summaryrefslogtreecommitdiffstats
path: root/js/src/jit/MacroAssembler.h
blob: 361de3ac5f622dde11555ad037bc2be12ba03c67 (plain)
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
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*-
 * vim: set ts=8 sts=2 et sw=2 tw=80:
 * This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

#ifndef jit_MacroAssembler_h
#define jit_MacroAssembler_h

#include "mozilla/EndianUtils.h"
#include "mozilla/MacroForEach.h"
#include "mozilla/MathAlgorithms.h"
#include "mozilla/Maybe.h"
#include "mozilla/Variant.h"

#include <utility>

#if defined(JS_CODEGEN_X86)
#  include "jit/x86/MacroAssembler-x86.h"
#elif defined(JS_CODEGEN_X64)
#  include "jit/x64/MacroAssembler-x64.h"
#elif defined(JS_CODEGEN_ARM)
#  include "jit/arm/MacroAssembler-arm.h"
#elif defined(JS_CODEGEN_ARM64)
#  include "jit/arm64/MacroAssembler-arm64.h"
#elif defined(JS_CODEGEN_MIPS32)
#  include "jit/mips32/MacroAssembler-mips32.h"
#elif defined(JS_CODEGEN_MIPS64)
#  include "jit/mips64/MacroAssembler-mips64.h"
#elif defined(JS_CODEGEN_LOONG64)
#  include "jit/loong64/MacroAssembler-loong64.h"
#elif defined(JS_CODEGEN_RISCV64)
#  include "jit/riscv64/MacroAssembler-riscv64.h"
#elif defined(JS_CODEGEN_WASM32)
#  include "jit/wasm32/MacroAssembler-wasm32.h"
#elif defined(JS_CODEGEN_NONE)
#  include "jit/none/MacroAssembler-none.h"
#else
#  error "Unknown architecture!"
#endif
#include "jit/ABIArgGenerator.h"
#include "jit/ABIFunctions.h"
#include "jit/AtomicOp.h"
#include "jit/IonTypes.h"
#include "jit/MoveResolver.h"
#include "jit/VMFunctions.h"
#include "js/ScalarType.h"  // js::Scalar::Type
#include "util/Memory.h"
#include "vm/FunctionFlags.h"
#include "vm/Opcodes.h"
#include "vm/RealmFuses.h"
#include "wasm/WasmCodegenTypes.h"
#include "wasm/WasmFrame.h"

// [SMDOC] MacroAssembler multi-platform overview
//
// * How to read/write MacroAssembler method declarations:
//
// The following macros are made to avoid #ifdef around each method declarations
// of the Macro Assembler, and they are also used as an hint on the location of
// the implementations of each method.  For example, the following declaration
//
//   void Pop(FloatRegister t) DEFINED_ON(x86_shared, arm);
//
// suggests the MacroAssembler::Pop(FloatRegister) method is implemented in
// x86-shared/MacroAssembler-x86-shared.h, and also in arm/MacroAssembler-arm.h.
//
// - If there is no annotation, then there is only one generic definition in
//   MacroAssembler.cpp.
//
// - If the declaration is "inline", then the method definition(s) would be in
//   the "-inl.h" variant of the same file(s).
//
// The script check_macroassembler_style.py (which runs on every build) is
// used to verify that method definitions match the annotation on the method
// declarations.  If there is any difference, then you either forgot to define
// the method in one of the macro assembler, or you forgot to update the
// annotation of the macro assembler declaration.
//
// Some convenient short-cuts are used to avoid repeating the same list of
// architectures on each method declaration, such as PER_ARCH and
// PER_SHARED_ARCH.
//
// Functions that are architecture-agnostic and are the same for all
// architectures, that it's necessary to define inline *in this header* to
// avoid used-before-defined warnings/errors that would occur if the
// definitions were in MacroAssembler-inl.h, should use the OOL_IN_HEADER
// marker at end of the declaration:
//
//   inline uint32_t framePushed() const OOL_IN_HEADER;
//
// Such functions should then be defined immediately after MacroAssembler's
// definition, for example:
//
//   //{{{ check_macroassembler_style
//   inline uint32_t
//   MacroAssembler::framePushed() const
//   {
//       return framePushed_;
//   }
//   ////}}} check_macroassembler_style

#define ALL_ARCH mips32, mips64, arm, arm64, x86, x64, loong64, riscv64, wasm32
#define ALL_SHARED_ARCH \
  arm, arm64, loong64, riscv64, x86_shared, mips_shared, wasm32

// * How this macro works:
//
// DEFINED_ON is a macro which check if, for the current architecture, the
// method is defined on the macro assembler or not.
//
// For each architecture, we have a macro named DEFINED_ON_arch.  This macro is
// empty if this is not the current architecture.  Otherwise it must be either
// set to "define" or "crash" (only used for the none target so far).
//
// The DEFINED_ON macro maps the list of architecture names given as arguments
// to a list of macro names.  For example,
//
//   DEFINED_ON(arm, x86_shared)
//
// is expanded to
//
//   DEFINED_ON_none DEFINED_ON_arm DEFINED_ON_x86_shared
//
// which are later expanded on ARM, x86, x64 by DEFINED_ON_EXPAND_ARCH_RESULTS
// to
//
//   define
//
// or if the JIT is disabled or set to no architecture to
//
//   crash
//
// or to nothing, if the current architecture is not listed in the list of
// arguments of DEFINED_ON.  Note, only one of the DEFINED_ON_arch macro
// contributes to the non-empty result, which is the macro of the current
// architecture if it is listed in the arguments of DEFINED_ON.
//
// This result is appended to DEFINED_ON_RESULT_ before expanding the macro,
// which results in either no annotation, a MOZ_CRASH(), or a "= delete"
// annotation on the method declaration.

#define DEFINED_ON_x86
#define DEFINED_ON_x64
#define DEFINED_ON_x86_shared
#define DEFINED_ON_arm
#define DEFINED_ON_arm64
#define DEFINED_ON_mips32
#define DEFINED_ON_mips64
#define DEFINED_ON_mips_shared
#define DEFINED_ON_loong64
#define DEFINED_ON_riscv64
#define DEFINED_ON_wasm32
#define DEFINED_ON_none

// Specialize for each architecture.
#if defined(JS_CODEGEN_X86)
#  undef DEFINED_ON_x86
#  define DEFINED_ON_x86 define
#  undef DEFINED_ON_x86_shared
#  define DEFINED_ON_x86_shared define
#elif defined(JS_CODEGEN_X64)
#  undef DEFINED_ON_x64
#  define DEFINED_ON_x64 define
#  undef DEFINED_ON_x86_shared
#  define DEFINED_ON_x86_shared define
#elif defined(JS_CODEGEN_ARM)
#  undef DEFINED_ON_arm
#  define DEFINED_ON_arm define
#elif defined(JS_CODEGEN_ARM64)
#  undef DEFINED_ON_arm64
#  define DEFINED_ON_arm64 define
#elif defined(JS_CODEGEN_MIPS32)
#  undef DEFINED_ON_mips32
#  define DEFINED_ON_mips32 define
#  undef DEFINED_ON_mips_shared
#  define DEFINED_ON_mips_shared define
#elif defined(JS_CODEGEN_MIPS64)
#  undef DEFINED_ON_mips64
#  define DEFINED_ON_mips64 define
#  undef DEFINED_ON_mips_shared
#  define DEFINED_ON_mips_shared define
#elif defined(JS_CODEGEN_LOONG64)
#  undef DEFINED_ON_loong64
#  define DEFINED_ON_loong64 define
#elif defined(JS_CODEGEN_RISCV64)
#  undef DEFINED_ON_riscv64
#  define DEFINED_ON_riscv64 define
#elif defined(JS_CODEGEN_WASM32)
#  undef DEFINED_ON_wasm32
#  define DEFINED_ON_wasm32 define
#elif defined(JS_CODEGEN_NONE)
#  undef DEFINED_ON_none
#  define DEFINED_ON_none crash
#else
#  error "Unknown architecture!"
#endif

#define DEFINED_ON_RESULT_crash \
  { MOZ_CRASH(); }
#define DEFINED_ON_RESULT_define
#define DEFINED_ON_RESULT_ = delete

#define DEFINED_ON_DISPATCH_RESULT_2(Macro, Result) Macro##Result
#define DEFINED_ON_DISPATCH_RESULT(...) \
  DEFINED_ON_DISPATCH_RESULT_2(DEFINED_ON_RESULT_, __VA_ARGS__)

// We need to let the evaluation of MOZ_FOR_EACH terminates.
#define DEFINED_ON_EXPAND_ARCH_RESULTS_3(ParenResult) \
  DEFINED_ON_DISPATCH_RESULT ParenResult
#define DEFINED_ON_EXPAND_ARCH_RESULTS_2(ParenResult) \
  DEFINED_ON_EXPAND_ARCH_RESULTS_3(ParenResult)
#define DEFINED_ON_EXPAND_ARCH_RESULTS(ParenResult) \
  DEFINED_ON_EXPAND_ARCH_RESULTS_2(ParenResult)

#define DEFINED_ON_FWDARCH(Arch) DEFINED_ON_##Arch
#define DEFINED_ON_MAP_ON_ARCHS(ArchList) \
  DEFINED_ON_EXPAND_ARCH_RESULTS(         \
      (MOZ_FOR_EACH(DEFINED_ON_FWDARCH, (), ArchList)))

#define DEFINED_ON(...) DEFINED_ON_MAP_ON_ARCHS((none, __VA_ARGS__))

#define PER_ARCH DEFINED_ON(ALL_ARCH)
#define PER_SHARED_ARCH DEFINED_ON(ALL_SHARED_ARCH)
#define OOL_IN_HEADER

class JSLinearString;

namespace JS {
struct ExpandoAndGeneration;
}

namespace js {

class StaticStrings;
class FixedLengthTypedArrayObject;

enum class NativeIteratorIndices : uint32_t;

namespace wasm {
class CalleeDesc;
class CallSiteDesc;
class BytecodeOffset;
class MemoryAccessDesc;

struct ModuleEnvironment;

enum class FailureMode : uint8_t;
enum class SimdOp;
enum class SymbolicAddress;
enum class Trap;
}  // namespace wasm

namespace jit {

// Defined in JitFrames.h
enum class ExitFrameType : uint8_t;

class AutoSaveLiveRegisters;
class CompileZone;
class TemplateNativeObject;
class TemplateObject;

enum class CheckUnsafeCallWithABI {
  // Require the callee to use AutoUnsafeCallWithABI.
  Check,

  // We pushed an exit frame so this callWithABI can safely GC and walk the
  // stack.
  DontCheckHasExitFrame,

  // Don't check this callWithABI uses AutoUnsafeCallWithABI, for instance
  // because we're calling a simple helper function (like malloc or js_free)
  // that we can't change and/or that we know won't GC.
  DontCheckOther,
};

// This is a global function made to create the DynFn type in a controlled
// environment which would check if the function signature has been registered
// as an ABI function signature.
template <typename Sig>
static inline DynFn DynamicFunction(Sig fun);

enum class CharEncoding { Latin1, TwoByte };

constexpr uint32_t WasmCallerInstanceOffsetBeforeCall =
    wasm::FrameWithInstances::callerInstanceOffsetWithoutFrame();
constexpr uint32_t WasmCalleeInstanceOffsetBeforeCall =
    wasm::FrameWithInstances::calleeInstanceOffsetWithoutFrame();

// Allocation sites may be passed to GC thing allocation methods either via a
// register (for baseline compilation) or an enum indicating one of the
// catch-all allocation sites (for optimized compilation).
struct AllocSiteInput
    : public mozilla::Variant<Register, gc::CatchAllAllocSite> {
  using Base = mozilla::Variant<Register, gc::CatchAllAllocSite>;
  AllocSiteInput() : Base(gc::CatchAllAllocSite::Unknown) {}
  explicit AllocSiteInput(gc::CatchAllAllocSite catchAll) : Base(catchAll) {}
  explicit AllocSiteInput(Register reg) : Base(reg) {}
};

#ifdef ENABLE_WASM_TAIL_CALLS
// Instance slots (including ShadowStackArea) and arguments size information
// from two neighboring frames.
// Used in Wasm tail calls to remove frame.
struct ReturnCallAdjustmentInfo {
  uint32_t newSlotsAndStackArgBytes;
  uint32_t oldSlotsAndStackArgBytes;

  ReturnCallAdjustmentInfo(uint32_t newSlotsAndStackArgBytes,
                           uint32_t oldSlotsAndStackArgBytes)
      : newSlotsAndStackArgBytes(newSlotsAndStackArgBytes),
        oldSlotsAndStackArgBytes(oldSlotsAndStackArgBytes) {}
};
#endif  // ENABLE_WASM_TAIL_CALLS

struct BranchWasmRefIsSubtypeRegisters {
  bool needSuperSTV;
  bool needScratch1;
  bool needScratch2;
};

// [SMDOC] Code generation invariants (incomplete)
//
// ## 64-bit GPRs carrying 32-bit values
//
// At least at the end of every JS or Wasm operation (= SpiderMonkey bytecode or
// Wasm bytecode; this is necessarily a little vague), if a 64-bit GPR has a
// 32-bit value, then the upper 32 bits of the register may be predictable in
// accordance with platform-specific rules, as follows.
//
// - On x64 and arm64, the upper bits are zero
// - On mips64 and loongarch64 the upper bits are the sign extension of the
//   lower bits
// - (On risc-v we have no rule, having no port yet.  Sign extension is the most
//   likely rule, but "unpredictable" is an option.)
//
// In most cases no extra work needs to be done to maintain the invariant:
//
// - 32-bit operations on x64 and arm64 zero-extend the result to 64 bits.
//   These operations ignore the upper bits of the inputs.
// - 32-bit operations on mips64 sign-extend the result to 64 bits (even many
//   that are labeled as "unsigned", eg ADDU, though not all, eg LU).
//   Additionally, the inputs to many 32-bit operations must be properly
//   sign-extended to avoid "unpredictable" behavior, and our simulators check
//   that inputs conform.
// - (32-bit operations on risc-v and loongarch64 sign-extend, much as mips, but
//   appear to ignore the upper bits of the inputs.)
//
// The upshot of these invariants is, among other things, that:
//
// - No code needs to be generated when a 32-bit value is extended to 64 bits
//   or a 64-bit value is wrapped to 32 bits, if the upper bits are known to be
//   correct because they resulted from an operation that produced them
//   predictably.
// - Literal loads must be careful to avoid instructions that might extend the
//   literal in the wrong way.
// - Code that produces values using intermediate values with non-canonical
//   extensions must extend according to platform conventions before being
//   "done".
//
// All optimizations are necessarily platform-specific and should only be used
// in platform-specific code.  We may add architectures in the future that do
// not follow the patterns of the few architectures we already have.
//
// Also see MacroAssembler::debugAssertCanonicalInt32().

// The public entrypoint for emitting assembly. Note that a MacroAssembler can
// use cx->lifoAlloc, so take care not to interleave masm use with other
// lifoAlloc use if one will be destroyed before the other.
class MacroAssembler : public MacroAssemblerSpecific {
 private:
  // Information about the current JSRuntime. This is nullptr only for Wasm
  // compilations.
  CompileRuntime* maybeRuntime_ = nullptr;

  // Information about the current Realm. This is nullptr for Wasm compilations
  // and when compiling JitRuntime trampolines.
  CompileRealm* maybeRealm_ = nullptr;

  // Labels for handling exceptions and failures.
  NonAssertingLabel failureLabel_;

 protected:
  // Constructor is protected. Use one of the derived classes!
  explicit MacroAssembler(TempAllocator& alloc,
                          CompileRuntime* maybeRuntime = nullptr,
                          CompileRealm* maybeRealm = nullptr);

 public:
  MoveResolver& moveResolver() {
    // As an optimization, the MoveResolver is a persistent data structure
    // shared between visitors in the CodeGenerator. This assertion
    // checks that state is not leaking from visitor to visitor
    // via an unresolved addMove().
    MOZ_ASSERT(moveResolver_.hasNoPendingMoves());
    return moveResolver_;
  }

  size_t instructionsSize() const { return size(); }

  CompileRealm* realm() const {
    MOZ_ASSERT(maybeRealm_);
    return maybeRealm_;
  }
  CompileRuntime* runtime() const {
    MOZ_ASSERT(maybeRuntime_);
    return maybeRuntime_;
  }

#ifdef JS_HAS_HIDDEN_SP
  void Push(RegisterOrSP reg);
#endif

#ifdef ENABLE_WASM_SIMD
  // `op` should be a shift operation. Return true if a variable-width shift
  // operation on this architecture should pre-mask the shift count, and if so,
  // return the mask in `*mask`.
  static bool MustMaskShiftCountSimd128(wasm::SimdOp op, int32_t* mask);
#endif

  //{{{ check_macroassembler_decl_style
 public:
  // ===============================================================
  // MacroAssembler high-level usage.

