1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
|
// Copyright 2018 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <map>
#include <optional>
#include <set>
#include <string>
#include <string_view>
#include <unordered_set>
#include <fmt/format.h>
#include "common/assert.h"
#include "common/common_types.h"
#include "video_core/engines/shader_bytecode.h"
#include "video_core/engines/shader_header.h"
#include "video_core/renderer_opengl/gl_rasterizer.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
namespace OpenGL::GLShader::Decompiler {
using Tegra::Shader::Attribute;
using Tegra::Shader::Instruction;
using Tegra::Shader::LogicOperation;
using Tegra::Shader::OpCode;
using Tegra::Shader::Register;
using Tegra::Shader::Sampler;
using Tegra::Shader::SubOp;
constexpr u32 PROGRAM_END = MAX_PROGRAM_CODE_LENGTH;
constexpr u32 PROGRAM_HEADER_SIZE = sizeof(Tegra::Shader::Header);
constexpr u32 MAX_GEOMETRY_BUFFERS = 6;
constexpr u32 MAX_ATTRIBUTES = 0x100; // Size in vec4s, this value is untested
static const char* INTERNAL_FLAG_NAMES[] = {"zero_flag", "sign_flag", "carry_flag",
"overflow_flag"};
enum class InternalFlag : u64 {
ZeroFlag = 0,
SignFlag = 1,
CarryFlag = 2,
OverflowFlag = 3,
Amount
};
class DecompileFail : public std::runtime_error {
public:
using std::runtime_error::runtime_error;
};
/// Translate topology
static std::string GetTopologyName(Tegra::Shader::OutputTopology topology) {
switch (topology) {
case Tegra::Shader::OutputTopology::PointList:
return "points";
case Tegra::Shader::OutputTopology::LineStrip:
return "line_strip";
case Tegra::Shader::OutputTopology::TriangleStrip:
return "triangle_strip";
default:
UNIMPLEMENTED_MSG("Unknown output topology: {}", static_cast<u32>(topology));
return "points";
}
}
/// Describes the behaviour of code path of a given entry point and a return point.
enum class ExitMethod {
Undetermined, ///< Internal value. Only occur when analyzing JMP loop.
AlwaysReturn, ///< All code paths reach the return point.
Conditional, ///< Code path reaches the return point or an END instruction conditionally.
AlwaysEnd, ///< All code paths reach a END instruction.
};
/// A subroutine is a range of code refereced by a CALL, IF or LOOP instruction.
struct Subroutine {
/// Generates a name suitable for GLSL source code.
std::string GetName() const {
return "sub_" + std::to_string(begin) + '_' + std::to_string(end) + '_' + suffix;
}
u32 begin; ///< Entry point of the subroutine.
u32 end; ///< Return point of the subroutine.
const std::string& suffix; ///< Suffix of the shader, used to make a unique subroutine name
ExitMethod exit_method; ///< Exit method of the subroutine.
std::set<u32> labels; ///< Addresses refereced by JMP instructions.
bool operator<(const Subroutine& rhs) const {
return std::tie(begin, end) < std::tie(rhs.begin, rhs.end);
}
};
/// Analyzes shader code and produces a set of subroutines.
class ControlFlowAnalyzer {
public:
ControlFlowAnalyzer(const ProgramCode& program_code, u32 main_offset, const std::string& suffix)
: program_code(program_code), shader_coverage_begin(main_offset),
shader_coverage_end(main_offset + 1) {
// Recursively finds all subroutines.
const Subroutine& program_main = AddSubroutine(main_offset, PROGRAM_END, suffix);
if (program_main.exit_method != ExitMethod::AlwaysEnd)
throw DecompileFail("Program does not always end");
}
std::set<Subroutine> GetSubroutines() {
return std::move(subroutines);
}
std::size_t GetShaderLength() const {
return shader_coverage_end * sizeof(u64);
}
private:
const ProgramCode& program_code;
std::set<Subroutine> subroutines;
std::map<std::pair<u32, u32>, ExitMethod> exit_method_map;
u32 shader_coverage_begin;
u32 shader_coverage_end;
/// Adds and analyzes a new subroutine if it is not added yet.
const Subroutine& AddSubroutine(u32 begin, u32 end, const std::string& suffix) {
Subroutine subroutine{begin, end, suffix, ExitMethod::Undetermined, {}};
const auto iter = subroutines.find(subroutine);
if (iter != subroutines.end()) {
return *iter;
}
subroutine.exit_method = Scan(begin, end, subroutine.labels);
if (subroutine.exit_method == ExitMethod::Undetermined) {
throw DecompileFail("Recursive function detected");
}
return *subroutines.insert(std::move(subroutine)).first;
}
/// Merges exit method of two parallel branches.
static ExitMethod ParallelExit(ExitMethod a, ExitMethod b) {
if (a == ExitMethod::Undetermined) {
return b;
}
if (b == ExitMethod::Undetermined) {
return a;
}
if (a == b) {
return a;
}
return ExitMethod::Conditional;
}
/// Scans a range of code for labels and determines the exit method.
ExitMethod Scan(u32 begin, u32 end, std::set<u32>& labels) {
const auto [iter, inserted] =
exit_method_map.emplace(std::make_pair(begin, end), ExitMethod::Undetermined);
ExitMethod& exit_method = iter->second;
if (!inserted)
return exit_method;
for (u32 offset = begin; offset != end && offset != PROGRAM_END; ++offset) {
shader_coverage_begin = std::min(shader_coverage_begin, offset);
shader_coverage_end = std::max(shader_coverage_end, offset + 1);
const Instruction instr = {program_code[offset]};
if (const auto opcode = OpCode::Decode(instr)) {
switch (opcode->get().GetId()) {
case OpCode::Id::EXIT: {
// The EXIT instruction can be predicated, which means that the shader can
// conditionally end on this instruction. We have to consider the case where the
// condition is not met and check the exit method of that other basic block.
using Tegra::Shader::Pred;
if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
return exit_method = ExitMethod::AlwaysEnd;
} else {
const ExitMethod not_met = Scan(offset + 1, end, labels);
return exit_method = ParallelExit(ExitMethod::AlwaysEnd, not_met);
}
}
case OpCode::Id::BRA: {
const u32 target = offset + instr.bra.GetBranchTarget();
labels.insert(target);
const ExitMethod no_jmp = Scan(offset + 1, end, labels);
const ExitMethod jmp = Scan(target, end, labels);
return exit_method = ParallelExit(no_jmp, jmp);
}
case OpCode::Id::SSY:
case OpCode::Id::PBK: {
// The SSY and PBK use a similar encoding as the BRA instruction.
UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
"Constant buffer branching is not supported");
const u32 target = offset + instr.bra.GetBranchTarget();
labels.insert(target);
// Continue scanning for an exit method.
break;
}
}
}
}
return exit_method = ExitMethod::AlwaysReturn;
}
};
class ShaderWriter {
public:
void AddLine(std::string_view text) {
DEBUG_ASSERT(scope >= 0);
if (!text.empty()) {
AppendIndentation();
}
shader_source += text;
AddNewLine();
}
void AddLine(char character) {
DEBUG_ASSERT(scope >= 0);
AppendIndentation();
shader_source += character;
AddNewLine();
}
void AddNewLine() {
DEBUG_ASSERT(scope >= 0);
shader_source += '\n';
}
std::string GetResult() {
return std::move(shader_source);
}
int scope = 0;
private:
void AppendIndentation() {
shader_source.append(static_cast<std::size_t>(scope) * 4, ' ');
}
std::string shader_source;
};
/**
* Represents an emulated shader register, used to track the state of that register for emulation
* with GLSL. At this time, a register can be used as a float or an integer. This class is used for
* bookkeeping within the GLSL program.
*/
class GLSLRegister {
public:
enum class Type {
Float,
Integer,
UnsignedInteger,
};
GLSLRegister(std::size_t index, const std::string& suffix) : index{index}, suffix{suffix} {}
/// Gets the GLSL type string for a register
static std::string GetTypeString() {
return "float";
}
/// Gets the GLSL register prefix string, used for declarations and referencing
static std::string GetPrefixString() {
return "reg_";
}
/// Returns a GLSL string representing the current state of the register
std::string GetString() const {
return GetPrefixString() + std::to_string(index) + '_' + suffix;
}
/// Returns the index of the register
std::size_t GetIndex() const {
return index;
}
private:
const std::size_t index;
const std::string& suffix;
};
/**
* Used to manage shader registers that are emulated with GLSL. This class keeps track of the state
* of all registers (e.g. whether they are currently being used as Floats or Integers), and
* generates the necessary GLSL code to perform conversions as needed. This class is used for
* bookkeeping within the GLSL program.
*/
class GLSLRegisterManager {
public:
GLSLRegisterManager(ShaderWriter& shader, ShaderWriter& declarations,
const Maxwell3D::Regs::ShaderStage& stage, const std::string& suffix,
const Tegra::Shader::Header& header)
: shader{shader}, declarations{declarations}, stage{stage}, suffix{suffix}, header{header},
fixed_pipeline_output_attributes_used{}, local_memory_size{0} {
BuildRegisterList();
BuildInputList();
}
/**
* Returns code that does an integer size conversion for the specified size.
* @param value Value to perform integer size conversion on.
* @param size Register size to use for conversion instructions.
* @returns GLSL string corresponding to the value converted to the specified size.
*/
static std::string ConvertIntegerSize(const std::string& value, Register::Size size) {
switch (size) {
case Register::Size::Byte:
return "((" + value + " << 24) >> 24)";
case Register::Size::Short:
return "((" + value + " << 16) >> 16)";
case Register::Size::Word:
// Default - do nothing
return value;
default:
UNIMPLEMENTED_MSG("Unimplemented conversion size: {}", static_cast<u32>(size));
}
}
/**
* Gets a register as an float.
* @param reg The register to get.
* @param elem The element to use for the operation.
* @returns GLSL string corresponding to the register as a float.
*/
std::string GetRegisterAsFloat(const Register& reg, unsigned elem = 0) {
return GetRegister(reg, elem);
}
/**
* Gets a register as an integer.
* @param reg The register to get.
* @param elem The element to use for the operation.
* @param is_signed Whether to get the register as a signed (or unsigned) integer.
* @param size Register size to use for conversion instructions.
* @returns GLSL string corresponding to the register as an integer.
*/
std::string GetRegisterAsInteger(const Register& reg, unsigned elem = 0, bool is_signed = true,
Register::Size size = Register::Size::Word) {
const std::string func{is_signed ? "floatBitsToInt" : "floatBitsToUint"};
const std::string value{func + '(' + GetRegister(reg, elem) + ')'};
return ConvertIntegerSize(value, size);
}
/**
* Writes code that does a register assignment to float value operation.
* @param reg The destination register to use.
* @param elem The element to use for the operation.
* @param value The code representing the value to assign.
* @param dest_num_components Number of components in the destination.
* @param value_num_components Number of components in the value.
* @param is_saturated Optional, when True, saturates the provided value.
* @param dest_elem Optional, the destination element to use for the operation.
*/
void SetRegisterToFloat(const Register& reg, u64 elem, const std::string& value,
u64 dest_num_components, u64 value_num_components,
bool is_saturated = false, u64 dest_elem = 0, bool precise = false) {
SetRegister(reg, elem, is_saturated ? "clamp(" + value + ", 0.0, 1.0)" : value,
dest_num_components, value_num_components, dest_elem, precise);
}
/**
* Writes code that does a register assignment to integer value operation.
* @param reg The destination register to use.
* @param elem The element to use for the operation.
* @param value The code representing the value to assign.
* @param dest_num_components Number of components in the destination.
* @param value_num_components Number of components in the value.
* @param is_saturated Optional, when True, saturates the provided value.
* @param dest_elem Optional, the destination element to use for the operation.
* @param size Register size to use for conversion instructions.
*/
void SetRegisterToInteger(const Register& reg, bool is_signed, u64 elem,
const std::string& value, u64 dest_num_components,
u64 value_num_components, bool is_saturated = false,
u64 dest_elem = 0, Register::Size size = Register::Size::Word,
bool sets_cc = false) {
UNIMPLEMENTED_IF(is_saturated);
const std::string func{is_signed ? "intBitsToFloat" : "uintBitsToFloat"};
SetRegister(reg, elem, func + '(' + ConvertIntegerSize(value, size) + ')',
dest_num_components, value_num_components, dest_elem, false);
if (sets_cc) {
const std::string zero_condition = "( " + ConvertIntegerSize(value, size) + " == 0 )";
SetInternalFlag(InternalFlag::ZeroFlag, zero_condition);
LOG_WARNING(HW_GPU, "Condition codes implementation is incomplete.");
}
}
/**
* Writes code that does a register assignment to a half float value operation.
* @param reg The destination register to use.
* @param elem The element to use for the operation.
* @param value The code representing the value to assign. Type has to be half float.
* @param merge Half float kind of assignment.
* @param dest_num_components Number of components in the destination.
* @param value_num_components Number of components in the value.
* @param is_saturated Optional, when True, saturates the provided value.
* @param dest_elem Optional, the destination element to use for the operation.
*/
void SetRegisterToHalfFloat(const Register& reg, u64 elem, const std::string& value,
Tegra::Shader::HalfMerge merge, u64 dest_num_components,
u64 value_num_components, bool is_saturated = false,
u64 dest_elem = 0) {
UNIMPLEMENTED_IF(is_saturated);
const std::string result = [&]() {
switch (merge) {
case Tegra::Shader::HalfMerge::H0_H1:
return "uintBitsToFloat(packHalf2x16(" + value + "))";
case Tegra::Shader::HalfMerge::F32:
// Half float instructions take the first component when doing a float cast.
return "float(" + value + ".x)";
case Tegra::Shader::HalfMerge::Mrg_H0:
// TODO(Rodrigo): I guess Mrg_H0 and Mrg_H1 take their respective component from the
// pack. I couldn't test this on hardware but it shouldn't really matter since most
// of the time when a Mrg_* flag is used both components will be mirrored. That
// being said, it deserves a test.
return "((" + GetRegisterAsInteger(reg, 0, false) +
" & 0xffff0000) | (packHalf2x16(" + value + ") & 0x0000ffff))";
case Tegra::Shader::HalfMerge::Mrg_H1:
return "((" + GetRegisterAsInteger(reg, 0, false) +
" & 0x0000ffff) | (packHalf2x16(" + value + ") & 0xffff0000))";
default:
UNREACHABLE();
return std::string("0");
}
}();
SetRegister(reg, elem, result, dest_num_components, value_num_components, dest_elem, false);
}
/**
* Writes code that does a register assignment to input attribute operation. Input attributes
* are stored as floats, so this may require conversion.
* @param reg The destination register to use.
* @param elem The element to use for the operation.
* @param attribute The input attribute to use as the source value.
* @param input_mode The input mode.
* @param vertex The register that decides which vertex to read from (used in GS).
*/
void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute,
const Tegra::Shader::IpaMode& input_mode,
std::optional<Register> vertex = {}) {
const std::string dest = GetRegisterAsFloat(reg);
const std::string src = GetInputAttribute(attribute, input_mode, vertex) + GetSwizzle(elem);
shader.AddLine(dest + " = " + src + ';');
}
std::string GetLocalMemoryAsFloat(const std::string& index) {
return "lmem[" + index + ']';
}
std::string GetLocalMemoryAsInteger(const std::string& index, bool is_signed = false) {
const std::string func{is_signed ? "floatToIntBits" : "floatBitsToUint"};
return func + "(lmem[" + index + "])";
}
void SetLocalMemoryAsFloat(const std::string& index, const std::string& value) {
shader.AddLine("lmem[" + index + "] = " + value + ';');
}
void SetLocalMemoryAsInteger(const std::string& index, const std::string& value,
bool is_signed = false) {
const std::string func{is_signed ? "intBitsToFloat" : "uintBitsToFloat"};
shader.AddLine("lmem[" + index + "] = " + func + '(' + value + ");");
}
std::string GetConditionCode(const Tegra::Shader::ConditionCode cc) const {
switch (cc) {
case Tegra::Shader::ConditionCode::NEU:
return "!(" + GetInternalFlag(InternalFlag::ZeroFlag) + ')';
default:
UNIMPLEMENTED_MSG("Unimplemented condition code: {}", static_cast<u32>(cc));
return "false";
}
}
std::string GetInternalFlag(const InternalFlag flag) const {
const auto index = static_cast<u32>(flag);
ASSERT(index < static_cast<u32>(InternalFlag::Amount));
return std::string(INTERNAL_FLAG_NAMES[index]) + '_' + suffix;
}
void SetInternalFlag(const InternalFlag flag, const std::string& value) const {
shader.AddLine(GetInternalFlag(flag) + " = " + value + ';');
}
/**
* Writes code that does a output attribute assignment to register operation. Output attributes
* are stored as floats, so this may require conversion.
