal.u8 pixels, g_whole_8, d64_4; \
2278 vmlal.u8 pixels, b_whole_8, d64_128; \
2280 vadd.u32 r_block, r_block, dx8; \
2281 vdup.u32 dx8, rg_dx8[1]; \
2283 vadd.u32 g_block, g_block, dx8; \
2284 vdup.u32 dx8, b_dx8; \
2286 vadd.u32 b_block, b_block, dx8; \
2287 add fb_ptr, fb_ptr, #16; \
2289 vmovn.u16 r_whole_8, r_whole; \
2290 vmovn.u16 g_whole_8, g_whole; \
2291 vmovn.u16 b_whole_8, b_whole; \
2293 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2294 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2298 subs span_num_blocks, span_num_blocks, #1; \
2302 str fb_ptr, [block_ptr_a, #44]; \
2303 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2305 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2306 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2308 vshr.u8 r_whole_8, r_whole_8, #3; \
2309 vdup.u8 draw_mask, right_mask; \
2311 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2312 vld1.u32 { test_mask }, [psx_gpu, :128]; \
2314 vtst.u16 draw_mask, draw_mask, test_mask; \
2316 vmull.u8 pixels, r_whole_8, d64_1; \
2317 vmlal.u8 pixels, g_whole_8, d64_4; \
2318 vmlal.u8 pixels, b_whole_8, d64_128; \
2320 vst1.u32 { draw_mask }, [block_ptr_a, :128], c_64; \
2321 vst1.u32 { pixels }, [block_ptr_b, :128], c_64; \
2324 add span_uvrg_offset, span_uvrg_offset, #16; \
2325 add span_b_offset, span_b_offset, #4; \
2327 add span_edge_data, span_edge_data, #8; \
2328 subs num_spans, num_spans, #1; \
2330 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
2333 restore_abi_regs(); \
2334 ldmia sp!, { r4 - r11, pc }; \
2337 /* TODO: Load from psx_gpu instead of saving/restoring these */\
2340 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2341 bl flush_render_block_buffer; \
2342 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }; \
2346 vmov.u8 d64_1, #1; \
2347 vmov.u8 d128_4, #4; \
2348 vmov.u8 d64_128, #128; \
2349 vmov.u8 d128_0x7, #0x7; \
2351 vadd.u32 rg_dx8, rg_dx4, rg_dx4; \
2353 mov num_blocks, span_num_blocks; \
2354 add block_ptr_a, psx_gpu, #psx_gpu_blocks_offset; \
2358 setup_blocks_shaded_untextured_indirect_builder(undithered)
2359 setup_blocks_shaded_untextured_indirect_builder(dithered)
2364 #define mask_msb_ptr r14
2366 #define draw_mask q0
2367 #define pixels_low d16
2368 #define pixels_high d17
2372 #define setup_blocks_shaded_untextured_direct_builder(dithering) \
2375 function(setup_blocks_shaded_untextured_##dithering##_unswizzled_direct) \
2376 ldrh num_spans, [psx_gpu, #psx_gpu_num_spans_offset]; \
2377 add rg_dx_ptr, psx_gpu, #(psx_gpu_uvrg_dx_offset + 8); \
2379 vld1.u32 { rg_dx }, [rg_dx_ptr, :64]; \
2381 cmp num_spans, #0; \
2384 stmdb sp!, { r4 - r11, r14 }; \
2386 vshl.u32 rg_dx4, rg_dx, #2; \
2388 ldr b_dx, [psx_gpu, #psx_gpu_b_dx_offset]; \
2389 vshl.u32 rg_dx8, rg_dx, #3; \
2391 add span_uvrg_offset, psx_gpu, #psx_gpu_span_uvrg_offset_offset; \
2392 add span_edge_data, psx_gpu, #psx_gpu_span_edge_data_offset; \
2394 add span_b_offset, psx_gpu, #psx_gpu_span_b_offset_offset; \
2395 vmov.u8 d64_1, #1; \
2397 vmov.u8 d128_4, #4; \
2398 vmov.u8 d64_128, #128; \
2400 vmov.u8 d128_0x7, #0x7; \
2401 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
2402 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
2405 ldrh span_num_blocks, [span_edge_data, #edge_data_num_blocks_offset]; \
2406 add dither_offset_ptr, psx_gpu, #psx_gpu_dither_table_offset; \
2408 ldrh y, [span_edge_data, #edge_data_y_offset]; \
2409 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
2411 cmp span_num_blocks, #0; \
2414 ldrh left_x, [span_edge_data, #edge_data_left_x_offset]; \
2415 add fb_ptr, fb_ptr, y, lsl #11; \
2417 ldr b, [span_b_offset]; \
2418 vdup.u32 v_left_x, left_x; \
2421 ldr dither_row, [dither_offset_ptr, y, lsl #2]; \
2422 add fb_ptr, fb_ptr, left_x, lsl #1; \
2424 mla b, b_dx, left_x, b; \
2425 and dither_shift, left_x, #0x03; \
2427 vld1.u32 { uvrg }, [span_uvrg_offset, :128]; \
2428 vshr.u32 rg_dx, rg_dx4, #2; \
2430 mov dither_shift, dither_shift, lsl #3; \
2431 vmla.u32 rg, rg_dx, v_left_x; \
2433 subs span_num_blocks, span_num_blocks, #1; \
2435 mov dither_row, dither_row, ror dither_shift; \
2436 mov b_dx4, b_dx, lsl #2; \
2438 vdup.u32 dither_offsets, dither_row; \
2439 add block_span_ptr, psx_gpu, #psx_gpu_r_block_span_offset; \
2441 vdup.u32 b_block, b; \
2442 vadd.u8 dither_offsets, dither_offsets, d128_4; \
2444 mov b_dx8, b_dx, lsl #3; \
2445 vdup.u32 r_block, rg[0]; \
2446 vdup.u32 g_block, rg[1]; \
2448 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2450 vadd.u32 r_block, r_block, block_span; \
2451 vld1.u32 { block_span }, [block_span_ptr, :128]!; \
2453 vadd.u32 g_block, g_block, block_span; \
2454 vld1.u32 { block_span }, [block_span_ptr, :128]; \
2456 vadd.u32 b_block, b_block, block_span; \
2457 add block_ptr_b, block_ptr_a, #16; \
2459 vshrn.u32 r_whole_low, r_block, #16; \
2460 vshrn.u32 g_whole_low, g_block, #16; \
2461 vshrn.u32 b_whole_low, b_block, #16; \
2462 vdup.u32 dx4, rg_dx4[0]; \
2464 vaddhn.u32 r_whole_high, r_block, dx4; \
2465 vdup.u32 dx4, rg_dx4[1]; \
2467 vaddhn.u32 g_whole_high, g_block, dx4; \
2468 vdup.u32 dx4, b_dx4; \
2470 vaddhn.u32 b_whole_high, b_block, dx4; \
2471 vdup.u32 dx8, rg_dx8[0]; \
2473 vadd.u32 r_block, r_block, dx8; \
2474 vdup.u32 dx8, rg_dx8[1]; \
2476 vadd.u32 g_block, g_block, dx8; \
2477 vdup.u32 dx8, b_dx8; \
2479 vadd.u32 b_block, b_block, dx8; \
2481 vmovn.u16 r_whole_8, r_whole; \
2482 vmovn.u16 g_whole_8, g_whole; \
2483 vmovn.u16 b_whole_8, b_whole; \
2488 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2489 vshrn.u32 r_whole_low, r_block, #16; \
2491 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2492 vshrn.u32 g_whole_low, g_block, #16; \
2494 vshrn.u32 b_whole_low, b_block, #16; \
2496 vdup.u32 dx4, rg_dx4[0]; \
2497 vshr.u8 r_whole_8, r_whole_8, #3; \
2498 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2500 vaddhn.u32 r_whole_high, r_block, dx4; \
2501 vdup.u32 dx4, rg_dx4[1]; \
2503 vmov pixels, msb_mask; \
2504 vaddhn.u32 g_whole_high, g_block, dx4; \
2505 vdup.u32 dx4, b_dx4; \
2507 vaddhn.u32 b_whole_high, b_block, dx4; \
2508 vdup.u32 dx8, rg_dx8[0]; \
2510 vmlal.u8 pixels, r_whole_8, d64_1; \
2511 vmlal.u8 pixels, g_whole_8, d64_4; \
2512 vmlal.u8 pixels, b_whole_8, d64_128; \
2514 vadd.u32 r_block, r_block, dx8; \
2515 vdup.u32 dx8, rg_dx8[1]; \
2517 vadd.u32 g_block, g_block, dx8; \
2518 vdup.u32 dx8, b_dx8; \
2520 vadd.u32 b_block, b_block, dx8; \
2522 vmovn.u16 r_whole_8, r_whole; \
2523 vmovn.u16 g_whole_8, g_whole; \
2524 vmovn.u16 b_whole_8, b_whole; \
2526 vst1.u32 { pixels }, [fb_ptr]!; \
2527 subs span_num_blocks, span_num_blocks, #1; \
2531 setup_blocks_shaded_untextured_dither_a_##dithering(); \
2533 ldrh right_mask, [span_edge_data, #edge_data_right_mask_offset]; \
2534 setup_blocks_shaded_untextured_dither_b_##dithering(); \
2536 vshr.u8 r_whole_8, r_whole_8, #3; \
2537 rbit right_mask, right_mask; \
2538 vmov pixels, msb_mask; \
2539 vbic.u8 gb_whole_8, gb_whole_8, d128_0x7; \
2540 clz right_mask, right_mask; \
2542 vmlal.u8 pixels, r_whole_8, d64_1; \
2543 vmlal.u8 pixels, g_whole_8, d64_4; \
2544 vmlal.u8 pixels, b_whole_8, d64_128; \
2546 JT_OP_REL(100f, right_mask, temp); \
2547 JT_OP(ldr pc, [pc, right_mask, lsl #2]); \
2551 .word JTE(100b, 4f); \
2552 .word JTE(100b, 5f); \
2553 .word JTE(100b, 6f); \
2554 .word JTE(100b, 7f); \
2555 .word JTE(100b, 8f); \
2556 .word JTE(100b, 9f); \
2557 .word JTE(100b, 10f); \
2558 .word JTE(100b, 11f); \
2561 vst1.u16 { pixels_low[0] }, [fb_ptr]; \
2565 vst1.u32 { pixels_low[0] }, [fb_ptr]; \
2569 vst1.u32 { pixels_low[0] }, [fb_ptr]!; \
2570 vst1.u16 { pixels_low[2] }, [fb_ptr]; \
2574 vst1.u32 { pixels_low }, [fb_ptr]; \
2578 vst1.u32 { pixels_low }, [fb_ptr]!; \
2579 vst1.u16 { pixels_high[0] }, [fb_ptr]; \
2583 vst1.u32 { pixels_low }, [fb_ptr]!; \
2584 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2588 vst1.u32 { pixels_low }, [fb_ptr]!; \
2589 vst1.u32 { pixels_high[0] }, [fb_ptr]!; \
2590 vst1.u16 { pixels_high[2] }, [fb_ptr]; \
2594 vst1.u32 { pixels }, [fb_ptr]; \
2598 add span_uvrg_offset, span_uvrg_offset, #16; \
2599 add span_b_offset, span_b_offset, #4; \
2601 add span_edge_data, span_edge_data, #8; \
2602 subs num_spans, num_spans, #1; \
2606 restore_abi_regs(); \
2607 ldmia sp!, { r4 - r11, pc } \
2609 setup_blocks_shaded_untextured_direct_builder(undithered)
2610 setup_blocks_shaded_untextured_direct_builder(dithered)
2619 #define block_ptr r1
2620 #define num_blocks r2
2633 #define texture_ptr r11
2647 #define pixels_d r10
2651 #define clut_ptr r12
2652 #define current_texture_mask r5
2653 #define dirty_textures_mask r6
2657 #define clut_low_a d2
2658 #define clut_low_b d3
2659 #define clut_high_a d4
2660 #define clut_high_b d5
2665 #define texels_low d6
2666 #define texels_high d7
2670 function(texture_blocks_untextured)
2676 function(texture_blocks_4bpp)
2677 stmdb sp!, { r3 - r11, r14 }
2678 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2680 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2681 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2683 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2684 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]
2686 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2687 vuzp.u8 clut_a, clut_b
2689 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
2690 tst dirty_textures_mask, current_texture_mask
2696 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2698 uxtah uv_0, texture_ptr, uv_01
2699 uxtah uv_1, texture_ptr, uv_01, ror #16
2701 uxtah uv_2, texture_ptr, uv_23
2702 uxtah uv_3, texture_ptr, uv_23, ror #16
2704 uxtah uv_4, texture_ptr, uv_45
2705 ldrb pixel_0, [uv_0]
2707 uxtah uv_5, texture_ptr, uv_45, ror #16
2708 ldrb pixel_1, [uv_1]
2710 uxtah uv_6, texture_ptr, uv_67
2711 ldrb pixel_2, [uv_2]
2713 uxtah uv_7, texture_ptr, uv_67, ror #16
2714 ldrb pixel_3, [uv_3]
2716 ldrb pixel_4, [uv_4]
2717 subs num_blocks, num_blocks, #1
2719 ldrb pixel_5, [uv_5]
2720 orr pixels_a, pixel_0, pixel_1, lsl #8
2722 ldrb pixel_6, [uv_6]
2723 orr pixels_b, pixel_4, pixel_5, lsl #8
2725 ldrb pixel_7, [uv_7]
2726 orr pixels_a, pixels_a, pixel_2, lsl #16
2728 orr pixels_b, pixels_b, pixel_6, lsl #16
2729 orr pixels_a, pixels_a, pixel_3, lsl #24
2731 orr pixels_b, pixels_b, pixel_7, lsl #24
2732 vmov texels, pixels_a, pixels_b
2734 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels
2735 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels
2737 vst2.u8 { texels_low, texels_high }, [block_ptr, :128], c_64
2740 ldmia sp!, { r3 - r11, pc }
2743 stmdb sp!, { r1 - r2 }
2744 bl update_texture_4bpp_cache
2747 ldmia sp!, { r1 - r2 }
2753 function(texture_blocks_8bpp)
2754 stmdb sp!, { r3 - r11, r14 }
2755 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2757 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2758 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2760 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
2761 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
2763 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]
2764 tst dirty_textures_mask, current_texture_mask
2770 ldm block_ptr, { uv_01, uv_23, uv_45, uv_67 }
2772 uxtah uv_0, texture_ptr, uv_01
2773 uxtah uv_1, texture_ptr, uv_01, ror #16
2775 uxtah uv_2, texture_ptr, uv_23
2776 uxtah uv_3, texture_ptr, uv_23, ror #16
2778 uxtah uv_4, texture_ptr, uv_45
2779 ldrb pixel_0, [uv_0]
2781 uxtah uv_5, texture_ptr, uv_45, ror #16
2782 ldrb pixel_1, [uv_1]
2784 uxtah uv_6, texture_ptr, uv_67
2785 ldrb pixel_2, [uv_2]
2787 uxtah uv_7, texture_ptr, uv_67, ror #16
2788 ldrb pixel_3, [uv_3]
2790 ldrb pixel_4, [uv_4]
2791 add pixel_0, pixel_0, pixel_0
2793 ldrb pixel_5, [uv_5]
2794 add pixel_1, pixel_1, pixel_1
2796 ldrb pixel_6, [uv_6]
2797 add pixel_2, pixel_2, pixel_2
2799 ldrb pixel_7, [uv_7]
2800 add pixel_3, pixel_3, pixel_3
2802 ldrh pixel_0, [clut_ptr, pixel_0]
2803 add pixel_4, pixel_4, pixel_4
2805 ldrh pixel_1, [clut_ptr, pixel_1]
2806 add pixel_5, pixel_5, pixel_5
2808 ldrh pixel_2, [clut_ptr, pixel_2]
2809 add pixel_6, pixel_6, pixel_6
2811 ldrh pixel_3, [clut_ptr, pixel_3]
2812 add pixel_7, pixel_7, pixel_7
2814 ldrh pixel_4, [clut_ptr, pixel_4]
2815 orr pixels_a, pixel_0, pixel_1, lsl #16
2817 ldrh pixel_5, [clut_ptr, pixel_5]
2818 orr pixels_c, pixel_2, pixel_3, lsl #16
2820 ldrh pixel_6, [clut_ptr, pixel_6]
2821 subs num_blocks, num_blocks, #1
2823 ldrh pixel_7, [clut_ptr, pixel_7]
2824 orr pixels_b, pixel_4, pixel_5, lsl #16
2826 orr pixels_d, pixel_6, pixel_7, lsl #16
2827 stm block_ptr, { pixels_a, pixels_c, pixels_b, pixels_d }
2829 add block_ptr, block_ptr, #64
2832 ldmia sp!, { r3 - r11, pc }
2835 stmdb sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2837 bl update_texture_8bpp_cache
2839 ldmia sp!, { r1 - r2, EXTRA_UNSAVED_REGS r12 }