  // Flushes the assembly buffer, on platforms that need it.
  void flush() PER_SHARED_ARCH;

  // Add a comment that is visible in the pretty printed assembly code.
  void comment(const char* msg) PER_SHARED_ARCH;

  // ===============================================================
  // Frame manipulation functions.

  inline uint32_t framePushed() const OOL_IN_HEADER;
  inline void setFramePushed(uint32_t framePushed) OOL_IN_HEADER;
  inline void adjustFrame(int32_t value) OOL_IN_HEADER;

  // Adjust the frame, to account for implicit modification of the stack
  // pointer, such that callee can remove arguments on the behalf of the
  // caller.
  inline void implicitPop(uint32_t bytes) OOL_IN_HEADER;

 private:
  // This field is used to statically (at compilation time) emulate a frame
  // pointer by keeping track of stack manipulations.
  //
  // It is maintained by all stack manipulation functions below.
  uint32_t framePushed_;

 public:
  // ===============================================================
  // Stack manipulation functions -- sets of registers.

  // Approximately speaking, the following routines must use the same memory
  // layout.  Any inconsistencies will certainly lead to crashing in generated
  // code:
  //
  //   MacroAssembler::PushRegsInMaskSizeInBytes
  //   MacroAssembler::PushRegsInMask
  //   MacroAssembler::storeRegsInMask
  //   MacroAssembler::PopRegsInMask
  //   MacroAssembler::PopRegsInMaskIgnore
  //   FloatRegister::getRegisterDumpOffsetInBytes
  //   (no class) PushRegisterDump
  //   (union) RegisterContent
  //   JitRuntime::generateInvalidator
  //   JitRuntime::generateBailoutHandler
  //   JSJitFrameIter::machineState
  //
  // To be more exact, the invariants are:
  //
  // * The save area is conceptually viewed as starting at a highest address
  //   (really, at "highest address - 1") and working down to some lower
  //   address.
  //
  // * PushRegsInMask, storeRegsInMask and PopRegsInMask{Ignore} must use
  //   exactly the same memory layout, when starting from the abovementioned
  //   highest address.
  //
  // * PushRegsInMaskSizeInBytes must produce a value which is exactly equal
  //   to the change in the machine's stack pointer register as a result of
  //   calling PushRegsInMask or PopRegsInMask{Ignore}.  This value must be at
  //   least uintptr_t-aligned on the target, and may be more aligned than that.
  //
  // * PushRegsInMaskSizeInBytes must produce a value which is greater than or
  //   equal to the amount of space used by storeRegsInMask.
  //
  // * Hence, regardless of whether the save area is created with
  //   storeRegsInMask or PushRegsInMask, it is guaranteed to fit inside an
  //   area of size calculated by PushRegsInMaskSizeInBytes.
  //
  // * For the `ignore` argument of PopRegsInMaskIgnore, equality checking
  //   for the floating point/SIMD registers is done on the basis of the
  //   underlying physical register, regardless of width.  For example, if the
  //   to-restore set contains v17 (the SIMD register with encoding 17) and
  //   the ignore set contains d17 (the double register with encoding 17) then
  //   no part of the physical register with encoding 17 will be restored.
  //   (This is probably not true on arm32, since that has aliased float32
  //   registers; but none of our other targets do.)
  //
  // * {Push,store}RegsInMask/storeRegsInMask are further constrained as
  //   follows: when given the argument AllFloatRegisters, the resulting
  //   memory area must contain exactly all the SIMD/FP registers for the
  //   target at their widest width (that we care about).  [We have no targets
  //   where the SIMD registers and FP register sets are disjoint.]  They must
  //   be packed end-to-end with no holes, with the register with the lowest
  //   encoding number (0), as returned by FloatRegister::encoding(), at the
  //   abovementioned highest address, register 1 just below that, etc.
  //
  //   Furthermore the sizeof(RegisterContent) must equal the size of a SIMD
  //   register in the abovementioned array.
  //
  //   Furthermore the value returned by
  //   FloatRegister::getRegisterDumpOffsetInBytes must be a correct index
  //   into the abovementioned array.  Given the constraints, the only correct
  //   value is `reg.encoding() * sizeof(RegisterContent)`.
  //
  // Note that some of the routines listed above are JS-only, and do not support
  // SIMD registers. They are otherwise part of the same equivalence class.
  // Register spilling for e.g. OOL VM calls is implemented using
  // PushRegsInMask, and recovered on bailout using machineState. This requires
  // the same layout to be used in machineState, and therefore in all other code
  // that can spill registers that are recovered on bailout. Implementations of
  // JitRuntime::generate{Invalidator,BailoutHandler} should either call
  // PushRegsInMask, or check carefully to be sure that they generate the same
  // layout.

  // The size of the area used by PushRegsInMask.
  static size_t PushRegsInMaskSizeInBytes(LiveRegisterSet set)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  void PushRegsInMask(LiveRegisterSet set)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);
  void PushRegsInMask(LiveGeneralRegisterSet set);

  // Like PushRegsInMask, but instead of pushing the registers, store them to
  // |dest|. |dest| should point to the end of the reserved space, so the
  // first register will be stored at |dest.offset - sizeof(register)|.  It is
  // required that |dest.offset| is at least as large as the value computed by
  // PushRegsInMaskSizeInBytes for this |set|.  In other words, |dest.base|
  // must point to either the lowest address in the save area, or some address
  // below that.
  void storeRegsInMask(LiveRegisterSet set, Address dest, Register scratch)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  void PopRegsInMask(LiveRegisterSet set);
  void PopRegsInMask(LiveGeneralRegisterSet set);
  void PopRegsInMaskIgnore(LiveRegisterSet set, LiveRegisterSet ignore)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  // ===============================================================
  // Stack manipulation functions -- single registers/values.

  void Push(const Operand op) DEFINED_ON(x86_shared);
  void Push(Register reg) PER_SHARED_ARCH;
  void Push(Register reg1, Register reg2, Register reg3, Register reg4)
      DEFINED_ON(arm64);
  void Push(const Imm32 imm) PER_SHARED_ARCH;
  void Push(const ImmWord imm) PER_SHARED_ARCH;
  void Push(const ImmPtr imm) PER_SHARED_ARCH;
  void Push(const ImmGCPtr ptr) PER_SHARED_ARCH;
  void Push(FloatRegister reg) PER_SHARED_ARCH;
  void PushBoxed(FloatRegister reg) PER_ARCH;
  void PushFlags() DEFINED_ON(x86_shared);
  void Push(PropertyKey key, Register scratchReg);
  void Push(const Address& addr);
  void Push(TypedOrValueRegister v);
  void Push(const ConstantOrRegister& v);
  void Push(const ValueOperand& val);
  void Push(const Value& val);
  void Push(JSValueType type, Register reg);
  void Push(const Register64 reg);
  void PushEmptyRooted(VMFunctionData::RootType rootType);
  inline CodeOffset PushWithPatch(ImmWord word);
  inline CodeOffset PushWithPatch(ImmPtr imm);

  void Pop(const Operand op) DEFINED_ON(x86_shared);
  void Pop(Register reg) PER_SHARED_ARCH;
  void Pop(FloatRegister t) PER_SHARED_ARCH;
  void Pop(const ValueOperand& val) PER_SHARED_ARCH;
  void PopFlags() DEFINED_ON(x86_shared);
  void PopStackPtr()
      DEFINED_ON(arm, mips_shared, x86_shared, loong64, riscv64, wasm32);

  // Move the stack pointer based on the requested amount.
  void adjustStack(int amount);
  void freeStack(uint32_t amount);

  // Move the stack pointer to the specified position. It assumes the SP
  // register is not valid -- it uses FP to set the position.
  void freeStackTo(uint32_t framePushed)
      DEFINED_ON(x86_shared, arm, arm64, loong64, mips64, riscv64);

  // Warning: This method does not update the framePushed() counter.
  void freeStack(Register amount);

 private:
  // ===============================================================
  // Register allocation fields.
#ifdef DEBUG
  friend AutoRegisterScope;
  friend AutoFloatRegisterScope;
  // Used to track register scopes for debug builds.
  // Manipulated by the AutoGenericRegisterScope class.
  AllocatableRegisterSet debugTrackedRegisters_;
#endif  // DEBUG

 public:
  // ===============================================================
  // Simple call functions.

  // The returned CodeOffset is the assembler offset for the instruction
  // immediately following the call; that is, for the return point.
  CodeOffset call(Register reg) PER_SHARED_ARCH;
  CodeOffset call(Label* label) PER_SHARED_ARCH;

  void call(const Address& addr) PER_SHARED_ARCH;
  void call(ImmWord imm) PER_SHARED_ARCH;
  // Call a target native function, which is neither traceable nor movable.
  void call(ImmPtr imm) PER_SHARED_ARCH;
  CodeOffset call(wasm::SymbolicAddress imm) PER_SHARED_ARCH;
  inline CodeOffset call(const wasm::CallSiteDesc& desc,
                         wasm::SymbolicAddress imm);

  // Call a target JitCode, which must be traceable, and may be movable.
  void call(JitCode* c) PER_SHARED_ARCH;

  inline void call(TrampolinePtr code);

  inline CodeOffset call(const wasm::CallSiteDesc& desc, const Register reg);
  inline CodeOffset call(const wasm::CallSiteDesc& desc, uint32_t funcDefIndex);
  inline void call(const wasm::CallSiteDesc& desc, wasm::Trap trap);

  CodeOffset callWithPatch() PER_SHARED_ARCH;
  void patchCall(uint32_t callerOffset, uint32_t calleeOffset) PER_SHARED_ARCH;

  // Push the return address and make a call. On platforms where this function
  // is not defined, push the link register (pushReturnAddress) at the entry
  // point of the callee.
  void callAndPushReturnAddress(Register reg) DEFINED_ON(x86_shared);
  void callAndPushReturnAddress(Label* label) DEFINED_ON(x86_shared);

  // These do not adjust framePushed().
  void pushReturnAddress()
      DEFINED_ON(mips_shared, arm, arm64, loong64, riscv64, wasm32);
  void popReturnAddress()
      DEFINED_ON(mips_shared, arm, arm64, loong64, riscv64, wasm32);

  // Useful for dealing with two-valued returns.
  void moveRegPair(Register src0, Register src1, Register dst0, Register dst1,
                   MoveOp::Type type = MoveOp::GENERAL);

  void reserveVMFunctionOutParamSpace(const VMFunctionData& f);
  void loadVMFunctionOutParam(const VMFunctionData& f, const Address& addr);

 public:
  // ===============================================================
  // Patchable near/far jumps.

  // "Far jumps" provide the ability to jump to any uint32_t offset from any
  // other uint32_t offset without using a constant pool (thus returning a
  // simple CodeOffset instead of a CodeOffsetJump).
  CodeOffset farJumpWithPatch() PER_SHARED_ARCH;
  void patchFarJump(CodeOffset farJump, uint32_t targetOffset) PER_SHARED_ARCH;

  // Emit a nop that can be patched to and from a nop and a call with int32
  // relative displacement.
  CodeOffset nopPatchableToCall() PER_SHARED_ARCH;
  void nopPatchableToCall(const wasm::CallSiteDesc& desc);
  static void patchNopToCall(uint8_t* callsite,
                             uint8_t* target) PER_SHARED_ARCH;
  static void patchCallToNop(uint8_t* callsite) PER_SHARED_ARCH;

  // These methods are like movWithPatch/PatchDataWithValueCheck but allow
  // using pc-relative addressing on certain platforms (RIP-relative LEA on x64,
  // ADR instruction on arm64).
  //
  // Note: "Near" applies to ARM64 where the target must be within 1 MB (this is
  // release-asserted).
  CodeOffset moveNearAddressWithPatch(Register dest) PER_ARCH;
  static void patchNearAddressMove(CodeLocationLabel loc,
                                   CodeLocationLabel target)
      DEFINED_ON(x86, x64, arm, arm64, loong64, riscv64, wasm32, mips_shared);

 public:
  // ===============================================================
  // [SMDOC] JIT-to-C++ Function Calls (callWithABI)
  //
  // callWithABI is used to make a call using the standard C/C++ system ABI.
  //
  // callWithABI is a low level interface for making calls, as such every call
  // made with callWithABI should be organized with 6 steps: spilling live
  // registers, aligning the stack, listing arguments of the called function,
  // calling a function pointer, extracting the returned value and restoring
  // live registers.
  //
  // A more detailed example of the six stages:
  //
  // 1) Saving of registers that are live. This will vary depending on which
  //    SpiderMonkey compiler you are working on. Registers that shouldn't be
  //    restored can be excluded.
  //
  //      LiveRegisterSet volatileRegs(...);
  //      volatileRegs.take(scratch);
  //      masm.PushRegsInMask(volatileRegs);
  //
  // 2) Align the stack to perform the call with the correct stack alignment.
  //
  //    When the stack pointer alignment is unknown and cannot be corrected
  //    when generating the code, setupUnalignedABICall must be used to
  //    dynamically align the stack pointer to the expectation of the ABI.
  //    When the stack pointer is known at JIT compilation time, the stack can
  //    be fixed manually and setupAlignedABICall and setupWasmABICall can be
  //    used.
  //
  //    setupWasmABICall is a special case of setupAlignedABICall as
  //    SpiderMonkey's WebAssembly implementation mostly follow the system
  //    ABI, except for float/double arguments, which always use floating
  //    point registers, even if this is not supported by the system ABI.
  //
  //      masm.setupUnalignedABICall(scratch);
  //
  // 3) Passing arguments. Arguments are passed left-to-right.
  //
  //      masm.passABIArg(scratch);
  //      masm.passABIArg(FloatOp0, ABIType::Float64);
  //
  //    Note how float register arguments are annotated with ABIType::Float64.
  //
  //    Concerning stack-relative address, see the note on passABIArg.
  //
  // 4) Make the call:
  //
  //      using Fn = int32_t (*)(int32_t)
  //      masm.callWithABI<Fn, Callee>();
  //
  //    In the case where the call returns a double, that needs to be
  //    indicated to the callWithABI like this:
  //
  //      using Fn = double (*)(int32_t)
  //      masm.callWithABI<Fn, Callee>(ABIType::Float64);
  //
  //    There are overloads to allow calls to registers and addresses.
  //
  // 5) Take care of the result
  //
  //      masm.storeCallPointerResult(scratch1);
  //      masm.storeCallBoolResult(scratch1);
  //      masm.storeCallInt32Result(scratch1);
  //      masm.storeCallFloatResult(scratch1);
  //
  // 6) Restore the potentially clobbered volatile registers
  //
  //      masm.PopRegsInMask(volatileRegs);
  //
  //    If expecting a returned value, this call should use
  //    PopRegsInMaskIgnore to filter out the registers which are containing
  //    the returned value.
  //
  // Unless an exit frame is pushed prior to the setupABICall, the callee
  // should not GC. To ensure this is the case callWithABI is instrumented to
  // make sure that in the default case callees are annotated with an
  // AutoUnsafeCallWithABI on the stack.
  //
  // A callWithABI can opt out of checking, if for example it is known there
  // is an exit frame, or the callee is known not to GC.
  //
  // If your callee needs to be able to GC, consider using a VMFunction, or
  // create a fake exit frame, and instrument the TraceJitExitFrame
  // accordingly.