* @param attribute The destination output attribute.
* @param elem The element to use for the operation.
* @param val_reg The register to use as the source value.
* @param buf_reg The register that tells which buffer to write to (used in geometry shaders).
*/
void SetOutputAttributeToRegister(Attribute::Index attribute, u64 elem, const Register& val_reg,
const Register& buf_reg) {
const std::string dest = GetOutputAttribute(attribute);
const std::string src = GetRegisterAsFloat(val_reg);
if (dest.empty())
return;
// Can happen with unknown/unimplemented output attributes, in which case we ignore the
// instruction for now.
if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
// TODO(Rodrigo): nouveau sets some attributes after setting emitting a geometry
// shader. These instructions use a dirty register as buffer index, to avoid some
// drivers from complaining about out of boundary writes, guard them.
const std::string buf_index{"((" + GetRegisterAsInteger(buf_reg) + ") % " +
std::to_string(MAX_GEOMETRY_BUFFERS) + ')'};
shader.AddLine("amem[" + buf_index + "][" +
std::to_string(static_cast<u32>(attribute)) + ']' + GetSwizzle(elem) +
" = " + src + ';');
return;
}
switch (attribute) {
case Attribute::Index::ClipDistances0123:
case Attribute::Index::ClipDistances4567: {
const u64 index = attribute == Attribute::Index::ClipDistances4567 ? 4 : 0 + elem;
UNIMPLEMENTED_IF_MSG(
((header.vtg.clip_distances >> index) & 1) == 0,
"Shader is setting gl_ClipDistance{} without enabling it in the header", index);
fixed_pipeline_output_attributes_used.insert(attribute);
shader.AddLine(dest + '[' + std::to_string(index) + "] = " + src + ';');
break;
}
case Attribute::Index::PointSize:
fixed_pipeline_output_attributes_used.insert(attribute);
shader.AddLine(dest + " = " + src + ';');
break;
default:
shader.AddLine(dest + GetSwizzle(elem) + " = " + src + ';');
break;
}
}
/// Generates code representing a uniform (C buffer) register, interpreted as the input type.
std::string GetUniform(u64 index, u64 offset, GLSLRegister::Type type,
Register::Size size = Register::Size::Word) {
declr_const_buffers[index].MarkAsUsed(index, offset, stage);
std::string value = 'c' + std::to_string(index) + '[' + std::to_string(offset / 4) + "][" +
std::to_string(offset % 4) + ']';
if (type == GLSLRegister::Type::Float) {
// Do nothing, default
} else if (type == GLSLRegister::Type::Integer) {
value = "floatBitsToInt(" + value + ')';
} else if (type == GLSLRegister::Type::UnsignedInteger) {
value = "floatBitsToUint(" + value + ')';
} else {
UNREACHABLE();
}
return ConvertIntegerSize(value, size);
}
std::string GetUniformIndirect(u64 cbuf_index, s64 offset, const std::string& index_str,
GLSLRegister::Type type) {
declr_const_buffers[cbuf_index].MarkAsUsedIndirect(cbuf_index, stage);
const std::string final_offset = fmt::format("({} + {})", index_str, offset / 4);
const std::string value = 'c' + std::to_string(cbuf_index) + '[' + final_offset + " / 4][" +
final_offset + " % 4]";
if (type == GLSLRegister::Type::Float) {
return value;
} else if (type == GLSLRegister::Type::Integer) {
return "floatBitsToInt(" + value + ')';
} else {
UNREACHABLE();
}
}
/// Add declarations.
void GenerateDeclarations(const std::string& suffix) {
GenerateVertex();
GenerateRegisters(suffix);
GenerateLocalMemory();
GenerateInternalFlags();
GenerateInputAttrs();
GenerateOutputAttrs();
GenerateConstBuffers();
GenerateSamplers();
GenerateGeometry();
}
/// Returns a list of constant buffer declarations.
std::vector<ConstBufferEntry> GetConstBuffersDeclarations() const {
std::vector<ConstBufferEntry> result;
std::copy_if(declr_const_buffers.begin(), declr_const_buffers.end(),
std::back_inserter(result), [](const auto& entry) { return entry.IsUsed(); });
return result;
}
/// Returns a list of samplers used in the shader.
const std::vector<SamplerEntry>& GetSamplers() const {
return used_samplers;
}
/// Returns the GLSL sampler used for the input shader sampler, and creates a new one if
/// necessary.
std::string AccessSampler(const Sampler& sampler, Tegra::Shader::TextureType type,
bool is_array, bool is_shadow) {
const auto offset = static_cast<std::size_t>(sampler.index.Value());
// If this sampler has already been used, return the existing mapping.
const auto itr =
std::find_if(used_samplers.begin(), used_samplers.end(),
[&](const SamplerEntry& entry) { return entry.GetOffset() == offset; });
if (itr != used_samplers.end()) {
ASSERT(itr->GetType() == type && itr->IsArray() == is_array &&
itr->IsShadow() == is_shadow);
return itr->GetName();
}
// Otherwise create a new mapping for this sampler
const std::size_t next_index = used_samplers.size();
const SamplerEntry entry{stage, offset, next_index, type, is_array, is_shadow};
used_samplers.emplace_back(entry);
return entry.GetName();
}
void SetLocalMemory(u64 lmem) {
local_memory_size = lmem;
}
private:
/// Generates declarations for registers.
void GenerateRegisters(const std::string& suffix) {
for (const auto& reg : regs) {
declarations.AddLine(GLSLRegister::GetTypeString() + ' ' + reg.GetPrefixString() +
std::to_string(reg.GetIndex()) + '_' + suffix + " = 0;");
}
declarations.AddNewLine();
}
/// Generates declarations for local memory.
void GenerateLocalMemory() {
if (local_memory_size > 0) {
declarations.AddLine("float lmem[" + std::to_string((local_memory_size - 1 + 4) / 4) +
"];");
declarations.AddNewLine();
}
}
/// Generates declarations for internal flags.
void GenerateInternalFlags() {
for (u32 flag = 0; flag < static_cast<u32>(InternalFlag::Amount); flag++) {
const InternalFlag code = static_cast<InternalFlag>(flag);
declarations.AddLine("bool " + GetInternalFlag(code) + " = false;");
}
declarations.AddNewLine();
}
/// Generates declarations for input attributes.
void GenerateInputAttrs() {
for (const auto element : declr_input_attribute) {
// TODO(bunnei): Use proper number of elements for these
u32 idx =
static_cast<u32>(element.first) - static_cast<u32>(Attribute::Index::Attribute_0);
if (stage != Maxwell3D::Regs::ShaderStage::Vertex) {
// If inputs are varyings, add an offset
idx += GENERIC_VARYING_START_LOCATION;
}
std::string attr{GetInputAttribute(element.first, element.second)};
if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
attr = "gs_" + attr + "[]";
}
declarations.AddLine("layout (location = " + std::to_string(idx) + ") " +
GetInputFlags(element.first) + "in vec4 " + attr + ';');
}
declarations.AddNewLine();
}
/// Generates declarations for output attributes.
void GenerateOutputAttrs() {
for (const auto& index : declr_output_attribute) {
// TODO(bunnei): Use proper number of elements for these
const u32 idx = static_cast<u32>(index) -
static_cast<u32>(Attribute::Index::Attribute_0) +
GENERIC_VARYING_START_LOCATION;
declarations.AddLine("layout (location = " + std::to_string(idx) + ") out vec4 " +
GetOutputAttribute(index) + ';');
}
declarations.AddNewLine();
}
/// Generates declarations for constant buffers.
void GenerateConstBuffers() {
for (const auto& entry : GetConstBuffersDeclarations()) {
declarations.AddLine("layout (std140) uniform " + entry.GetName());
declarations.AddLine('{');
declarations.AddLine(" vec4 c" + std::to_string(entry.GetIndex()) +
"[MAX_CONSTBUFFER_ELEMENTS];");
declarations.AddLine("};");
declarations.AddNewLine();
}
declarations.AddNewLine();
}
/// Generates declarations for samplers.
void GenerateSamplers() {
const auto& samplers = GetSamplers();
for (const auto& sampler : samplers) {
declarations.AddLine("uniform " + sampler.GetTypeString() + ' ' + sampler.GetName() +
';');
}
declarations.AddNewLine();
}
/// Generates declarations used for geometry shaders.
void GenerateGeometry() {
if (stage != Maxwell3D::Regs::ShaderStage::Geometry)
return;
declarations.AddLine(
"layout (" + GetTopologyName(header.common3.output_topology) +
", max_vertices = " + std::to_string(header.common4.max_output_vertices) + ") out;");
declarations.AddNewLine();
declarations.AddLine("vec4 amem[" + std::to_string(MAX_GEOMETRY_BUFFERS) + "][" +
std::to_string(MAX_ATTRIBUTES) + "];");
declarations.AddNewLine();
constexpr char buffer[] = "amem[output_buffer]";
declarations.AddLine("void emit_vertex(uint output_buffer) {");
++declarations.scope;
for (const auto element : declr_output_attribute) {
declarations.AddLine(GetOutputAttribute(element) + " = " + buffer + '[' +
std::to_string(static_cast<u32>(element)) + "];");
}
declarations.AddLine("position = " + std::string(buffer) + '[' +
std::to_string(static_cast<u32>(Attribute::Index::Position)) + "];");
// If a geometry shader is attached, it will always flip (it's the last stage before
// fragment). For more info about flipping, refer to gl_shader_gen.cpp.
declarations.AddLine("position.xy *= viewport_flip.xy;");
declarations.AddLine("gl_Position = position;");
declarations.AddLine("position.w = 1.0;");
declarations.AddLine("EmitVertex();");
--declarations.scope;
declarations.AddLine('}');
declarations.AddNewLine();
}
void GenerateVertex() {
if (stage != Maxwell3D::Regs::ShaderStage::Vertex)
return;
bool clip_distances_declared = false;
declarations.AddLine("out gl_PerVertex {");
++declarations.scope;
declarations.AddLine("vec4 gl_Position;");
for (auto& o : fixed_pipeline_output_attributes_used) {
if (o == Attribute::Index::PointSize)
declarations.AddLine("float gl_PointSize;");
if (!clip_distances_declared && (o == Attribute::Index::ClipDistances0123 ||
o == Attribute::Index::ClipDistances4567)) {
declarations.AddLine("float gl_ClipDistance[];");
clip_distances_declared = true;
}
}
--declarations.scope;
declarations.AddLine("};");
}
/// Generates code representing a temporary (GPR) register.
std::string GetRegister(const Register& reg, unsigned elem) {
if (reg == Register::ZeroIndex) {
return "0";
}
return regs[reg.GetSwizzledIndex(elem)].GetString();
}
/**
* Writes code that does a register assignment to value operation.
* @param reg The destination register to use.
* @param elem The element to use for the operation.
* @param value The code representing the value to assign.
* @param dest_num_components Number of components in the destination.
* @param value_num_components Number of components in the value.
* @param dest_elem Optional, the destination element to use for the operation.
*/
void SetRegister(const Register& reg, u64 elem, const std::string& value,
u64 dest_num_components, u64 value_num_components, u64 dest_elem,
bool precise) {
if (reg == Register::ZeroIndex) {
// Setting RZ is a nop in hardware.
return;
}
std::string dest = GetRegister(reg, static_cast<u32>(dest_elem));
if (dest_num_components > 1) {
dest += GetSwizzle(elem);
}
std::string src = '(' + value + ')';
if (value_num_components > 1) {
src += GetSwizzle(elem);
}
if (precise && stage != Maxwell3D::Regs::ShaderStage::Fragment) {
shader.AddLine('{');
++shader.scope;
// This avoids optimizations of constant propagation and keeps the code as the original
// Sadly using the precise keyword causes "linking" errors on fragment shaders.
shader.AddLine("precise float tmp = " + src + ';');
shader.AddLine(dest + " = tmp;");
--shader.scope;
shader.AddLine('}');
} else {
shader.AddLine(dest + " = " + src + ';');
}
}
/// Build the GLSL register list.
void BuildRegisterList() {
regs.reserve(Register::NumRegisters);
for (std::size_t index = 0; index < Register::NumRegisters; ++index) {
regs.emplace_back(index, suffix);
}
}
void BuildInputList() {
const u32 size = static_cast<u32>(Attribute::Index::Attribute_31) -
static_cast<u32>(Attribute::Index::Attribute_0) + 1;
declr_input_attribute.reserve(size);
}
/// Generates code representing an input attribute register.
std::string GetInputAttribute(Attribute::Index attribute,
const Tegra::Shader::IpaMode& input_mode,
std::optional<Register> vertex = {}) {
auto GeometryPass = [&](const std::string& name) {
if (stage == Maxwell3D::Regs::ShaderStage::Geometry && vertex) {
// TODO(Rodrigo): Guard geometry inputs against out of bound reads. Some games set
// an 0x80000000 index for those and the shader fails to build. Find out why this
// happens and what's its intent.
return "gs_" + name + '[' + GetRegisterAsInteger(*vertex, 0, false) +
" % MAX_VERTEX_INPUT]";
}
return name;
};
switch (attribute) {
case Attribute::Index::Position:
if (stage != Maxwell3D::Regs::ShaderStage::Fragment) {
return GeometryPass("position");
} else {
return "vec4(gl_FragCoord.x, gl_FragCoord.y, gl_FragCoord.z, 1.0)";
}
case Attribute::Index::PointCoord:
return "vec4(gl_PointCoord.x, gl_PointCoord.y, 0, 0)";
case Attribute::Index::TessCoordInstanceIDVertexID:
// TODO(Subv): Find out what the values are for the first two elements when inside a
// vertex shader, and what's the value of the fourth element when inside a Tess Eval
// shader.
ASSERT(stage == Maxwell3D::Regs::ShaderStage::Vertex);
return "vec4(0, 0, uintBitsToFloat(instance_id.x), uintBitsToFloat(gl_VertexID))";
case Attribute::Index::FrontFacing:
// TODO(Subv): Find out what the values are for the other elements.