2869 #define block_ptr r1
2870 #define num_blocks r2
2914 #define texture_ptr r12
2919 function(texture_blocks_16bpp)
2920 stmdb sp!, { r3 - r11, r14 }
2921 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
2923 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
2924 ldr texture_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
2927 ldrh uv_0, [block_ptr]
2928 subs num_blocks, num_blocks, #1
2930 ldrh uv_1, [block_ptr, #2]
2932 and v_0, uv_0, #0xFF00
2933 and v_1, uv_1, #0xFF00
2935 and u_0, uv_0, #0xFF
2936 and u_1, uv_1, #0xFF
2938 add uv_0, u_0, v_0, lsl #2
2939 ldrh uv_2, [block_ptr, #4]
2941 add uv_1, u_1, v_1, lsl #2
2942 ldrh uv_3, [block_ptr, #6]
2944 add uv_0, uv_0, uv_0
2945 add uv_1, uv_1, uv_1
2947 and v_2, uv_2, #0xFF00
2948 and v_3, uv_3, #0xFF00
2950 and u_2, uv_2, #0xFF
2951 and u_3, uv_3, #0xFF
2953 add uv_2, u_2, v_2, lsl #2
2954 ldrh uv_4, [block_ptr, #8]
2956 add uv_3, u_3, v_3, lsl #2
2957 ldrh uv_5, [block_ptr, #10]
2959 add uv_2, uv_2, uv_2
2960 add uv_3, uv_3, uv_3
2962 and v_4, uv_4, #0xFF00
2963 and v_5, uv_5, #0xFF00
2965 and u_4, uv_4, #0xFF
2966 and u_5, uv_5, #0xFF
2968 add uv_4, u_4, v_4, lsl #2
2969 ldrh uv_6, [block_ptr, #12]
2971 add uv_5, u_5, v_5, lsl #2
2972 ldrh uv_7, [block_ptr, #14]
2974 add uv_4, uv_4, uv_4
2975 ldrh pixel_0, [texture_ptr, uv_0]
2977 add uv_5, uv_5, uv_5
2978 ldrh pixel_1, [texture_ptr, uv_1]
2980 and v_6, uv_6, #0xFF00
2981 ldrh pixel_2, [texture_ptr, uv_2]
2983 and v_7, uv_7, #0xFF00
2984 ldrh pixel_3, [texture_ptr, uv_3]
2986 and u_6, uv_6, #0xFF
2987 ldrh pixel_4, [texture_ptr, uv_4]
2989 and u_7, uv_7, #0xFF
2990 ldrh pixel_5, [texture_ptr, uv_5]
2992 add uv_6, u_6, v_6, lsl #2
2993 add uv_7, u_7, v_7, lsl #2
2995 add uv_6, uv_6, uv_6
2996 add uv_7, uv_7, uv_7
2998 orr pixels_a, pixel_0, pixel_1, lsl #16
2999 orr pixels_b, pixel_2, pixel_3, lsl #16
3001 ldrh pixel_6, [texture_ptr, uv_6]
3002 orr pixels_c, pixel_4, pixel_5, lsl #16
3004 ldrh pixel_7, [texture_ptr, uv_7]
3005 orr pixels_d, pixel_6, pixel_7, lsl #16
3007 stm block_ptr, { pixels_a, pixels_b, pixels_c, pixels_d }
3008 add block_ptr, block_ptr, #64
3012 ldmia sp!, { r3 - r11, pc }
3027 #undef msb_mask_high
3035 #define num_blocks r1
3036 #define color_ptr r2
3037 #define colors_scalar r2
3038 #define colors_scalar_compare r3
3039 #define mask_msb_ptr r2
3041 #define block_ptr_load_a r0
3042 #define block_ptr_store r3
3043 #define block_ptr_load_b r12
3048 #define draw_mask_bits_scalar r5
3050 #define d128_0x07 q0
3051 #define d128_0x1F q1
3052 #define d128_0x8000 q2
3053 #define test_mask q3
3055 #define colors_rg q5
3056 #define colors_b_dm_bits q6
3057 #define texels_rg q7
3060 #define pixels_b q10
3062 #define zero_mask q4
3063 #define draw_mask q12
3064 #define msb_mask q13
3066 #define fb_pixels q8
3068 #define pixels_gb_low q9
3070 #define colors_r d10
3071 #define colors_g d11
3072 #define colors_b d12
3073 #define draw_mask_bits d13
3074 #define texels_r d14
3075 #define texels_g d15
3076 #define pixels_r_low d16
3077 #define pixels_g_low d18
3078 #define pixels_b_low d19
3079 #define msb_mask_low d26
3080 #define msb_mask_high d27
3085 #define texels_b d31
3087 #define shade_blocks_textured_modulated_prologue_indirect() \
3089 add block_ptr_store, psx_gpu, #psx_gpu_blocks_offset \
3091 #define shade_blocks_textured_modulated_prologue_direct() \
3092 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3093 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16] \
3096 #define shade_blocks_textured_modulated_prologue_shaded(dithering, target) \
3098 #define shade_blocks_textured_false_modulation_check_undithered(target) \
3099 ldr colors_scalar, [psx_gpu, #psx_gpu_triangle_color_offset]; \
3100 movw colors_scalar_compare, #0x8080; \
3102 movt colors_scalar_compare, #0x80; \
3103 cmp colors_scalar, colors_scalar_compare; \
3104 beq shade_blocks_textured_unmodulated_##target \
3106 #define shade_blocks_textured_false_modulation_check_dithered(target) \
3108 #define shade_blocks_textured_modulated_prologue_unshaded(dithering, target) \
3109 shade_blocks_textured_false_modulation_check_##dithering(target); \
3110 add color_ptr, psx_gpu, #psx_gpu_triangle_color_offset; \
3111 vld1.u32 { colors_r[] }, [color_ptr, :32]; \
3112 vdup.u8 colors_g, colors_r[1]; \
3113 vdup.u8 colors_b, colors_r[2]; \
3114 vdup.u8 colors_r, colors_r[0] \
3117 #define shade_blocks_textured_modulated_load_dithered(target) \
3118 vld1.u32 { target }, [block_ptr_load_b, :128] \
3120 #define shade_blocks_textured_modulated_load_last_dithered(target) \
3121 vld1.u32 { target }, [block_ptr_load_b, :128], c_32 \
3123 #define shade_blocks_textured_modulated_load_undithered(target) \
3125 #define shade_blocks_textured_modulated_load_last_undithered(target) \
3126 add block_ptr_load_b, block_ptr_load_b, #32 \
3128 #define shade_blocks_textured_modulate_dithered(channel) \
3129 vmlal.u8 pixels_##channel, texels_##channel, colors_##channel \
3131 #define shade_blocks_textured_modulate_undithered(channel) \
3132 vmull.u8 pixels_##channel, texels_##channel, colors_##channel \
3135 #define shade_blocks_textured_modulated_store_draw_mask_indirect(offset) \
3136 vst1.u32 { draw_mask }, [block_ptr_store, :128]! \
3138 #define shade_blocks_textured_modulated_store_draw_mask_direct(offset) \
3139 ldr fb_ptr, [block_ptr_load_b, #(offset - 64)]; \
3140 vld1.u32 { fb_pixels }, [fb_ptr]; \
3141 vbit.u16 pixels, fb_pixels, draw_mask \
3143 #define shade_blocks_textured_modulated_store_pixels_indirect() \
3144 vst1.u32 { pixels }, [block_ptr_store, :128], c_48 \
3146 #define shade_blocks_textured_modulated_store_pixels_direct() \
3147 vst1.u32 { pixels }, [fb_ptr] \
3150 #define shade_blocks_textured_modulated_load_rg_shaded() \
3151 vld1.u32 { colors_r, colors_g }, [block_ptr_load_b, :128], c_32 \
3153 #define shade_blocks_textured_modulated_load_rg_unshaded() \
3154 add block_ptr_load_b, block_ptr_load_b, #32 \
3156 #define shade_blocks_textured_modulated_load_bdm_shaded() \
3157 vld1.u32 { colors_b, draw_mask_bits }, [block_ptr_load_a, :128], c_32 \
3159 #define shade_blocks_textured_modulated_load_bdm_unshaded() \
3160 ldr draw_mask_bits_scalar, [block_ptr_load_a, #8]; \
3161 add block_ptr_load_a, block_ptr_load_a, #32 \
3163 #define shade_blocks_textured_modulated_expand_draw_mask_shaded() \
3164 vdup.u16 draw_mask, draw_mask_bits[0] \
3166 #define shade_blocks_textured_modulated_expand_draw_mask_unshaded() \
3167 vdup.u16 draw_mask, draw_mask_bits_scalar \
3170 #define shade_blocks_textured_modulated_apply_msb_mask_indirect() \
3172 #define shade_blocks_textured_modulated_apply_msb_mask_direct() \
3173 vorr.u16 pixels, pixels, msb_mask \
3176 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
3179 function(shade_blocks_##shading##_textured_modulated_##dithering##_##target) \
3181 shade_blocks_textured_modulated_prologue_##shading(dithering, target); \
3182 stmdb sp!, { r4 - r5, lr }; \
3183 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3185 vld1.u32 { test_mask }, [psx_gpu, :128]; \
3187 shade_blocks_textured_modulated_prologue_##target(); \
3189 add block_ptr_load_a, psx_gpu, #psx_gpu_blocks_offset; \
3192 add block_ptr_load_b, block_ptr_load_a, #16; \
3193 vmov.u8 d64_1, #1; \
3194 vmov.u8 d64_4, #4; \
3195 vmov.u8 d64_128, #128; \
3197 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3198 vmov.u8 d128_0x07, #0x07; \
3200 shade_blocks_textured_modulated_load_rg_##shading(); \
3201 vmov.u8 d128_0x1F, #0x1F; \
3203 shade_blocks_textured_modulated_load_bdm_##shading(); \
3204 vmov.u16 d128_0x8000, #0x8000; \
3206 vmovn.u16 texels_r, texels; \
3207 vshrn.u16 texels_g, texels, #5; \
3209 vshrn.u16 texels_b, texels, #7; \
3210 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3212 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3213 vtst.u16 draw_mask, draw_mask, test_mask; \
3215 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3216 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3218 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3219 vshr.u8 texels_b, texels_b, #3; \
3221 shade_blocks_textured_modulate_##dithering(r); \
3222 shade_blocks_textured_modulate_##dithering(g); \
3223 shade_blocks_textured_modulate_##dithering(b); \
3225 vand.u16 pixels, texels, d128_0x8000; \
3226 vceq.u16 zero_mask, texels, #0; \
3228 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3229 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3230 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3232 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3233 vorr.u16 draw_mask, draw_mask, zero_mask; \
3234 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3235 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3237 subs num_blocks, num_blocks, #1; \
3243 vld1.u32 { texels }, [block_ptr_load_a, :128], c_32; \
3244 shade_blocks_textured_modulated_load_rg_##shading(); \
3245 vshrn.u16 texels_g, texels, #5; \
3247 shade_blocks_textured_modulated_load_bdm_##shading(); \
3248 vshrn.u16 texels_b, texels, #7; \
3250 pld [block_ptr_load_a]; \
3251 vmovn.u16 texels_r, texels; \
3252 vmlal.u8 pixels, pixels_r_low, d64_1; \
3254 vmlal.u8 pixels, pixels_g_low, d64_4; \
3255 vmlal.u8 pixels, pixels_b_low, d64_128; \
3256 shade_blocks_textured_modulated_store_draw_mask_##target(-4); \
3258 shade_blocks_textured_modulated_load_##dithering(pixels_r); \
3259 shade_blocks_textured_modulated_expand_draw_mask_##shading(); \
3261 shade_blocks_textured_modulated_load_##dithering(pixels_g); \
3262 vand.u8 texels_rg, texels_rg, d128_0x1F; \
3264 shade_blocks_textured_modulated_load_last_##dithering(pixels_b); \
3265 vtst.u16 draw_mask, draw_mask, test_mask; \
3267 shade_blocks_textured_modulated_store_pixels_##target(); \
3268 vshr.u8 texels_b, texels_b, #3; \
3270 shade_blocks_textured_modulate_##dithering(r); \
3271 shade_blocks_textured_modulate_##dithering(g); \
3272 shade_blocks_textured_modulate_##dithering(b); \
3274 vand.u16 pixels, texels, d128_0x8000; \
3275 vceq.u16 zero_mask, texels, #0; \
3277 subs num_blocks, num_blocks, #1; \
3279 vqshrun.s16 pixels_r_low, pixels_r, #4; \
3280 vqshrun.s16 pixels_g_low, pixels_g, #4; \
3281 vqshrun.s16 pixels_b_low, pixels_b, #4; \
3283 shade_blocks_textured_modulated_apply_msb_mask_##target(); \
3284 vorr.u16 draw_mask, draw_mask, zero_mask; \
3285 vshr.u8 pixels_r_low, pixels_r_low, #3; \
3286 vbic.u8 pixels_gb_low, pixels_gb_low, d128_0x07; \
3291 vmlal.u8 pixels, pixels_r_low, d64_1; \
3292 vmlal.u8 pixels, pixels_g_low, d64_4; \
3293 vmlal.u8 pixels, pixels_b_low, d64_128; \
3295 shade_blocks_textured_modulated_store_draw_mask_##target(28); \
3296 shade_blocks_textured_modulated_store_pixels_##target(); \
3298 ldmia sp!, { r4 - r5, lr }; \
3299 restore_abi_regs(); \
3303 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
3304 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
3305 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
3306 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
3308 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
3309 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
3310 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
3311 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
3329 #undef msb_mask_high
3332 #define num_blocks r1
3333 #define mask_msb_ptr r2
3334 #define color_ptr r3
3336 #define block_ptr_load r0
3337 #define draw_mask_store_ptr r3
3338 #define draw_mask_bits_ptr r12
3339 #define draw_mask_ptr r12
3340 #define pixel_store_ptr r14
3342 #define fb_ptr_cmp r4
3345 #define fb_ptr_next r14
3349 #define test_mask q0
3351 #define draw_mask q2
3352 #define zero_mask q3
3353 #define draw_mask_combined q4
3354 #define fb_pixels q5
3355 #define fb_pixels_next q6
3358 #define draw_mask_low d4
3359 #define draw_mask_high d5
3360 #define msb_mask_low d14
3361 #define msb_mask_high d15
3364 function(shade_blocks_textured_unmodulated_indirect)
3365 stmdb sp!, { r4, r14 }
3367 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3369 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3370 add pixel_store_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3372 vld1.u32 { test_mask }, [psx_gpu, :128]
3373 add draw_mask_store_ptr, psx_gpu, #psx_gpu_blocks_offset
3376 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3378 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3379 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3380 [draw_mask_bits_ptr, :16], c_64
3381 vceq.u16 zero_mask, pixels, #0
3383 vtst.u16 draw_mask, draw_mask, test_mask
3384 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3386 subs num_blocks, num_blocks, #1
3390 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3391 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3393 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3394 [draw_mask_bits_ptr, :16], c_64