  // Setup a call to C/C++ code, given the assumption that the framePushed
  // accurately defines the state of the stack, and that the top of the stack
  // was properly aligned. Note that this only supports cdecl.
  //
  // As a rule of thumb, this can be used in CodeGenerator but not in CacheIR or
  // Baseline code (because the stack is not aligned to ABIStackAlignment).
  void setupAlignedABICall();

  // As setupAlignedABICall, but for WebAssembly native ABI calls, which pass
  // through a builtin thunk that uses the wasm ABI. All the wasm ABI calls
  // can be native, since we always know the stack alignment a priori.
  void setupWasmABICall();

  // Setup an ABI call for when the alignment is not known. This may need a
  // scratch register.
  void setupUnalignedABICall(Register scratch) PER_ARCH;

  // Like setupUnalignedABICall, but more efficient because it doesn't push/pop
  // the unaligned stack pointer. The caller is responsible for restoring SP
  // after the callWithABI, for example using the frame pointer register.
  void setupUnalignedABICallDontSaveRestoreSP();

  // Arguments must be assigned to a C/C++ call in order. They are moved
  // in parallel immediately before performing the call. This process may
  // temporarily use more stack, in which case esp-relative addresses will be
  // automatically adjusted. It is extremely important that esp-relative
  // addresses are computed *after* setupABICall(). Furthermore, no
  // operations should be emitted while setting arguments.
  void passABIArg(const MoveOperand& from, ABIType type);
  inline void passABIArg(Register reg);
  inline void passABIArg(FloatRegister reg, ABIType type);

  inline void callWithABI(
      DynFn fun, ABIType result = ABIType::General,
      CheckUnsafeCallWithABI check = CheckUnsafeCallWithABI::Check);
  template <typename Sig, Sig fun>
  inline void callWithABI(
      ABIType result = ABIType::General,
      CheckUnsafeCallWithABI check = CheckUnsafeCallWithABI::Check);
  inline void callWithABI(Register fun, ABIType result = ABIType::General);
  inline void callWithABI(const Address& fun,
                          ABIType result = ABIType::General);

  CodeOffset callWithABI(wasm::BytecodeOffset offset, wasm::SymbolicAddress fun,
                         mozilla::Maybe<int32_t> instanceOffset,
                         ABIType result = ABIType::General);
  void callDebugWithABI(wasm::SymbolicAddress fun,
                        ABIType result = ABIType::General);

 private:
  // Reinitialize the variables which have to be cleared before making a call
  // with callWithABI.
  template <class ABIArgGeneratorT>
  void setupABICallHelper();

  // Reinitialize the variables which have to be cleared before making a call
  // with native abi.
  void setupNativeABICall();

  // Reserve the stack and resolve the arguments move.
  void callWithABIPre(uint32_t* stackAdjust,
                      bool callFromWasm = false) PER_ARCH;

  // Emits a call to a C/C++ function, resolving all argument moves.
  void callWithABINoProfiler(void* fun, ABIType result,
                             CheckUnsafeCallWithABI check);
  void callWithABINoProfiler(Register fun, ABIType result) PER_ARCH;
  void callWithABINoProfiler(const Address& fun, ABIType result) PER_ARCH;

  // Restore the stack to its state before the setup function call.
  void callWithABIPost(uint32_t stackAdjust, ABIType result,
                       bool callFromWasm = false) PER_ARCH;

  // Create the signature to be able to decode the arguments of a native
  // function, when calling a function within the simulator.
  inline void appendSignatureType(ABIType type);
  inline ABIFunctionType signature() const;

  // Private variables used to handle moves between registers given as
  // arguments to passABIArg and the list of ABI registers expected for the
  // signature of the function.
  MoveResolver moveResolver_;

  // Architecture specific implementation which specify how registers & stack
  // offsets are used for calling a function.
  ABIArgGenerator abiArgs_;

#ifdef DEBUG
  // Flag use to assert that we use ABI function in the right context.
  bool inCall_;
#endif

  // If set by setupUnalignedABICall then callWithABI will pop the stack
  // register which is on the stack.
  bool dynamicAlignment_;

#ifdef JS_SIMULATOR
  // The signature is used to accumulate all types of arguments which are used
  // by the caller. This is used by the simulators to decode the arguments
  // properly, and cast the function pointer to the right type.
  uint32_t signature_;
#endif

 public:
  // ===============================================================
  // Jit Frames.
  //
  // These functions are used to build the content of the Jit frames.  See
  // CommonFrameLayout class, and all its derivatives. The content should be
  // pushed in the opposite order as the fields of the structures, such that
  // the structures can be used to interpret the content of the stack.

  // Call the Jit function, and push the return address (or let the callee
  // push the return address).
  //
  // These functions return the offset of the return address, in order to use
  // the return address to index the safepoints, which are used to list all
  // live registers.
  inline uint32_t callJitNoProfiler(Register callee);
  inline uint32_t callJit(Register callee);
  inline uint32_t callJit(JitCode* code);
  inline uint32_t callJit(TrampolinePtr code);
  inline uint32_t callJit(ImmPtr callee);

  // The frame descriptor is the second field of all Jit frames, pushed before
  // calling the Jit function. See CommonFrameLayout::descriptor_.
  inline void pushFrameDescriptor(FrameType type);
  inline void PushFrameDescriptor(FrameType type);

  // For JitFrameLayout, the descriptor also stores the number of arguments
  // passed by the caller. See MakeFrameDescriptorForJitCall.
  inline void pushFrameDescriptorForJitCall(FrameType type, uint32_t argc);
  inline void pushFrameDescriptorForJitCall(FrameType type, Register argc,
                                            Register scratch);
  inline void PushFrameDescriptorForJitCall(FrameType type, uint32_t argc);
  inline void PushFrameDescriptorForJitCall(FrameType type, Register argc,
                                            Register scratch);

  // Load the number of actual arguments from the frame's JitFrameLayout.
  inline void loadNumActualArgs(Register framePtr, Register dest);

  // Push the callee token of a JSFunction which pointer is stored in the
  // |callee| register. The callee token is packed with a |constructing| flag
  // which correspond to the fact that the JS function is called with "new" or
  // not.
  inline void PushCalleeToken(Register callee, bool constructing);

  // Unpack a callee token located at the |token| address, and return the
  // JSFunction pointer in the |dest| register.
  inline void loadFunctionFromCalleeToken(Address token, Register dest);

  // This function emulates a call by pushing an exit frame on the stack,
  // except that the fake-function is inlined within the body of the caller.
  //
  // This function assumes that the current frame is an IonJS frame.
  //
  // This function returns the offset of the /fake/ return address, in order to
  // use the return address to index the safepoints, which are used to list all
  // live registers.
  //
  // This function should be balanced with a call to adjustStack, to pop the
  // exit frame and emulate the return statement of the inlined function.
  inline uint32_t buildFakeExitFrame(Register scratch);

 private:
  // This function is used by buildFakeExitFrame to push a fake return address
  // on the stack. This fake return address should never be used for resuming
  // any execution, and can even be an invalid pointer into the instruction
  // stream, as long as it does not alias any other.
  uint32_t pushFakeReturnAddress(Register scratch) PER_SHARED_ARCH;

 public:
  // ===============================================================
  // Exit frame footer.
  //
  // When calling outside the Jit we push an exit frame. To mark the stack
  // correctly, we have to push additional information, called the Exit frame
  // footer, which is used to identify how the stack is marked.
  //
  // See JitFrames.h, and TraceJitExitFrame in JitFrames.cpp.

  // Links the exit frame and pushes the ExitFooterFrame.
  inline void enterExitFrame(Register cxreg, Register scratch, VMFunctionId f);

  // Push an exit frame token to identify which fake exit frame this footer
  // corresponds to.
  inline void enterFakeExitFrame(Register cxreg, Register scratch,
                                 ExitFrameType type);

  // Push an exit frame token for a native call.
  inline void enterFakeExitFrameForNative(Register cxreg, Register scratch,
                                          bool isConstructing);

  // Pop ExitFrame footer in addition to the extra frame.
  inline void leaveExitFrame(size_t extraFrame = 0);

 private:
  // Save the top of the stack into JitActivation::packedExitFP of the
  // current thread, which should be the location of the latest exit frame.
  void linkExitFrame(Register cxreg, Register scratch);

 public:
  // ===============================================================
  // Move instructions

  inline void move64(Imm64 imm, Register64 dest) PER_ARCH;
  inline void move64(Register64 src, Register64 dest) PER_ARCH;

  inline void moveFloat32ToGPR(FloatRegister src,
                               Register dest) PER_SHARED_ARCH;
  inline void moveGPRToFloat32(Register src,
                               FloatRegister dest) PER_SHARED_ARCH;

  inline void moveDoubleToGPR64(FloatRegister src, Register64 dest) PER_ARCH;
  inline void moveGPR64ToDouble(Register64 src, FloatRegister dest) PER_ARCH;

  inline void move8ZeroExtend(Register src, Register dest) PER_SHARED_ARCH;

  inline void move8SignExtend(Register src, Register dest) PER_SHARED_ARCH;
  inline void move16SignExtend(Register src, Register dest) PER_SHARED_ARCH;

  // move64To32 will clear the high bits of `dest` on 64-bit systems.
  inline void move64To32(Register64 src, Register dest) PER_ARCH;

  inline void move32To64ZeroExtend(Register src, Register64 dest) PER_ARCH;

  inline void move8To64SignExtend(Register src, Register64 dest) PER_ARCH;
  inline void move16To64SignExtend(Register src, Register64 dest) PER_ARCH;
  inline void move32To64SignExtend(Register src, Register64 dest) PER_ARCH;

  inline void move32SignExtendToPtr(Register src, Register dest) PER_ARCH;
  inline void move32ZeroExtendToPtr(Register src, Register dest) PER_ARCH;

  // Copy a constant, typed-register, or a ValueOperand into a ValueOperand
  // destination.
  inline void moveValue(const ConstantOrRegister& src,
                        const ValueOperand& dest);
  void moveValue(const TypedOrValueRegister& src,
                 const ValueOperand& dest) PER_ARCH;
  void moveValue(const ValueOperand& src, const ValueOperand& dest) PER_ARCH;
  void moveValue(const Value& src, const ValueOperand& dest) PER_ARCH;

  void movePropertyKey(PropertyKey key, Register dest);

  // ===============================================================
  // Load instructions

  inline void load32SignExtendToPtr(const Address& src, Register dest) PER_ARCH;

  inline void loadAbiReturnAddress(Register dest) PER_SHARED_ARCH;

  // ===============================================================
  // Copy instructions

  inline void copy64(const Address& src, const Address& dest, Register scratch);

 public:
  // ===============================================================
  // Logical instructions

  inline void not32(Register reg) PER_SHARED_ARCH;
  inline void notPtr(Register reg) PER_ARCH;

  inline void and32(Register src, Register dest) PER_SHARED_ARCH;
  inline void and32(Imm32 imm, Register dest) PER_SHARED_ARCH;
  inline void and32(Imm32 imm, Register src, Register dest) DEFINED_ON(arm64);
  inline void and32(Imm32 imm, const Address& dest) PER_SHARED_ARCH;
  inline void and32(const Address& src, Register dest) PER_SHARED_ARCH;

  inline void andPtr(Register src, Register dest) PER_ARCH;
  inline void andPtr(Imm32 imm, Register dest) PER_ARCH;

  inline void and64(Imm64 imm, Register64 dest) PER_ARCH;
  inline void or64(Imm64 imm, Register64 dest) PER_ARCH;
  inline void xor64(Imm64 imm, Register64 dest) PER_ARCH;

  inline void or32(Register src, Register dest) PER_SHARED_ARCH;
  inline void or32(Imm32 imm, Register dest) PER_SHARED_ARCH;
  inline void or32(Imm32 imm, const Address& dest) PER_SHARED_ARCH;

  inline void orPtr(Register src, Register dest) PER_ARCH;
  inline void orPtr(Imm32 imm, Register dest) PER_ARCH;

  inline void and64(Register64 src, Register64 dest) PER_ARCH;
  inline void or64(Register64 src, Register64 dest) PER_ARCH;
  inline void xor64(Register64 src, Register64 dest) PER_ARCH;

  inline void xor32(Register src, Register dest) PER_SHARED_ARCH;
  inline void xor32(Imm32 imm, Register dest) PER_SHARED_ARCH;
  inline void xor32(Imm32 imm, const Address& dest) PER_SHARED_ARCH;
  inline void xor32(const Address& src, Register dest) PER_SHARED_ARCH;

  inline void xorPtr(Register src, Register dest) PER_ARCH;
  inline void xorPtr(Imm32 imm, Register dest) PER_ARCH;

  inline void and64(const Operand& src, Register64 dest)
      DEFINED_ON(x64, mips64, loong64, riscv64);
  inline void or64(const Operand& src, Register64 dest)
      DEFINED_ON(x64, mips64, loong64, riscv64);
  inline void xor64(const Operand& src, Register64 dest)
      DEFINED_ON(x64, mips64, loong64, riscv64);

  // ===============================================================
  // Swap instructions

  // Swap the two lower bytes and sign extend the result to 32-bit.
  inline void byteSwap16SignExtend(Register reg) PER_SHARED_ARCH;

  // Swap the two lower bytes and zero extend the result to 32-bit.
  inline void byteSwap16ZeroExtend(Register reg) PER_SHARED_ARCH;

  // Swap all four bytes in a 32-bit integer.
  inline void byteSwap32(Register reg) PER_SHARED_ARCH;

  // Swap all eight bytes in a 64-bit integer.
  inline void byteSwap64(Register64 reg) PER_ARCH;

  // ===============================================================
  // Arithmetic functions

  // Condition flags aren't guaranteed to be set by these functions, for example
  // x86 will always set condition flags, but ARM64 won't do it unless
  // explicitly requested. Instead use branch(Add|Sub|Mul|Neg) to test for
  // condition flags after performing arithmetic operations.

  inline void add32(Register src, Register dest) PER_SHARED_ARCH;
  inline void add32(Imm32 imm, Register dest) PER_SHARED_ARCH;
  inline void add32(Imm32 imm, Register src, Register dest) PER_SHARED_ARCH;
  inline void add32(Imm32 imm, const Address& dest) PER_SHARED_ARCH;
  inline void add32(Imm32 imm, const AbsoluteAddress& dest)
      DEFINED_ON(x86_shared);

  inline void addPtr(Register src, Register dest) PER_ARCH;
  inline void addPtr(Register src1, Register src2, Register dest)
      DEFINED_ON(arm64);
  inline void addPtr(Imm32 imm, Register dest) PER_ARCH;
  inline void addPtr(Imm32 imm, Register src, Register dest) DEFINED_ON(arm64);
  inline void addPtr(ImmWord imm, Register dest) PER_ARCH;
  inline void addPtr(ImmPtr imm, Register dest);
  inline void addPtr(Imm32 imm, const Address& dest)
      DEFINED_ON(mips_shared, arm, arm64, x86, x64, loong64, riscv64, wasm32);
  inline void addPtr(Imm32 imm, const AbsoluteAddress& dest)
      DEFINED_ON(x86, x64);
  inline void addPtr(const Address& src, Register dest)
      DEFINED_ON(mips_shared, arm, arm64, x86, x64, loong64, riscv64, wasm32);

  inline void add64(Register64 src, Register64 dest) PER_ARCH;
  inline void add64(Imm32 imm, Register64 dest) PER_ARCH;
  inline void add64(Imm64 imm, Register64 dest) PER_ARCH;
  inline void add64(const Operand& src, Register64 dest)
      DEFINED_ON(x64, mips64, loong64, riscv64);

  inline void addFloat32(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  // Compute dest=SP-imm where dest is a pointer registers and not SP.  The
  // offset returned from sub32FromStackPtrWithPatch() must be passed to
  // patchSub32FromStackPtr().
  inline CodeOffset sub32FromStackPtrWithPatch(Register dest) PER_ARCH;
  inline void patchSub32FromStackPtr(CodeOffset offset, Imm32 imm) PER_ARCH;

  inline void addDouble(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;
  inline void addConstantDouble(double d, FloatRegister dest) DEFINED_ON(x86);

  inline void sub32(const Address& src, Register dest) PER_SHARED_ARCH;
  inline void sub32(Register src, Register dest) PER_SHARED_ARCH;
  inline void sub32(Imm32 imm, Register dest) PER_SHARED_ARCH;

  inline void subPtr(Register src, Register dest) PER_ARCH;
  inline void subPtr(Register src, const Address& dest)
      DEFINED_ON(mips_shared, arm, arm64, x86, x64, loong64, riscv64, wasm32);
  inline void subPtr(Imm32 imm, Register dest) PER_ARCH;
  inline void subPtr(ImmWord imm, Register dest) DEFINED_ON(x64);
  inline void subPtr(const Address& addr, Register dest)
      DEFINED_ON(mips_shared, arm, arm64, x86, x64, loong64, riscv64, wasm32);

  inline void sub64(Register64 src, Register64 dest) PER_ARCH;
  inline void sub64(Imm64 imm, Register64 dest) PER_ARCH;
  inline void sub64(const Operand& src, Register64 dest)
      DEFINED_ON(x64, mips64, loong64, riscv64);

  inline void subFloat32(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  inline void subDouble(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  inline void mul32(Register rhs, Register srcDest) PER_SHARED_ARCH;
  inline void mul32(Imm32 imm, Register srcDest) PER_SHARED_ARCH;

  inline void mul32(Register src1, Register src2, Register dest, Label* onOver)
      DEFINED_ON(arm64);

  // Return the high word of the unsigned multiplication into |dest|.
  inline void mulHighUnsigned32(Imm32 imm, Register src,
                                Register dest) PER_ARCH;

  inline void mulPtr(Register rhs, Register srcDest) PER_ARCH;

  inline void mul64(const Operand& src, const Register64& dest) DEFINED_ON(x64);
  inline void mul64(const Operand& src, const Register64& dest,
                    const Register temp)
      DEFINED_ON(x64, mips64, loong64, riscv64);
  inline void mul64(Imm64 imm, const Register64& dest) PER_ARCH;
  inline void mul64(Imm64 imm, const Register64& dest, const Register temp)
      DEFINED_ON(x86, x64, arm, mips32, mips64, loong64, riscv64);
  inline void mul64(const Register64& src, const Register64& dest,
                    const Register temp) PER_ARCH;
  inline void mul64(const Register64& src1, const Register64& src2,
                    const Register64& dest) DEFINED_ON(arm64);
  inline void mul64(Imm64 src1, const Register64& src2, const Register64& dest)
      DEFINED_ON(arm64);

  inline void mulBy3(Register src, Register dest) PER_ARCH;

  inline void mulFloat32(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;
  inline void mulDouble(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  inline void mulDoublePtr(ImmPtr imm, Register temp, FloatRegister dest)
      DEFINED_ON(mips_shared, arm, arm64, x86, x64, loong64, riscv64, wasm32);

  // Perform an integer division, returning the integer part rounded toward
  // zero. rhs must not be zero, and the division must not overflow.
  //
  // On ARM, the chip must have hardware division instructions.
  inline void quotient32(Register rhs, Register srcDest, bool isUnsigned)
      DEFINED_ON(mips_shared, arm, arm64, loong64, riscv64, wasm32);

  // As above, but srcDest must be eax and tempEdx must be edx.
  inline void quotient32(Register rhs, Register srcDest, Register tempEdx,
                         bool isUnsigned) DEFINED_ON(x86_shared);