ASSERT(stage == Maxwell3D::Regs::ShaderStage::Fragment);
return "vec4(0, 0, 0, uintBitsToFloat(gl_FrontFacing ? 1 : 0))";
default:
const u32 index{static_cast<u32>(attribute) -
static_cast<u32>(Attribute::Index::Attribute_0)};
if (attribute >= Attribute::Index::Attribute_0 &&
attribute <= Attribute::Index::Attribute_31) {
if (declr_input_attribute.count(attribute) == 0) {
declr_input_attribute[attribute] = input_mode;
} else {
UNIMPLEMENTED_IF_MSG(declr_input_attribute[attribute] != input_mode,
"Multiple input modes for the same attribute");
}
return GeometryPass("input_attribute_" + std::to_string(index));
}
UNIMPLEMENTED_MSG("Unhandled input attribute: {}", static_cast<u32>(attribute));
}
return "vec4(0, 0, 0, 0)";
}
std::string GetInputFlags(const Attribute::Index attribute) {
const Tegra::Shader::IpaSampleMode sample_mode =
declr_input_attribute[attribute].sampling_mode;
const Tegra::Shader::IpaInterpMode interp_mode =
declr_input_attribute[attribute].interpolation_mode;
std::string out;
switch (interp_mode) {
case Tegra::Shader::IpaInterpMode::Flat: {
out += "flat ";
break;
}
case Tegra::Shader::IpaInterpMode::Linear: {
out += "noperspective ";
break;
}
case Tegra::Shader::IpaInterpMode::Perspective: {
// Default, Smooth
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled IPA interp mode: {}", static_cast<u32>(interp_mode));
}
}
switch (sample_mode) {
case Tegra::Shader::IpaSampleMode::Centroid:
// It can be implemented with the "centroid " keyword in glsl
UNIMPLEMENTED_MSG("Unimplemented IPA sampler mode centroid");
break;
case Tegra::Shader::IpaSampleMode::Default:
// Default, n/a
break;
default: {
UNIMPLEMENTED_MSG("Unimplemented IPA sampler mode: {}", static_cast<u32>(sample_mode));
break;
}
}
return out;
}
/// Generates code representing the declaration name of an output attribute register.
std::string GetOutputAttribute(Attribute::Index attribute) {
switch (attribute) {
case Attribute::Index::PointSize:
return "gl_PointSize";
case Attribute::Index::Position:
return "position";
case Attribute::Index::ClipDistances0123:
case Attribute::Index::ClipDistances4567: {
return "gl_ClipDistance";
}
default:
const u32 index{static_cast<u32>(attribute) -
static_cast<u32>(Attribute::Index::Attribute_0)};
if (attribute >= Attribute::Index::Attribute_0) {
declr_output_attribute.insert(attribute);
return "output_attribute_" + std::to_string(index);
}
UNIMPLEMENTED_MSG("Unhandled output attribute={}", index);
return {};
}
}
/// Generates code to use for a swizzle operation.
static std::string GetSwizzle(u64 elem) {
ASSERT(elem <= 3);
std::string swizzle = ".";
swizzle += "xyzw"[elem];
return swizzle;
}
ShaderWriter& shader;
ShaderWriter& declarations;
std::vector<GLSLRegister> regs;
std::unordered_map<Attribute::Index, Tegra::Shader::IpaMode> declr_input_attribute;
std::set<Attribute::Index> declr_output_attribute;
std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
std::vector<SamplerEntry> used_samplers;
const Maxwell3D::Regs::ShaderStage& stage;
const std::string& suffix;
const Tegra::Shader::Header& header;
std::unordered_set<Attribute::Index> fixed_pipeline_output_attributes_used;
u64 local_memory_size;
};
class GLSLGenerator {
public:
GLSLGenerator(const std::set<Subroutine>& subroutines, const ProgramCode& program_code,
u32 main_offset, Maxwell3D::Regs::ShaderStage stage, const std::string& suffix,
std::size_t shader_length)
: subroutines(subroutines), program_code(program_code), main_offset(main_offset),
stage(stage), suffix(suffix), shader_length(shader_length) {
std::memcpy(&header, program_code.data(), sizeof(Tegra::Shader::Header));
local_memory_size = header.GetLocalMemorySize();
regs.SetLocalMemory(local_memory_size);
Generate(suffix);
}
std::string GetShaderCode() {
return declarations.GetResult() + shader.GetResult();
}
/// Returns entries in the shader that are useful for external functions
ShaderEntries GetEntries() const {
return {regs.GetConstBuffersDeclarations(), regs.GetSamplers(), shader_length};
}
private:
/// Gets the Subroutine object corresponding to the specified address.
const Subroutine& GetSubroutine(u32 begin, u32 end) const {
const auto iter = subroutines.find(Subroutine{begin, end, suffix});
ASSERT(iter != subroutines.end());
return *iter;
}
/// Generates code representing a 19-bit immediate value
static std::string GetImmediate19(const Instruction& instr) {
return fmt::format("uintBitsToFloat({})", instr.alu.GetImm20_19());
}
/// Generates code representing a 32-bit immediate value
static std::string GetImmediate32(const Instruction& instr) {
return fmt::format("uintBitsToFloat({})", instr.alu.GetImm20_32());
}
/// Generates code representing a vec2 pair unpacked from a half float immediate
static std::string UnpackHalfImmediate(const Instruction& instr, bool negate) {
const std::string immediate = GetHalfFloat(std::to_string(instr.half_imm.PackImmediates()));
if (!negate) {
return immediate;
}
const std::string negate_first = instr.half_imm.first_negate != 0 ? "-" : "";
const std::string negate_second = instr.half_imm.second_negate != 0 ? "-" : "";
const std::string negate_vec = "vec2(" + negate_first + "1, " + negate_second + "1)";
return '(' + immediate + " * " + negate_vec + ')';
}
/// Generates code representing a texture sampler.
std::string GetSampler(const Sampler& sampler, Tegra::Shader::TextureType type, bool is_array,
bool is_shadow) {
return regs.AccessSampler(sampler, type, is_array, is_shadow);
}
/**
* Adds code that calls a subroutine.
* @param subroutine the subroutine to call.
*/
void CallSubroutine(const Subroutine& subroutine) {
if (subroutine.exit_method == ExitMethod::AlwaysEnd) {
shader.AddLine(subroutine.GetName() + "();");
shader.AddLine("return true;");
} else if (subroutine.exit_method == ExitMethod::Conditional) {
shader.AddLine("if (" + subroutine.GetName() + "()) { return true; }");
} else {
shader.AddLine(subroutine.GetName() + "();");
}
}
/*
* Writes code that assigns a predicate boolean variable.
* @param pred The id of the predicate to write to.
* @param value The expression value to assign to the predicate.
*/
void SetPredicate(u64 pred, const std::string& value) {
using Tegra::Shader::Pred;
// Can't assign to the constant predicate.
ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
shader.AddLine(variable + " = " + value + ';');
declr_predicates.insert(std::move(variable));
}
/*
* Returns the condition to use in the 'if' for a predicated instruction.
* @param instr Instruction to generate the if condition for.
* @returns string containing the predicate condition.
*/
std::string GetPredicateCondition(u64 index, bool negate) {
using Tegra::Shader::Pred;
std::string variable;
// Index 7 is used as an 'Always True' condition.
if (index == static_cast<u64>(Pred::UnusedIndex)) {
variable = "true";
} else {
variable = 'p' + std::to_string(index) + '_' + suffix;
declr_predicates.insert(variable);
}
if (negate) {
return "!(" + variable + ')';
}
return variable;
}
/**
* Returns the comparison string to use to compare two values in the 'set' family of
* instructions.
* @param condition The condition used in the 'set'-family instruction.
* @param op_a First operand to use for the comparison.
* @param op_b Second operand to use for the comparison.
* @returns String corresponding to the GLSL operator that matches the desired comparison.
*/
std::string GetPredicateComparison(Tegra::Shader::PredCondition condition,
const std::string& op_a, const std::string& op_b) const {
using Tegra::Shader::PredCondition;
static const std::unordered_map<PredCondition, const char*> PredicateComparisonStrings = {
{PredCondition::LessThan, "<"},
{PredCondition::Equal, "=="},
{PredCondition::LessEqual, "<="},
{PredCondition::GreaterThan, ">"},
{PredCondition::NotEqual, "!="},
{PredCondition::GreaterEqual, ">="},
{PredCondition::LessThanWithNan, "<"},
{PredCondition::NotEqualWithNan, "!="},
{PredCondition::LessEqualWithNan, "<="},
{PredCondition::GreaterThanWithNan, ">"},
{PredCondition::GreaterEqualWithNan, ">="}};
const auto& comparison{PredicateComparisonStrings.find(condition)};
UNIMPLEMENTED_IF_MSG(comparison == PredicateComparisonStrings.end(),
"Unknown predicate comparison operation");
std::string predicate{'(' + op_a + ") " + comparison->second + " (" + op_b + ')'};
if (condition == PredCondition::LessThanWithNan ||
condition == PredCondition::NotEqualWithNan ||
condition == PredCondition::LessEqualWithNan ||
condition == PredCondition::GreaterThanWithNan ||
condition == PredCondition::GreaterEqualWithNan) {
predicate += " || isnan(" + op_a + ") || isnan(" + op_b + ')';
}
return predicate;
}
/**
* Returns the operator string to use to combine two predicates in the 'setp' family of
* instructions.
* @params operation The operator used in the 'setp'-family instruction.
* @returns String corresponding to the GLSL operator that matches the desired operator.
*/
std::string GetPredicateCombiner(Tegra::Shader::PredOperation operation) const {
using Tegra::Shader::PredOperation;
static const std::unordered_map<PredOperation, const char*> PredicateOperationStrings = {
{PredOperation::And, "&&"},
{PredOperation::Or, "||"},
{PredOperation::Xor, "^^"},
};
auto op = PredicateOperationStrings.find(operation);
UNIMPLEMENTED_IF_MSG(op == PredicateOperationStrings.end(), "Unknown predicate operation");
return op->second;
}
/**
* Transforms the input string GLSL operand into one that applies the abs() function and negates
* the output if necessary. When both abs and neg are true, the negation will be applied after
* taking the absolute value.
* @param operand The input operand to take the abs() of, negate, or both.
* @param abs Whether to apply the abs() function to the input operand.
* @param neg Whether to negate the input operand.
* @returns String corresponding to the operand after being transformed by the abs() and
* negation operations.
*/
static std::string GetOperandAbsNeg(const std::string& operand, bool abs, bool neg) {
std::string result = operand;
if (abs) {
result = "abs(" + result + ')';
}
if (neg) {
result = "-(" + result + ')';
}
return result;
}
/*
* Transforms the input string GLSL operand into an unpacked half float pair.
* @note This function returns a float type pair instead of a half float pair. This is because
* real half floats are not standardized in GLSL but unpackHalf2x16 (which returns a vec2) is.
* @param operand Input operand. It has to be an unsigned integer.
* @param type How to unpack the unsigned integer to a half float pair.
* @param abs Get the absolute value of unpacked half floats.
* @param neg Get the negative value of unpacked half floats.
* @returns String corresponding to a half float pair.
*/
static std::string GetHalfFloat(const std::string& operand,
Tegra::Shader::HalfType type = Tegra::Shader::HalfType::H0_H1,
bool abs = false, bool neg = false) {
// "vec2" calls emitted in this function are intended to alias components.
const std::string value = [&]() {
switch (type) {
case Tegra::Shader::HalfType::H0_H1:
return "unpackHalf2x16(" + operand + ')';
case Tegra::Shader::HalfType::F32:
return "vec2(uintBitsToFloat(" + operand + "))";
case Tegra::Shader::HalfType::H0_H0:
case Tegra::Shader::HalfType::H1_H1: {
const bool high = type == Tegra::Shader::HalfType::H1_H1;
const char unpack_index = "xy"[high ? 1 : 0];
return "vec2(unpackHalf2x16(" + operand + ")." + unpack_index + ')';
}
default:
UNREACHABLE();
return std::string("vec2(0)");
}
}();
return GetOperandAbsNeg(value, abs, neg);
}
/*
* Returns whether the instruction at the specified offset is a 'sched' instruction.
* Sched instructions always appear before a sequence of 3 instructions.
*/
bool IsSchedInstruction(u32 offset) const {
// sched instructions appear once every 4 instructions.
static constexpr std::size_t SchedPeriod = 4;
u32 absolute_offset = offset - main_offset;
return (absolute_offset % SchedPeriod) == 0;
}
void WriteLogicOperation(Register dest, LogicOperation logic_op, const std::string& op_a,
const std::string& op_b,
Tegra::Shader::PredicateResultMode predicate_mode,
Tegra::Shader::Pred predicate) {
std::string result{};
switch (logic_op) {
case LogicOperation::And: {
result = '(' + op_a + " & " + op_b + ')';
break;
}
case LogicOperation::Or: {
result = '(' + op_a + " | " + op_b + ')';
break;
}
case LogicOperation::Xor: {
result = '(' + op_a + " ^ " + op_b + ')';
break;
}
case LogicOperation::PassB: {
result = op_b;
break;
}
default:
UNIMPLEMENTED_MSG("Unimplemented logic operation={}", static_cast<u32>(logic_op));
}
if (dest != Tegra::Shader::Register::ZeroIndex) {
regs.SetRegisterToInteger(dest, true, 0, result, 1, 1);
}
using Tegra::Shader::PredicateResultMode;
// Write the predicate value depending on the predicate mode.
switch (predicate_mode) {
case PredicateResultMode::None:
// Do nothing.
return;
case PredicateResultMode::NotZero:
// Set the predicate to true if the result is not zero.
SetPredicate(static_cast<u64>(predicate), '(' + result + ") != 0");
break;
default:
UNIMPLEMENTED_MSG("Unimplemented predicate result mode: {}",
static_cast<u32>(predicate_mode));
}
}
void WriteLop3Instruction(Register dest, const std::string& op_a, const std::string& op_b,
const std::string& op_c, const std::string& imm_lut) {
if (dest == Tegra::Shader::Register::ZeroIndex) {
return;
}
static constexpr std::array<const char*, 32> shift_amounts = {
"0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "10",
"11", "12", "13", "14", "15", "16", "17", "18", "19", "20", "21",
"22", "23", "24", "25", "26", "27", "28", "29", "30", "31"};
std::string result;
result += '(';
for (std::size_t i = 0; i < shift_amounts.size(); ++i) {
if (i)
result += '|';
result += "(((" + imm_lut + " >> (((" + op_c + " >> " + shift_amounts[i] +
") & 1) | ((" + op_b + " >> " + shift_amounts[i] + ") & 1) << 1 | ((" + op_a +
" >> " + shift_amounts[i] + ") & 1) << 2)) & 1) << " + shift_amounts[i] + ")";
}
result += ')';
regs.SetRegisterToInteger(dest, true, 0, result, 1, 1);
}
void WriteTexsInstruction(const Instruction& instr, const std::string& coord,
const std::string& texture) {
// Add an extra scope and declare the texture coords inside to prevent
// overwriting them in case they are used as outputs of the texs instruction.
shader.AddLine('{');
++shader.scope;
shader.AddLine(coord);
shader.AddLine("vec4 texture_tmp = " + texture + ';');
// TEXS has two destination registers and a swizzle. The first two elements in the swizzle
// go into gpr0+0 and gpr0+1, and the rest goes into gpr28+0 and gpr28+1
std::size_t written_components = 0;
for (u32 component = 0; component < 4; ++component) {
if (!instr.texs.IsComponentEnabled(component)) {
continue;
}
if (written_components < 2) {
// Write the first two swizzle components to gpr0 and gpr0+1
regs.SetRegisterToFloat(instr.gpr0, component, "texture_tmp", 1, 4, false,
written_components % 2);
} else {
ASSERT(instr.texs.HasTwoDestinations());
// Write the rest of the swizzle components to gpr28 and gpr28+1
regs.SetRegisterToFloat(instr.gpr28, component, "texture_tmp", 1, 4, false,
written_components % 2);
}
++written_components;
}
--shader.scope;
shader.AddLine('}');
}
static u32 TextureCoordinates(Tegra::Shader::TextureType texture_type) {
switch (texture_type) {
case Tegra::Shader::TextureType::Texture1D:
return 1;
case Tegra::Shader::TextureType::Texture2D:
return 2;
case Tegra::Shader::TextureType::Texture3D:
case Tegra::Shader::TextureType::TextureCube:
return 3;
default:
UNIMPLEMENTED_MSG("Unhandled texture type: {}", static_cast<u32>(texture_type));
return 0;
}
}
/*
* Emits code to push the input target address to the flow address stack, incrementing the stack
* top.
*/
void EmitPushToFlowStack(u32 target) {
shader.AddLine('{');
++shader.scope;
shader.AddLine("flow_stack[flow_stack_top] = " + std::to_string(target) + "u;");
shader.AddLine("flow_stack_top++;");
--shader.scope;
shader.AddLine('}');
}
/*
* Emits code to pop an address from the flow address stack, setting the jump address to the
* popped address and decrementing the stack top.