3395 vceq.u16 zero_mask, pixels, #0
3397 vtst.u16 draw_mask, draw_mask, test_mask
3398 vst1.u32 { pixels }, [pixel_store_ptr, :128], c_64
3400 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3401 subs num_blocks, num_blocks, #1
3406 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3407 vst1.u32 { draw_mask_combined }, [draw_mask_store_ptr, :128], c_64
3410 ldmia sp!, { r4, pc }
3415 function(shade_blocks_textured_unmodulated_direct)
3416 stmdb sp!, { r4, r14 }
3418 add draw_mask_bits_ptr, psx_gpu, #(psx_gpu_blocks_offset + 40)
3420 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3421 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3423 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3426 vld1.u32 { test_mask }, [psx_gpu, :128]
3427 add block_ptr_load, psx_gpu, #psx_gpu_blocks_offset
3429 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3430 [draw_mask_bits_ptr, :16], c_64
3431 ldr fb_ptr_next, [block_ptr_load, #44]
3433 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3434 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3435 vceq.u16 zero_mask, pixels, #0
3436 vtst.u16 draw_mask, draw_mask, test_mask
3438 subs num_blocks, num_blocks, #1
3442 mov fb_ptr, fb_ptr_next
3443 ldr fb_ptr_next, [block_ptr_load, #44]
3445 vorr.u16 pixels, pixels, msb_mask
3447 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3448 vmov fb_pixels, fb_pixels_next
3450 vld1.u16 { draw_mask_low[], draw_mask_high[] }, \
3451 [draw_mask_bits_ptr, :16], c_64
3452 vbif.u16 fb_pixels, pixels, draw_mask_combined
3454 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3455 pld [fb_ptr_next, #64]
3457 add fb_ptr_cmp, fb_ptr_cmp, #14
3458 vld1.u32 { pixels }, [block_ptr_load, :128], c_64
3463 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3464 vceq.u16 zero_mask, pixels, #0
3466 vst1.u16 { fb_pixels }, [fb_ptr]
3467 vtst.u16 draw_mask, draw_mask, test_mask
3470 subs num_blocks, num_blocks, #1
3474 vorr.u16 draw_mask_combined, draw_mask, zero_mask
3475 vbif.u16 fb_pixels_next, pixels, draw_mask_combined
3477 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3480 ldmia sp!, { r4, pc }
3483 vst1.u16 { fb_pixels }, [fb_ptr]
3484 vceq.u16 zero_mask, pixels, #0
3486 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3487 vtst.u16 draw_mask, draw_mask, test_mask
3492 function(shade_blocks_unshaded_untextured_indirect)
3497 function(shade_blocks_unshaded_untextured_direct)
3498 stmdb sp!, { r4, r14 }
3500 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
3502 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
3503 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
3505 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
3506 add color_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
3508 add block_ptr_load, psx_gpu, #(psx_gpu_blocks_offset + 44)
3509 vld1.u16 { pixels }, [color_ptr, :128]
3512 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3514 vorr.u16 pixels, pixels, msb_mask
3515 subs num_blocks, num_blocks, #1
3517 ldr fb_ptr_next, [block_ptr_load], #64
3519 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3523 vmov fb_pixels, fb_pixels_next
3524 mov fb_ptr, fb_ptr_next
3525 ldr fb_ptr_next, [block_ptr_load], #64
3527 vbif.u16 fb_pixels, pixels, draw_mask
3528 vld1.u16 { draw_mask }, [draw_mask_ptr, :128], c_64
3530 sub fb_ptr_cmp, fb_ptr_next, fb_ptr
3531 add fb_ptr_cmp, fb_ptr_cmp, #14
3535 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3536 vst1.u16 { fb_pixels }, [fb_ptr]
3539 subs num_blocks, num_blocks, #1
3543 vbif.u16 fb_pixels_next, pixels, draw_mask
3544 vst1.u16 { fb_pixels_next }, [fb_ptr_next]
3547 ldmia sp!, { r4, pc }
3550 vst1.u16 { fb_pixels }, [fb_ptr]
3551 vld1.u16 { fb_pixels_next }, [fb_ptr_next]
3555 #undef draw_mask_ptr
3562 #define num_blocks r1
3563 #define msb_mask_ptr r2
3564 #define pixel_ptr r3
3565 #define draw_mask_ptr r0
3568 #define fb_ptr_next r14
3569 #define fb_ptr_cmp r4
3577 #undef msb_mask_high
3578 #undef draw_mask_next
3581 #undef fb_pixels_next
3584 #define draw_mask q1
3586 #define fb_pixels q3
3587 #define blend_pixels q4
3588 #define pixels_no_msb q5
3589 #define blend_mask q6
3590 #define fb_pixels_no_msb q7
3591 #define d128_0x8000 q8
3592 #define d128_0x0421 q9
3593 #define fb_pixels_next q10
3594 #define blend_pixels_next q11
3595 #define pixels_next q12
3596 #define draw_mask_next q13
3597 #define write_mask q14
3599 #define pixels_rb q5
3600 #define pixels_mg q7
3602 #define d128_0x7C1F q8
3603 #define d128_0x03E0 q9
3604 #define fb_pixels_rb q10
3605 #define fb_pixels_g q11
3606 #define fb_pixels_masked q11
3607 #define d128_0x83E0 q15
3608 #define pixels_fourth q7
3609 #define d128_0x1C07 q12
3610 #define d128_0x00E0 q13
3611 #define d128_0x80E0 q13
3613 #define msb_mask_low d0
3614 #define msb_mask_high d1
3616 #define blend_blocks_average_set_blend_mask_textured(source) \
3617 vclt.s16 blend_mask, source, #0 \
3619 #define blend_blocks_average_set_stp_bit_textured() \
3620 vorr.u16 blend_pixels, #0x8000 \
3622 #define blend_blocks_average_combine_textured(source) \
3623 vbif.u16 blend_pixels, source, blend_mask \
3625 #define blend_blocks_average_set_blend_mask_untextured(source) \
3627 #define blend_blocks_average_set_stp_bit_untextured() \
3629 #define blend_blocks_average_combine_untextured(source) \
3631 #define blend_blocks_average_mask_set_on() \
3632 vclt.s16 write_mask, fb_pixels_next, #0 \
3634 #define blend_blocks_average_mask_copy_on() \
3635 vorr.u16 draw_mask, draw_mask_next, write_mask \
3637 #define blend_blocks_average_mask_copy_b_on() \
3638 vorr.u16 draw_mask_next, draw_mask_next, write_mask \
3640 #define blend_blocks_average_mask_set_off() \
3642 #define blend_blocks_average_mask_copy_off() \
3643 vmov draw_mask, draw_mask_next \
3645 #define blend_blocks_average_mask_copy_b_off() \
3647 #define blend_blocks_average_builder(texturing, mask_evaluate) \
3650 function(blend_blocks_##texturing##_average_##mask_evaluate) \
3651 stmdb sp!, { r4, r14 }; \
3653 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3654 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3656 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3657 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3659 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3662 vmov.u16 d128_0x8000, #0x8000; \
3663 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3664 ldr fb_ptr_next, [pixel_ptr, #28]; \
3666 vmov.u16 d128_0x0421, #0x0400; \
3667 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3669 vorr.u16 d128_0x0421, #0x0021; \
3670 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3672 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3673 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3674 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3675 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3676 blend_blocks_average_mask_set_##mask_evaluate(); \
3677 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3679 subs num_blocks, num_blocks, #1; \
3683 mov fb_ptr, fb_ptr_next; \
3684 ldr fb_ptr_next, [pixel_ptr, #28]; \
3686 vmov pixels, pixels_next; \
3687 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
3689 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3691 blend_blocks_average_mask_copy_##mask_evaluate(); \
3692 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
3694 blend_blocks_average_set_blend_mask_##texturing(pixels); \
3695 blend_blocks_average_set_stp_bit_##texturing(); \
3696 vmov fb_pixels, fb_pixels_next; \
3697 blend_blocks_average_combine_##texturing(pixels); \
3699 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3700 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3701 cmp fb_ptr_cmp, #28; \
3704 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3705 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3707 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3708 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3710 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3711 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3713 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3714 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3715 blend_blocks_average_mask_set_##mask_evaluate(); \
3716 vst1.u16 { fb_pixels }, [fb_ptr]; \
3719 subs num_blocks, num_blocks, #1; \
3723 blend_blocks_average_mask_copy_b_##mask_evaluate(); \
3724 vhadd.u16 blend_pixels, fb_pixels_no_msb, blend_pixels_next; \
3726 blend_blocks_average_set_blend_mask_##texturing(pixels_next); \
3727 blend_blocks_average_set_stp_bit_##texturing(); \
3728 blend_blocks_average_combine_##texturing(pixels_next); \
3730 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3731 vbif.u16 fb_pixels_next, blend_pixels, draw_mask_next; \
3732 vst1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3734 restore_abi_regs(); \
3735 ldmia sp!, { r4, pc }; \
3738 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3739 vbif.u16 fb_pixels, blend_pixels, draw_mask; \
3740 vst1.u16 { fb_pixels }, [fb_ptr]; \
3742 vld1.u16 { fb_pixels_next }, [fb_ptr_next]; \
3743 veor.u16 blend_pixels_next, pixels_next, fb_pixels_next; \
3744 vbic.u16 pixels_no_msb, pixels_next, d128_0x8000; \
3745 vand.u16 blend_pixels_next, blend_pixels_next, d128_0x0421; \
3746 vsub.u16 blend_pixels_next, pixels_no_msb, blend_pixels_next; \
3747 vbic.u16 fb_pixels_no_msb, fb_pixels_next, d128_0x8000; \
3751 blend_blocks_average_builder(textured, off)
3752 blend_blocks_average_builder(untextured, off)
3753 blend_blocks_average_builder(textured, on)
3754 blend_blocks_average_builder(untextured, on)
3757 #define blend_blocks_add_mask_set_on() \
3758 vclt.s16 write_mask, fb_pixels, #0 \
3760 #define blend_blocks_add_mask_copy_on() \
3761 vorr.u16 draw_mask, draw_mask, write_mask \
3763 #define blend_blocks_add_mask_set_off() \
3765 #define blend_blocks_add_mask_copy_off() \
3768 #define blend_blocks_add_textured_builder(mask_evaluate) \
3771 function(blend_blocks_textured_add_##mask_evaluate) \
3772 stmdb sp!, { r4, r14 }; \
3774 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3775 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3777 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3778 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3780 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3783 vmov.u16 d128_0x7C1F, #0x7C00; \
3784 vmov.u16 d128_0x03E0, #0x0300; \
3785 vmov.u16 d128_0x83E0, #0x8000; \
3786 vorr.u16 d128_0x03E0, #0x00E0; \
3787 vorr.u16 d128_0x7C1F, #0x001F; \
3788 vorr.u16 d128_0x83E0, d128_0x83E0, d128_0x03E0; \
3790 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3791 ldr fb_ptr_next, [pixel_ptr, #28]; \
3792 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3793 vclt.s16 blend_mask, pixels, #0; \
3794 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3795 blend_blocks_add_mask_set_##mask_evaluate(); \
3796 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3798 blend_blocks_add_mask_copy_##mask_evaluate(); \
3799 vorr.u16 pixels, pixels, msb_mask; \
3800 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3801 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3802 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3803 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3804 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3805 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3806 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3807 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3809 subs num_blocks, num_blocks, #1; \
3813 mov fb_ptr, fb_ptr_next; \
3815 ldr fb_ptr_next, [pixel_ptr, #28]; \
3817 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3818 vclt.s16 blend_mask, pixels, #0; \
3820 vorr.u16 pixels, pixels, msb_mask; \
3821 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3822 vand.u16 pixels_mg, pixels, d128_0x83E0; \
3824 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3825 pld [fb_ptr_next, #64]; \
3827 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3828 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3830 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3831 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3833 cmp fb_ptr_cmp, #28; \
3836 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3837 blend_blocks_add_mask_set_##mask_evaluate(); \
3838 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3839 blend_blocks_add_mask_copy_##mask_evaluate(); \
3840 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3841 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3842 vst1.u16 { blend_pixels }, [fb_ptr]; \
3845 vand.u16 fb_pixels_g, fb_pixels_masked, d128_0x03E0; \
3846 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3847 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_mg; \