  // Perform an integer division, returning the remainder part.
  // rhs must not be zero, and the division must not overflow.
  //
  // On ARM, the chip must have hardware division instructions.
  inline void remainder32(Register rhs, Register srcDest, bool isUnsigned)
      DEFINED_ON(mips_shared, arm, arm64, loong64, riscv64, wasm32);

  // As above, but srcDest must be eax and tempEdx must be edx.
  inline void remainder32(Register rhs, Register srcDest, Register tempEdx,
                          bool isUnsigned) DEFINED_ON(x86_shared);

  // Perform an integer division, returning the integer part rounded toward
  // zero. rhs must not be zero, and the division must not overflow.
  //
  // This variant preserves registers, and doesn't require hardware division
  // instructions on ARM (will call out to a runtime routine).
  //
  // rhs is preserved, srdDest is clobbered.
  void flexibleRemainder32(Register rhs, Register srcDest, bool isUnsigned,
                           const LiveRegisterSet& volatileLiveRegs)
      DEFINED_ON(mips_shared, arm, arm64, x86_shared, loong64, riscv64, wasm32);

  // Perform an integer division, returning the integer part rounded toward
  // zero. rhs must not be zero, and the division must not overflow.
  //
  // This variant preserves registers, and doesn't require hardware division
  // instructions on ARM (will call out to a runtime routine).
  //
  // rhs is preserved, srdDest is clobbered.
  void flexibleQuotient32(Register rhs, Register srcDest, bool isUnsigned,
                          const LiveRegisterSet& volatileLiveRegs)
      DEFINED_ON(mips_shared, arm, arm64, x86_shared, loong64, riscv64);

  // Perform an integer division, returning the integer part rounded toward
  // zero. rhs must not be zero, and the division must not overflow. The
  // remainder is stored into the third argument register here.
  //
  // This variant preserves registers, and doesn't require hardware division
  // instructions on ARM (will call out to a runtime routine).
  //
  // rhs is preserved, srdDest and remOutput are clobbered.
  void flexibleDivMod32(Register rhs, Register srcDest, Register remOutput,
                        bool isUnsigned,
                        const LiveRegisterSet& volatileLiveRegs)
      DEFINED_ON(mips_shared, arm, arm64, x86_shared, loong64, riscv64, wasm32);

  inline void divFloat32(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;
  inline void divDouble(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  inline void inc64(AbsoluteAddress dest) PER_ARCH;

  inline void neg32(Register reg) PER_SHARED_ARCH;
  inline void neg64(Register64 reg) PER_ARCH;
  inline void negPtr(Register reg) PER_ARCH;

  inline void negateFloat(FloatRegister reg) PER_SHARED_ARCH;

  inline void negateDouble(FloatRegister reg) PER_SHARED_ARCH;

  inline void abs32(Register src, Register dest) PER_SHARED_ARCH;
  inline void absFloat32(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;
  inline void absDouble(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  inline void sqrtFloat32(FloatRegister src,
                          FloatRegister dest) PER_SHARED_ARCH;
  inline void sqrtDouble(FloatRegister src, FloatRegister dest) PER_SHARED_ARCH;

  void floorFloat32ToInt32(FloatRegister src, Register dest,
                           Label* fail) PER_SHARED_ARCH;
  void floorDoubleToInt32(FloatRegister src, Register dest,
                          Label* fail) PER_SHARED_ARCH;

  void ceilFloat32ToInt32(FloatRegister src, Register dest,
                          Label* fail) PER_SHARED_ARCH;
  void ceilDoubleToInt32(FloatRegister src, Register dest,
                         Label* fail) PER_SHARED_ARCH;

  void roundFloat32ToInt32(FloatRegister src, Register dest, FloatRegister temp,
                           Label* fail) PER_SHARED_ARCH;
  void roundDoubleToInt32(FloatRegister src, Register dest, FloatRegister temp,
                          Label* fail) PER_SHARED_ARCH;

  void truncFloat32ToInt32(FloatRegister src, Register dest,
                           Label* fail) PER_SHARED_ARCH;
  void truncDoubleToInt32(FloatRegister src, Register dest,
                          Label* fail) PER_SHARED_ARCH;

  void nearbyIntDouble(RoundingMode mode, FloatRegister src,
                       FloatRegister dest) PER_SHARED_ARCH;
  void nearbyIntFloat32(RoundingMode mode, FloatRegister src,
                        FloatRegister dest) PER_SHARED_ARCH;

  void signInt32(Register input, Register output);
  void signDouble(FloatRegister input, FloatRegister output);
  void signDoubleToInt32(FloatRegister input, Register output,
                         FloatRegister temp, Label* fail);

  void copySignDouble(FloatRegister lhs, FloatRegister rhs,
                      FloatRegister output) PER_SHARED_ARCH;
  void copySignFloat32(FloatRegister lhs, FloatRegister rhs,
                       FloatRegister output) DEFINED_ON(x86_shared, arm64);

  // Returns a random double in range [0, 1) in |dest|. The |rng| register must
  // hold a pointer to a mozilla::non_crypto::XorShift128PlusRNG.
  void randomDouble(Register rng, FloatRegister dest, Register64 temp0,
                    Register64 temp1);

  // srcDest = {min,max}{Float32,Double}(srcDest, other)
  // For min and max, handle NaN specially if handleNaN is true.

  inline void minFloat32(FloatRegister other, FloatRegister srcDest,
                         bool handleNaN) PER_SHARED_ARCH;
  inline void minDouble(FloatRegister other, FloatRegister srcDest,
                        bool handleNaN) PER_SHARED_ARCH;

  inline void maxFloat32(FloatRegister other, FloatRegister srcDest,
                         bool handleNaN) PER_SHARED_ARCH;
  inline void maxDouble(FloatRegister other, FloatRegister srcDest,
                        bool handleNaN) PER_SHARED_ARCH;

  void minMaxArrayInt32(Register array, Register result, Register temp1,
                        Register temp2, Register temp3, bool isMax,
                        Label* fail);
  void minMaxArrayNumber(Register array, FloatRegister result,
                         FloatRegister floatTemp, Register temp1,
                         Register temp2, bool isMax, Label* fail);

  // Compute |pow(base, power)| and store the result in |dest|. If the result
  // exceeds the int32 range, jumps to |onOver|.
  // |base| and |power| are preserved, the other input registers are clobbered.
  void pow32(Register base, Register power, Register dest, Register temp1,
             Register temp2, Label* onOver);

  void sameValueDouble(FloatRegister left, FloatRegister right,
                       FloatRegister temp, Register dest);

  void branchIfNotRegExpPrototypeOptimizable(Register proto, Register temp,
                                             const GlobalObject* maybeGlobal,
                                             Label* label);
  void branchIfNotRegExpInstanceOptimizable(Register regexp, Register temp,
                                            const GlobalObject* maybeGlobal,
                                            Label* label);

  void loadRegExpLastIndex(Register regexp, Register string, Register lastIndex,
                           Label* notFoundZeroLastIndex);

  void loadAndClearRegExpSearcherLastLimit(Register result, Register scratch);

  void loadParsedRegExpShared(Register regexp, Register result,
                              Label* unparsed);

  // ===============================================================
  // Shift functions

  // For shift-by-register there may be platform-specific variations, for
  // example, x86 will perform the shift mod 32 but ARM will perform the shift
  // mod 256.
  //
  // For shift-by-immediate the platform assembler may restrict the immediate,
  // for example, the ARM assembler requires the count for 32-bit shifts to be
  // in the range [0,31].

  inline void lshift32(Imm32 shift, Register srcDest) PER_SHARED_ARCH;
  inline void rshift32(Imm32 shift, Register srcDest) PER_SHARED_ARCH;
  inline void rshift32Arithmetic(Imm32 shift, Register srcDest) PER_SHARED_ARCH;

  inline void lshiftPtr(Imm32 imm, Register dest) PER_ARCH;
  inline void rshiftPtr(Imm32 imm, Register dest) PER_ARCH;
  inline void rshiftPtr(Imm32 imm, Register src, Register dest)
      DEFINED_ON(arm64);
  inline void rshiftPtrArithmetic(Imm32 imm, Register dest) PER_ARCH;

  inline void lshift64(Imm32 imm, Register64 dest) PER_ARCH;
  inline void rshift64(Imm32 imm, Register64 dest) PER_ARCH;
  inline void rshift64Arithmetic(Imm32 imm, Register64 dest) PER_ARCH;

  // On x86_shared these have the constraint that shift must be in CL.
  inline void lshift32(Register shift, Register srcDest) PER_SHARED_ARCH;
  inline void rshift32(Register shift, Register srcDest) PER_SHARED_ARCH;
  inline void rshift32Arithmetic(Register shift,
                                 Register srcDest) PER_SHARED_ARCH;
  inline void lshiftPtr(Register shift, Register srcDest) PER_ARCH;
  inline void rshiftPtr(Register shift, Register srcDest) PER_ARCH;

  // These variants do not have the above constraint, but may emit some extra
  // instructions on x86_shared. They also handle shift >= 32 consistently by
  // masking with 0x1F (either explicitly or relying on the hardware to do
  // that).
  inline void flexibleLshift32(Register shift,
                               Register srcDest) PER_SHARED_ARCH;
  inline void flexibleRshift32(Register shift,
                               Register srcDest) PER_SHARED_ARCH;
  inline void flexibleRshift32Arithmetic(Register shift,
                                         Register srcDest) PER_SHARED_ARCH;

  inline void lshift64(Register shift, Register64 srcDest) PER_ARCH;
  inline void rshift64(Register shift, Register64 srcDest) PER_ARCH;
  inline void rshift64Arithmetic(Register shift, Register64 srcDest) PER_ARCH;

  // ===============================================================
  // Rotation functions
  // Note: - on x86 and x64 the count register must be in CL.
  //       - on x64 the temp register should be InvalidReg.

  inline void rotateLeft(Imm32 count, Register input,
                         Register dest) PER_SHARED_ARCH;
  inline void rotateLeft(Register count, Register input,
                         Register dest) PER_SHARED_ARCH;
  inline void rotateLeft64(Imm32 count, Register64 input, Register64 dest)
      DEFINED_ON(x64);
  inline void rotateLeft64(Register count, Register64 input, Register64 dest)
      DEFINED_ON(x64);
  inline void rotateLeft64(Imm32 count, Register64 input, Register64 dest,
                           Register temp) PER_ARCH;
  inline void rotateLeft64(Register count, Register64 input, Register64 dest,
                           Register temp) PER_ARCH;

  inline void rotateRight(Imm32 count, Register input,
                          Register dest) PER_SHARED_ARCH;
  inline void rotateRight(Register count, Register input,
                          Register dest) PER_SHARED_ARCH;
  inline void rotateRight64(Imm32 count, Register64 input, Register64 dest)
      DEFINED_ON(x64);
  inline void rotateRight64(Register count, Register64 input, Register64 dest)
      DEFINED_ON(x64);
  inline void rotateRight64(Imm32 count, Register64 input, Register64 dest,
                            Register temp) PER_ARCH;
  inline void rotateRight64(Register count, Register64 input, Register64 dest,
                            Register temp) PER_ARCH;

  // ===============================================================
  // Bit counting functions

  // knownNotZero may be true only if the src is known not to be zero.
  inline void clz32(Register src, Register dest,
                    bool knownNotZero) PER_SHARED_ARCH;
  inline void ctz32(Register src, Register dest,
                    bool knownNotZero) PER_SHARED_ARCH;

  inline void clz64(Register64 src, Register dest) PER_ARCH;
  inline void ctz64(Register64 src, Register dest) PER_ARCH;

  // On x86_shared, temp may be Invalid only if the chip has the POPCNT
  // instruction. On ARM, temp may never be Invalid.
  inline void popcnt32(Register src, Register dest,
                       Register temp) PER_SHARED_ARCH;

  // temp may be invalid only if the chip has the POPCNT instruction.
  inline void popcnt64(Register64 src, Register64 dest, Register temp) PER_ARCH;

  // ===============================================================
  // Condition functions

  inline void cmp8Set(Condition cond, Address lhs, Imm32 rhs,
                      Register dest) PER_SHARED_ARCH;

  inline void cmp16Set(Condition cond, Address lhs, Imm32 rhs,
                       Register dest) PER_SHARED_ARCH;

  template <typename T1, typename T2>
  inline void cmp32Set(Condition cond, T1 lhs, T2 rhs, Register dest)
      DEFINED_ON(x86_shared, arm, arm64, mips32, mips64, loong64, riscv64,
                 wasm32);

  // Only the NotEqual and Equal conditions are allowed.
  inline void cmp64Set(Condition cond, Address lhs, Imm64 rhs,
                       Register dest) PER_ARCH;

  template <typename T1, typename T2>
  inline void cmpPtrSet(Condition cond, T1 lhs, T2 rhs, Register dest) PER_ARCH;

  // ===============================================================
  // Branch functions

  inline void branch8(Condition cond, const Address& lhs, Imm32 rhs,
                      Label* label) PER_SHARED_ARCH;

  // Compares the byte in |lhs| against |rhs| using a 8-bit comparison on
  // x86/x64 or a 32-bit comparison (all other platforms). The caller should
  // ensure |rhs| is a zero- resp. sign-extended byte value for cross-platform
  // compatible code.
  inline void branch8(Condition cond, const BaseIndex& lhs, Register rhs,
                      Label* label) PER_SHARED_ARCH;

  inline void branch16(Condition cond, const Address& lhs, Imm32 rhs,
                       Label* label) PER_SHARED_ARCH;

  template <class L>
  inline void branch32(Condition cond, Register lhs, Register rhs,
                       L label) PER_SHARED_ARCH;
  template <class L>
  inline void branch32(Condition cond, Register lhs, Imm32 rhs,
                       L label) PER_SHARED_ARCH;

  inline void branch32(Condition cond, Register lhs, const Address& rhs,
                       Label* label) DEFINED_ON(arm64);

  inline void branch32(Condition cond, const Address& lhs, Register rhs,
                       Label* label) PER_SHARED_ARCH;
  inline void branch32(Condition cond, const Address& lhs, Imm32 rhs,
                       Label* label) PER_SHARED_ARCH;

  inline void branch32(Condition cond, const AbsoluteAddress& lhs, Register rhs,
                       Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  inline void branch32(Condition cond, const AbsoluteAddress& lhs, Imm32 rhs,
                       Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  inline void branch32(Condition cond, const BaseIndex& lhs, Register rhs,
                       Label* label) DEFINED_ON(arm, x86_shared);
  inline void branch32(Condition cond, const BaseIndex& lhs, Imm32 rhs,
                       Label* label) PER_SHARED_ARCH;

  inline void branch32(Condition cond, const Operand& lhs, Register rhs,
                       Label* label) DEFINED_ON(x86_shared);
  inline void branch32(Condition cond, const Operand& lhs, Imm32 rhs,
                       Label* label) DEFINED_ON(x86_shared);

  inline void branch32(Condition cond, wasm::SymbolicAddress lhs, Imm32 rhs,
                       Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  // The supported condition are Equal, NotEqual, LessThan(orEqual),
  // GreaterThan(orEqual), Below(orEqual) and Above(orEqual). When a fail label
  // is not defined it will fall through to next instruction, else jump to the
  // fail label.
  inline void branch64(Condition cond, Register64 lhs, Imm64 val,
                       Label* success, Label* fail = nullptr) PER_ARCH;
  inline void branch64(Condition cond, Register64 lhs, Register64 rhs,
                       Label* success, Label* fail = nullptr) PER_ARCH;
  // Only the NotEqual and Equal conditions are allowed for the branch64
  // variants with Address as lhs.
  inline void branch64(Condition cond, const Address& lhs, Imm64 val,
                       Label* label) PER_ARCH;
  inline void branch64(Condition cond, const Address& lhs, Register64 rhs,
                       Label* label) PER_ARCH;

  // Compare the value at |lhs| with the value at |rhs|.  The scratch
  // register *must not* be the base of |lhs| or |rhs|.
  inline void branch64(Condition cond, const Address& lhs, const Address& rhs,
                       Register scratch, Label* label) PER_ARCH;

  template <class L>
  inline void branchPtr(Condition cond, Register lhs, Register rhs,
                        L label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, Register lhs, Imm32 rhs,
                        Label* label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, Register lhs, ImmPtr rhs,
                        Label* label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, Register lhs, ImmGCPtr rhs,
                        Label* label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, Register lhs, ImmWord rhs,
                        Label* label) PER_SHARED_ARCH;

  template <class L>
  inline void branchPtr(Condition cond, const Address& lhs, Register rhs,
                        L label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, const Address& lhs, ImmPtr rhs,
                        Label* label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, const Address& lhs, ImmGCPtr rhs,
                        Label* label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, const Address& lhs, ImmWord rhs,
                        Label* label) PER_SHARED_ARCH;

  inline void branchPtr(Condition cond, const BaseIndex& lhs, ImmWord rhs,
                        Label* label) PER_SHARED_ARCH;
  inline void branchPtr(Condition cond, const BaseIndex& lhs, Register rhs,
                        Label* label) PER_SHARED_ARCH;

  inline void branchPtr(Condition cond, const AbsoluteAddress& lhs,
                        Register rhs, Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  inline void branchPtr(Condition cond, const AbsoluteAddress& lhs, ImmWord rhs,
                        Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  inline void branchPtr(Condition cond, wasm::SymbolicAddress lhs, Register rhs,
                        Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  // Given a pointer to a GC Cell, retrieve the StoreBuffer pointer from its
  // chunk header, or nullptr if it is in the tenured heap.
  void loadStoreBuffer(Register ptr, Register buffer) PER_ARCH;

  void branchPtrInNurseryChunk(Condition cond, Register ptr, Register temp,
                               Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  void branchPtrInNurseryChunk(Condition cond, const Address& address,
                               Register temp, Label* label) DEFINED_ON(x86);
  void branchValueIsNurseryCell(Condition cond, const Address& address,
                                Register temp, Label* label) PER_ARCH;
  void branchValueIsNurseryCell(Condition cond, ValueOperand value,
                                Register temp, Label* label) PER_ARCH;