*/
void EmitPopFromFlowStack() {
shader.AddLine('{');
++shader.scope;
shader.AddLine("flow_stack_top--;");
shader.AddLine("jmp_to = flow_stack[flow_stack_top];");
shader.AddLine("break;");
--shader.scope;
shader.AddLine('}');
}
/// Writes the output values from a fragment shader to the corresponding GLSL output variables.
void EmitFragmentOutputsWrite() {
ASSERT(stage == Maxwell3D::Regs::ShaderStage::Fragment);
UNIMPLEMENTED_IF_MSG(header.ps.omap.sample_mask != 0, "Samplemask write is unimplemented");
shader.AddLine("if (alpha_test[0] != 0) {");
++shader.scope;
// We start on the register containing the alpha value in the first RT.
u32 current_reg = 3;
for (u32 render_target = 0; render_target < Maxwell3D::Regs::NumRenderTargets;
++render_target) {
// TODO(Blinkhawk): verify the behavior of alpha testing on hardware when
// multiple render targets are used.
if (header.ps.IsColorComponentOutputEnabled(render_target, 0) ||
header.ps.IsColorComponentOutputEnabled(render_target, 1) ||
header.ps.IsColorComponentOutputEnabled(render_target, 2) ||
header.ps.IsColorComponentOutputEnabled(render_target, 3)) {
shader.AddLine(fmt::format("if (!AlphaFunc({})) discard;",
regs.GetRegisterAsFloat(current_reg)));
current_reg += 4;
}
}
--shader.scope;
shader.AddLine('}');
// Write the color outputs using the data in the shader registers, disabled
// rendertargets/components are skipped in the register assignment.
current_reg = 0;
for (u32 render_target = 0; render_target < Maxwell3D::Regs::NumRenderTargets;
++render_target) {
// TODO(Subv): Figure out how dual-source blending is configured in the Switch.
for (u32 component = 0; component < 4; ++component) {
if (header.ps.IsColorComponentOutputEnabled(render_target, component)) {
shader.AddLine(fmt::format("FragColor{}[{}] = {};", render_target, component,
regs.GetRegisterAsFloat(current_reg)));
++current_reg;
}
}
}
if (header.ps.omap.depth) {
// The depth output is always 2 registers after the last color output, and current_reg
// already contains one past the last color register.
shader.AddLine(
"gl_FragDepth = " +
regs.GetRegisterAsFloat(static_cast<Tegra::Shader::Register>(current_reg) + 1) +
';');
}
}
/// Unpacks a video instruction operand (e.g. VMAD).
std::string GetVideoOperand(const std::string& op, bool is_chunk, bool is_signed,
Tegra::Shader::VideoType type, u64 byte_height) {
const std::string value = [&]() {
if (!is_chunk) {
const auto offset = static_cast<u32>(byte_height * 8);
return "((" + op + " >> " + std::to_string(offset) + ") & 0xff)";
}
const std::string zero = "0";
switch (type) {
case Tegra::Shader::VideoType::Size16_Low:
return '(' + op + " & 0xffff)";
case Tegra::Shader::VideoType::Size16_High:
return '(' + op + " >> 16)";
case Tegra::Shader::VideoType::Size32:
// TODO(Rodrigo): From my hardware tests it becomes a bit "mad" when
// this type is used (1 * 1 + 0 == 0x5b800000). Until a better
// explanation is found: abort.
UNIMPLEMENTED();
return zero;
case Tegra::Shader::VideoType::Invalid:
UNREACHABLE_MSG("Invalid instruction encoding");
return zero;
default:
UNREACHABLE();
return zero;
}
}();
if (is_signed) {
return "int(" + value + ')';
}
return value;
};
/// Gets the A operand for a video instruction.
std::string GetVideoOperandA(Instruction instr) {
return GetVideoOperand(regs.GetRegisterAsInteger(instr.gpr8, 0, false),
instr.video.is_byte_chunk_a != 0, instr.video.signed_a,
instr.video.type_a, instr.video.byte_height_a);
}
/// Gets the B operand for a video instruction.
std::string GetVideoOperandB(Instruction instr) {
if (instr.video.use_register_b) {
return GetVideoOperand(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.video.is_byte_chunk_b != 0, instr.video.signed_b,
instr.video.type_b, instr.video.byte_height_b);
} else {
return '(' +
std::to_string(instr.video.signed_b ? static_cast<s16>(instr.alu.GetImm20_16())
: instr.alu.GetImm20_16()) +
')';
}
}
/**
* Compiles a single instruction from Tegra to GLSL.
* @param offset the offset of the Tegra shader instruction.
* @return the offset of the next instruction to execute. Usually it is the current offset
* + 1. If the current instruction always terminates the program, returns PROGRAM_END.
*/
u32 CompileInstr(u32 offset) {
// Ignore sched instructions when generating code.
if (IsSchedInstruction(offset)) {
return offset + 1;
}
const Instruction instr = {program_code[offset]};
const auto opcode = OpCode::Decode(instr);
// Decoding failure
if (!opcode) {
UNIMPLEMENTED_MSG("Unhandled instruction: {0:x}", instr.value);
return offset + 1;
}
shader.AddLine(
fmt::format("// {}: {} (0x{:016x})", offset, opcode->get().GetName(), instr.value));
using Tegra::Shader::Pred;
UNIMPLEMENTED_IF_MSG(instr.pred.full_pred == Pred::NeverExecute,
"NeverExecute predicate not implemented");
// Some instructions (like SSY) don't have a predicate field, they are always
// unconditionally executed.
bool can_be_predicated = OpCode::IsPredicatedInstruction(opcode->get().GetId());
if (can_be_predicated && instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex)) {
shader.AddLine("if (" +
GetPredicateCondition(instr.pred.pred_index, instr.negate_pred != 0) +
')');
shader.AddLine('{');
++shader.scope;
}
switch (opcode->get().GetType()) {
case OpCode::Type::Arithmetic: {
std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
std::string op_b;
if (instr.is_b_imm) {
op_b = GetImmediate19(instr);
} else {
if (instr.is_b_gpr) {
op_b = regs.GetRegisterAsFloat(instr.gpr20);
} else {
op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
}
}
switch (opcode->get().GetId()) {
case OpCode::Id::MOV_C:
case OpCode::Id::MOV_R: {
// MOV does not have neither 'abs' nor 'neg' bits.
regs.SetRegisterToFloat(instr.gpr0, 0, op_b, 1, 1);
break;
}
case OpCode::Id::FMUL_C:
case OpCode::Id::FMUL_R:
case OpCode::Id::FMUL_IMM: {
// FMUL does not have 'abs' bits and only the second operand has a 'neg' bit.
UNIMPLEMENTED_IF_MSG(instr.fmul.tab5cb8_2 != 0,
"FMUL tab5cb8_2({}) is not implemented",
instr.fmul.tab5cb8_2.Value());
UNIMPLEMENTED_IF_MSG(instr.fmul.tab5c68_1 != 0,
"FMUL tab5cb8_1({}) is not implemented",
instr.fmul.tab5c68_1.Value());
UNIMPLEMENTED_IF_MSG(
instr.fmul.tab5c68_0 != 1, "FMUL tab5cb8_0({}) is not implemented",
instr.fmul.tab5c68_0
.Value()); // SMO typical sends 1 here which seems to be the default
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in FMUL is not implemented");
op_b = GetOperandAbsNeg(op_b, false, instr.fmul.negate_b);
regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " * " + op_b, 1, 1,
instr.alu.saturate_d, 0, true);
break;
}
case OpCode::Id::FADD_C:
case OpCode::Id::FADD_R:
case OpCode::Id::FADD_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in FADD is not implemented");
op_a = GetOperandAbsNeg(op_a, instr.alu.abs_a, instr.alu.negate_a);
op_b = GetOperandAbsNeg(op_b, instr.alu.abs_b, instr.alu.negate_b);
regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1,
instr.alu.saturate_d, 0, true);
break;
}
case OpCode::Id::MUFU: {
op_a = GetOperandAbsNeg(op_a, instr.alu.abs_a, instr.alu.negate_a);
switch (instr.sub_op) {
case SubOp::Cos:
regs.SetRegisterToFloat(instr.gpr0, 0, "cos(" + op_a + ')', 1, 1,
instr.alu.saturate_d, 0, true);
break;
case SubOp::Sin:
regs.SetRegisterToFloat(instr.gpr0, 0, "sin(" + op_a + ')', 1, 1,
instr.alu.saturate_d, 0, true);
break;
case SubOp::Ex2:
regs.SetRegisterToFloat(instr.gpr0, 0, "exp2(" + op_a + ')', 1, 1,
instr.alu.saturate_d, 0, true);
break;
case SubOp::Lg2:
regs.SetRegisterToFloat(instr.gpr0, 0, "log2(" + op_a + ')', 1, 1,
instr.alu.saturate_d, 0, true);
break;
case SubOp::Rcp:
regs.SetRegisterToFloat(instr.gpr0, 0, "1.0 / " + op_a, 1, 1,
instr.alu.saturate_d, 0, true);
break;
case SubOp::Rsq:
regs.SetRegisterToFloat(instr.gpr0, 0, "inversesqrt(" + op_a + ')', 1, 1,
instr.alu.saturate_d, 0, true);
break;
case SubOp::Sqrt:
regs.SetRegisterToFloat(instr.gpr0, 0, "sqrt(" + op_a + ')', 1, 1,
instr.alu.saturate_d, 0, true);
break;
default:
UNIMPLEMENTED_MSG("Unhandled MUFU sub op={0:x}",
static_cast<unsigned>(instr.sub_op.Value()));
}
break;
}
case OpCode::Id::FMNMX_C:
case OpCode::Id::FMNMX_R:
case OpCode::Id::FMNMX_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in FMNMX is not implemented");
op_a = GetOperandAbsNeg(op_a, instr.alu.abs_a, instr.alu.negate_a);
op_b = GetOperandAbsNeg(op_b, instr.alu.abs_b, instr.alu.negate_b);
std::string condition =
GetPredicateCondition(instr.alu.fmnmx.pred, instr.alu.fmnmx.negate_pred != 0);
std::string parameters = op_a + ',' + op_b;
regs.SetRegisterToFloat(instr.gpr0, 0,
'(' + condition + ") ? min(" + parameters + ") : max(" +
parameters + ')',
1, 1, false, 0, true);
break;
}
case OpCode::Id::RRO_C:
case OpCode::Id::RRO_R:
case OpCode::Id::RRO_IMM: {
// Currently RRO is only implemented as a register move.
op_b = GetOperandAbsNeg(op_b, instr.alu.abs_b, instr.alu.negate_b);
regs.SetRegisterToFloat(instr.gpr0, 0, op_b, 1, 1);
LOG_WARNING(HW_GPU, "RRO instruction is incomplete");
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled arithmetic instruction: {}", opcode->get().GetName());
}
}
break;
}
case OpCode::Type::ArithmeticImmediate: {
switch (opcode->get().GetId()) {
case OpCode::Id::MOV32_IMM: {
regs.SetRegisterToFloat(instr.gpr0, 0, GetImmediate32(instr), 1, 1);
break;
}
case OpCode::Id::FMUL32_IMM: {
UNIMPLEMENTED_IF_MSG(instr.op_32.generates_cc,
"Condition codes generation in FMUL32 is not implemented");
regs.SetRegisterToFloat(instr.gpr0, 0,
regs.GetRegisterAsFloat(instr.gpr8) + " * " +
GetImmediate32(instr),
1, 1, instr.fmul32.saturate, 0, true);
break;
}
case OpCode::Id::FADD32I: {
UNIMPLEMENTED_IF_MSG(instr.op_32.generates_cc,
"Condition codes generation in FADD32I is not implemented");
std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
std::string op_b = GetImmediate32(instr);
if (instr.fadd32i.abs_a) {
op_a = "abs(" + op_a + ')';
}
if (instr.fadd32i.negate_a) {
op_a = "-(" + op_a + ')';
}
if (instr.fadd32i.abs_b) {
op_b = "abs(" + op_b + ')';
}
if (instr.fadd32i.negate_b) {
op_b = "-(" + op_b + ')';
}
regs.SetRegisterToFloat(instr.gpr0, 0, op_a + " + " + op_b, 1, 1, false, 0, true);
break;
}
}
break;
}
case OpCode::Type::Bfe: {
UNIMPLEMENTED_IF(instr.bfe.negate_b);
std::string op_a = instr.bfe.negate_a ? "-" : "";
op_a += regs.GetRegisterAsInteger(instr.gpr8);
switch (opcode->get().GetId()) {
case OpCode::Id::BFE_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in BFE is not implemented");
std::string inner_shift =
'(' + op_a + " << " + std::to_string(instr.bfe.GetLeftShiftValue()) + ')';
std::string outer_shift =
'(' + inner_shift + " >> " +
std::to_string(instr.bfe.GetLeftShiftValue() + instr.bfe.shift_position) + ')';
regs.SetRegisterToInteger(instr.gpr0, true, 0, outer_shift, 1, 1);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled BFE instruction: {}", opcode->get().GetName());
}
}
break;
}
case OpCode::Type::Bfi: {
UNIMPLEMENTED_IF(instr.generates_cc);
const auto [base, packed_shift] = [&]() -> std::tuple<std::string, std::string> {
switch (opcode->get().GetId()) {
case OpCode::Id::BFI_IMM_R:
return {regs.GetRegisterAsInteger(instr.gpr39, 0, false),
std::to_string(instr.alu.GetSignedImm20_20())};
default:
UNREACHABLE();
}
}();
const std::string offset = '(' + packed_shift + " & 0xff)";
const std::string bits = "((" + packed_shift + " >> 8) & 0xff)";
const std::string insert = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
regs.SetRegisterToInteger(
instr.gpr0, false, 0,
"bitfieldInsert(" + base + ", " + insert + ", " + offset + ", " + bits + ')', 1, 1);
break;
}
case OpCode::Type::Shift: {
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true);
std::string op_b;
if (instr.is_b_imm) {
op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
} else {
if (instr.is_b_gpr) {
op_b += regs.GetRegisterAsInteger(instr.gpr20);
} else {
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Integer);
}
}
switch (opcode->get().GetId()) {
case OpCode::Id::SHR_C:
case OpCode::Id::SHR_R:
case OpCode::Id::SHR_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in SHR is not implemented");
if (!instr.shift.is_signed) {
// Logical shift right
op_a = "uint(" + op_a + ')';
}
// Cast to int is superfluous for arithmetic shift, it's only for a logical shift
regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(" + op_a + " >> " + op_b + ')',
1, 1);
break;
}
case OpCode::Id::SHL_C:
case OpCode::Id::SHL_R:
case OpCode::Id::SHL_IMM:
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in SHL is not implemented");
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " << " + op_b, 1, 1);
break;
default: {
UNIMPLEMENTED_MSG("Unhandled shift instruction: {}", opcode->get().GetName());
}
}
break;
}
case OpCode::Type::ArithmeticIntegerImmediate: {
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
std::string op_b = std::to_string(instr.alu.imm20_32.Value());
switch (opcode->get().GetId()) {
case OpCode::Id::IADD32I:
UNIMPLEMENTED_IF_MSG(instr.op_32.generates_cc,
"Condition codes generation in IADD32I is not implemented");
if (instr.iadd32i.negate_a)
op_a = "-(" + op_a + ')';
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
instr.iadd32i.saturate != 0);
break;
case OpCode::Id::LOP32I: {
UNIMPLEMENTED_IF_MSG(instr.op_32.generates_cc,
"Condition codes generation in LOP32I is not implemented");
if (instr.alu.lop32i.invert_a)
op_a = "~(" + op_a + ')';
if (instr.alu.lop32i.invert_b)
op_b = "~(" + op_b + ')';
WriteLogicOperation(instr.gpr0, instr.alu.lop32i.operation, op_a, op_b,
Tegra::Shader::PredicateResultMode::None,
Tegra::Shader::Pred::UnusedIndex);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled ArithmeticIntegerImmediate instruction: {}",
opcode->get().GetName());
}
}
break;
}
case OpCode::Type::ArithmeticInteger: {
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
std::string op_b;
if (instr.is_b_imm) {
op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
} else {
if (instr.is_b_gpr) {
op_b += regs.GetRegisterAsInteger(instr.gpr20);
} else {
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Integer);
}
}
switch (opcode->get().GetId()) {
case OpCode::Id::IADD_C:
case OpCode::Id::IADD_R:
case OpCode::Id::IADD_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in IADD is not implemented");
if (instr.alu_integer.negate_a)
op_a = "-(" + op_a + ')';
if (instr.alu_integer.negate_b)
op_b = "-(" + op_b + ')';
regs.SetRegisterToInteger(instr.gpr0, true, 0, op_a + " + " + op_b, 1, 1,
instr.alu.saturate_d);
break;
}
case OpCode::Id::IADD3_C:
case OpCode::Id::IADD3_R:
case OpCode::Id::IADD3_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in IADD3 is not implemented");
std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
auto apply_height = [](auto height, auto& oprand) {
switch (height) {
case Tegra::Shader::IAdd3Height::None:
break;
case Tegra::Shader::IAdd3Height::LowerHalfWord:
oprand = "((" + oprand + ") & 0xFFFF)";
break;
case Tegra::Shader::IAdd3Height::UpperHalfWord:
oprand = "((" + oprand + ") >> 16)";
break;
default:
UNIMPLEMENTED_MSG("Unhandled IADD3 height: {}",
static_cast<u32>(height.Value()));
}
};
if (opcode->get().GetId() == OpCode::Id::IADD3_R) {
apply_height(instr.iadd3.height_a, op_a);
apply_height(instr.iadd3.height_b, op_b);
apply_height(instr.iadd3.height_c, op_c);
}
if (instr.iadd3.neg_a)
op_a = "-(" + op_a + ')';
if (instr.iadd3.neg_b)
op_b = "-(" + op_b + ')';
if (instr.iadd3.neg_c)
op_c = "-(" + op_c + ')';
std::string result;
if (opcode->get().GetId() == OpCode::Id::IADD3_R) {
switch (instr.iadd3.mode) {
case Tegra::Shader::IAdd3Mode::RightShift:
// TODO(tech4me): According to
// https://envytools.readthedocs.io/en/latest/hw/graph/maxwell/cuda/int.html?highlight=iadd3
// The addition between op_a and op_b should be done in uint33, more
// investigation required
result = "(((" + op_a + " + " + op_b + ") >> 16) + " + op_c + ')';
break;
case Tegra::Shader::IAdd3Mode::LeftShift:
result = "(((" + op_a + " + " + op_b + ") << 16) + " + op_c + ')';
break;
default:
result = '(' + op_a + " + " + op_b + " + " + op_c + ')';
break;
}
} else {
result = '(' + op_a + " + " + op_b + " + " + op_c + ')';
}
regs.SetRegisterToInteger(instr.gpr0, true, 0, result, 1, 1);
break;
}
case OpCode::Id::ISCADD_C:
case OpCode::Id::ISCADD_R:
case OpCode::Id::ISCADD_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in ISCADD is not implemented");
if (instr.alu_integer.negate_a)
op_a = "-(" + op_a + ')';
if (instr.alu_integer.negate_b)
op_b = "-(" + op_b + ')';
const std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
regs.SetRegisterToInteger(instr.gpr0, true, 0,
"((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
break;
}
case OpCode::Id::POPC_C:
case OpCode::Id::POPC_R:
case OpCode::Id::POPC_IMM: {
if (instr.popc.invert) {
op_b = "~(" + op_b + ')';
}
regs.SetRegisterToInteger(instr.gpr0, true, 0, "bitCount(" + op_b + ')', 1, 1);
break;
}
case OpCode::Id::SEL_C:
case OpCode::Id::SEL_R:
case OpCode::Id::SEL_IMM: {
const std::string condition =
GetPredicateCondition(instr.sel.pred, instr.sel.neg_pred != 0);
regs.SetRegisterToInteger(instr.gpr0, true, 0,
'(' + condition + ") ? " + op_a + " : " + op_b, 1, 1);
break;
}
case OpCode::Id::LOP_C:
case OpCode::Id::LOP_R:
case OpCode::Id::LOP_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in LOP is not implemented");
if (instr.alu.lop.invert_a)
op_a = "~(" + op_a + ')';
if (instr.alu.lop.invert_b)
op_b = "~(" + op_b + ')';
WriteLogicOperation(instr.gpr0, instr.alu.lop.operation, op_a, op_b,
instr.alu.lop.pred_result_mode, instr.alu.lop.pred48);
break;
}
case OpCode::Id::LOP3_C:
case OpCode::Id::LOP3_R:
case OpCode::Id::LOP3_IMM: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in LOP3 is not implemented");
const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
std::string lut;
if (opcode->get().GetId() == OpCode::Id::LOP3_R) {
lut = '(' + std::to_string(instr.alu.lop3.GetImmLut28()) + ')';
} else {
lut = '(' + std::to_string(instr.alu.lop3.GetImmLut48()) + ')';
}
WriteLop3Instruction(instr.gpr0, op_a, op_b, op_c, lut);
break;
}
case OpCode::Id::IMNMX_C:
case OpCode::Id::IMNMX_R:
case OpCode::Id::IMNMX_IMM: {
UNIMPLEMENTED_IF(instr.imnmx.exchange != Tegra::Shader::IMinMaxExchange::None);
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in IMNMX is not implemented");
const std::string condition =
GetPredicateCondition(instr.imnmx.pred, instr.imnmx.negate_pred != 0);
const std::string parameters = op_a + ',' + op_b;
regs.SetRegisterToInteger(instr.gpr0, instr.imnmx.is_signed, 0,
'(' + condition + ") ? min(" + parameters + ") : max(" +
parameters + ')',
1, 1);
break;
}
case OpCode::Id::LEA_R2:
case OpCode::Id::LEA_R1:
case OpCode::Id::LEA_IMM:
case OpCode::Id::LEA_RZ:
case OpCode::Id::LEA_HI: {
std::string op_c;
switch (opcode->get().GetId()) {
case OpCode::Id::LEA_R2: {
op_a = regs.GetRegisterAsInteger(instr.gpr20);
op_b = regs.GetRegisterAsInteger(instr.gpr39);
op_c = std::to_string(instr.lea.r2.entry_a);
break;
}
case OpCode::Id::LEA_R1: {
const bool neg = instr.lea.r1.neg != 0;
op_a = regs.GetRegisterAsInteger(instr.gpr8);
if (neg)
op_a = "-(" + op_a + ')';
op_b = regs.GetRegisterAsInteger(instr.gpr20);
op_c = std::to_string(instr.lea.r1.entry_a);
break;
}
case OpCode::Id::LEA_IMM: {
const bool neg = instr.lea.imm.neg != 0;
op_b = regs.GetRegisterAsInteger(instr.gpr8);
if (neg)
op_b = "-(" + op_b + ')';
op_a = std::to_string(instr.lea.imm.entry_a);
op_c = std::to_string(instr.lea.imm.entry_b);
break;
}
case OpCode::Id::LEA_RZ: {
const bool neg = instr.lea.rz.neg != 0;
op_b = regs.GetRegisterAsInteger(instr.gpr8);
if (neg)
op_b = "-(" + op_b + ')';
op_a = regs.GetUniform(instr.lea.rz.cb_index, instr.lea.rz.cb_offset,
GLSLRegister::Type::Integer);
op_c = std::to_string(instr.lea.rz.entry_a);
break;
}
case OpCode::Id::LEA_HI:
default: {
op_b = regs.GetRegisterAsInteger(instr.gpr8);
op_a = std::to_string(instr.lea.imm.entry_a);
op_c = std::to_string(instr.lea.imm.entry_b);
UNIMPLEMENTED_MSG("Unhandled LEA subinstruction: {}", opcode->get().GetName());
}
}
UNIMPLEMENTED_IF_MSG(instr.lea.pred48 != static_cast<u64>(Pred::UnusedIndex),
"Unhandled LEA Predicate");
const std::string value = '(' + op_a + " + (" + op_b + "*(1 << " + op_c + ")))";
regs.SetRegisterToInteger(instr.gpr0, true, 0, value, 1, 1);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled ArithmeticInteger instruction: {}",
opcode->get().GetName());
}
}
break;
}
case OpCode::Type::ArithmeticHalf: {
if (opcode->get().GetId() == OpCode::Id::HADD2_C ||
opcode->get().GetId() == OpCode::Id::HADD2_R) {
UNIMPLEMENTED_IF(instr.alu_half.ftz != 0);
}
const bool negate_a =
opcode->get().GetId() != OpCode::Id::HMUL2_R && instr.alu_half.negate_a != 0;
const bool negate_b =
opcode->get().GetId() != OpCode::Id::HMUL2_C && instr.alu_half.negate_b != 0;
const std::string op_a =
GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.alu_half.type_a,
instr.alu_half.abs_a != 0, negate_a);
std::string op_b;
switch (opcode->get().GetId()) {
case OpCode::Id::HADD2_C:
case OpCode::Id::HMUL2_C:
op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::UnsignedInteger);
break;
case OpCode::Id::HADD2_R:
case OpCode::Id::HMUL2_R:
op_b = regs.GetRegisterAsInteger(instr.gpr20, 0, false);
break;
default:
UNREACHABLE();
op_b = "0";
break;
}
op_b = GetHalfFloat(op_b, instr.alu_half.type_b, instr.alu_half.abs_b != 0, negate_b);
const std::string result = [&]() {
switch (opcode->get().GetId()) {
case OpCode::Id::HADD2_C:
case OpCode::Id::HADD2_R:
return '(' + op_a + " + " + op_b + ')';
case OpCode::Id::HMUL2_C:
case OpCode::Id::HMUL2_R:
return '(' + op_a + " * " + op_b + ')';
default:
UNIMPLEMENTED_MSG("Unhandled half float instruction: {}",
opcode->get().GetName());
return std::string("0");
}
}();
regs.SetRegisterToHalfFloat(instr.gpr0, 0, result, instr.alu_half.merge, 1, 1,
instr.alu_half.saturate != 0);
break;
}
case OpCode::Type::ArithmeticHalfImmediate: {
if (opcode->get().GetId() == OpCode::Id::HADD2_IMM) {
UNIMPLEMENTED_IF(instr.alu_half_imm.ftz != 0);
} else {
UNIMPLEMENTED_IF(instr.alu_half_imm.precision !=
Tegra::Shader::HalfPrecision::None);
}
const std::string op_a = GetHalfFloat(
regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.alu_half_imm.type_a,
instr.alu_half_imm.abs_a != 0, instr.alu_half_imm.negate_a != 0);
const std::string op_b = UnpackHalfImmediate(instr, true);
const std::string result = [&]() {
switch (opcode->get().GetId()) {
case OpCode::Id::HADD2_IMM:
return op_a + " + " + op_b;
case OpCode::Id::HMUL2_IMM:
return op_a + " * " + op_b;
default:
UNREACHABLE();
return std::string("0");
}
}();
regs.SetRegisterToHalfFloat(instr.gpr0, 0, result, instr.alu_half_imm.merge, 1, 1,
instr.alu_half_imm.saturate != 0);
break;
}
case OpCode::Type::Ffma: {
const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
std::string op_b = instr.ffma.negate_b ? "-" : "";
std::string op_c = instr.ffma.negate_c ? "-" : "";
UNIMPLEMENTED_IF_MSG(instr.ffma.cc != 0, "FFMA cc not implemented");
UNIMPLEMENTED_IF_MSG(
instr.ffma.tab5980_0 != 1, "FFMA tab5980_0({}) not implemented",
instr.ffma.tab5980_0.Value()); // Seems to be 1 by default based on SMO
UNIMPLEMENTED_IF_MSG(instr.ffma.tab5980_1 != 0, "FFMA tab5980_1({}) not implemented",
instr.ffma.tab5980_1.Value());
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in FFMA is not implemented");
switch (opcode->get().GetId()) {
case OpCode::Id::FFMA_CR: {
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
op_c += regs.GetRegisterAsFloat(instr.gpr39);
break;
}
case OpCode::Id::FFMA_RR: {
op_b += regs.GetRegisterAsFloat(instr.gpr20);
op_c += regs.GetRegisterAsFloat(instr.gpr39);
break;
}
case OpCode::Id::FFMA_RC: {
op_b += regs.GetRegisterAsFloat(instr.gpr39);
op_c += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
break;
}
case OpCode::Id::FFMA_IMM: {
op_b += GetImmediate19(instr);
op_c += regs.GetRegisterAsFloat(instr.gpr39);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled FFMA instruction: {}", opcode->get().GetName());
}
}
regs.SetRegisterToFloat(instr.gpr0, 0, "fma(" + op_a + ", " + op_b + ", " + op_c + ')',
1, 1, instr.alu.saturate_d, 0, true);
break;
}
case OpCode::Type::Hfma2: {
if (opcode->get().GetId() == OpCode::Id::HFMA2_RR) {
UNIMPLEMENTED_IF(instr.hfma2.rr.precision != Tegra::Shader::HalfPrecision::None);
} else {
UNIMPLEMENTED_IF(instr.hfma2.precision != Tegra::Shader::HalfPrecision::None);
}
const bool saturate = opcode->get().GetId() == OpCode::Id::HFMA2_RR
? instr.hfma2.rr.saturate != 0
: instr.hfma2.saturate != 0;
const std::string op_a =
GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.hfma2.type_a);
std::string op_b, op_c;
switch (opcode->get().GetId()) {
case OpCode::Id::HFMA2_CR:
op_b = GetHalfFloat(regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::UnsignedInteger),
instr.hfma2.type_b, false, instr.hfma2.negate_b);
op_c = GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr39, 0, false),
instr.hfma2.type_reg39, false, instr.hfma2.negate_c);
break;
case OpCode::Id::HFMA2_RC:
op_b = GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr39, 0, false),
instr.hfma2.type_reg39, false, instr.hfma2.negate_b);
op_c = GetHalfFloat(regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::UnsignedInteger),
instr.hfma2.type_b, false, instr.hfma2.negate_c);
break;
case OpCode::Id::HFMA2_RR:
op_b = GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.hfma2.type_b, false, instr.hfma2.negate_b);
op_c = GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr39, 0, false),
instr.hfma2.rr.type_c, false, instr.hfma2.rr.negate_c);
break;
case OpCode::Id::HFMA2_IMM_R:
op_b = UnpackHalfImmediate(instr, true);
op_c = GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr39, 0, false),
instr.hfma2.type_reg39, false, instr.hfma2.negate_c);
break;
default:
UNREACHABLE();
op_c = op_b = "vec2(0)";
break;
}
const std::string result = '(' + op_a + " * " + op_b + " + " + op_c + ')';
regs.SetRegisterToHalfFloat(instr.gpr0, 0, result, instr.hfma2.merge, 1, 1, saturate);
break;
}
case OpCode::Type::Conversion: {
switch (opcode->get().GetId()) {
case OpCode::Id::I2I_R: {
UNIMPLEMENTED_IF(instr.conversion.selector);
std::string op_a = regs.GetRegisterAsInteger(
instr.gpr20, 0, instr.conversion.is_input_signed, instr.conversion.src_size);
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
}
if (instr.conversion.negate_a) {
op_a = "-(" + op_a + ')';
}
regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
1, instr.alu.saturate_d, 0, instr.conversion.dest_size,
instr.generates_cc.Value() != 0);
break;
}
case OpCode::Id::I2F_R:
case OpCode::Id::I2F_C: {
UNIMPLEMENTED_IF(instr.conversion.dest_size != Register::Size::Word);
UNIMPLEMENTED_IF(instr.conversion.selector);
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in I2F is not implemented");
std::string op_a{};
if (instr.is_b_gpr) {
op_a =
regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_input_signed,
instr.conversion.src_size);
} else {
op_a = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
instr.conversion.is_input_signed
? GLSLRegister::Type::Integer
: GLSLRegister::Type::UnsignedInteger,
instr.conversion.src_size);
}
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
}
if (instr.conversion.negate_a) {
op_a = "-(" + op_a + ')';
}
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
break;
}
case OpCode::Id::F2F_R: {
UNIMPLEMENTED_IF(instr.conversion.dest_size != Register::Size::Word);
UNIMPLEMENTED_IF(instr.conversion.src_size != Register::Size::Word);
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in F2F is not implemented");
std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
}
if (instr.conversion.negate_a) {
op_a = "-(" + op_a + ')';
}
switch (instr.conversion.f2f.rounding) {
case Tegra::Shader::F2fRoundingOp::None:
break;
case Tegra::Shader::F2fRoundingOp::Round:
op_a = "roundEven(" + op_a + ')';
break;
case Tegra::Shader::F2fRoundingOp::Floor:
op_a = "floor(" + op_a + ')';
break;
case Tegra::Shader::F2fRoundingOp::Ceil:
op_a = "ceil(" + op_a + ')';
break;
case Tegra::Shader::F2fRoundingOp::Trunc:
op_a = "trunc(" + op_a + ')';
break;
default:
UNIMPLEMENTED_MSG("Unimplemented F2F rounding mode {}",
static_cast<u32>(instr.conversion.f2f.rounding.Value()));
break;
}
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1, instr.alu.saturate_d);
break;
}
case OpCode::Id::F2I_R:
case OpCode::Id::F2I_C: {
UNIMPLEMENTED_IF(instr.conversion.src_size != Register::Size::Word);
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in F2I is not implemented");
std::string op_a{};
if (instr.is_b_gpr) {
op_a = regs.GetRegisterAsFloat(instr.gpr20);
} else {
op_a = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
}
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
}
if (instr.conversion.negate_a) {
op_a = "-(" + op_a + ')';
}
switch (instr.conversion.f2i.rounding) {
case Tegra::Shader::F2iRoundingOp::None:
break;
case Tegra::Shader::F2iRoundingOp::Floor:
op_a = "floor(" + op_a + ')';
break;
case Tegra::Shader::F2iRoundingOp::Ceil:
op_a = "ceil(" + op_a + ')';
break;
case Tegra::Shader::F2iRoundingOp::Trunc:
op_a = "trunc(" + op_a + ')';
break;
default:
UNIMPLEMENTED_MSG("Unimplemented F2I rounding mode {}",
static_cast<u32>(instr.conversion.f2i.rounding.Value()));
break;
}
if (instr.conversion.is_output_signed) {
op_a = "int(" + op_a + ')';
} else {
op_a = "uint(" + op_a + ')';
}
regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
1, false, 0, instr.conversion.dest_size);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled conversion instruction: {}", opcode->get().GetName());
}
}
break;
}
case OpCode::Type::Memory: {
switch (opcode->get().GetId()) {
case OpCode::Id::LD_A: {
// Note: Shouldn't this be interp mode flat? As in no interpolation made.