3848 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3849 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x83E0; \
3851 subs num_blocks, num_blocks, #1; \
3855 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3856 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3857 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3859 restore_abi_regs(); \
3860 ldmia sp!, { r4, pc }; \
3863 vst1.u16 { blend_pixels }, [fb_ptr]; \
3864 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3866 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3867 blend_blocks_add_mask_set_##mask_evaluate(); \
3868 vand.u16 fb_pixels_masked, fb_pixels, blend_mask; \
3869 blend_blocks_add_mask_copy_##mask_evaluate(); \
3870 vand.u16 fb_pixels_rb, fb_pixels_masked, d128_0x7C1F; \
3874 #define blend_blocks_add_untextured_builder(mask_evaluate) \
3877 function(blend_blocks_untextured_add_##mask_evaluate) \
3878 stmdb sp!, { r4, r14 }; \
3880 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
3881 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
3883 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
3884 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
3886 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
3889 vmov.u16 d128_0x7C1F, #0x7C00; \
3890 vmov.u16 d128_0x03E0, #0x0300; \
3891 vorr.u16 d128_0x7C1F, #0x001F; \
3892 vorr.u16 d128_0x03E0, #0x00E0; \
3894 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3895 ldr fb_ptr_next, [pixel_ptr, #28]; \
3896 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3897 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3898 blend_blocks_add_mask_set_##mask_evaluate(); \
3899 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3901 blend_blocks_add_mask_copy_##mask_evaluate(); \
3902 vand.u16 pixels_g, pixels, d128_0x03E0; \
3903 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3904 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3905 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3906 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3907 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3908 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3910 subs num_blocks, num_blocks, #1; \
3914 mov fb_ptr, fb_ptr_next; \
3916 ldr fb_ptr_next, [pixel_ptr, #28]; \
3918 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
3920 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3921 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3922 vand.u16 pixels_g, pixels, d128_0x03E0; \
3924 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3925 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
3927 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
3928 add fb_ptr_cmp, fb_ptr_cmp, #14; \
3929 cmp fb_ptr_cmp, #28; \
3932 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3933 blend_blocks_add_mask_set_##mask_evaluate(); \
3934 blend_blocks_add_mask_copy_##mask_evaluate(); \
3935 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3936 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3937 vst1.u16 { blend_pixels }, [fb_ptr]; \
3940 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
3941 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
3942 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
3943 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
3944 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
3946 subs num_blocks, num_blocks, #1; \
3950 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
3951 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
3952 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
3953 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
3955 restore_abi_regs(); \
3956 ldmia sp!, { r4, pc }; \
3959 vst1.u16 { blend_pixels }, [fb_ptr]; \
3960 vand.u16 pixels_rb, pixels, d128_0x7C1F; \
3962 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
3963 blend_blocks_add_mask_set_##mask_evaluate(); \
3964 blend_blocks_add_mask_copy_##mask_evaluate(); \
3965 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
3969 blend_blocks_add_textured_builder(off)
3970 blend_blocks_add_textured_builder(on)
3971 blend_blocks_add_untextured_builder(off)
3972 blend_blocks_add_untextured_builder(on)
3974 #define blend_blocks_subtract_set_blend_mask_textured() \
3975 vclt.s16 blend_mask, pixels_next, #0 \
3977 #define blend_blocks_subtract_combine_textured() \
3978 vbif.u16 blend_pixels, pixels, blend_mask \
3980 #define blend_blocks_subtract_set_stp_textured() \
3981 vorr.u16 blend_pixels, #0x8000 \
3983 #define blend_blocks_subtract_msb_mask_textured() \
3984 vorr.u16 pixels, pixels_next, msb_mask \
3986 #define blend_blocks_subtract_set_blend_mask_untextured() \
3988 #define blend_blocks_subtract_combine_untextured() \
3990 #define blend_blocks_subtract_set_stp_untextured() \
3991 vorr.u16 blend_pixels, blend_pixels, msb_mask \
3993 #define blend_blocks_subtract_msb_mask_untextured() \
3996 #define blend_blocks_subtract_mask_set_on() \
3997 vclt.s16 write_mask, fb_pixels, #0 \
3999 #define blend_blocks_subtract_mask_copy_on() \
4000 vorr.u16 draw_mask, draw_mask_next, write_mask \
4002 #define blend_blocks_subtract_mask_set_off() \
4004 #define blend_blocks_subtract_mask_copy_off() \
4005 vmov draw_mask, draw_mask_next \
4008 #define blend_blocks_subtract_builder(texturing, mask_evaluate) \
4011 function(blend_blocks_##texturing##_subtract_##mask_evaluate) \
4012 stmdb sp!, { r4, r14 }; \
4014 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4015 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4017 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4018 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4020 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4023 vmov.u16 d128_0x7C1F, #0x7C00; \
4024 vmov.u16 d128_0x03E0, #0x0300; \
4025 vorr.u16 d128_0x7C1F, #0x001F; \
4026 vorr.u16 d128_0x03E0, #0x00E0; \
4028 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4029 ldr fb_ptr_next, [pixel_ptr, #28]; \
4030 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4031 blend_blocks_subtract_set_blend_mask_##texturing(); \
4032 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4033 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4034 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4036 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4037 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4038 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4039 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4040 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4042 subs num_blocks, num_blocks, #1; \
4046 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4047 mov fb_ptr, fb_ptr_next; \
4048 ldr fb_ptr_next, [pixel_ptr, #28]; \
4050 vld1.u32 { draw_mask_next }, [draw_mask_ptr, :128], c_64; \
4051 blend_blocks_subtract_msb_mask_##texturing(); \
4053 vld1.u32 { pixels_next }, [pixel_ptr, :128], c_64; \
4054 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4055 vand.u16 pixels_rb, pixels_next, d128_0x7C1F; \
4056 blend_blocks_subtract_set_stp_##texturing(); \
4057 vand.u16 pixels_g, pixels_next, d128_0x03E0; \
4058 blend_blocks_subtract_combine_##texturing(); \
4059 blend_blocks_subtract_set_blend_mask_##texturing(); \
4060 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4062 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4063 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4064 cmp fb_ptr_cmp, #28; \
4067 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4068 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4069 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4070 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4071 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4072 vst1.u16 { blend_pixels }, [fb_ptr]; \
4073 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4076 subs num_blocks, num_blocks, #1; \
4080 blend_blocks_subtract_mask_copy_##mask_evaluate(); \
4082 blend_blocks_subtract_msb_mask_##texturing(); \
4083 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4084 blend_blocks_subtract_set_stp_##texturing(); \
4085 blend_blocks_subtract_combine_##texturing(); \
4086 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4087 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4089 restore_abi_regs(); \
4090 ldmia sp!, { r4, pc }; \
4093 vst1.u16 { blend_pixels }, [fb_ptr]; \
4094 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4095 blend_blocks_subtract_mask_set_##mask_evaluate(); \
4096 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4097 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4098 vqsub.u8 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4099 vqsub.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4103 blend_blocks_subtract_builder(textured, off)
4104 blend_blocks_subtract_builder(textured, on)
4105 blend_blocks_subtract_builder(untextured, off)
4106 blend_blocks_subtract_builder(untextured, on)
4109 #define blend_blocks_add_fourth_textured_builder(mask_evaluate) \
4112 function(blend_blocks_textured_add_fourth_##mask_evaluate) \
4113 stmdb sp!, { r4, r14 }; \
4115 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4116 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4118 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4119 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4121 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4124 vmov.u16 d128_0x7C1F, #0x7C00; \
4125 vmov.u16 d128_0x03E0, #0x0300; \
4126 vmov.u16 d128_0x1C07, #0x1C00; \
4127 vmov.u16 d128_0x00E0, #0x00E0; \
4128 vorr.u16 d128_0x7C1F, #0x001F; \
4129 vorr.u16 d128_0x03E0, #0x00E0; \
4130 vorr.u16 d128_0x1C07, #0x0007; \
4132 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4133 ldr fb_ptr_next, [pixel_ptr, #28]; \
4134 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4135 vclt.s16 blend_mask, pixels, #0; \
4136 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4137 blend_blocks_add_mask_set_##mask_evaluate(); \
4138 vshr.s16 pixels_fourth, pixels, #2; \
4139 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4141 blend_blocks_add_mask_copy_##mask_evaluate(); \
4142 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4143 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4144 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4145 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4146 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4147 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4148 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4150 subs num_blocks, num_blocks, #1; \
4154 mov fb_ptr, fb_ptr_next; \
4155 ldr fb_ptr_next, [pixel_ptr, #28]; \
4157 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4158 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4159 vbif.u16 blend_pixels, pixels, blend_mask; \
4161 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4162 vclt.s16 blend_mask, pixels, #0; \
4163 vshr.s16 pixels_fourth, pixels, #2; \
4164 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4165 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4167 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4168 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4170 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4171 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4172 cmp fb_ptr_cmp, #28; \
4175 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4176 blend_blocks_add_mask_set_##mask_evaluate(); \
4177 blend_blocks_add_mask_copy_##mask_evaluate(); \
4178 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4179 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4180 vst1.u16 { blend_pixels }, [fb_ptr]; \
4183 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4184 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4185 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4186 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4187 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4189 subs num_blocks, num_blocks, #1; \
4193 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4194 vorr.u16 blend_pixels, #0x8000; /* stp */ \
4195 vbif.u16 blend_pixels, pixels, blend_mask; \
4196 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4197 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4198 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4200 restore_abi_regs(); \
4201 ldmia sp!, { r4, pc }; \
4204 vst1.u16 { blend_pixels }, [fb_ptr]; \
4205 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4207 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4208 blend_blocks_add_mask_set_##mask_evaluate(); \
4209 blend_blocks_add_mask_copy_##mask_evaluate(); \
4210 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4215 #define blend_blocks_add_fourth_untextured_builder(mask_evaluate) \
4218 function(blend_blocks_untextured_add_fourth_##mask_evaluate) \