  // This function compares a Value (lhs) which is having a private pointer
  // boxed inside a js::Value, with a raw pointer (rhs).
  inline void branchPrivatePtr(Condition cond, const Address& lhs, Register rhs,
                               Label* label) PER_ARCH;

  inline void branchFloat(DoubleCondition cond, FloatRegister lhs,
                          FloatRegister rhs, Label* label) PER_SHARED_ARCH;

  // Truncate a double/float32 to int32 and when it doesn't fit an int32 it will
  // jump to the failure label. This particular variant is allowed to return the
  // value module 2**32, which isn't implemented on all architectures. E.g. the
  // x64 variants will do this only in the int64_t range.
  inline void branchTruncateFloat32MaybeModUint32(FloatRegister src,
                                                  Register dest, Label* fail)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  inline void branchTruncateDoubleMaybeModUint32(FloatRegister src,
                                                 Register dest, Label* fail)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  // Truncate a double/float32 to intptr and when it doesn't fit jump to the
  // failure label.
  inline void branchTruncateFloat32ToPtr(FloatRegister src, Register dest,
                                         Label* fail) DEFINED_ON(x86, x64);
  inline void branchTruncateDoubleToPtr(FloatRegister src, Register dest,
                                        Label* fail) DEFINED_ON(x86, x64);

  // Truncate a double/float32 to int32 and when it doesn't fit jump to the
  // failure label.
  inline void branchTruncateFloat32ToInt32(FloatRegister src, Register dest,
                                           Label* fail)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  inline void branchTruncateDoubleToInt32(FloatRegister src, Register dest,
                                          Label* fail) PER_ARCH;

  inline void branchDouble(DoubleCondition cond, FloatRegister lhs,
                           FloatRegister rhs, Label* label) PER_SHARED_ARCH;

  inline void branchDoubleNotInInt64Range(Address src, Register temp,
                                          Label* fail);
  inline void branchDoubleNotInUInt64Range(Address src, Register temp,
                                           Label* fail);
  inline void branchFloat32NotInInt64Range(Address src, Register temp,
                                           Label* fail);
  inline void branchFloat32NotInUInt64Range(Address src, Register temp,
                                            Label* fail);

  template <typename T>
  inline void branchAdd32(Condition cond, T src, Register dest,
                          Label* label) PER_SHARED_ARCH;
  template <typename T>
  inline void branchSub32(Condition cond, T src, Register dest,
                          Label* label) PER_SHARED_ARCH;
  template <typename T>
  inline void branchMul32(Condition cond, T src, Register dest,
                          Label* label) PER_SHARED_ARCH;
  template <typename T>
  inline void branchRshift32(Condition cond, T src, Register dest,
                             Label* label) PER_SHARED_ARCH;

  inline void branchNeg32(Condition cond, Register reg,
                          Label* label) PER_SHARED_ARCH;

  inline void branchAdd64(Condition cond, Imm64 imm, Register64 dest,
                          Label* label) DEFINED_ON(x86, arm, wasm32);

  template <typename T>
  inline void branchAddPtr(Condition cond, T src, Register dest,
                           Label* label) PER_SHARED_ARCH;

  template <typename T>
  inline void branchSubPtr(Condition cond, T src, Register dest,
                           Label* label) PER_SHARED_ARCH;

  inline void branchMulPtr(Condition cond, Register src, Register dest,
                           Label* label) PER_SHARED_ARCH;

  inline void decBranchPtr(Condition cond, Register lhs, Imm32 rhs,
                           Label* label) PER_SHARED_ARCH;

  template <class L>
  inline void branchTest32(Condition cond, Register lhs, Register rhs,
                           L label) PER_SHARED_ARCH;
  template <class L>
  inline void branchTest32(Condition cond, Register lhs, Imm32 rhs,
                           L label) PER_SHARED_ARCH;
  inline void branchTest32(Condition cond, const Address& lhs, Imm32 rhh,
                           Label* label) PER_SHARED_ARCH;
  inline void branchTest32(Condition cond, const AbsoluteAddress& lhs,
                           Imm32 rhs, Label* label)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  template <class L>
  inline void branchTestPtr(Condition cond, Register lhs, Register rhs,
                            L label) PER_SHARED_ARCH;
  inline void branchTestPtr(Condition cond, Register lhs, Imm32 rhs,
                            Label* label) PER_SHARED_ARCH;
  inline void branchTestPtr(Condition cond, const Address& lhs, Imm32 rhs,
                            Label* label) PER_SHARED_ARCH;

  template <class L>
  inline void branchTest64(Condition cond, Register64 lhs, Register64 rhs,
                           Register temp, L label) PER_ARCH;

  // Branches to |label| if |reg| is false. |reg| should be a C++ bool.
  template <class L>
  inline void branchIfFalseBool(Register reg, L label);

  // Branches to |label| if |reg| is true. |reg| should be a C++ bool.
  inline void branchIfTrueBool(Register reg, Label* label);

  inline void branchIfRope(Register str, Label* label);
  inline void branchIfNotRope(Register str, Label* label);

  inline void branchLatin1String(Register string, Label* label);
  inline void branchTwoByteString(Register string, Label* label);

  inline void branchIfBigIntIsNegative(Register bigInt, Label* label);
  inline void branchIfBigIntIsNonNegative(Register bigInt, Label* label);
  inline void branchIfBigIntIsZero(Register bigInt, Label* label);
  inline void branchIfBigIntIsNonZero(Register bigInt, Label* label);

  inline void branchTestFunctionFlags(Register fun, uint32_t flags,
                                      Condition cond, Label* label);

  inline void branchIfNotFunctionIsNonBuiltinCtor(Register fun,
                                                  Register scratch,
                                                  Label* label);

  inline void branchIfFunctionHasNoJitEntry(Register fun, bool isConstructing,
                                            Label* label);
  inline void branchIfFunctionHasJitEntry(Register fun, bool isConstructing,
                                          Label* label);

  inline void branchIfScriptHasJitScript(Register script, Label* label);
  inline void branchIfScriptHasNoJitScript(Register script, Label* label);
  inline void loadJitScript(Register script, Register dest);

  // Loads the function's argument count.
  inline void loadFunctionArgCount(Register func, Register output);

  // Loads the function length. This handles interpreted, native, and bound
  // functions. The caller is responsible for checking that INTERPRETED_LAZY and
  // RESOLVED_LENGTH flags are not set.
  void loadFunctionLength(Register func, Register funFlagsAndArgCount,
                          Register output, Label* slowPath);

  // Loads the function name. This handles interpreted, native, and bound
  // functions.
  void loadFunctionName(Register func, Register output, ImmGCPtr emptyString,
                        Label* slowPath);

  void assertFunctionIsExtended(Register func);

  inline void branchFunctionKind(Condition cond,
                                 FunctionFlags::FunctionKind kind, Register fun,
                                 Register scratch, Label* label);

  inline void branchIfObjectEmulatesUndefined(Register objReg, Register scratch,
                                              Label* slowCheck, Label* label);

  // For all methods below: spectreRegToZero is a register that will be zeroed
  // on speculatively executed code paths (when the branch should be taken but
  // branch prediction speculates it isn't). Usually this will be the object
  // register but the caller may pass a different register.

  inline void branchTestObjClass(Condition cond, Register obj,
                                 const JSClass* clasp, Register scratch,
                                 Register spectreRegToZero, Label* label);
  inline void branchTestObjClassNoSpectreMitigations(Condition cond,
                                                     Register obj,
                                                     const JSClass* clasp,
                                                     Register scratch,
                                                     Label* label);

  inline void branchTestObjClass(Condition cond, Register obj,
                                 const Address& clasp, Register scratch,
                                 Register spectreRegToZero, Label* label);
  inline void branchTestObjClassNoSpectreMitigations(Condition cond,
                                                     Register obj,
                                                     const Address& clasp,
                                                     Register scratch,
                                                     Label* label);

  inline void branchTestObjClass(Condition cond, Register obj, Register clasp,
                                 Register scratch, Register spectreRegToZero,
                                 Label* label);

 private:
  inline void branchTestClass(Condition cond, Register clasp,
                              std::pair<const JSClass*, const JSClass*> classes,
                              Label* label);

 public:
  inline void branchTestObjClass(
      Condition cond, Register obj,
      std::pair<const JSClass*, const JSClass*> classes, Register scratch,
      Register spectreRegToZero, Label* label);
  inline void branchTestObjClassNoSpectreMitigations(
      Condition cond, Register obj,
      std::pair<const JSClass*, const JSClass*> classes, Register scratch,
      Label* label);

  inline void branchTestObjShape(Condition cond, Register obj,
                                 const Shape* shape, Register scratch,
                                 Register spectreRegToZero, Label* label);
  inline void branchTestObjShapeNoSpectreMitigations(Condition cond,
                                                     Register obj,
                                                     const Shape* shape,
                                                     Label* label);

  void branchTestObjShapeList(Condition cond, Register obj,
                              Register shapeElements, Register shapeScratch,
                              Register endScratch, Register spectreScratch,
                              Label* label);

  inline void branchTestClassIsFunction(Condition cond, Register clasp,
                                        Label* label);
  inline void branchTestObjIsFunction(Condition cond, Register obj,
                                      Register scratch,
                                      Register spectreRegToZero, Label* label);
  inline void branchTestObjIsFunctionNoSpectreMitigations(Condition cond,
                                                          Register obj,
                                                          Register scratch,
                                                          Label* label);

  inline void branchTestObjShape(Condition cond, Register obj, Register shape,
                                 Register scratch, Register spectreRegToZero,
                                 Label* label);
  inline void branchTestObjShapeNoSpectreMitigations(Condition cond,
                                                     Register obj,
                                                     Register shape,
                                                     Label* label);

  // TODO: audit/fix callers to be Spectre safe.
  inline void branchTestObjShapeUnsafe(Condition cond, Register obj,
                                       Register shape, Label* label);

  void branchTestObjCompartment(Condition cond, Register obj,
                                const Address& compartment, Register scratch,
                                Label* label);
  void branchTestObjCompartment(Condition cond, Register obj,
                                const JS::Compartment* compartment,
                                Register scratch, Label* label);

  void branchIfNonNativeObj(Register obj, Register scratch, Label* label);

  void branchIfObjectNotExtensible(Register obj, Register scratch,
                                   Label* label);

  void branchTestObjectNeedsProxyResultValidation(Condition condition,
                                                  Register obj,
                                                  Register scratch,
                                                  Label* label);

  inline void branchTestClassIsProxy(bool proxy, Register clasp, Label* label);

  inline void branchTestObjectIsProxy(bool proxy, Register object,
                                      Register scratch, Label* label);

  inline void branchTestProxyHandlerFamily(Condition cond, Register proxy,
                                           Register scratch,
                                           const void* handlerp, Label* label);

  inline void branchTestNeedsIncrementalBarrier(Condition cond, Label* label);
  inline void branchTestNeedsIncrementalBarrierAnyZone(Condition cond,
                                                       Label* label,
                                                       Register scratch);

  // Perform a type-test on a tag of a Value (32bits boxing), or the tagged
  // value (64bits boxing).
  inline void branchTestUndefined(Condition cond, Register tag,
                                  Label* label) PER_SHARED_ARCH;
  inline void branchTestInt32(Condition cond, Register tag,
                              Label* label) PER_SHARED_ARCH;
  inline void branchTestDouble(Condition cond, Register tag, Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);
  inline void branchTestNumber(Condition cond, Register tag,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestBoolean(Condition cond, Register tag,
                                Label* label) PER_SHARED_ARCH;
  inline void branchTestString(Condition cond, Register tag,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestSymbol(Condition cond, Register tag,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestBigInt(Condition cond, Register tag,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestNull(Condition cond, Register tag,
                             Label* label) PER_SHARED_ARCH;
  inline void branchTestObject(Condition cond, Register tag,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestPrimitive(Condition cond, Register tag,
                                  Label* label) PER_SHARED_ARCH;
  inline void branchTestMagic(Condition cond, Register tag,
                              Label* label) PER_SHARED_ARCH;
  void branchTestType(Condition cond, Register tag, JSValueType type,
                      Label* label);

  // Perform a type-test on a Value, addressed by Address or BaseIndex, or
  // loaded into ValueOperand.
  // BaseIndex and ValueOperand variants clobber the ScratchReg on x64.
  // All Variants clobber the ScratchReg on arm64.
  inline void branchTestUndefined(Condition cond, const Address& address,
                                  Label* label) PER_SHARED_ARCH;
  inline void branchTestUndefined(Condition cond, const BaseIndex& address,
                                  Label* label) PER_SHARED_ARCH;
  inline void branchTestUndefined(Condition cond, const ValueOperand& value,
                                  Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestInt32(Condition cond, const Address& address,
                              Label* label) PER_SHARED_ARCH;
  inline void branchTestInt32(Condition cond, const BaseIndex& address,
                              Label* label) PER_SHARED_ARCH;
  inline void branchTestInt32(Condition cond, const ValueOperand& value,
                              Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestDouble(Condition cond, const Address& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestDouble(Condition cond, const BaseIndex& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestDouble(Condition cond, const ValueOperand& value,
                               Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestNumber(Condition cond, const ValueOperand& value,
                               Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestBoolean(Condition cond, const Address& address,
                                Label* label) PER_SHARED_ARCH;
  inline void branchTestBoolean(Condition cond, const BaseIndex& address,
                                Label* label) PER_SHARED_ARCH;
  inline void branchTestBoolean(Condition cond, const ValueOperand& value,
                                Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestString(Condition cond, const Address& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestString(Condition cond, const BaseIndex& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestString(Condition cond, const ValueOperand& value,
                               Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestSymbol(Condition cond, const Address& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestSymbol(Condition cond, const BaseIndex& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestSymbol(Condition cond, const ValueOperand& value,
                               Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestBigInt(Condition cond, const Address& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestBigInt(Condition cond, const BaseIndex& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestBigInt(Condition cond, const ValueOperand& value,
                               Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestNull(Condition cond, const Address& address,
                             Label* label) PER_SHARED_ARCH;
  inline void branchTestNull(Condition cond, const BaseIndex& address,
                             Label* label) PER_SHARED_ARCH;
  inline void branchTestNull(Condition cond, const ValueOperand& value,
                             Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  // Clobbers the ScratchReg on x64.
  inline void branchTestObject(Condition cond, const Address& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestObject(Condition cond, const BaseIndex& address,
                               Label* label) PER_SHARED_ARCH;
  inline void branchTestObject(Condition cond, const ValueOperand& value,
                               Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestGCThing(Condition cond, const Address& address,
                                Label* label) PER_SHARED_ARCH;
  inline void branchTestGCThing(Condition cond, const BaseIndex& address,
                                Label* label) PER_SHARED_ARCH;
  inline void branchTestGCThing(Condition cond, const ValueOperand& value,
                                Label* label) PER_SHARED_ARCH;

  inline void branchTestPrimitive(Condition cond, const ValueOperand& value,
                                  Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestMagic(Condition cond, const Address& address,
                              Label* label) PER_SHARED_ARCH;
  inline void branchTestMagic(Condition cond, const BaseIndex& address,
                              Label* label) PER_SHARED_ARCH;
  template <class L>
  inline void branchTestMagic(Condition cond, const ValueOperand& value,
                              L label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  inline void branchTestMagic(Condition cond, const Address& valaddr,
                              JSWhyMagic why, Label* label) PER_ARCH;

  inline void branchTestMagicValue(Condition cond, const ValueOperand& val,
                                   JSWhyMagic why, Label* label);

  void branchTestValue(Condition cond, const ValueOperand& lhs,
                       const Value& rhs, Label* label) PER_ARCH;

  inline void branchTestValue(Condition cond, const BaseIndex& lhs,
                              const ValueOperand& rhs, Label* label) PER_ARCH;

  // Checks if given Value is evaluated to true or false in a condition.
  // The type of the value should match the type of the method.
  inline void branchTestInt32Truthy(bool truthy, const ValueOperand& value,
                                    Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, x86_shared,
                 wasm32);
  inline void branchTestDoubleTruthy(bool truthy, FloatRegister reg,
                                     Label* label) PER_SHARED_ARCH;
  inline void branchTestBooleanTruthy(bool truthy, const ValueOperand& value,
                                      Label* label) PER_ARCH;
  inline void branchTestStringTruthy(bool truthy, const ValueOperand& value,
                                     Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);
  inline void branchTestBigIntTruthy(bool truthy, const ValueOperand& value,
                                     Label* label)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, wasm32,
                 x86_shared);