UNIMPLEMENTED_IF_MSG(instr.gpr8.Value() != Register::ZeroIndex,
"Indirect attribute loads are not supported");
UNIMPLEMENTED_IF_MSG((instr.attribute.fmt20.immediate.Value() % sizeof(u32)) != 0,
"Unaligned attribute loads are not supported");
Tegra::Shader::IpaMode input_mode{Tegra::Shader::IpaInterpMode::Perspective,
Tegra::Shader::IpaSampleMode::Default};
u64 next_element = instr.attribute.fmt20.element;
u64 next_index = static_cast<u64>(instr.attribute.fmt20.index.Value());
const auto LoadNextElement = [&](u32 reg_offset) {
regs.SetRegisterToInputAttibute(instr.gpr0.Value() + reg_offset, next_element,
static_cast<Attribute::Index>(next_index),
input_mode, instr.gpr39.Value());
// Load the next attribute element into the following register. If the element
// to load goes beyond the vec4 size, load the first element of the next
// attribute.
next_element = (next_element + 1) % 4;
next_index = next_index + (next_element == 0 ? 1 : 0);
};
const u32 num_words = static_cast<u32>(instr.attribute.fmt20.size.Value()) + 1;
for (u32 reg_offset = 0; reg_offset < num_words; ++reg_offset) {
LoadNextElement(reg_offset);
}
break;
}
case OpCode::Id::LD_C: {
UNIMPLEMENTED_IF(instr.ld_c.unknown != 0);
// Add an extra scope and declare the index register inside to prevent
// overwriting it in case it is used as an output of the LD instruction.
shader.AddLine("{");
++shader.scope;
shader.AddLine("uint index = (" + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
" / 4) & (MAX_CONSTBUFFER_ELEMENTS - 1);");
const std::string op_a =
regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, "index",
GLSLRegister::Type::Float);
switch (instr.ld_c.type.Value()) {
case Tegra::Shader::UniformType::Single:
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
break;
case Tegra::Shader::UniformType::Double: {
const std::string op_b =
regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4,
"index", GLSLRegister::Type::Float);
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
regs.SetRegisterToFloat(instr.gpr0.Value() + 1, 0, op_b, 1, 1);
break;
}
default:
UNIMPLEMENTED_MSG("Unhandled type: {}",
static_cast<unsigned>(instr.ld_c.type.Value()));
}
--shader.scope;
shader.AddLine("}");
break;
}
case OpCode::Id::LD_L: {
UNIMPLEMENTED_IF_MSG(instr.ld_l.unknown == 1, "LD_L Unhandled mode: {}",
static_cast<unsigned>(instr.ld_l.unknown.Value()));
// Add an extra scope and declare the index register inside to prevent
// overwriting it in case it is used as an output of the LD instruction.
shader.AddLine('{');
++shader.scope;
std::string op = '(' + regs.GetRegisterAsInteger(instr.gpr8, 0, false) + " + " +
std::to_string(instr.smem_imm.Value()) + ')';
shader.AddLine("uint index = (" + op + " / 4);");
const std::string op_a = regs.GetLocalMemoryAsFloat("index");
switch (instr.ldst_sl.type.Value()) {
case Tegra::Shader::StoreType::Bytes32:
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
break;
default:
UNIMPLEMENTED_MSG("LD_L Unhandled type: {}",
static_cast<unsigned>(instr.ldst_sl.type.Value()));
}
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::ST_A: {
UNIMPLEMENTED_IF_MSG(instr.gpr8.Value() != Register::ZeroIndex,
"Indirect attribute loads are not supported");
UNIMPLEMENTED_IF_MSG((instr.attribute.fmt20.immediate.Value() % sizeof(u32)) != 0,
"Unaligned attribute loads are not supported");
u64 next_element = instr.attribute.fmt20.element;
u64 next_index = static_cast<u64>(instr.attribute.fmt20.index.Value());
const auto StoreNextElement = [&](u32 reg_offset) {
regs.SetOutputAttributeToRegister(static_cast<Attribute::Index>(next_index),
next_element, instr.gpr0.Value() + reg_offset,
instr.gpr39.Value());
// Load the next attribute element into the following register. If the element
// to load goes beyond the vec4 size, load the first element of the next
// attribute.
next_element = (next_element + 1) % 4;
next_index = next_index + (next_element == 0 ? 1 : 0);
};
const u32 num_words = static_cast<u32>(instr.attribute.fmt20.size.Value()) + 1;
for (u32 reg_offset = 0; reg_offset < num_words; ++reg_offset) {
StoreNextElement(reg_offset);
}
break;
}
case OpCode::Id::ST_L: {
UNIMPLEMENTED_IF_MSG(instr.st_l.unknown == 0, "ST_L Unhandled mode: {}",
static_cast<unsigned>(instr.st_l.unknown.Value()));
// Add an extra scope and declare the index register inside to prevent
// overwriting it in case it is used as an output of the LD instruction.
shader.AddLine('{');
++shader.scope;
std::string op = '(' + regs.GetRegisterAsInteger(instr.gpr8, 0, false) + " + " +
std::to_string(instr.smem_imm.Value()) + ')';
shader.AddLine("uint index = (" + op + " / 4);");
switch (instr.ldst_sl.type.Value()) {
case Tegra::Shader::StoreType::Bytes32:
regs.SetLocalMemoryAsFloat("index", regs.GetRegisterAsFloat(instr.gpr0));
break;
default:
UNIMPLEMENTED_MSG("ST_L Unhandled type: {}",
static_cast<unsigned>(instr.ldst_sl.type.Value()));
}
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::TEX: {
Tegra::Shader::TextureType texture_type{instr.tex.texture_type};
std::string coord;
const bool is_array = instr.tex.array != 0;
UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
"AOFFI is not implemented");
const bool depth_compare =
instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
u32 num_coordinates = TextureCoordinates(texture_type);
u32 start_index = 0;
std::string array_elem;
if (is_array) {
array_elem = regs.GetRegisterAsInteger(instr.gpr8);
start_index = 1;
}
const auto process_mode = instr.tex.GetTextureProcessMode();
u32 start_index_b = 0;
std::string lod_value;
if (process_mode != Tegra::Shader::TextureProcessMode::LZ &&
process_mode != Tegra::Shader::TextureProcessMode::None) {
start_index_b = 1;
lod_value = regs.GetRegisterAsFloat(instr.gpr20);
}
std::string depth_value;
if (depth_compare) {
depth_value = regs.GetRegisterAsFloat(instr.gpr20.Value() + start_index_b);
}
bool depth_compare_extra = false;
switch (num_coordinates) {
case 1: {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index);
if (is_array) {
if (depth_compare) {
coord = "vec3 coords = vec3(" + x + ", " + depth_value + ", " +
array_elem + ");";
} else {
coord = "vec2 coords = vec2(" + x + ", " + array_elem + ");";
}
} else {
if (depth_compare) {
coord = "vec2 coords = vec2(" + x + ", " + depth_value + ");";
} else {
coord = "float coords = " + x + ';';
}
}
break;
}
case 2: {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index);
const std::string y =
regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 1);
if (is_array) {
if (depth_compare) {
coord = "vec4 coords = vec4(" + x + ", " + y + ", " + depth_value +
", " + array_elem + ");";
} else {
coord = "vec3 coords = vec3(" + x + ", " + y + ", " + array_elem + ");";
}
} else {
if (depth_compare) {
coord =
"vec3 coords = vec3(" + x + ", " + y + ", " + depth_value + ");";
} else {
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
}
}
break;
}
case 3: {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index);
const std::string y =
regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 1);
const std::string z =
regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 2);
if (is_array) {
depth_compare_extra = depth_compare;
coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " +
array_elem + ");";
} else {
if (depth_compare) {
coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " +
depth_value + ");";
} else {
coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
}
}
break;
}
default:
UNIMPLEMENTED_MSG("Unhandled coordinates number {}",
static_cast<u32>(num_coordinates));
// Fallback to interpreting as a 2D texture for now
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
texture_type = Tegra::Shader::TextureType::Texture2D;
}
const std::string sampler =
GetSampler(instr.sampler, texture_type, is_array, depth_compare);
// Add an extra scope and declare the texture coords inside to prevent
// overwriting them in case they are used as outputs of the texs instruction.
shader.AddLine('{');
++shader.scope;
shader.AddLine(coord);
std::string texture;
switch (instr.tex.GetTextureProcessMode()) {
case Tegra::Shader::TextureProcessMode::None: {
if (!depth_compare_extra) {
texture = "texture(" + sampler + ", coords)";
} else {
texture = "texture(" + sampler + ", coords, " + depth_value + ')';
}
break;
}
case Tegra::Shader::TextureProcessMode::LZ: {
if (!depth_compare_extra) {
texture = "textureLod(" + sampler + ", coords, 0.0)";
} else {
texture = "texture(" + sampler + ", coords, " + depth_value + ')';
}
break;
}
case Tegra::Shader::TextureProcessMode::LB:
case Tegra::Shader::TextureProcessMode::LBA: {
// TODO: Figure if A suffix changes the equation at all.
if (!depth_compare_extra) {
texture = "texture(" + sampler + ", coords, " + lod_value + ')';
} else {
texture = "texture(" + sampler + ", coords, " + depth_value + ')';
LOG_WARNING(HW_GPU,
"OpenGL Limitation: can't set bias value along depth compare");
}
break;
}
case Tegra::Shader::TextureProcessMode::LL:
case Tegra::Shader::TextureProcessMode::LLA: {
// TODO: Figure if A suffix changes the equation at all.
if (!depth_compare_extra) {
texture = "textureLod(" + sampler + ", coords, " + lod_value + ')';
} else {
texture = "texture(" + sampler + ", coords, " + depth_value + ')';
LOG_WARNING(HW_GPU,
"OpenGL Limitation: can't set lod value along depth compare");
}
break;
}
default: {
if (!depth_compare_extra) {
texture = "texture(" + sampler + ", coords)";
} else {
texture = "texture(" + sampler + ", coords, " + depth_value + ')';
}
UNIMPLEMENTED_MSG("Unhandled texture process mode {}",
static_cast<u32>(instr.tex.GetTextureProcessMode()));
}
}
if (!depth_compare) {
shader.AddLine("vec4 texture_tmp = " + texture + ';');
std::size_t dest_elem{};
for (std::size_t elem = 0; elem < 4; ++elem) {
if (!instr.tex.IsComponentEnabled(elem)) {
// Skip disabled components
continue;
}
regs.SetRegisterToFloat(instr.gpr0, elem, "texture_tmp", 1, 4, false,
dest_elem);
++dest_elem;
}
} else {
regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false);
}
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::TEXS: {
Tegra::Shader::TextureType texture_type{instr.texs.GetTextureType()};
bool is_array{instr.texs.IsArrayTexture()};
UNIMPLEMENTED_IF_MSG(instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
const bool depth_compare =
instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
u32 num_coordinates = TextureCoordinates(texture_type);
const auto process_mode = instr.texs.GetTextureProcessMode();
std::string lod_value;
std::string coord;
u32 lod_offset = 0;
if (process_mode == Tegra::Shader::TextureProcessMode::LL) {
if (num_coordinates > 2) {
lod_value = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
lod_offset = 2;
} else {
lod_value = regs.GetRegisterAsFloat(instr.gpr20);
lod_offset = 1;
}
}
switch (num_coordinates) {
case 1: {
coord = "float coords = " + regs.GetRegisterAsFloat(instr.gpr8) + ';';
break;
}
case 2: {
if (is_array) {
if (depth_compare) {
const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
const std::string z = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
");";
} else {
const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
}
} else {
if (lod_offset != 0) {
if (depth_compare) {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y =
regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
const std::string z =
regs.GetRegisterAsFloat(instr.gpr20.Value() + lod_offset);
coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
} else {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y =
regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
}
} else {
if (depth_compare) {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y =
regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
} else {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
}
}
}
break;
}
case 3: {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
break;
}
default:
UNIMPLEMENTED_MSG("Unhandled coordinates number {}",
static_cast<u32>(num_coordinates));
// Fallback to interpreting as a 2D texture for now
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
texture_type = Tegra::Shader::TextureType::Texture2D;
is_array = false;
}
const std::string sampler =
GetSampler(instr.sampler, texture_type, is_array, depth_compare);
std::string texture;
switch (process_mode) {
case Tegra::Shader::TextureProcessMode::None: {
texture = "texture(" + sampler + ", coords)";
break;
}
case Tegra::Shader::TextureProcessMode::LZ: {
if (depth_compare && is_array) {
texture = "texture(" + sampler + ", coords)";
} else {
texture = "textureLod(" + sampler + ", coords, 0.0)";
}
break;
}
case Tegra::Shader::TextureProcessMode::LL: {
texture = "textureLod(" + sampler + ", coords, " + lod_value + ')';
break;
}
default: {
texture = "texture(" + sampler + ", coords)";
UNIMPLEMENTED_MSG("Unhandled texture process mode {}",
static_cast<u32>(instr.texs.GetTextureProcessMode()));
}
}
if (!depth_compare) {
WriteTexsInstruction(instr, coord, texture);
} else {
WriteTexsInstruction(instr, coord, "vec4(" + texture + ')');
}
break;
}
case OpCode::Id::TLDS: {
const Tegra::Shader::TextureType texture_type{instr.tlds.GetTextureType()};
const bool is_array{instr.tlds.IsArrayTexture()};
ASSERT(texture_type == Tegra::Shader::TextureType::Texture2D);
ASSERT(is_array == false);
UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
"AOFFI is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(Tegra::Shader::TextureMiscMode::MZ),
"MZ is not implemented");
u32 extra_op_offset = 0;
// Scope to avoid variable name overlaps.
shader.AddLine('{');
++shader.scope;
std::string coords;
switch (texture_type) {
case Tegra::Shader::TextureType::Texture1D: {
const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
coords = "float coords = " + x + ';';
break;
}
case Tegra::Shader::TextureType::Texture2D: {
UNIMPLEMENTED_IF_MSG(is_array, "Unhandled 2d array texture");
const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
const std::string y = regs.GetRegisterAsInteger(instr.gpr20);
// shader.AddLine("ivec2 coords = ivec2(" + x + ", " + y + ");");
coords = "ivec2 coords = ivec2(" + x + ", " + y + ");";
extra_op_offset = 1;
break;
}
default:
UNIMPLEMENTED_MSG("Unhandled texture type {}", static_cast<u32>(texture_type));
}
const std::string sampler =
GetSampler(instr.sampler, texture_type, is_array, false);
std::string texture = "texelFetch(" + sampler + ", coords, 0)";
switch (instr.tlds.GetTextureProcessMode()) {
case Tegra::Shader::TextureProcessMode::LZ: {
texture = "texelFetch(" + sampler + ", coords, 0)";
break;
}
case Tegra::Shader::TextureProcessMode::LL: {
shader.AddLine(
"float lod = " +
regs.GetRegisterAsInteger(instr.gpr20.Value() + extra_op_offset) + ';');
texture = "texelFetch(" + sampler + ", coords, lod)";
break;
}
default: {
texture = "texelFetch(" + sampler + ", coords, 0)";
UNIMPLEMENTED_MSG("Unhandled texture process mode {}",
static_cast<u32>(instr.tlds.GetTextureProcessMode()));
}
}
WriteTexsInstruction(instr, coords, texture);
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::TLD4: {
ASSERT(instr.tld4.texture_type == Tegra::Shader::TextureType::Texture2D);
ASSERT(instr.tld4.array == 0);
std::string coord;
UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
"AOFFI is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
"NDV is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::PTP),
"PTP is not implemented");
const bool depth_compare =
instr.tld4.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
auto texture_type = instr.tld4.texture_type.Value();
u32 num_coordinates = TextureCoordinates(texture_type);
if (depth_compare)
num_coordinates += 1;
// Add an extra scope and declare the texture coords inside to prevent
// overwriting them in case they are used as outputs of the texs instruction.
shader.AddLine('{');
++shader.scope;
switch (num_coordinates) {
case 2: {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");");
break;
}
case 3: {
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + ");");
break;
}
default:
UNIMPLEMENTED_MSG("Unhandled coordinates number {}",
static_cast<u32>(num_coordinates));
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");");
texture_type = Tegra::Shader::TextureType::Texture2D;
}
const std::string sampler =
GetSampler(instr.sampler, texture_type, false, depth_compare);
const std::string texture = "textureGather(" + sampler + ", coords, " +
std::to_string(instr.tld4.component) + ')';
if (!depth_compare) {
shader.AddLine("vec4 texture_tmp = " + texture + ';');
std::size_t dest_elem{};
for (std::size_t elem = 0; elem < 4; ++elem) {
if (!instr.tex.IsComponentEnabled(elem)) {
// Skip disabled components
continue;
}
regs.SetRegisterToFloat(instr.gpr0, elem, "texture_tmp", 1, 4, false,
dest_elem);
++dest_elem;
}
} else {
regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false);
}
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::TLD4S: {
UNIMPLEMENTED_IF_MSG(
instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
UNIMPLEMENTED_IF_MSG(
instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::AOFFI),
"AOFFI is not implemented");
// Scope to avoid variable name overlaps.
shader.AddLine('{');
++shader.scope;
std::string coords;
const bool depth_compare =
instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
// TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
const std::string sampler = GetSampler(
instr.sampler, Tegra::Shader::TextureType::Texture2D, false, depth_compare);
if (!depth_compare) {
coords = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
} else {
// Note: TLD4S coordinate encoding works just like TEXS's
const std::string op_y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
coords = "vec3 coords = vec3(" + op_a + ", " + op_y + ", " + op_b + ");";
}
const std::string texture = "textureGather(" + sampler + ", coords, " +
std::to_string(instr.tld4s.component) + ')';
if (!depth_compare) {
WriteTexsInstruction(instr, coords, texture);
} else {
WriteTexsInstruction(instr, coords, "vec4(" + texture + ')');
}
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::TXQ: {
UNIMPLEMENTED_IF_MSG(instr.txq.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
++shader.scope;
shader.AddLine('{');
// TODO: the new commits on the texture refactor, change the way samplers work.