4219 stmdb sp!, { r4, r14 }; \
4221 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset; \
4222 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
4224 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16); \
4225 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]; \
4227 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset; \
4230 vmov.u16 d128_0x7C1F, #0x7C00; \
4231 vmov.u16 d128_0x03E0, #0x0300; \
4232 vmov.u16 d128_0x1C07, #0x1C00; \
4233 vmov.u16 d128_0x00E0, #0x00E0; \
4234 vorr.u16 d128_0x7C1F, #0x001F; \
4235 vorr.u16 d128_0x03E0, #0x00E0; \
4236 vorr.u16 d128_0x1C07, #0x0007; \
4238 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4239 ldr fb_ptr_next, [pixel_ptr, #28]; \
4240 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4241 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4242 blend_blocks_add_mask_set_##mask_evaluate(); \
4243 vshr.s16 pixels_fourth, pixels, #2; \
4244 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4246 blend_blocks_add_mask_copy_##mask_evaluate(); \
4247 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4248 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4249 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4250 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4251 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4252 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4253 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4255 subs num_blocks, num_blocks, #1; \
4259 mov fb_ptr, fb_ptr_next; \
4260 ldr fb_ptr_next, [pixel_ptr, #28]; \
4262 vld1.u32 { pixels }, [pixel_ptr, :128], c_64; \
4264 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4265 vshr.s16 pixels_fourth, pixels, #2; \
4266 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4267 vand.u16 pixels_rb, pixels_fourth, d128_0x1C07; \
4269 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4270 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64; \
4272 sub fb_ptr_cmp, fb_ptr_next, fb_ptr; \
4273 add fb_ptr_cmp, fb_ptr_cmp, #14; \
4274 cmp fb_ptr_cmp, #28; \
4277 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4278 blend_blocks_add_mask_set_##mask_evaluate(); \
4279 blend_blocks_add_mask_copy_##mask_evaluate(); \
4280 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4281 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4282 vst1.u16 { blend_pixels }, [fb_ptr]; \
4285 vand.u16 fb_pixels_g, fb_pixels, d128_0x03E0; \
4286 vadd.u16 fb_pixels_rb, fb_pixels_rb, pixels_rb; \
4287 vadd.u16 fb_pixels_g, fb_pixels_g, pixels_g; \
4288 vmin.u8 fb_pixels_rb, fb_pixels_rb, d128_0x7C1F; \
4289 vmin.u16 fb_pixels_g, fb_pixels_g, d128_0x03E0; \
4291 subs num_blocks, num_blocks, #1; \
4295 vorr.u16 blend_pixels, fb_pixels_rb, fb_pixels_g; \
4296 vorr.u16 blend_pixels, blend_pixels, msb_mask; \
4297 vbit.u16 blend_pixels, fb_pixels, draw_mask; \
4298 vst1.u16 { blend_pixels }, [fb_ptr_next]; \
4300 restore_abi_regs(); \
4301 ldmia sp!, { r4, pc }; \
4304 vst1.u16 { blend_pixels }, [fb_ptr]; \
4305 vand.u16 pixels_g, pixels_fourth, d128_0x00E0; \
4307 vld1.u16 { fb_pixels }, [fb_ptr_next]; \
4308 blend_blocks_add_mask_set_##mask_evaluate(); \
4309 blend_blocks_add_mask_copy_##mask_evaluate(); \
4310 vand.u16 fb_pixels_rb, fb_pixels, d128_0x7C1F; \
4314 blend_blocks_add_fourth_textured_builder(off)
4315 blend_blocks_add_fourth_textured_builder(on)
4316 blend_blocks_add_fourth_untextured_builder(off)
4317 blend_blocks_add_fourth_untextured_builder(on)
4319 // TODO: Optimize this more. Need a scene that actually uses it for
4324 function(blend_blocks_textured_unblended_on)
4325 stmdb sp!, { r4, r14 }
4327 add mask_msb_ptr, psx_gpu, #psx_gpu_mask_msb_offset
4328 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
4330 add pixel_ptr, psx_gpu, #(psx_gpu_blocks_offset + 16)
4331 vld1.u16 { msb_mask_low[], msb_mask_high[] }, [mask_msb_ptr, :16]
4333 add draw_mask_ptr, psx_gpu, #psx_gpu_blocks_offset
4336 ldr fb_ptr, [pixel_ptr, #28]
4337 vld1.u16 { fb_pixels }, [fb_ptr]
4338 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4339 vclt.s16 write_mask, fb_pixels, #0
4340 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4342 subs num_blocks, num_blocks, #1
4346 vorr.u16 pixels, pixels, msb_mask
4347 vorr.u16 draw_mask, draw_mask, write_mask
4348 vbif.u16 fb_pixels, pixels, draw_mask
4349 vst1.u16 { fb_pixels }, [fb_ptr]
4351 ldr fb_ptr, [pixel_ptr, #28]
4352 vld1.u16 { fb_pixels }, [fb_ptr]
4353 vld1.u32 { draw_mask }, [draw_mask_ptr, :128], c_64
4354 vclt.s16 write_mask, fb_pixels, #0
4355 vld1.u32 { pixels }, [pixel_ptr, :128], c_64
4357 subs num_blocks, num_blocks, #1
4361 vorr.u16 pixels, pixels, msb_mask
4362 vorr.u16 draw_mask, draw_mask, write_mask
4363 vbif.u16 fb_pixels, pixels, draw_mask
4364 vst1.u16 { fb_pixels }, [fb_ptr]
4367 ldmia sp!, { r4, pc }
4370 function(blend_blocks_textured_unblended_off)
4380 vld1.u32 { u_whole_8, v_whole_8 }, [r1, :128], r3
4400 #define num_width r12
4410 function(render_block_fill_body)
4411 vdup.u16 colors_a, color
4414 vmov colors_b, colors_a
4415 sub pitch, pitch, width, lsl #1
4417 mov num_width, width
4420 vst1.u32 { colors_a, colors_b }, [vram_ptr, :256]!
4422 subs num_width, num_width, #16
4425 add vram_ptr, vram_ptr, pitch
4426 mov num_width, width
4428 subs height, height, #1
4442 #undef dirty_textures_mask
4444 #undef current_texture_mask
4455 #define offset_u_right r10
4456 #define width_rounded r11
4457 #define height_rounded r12
4459 #define texture_offset_base r1
4460 #define tile_width r2
4461 #define tile_height r3
4462 #define num_blocks r4
4464 #define sub_tile_height r6
4466 #define texture_mask r8
4467 #define column_data r9
4468 #define texture_offset r10
4469 #define tiles_remaining r11
4470 #define fb_ptr_advance_column r12
4471 #define texture_block_ptr r14
4475 #define texture_page_ptr r3
4476 #define left_block_mask r4
4477 #define right_block_mask r5
4478 #define texture_mask_rev r10
4479 #define control_mask r11
4481 #define dirty_textures_mask r4
4483 #define current_texture_mask r6
4497 #define draw_masks_fb_ptrs q1
4499 #define draw_mask_fb_ptr_left d2
4500 #define draw_mask_fb_ptr_right d3
4502 #define draw_mask_fb_ptr_left_a d2
4503 #define draw_mask_fb_ptr_left_b d3
4504 #define draw_mask_fb_ptr_right_a d10
4505 #define draw_mask_fb_ptr_right_b d11
4506 #define draw_masks_fb_ptrs2 q5
4508 #define clut_low_a d4
4509 #define clut_low_b d5
4510 #define clut_high_a d6
4511 #define clut_high_b d7
4513 #define block_masks d8
4514 #define block_masks_shifted d9
4519 #define texels_low d12
4520 #define texels_high d13
4522 #define texels_wide_low d14
4523 #define texels_wide_high d15
4524 #define texels_wide q7
4527 setup_sprite_flush_blocks:
4530 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4531 bl flush_render_block_buffer
4532 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
4536 add block, psx_gpu, #psx_gpu_blocks_offset
4540 setup_sprite_update_texture_4bpp_cache:
4541 stmdb sp!, { r0 - r3, r14 }
4542 bl update_texture_4bpp_cache
4543 ldmia sp!, { r0 - r3, pc }
4546 setup_sprite_update_texture_8bpp_cache:
4547 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r14 }
4548 bl update_texture_8bpp_cache
4549 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS pc }
4552 #define setup_sprite_tiled_initialize_4bpp() \
4553 ldr dirty_textures_mask, \
4554 [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]; \
4555 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4557 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4558 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4560 tst current_texture_mask, dirty_textures_mask; \
4561 vuzp.u8 clut_a, clut_b; \
4563 blne setup_sprite_update_texture_4bpp_cache \
4565 #define setup_sprite_tiled_initialize_8bpp() \
4566 ldr dirty_textures_mask, \
4567 [psx_gpu, #psx_gpu_dirty_textures_8bpp_mask_offset]; \
4568 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]; \
4570 tst current_texture_mask, dirty_textures_mask; \
4571 blne setup_sprite_update_texture_8bpp_cache \
4574 #define setup_sprite_block_count_single() \
4577 #define setup_sprite_block_count_double() \
4578 sub_tile_height, lsl #1 \
4580 #define setup_sprite_tile_add_blocks(type) \
4581 add num_blocks, num_blocks, setup_sprite_block_count_##type(); \
4582 cmp num_blocks, #MAX_BLOCKS; \
4584 movgt num_blocks, setup_sprite_block_count_##type(); \
4585 blgt setup_sprite_flush_blocks \
4588 #define setup_sprite_tile_full_4bpp(edge) \
4589 setup_sprite_tile_add_blocks(double); \
4592 and texture_block_ptr, texture_offset, texture_mask; \
4593 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4596 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4597 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4599 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4600 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4602 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4603 add texture_block_ptr, texture_offset, #8; \
4605 and texture_block_ptr, texture_block_ptr, texture_mask; \
4606 add block, block, #40; \
4608 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4609 add fb_ptr, fb_ptr, #16; \
4611 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4612 add block, block, #24; \
4614 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4615 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4618 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4619 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4621 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4622 add block, block, #40; \
4624 add texture_offset, texture_offset, #0x10; \
4625 add fb_ptr, fb_ptr, #(2048 - 16); \
4627 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4628 add block, block, #24; \
4630 subs sub_tile_height, sub_tile_height, #1; \
4633 add texture_offset, texture_offset, #0xF00; \
4634 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4637 #define setup_sprite_tile_half_4bpp(edge) \
4638 setup_sprite_tile_add_blocks(single); \
4641 and texture_block_ptr, texture_offset, texture_mask; \
4642 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4645 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4646 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4648 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
4649 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
4651 vst2.u8 { texels_low, texels_high }, [block, :128]; \
4652 add block, block, #40; \
4654 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4655 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4657 add block, block, #24; \
4658 add texture_offset, texture_offset, #0x10; \
4660 add fb_ptr, fb_ptr, #2048; \
4661 subs sub_tile_height, sub_tile_height, #1; \
4665 add texture_offset, texture_offset, #0xF00; \
4666 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4669 #define setup_sprite_tile_full_8bpp(edge) \
4670 setup_sprite_tile_add_blocks(double); \
4671 add block, block, #16; \
4674 and texture_block_ptr, texture_offset, texture_mask; \
4675 vmov.u32 draw_mask_fb_ptr_left[1], fb_ptr; \
4678 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4679 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4681 add texture_block_ptr, texture_offset, #8; \
4682 vst1.u32 { texels }, [block, :64]; \
4684 and texture_block_ptr, texture_block_ptr, texture_mask; \
4685 add block, block, #24; \
4687 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4689 add fb_ptr, fb_ptr, #16; \
4690 vst1.u32 { draw_mask_fb_ptr_left }, [block, :64]; \
4692 add block, block, #40; \
4693 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4696 vmov.u32 draw_mask_fb_ptr_right[1], fb_ptr; \
4697 vst1.u32 { texels }, [block, :64]; \
4698 add block, block, #24; \
4700 add texture_offset, texture_offset, #0x10; \
4701 add fb_ptr, fb_ptr, #(2048 - 16); \
4703 vst1.u32 { draw_mask_fb_ptr_right }, [block, :64]; \
4704 add block, block, #40; \
4706 subs sub_tile_height, sub_tile_height, #1; \
4709 sub block, block, #16; \
4710 add texture_offset, texture_offset, #0xF00; \
4711 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4714 #define setup_sprite_tile_half_8bpp(edge) \
4715 setup_sprite_tile_add_blocks(single); \
4716 add block, block, #16; \
4719 and texture_block_ptr, texture_offset, texture_mask; \
4720 vmov.u32 draw_mask_fb_ptr_##edge[1], fb_ptr; \
4723 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4724 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4726 vst1.u32 { texels }, [block, :64]; \
4727 add block, block, #24; \
4729 vst1.u32 { draw_mask_fb_ptr_##edge }, [block, :64]; \
4730 add block, block, #40; \
4732 add texture_offset, texture_offset, #0x10; \
4733 add fb_ptr, fb_ptr, #2048; \
4735 subs sub_tile_height, sub_tile_height, #1; \
4738 sub block, block, #16; \