  // Create an unconditional branch to the address given as argument.
  inline void branchToComputedAddress(const BaseIndex& address) PER_ARCH;

 private:
  template <typename T, typename S, typename L>
  inline void branchPtrImpl(Condition cond, const T& lhs, const S& rhs, L label)
      DEFINED_ON(x86_shared);

  void branchPtrInNurseryChunkImpl(Condition cond, Register ptr, Label* label)
      DEFINED_ON(x86);
  template <typename T>
  void branchValueIsNurseryCellImpl(Condition cond, const T& value,
                                    Register temp, Label* label)
      DEFINED_ON(arm64, x64, mips64, loong64, riscv64);

  template <typename T>
  inline void branchTestUndefinedImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestInt32Impl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestDoubleImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestNumberImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestBooleanImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestStringImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestSymbolImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestBigIntImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestNullImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestObjectImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T>
  inline void branchTestGCThingImpl(Condition cond, const T& t,
                                    Label* label) PER_SHARED_ARCH;
  template <typename T>
  inline void branchTestPrimitiveImpl(Condition cond, const T& t, Label* label)
      DEFINED_ON(arm, arm64, x86_shared);
  template <typename T, class L>
  inline void branchTestMagicImpl(Condition cond, const T& t, L label)
      DEFINED_ON(arm, arm64, x86_shared);

 public:
  template <typename T>
  inline void testNumberSet(Condition cond, const T& src,
                            Register dest) PER_SHARED_ARCH;
  template <typename T>
  inline void testBooleanSet(Condition cond, const T& src,
                             Register dest) PER_SHARED_ARCH;
  template <typename T>
  inline void testStringSet(Condition cond, const T& src,
                            Register dest) PER_SHARED_ARCH;
  template <typename T>
  inline void testSymbolSet(Condition cond, const T& src,
                            Register dest) PER_SHARED_ARCH;
  template <typename T>
  inline void testBigIntSet(Condition cond, const T& src,
                            Register dest) PER_SHARED_ARCH;

 public:
  // The fallibleUnbox* methods below combine a Value type check with an unbox.
  // Especially on 64-bit platforms this can be implemented more efficiently
  // than a separate branch + unbox.
  //
  // |src| and |dest| can be the same register, but |dest| may hold garbage on
  // failure.
  inline void fallibleUnboxPtr(const ValueOperand& src, Register dest,
                               JSValueType type, Label* fail) PER_ARCH;
  inline void fallibleUnboxPtr(const Address& src, Register dest,
                               JSValueType type, Label* fail) PER_ARCH;
  inline void fallibleUnboxPtr(const BaseIndex& src, Register dest,
                               JSValueType type, Label* fail) PER_ARCH;
  template <typename T>
  inline void fallibleUnboxInt32(const T& src, Register dest, Label* fail);
  template <typename T>
  inline void fallibleUnboxBoolean(const T& src, Register dest, Label* fail);
  template <typename T>
  inline void fallibleUnboxObject(const T& src, Register dest, Label* fail);
  template <typename T>
  inline void fallibleUnboxString(const T& src, Register dest, Label* fail);
  template <typename T>
  inline void fallibleUnboxSymbol(const T& src, Register dest, Label* fail);
  template <typename T>
  inline void fallibleUnboxBigInt(const T& src, Register dest, Label* fail);

  inline void cmp32Move32(Condition cond, Register lhs, Imm32 rhs, Register src,
                          Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86_shared);

  inline void cmp32Move32(Condition cond, Register lhs, Register rhs,
                          Register src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86_shared);

  inline void cmp32Move32(Condition cond, Register lhs, const Address& rhs,
                          Register src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86_shared);

  inline void cmpPtrMovePtr(Condition cond, Register lhs, Register rhs,
                            Register src, Register dest) PER_ARCH;

  inline void cmpPtrMovePtr(Condition cond, Register lhs, const Address& rhs,
                            Register src, Register dest) PER_ARCH;

  inline void cmp32Load32(Condition cond, Register lhs, const Address& rhs,
                          const Address& src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, mips_shared, x86_shared);

  inline void cmp32Load32(Condition cond, Register lhs, Register rhs,
                          const Address& src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, mips_shared, x86_shared);

  inline void cmp32Load32(Condition cond, Register lhs, Imm32 rhs,
                          const Address& src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86_shared);

  inline void cmp32LoadPtr(Condition cond, const Address& lhs, Imm32 rhs,
                           const Address& src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86, x64);

  inline void cmp32MovePtr(Condition cond, Register lhs, Imm32 rhs,
                           Register src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86, x64);

  inline void test32LoadPtr(Condition cond, const Address& addr, Imm32 mask,
                            const Address& src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86, x64);

  inline void test32MovePtr(Condition cond, const Address& addr, Imm32 mask,
                            Register src, Register dest)
      DEFINED_ON(arm, arm64, loong64, riscv64, wasm32, mips_shared, x86, x64);

  // Conditional move for Spectre mitigations.
  inline void spectreMovePtr(Condition cond, Register src, Register dest)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  // Zeroes dest if the condition is true.
  inline void spectreZeroRegister(Condition cond, Register scratch,
                                  Register dest)
      DEFINED_ON(arm, arm64, mips_shared, x86_shared, loong64, riscv64, wasm32);

  // Performs a bounds check and zeroes the index register if out-of-bounds
  // (to mitigate Spectre).
 private:
  inline void spectreBoundsCheck32(Register index, const Operand& length,
                                   Register maybeScratch, Label* failure)
      DEFINED_ON(x86);

 public:
  inline void spectreBoundsCheck32(Register index, Register length,
                                   Register maybeScratch, Label* failure)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  inline void spectreBoundsCheck32(Register index, const Address& length,
                                   Register maybeScratch, Label* failure)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  inline void spectreBoundsCheckPtr(Register index, Register length,
                                    Register maybeScratch, Label* failure)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);
  inline void spectreBoundsCheckPtr(Register index, const Address& length,
                                    Register maybeScratch, Label* failure)
      DEFINED_ON(arm, arm64, mips_shared, x86, x64, loong64, riscv64, wasm32);

  // ========================================================================
  // Canonicalization primitives.
  inline void canonicalizeDouble(FloatRegister reg);
  inline void canonicalizeDoubleIfDeterministic(FloatRegister reg);

  inline void canonicalizeFloat(FloatRegister reg);
  inline void canonicalizeFloatIfDeterministic(FloatRegister reg);

 public:
  // ========================================================================
  // Memory access primitives.
  inline FaultingCodeOffset storeUncanonicalizedDouble(FloatRegister src,
                                                       const Address& dest)
      DEFINED_ON(x86_shared, arm, arm64, mips32, mips64, loong64, riscv64,
                 wasm32);
  inline FaultingCodeOffset storeUncanonicalizedDouble(FloatRegister src,
                                                       const BaseIndex& dest)
      DEFINED_ON(x86_shared, arm, arm64, mips32, mips64, loong64, riscv64,
                 wasm32);
  inline FaultingCodeOffset storeUncanonicalizedDouble(FloatRegister src,
                                                       const Operand& dest)
      DEFINED_ON(x86_shared);

  template <class T>
  inline FaultingCodeOffset storeDouble(FloatRegister src, const T& dest);

  template <class T>
  inline void boxDouble(FloatRegister src, const T& dest);

  using MacroAssemblerSpecific::boxDouble;

  inline FaultingCodeOffset storeUncanonicalizedFloat32(FloatRegister src,
                                                        const Address& dest)
      DEFINED_ON(x86_shared, arm, arm64, mips32, mips64, loong64, riscv64,
                 wasm32);
  inline FaultingCodeOffset storeUncanonicalizedFloat32(FloatRegister src,
                                                        const BaseIndex& dest)
      DEFINED_ON(x86_shared, arm, arm64, mips32, mips64, loong64, riscv64,
                 wasm32);
  inline FaultingCodeOffset storeUncanonicalizedFloat32(FloatRegister src,
                                                        const Operand& dest)
      DEFINED_ON(x86_shared);

  template <class T>
  inline FaultingCodeOffset storeFloat32(FloatRegister src, const T& dest);

  template <typename T>
  void storeUnboxedValue(const ConstantOrRegister& value, MIRType valueType,
                         const T& dest) PER_ARCH;

  inline void memoryBarrier(MemoryBarrierBits barrier) PER_SHARED_ARCH;

 public:
  // ========================================================================
  // Wasm SIMD
  //
  // Naming is "operationSimd128" when operate on the whole vector, otherwise
  // it's "operation<Type><Size>x<Lanes>".
  //
  // For microarchitectural reasons we can in principle get a performance win by
  // using int or float specific instructions in the operationSimd128 case when
  // we know that subsequent operations on the result are int or float oriented.
  // In practice, we don't care about that yet.
  //
  // The order of operations here follows those in the SIMD overview document,
  // https://github.com/WebAssembly/simd/blob/master/proposals/simd/SIMD.md.
  //
  // Since we must target Intel SSE indefinitely and SSE is one-address or
  // two-address, the x86 porting interfaces are nearly all one-address or
  // two-address.  Likewise there are two-address ARM64 interfaces to support
  // the baseline compiler.  But there are also three-address ARM64 interfaces
  // as the ARM64 Ion back-end can use those.  In the future, they may support
  // AVX2 or similar for x86.
  //
  // Conventions for argument order and naming and semantics:
  //  - Condition codes come first.
  //  - Other immediates (masks, shift counts) come next.
  //  - Operands come next:
  //    - For a binary two-address operator where the left-hand-side has the
  //      same type as the result, one register parameter is normally named
  //      `lhsDest` and is both the left-hand side and destination; the other
  //      parameter is named `rhs` and is the right-hand side.  `rhs` comes
  //      first, `lhsDest` second.  `rhs` and `lhsDest` may be the same register
  //      (if rhs is a register).
  //    - For a binary three-address operator the order is `lhs`, `rhs`, `dest`,
  //      and generally these registers may be the same.
  //    - For a unary operator, the input is named `src` and the output is named
  //      `dest`.  `src` comes first, `dest` second.  `src` and `dest` may be
  //      the same register (if `src` is a register).
  //  - Temp registers follow operands and are named `temp` if there's only one,
  //    otherwise `temp1`, `temp2`, etc regardless of type.  GPR temps precede
  //    FPU temps.  If there are several temps then they must be distinct
  //    registers, and they must be distinct from the operand registers unless
  //    noted.

  // Moves

  inline void moveSimd128(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Constants

  inline void loadConstantSimd128(const SimdConstant& v, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Splat

  inline void splatX16(Register src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void splatX16(uint32_t srcLane, FloatRegister src, FloatRegister dest)
      DEFINED_ON(arm64);

  inline void splatX8(Register src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void splatX8(uint32_t srcLane, FloatRegister src, FloatRegister dest)
      DEFINED_ON(arm64);

  inline void splatX4(Register src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void splatX4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void splatX2(Register64 src, FloatRegister dest)
      DEFINED_ON(x86, x64, arm64);

  inline void splatX2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Extract lane as scalar.  Float extraction does not canonicalize the value.

  inline void extractLaneInt8x16(uint32_t lane, FloatRegister src,
                                 Register dest) DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtractLaneInt8x16(uint32_t lane, FloatRegister src,
                                         Register dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extractLaneInt16x8(uint32_t lane, FloatRegister src,
                                 Register dest) DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtractLaneInt16x8(uint32_t lane, FloatRegister src,
                                         Register dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extractLaneInt32x4(uint32_t lane, FloatRegister src,
                                 Register dest) DEFINED_ON(x86_shared, arm64);

  inline void extractLaneInt64x2(uint32_t lane, FloatRegister src,
                                 Register64 dest) DEFINED_ON(x86, x64, arm64);

  inline void extractLaneFloat32x4(uint32_t lane, FloatRegister src,
                                   FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extractLaneFloat64x2(uint32_t lane, FloatRegister src,
                                   FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Replace lane value

  inline void replaceLaneInt8x16(unsigned lane, FloatRegister lhs, Register rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void replaceLaneInt8x16(unsigned lane, Register rhs,
                                 FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void replaceLaneInt16x8(unsigned lane, FloatRegister lhs, Register rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void replaceLaneInt16x8(unsigned lane, Register rhs,
                                 FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void replaceLaneInt32x4(unsigned lane, FloatRegister lhs, Register rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void replaceLaneInt32x4(unsigned lane, Register rhs,
                                 FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void replaceLaneInt64x2(unsigned lane, FloatRegister lhs,
                                 Register64 rhs, FloatRegister dest)
      DEFINED_ON(x86, x64);

  inline void replaceLaneInt64x2(unsigned lane, Register64 rhs,
                                 FloatRegister lhsDest)
      DEFINED_ON(x86, x64, arm64);

  inline void replaceLaneFloat32x4(unsigned lane, FloatRegister lhs,
                                   FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void replaceLaneFloat32x4(unsigned lane, FloatRegister rhs,
                                   FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void replaceLaneFloat64x2(unsigned lane, FloatRegister lhs,
                                   FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void replaceLaneFloat64x2(unsigned lane, FloatRegister rhs,
                                   FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  // Shuffle - blend and permute with immediate indices, and its many
  // specializations.  Lane values other than those mentioned are illegal.

  // lane values 0..31
  inline void shuffleInt8x16(const uint8_t lanes[16], FloatRegister rhs,
                             FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void shuffleInt8x16(const uint8_t lanes[16], FloatRegister lhs,
                             FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Lane values must be 0 (select from lhs) or FF (select from rhs).
  // The behavior is undefined for lane values that are neither 0 nor FF.
  // on x86_shared: it is required that lhs == dest.
  inline void blendInt8x16(const uint8_t lanes[16], FloatRegister lhs,
                           FloatRegister rhs, FloatRegister dest,
                           FloatRegister temp) DEFINED_ON(x86_shared);

  // Lane values must be 0 (select from lhs) or FF (select from rhs).
  // The behavior is undefined for lane values that are neither 0 nor FF.
  inline void blendInt8x16(const uint8_t lanes[16], FloatRegister lhs,
                           FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(arm64);

  // Lane values must be 0 (select from lhs) or FFFF (select from rhs).
  // The behavior is undefined for lane values that are neither 0 nor FFFF.
  // on x86_shared: it is required that lhs == dest.
  inline void blendInt16x8(const uint16_t lanes[8], FloatRegister lhs,
                           FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Mask lane values must be ~0 or 0. The former selects from lhs and the
  // latter from rhs.
  // The implementation works effectively for I8x16, I16x8, I32x4, and I64x2.
  inline void laneSelectSimd128(FloatRegister mask, FloatRegister lhs,
                                FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveHighInt8x16(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveHighInt16x8(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveHighInt32x4(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveHighInt64x2(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveLowInt8x16(FloatRegister lhs, FloatRegister rhs,
                                   FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveLowInt16x8(FloatRegister lhs, FloatRegister rhs,
                                   FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveLowInt32x4(FloatRegister lhs, FloatRegister rhs,
                                   FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void interleaveLowInt64x2(FloatRegister lhs, FloatRegister rhs,
                                   FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Permute - permute with immediate indices.