// Sadly, not all texture instructions specify the type of texture their sampler
// uses. This must be fixed at a later instance.
const std::string sampler =
GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false, false);
switch (instr.txq.query_type) {
case Tegra::Shader::TextureQueryType::Dimension: {
const std::string texture = "textureSize(" + sampler + ", " +
regs.GetRegisterAsInteger(instr.gpr8) + ')';
const std::string mip_level = "textureQueryLevels(" + sampler + ')';
shader.AddLine("ivec2 sizes = " + texture + ';');
regs.SetRegisterToInteger(instr.gpr0, true, 0, "sizes.x", 1, 1);
regs.SetRegisterToInteger(instr.gpr0.Value() + 1, true, 0, "sizes.y", 1, 1);
regs.SetRegisterToInteger(instr.gpr0.Value() + 2, true, 0, "0", 1, 1);
regs.SetRegisterToInteger(instr.gpr0.Value() + 3, true, 0, mip_level, 1, 1);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled texture query type: {}",
static_cast<u32>(instr.txq.query_type.Value()));
}
}
--shader.scope;
shader.AddLine('}');
break;
}
case OpCode::Id::TMML: {
UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NODEP),
"NODEP is not implemented");
UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
"NDV is not implemented");
const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
const bool is_array = instr.tmml.array != 0;
auto texture_type = instr.tmml.texture_type.Value();
const std::string sampler =
GetSampler(instr.sampler, texture_type, is_array, false);
// TODO: add coordinates for different samplers once other texture types are
// implemented.
std::string coord;
switch (texture_type) {
case Tegra::Shader::TextureType::Texture1D: {
coord = "float coords = " + x + ';';
break;
}
case Tegra::Shader::TextureType::Texture2D: {
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
break;
}
default:
UNIMPLEMENTED_MSG("Unhandled texture type {}", static_cast<u32>(texture_type));
// Fallback to interpreting as a 2D texture for now
const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
coord = "vec2 coords = vec2(" + x + ", " + y + ");";
texture_type = Tegra::Shader::TextureType::Texture2D;
}
// Add an extra scope and declare the texture coords inside to prevent
// overwriting them in case they are used as outputs of the texs instruction.
shader.AddLine('{');
++shader.scope;
shader.AddLine(coord);
const std::string texture = "textureQueryLod(" + sampler + ", coords)";
const std::string tmp = "vec2 tmp = " + texture + "*vec2(256.0, 256.0);";
shader.AddLine(tmp);
regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(tmp.y)", 1, 1);
regs.SetRegisterToInteger(instr.gpr0.Value() + 1, false, 0, "uint(tmp.x)", 1, 1);
--shader.scope;
shader.AddLine('}');
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled memory instruction: {}", opcode->get().GetName());
}
}
break;
}
case OpCode::Type::FloatSetPredicate: {
const std::string op_a =
GetOperandAbsNeg(regs.GetRegisterAsFloat(instr.gpr8), instr.fsetp.abs_a != 0,
instr.fsetp.neg_a != 0);
std::string op_b;
if (instr.is_b_imm) {
op_b += '(' + GetImmediate19(instr) + ')';
} else {
if (instr.is_b_gpr) {
op_b += regs.GetRegisterAsFloat(instr.gpr20);
} else {
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
}
}
if (instr.fsetp.abs_b) {
op_b = "abs(" + op_b + ')';
}
// We can't use the constant predicate as destination.
ASSERT(instr.fsetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
const std::string second_pred =
GetPredicateCondition(instr.fsetp.pred39, instr.fsetp.neg_pred != 0);
const std::string combiner = GetPredicateCombiner(instr.fsetp.op);
const std::string predicate = GetPredicateComparison(instr.fsetp.cond, op_a, op_b);
// Set the primary predicate to the result of Predicate OP SecondPredicate
SetPredicate(instr.fsetp.pred3,
'(' + predicate + ") " + combiner + " (" + second_pred + ')');
if (instr.fsetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
// Set the secondary predicate to the result of !Predicate OP SecondPredicate,
// if enabled
SetPredicate(instr.fsetp.pred0,
"!(" + predicate + ") " + combiner + " (" + second_pred + ')');
}
break;
}
case OpCode::Type::IntegerSetPredicate: {
const std::string op_a =
regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
std::string op_b;
if (instr.is_b_imm) {
op_b += '(' + std::to_string(instr.alu.GetSignedImm20_20()) + ')';
} else {
if (instr.is_b_gpr) {
op_b += regs.GetRegisterAsInteger(instr.gpr20, 0, instr.isetp.is_signed);
} else {
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Integer);
}
}
// We can't use the constant predicate as destination.
ASSERT(instr.isetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
const std::string second_pred =
GetPredicateCondition(instr.isetp.pred39, instr.isetp.neg_pred != 0);
const std::string combiner = GetPredicateCombiner(instr.isetp.op);
const std::string predicate = GetPredicateComparison(instr.isetp.cond, op_a, op_b);
// Set the primary predicate to the result of Predicate OP SecondPredicate
SetPredicate(instr.isetp.pred3,
'(' + predicate + ") " + combiner + " (" + second_pred + ')');
if (instr.isetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
// Set the secondary predicate to the result of !Predicate OP SecondPredicate,
// if enabled
SetPredicate(instr.isetp.pred0,
"!(" + predicate + ") " + combiner + " (" + second_pred + ')');
}
break;
}
case OpCode::Type::HalfSetPredicate: {
UNIMPLEMENTED_IF(instr.hsetp2.ftz != 0);
const std::string op_a =
GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.hsetp2.type_a,
instr.hsetp2.abs_a, instr.hsetp2.negate_a);
const std::string op_b = [&]() {
switch (opcode->get().GetId()) {
case OpCode::Id::HSETP2_R:
return GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.hsetp2.type_b, instr.hsetp2.abs_a,
instr.hsetp2.negate_b);
default:
UNREACHABLE();
return std::string("vec2(0)");
}
}();
// We can't use the constant predicate as destination.
ASSERT(instr.hsetp2.pred3 != static_cast<u64>(Pred::UnusedIndex));
const std::string second_pred =
GetPredicateCondition(instr.hsetp2.pred39, instr.hsetp2.neg_pred != 0);
const std::string combiner = GetPredicateCombiner(instr.hsetp2.op);
const std::string component_combiner = instr.hsetp2.h_and ? "&&" : "||";
const std::string predicate =
'(' + GetPredicateComparison(instr.hsetp2.cond, op_a + ".x", op_b + ".x") + ' ' +
component_combiner + ' ' +
GetPredicateComparison(instr.hsetp2.cond, op_a + ".y", op_b + ".y") + ')';
// Set the primary predicate to the result of Predicate OP SecondPredicate
SetPredicate(instr.hsetp2.pred3,
'(' + predicate + ") " + combiner + " (" + second_pred + ')');
if (instr.hsetp2.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
// Set the secondary predicate to the result of !Predicate OP SecondPredicate,
// if enabled
SetPredicate(instr.hsetp2.pred0,
"!(" + predicate + ") " + combiner + " (" + second_pred + ')');
}
break;
}
case OpCode::Type::PredicateSetRegister: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in PSET is not implemented");
const std::string op_a =
GetPredicateCondition(instr.pset.pred12, instr.pset.neg_pred12 != 0);
const std::string op_b =
GetPredicateCondition(instr.pset.pred29, instr.pset.neg_pred29 != 0);
const std::string second_pred =
GetPredicateCondition(instr.pset.pred39, instr.pset.neg_pred39 != 0);
const std::string combiner = GetPredicateCombiner(instr.pset.op);
const std::string predicate =
'(' + op_a + ") " + GetPredicateCombiner(instr.pset.cond) + " (" + op_b + ')';
const std::string result = '(' + predicate + ") " + combiner + " (" + second_pred + ')';
if (instr.pset.bf == 0) {
const std::string value = '(' + result + ") ? 0xFFFFFFFF : 0";
regs.SetRegisterToInteger(instr.gpr0, false, 0, value, 1, 1);
} else {
const std::string value = '(' + result + ") ? 1.0 : 0.0";
regs.SetRegisterToFloat(instr.gpr0, 0, value, 1, 1);
}
break;
}
case OpCode::Type::PredicateSetPredicate: {
switch (opcode->get().GetId()) {
case OpCode::Id::PSETP: {
const std::string op_a =
GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
const std::string op_b =
GetPredicateCondition(instr.psetp.pred29, instr.psetp.neg_pred29 != 0);
// We can't use the constant predicate as destination.
ASSERT(instr.psetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
const std::string second_pred =
GetPredicateCondition(instr.psetp.pred39, instr.psetp.neg_pred39 != 0);
const std::string combiner = GetPredicateCombiner(instr.psetp.op);
const std::string predicate =
'(' + op_a + ") " + GetPredicateCombiner(instr.psetp.cond) + " (" + op_b + ')';
// Set the primary predicate to the result of Predicate OP SecondPredicate
SetPredicate(instr.psetp.pred3,
'(' + predicate + ") " + combiner + " (" + second_pred + ')');
if (instr.psetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
// Set the secondary predicate to the result of !Predicate OP SecondPredicate,
// if enabled
SetPredicate(instr.psetp.pred0,
"!(" + predicate + ") " + combiner + " (" + second_pred + ')');
}
break;
}
case OpCode::Id::CSETP: {
const std::string pred =
GetPredicateCondition(instr.csetp.pred39, instr.csetp.neg_pred39 != 0);
const std::string combiner = GetPredicateCombiner(instr.csetp.op);
const std::string condition_code = regs.GetConditionCode(instr.csetp.cc);
if (instr.csetp.pred3 != static_cast<u64>(Pred::UnusedIndex)) {
SetPredicate(instr.csetp.pred3,
'(' + condition_code + ") " + combiner + " (" + pred + ')');
}
if (instr.csetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
SetPredicate(instr.csetp.pred0,
"!(" + condition_code + ") " + combiner + " (" + pred + ')');
}
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled predicate instruction: {}", opcode->get().GetName());
}
}
break;
}
case OpCode::Type::RegisterSetPredicate: {
UNIMPLEMENTED_IF(instr.r2p.mode != Tegra::Shader::R2pMode::Pr);
const std::string apply_mask = [&]() {
switch (opcode->get().GetId()) {
case OpCode::Id::R2P_IMM:
return std::to_string(instr.r2p.immediate_mask);
default:
UNREACHABLE();
}
}();
const std::string mask = '(' + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
" >> " + std::to_string(instr.r2p.byte) + ')';
constexpr u64 programmable_preds = 7;
for (u64 pred = 0; pred < programmable_preds; ++pred) {
const auto shift = std::to_string(1 << pred);
shader.AddLine("if ((" + apply_mask + " & " + shift + ") != 0) {");
++shader.scope;
SetPredicate(pred, '(' + mask + " & " + shift + ") != 0");
--shader.scope;
shader.AddLine('}');
}
break;
}
case OpCode::Type::FloatSet: {
const std::string op_a = GetOperandAbsNeg(regs.GetRegisterAsFloat(instr.gpr8),
instr.fset.abs_a != 0, instr.fset.neg_a != 0);
std::string op_b;
if (instr.is_b_imm) {
const std::string imm = GetImmediate19(instr);
op_b = imm;
} else {
if (instr.is_b_gpr) {
op_b = regs.GetRegisterAsFloat(instr.gpr20);
} else {
op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Float);
}
}
op_b = GetOperandAbsNeg(op_b, instr.fset.abs_b != 0, instr.fset.neg_b != 0);
// The fset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
// condition is true, and to 0 otherwise.
const std::string second_pred =
GetPredicateCondition(instr.fset.pred39, instr.fset.neg_pred != 0);
const std::string combiner = GetPredicateCombiner(instr.fset.op);
const std::string predicate = "((" +
GetPredicateComparison(instr.fset.cond, op_a, op_b) +
") " + combiner + " (" + second_pred + "))";
if (instr.fset.bf) {
regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
} else {
regs.SetRegisterToInteger(instr.gpr0, false, 0, predicate + " ? 0xFFFFFFFF : 0", 1,
1);
}
break;
}
case OpCode::Type::IntegerSet: {
const std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
std::string op_b;
if (instr.is_b_imm) {
op_b = std::to_string(instr.alu.GetSignedImm20_20());
} else {
if (instr.is_b_gpr) {
op_b = regs.GetRegisterAsInteger(instr.gpr20, 0, instr.iset.is_signed);
} else {
op_b = regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
GLSLRegister::Type::Integer);
}
}
// The iset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
// condition is true, and to 0 otherwise.
const std::string second_pred =
GetPredicateCondition(instr.iset.pred39, instr.iset.neg_pred != 0);
const std::string combiner = GetPredicateCombiner(instr.iset.op);
const std::string predicate = "((" +
GetPredicateComparison(instr.iset.cond, op_a, op_b) +
") " + combiner + " (" + second_pred + "))";
if (instr.iset.bf) {
regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
} else {
regs.SetRegisterToInteger(instr.gpr0, false, 0, predicate + " ? 0xFFFFFFFF : 0", 1,
1);
}
break;
}
case OpCode::Type::HalfSet: {
UNIMPLEMENTED_IF(instr.hset2.ftz != 0);
const std::string op_a =
GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr8, 0, false), instr.hset2.type_a,
instr.hset2.abs_a != 0, instr.hset2.negate_a != 0);
const std::string op_b = [&]() {
switch (opcode->get().GetId()) {
case OpCode::Id::HSET2_R:
return GetHalfFloat(regs.GetRegisterAsInteger(instr.gpr20, 0, false),
instr.hset2.type_b, instr.hset2.abs_b != 0,
instr.hset2.negate_b != 0);
default:
UNREACHABLE();
return std::string("vec2(0)");
}
}();
const std::string second_pred =
GetPredicateCondition(instr.hset2.pred39, instr.hset2.neg_pred != 0);
const std::string combiner = GetPredicateCombiner(instr.hset2.op);
// HSET2 operates on each half float in the pack.