4739 add texture_offset, texture_offset, #0xF00; \
4740 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
4743 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
4744 add texture_offset, texture_offset_base, #8; \
4745 add fb_ptr, fb_ptr, #16 \
4747 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
4748 mov texture_offset, texture_offset_base \
4750 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
4751 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
4753 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
4754 mov texture_offset, texture_offset_base \
4756 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
4757 sub fb_ptr, fb_ptr, #16 \
4759 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
4761 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
4762 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
4764 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
4767 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
4769 mov sub_tile_height, column_data; \
4770 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4771 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4772 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4774 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
4776 and sub_tile_height, column_data, #0xFF; \
4777 mov tiles_remaining, column_data, lsr #16; \
4778 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
4779 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4781 subs tiles_remaining, tiles_remaining, #1; \
4785 mov sub_tile_height, #16; \
4786 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4787 subs tiles_remaining, tiles_remaining, #1; \
4791 uxtb sub_tile_height, column_data, ror #8; \
4792 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
4793 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge) \
4796 #define setup_sprite_column_data_single() \
4797 mov column_data, height; \
4798 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset] \
4800 #define setup_sprite_column_data_multi() \
4801 and height_rounded, height_rounded, #0xF; \
4802 rsb column_data, offset_v, #16; \
4804 add height_rounded, height_rounded, #1; \
4805 sub tile_height, tile_height, #1; \
4807 orr column_data, column_data, tile_height, lsl #16; \
4808 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]; \
4810 orr column_data, column_data, height_rounded, lsl #8 \
4812 #define setup_sprite_setup_left_draw_mask_fb_ptr() \
4813 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4814 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4816 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column() \
4817 mov fb_ptr_advance_column, #32; \
4818 vdup.u8 draw_mask_fb_ptr_left, block_masks[0]; \
4820 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11; \
4821 vdup.u8 draw_mask_fb_ptr_right, block_masks[1] \
4823 #define setup_sprite_setup_right_draw_mask_fb_ptr() \
4824 vdup.u8 draw_mask_fb_ptr_left, block_masks[4]; \
4825 vdup.u8 draw_mask_fb_ptr_right, block_masks[5] \
4827 #define setup_sprite_tile_column_width_single(tm, multi_height, edge_mode, \
4829 setup_sprite_##tm##_single_##multi_height##_##edge_mode##_##edge##x4mode: \
4830 setup_sprite_column_data_##multi_height(); \
4831 vext.32 block_masks_shifted, block_masks, block_masks, #1; \
4832 vorr.u32 block_masks, block_masks, block_masks_shifted; \
4833 setup_sprite_setup_left_draw_mask_fb_ptr##x4mode(); \
4835 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, tm, x4mode); \
4836 restore_abi_regs(); \
4837 ldmia sp!, { r4 - r11, pc } \
4839 #define setup_sprite_tiled_advance_column() \
4840 add texture_offset_base, texture_offset_base, #0x100; \
4841 tst texture_offset_base, #0xF00; \
4842 subeq texture_offset_base, texture_offset_base, #(0x100 + 0xF00) \
4844 #define setup_sprite_tile_column_width_multi(tm, multi_height, left_mode, \
4845 right_mode, x4mode) \
4846 setup_sprite_##tm##_multi_##multi_height##_##left_mode##_##right_mode##x4mode:\
4847 setup_sprite_column_data_##multi_height(); \
4849 setup_sprite_setup_left_draw_mask_fb_ptr_advance_column##x4mode(); \
4851 setup_sprite_tile_column_height_##multi_height(left_mode, right, tm, x4mode);\
4853 subs tile_width, tile_width, #2; \
4854 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4858 vmov.u8 draw_masks_fb_ptrs, #0; \
4859 vmov.u8 draw_masks_fb_ptrs2, #0; \
4862 setup_sprite_tiled_advance_column(); \
4863 setup_sprite_tile_column_height_##multi_height(full, none, tm, x4mode); \
4864 add fb_ptr, fb_ptr, fb_ptr_advance_column; \
4865 subs tile_width, tile_width, #1; \
4869 setup_sprite_setup_right_draw_mask_fb_ptr##x4mode(); \
4871 setup_sprite_tiled_advance_column(); \
4872 setup_sprite_tile_column_height_##multi_height(right_mode, left, tm, x4mode);\
4873 restore_abi_regs(); \
4874 ldmia sp!, { r4 - r11, pc } \
4877 #define setup_sprite_offset_u_adjust() \
4879 #define setup_sprite_get_left_block_mask() \
4880 and left_block_mask, left_block_mask, #0xFF \
4882 #define setup_sprite_compare_left_block_mask() \
4883 cmp left_block_mask, #0xFF \
4885 #define setup_sprite_get_right_block_mask() \
4886 uxtb right_block_mask, right_block_mask, ror #8 \
4888 #define setup_sprite_compare_right_block_mask() \
4889 cmp right_block_mask, #0xFF \
4894 #define fb_ptr2 column_data
4896 #define setup_sprite_offset_u_adjust_4x() \
4897 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
4898 lsl offset_u_right, #1; \
4900 add offset_u_right, #1 \
4902 #define setup_sprite_get_left_block_mask_4x() \
4903 sxth left_block_mask, left_block_mask \
4905 #define setup_sprite_compare_left_block_mask_4x() \
4906 cmp left_block_mask, #0xFFFFFFFF \
4908 #define setup_sprite_get_right_block_mask_4x() \
4909 sxth right_block_mask, right_block_mask, ror #16 \
4911 #define setup_sprite_compare_right_block_mask_4x() \
4912 cmp right_block_mask, #0xFFFFFFFF \
4915 #define widen_texels_16bpp(texels_) \
4916 vmov texels_wide_low, texels_; \
4917 vmov texels_wide_high, texels_; \
4918 vzip.16 texels_wide_low, texels_wide_high \
4920 #define widen_texels_8bpp(texels_) \
4921 vmov texels_wide_low, texels_; \
4922 vmov texels_wide_high, texels_; \
4923 vzip.8 texels_wide_low, texels_wide_high \
4925 #define write_block_16bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4926 vst1.u32 { texels_ }, [block_, :128]; \
4927 add block_, block_, #40; \
4929 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4930 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4931 add block_, block_, #24 \
4933 /* assumes 16-byte offset already added to block_ */
4934 #define write_block_8bpp(texels_, block_, draw_mask_fb_ptr_, fb_ptr_) \
4935 vst1.u32 { texels_ }, [block_, :64]; \
4936 add block_, block_, #24; \
4938 vmov.u32 draw_mask_fb_ptr_[1], fb_ptr_; \
4939 vst1.u32 { draw_mask_fb_ptr_ }, [block_, :64]; \
4940 add block_, block_, #40 \
4942 #define do_texture_block_16bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4943 draw_mask_fb_ptr_b_) \
4944 widen_texels_16bpp(texels_low); \
4945 add fb_ptr_tmp, fb_ptr, #1024*2; \
4947 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr); \
4949 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4950 widen_texels_16bpp(texels_high); \
4952 add fb_ptr_tmp, fb_ptr, #8*2; \
4953 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4955 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4956 write_block_16bpp(texels_wide, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4958 #define do_texture_block_8bpp_4x(fb_ptr_tmp, draw_mask_fb_ptr_a_, \
4959 draw_mask_fb_ptr_b_) \
4960 widen_texels_8bpp(texels); \
4961 add fb_ptr_tmp, fb_ptr, #1024*2; \
4963 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr); \
4964 write_block_8bpp(texels_wide_low, block, draw_mask_fb_ptr_a_, fb_ptr_tmp); \
4966 add fb_ptr_tmp, fb_ptr, #8*2; \
4967 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp); \
4969 add fb_ptr_tmp, fb_ptr_tmp, #1024*2; \
4970 write_block_8bpp(texels_wide_high, block, draw_mask_fb_ptr_b_, fb_ptr_tmp) \
4973 #define setup_sprite_tiled_initialize_4bpp_4x() \
4974 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]; \
4975 vld1.u32 { clut_a, clut_b }, [clut_ptr, :128]; \
4977 vuzp.u8 clut_a, clut_b \
4979 #define setup_sprite_tiled_initialize_8bpp_4x() \
4982 #define setup_sprite_block_count_single_4x() \
4983 sub_tile_height, lsl #2 \
4985 #define setup_sprite_block_count_double_4x() \
4986 sub_tile_height, lsl #(1+2) \
4988 #define setup_sprite_tile_full_4bpp_4x(edge) \
4989 setup_sprite_tile_add_blocks(double_4x); \
4990 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
4993 and texture_block_ptr, texture_offset, texture_mask; \
4996 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
4997 vld1.u32 { texels }, [texture_block_ptr, :64]; \
4999 add texture_block_ptr, texture_offset, #8; \
5000 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5002 and texture_block_ptr, texture_block_ptr, texture_mask; \
5003 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5005 vzip.8 texels_low, texels_high; \
5006 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5007 draw_mask_fb_ptr_left_b); \
5009 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5010 pld [fb_ptr, #2048]; \
5012 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5013 add fb_ptr, fb_ptr, #16*2; \
5015 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5016 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5018 vzip.8 texels_low, texels_high; \
5019 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5020 draw_mask_fb_ptr_right_b); \
5022 add texture_offset, texture_offset, #0x10; \
5023 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5025 subs sub_tile_height, sub_tile_height, #1; \
5028 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5029 add texture_offset, texture_offset, #0xF00; \
5030 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5033 #define setup_sprite_tile_half_4bpp_4x(edge) \
5034 setup_sprite_tile_add_blocks(single_4x); \
5035 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5038 and texture_block_ptr, texture_offset, texture_mask; \
5041 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5042 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5044 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5045 vtbl.8 texels_low, { clut_low_a, clut_low_b }, texels; \
5047 vtbl.8 texels_high, { clut_high_a, clut_high_b }, texels; \
5048 add texture_offset, texture_offset, #0x10; \
5050 vzip.8 texels_low, texels_high; \
5051 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5052 draw_mask_fb_ptr_##edge##_b); \
5054 pld [fb_ptr, #2048]; \
5055 add fb_ptr, fb_ptr, #2048 * 2; \
5057 subs sub_tile_height, sub_tile_height, #1; \
5060 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5061 add texture_offset, texture_offset, #0xF00; \
5062 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5065 #define setup_sprite_tile_full_8bpp_4x(edge) \
5066 setup_sprite_tile_add_blocks(double_4x); \
5067 add block, block, #16; \
5068 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5071 and texture_block_ptr, texture_offset, texture_mask; \
5074 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5075 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5077 add texture_block_ptr, texture_offset, #8; \
5078 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_left_a, \
5079 draw_mask_fb_ptr_left_b); \
5081 pld [fb_ptr, #2048]; \
5082 and texture_block_ptr, texture_block_ptr, texture_mask; \
5084 add fb_ptr, fb_ptr, #16*2; \
5085 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5087 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5089 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_right_a, \
5090 draw_mask_fb_ptr_right_b); \
5092 add texture_offset, texture_offset, #0x10; \
5093 add fb_ptr, fb_ptr, #(2048 - 16) * 2; \
5095 subs sub_tile_height, sub_tile_height, #1; \
5098 sub block, block, #16; \
5099 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5100 add texture_offset, texture_offset, #0xF00; \
5101 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5104 #define setup_sprite_tile_half_8bpp_4x(edge) \
5105 setup_sprite_tile_add_blocks(single_4x); \
5106 add block, block, #16; \
5107 str column_data, [sp, #-8]!; /* fb_ptr2 */ \
5110 and texture_block_ptr, texture_offset, texture_mask; \
5113 add texture_block_ptr, texture_page_ptr, texture_block_ptr; \
5114 vld1.u32 { texels }, [texture_block_ptr, :64]; \
5116 pld [fb_ptr, #2048]; \
5117 do_texture_block_8bpp_4x(fb_ptr2, draw_mask_fb_ptr_##edge##_a, \
5118 draw_mask_fb_ptr_##edge##_b); \
5120 add texture_offset, texture_offset, #0x10; \
5121 add fb_ptr, fb_ptr, #2048 * 2; \
5123 subs sub_tile_height, sub_tile_height, #1; \
5126 sub block, block, #16; \
5127 ldr column_data, [sp], #8; /* fb_ptr2 */ \
5128 add texture_offset, texture_offset, #0xF00; \