  // lane values 0..15
  inline void permuteInt8x16(const uint8_t lanes[16], FloatRegister src,
                             FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  // lane values 0..7
  inline void permuteInt16x8(const uint16_t lanes[8], FloatRegister src,
                             FloatRegister dest) DEFINED_ON(arm64);

  // lane values 0..3 [sic].
  inline void permuteHighInt16x8(const uint16_t lanes[4], FloatRegister src,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  // lane values 0..3.
  inline void permuteLowInt16x8(const uint16_t lanes[4], FloatRegister src,
                                FloatRegister dest) DEFINED_ON(x86_shared);

  // lane values 0..3
  inline void permuteInt32x4(const uint32_t lanes[4], FloatRegister src,
                             FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  // Funnel shift by immediate count:
  //   low_16_bytes_of((lhs ++ rhs) >> shift*8), shift must be < 16
  inline void concatAndRightShiftSimd128(FloatRegister lhs, FloatRegister rhs,
                                         FloatRegister dest, uint32_t shift)
      DEFINED_ON(x86_shared, arm64);

  // Rotate right by immediate count:
  //   low_16_bytes_of((src ++ src) >> shift*8), shift must be < 16
  inline void rotateRightSimd128(FloatRegister src, FloatRegister dest,
                                 uint32_t shift) DEFINED_ON(arm64);

  // Shift bytes with immediate count, shifting in zeroes.  Shift count 0..15.

  inline void leftShiftSimd128(Imm32 count, FloatRegister src,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void rightShiftSimd128(Imm32 count, FloatRegister src,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Zero extend int values.

  inline void zeroExtend8x16To16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);
  inline void zeroExtend8x16To32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);
  inline void zeroExtend8x16To64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);
  inline void zeroExtend16x8To32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);
  inline void zeroExtend16x8To64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);
  inline void zeroExtend32x4To64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Reverse bytes in lanes.

  inline void reverseInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void reverseInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void reverseInt64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Swizzle - permute with variable indices.  `rhs` holds the lanes parameter.

  inline void swizzleInt8x16(FloatRegister lhs, FloatRegister rhs,
                             FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void swizzleInt8x16Relaxed(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Integer Add

  inline void addInt8x16(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void addInt16x8(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void addInt32x4(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void addInt64x2(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addInt64x2(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  // Integer Subtract

  inline void subInt8x16(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void subInt16x8(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void subInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void subInt32x4(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subInt64x2(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void subInt64x2(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  // Integer Multiply

  inline void mulInt16x8(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void mulInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void mulInt32x4(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void mulInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  // On x86_shared, it is required lhs == dest
  inline void mulInt64x2(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest, FloatRegister temp)
      DEFINED_ON(x86_shared);

  inline void mulInt64x2(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest, FloatRegister temp)
      DEFINED_ON(x86_shared);

  inline void mulInt64x2(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest, FloatRegister temp1,
                         FloatRegister temp2) DEFINED_ON(arm64);

  // Note for the extMul opcodes, the NxM designation is for the input lanes;
  // the output lanes are twice as wide.
  inline void extMulLowInt8x16(FloatRegister lhs, FloatRegister rhs,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extMulHighInt8x16(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtMulLowInt8x16(FloatRegister lhs, FloatRegister rhs,
                                       FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtMulHighInt8x16(FloatRegister lhs, FloatRegister rhs,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extMulLowInt16x8(FloatRegister lhs, FloatRegister rhs,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extMulHighInt16x8(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtMulLowInt16x8(FloatRegister lhs, FloatRegister rhs,
                                       FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtMulHighInt16x8(FloatRegister lhs, FloatRegister rhs,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extMulLowInt32x4(FloatRegister lhs, FloatRegister rhs,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extMulHighInt32x4(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtMulLowInt32x4(FloatRegister lhs, FloatRegister rhs,
                                       FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtMulHighInt32x4(FloatRegister lhs, FloatRegister rhs,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void q15MulrSatInt16x8(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Integer Negate

  inline void negInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void negInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void negInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void negInt64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Saturating integer add

  inline void addSatInt8x16(FloatRegister lhs, FloatRegister rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addSatInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedAddSatInt8x16(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedAddSatInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                                    FloatRegister dest) DEFINED_ON(x86_shared);

  inline void addSatInt16x8(FloatRegister lhs, FloatRegister rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addSatInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedAddSatInt16x8(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedAddSatInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                                    FloatRegister dest) DEFINED_ON(x86_shared);

  // Saturating integer subtract

  inline void subSatInt8x16(FloatRegister lhs, FloatRegister rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subSatInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedSubSatInt8x16(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedSubSatInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                                    FloatRegister dest) DEFINED_ON(x86_shared);

  inline void subSatInt16x8(FloatRegister lhs, FloatRegister rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subSatInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedSubSatInt16x8(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedSubSatInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                                    FloatRegister dest) DEFINED_ON(x86_shared);

  // Lane-wise integer minimum

  inline void minInt8x16(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void minInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedMinInt8x16(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedMinInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void minInt16x8(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void minInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedMinInt16x8(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedMinInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void minInt32x4(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void minInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedMinInt32x4(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedMinInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  // Lane-wise integer maximum

  inline void maxInt8x16(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void maxInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedMaxInt8x16(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedMaxInt8x16(FloatRegister lhs, const SimdConstant& rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void maxInt16x8(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void maxInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedMaxInt16x8(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedMaxInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  inline void maxInt32x4(FloatRegister lhs, FloatRegister rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void maxInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                         FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedMaxInt32x4(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedMaxInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                                 FloatRegister dest) DEFINED_ON(x86_shared);

  // Lane-wise integer rounding average

  inline void unsignedAverageInt8x16(FloatRegister lhs, FloatRegister rhs,
                                     FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedAverageInt16x8(FloatRegister lhs, FloatRegister rhs,
                                     FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Lane-wise integer absolute value

  inline void absInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void absInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void absInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void absInt64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Left shift by scalar. Immediates and variable shifts must have been
  // masked; shifts of zero will work but may or may not generate code.

  inline void leftShiftInt8x16(Register rhs, FloatRegister lhsDest,
                               FloatRegister temp) DEFINED_ON(x86_shared);

  inline void leftShiftInt8x16(FloatRegister lhs, Register rhs,
                               FloatRegister dest) DEFINED_ON(arm64);

  inline void leftShiftInt8x16(Imm32 count, FloatRegister src,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void leftShiftInt16x8(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void leftShiftInt16x8(FloatRegister lhs, Register rhs,
                               FloatRegister dest) DEFINED_ON(arm64);

  inline void leftShiftInt16x8(Imm32 count, FloatRegister src,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void leftShiftInt32x4(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void leftShiftInt32x4(FloatRegister lhs, Register rhs,
                               FloatRegister dest) DEFINED_ON(arm64);

  inline void leftShiftInt32x4(Imm32 count, FloatRegister src,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void leftShiftInt64x2(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void leftShiftInt64x2(FloatRegister lhs, Register rhs,
                               FloatRegister dest) DEFINED_ON(arm64);

  inline void leftShiftInt64x2(Imm32 count, FloatRegister src,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Right shift by scalar. Immediates and variable shifts must have been
  // masked; shifts of zero will work but may or may not generate code.

  inline void rightShiftInt8x16(Register rhs, FloatRegister lhsDest,
                                FloatRegister temp) DEFINED_ON(x86_shared);

  inline void rightShiftInt8x16(FloatRegister lhs, Register rhs,
                                FloatRegister dest) DEFINED_ON(arm64);

  inline void rightShiftInt8x16(Imm32 count, FloatRegister src,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedRightShiftInt8x16(Register rhs, FloatRegister lhsDest,
                                        FloatRegister temp)
      DEFINED_ON(x86_shared);

  inline void unsignedRightShiftInt8x16(FloatRegister lhs, Register rhs,
                                        FloatRegister dest) DEFINED_ON(arm64);

  inline void unsignedRightShiftInt8x16(Imm32 count, FloatRegister src,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void rightShiftInt16x8(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void rightShiftInt16x8(FloatRegister lhs, Register rhs,
                                FloatRegister dest) DEFINED_ON(arm64);

  inline void rightShiftInt16x8(Imm32 count, FloatRegister src,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedRightShiftInt16x8(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void unsignedRightShiftInt16x8(FloatRegister lhs, Register rhs,
                                        FloatRegister dest) DEFINED_ON(arm64);

  inline void unsignedRightShiftInt16x8(Imm32 count, FloatRegister src,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void rightShiftInt32x4(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void rightShiftInt32x4(FloatRegister lhs, Register rhs,
                                FloatRegister dest) DEFINED_ON(arm64);

  inline void rightShiftInt32x4(Imm32 count, FloatRegister src,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedRightShiftInt32x4(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void unsignedRightShiftInt32x4(FloatRegister lhs, Register rhs,
                                        FloatRegister dest) DEFINED_ON(arm64);

  inline void unsignedRightShiftInt32x4(Imm32 count, FloatRegister src,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void rightShiftInt64x2(Register rhs, FloatRegister lhsDest,
                                FloatRegister temp) DEFINED_ON(x86_shared);

  inline void rightShiftInt64x2(Imm32 count, FloatRegister src,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void rightShiftInt64x2(FloatRegister lhs, Register rhs,
                                FloatRegister dest) DEFINED_ON(arm64);

  inline void unsignedRightShiftInt64x2(Register rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  inline void unsignedRightShiftInt64x2(FloatRegister lhs, Register rhs,
                                        FloatRegister dest) DEFINED_ON(arm64);

  inline void unsignedRightShiftInt64x2(Imm32 count, FloatRegister src,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Sign replication operation

  inline void signReplicationInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void signReplicationInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void signReplicationInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void signReplicationInt64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared);

  // Bitwise and, or, xor, not

  inline void bitwiseAndSimd128(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseAndSimd128(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseAndSimd128(FloatRegister lhs, const SimdConstant& rhs,
                                FloatRegister dest) DEFINED_ON(x86_shared);

  inline void bitwiseOrSimd128(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseOrSimd128(FloatRegister lhs, FloatRegister rhs,
                               FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseOrSimd128(FloatRegister lhs, const SimdConstant& rhs,
                               FloatRegister dest) DEFINED_ON(x86_shared);

  inline void bitwiseXorSimd128(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseXorSimd128(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseXorSimd128(FloatRegister lhs, const SimdConstant& rhs,
                                FloatRegister dest) DEFINED_ON(x86_shared);

  inline void bitwiseNotSimd128(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Bitwise AND with compliment: dest = lhs & ~rhs, note only arm64 can do it.
  inline void bitwiseAndNotSimd128(FloatRegister lhs, FloatRegister rhs,
                                   FloatRegister lhsDest) DEFINED_ON(arm64);

  // Bitwise AND with complement: dest = ~lhs & rhs, note this is not what Wasm
  // wants but what the x86 hardware offers.  Hence the name.

  inline void bitwiseNotAndSimd128(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void bitwiseNotAndSimd128(FloatRegister lhs, FloatRegister rhs,
                                   FloatRegister lhsDest)
      DEFINED_ON(x86_shared);

  // Bitwise select

  inline void bitwiseSelectSimd128(FloatRegister mask, FloatRegister onTrue,
                                   FloatRegister onFalse, FloatRegister dest,
                                   FloatRegister temp) DEFINED_ON(x86_shared);

  inline void bitwiseSelectSimd128(FloatRegister onTrue, FloatRegister onFalse,
                                   FloatRegister maskDest) DEFINED_ON(arm64);

  // Population count

  inline void popcntInt8x16(FloatRegister src, FloatRegister dest,
                            FloatRegister temp) DEFINED_ON(x86_shared);

  inline void popcntInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(arm64);

  // Any lane true, ie, any bit set

  inline void anyTrueSimd128(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared, arm64);

  // All lanes true

  inline void allTrueInt8x16(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared, arm64);

  inline void allTrueInt16x8(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared, arm64);

  inline void allTrueInt32x4(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared, arm64);

  inline void allTrueInt64x2(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared, arm64);

  // Bitmask, ie extract and compress high bits of all lanes

  inline void bitmaskInt8x16(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared);

  inline void bitmaskInt8x16(FloatRegister src, Register dest,
                             FloatRegister temp) DEFINED_ON(arm64);

  inline void bitmaskInt16x8(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared);

  inline void bitmaskInt16x8(FloatRegister src, Register dest,
                             FloatRegister temp) DEFINED_ON(arm64);

  inline void bitmaskInt32x4(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared);

  inline void bitmaskInt32x4(FloatRegister src, Register dest,
                             FloatRegister temp) DEFINED_ON(arm64);

  inline void bitmaskInt64x2(FloatRegister src, Register dest)
      DEFINED_ON(x86_shared);

  inline void bitmaskInt64x2(FloatRegister src, Register dest,
                             FloatRegister temp) DEFINED_ON(arm64);

  // Comparisons (integer and floating-point)

  inline void compareInt8x16(Assembler::Condition cond, FloatRegister rhs,
                             FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  // On x86_shared, limited to !=, ==, <=, >
  inline void compareInt8x16(Assembler::Condition cond, FloatRegister lhs,
                             const SimdConstant& rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  // On arm64, use any integer comparison condition.
  inline void compareInt8x16(Assembler::Condition cond, FloatRegister lhs,
                             FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void compareInt16x8(Assembler::Condition cond, FloatRegister rhs,
                             FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void compareInt16x8(Assembler::Condition cond, FloatRegister lhs,
                             FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // On x86_shared, limited to !=, ==, <=, >
  inline void compareInt16x8(Assembler::Condition cond, FloatRegister lhs,
                             const SimdConstant& rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  // On x86_shared, limited to !=, ==, <=, >
  inline void compareInt32x4(Assembler::Condition cond, FloatRegister rhs,
                             FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void compareInt32x4(Assembler::Condition cond, FloatRegister lhs,
                             const SimdConstant& rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  // On arm64, use any integer comparison condition.
  inline void compareInt32x4(Assembler::Condition cond, FloatRegister lhs,
                             FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void compareForEqualityInt64x2(Assembler::Condition cond,
                                        FloatRegister lhs, FloatRegister rhs,
                                        FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void compareForOrderingInt64x2(Assembler::Condition cond,
                                        FloatRegister lhs, FloatRegister rhs,
                                        FloatRegister dest, FloatRegister temp1,
                                        FloatRegister temp2)
      DEFINED_ON(x86_shared);

  inline void compareInt64x2(Assembler::Condition cond, FloatRegister rhs,
                             FloatRegister lhsDest) DEFINED_ON(arm64);

  inline void compareInt64x2(Assembler::Condition cond, FloatRegister lhs,
                             FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(arm64);

  inline void compareFloat32x4(Assembler::Condition cond, FloatRegister rhs,
                               FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  // On x86_shared, limited to ==, !=, <, <=
  inline void compareFloat32x4(Assembler::Condition cond, FloatRegister lhs,
                               const SimdConstant& rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  // On x86_shared, limited to ==, !=, <, <=
  // On arm64, use any float-point comparison condition.
  inline void compareFloat32x4(Assembler::Condition cond, FloatRegister lhs,
                               FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void compareFloat64x2(Assembler::Condition cond, FloatRegister rhs,
                               FloatRegister lhsDest)
      DEFINED_ON(x86_shared, arm64);

  // On x86_shared, limited to ==, !=, <, <=
  inline void compareFloat64x2(Assembler::Condition cond, FloatRegister lhs,
                               const SimdConstant& rhs, FloatRegister dest)
      DEFINED_ON(x86_shared);

  // On x86_shared, limited to ==, !=, <, <=
  // On arm64, use any float-point comparison condition.
  inline void compareFloat64x2(Assembler::Condition cond, FloatRegister lhs,
                               FloatRegister rhs, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Load

  inline void loadUnalignedSimd128(const Operand& src, FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline FaultingCodeOffset loadUnalignedSimd128(const Address& src,
                                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline FaultingCodeOffset loadUnalignedSimd128(const BaseIndex& src,
                                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Store

  inline FaultingCodeOffset storeUnalignedSimd128(FloatRegister src,
                                                  const Address& dest)
      DEFINED_ON(x86_shared, arm64);

  inline FaultingCodeOffset storeUnalignedSimd128(FloatRegister src,
                                                  const BaseIndex& dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating point negation

  inline void negFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void negFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating point absolute value

  inline void absFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void absFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // NaN-propagating minimum

  inline void minFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest, FloatRegister temp1,
                           FloatRegister temp2) DEFINED_ON(x86_shared);

  inline void minFloat32x4(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(arm64);

  inline void minFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(arm64);

  inline void minFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest, FloatRegister temp1,
                           FloatRegister temp2) DEFINED_ON(x86_shared);

  inline void minFloat64x2(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(arm64);

  inline void minFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(arm64);

  // NaN-propagating maximum

  inline void maxFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest, FloatRegister temp1,
                           FloatRegister temp2) DEFINED_ON(x86_shared);

  inline void maxFloat32x4(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(arm64);

  inline void maxFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(arm64);

  inline void maxFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest, FloatRegister temp1,
                           FloatRegister temp2) DEFINED_ON(x86_shared);

  inline void maxFloat64x2(FloatRegister rhs, FloatRegister lhsDest)
      DEFINED_ON(arm64);

  inline void maxFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(arm64);

  // Floating add

  inline void addFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addFloat32x4(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  inline void addFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void addFloat64x2(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  // Floating subtract

  inline void subFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subFloat32x4(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  inline void subFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void subFloat64x2(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  // Floating division

  inline void divFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void divFloat32x4(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  inline void divFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void divFloat64x2(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  // Floating Multiply

  inline void mulFloat32x4(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void mulFloat32x4(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  inline void mulFloat64x2(FloatRegister lhs, FloatRegister rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void mulFloat64x2(FloatRegister lhs, const SimdConstant& rhs,
                           FloatRegister dest) DEFINED_ON(x86_shared);