std::string result;
for (int i = 0; i < 2; ++i) {
const std::string float_value = i == 0 ? "0x00003c00" : "0x3c000000";
const std::string integer_value = i == 0 ? "0x0000ffff" : "0xffff0000";
const std::string value = instr.hset2.bf == 1 ? float_value : integer_value;
const std::string comp = std::string(".") + "xy"[i];
const std::string predicate =
"((" + GetPredicateComparison(instr.hset2.cond, op_a + comp, op_b + comp) +
") " + combiner + " (" + second_pred + "))";
result += '(' + predicate + " ? " + value + " : 0)";
if (i == 0) {
result += " | ";
}
}
regs.SetRegisterToInteger(instr.gpr0, false, 0, '(' + result + ')', 1, 1);
break;
}
case OpCode::Type::Xmad: {
UNIMPLEMENTED_IF(instr.xmad.sign_a);
UNIMPLEMENTED_IF(instr.xmad.sign_b);
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in XMAD is not implemented");
std::string op_a{regs.GetRegisterAsInteger(instr.gpr8, 0, instr.xmad.sign_a)};
std::string op_b;
std::string op_c;
// TODO(bunnei): Needs to be fixed once op_a or op_b is signed
UNIMPLEMENTED_IF(instr.xmad.sign_a != instr.xmad.sign_b);
const bool is_signed{instr.xmad.sign_a == 1};
bool is_merge{};
switch (opcode->get().GetId()) {
case OpCode::Id::XMAD_CR: {
is_merge = instr.xmad.merge_56;
op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
instr.xmad.sign_b ? GLSLRegister::Type::Integer
: GLSLRegister::Type::UnsignedInteger);
op_c += regs.GetRegisterAsInteger(instr.gpr39, 0, is_signed);
break;
}
case OpCode::Id::XMAD_RR: {
is_merge = instr.xmad.merge_37;
op_b += regs.GetRegisterAsInteger(instr.gpr20, 0, instr.xmad.sign_b);
op_c += regs.GetRegisterAsInteger(instr.gpr39, 0, is_signed);
break;
}
case OpCode::Id::XMAD_RC: {
op_b += regs.GetRegisterAsInteger(instr.gpr39, 0, instr.xmad.sign_b);
op_c += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
is_signed ? GLSLRegister::Type::Integer
: GLSLRegister::Type::UnsignedInteger);
break;
}
case OpCode::Id::XMAD_IMM: {
is_merge = instr.xmad.merge_37;
op_b += std::to_string(instr.xmad.imm20_16);
op_c += regs.GetRegisterAsInteger(instr.gpr39, 0, is_signed);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled XMAD instruction: {}", opcode->get().GetName());
}
}
// TODO(bunnei): Ensure this is right with signed operands
if (instr.xmad.high_a) {
op_a = "((" + op_a + ") >> 16)";
} else {
op_a = "((" + op_a + ") & 0xFFFF)";
}
std::string src2 = '(' + op_b + ')'; // Preserve original source 2
if (instr.xmad.high_b) {
op_b = '(' + src2 + " >> 16)";
} else {
op_b = '(' + src2 + " & 0xFFFF)";
}
std::string product = '(' + op_a + " * " + op_b + ')';
if (instr.xmad.product_shift_left) {
product = '(' + product + " << 16)";
}
switch (instr.xmad.mode) {
case Tegra::Shader::XmadMode::None:
break;
case Tegra::Shader::XmadMode::CLo:
op_c = "((" + op_c + ") & 0xFFFF)";
break;
case Tegra::Shader::XmadMode::CHi:
op_c = "((" + op_c + ") >> 16)";
break;
case Tegra::Shader::XmadMode::CBcc:
op_c = "((" + op_c + ") + (" + src2 + "<< 16))";
break;
default: {
UNIMPLEMENTED_MSG("Unhandled XMAD mode: {}",
static_cast<u32>(instr.xmad.mode.Value()));
}
}
std::string sum{'(' + product + " + " + op_c + ')'};
if (is_merge) {
sum = "((" + sum + " & 0xFFFF) | (" + src2 + "<< 16))";
}
regs.SetRegisterToInteger(instr.gpr0, is_signed, 0, sum, 1, 1);
break;
}
default: {
switch (opcode->get().GetId()) {
case OpCode::Id::EXIT: {
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T,
"EXIT condition code used: {}", static_cast<u32>(cc));
if (stage == Maxwell3D::Regs::ShaderStage::Fragment) {
EmitFragmentOutputsWrite();
}
switch (instr.flow.cond) {
case Tegra::Shader::FlowCondition::Always:
shader.AddLine("return true;");
if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
// If this is an unconditional exit then just end processing here,
// otherwise we have to account for the possibility of the condition
// not being met, so continue processing the next instruction.
offset = PROGRAM_END - 1;
}
break;
case Tegra::Shader::FlowCondition::Fcsm_Tr:
// TODO(bunnei): What is this used for? If we assume this conditon is not
// satisifed, dual vertex shaders in Farming Simulator make more sense
UNIMPLEMENTED_MSG("Skipping unknown FlowCondition::Fcsm_Tr");
break;
default:
UNIMPLEMENTED_MSG("Unhandled flow condition: {}",
static_cast<u32>(instr.flow.cond.Value()));
}
break;
}
case OpCode::Id::KIL: {
UNIMPLEMENTED_IF(instr.flow.cond != Tegra::Shader::FlowCondition::Always);
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T,
"KIL condition code used: {}", static_cast<u32>(cc));
// Enclose "discard" in a conditional, so that GLSL compilation does not complain
// about unexecuted instructions that may follow this.
shader.AddLine("if (true) {");
++shader.scope;
shader.AddLine("discard;");
--shader.scope;
shader.AddLine("}");
break;
}
case OpCode::Id::OUT_R: {
UNIMPLEMENTED_IF_MSG(instr.gpr20.Value() != Register::ZeroIndex,
"Stream buffer is not supported");
ASSERT_MSG(stage == Maxwell3D::Regs::ShaderStage::Geometry,
"OUT is expected to be used in a geometry shader.");
if (instr.out.emit) {
// gpr0 is used to store the next address. Hardware returns a pointer but
// we just return the next index with a cyclic cap.
const std::string current{regs.GetRegisterAsInteger(instr.gpr8, 0, false)};
const std::string next = "((" + current + " + 1" + ") % " +
std::to_string(MAX_GEOMETRY_BUFFERS) + ')';
shader.AddLine("emit_vertex(" + current + ");");
regs.SetRegisterToInteger(instr.gpr0, false, 0, next, 1, 1);
}
if (instr.out.cut) {
shader.AddLine("EndPrimitive();");
}
break;
}
case OpCode::Id::MOV_SYS: {
switch (instr.sys20) {
case Tegra::Shader::SystemVariable::InvocationInfo: {
LOG_WARNING(HW_GPU, "MOV_SYS instruction with InvocationInfo is incomplete");
regs.SetRegisterToInteger(instr.gpr0, false, 0, "0u", 1, 1);
break;
}
default: {
UNIMPLEMENTED_MSG("Unhandled system move: {}",
static_cast<u32>(instr.sys20.Value()));
}
}
break;
}
case OpCode::Id::ISBERD: {
UNIMPLEMENTED_IF(instr.isberd.o != 0);
UNIMPLEMENTED_IF(instr.isberd.skew != 0);
UNIMPLEMENTED_IF(instr.isberd.shift != Tegra::Shader::IsberdShift::None);
UNIMPLEMENTED_IF(instr.isberd.mode != Tegra::Shader::IsberdMode::None);
ASSERT_MSG(stage == Maxwell3D::Regs::ShaderStage::Geometry,
"ISBERD is expected to be used in a geometry shader.");
LOG_WARNING(HW_GPU, "ISBERD instruction is incomplete");
regs.SetRegisterToFloat(instr.gpr0, 0, regs.GetRegisterAsFloat(instr.gpr8), 1, 1);
break;
}
case OpCode::Id::BRA: {
UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
"BRA with constant buffers are not implemented");
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T,
"BRA condition code used: {}", static_cast<u32>(cc));
const u32 target = offset + instr.bra.GetBranchTarget();
shader.AddLine("{ jmp_to = " + std::to_string(target) + "u; break; }");
break;
}
case OpCode::Id::IPA: {
const auto& attribute = instr.attribute.fmt28;
const auto& reg = instr.gpr0;
Tegra::Shader::IpaMode input_mode{instr.ipa.interp_mode.Value(),
instr.ipa.sample_mode.Value()};
regs.SetRegisterToInputAttibute(reg, attribute.element, attribute.index,
input_mode);
if (instr.ipa.saturate) {
regs.SetRegisterToFloat(reg, 0, regs.GetRegisterAsFloat(reg), 1, 1, true);
}
break;
}
case OpCode::Id::SSY: {
// The SSY opcode tells the GPU where to re-converge divergent execution paths, it
// sets the target of the jump that the SYNC instruction will make. The SSY opcode
// has a similar structure to the BRA opcode.
UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
"Constant buffer flow is not supported");
const u32 target = offset + instr.bra.GetBranchTarget();
EmitPushToFlowStack(target);
break;
}
case OpCode::Id::PBK: {
// PBK pushes to a stack the address where BRK will jump to. This shares stack with
// SSY but using SYNC on a PBK address will kill the shader execution. We don't
// emulate this because it's very unlikely a driver will emit such invalid shader.
UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
"Constant buffer PBK is not supported");
const u32 target = offset + instr.bra.GetBranchTarget();
EmitPushToFlowStack(target);
break;
}
case OpCode::Id::SYNC: {
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T,
"SYNC condition code used: {}", static_cast<u32>(cc));
// The SYNC opcode jumps to the address previously set by the SSY opcode
EmitPopFromFlowStack();
break;
}
case OpCode::Id::BRK: {
// The BRK opcode jumps to the address previously set by the PBK opcode
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T,
"BRK condition code used: {}", static_cast<u32>(cc));
EmitPopFromFlowStack();
break;
}
case OpCode::Id::DEPBAR: {
// TODO(Subv): Find out if we actually have to care about this instruction or if
// the GLSL compiler takes care of that for us.
LOG_WARNING(HW_GPU, "DEPBAR instruction is stubbed");
break;
}
case OpCode::Id::VMAD: {
UNIMPLEMENTED_IF_MSG(instr.generates_cc,
"Condition codes generation in VMAD is not implemented");
const bool result_signed = instr.video.signed_a == 1 || instr.video.signed_b == 1;
const std::string op_a = GetVideoOperandA(instr);
const std::string op_b = GetVideoOperandB(instr);
const std::string op_c = regs.GetRegisterAsInteger(instr.gpr39, 0, result_signed);
std::string result = '(' + op_a + " * " + op_b + " + " + op_c + ')';
switch (instr.vmad.shr) {
case Tegra::Shader::VmadShr::Shr7:
result = '(' + result + " >> 7)";
break;
case Tegra::Shader::VmadShr::Shr15:
result = '(' + result + " >> 15)";
break;
}
regs.SetRegisterToInteger(instr.gpr0, result_signed, 1, result, 1, 1,
instr.vmad.saturate == 1, 0, Register::Size::Word,
instr.vmad.cc);
break;
}
case OpCode::Id::VSETP: {
const std::string op_a = GetVideoOperandA(instr);
const std::string op_b = GetVideoOperandB(instr);
// We can't use the constant predicate as destination.
ASSERT(instr.vsetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
const std::string second_pred = GetPredicateCondition(instr.vsetp.pred39, false);
const std::string combiner = GetPredicateCombiner(instr.vsetp.op);
const std::string predicate = GetPredicateComparison(instr.vsetp.cond, op_a, op_b);
// Set the primary predicate to the result of Predicate OP SecondPredicate
SetPredicate(instr.vsetp.pred3,
'(' + predicate + ") " + combiner + " (" + second_pred + ')');
if (instr.vsetp.pred0 != static_cast<u64>(Pred::UnusedIndex)) {
// Set the secondary predicate to the result of !Predicate OP SecondPredicate,
// if enabled
SetPredicate(instr.vsetp.pred0,
"!(" + predicate + ") " + combiner + " (" + second_pred + ')');
}
break;
}
default: { UNIMPLEMENTED_MSG("Unhandled instruction: {}", opcode->get().GetName()); }
}
break;
}
}
// Close the predicate condition scope.
if (can_be_predicated && instr.pred.pred_index != static_cast<u64>(Pred::UnusedIndex)) {
--shader.scope;
shader.AddLine('}');
}
return offset + 1;
}
/**
* Compiles a range of instructions from Tegra to GLSL.
* @param begin the offset of the starting instruction.
* @param end the offset where the compilation should stop (exclusive).
* @return the offset of the next instruction to compile. PROGRAM_END if the program
* terminates.
*/
u32 CompileRange(u32 begin, u32 end) {
u32 program_counter;
for (program_counter = begin; program_counter < (begin > end ? PROGRAM_END : end);) {
program_counter = CompileInstr(program_counter);
}
return program_counter;
}
void Generate(const std::string& suffix) {
// Add declarations for all subroutines
for (const auto& subroutine : subroutines) {
shader.AddLine("bool " + subroutine.GetName() + "();");
}
shader.AddNewLine();
// Add the main entry point
shader.AddLine("bool exec_" + suffix + "() {");
++shader.scope;
CallSubroutine(GetSubroutine(main_offset, PROGRAM_END));
--shader.scope;
shader.AddLine("}\n");
// Add definitions for all subroutines
for (const auto& subroutine : subroutines) {
std::set<u32> labels = subroutine.labels;
shader.AddLine("bool " + subroutine.GetName() + "() {");
++shader.scope;
if (labels.empty()) {
if (CompileRange(subroutine.begin, subroutine.end) != PROGRAM_END) {
shader.AddLine("return false;");
}
} else {
labels.insert(subroutine.begin);
shader.AddLine("uint jmp_to = " + std::to_string(subroutine.begin) + "u;");
// TODO(Subv): Figure out the actual depth of the flow stack, for now it seems
// unlikely that shaders will use 20 nested SSYs and PBKs.
constexpr u32 FLOW_STACK_SIZE = 20;
shader.AddLine("uint flow_stack[" + std::to_string(FLOW_STACK_SIZE) + "];");
shader.AddLine("uint flow_stack_top = 0u;");
shader.AddLine("while (true) {");
++shader.scope;
shader.AddLine("switch (jmp_to) {");
for (auto label : labels) {
shader.AddLine("case " + std::to_string(label) + "u: {");
++shader.scope;
const auto next_it = labels.lower_bound(label + 1);
const u32 next_label = next_it == labels.end() ? subroutine.end : *next_it;
const u32 compile_end = CompileRange(label, next_label);
if (compile_end > next_label && compile_end != PROGRAM_END) {
// This happens only when there is a label inside a IF/LOOP block
shader.AddLine(" jmp_to = " + std::to_string(compile_end) + "u; break; }");
labels.emplace(compile_end);
}
--shader.scope;
shader.AddLine('}');
}
shader.AddLine("default: return false;");
shader.AddLine('}');
--shader.scope;
shader.AddLine('}');
shader.AddLine("return false;");
}
--shader.scope;
shader.AddLine("}\n");
DEBUG_ASSERT(shader.scope == 0);
}
GenerateDeclarations();
}
/// Add declarations for registers
void GenerateDeclarations() {
regs.GenerateDeclarations(suffix);
for (const auto& pred : declr_predicates) {
declarations.AddLine("bool " + pred + " = false;");
}
declarations.AddNewLine();
}
private:
const std::set<Subroutine>& subroutines;
const ProgramCode& program_code;
Tegra::Shader::Header header;
const u32 main_offset;
Maxwell3D::Regs::ShaderStage stage;
const std::string& suffix;
u64 local_memory_size;
std::size_t shader_length;
ShaderWriter shader;
ShaderWriter declarations;
GLSLRegisterManager regs{shader, declarations, stage, suffix, header};
// Declarations
std::set<std::string> declr_predicates;
}; // namespace OpenGL::GLShader::Decompiler
std::string GetCommonDeclarations() {
return fmt::format("#define MAX_CONSTBUFFER_ELEMENTS {}\n",
RasterizerOpenGL::MaxConstbufferSize / sizeof(GLvec4));
}
std::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
Maxwell3D::Regs::ShaderStage stage,
const std::string& suffix) {
try {
ControlFlowAnalyzer analyzer(program_code, main_offset, suffix);
const auto subroutines = analyzer.GetSubroutines();
GLSLGenerator generator(subroutines, program_code, main_offset, stage, suffix,
analyzer.GetShaderLength());
return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
} catch (const DecompileFail& exception) {
LOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());
}
return {};
}
} // namespace OpenGL::GLShader::Decompiler
|