5129 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset] \
5132 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
5133 add texture_offset, texture_offset_base, #8; \
5134 add fb_ptr, fb_ptr, #16 * 2 \
5136 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
5137 mov texture_offset, texture_offset_base \
5139 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
5140 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
5142 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
5143 mov texture_offset, texture_offset_base \
5145 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
5146 sub fb_ptr, fb_ptr, #16 * 2 \
5148 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
5150 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
5151 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
5153 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
5156 #define setup_sprite_setup_left_draw_mask_fb_ptr_4x() \
5157 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5158 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5159 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5160 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5162 #define setup_sprite_setup_left_draw_mask_fb_ptr_advance_column_4x() \
5163 mov fb_ptr_advance_column, #32 * 2; \
5164 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[0]; \
5165 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[1]; \
5166 sub fb_ptr_advance_column, fb_ptr_advance_column, height, lsl #11 + 1; \
5167 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[2]; \
5168 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[3] \
5170 #define setup_sprite_setup_right_draw_mask_fb_ptr_4x() \
5171 vdup.u8 draw_mask_fb_ptr_left_a, block_masks[4]; \
5172 vdup.u8 draw_mask_fb_ptr_left_b, block_masks[5]; \
5173 vdup.u8 draw_mask_fb_ptr_right_a, block_masks[6]; \
5174 vdup.u8 draw_mask_fb_ptr_right_b, block_masks[7] \
5184 // [sp + 12]: color (unused)
5186 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
5188 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
5190 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
5192 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
5194 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
5196 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
5198 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
5200 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
5202 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
5204 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
5206 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
5208 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
5210 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
5212 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
5214 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
5219 function(setup_sprite_##texture_mode##x4mode) \
5220 stmdb sp!, { r4 - r11, r14 }; \
5221 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
5224 and offset_u, u, #0xF; \
5226 ldr width, [sp, #40]; \
5227 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]; \
5229 ldr height, [sp, #44]; \
5230 add fb_ptr, fb_ptr, y, lsl #11; \
5234 add fb_ptr, fb_ptr, x, lsl #1; \
5235 and offset_v, v, #0xF; \
5237 sub fb_ptr, fb_ptr, offset_u, lsl #1; \
5238 add width_rounded, offset_u, width; \
5240 add height_rounded, offset_v, height; \
5241 add width_rounded, width_rounded, #15; \
5243 add height_rounded, height_rounded, #15; \
5244 mov tile_width, width_rounded, lsr #4; \
5246 /* texture_offset_base = VH-VL-00-00 */\
5247 mov texture_offset_base, v, lsl #8; \
5248 and offset_u_right, width_rounded, #0xF; \
5250 /* texture_offset_base = VH-UH-UL-00 */\
5251 bfi texture_offset_base, u, #4, #8; \
5252 mov right_block_mask, #0xFFFFFFFE; \
5254 setup_sprite_offset_u_adjust##x4mode(); \
5256 /* texture_offset_base = VH-UH-VL-00 */\
5257 bfi texture_offset_base, v, #4, #4; \
5258 mov left_block_mask, #0xFFFFFFFF; \
5260 mov tile_height, height_rounded, lsr #4; \
5261 mvn left_block_mask, left_block_mask, lsl offset_u; \
5263 /* texture_mask = HH-HL-WH-WL */\
5264 ldrh texture_mask, [psx_gpu, #psx_gpu_texture_mask_width_offset]; \
5265 mov right_block_mask, right_block_mask, lsl offset_u_right; \
5267 /* texture_mask_rev = WH-WL-HH-HL */\
5268 rev16 texture_mask_rev, texture_mask; \
5269 vmov block_masks, left_block_mask, right_block_mask; \
5271 /* texture_mask = HH-HL-HL-WL */\
5272 bfi texture_mask, texture_mask_rev, #4, #4; \
5273 /* texture_mask_rev = 00-00-00-WH */\
5274 mov texture_mask_rev, texture_mask_rev, lsr #12; \
5276 /* texture_mask = HH-WH-HL-WL */\
5277 bfi texture_mask, texture_mask_rev, #8, #4; \
5278 setup_sprite_get_left_block_mask##x4mode(); \
5280 mov control_mask, #0; \
5281 setup_sprite_compare_left_block_mask##x4mode(); \
5283 setup_sprite_get_right_block_mask##x4mode(); \
5284 orreq control_mask, control_mask, #0x4; \
5286 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]; \
5287 setup_sprite_compare_right_block_mask##x4mode(); \
5289 orreq control_mask, control_mask, #0x8; \
5290 cmp tile_width, #1; \
5292 add block, psx_gpu, #psx_gpu_blocks_offset; \
5293 orreq control_mask, control_mask, #0x1; \
5295 cmp tile_height, #1; \
5296 add block, block, num_blocks, lsl #6; \
5298 orreq control_mask, control_mask, #0x2; \
5299 JT_OP_REL(9f, control_mask, temp); \
5300 JT_OP(ldr pc, [pc, control_mask, lsl #2]); \
5304 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_full##x4mode); \
5305 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_full_none##x4mode); \
5306 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_full##x4mode); \
5307 .word JTE(9b, setup_sprite_##texture_mode##_single_single_full_none##x4mode); \
5308 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_full##x4mode); \
5309 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_right##x4mode); \
5310 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_full##x4mode); \
5311 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_right##x4mode);\
5312 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_full_half##x4mode); \
5313 .word JTE(9b, setup_sprite_##texture_mode##_single_multi_half_left##x4mode); \
5314 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_full_half##x4mode); \
5315 .word JTE(9b, setup_sprite_##texture_mode##_single_single_half_left##x4mode); \
5316 .word JTE(9b, setup_sprite_##texture_mode##_multi_multi_half_half##x4mode); \
5318 .word JTE(9b, setup_sprite_##texture_mode##_multi_single_half_half##x4mode); \
5321 setup_sprite_tiled_builder(4bpp,);
5322 setup_sprite_tiled_builder(8bpp,);
5324 #undef draw_mask_fb_ptr_left
5325 #undef draw_mask_fb_ptr_right
5327 setup_sprite_tiled_builder(4bpp, _4x);
5328 setup_sprite_tiled_builder(8bpp, _4x);
5336 #define block_ptr r0
5337 #define num_blocks r1
5339 #define texel_shift_mask r3
5340 #define block_pixels_a r4
5341 #define block_pixels_b r5
5350 #define texels_01 r6
5351 #define texels_23 r7
5352 #define texels_45 r8
5353 #define texels_67 r9
5355 function(texture_sprite_blocks_8bpp)
5356 stmdb sp!, { r4 - r11, r14 }
5357 movw texel_shift_mask, #(0xFF << 1)
5359 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5360 ldr clut_ptr, [psx_gpu, #psx_gpu_clut_ptr_offset]
5362 add block_ptr, psx_gpu, #psx_gpu_blocks_offset
5363 ldr block_pixels_a, [block_ptr, #16]
5366 and texel_0, texel_shift_mask, block_pixels_a, lsl #1
5367 ldr block_pixels_b, [block_ptr, #20]
5369 and texel_1, texel_shift_mask, block_pixels_a, lsr #7
5370 ldrh texel_0, [clut_ptr, texel_0]
5372 and texel_2, texel_shift_mask, block_pixels_a, lsr #15
5373 ldrh texel_1, [clut_ptr, texel_1]
5375 and texel_3, texel_shift_mask, block_pixels_a, lsr #23
5376 ldr block_pixels_a, [block_ptr, #(64 + 16)]
5378 ldrh texel_2, [clut_ptr, texel_2]
5379 and texel_4, texel_shift_mask, block_pixels_b, lsl #1
5381 ldrh texel_3, [clut_ptr, texel_3]
5382 and texel_5, texel_shift_mask, block_pixels_b, lsr #7
5384 ldrh texel_4, [clut_ptr, texel_4]
5385 and texel_6, texel_shift_mask, block_pixels_b, lsr #15
5387 ldrh texel_5, [clut_ptr, texel_5]
5388 and texel_7, texel_shift_mask, block_pixels_b, lsr #23
5390 ldrh texel_6, [clut_ptr, texel_6]
5391 orr texels_01, texel_0, texel_1, lsl #16
5393 ldrh texel_7, [clut_ptr, texel_7]
5394 orr texels_23, texel_2, texel_3, lsl #16
5396 orr texels_45, texel_4, texel_5, lsl #16
5397 str texels_01, [block_ptr, #0]
5399 orr texels_67, texel_6, texel_7, lsl #16
5400 str texels_23, [block_ptr, #4]
5402 subs num_blocks, num_blocks, #1
5403 str texels_45, [block_ptr, #8]
5405 str texels_67, [block_ptr, #12]
5406 add block_ptr, block_ptr, #64
5410 ldmia sp!, { r4 - r11, pc }
5413 #undef width_rounded
5416 #undef texture_offset
5419 #undef texels_wide_low
5420 #undef texels_wide_high
5432 #define left_offset r8
5433 #define width_rounded r9
5434 #define right_width r10
5436 #define block_width r11
5438 #define texture_offset_base r1
5439 #define texture_mask r2
5440 #define texture_page_ptr r3
5441 #define num_blocks r4
5444 #define texture_offset r8
5445 #define blocks_remaining r9
5447 #define fb_ptr_pitch r12
5448 #define texture_block_ptr r14
5450 #define texture_mask_width r2
5451 #define texture_mask_height r3
5452 #define left_mask_bits r4
5453 #define right_mask_bits r5
5457 #undef block_masks_shifted
5460 #define block_masks d0
5461 #define block_masks_shifted d1
5462 #define draw_mask_fb_ptr d2
5465 #define draw_mask_fb_ptr_a d2
5466 #define draw_mask_fb_ptr_b d3
5467 #define texels_low d4
5468 #define texels_high d5
5469 #define texels_wide_low d6
5470 #define texels_wide_high d7
5471 #define texels_wide q3
5474 setup_sprites_16bpp_flush:
5477 stmdb sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5478 bl flush_render_block_buffer
5479 ldmia sp!, { r0 - r3, EXTRA_UNSAVED_REGS r12, r14 }
5483 add block, psx_gpu, #psx_gpu_blocks_offset
5484 mov num_blocks, block_width
5488 function(setup_sprite_16bpp)
5489 stmdb sp!, { r4 - r11, r14 }
5490 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5493 add fb_ptr, fb_ptr, y, lsl #11
5495 ldr width, [sp, #40]
5496 add fb_ptr, fb_ptr, x, lsl #1
5498 ldr height, [sp, #44]
5499 and left_offset, u, #0x7
5501 add texture_offset_base, u, u
5502 add width_rounded, width, #7
5504 add texture_offset_base, texture_offset_base, v, lsl #11
5505 mov left_mask_bits, #0xFF
5507 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5508 add width_rounded, width_rounded, left_offset
5510 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5511 sub fb_ptr, fb_ptr, left_offset, lsl #1
5513 add texture_mask, texture_mask_width, texture_mask_width
5514 mov right_mask_bits, #0xFE
5516 and right_width, width_rounded, #0x7
5517 mvn left_mask_bits, left_mask_bits, lsl left_offset
5519 add texture_mask, texture_mask, texture_mask_height, lsl #11
5520 mov block_width, width_rounded, lsr #3
5522 mov right_mask_bits, right_mask_bits, lsl right_width
5523 movw fb_ptr_pitch, #(2048 + 16)
5525 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4
5526 vmov block_masks, left_mask_bits, right_mask_bits
5528 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5529 add block, psx_gpu, #psx_gpu_blocks_offset
5531 bic texture_offset_base, texture_offset_base, #0xF
5534 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5535 add block, block, num_blocks, lsl #6
5539 vext.32 block_masks_shifted, block_masks, block_masks, #1
5540 vorr.u32 block_masks, block_masks, block_masks_shifted
5541 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5544 add num_blocks, num_blocks, #1
5545 cmp num_blocks, #MAX_BLOCKS
5546 blgt setup_sprites_16bpp_flush
5548 and texture_block_ptr, texture_offset_base, texture_mask
5549 subs height, height, #1
5551 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5552 vld1.u32 { texels }, [texture_block_ptr, :128]
5554 vst1.u32 { texels }, [block, :128]
5555 add block, block, #40
5557 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5560 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5562 add block, block, #24
5563 add texture_offset_base, texture_offset_base, #2048
5564 add fb_ptr, fb_ptr, #2048
5565 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5568 ldmia sp!, { r4 - r11, pc }
5571 add num_blocks, num_blocks, block_width
5572 mov texture_offset, texture_offset_base
5574 cmp num_blocks, #MAX_BLOCKS
5575 blgt setup_sprites_16bpp_flush
5577 add texture_offset_base, texture_offset_base, #2048
5578 and texture_block_ptr, texture_offset, texture_mask
5580 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5581 vld1.u32 { texels }, [texture_block_ptr, :128]
5583 vst1.u32 { texels }, [block, :128]
5584 add block, block, #40
5586 vdup.u8 draw_mask_fb_ptr, block_masks[0]
5587 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5590 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5591 subs blocks_remaining, block_width, #2
5593 add texture_offset, texture_offset, #16
5594 add fb_ptr, fb_ptr, #16
5596 vmov.u8 draw_mask_fb_ptr, #0
5598 add block, block, #24
5602 and texture_block_ptr, texture_offset, texture_mask
5603 subs blocks_remaining, blocks_remaining, #1
5605 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5606 vld1.u32 { texels }, [texture_block_ptr, :128]
5608 vst1.u32 { texels }, [block, :128]
5609 add block, block, #40
5611 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5614 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5616 add texture_offset, texture_offset, #16
5617 add fb_ptr, fb_ptr, #16
5619 add block, block, #24
5623 and texture_block_ptr, texture_offset, texture_mask
5624 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5626 vld1.u32 { texels }, [texture_block_ptr, :128]
5627 vdup.u8 draw_mask_fb_ptr, block_masks[4]
5629 vst1.u32 { texels }, [block, :128]
5630 add block, block, #40
5632 vmov.u32 draw_mask_fb_ptr[1], fb_ptr
5633 vst1.u32 { draw_mask_fb_ptr }, [block, :64]
5635 add block, block, #24
5636 subs height, height, #1
5638 add fb_ptr, fb_ptr, fb_ptr_pitch
5639 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5643 ldmia sp!, { r4 - r11, pc }
5647 // FIXME: duplicate code with normal version :(
5648 #undef draw_mask_fb_ptr
5650 function(setup_sprite_16bpp_4x)
5651 stmdb sp!, { r4 - r11, r14 }
5652 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5655 add fb_ptr, fb_ptr, y, lsl #11
5657 ldr width, [sp, #40]
5658 add fb_ptr, fb_ptr, x, lsl #1
5660 ldr height, [sp, #44]
5661 and left_offset, u, #0x7
5663 add texture_offset_base, u, u
5664 add width_rounded, width, #7
5666 add texture_offset_base, texture_offset_base, v, lsl #11
5667 movw left_mask_bits, #0xFFFF
5669 ldrb texture_mask_width, [psx_gpu, #psx_gpu_texture_mask_width_offset]
5670 add width_rounded, width_rounded, left_offset
5674 ldrb texture_mask_height, [psx_gpu, #psx_gpu_texture_mask_height_offset]
5675 sub fb_ptr, fb_ptr, left_offset, lsl #1
5677 add texture_mask, texture_mask_width, texture_mask_width
5678 movw right_mask_bits, #0xFFFC
5680 and right_width, width_rounded, #0x7
5681 mvn left_mask_bits, left_mask_bits, lsl left_offset
5685 add texture_mask, texture_mask, texture_mask_height, lsl #11
5686 mov block_width, width_rounded, lsr #3
5688 mov right_mask_bits, right_mask_bits, lsl right_width
5689 movw fb_ptr_pitch, #(2048 + 16) * 2
5691 sub fb_ptr_pitch, fb_ptr_pitch, block_width, lsl #4+1
5692 vmov block_masks, left_mask_bits, right_mask_bits
5694 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5695 add block, psx_gpu, #psx_gpu_blocks_offset
5697 bic texture_offset_base, texture_offset_base, #0xF
5700 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_ptr_offset]
5701 add block, block, num_blocks, lsl #6
5706 vext.32 block_masks_shifted, block_masks, block_masks, #1
5707 vorr.u32 block_masks, block_masks, block_masks_shifted
5708 vdup.u8 draw_mask_fb_ptr_a, block_masks[0]
5709 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5712 add num_blocks, num_blocks, block_width
5713 cmp num_blocks, #MAX_BLOCKS
5714 blgt setup_sprites_16bpp_flush
5716 and texture_block_ptr, texture_offset_base, texture_mask
5717 subs height, height, #1
5719 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5720 vld1.u32 { texels }, [texture_block_ptr, :128]
5722 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5724 add texture_offset_base, texture_offset_base, #2048
5725 add fb_ptr, fb_ptr, #2048*2
5726 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5729 ldmia sp!, { r4 - r11, pc }
5732 add num_blocks, num_blocks, block_width
5733 mov texture_offset, texture_offset_base
5735 vdup.u8 draw_mask_fb_ptr_a, block_masks[0] // left_mask_bits
5736 vdup.u8 draw_mask_fb_ptr_b, block_masks[1]
5738 cmp num_blocks, #MAX_BLOCKS
5739 blgt setup_sprites_16bpp_flush
5741 add texture_offset_base, texture_offset_base, #2048
5742 and texture_block_ptr, texture_offset, texture_mask
5744 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5745 vld1.u32 { texels }, [texture_block_ptr, :128]
5747 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5749 subs blocks_remaining, block_width, #2*4
5750 add texture_offset, texture_offset, #16
5752 vmov.u8 draw_mask_fb_ptr_a, #0
5753 vmov.u8 draw_mask_fb_ptr_b, #0
5755 add fb_ptr, fb_ptr, #16*2
5759 and texture_block_ptr, texture_offset, texture_mask
5760 subs blocks_remaining, blocks_remaining, #4
5762 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5763 vld1.u32 { texels }, [texture_block_ptr, :128]
5765 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5766 add texture_offset, texture_offset, #16
5768 add fb_ptr, fb_ptr, #16*2
5772 vdup.u8 draw_mask_fb_ptr_a, block_masks[4] // right_mask_bits
5773 vdup.u8 draw_mask_fb_ptr_b, block_masks[5]
5775 and texture_block_ptr, texture_offset, texture_mask
5776 add texture_block_ptr, texture_page_ptr, texture_block_ptr
5778 vld1.u32 { texels }, [texture_block_ptr, :128]
5780 do_texture_block_16bpp_4x(fb_ptr2, draw_mask_fb_ptr_a, draw_mask_fb_ptr_b)
5781 subs height, height, #1
5783 add fb_ptr, fb_ptr, fb_ptr_pitch
5784 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5788 ldmia sp!, { r4 - r11, pc }
5793 #undef right_mask_bits
5796 #undef blocks_remaining
5806 #define right_width r5
5807 #define right_mask_bits r6
5811 #define fb_ptr_pitch r12
5813 // referenced by setup_sprites_16bpp_flush
5814 #define num_blocks r4
5816 #define block_width r11
5821 #define blocks_remaining r6
5824 #define right_mask q1
5825 #define test_mask q2
5826 #define draw_mask q2
5827 #define draw_mask_bits_fb_ptr d6
5832 function(setup_sprite_untextured_512)
5833 stmdb sp!, { r4 - r11, r14 }
5835 ldr width, [sp, #40]
5836 ldr fb_ptr, [psx_gpu, #psx_gpu_vram_out_ptr_offset]
5838 ldr height, [sp, #44]
5839 add fb_ptr, fb_ptr, y, lsl #11
5841 add fb_ptr, fb_ptr, x, lsl #1
5842 sub right_width, width, #1
5844 ldr color, [sp, #48]
5847 add block_width, width, #7
5851 mov right_mask_bits, #0xff
5853 sub fb_ptr_pitch, block_width, #1
5854 lsl right_mask_bits, right_width
5856 lsl fb_ptr_pitch, #3+1
5857 ubfx color_r, color, #3, #5
5859 rsb fb_ptr_pitch, #1024*2
5860 ubfx color_g, color, #11, #5
5862 vld1.u32 { test_mask }, [psx_gpu, :128]
5863 ubfx color_b, color, #19, #5
5865 vdup.u16 right_mask, right_mask_bits
5866 orr color, color_r, color_b, lsl #10
5868 ldrh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5869 orr color, color, color_g, lsl #5
5871 vtst.u16 right_mask, right_mask, test_mask
5872 add block, psx_gpu, #psx_gpu_blocks_offset
5874 vdup.u16 colors, color
5875 add block, block, num_blocks, lsl #6
5878 setup_sprite_untextured_height_loop:
5879 add num_blocks, block_width
5880 sub blocks_remaining, block_width, #1
5882 cmp num_blocks, #MAX_BLOCKS
5883 blgt setup_sprites_16bpp_flush
5885 cmp blocks_remaining, #0
5888 vmov.u8 draw_mask, #0 /* zero_mask */
5889 vmov.u8 draw_mask_bits_fb_ptr, #0
5892 vst1.u32 { draw_mask }, [block, :128]!
5893 subs blocks_remaining, #1
5895 vst1.u32 { colors }, [block, :128]
5896 add block, block, #24
5898 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5899 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5901 add block, block, #24
5906 vst1.u32 { right_mask }, [block, :128]!
5909 vst1.u32 { colors }, [block, :128]
5910 add block, block, #24
5912 vmov.u32 draw_mask_bits_fb_ptr[1], fb_ptr
5913 vst1.u32 { draw_mask_bits_fb_ptr }, [block, :64]
5915 add block, block, #24
5916 add fb_ptr, fb_ptr_pitch
5918 strh num_blocks, [psx_gpu, #psx_gpu_num_blocks_offset]
5919 bgt setup_sprite_untextured_height_loop
5921 ldmia sp!, { r4 - r11, pc }
5925 #undef texture_page_ptr
5927 #undef dirty_textures_mask
5928 #undef current_texture_mask
5931 #define current_texture_page r1
5932 #define texture_page_ptr r2
5933 #define vram_ptr_a r3
5934 #define current_texture_page_x r12
5935 #define current_texture_page_y r4
5936 #define dirty_textures_mask r5
5940 #define current_texture_mask r9
5942 #define vram_ptr_b r11
5944 #define texel_block_a d0
5945 #define texel_block_b d1
5946 #define texel_block_expanded_a q1
5947 #define texel_block_expanded_b q2
5948 #define texel_block_expanded_ab q2
5949 #define texel_block_expanded_c q3
5950 #define texel_block_expanded_d q0
5951 #define texel_block_expanded_cd q3
5953 function(update_texture_4bpp_cache)
5954 stmdb sp!, { r4 - r11, r14 }
5957 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
5959 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
5960 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
5962 and current_texture_page_x, current_texture_page, #0xF
5963 ldr current_texture_mask, [psx_gpu, #psx_gpu_current_texture_mask_offset]
5965 mov current_texture_page_y, current_texture_page, lsr #4
5966 ldr dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5968 add vram_ptr_a, vram_ptr_a, current_texture_page_y, lsl #19
5971 add vram_ptr_a, vram_ptr_a, current_texture_page_x, lsl #7
5972 bic dirty_textures_mask, current_texture_mask
5975 str dirty_textures_mask, [psx_gpu, #psx_gpu_dirty_textures_4bpp_mask_offset]
5980 add vram_ptr_b, vram_ptr_a, #2048
5983 vld1.u32 { texel_block_a }, [vram_ptr_a, :64], c_4096
5984 vld1.u32 { texel_block_b }, [vram_ptr_b, :64], c_4096
5986 vmovl.u8 texel_block_expanded_a, texel_block_a
5987 vshll.u8 texel_block_expanded_b, texel_block_a, #4
5988 vmovl.u8 texel_block_expanded_c, texel_block_b
5989 vshll.u8 texel_block_expanded_d, texel_block_b, #4
5991 vbic.u16 texel_block_expanded_a, #0x00F0
5992 vbic.u16 texel_block_expanded_b, #0x00F0
5993 vbic.u16 texel_block_expanded_c, #0x00F0
5994 vbic.u16 texel_block_expanded_d, #0x00F0
5996 vorr.u16 texel_block_expanded_ab, texel_block_expanded_a, \
5997 texel_block_expanded_b
5998 vorr.u16 texel_block_expanded_cd, texel_block_expanded_c, \
5999 texel_block_expanded_d
6001 vst1.u32 { texel_block_expanded_ab, texel_block_expanded_cd }, \
6002 [texture_page_ptr, :256]!
6004 subs sub_y, sub_y, #1
6008 add vram_ptr_a, vram_ptr_a, #8
6009 add vram_ptr_b, vram_ptr_b, #8
6011 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6012 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6014 subs tile_x, tile_x, #1
6018 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6019 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6021 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6022 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6024 subs tile_y, tile_y, #1
6028 ldmia sp!, { r4 - r11, pc }
6031 #undef current_texture_page
6034 #define texture_page r1
6035 #define texture_page_ptr r2
6036 #define vram_ptr_a r3
6037 #define texture_page_x r12
6038 #define texture_page_y r4
6039 #define current_texture_page r5
6044 #define vram_ptr_b r11
6056 function(update_texture_8bpp_cache_slice)
6057 stmdb sp!, { r4 - r11, r14 }
6060 ldrb current_texture_page, [psx_gpu, #psx_gpu_current_texture_page_offset]
6061 ldr vram_ptr_a, [psx_gpu, #psx_gpu_vram_ptr_offset]
6063 ldr texture_page_ptr, [psx_gpu, #psx_gpu_texture_page_base_offset]
6066 and texture_page_x, texture_page, #0xF
6067 mov texture_page_y, texture_page, lsr #4
6069 add vram_ptr_a, vram_ptr_a, texture_page_x, lsl #7
6072 add vram_ptr_a, vram_ptr_a, texture_page_y, lsl #19
6073 eor current_texture_page, current_texture_page, texture_page
6075 ands current_texture_page, current_texture_page, #0x1
6078 addne texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6081 add vram_ptr_b, vram_ptr_a, #2048
6084 vld1.u32 { texels_a }, [vram_ptr_a, :128], c_4096
6085 vld1.u32 { texels_b }, [vram_ptr_b, :128], c_4096
6086 vld1.u32 { texels_c }, [vram_ptr_a, :128], c_4096
6087 vld1.u32 { texels_d }, [vram_ptr_b, :128], c_4096
6089 vst1.u32 { texels_a, texels_b }, [texture_page_ptr, :256]!
6090 vst1.u32 { texels_c, texels_d }, [texture_page_ptr, :256]!
6092 subs sub_y, sub_y, #1
6097 add vram_ptr_a, vram_ptr_a, #16
6098 add vram_ptr_b, vram_ptr_b, #16
6100 sub vram_ptr_a, vram_ptr_a, #(16 * 2048)
6101 sub vram_ptr_b, vram_ptr_b, #(16 * 2048)
6103 subs tile_x, tile_x, #1
6108 add vram_ptr_a, vram_ptr_a, #(16 * 2048)
6109 add vram_ptr_b, vram_ptr_b, #(16 * 2048)
6111 sub vram_ptr_a, vram_ptr_a, #(8 * 16)
6112 sub vram_ptr_b, vram_ptr_b, #(8 * 16)
6114 subs tile_y, tile_y, #1
6115 add texture_page_ptr, texture_page_ptr, #(8 * 16 * 16)
6120 ldmia sp!, { r4 - r11, pc }
6123 /* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
6124 function(scale2x_tiles8)
6128 add r12, r0, #1024*2
6133 vld1.u16 { q0 }, [r1, :128]!
6134 vld1.u16 { q2 }, [r1, :128]!
6140 vst1.u16 { q0, q1 }, [r0, :128]!
6141 vst1.u16 { q0, q1 }, [r12, :128]!
6143 vst1.u16 { q2, q3 }, [r0, :128]!
6144 vst1.u16 { q2, q3 }, [r12, :128]!
6151 sub r0, r0, r2, lsl #4+1
6153 add r12, r0, #1024*2
6159 // vim:filetype=armasm