  // Pairwise add

  inline void extAddPairwiseInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtAddPairwiseInt8x16(FloatRegister src,
                                            FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void extAddPairwiseInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedExtAddPairwiseInt16x8(FloatRegister src,
                                            FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating square root

  inline void sqrtFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void sqrtFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Integer to floating point with rounding

  inline void convertInt32x4ToFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedConvertInt32x4ToFloat32x4(FloatRegister src,
                                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void convertInt32x4ToFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedConvertInt32x4ToFloat64x2(FloatRegister src,
                                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating point to integer with saturation

  inline void truncSatFloat32x4ToInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedTruncSatFloat32x4ToInt32x4(FloatRegister src,
                                                 FloatRegister dest,
                                                 FloatRegister temp)
      DEFINED_ON(x86_shared);

  inline void unsignedTruncSatFloat32x4ToInt32x4(FloatRegister src,
                                                 FloatRegister dest)
      DEFINED_ON(arm64);

  inline void truncSatFloat64x2ToInt32x4(FloatRegister src, FloatRegister dest,
                                         FloatRegister temp)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedTruncSatFloat64x2ToInt32x4(FloatRegister src,
                                                 FloatRegister dest,
                                                 FloatRegister temp)
      DEFINED_ON(x86_shared, arm64);

  inline void truncFloat32x4ToInt32x4Relaxed(FloatRegister src,
                                             FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedTruncFloat32x4ToInt32x4Relaxed(FloatRegister src,
                                                     FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void truncFloat64x2ToInt32x4Relaxed(FloatRegister src,
                                             FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedTruncFloat64x2ToInt32x4Relaxed(FloatRegister src,
                                                     FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating point narrowing

  inline void convertFloat64x2ToFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating point widening

  inline void convertFloat32x4ToFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Integer to integer narrowing

  inline void narrowInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared);

  inline void narrowInt16x8(FloatRegister lhs, FloatRegister rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void unsignedNarrowInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                                    FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedNarrowInt16x8(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void narrowInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared);

  inline void narrowInt32x4(FloatRegister lhs, FloatRegister rhs,
                            FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void unsignedNarrowInt32x4(FloatRegister lhs, const SimdConstant& rhs,
                                    FloatRegister dest) DEFINED_ON(x86_shared);

  inline void unsignedNarrowInt32x4(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Integer to integer widening

  inline void widenLowInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void widenHighInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedWidenLowInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedWidenHighInt8x16(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void widenLowInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void widenHighInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedWidenLowInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedWidenHighInt16x8(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void widenLowInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedWidenLowInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void widenHighInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void unsignedWidenHighInt32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Compare-based minimum/maximum
  //
  // On x86, the signature is (rhsDest, lhs); on arm64 it is (rhs, lhsDest).
  //
  // The masm preprocessor can't deal with multiple declarations with identical
  // signatures even if they are on different platforms, hence the weird
  // argument names.

  inline void pseudoMinFloat32x4(FloatRegister rhsOrRhsDest,
                                 FloatRegister lhsOrLhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMinFloat32x4(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMinFloat64x2(FloatRegister rhsOrRhsDest,
                                 FloatRegister lhsOrLhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMinFloat64x2(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMaxFloat32x4(FloatRegister rhsOrRhsDest,
                                 FloatRegister lhsOrLhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMaxFloat32x4(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMaxFloat64x2(FloatRegister rhsOrRhsDest,
                                 FloatRegister lhsOrLhsDest)
      DEFINED_ON(x86_shared, arm64);

  inline void pseudoMaxFloat64x2(FloatRegister lhs, FloatRegister rhs,
                                 FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Widening/pairwise integer dot product

  inline void widenDotInt16x8(FloatRegister lhs, FloatRegister rhs,
                              FloatRegister dest) DEFINED_ON(x86_shared, arm64);

  inline void widenDotInt16x8(FloatRegister lhs, const SimdConstant& rhs,
                              FloatRegister dest) DEFINED_ON(x86_shared);

  inline void dotInt8x16Int7x16(FloatRegister lhs, FloatRegister rhs,
                                FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void dotInt8x16Int7x16ThenAdd(FloatRegister lhs, FloatRegister rhs,
                                       FloatRegister dest)
      DEFINED_ON(x86_shared);

  inline void dotInt8x16Int7x16ThenAdd(FloatRegister lhs, FloatRegister rhs,
                                       FloatRegister dest, FloatRegister temp)
      DEFINED_ON(arm64);

  // Floating point rounding

  inline void ceilFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void ceilFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void floorFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void floorFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void truncFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void truncFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void nearestFloat32x4(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void nearestFloat64x2(FloatRegister src, FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  // Floating multiply-accumulate: srcDest [+-]= src1 * src2

  inline void fmaFloat32x4(FloatRegister src1, FloatRegister src2,
                           FloatRegister srcDest) DEFINED_ON(x86_shared, arm64);

  inline void fnmaFloat32x4(FloatRegister src1, FloatRegister src2,
                            FloatRegister srcDest)
      DEFINED_ON(x86_shared, arm64);

  inline void fmaFloat64x2(FloatRegister src1, FloatRegister src2,
                           FloatRegister srcDest) DEFINED_ON(x86_shared, arm64);

  inline void fnmaFloat64x2(FloatRegister src1, FloatRegister src2,
                            FloatRegister srcDest)
      DEFINED_ON(x86_shared, arm64);

  inline void minFloat32x4Relaxed(FloatRegister src, FloatRegister srcDest)
      DEFINED_ON(x86_shared, arm64);

  inline void minFloat32x4Relaxed(FloatRegister lhs, FloatRegister rhs,
                                  FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void maxFloat32x4Relaxed(FloatRegister src, FloatRegister srcDest)
      DEFINED_ON(x86_shared, arm64);

  inline void maxFloat32x4Relaxed(FloatRegister lhs, FloatRegister rhs,
                                  FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void minFloat64x2Relaxed(FloatRegister src, FloatRegister srcDest)
      DEFINED_ON(x86_shared, arm64);

  inline void minFloat64x2Relaxed(FloatRegister lhs, FloatRegister rhs,
                                  FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void maxFloat64x2Relaxed(FloatRegister src, FloatRegister srcDest)
      DEFINED_ON(x86_shared, arm64);

  inline void maxFloat64x2Relaxed(FloatRegister lhs, FloatRegister rhs,
                                  FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

  inline void q15MulrInt16x8Relaxed(FloatRegister lhs, FloatRegister rhs,
                                    FloatRegister dest)
      DEFINED_ON(x86_shared, arm64);

 public:
  // ========================================================================
  // Truncate floating point.

  // Undefined behaviour when truncation is outside Int64 range.
  // Needs a temp register if SSE3 is not present.
  inline void truncateFloat32ToInt64(Address src, Address dest, Register temp)
      DEFINED_ON(x86_shared);
  inline void truncateFloat32ToUInt64(Address src, Address dest, Register temp,
                                      FloatRegister floatTemp)
      DEFINED_ON(x86, x64);
  inline void truncateDoubleToInt64(Address src, Address dest, Register temp)
      DEFINED_ON(x86_shared);
  inline void truncateDoubleToUInt64(Address src, Address dest, Register temp,
                                     FloatRegister floatTemp)
      DEFINED_ON(x86, x64);

 public:
  // ========================================================================
  // Convert floating point.

  // temp required on x86 and x64; must be undefined on mips64 and loong64.
  void convertUInt64ToFloat32(Register64 src, FloatRegister dest, Register temp)
      DEFINED_ON(arm64, mips64, loong64, riscv64, wasm32, x64, x86);

  void convertInt64ToFloat32(Register64 src, FloatRegister dest)
      DEFINED_ON(arm64, mips64, loong64, riscv64, wasm32, x64, x86);

  bool convertUInt64ToDoubleNeedsTemp() PER_ARCH;

  // temp required when convertUInt64ToDoubleNeedsTemp() returns true.
  void convertUInt64ToDouble(Register64 src, FloatRegister dest,
                             Register temp) PER_ARCH;

  void convertInt64ToDouble(Register64 src, FloatRegister dest) PER_ARCH;

  void convertIntPtrToDouble(Register src, FloatRegister dest) PER_ARCH;

 public:
  // ========================================================================
  // wasm support

  FaultingCodeOffset wasmTrapInstruction() PER_SHARED_ARCH;

  void wasmTrap(wasm::Trap trap, wasm::BytecodeOffset bytecodeOffset);

  // Load all pinned regs via InstanceReg.  If the trapOffset is something,
  // give the first load a trap descriptor with type IndirectCallToNull, so that
  // a null instance will cause a trap.
  void loadWasmPinnedRegsFromInstance(
      mozilla::Maybe<wasm::BytecodeOffset> trapOffset = mozilla::Nothing());

  // Returns a pair: the offset of the undefined (trapping) instruction, and
  // the number of extra bytes of stack allocated prior to the trap
  // instruction proper.
  std::pair<CodeOffset, uint32_t> wasmReserveStackChecked(
      uint32_t amount, wasm::BytecodeOffset trapOffset);

  // Emit a bounds check against the wasm heap limit, jumping to 'ok' if 'cond'
  // holds; this can be the label either of the access or of the trap.  The
  // label should name a code position greater than the position of the bounds
  // check.
  //
  // If JitOptions.spectreMaskIndex is true, a no-op speculation barrier is
  // emitted in the code stream after the check to prevent an OOB access from
  // being executed speculatively.  (On current tier-1 platforms the barrier is
  // a conditional saturation of 'index' to 'boundsCheckLimit', using the same
  // condition as the check.)  If the condition is such that the bounds check
  // branches out of line to the trap, the barrier will actually be executed
  // when the bounds check passes.
  //
  // On 32-bit systems for both wasm and asm.js, and on 64-bit systems for
  // asm.js, heap lengths are limited to 2GB.  On 64-bit systems for wasm,
  // 32-bit heap lengths are limited to 4GB, and 64-bit heap lengths will be
  // limited to something much larger.

  void wasmBoundsCheck32(Condition cond, Register index,
                         Register boundsCheckLimit, Label* ok)
      DEFINED_ON(arm, arm64, mips32, mips64, x86_shared, loong64, riscv64,
                 wasm32);

  void wasmBoundsCheck32(Condition cond, Register index,
                         Address boundsCheckLimit, Label* ok)
      DEFINED_ON(arm, arm64, mips32, mips64, x86_shared, loong64, riscv64,
                 wasm32);

  void wasmBoundsCheck64(Condition cond, Register64 index,
                         Register64 boundsCheckLimit, Label* ok)
      DEFINED_ON(arm64, mips64, x64, x86, arm, loong64, riscv64, wasm32);

  void wasmBoundsCheck64(Condition cond, Register64 index,
                         Address boundsCheckLimit, Label* ok)
      DEFINED_ON(arm64, mips64, x64, x86, arm, loong64, riscv64, wasm32);

  // Each wasm load/store instruction appends its own wasm::Trap::OutOfBounds.
  void wasmLoad(const wasm::MemoryAccessDesc& access, Operand srcAddr,
                AnyRegister out) DEFINED_ON(x86, x64);
  void wasmLoadI64(const wasm::MemoryAccessDesc& access, Operand srcAddr,
                   Register64 out) DEFINED_ON(x86, x64);
  void wasmStore(const wasm::MemoryAccessDesc& access, AnyRegister value,
                 Operand dstAddr) DEFINED_ON(x86, x64);
  void wasmStoreI64(const wasm::MemoryAccessDesc& access, Register64 value,
                    Operand dstAddr) DEFINED_ON(x86);

  // For all the ARM/MIPS/LOONG64 wasmLoad and wasmStore functions below, `ptr`
  // MUST equal `ptrScratch`, and that register will be updated based on
  // conditions listed below (where it is only mentioned as `ptr`).

  // `ptr` will be updated if access.offset() != 0 or access.type() ==
  // Scalar::Int64.
  void wasmLoad(const wasm::MemoryAccessDesc& access, Register memoryBase,
                Register ptr, Register ptrScratch, AnyRegister output)
      DEFINED_ON(arm, loong64, riscv64, mips_shared);
  void wasmLoadI64(const wasm::MemoryAccessDesc& access, Register memoryBase,
                   Register ptr, Register ptrScratch, Register64 output)
      DEFINED_ON(arm, mips32, mips64, loong64, riscv64);
  void wasmStore(const wasm::MemoryAccessDesc& access, AnyRegister value,
                 Register memoryBase, Register ptr, Register ptrScratch)
      DEFINED_ON(arm, loong64, riscv64, mips_shared);
  void wasmStoreI64(const wasm::MemoryAccessDesc& access, Register64 value,
                    Register memoryBase, Register ptr, Register ptrScratch)
      DEFINED_ON(arm, mips32, mips64, loong64, riscv64);

  // These accept general memoryBase + ptr + offset (in `access`); the offset is
  // always smaller than the guard region.  They will insert an additional add
  // if the offset is nonzero, and of course that add may require a temporary
  // register for the offset if the offset is large, and instructions to set it
  // up.
  void wasmLoad(const wasm::MemoryAccessDesc& access, Register memoryBase,
                Register ptr, AnyRegister output) DEFINED_ON(arm64);
  void wasmLoadI64(const wasm::MemoryAccessDesc& access, Register memoryBase,
                   Register ptr, Register64 output) DEFINED_ON(arm64);
  void wasmStore(const wasm::MemoryAccessDesc& access, AnyRegister value,
                 Register memoryBase, Register ptr) DEFINED_ON(arm64);
  void wasmStoreI64(const wasm::MemoryAccessDesc& access, Register64 value,
                    Register memoryBase, Register ptr) DEFINED_ON(arm64);

  // `ptr` will always be updated.
  void wasmUnalignedLoad(const wasm::MemoryAccessDesc& access,
                         Register memoryBase, Register ptr, Register ptrScratch,
                         Register output, Register tmp)
      DEFINED_ON(mips32, mips64);

  // MIPS: `ptr` will always be updated.
  void wasmUnalignedLoadFP(const wasm::MemoryAccessDesc& access,
                           Register memoryBase, Register ptr,
                           Register ptrScratch, FloatRegister output,
                           Register tmp1) DEFINED_ON(mips32, mips64);

  // `ptr` will always be updated.
  void wasmUnalignedLoadI64(const wasm::MemoryAccessDesc& access,
                            Register memoryBase, Register ptr,
                            Register ptrScratch, Register64 output,
                            Register tmp) DEFINED_ON(mips32, mips64);

  // MIPS: `ptr` will always be updated.
  void wasmUnalignedStore(const wasm::MemoryAccessDesc& access, Register value,
                          Register memoryBase, Register ptr,
                          Register ptrScratch, Register tmp)
      DEFINED_ON(mips32, mips64);

  // `ptr` will always be updated.
  void wasmUnalignedStoreFP(const wasm::MemoryAccessDesc& access,
                            FloatRegister floatValue, Register memoryBase,
                            Register ptr, Register ptrScratch, Register tmp)
      DEFINED_ON(mips32, mips64);

  // `ptr` will always be updated.
  void wasmUnalignedStoreI64(const wasm::MemoryAccessDesc& access,
                             Register64 value, Register memoryBase,
                             Register ptr, Register ptrScratch, Register tmp)
      DEFINED_ON(mips32, mips64);

  // wasm specific methods, used in both the wasm baseline compiler and ion.

  // The truncate-to-int32 methods do not bind the rejoin label; clients must
  // do so if oolWasmTruncateCheckF64ToI32() can jump to it.
  void wasmTruncateDoubleToUInt32(FloatRegister input, Register output,
                                  bool isSaturating, Label* oolEntry) PER_ARCH;
  void wasmTruncateDoubleToInt32(FloatRegister input, Register output,
                                 bool isSaturating,
                                 Label* oolEntry) PER_SHARED_ARCH;
  void oolWasmTruncateCheckF64ToI32(FloatRegister input, Register output,
                                    TruncFlags flags, wasm::BytecodeOffset off,
                                    Label* rejoin)
      DEFINED_ON(arm, arm64, x86_shared, mips_shared, loong64, riscv64, wasm32);

  void wasmTruncateFloat32ToUInt32(FloatRegister input, Register output,
                                   bool isSaturating, Label* oolEntry) PER_ARCH;
  void wasmTruncateFloat32ToInt32(FloatRegister input, Register output,
                                  bool isSaturating,
                                  Label* oolEntry) PER_SHARED_ARCH;
  void oolWasmTruncateCheckF32ToI32(FloatRegister input, Register output,
                                    TruncFlags flags, wasm::BytecodeOffset off,
                                    Label* rejoin)
      DEFINED_ON(arm, arm64, x86_shared, mips_shared, loong64, riscv64, wasm32);

  // The truncate-to-int64 methods will always bind the `oolRejoin` label
  // after the last emitted instruction.
  void wasmTruncateDoubleToInt64(FloatRegister input, Register64 output,
                                 bool isSaturating, Label* oolEntry,
                                 Label* oolRejoin, FloatRegister tempDouble)
      DEFINED_ON(arm64, x86, x64, mips64, loong64, riscv64, wasm32);
  void wasmTruncateDoubleToUInt64(FloatRegister input, Register64 output,
                                  bool isSaturating, Label* oolEntry,
                                  Label* oolRejoin, FloatRegister tempDouble)
      DEFINED_ON(arm64, x86, x64, mips64, loong64, riscv64, wasm32);
  void oolWasmTruncateCheckF64ToI64(FloatRegister input, Register64 output,
                                    TruncFlags flags, wasm::BytecodeOffset off,
                                    Label* rejoin)
      DEFINED_ON(arm, arm64, x86_shared, mips_shared, loong64, riscv64, wasm32);

  void wasmTruncateFloat32ToInt64(FloatRegister input, Register64 output,
                                  bool isSaturating, Label* oolEntry,
                                  Label* oolRejoin, FloatRegister tempDouble)
      DEFINED_ON(arm64, x86, x64, mips64, loong64, riscv64, wasm32);
  void wasmTruncateFloat32ToUInt64(FloatRegister input, Register64 output,
                                   bool isSaturating, Label* oolEntry,
                                   Label* oolRejoin, FloatRegister tempDouble)
      DEFINED_ON(arm64, x86, x64, mips64, loong64, riscv64, wasm32);
  void oolWasmTruncateCheckF32ToI64(FloatRegister input, Register64 output,
                                    TruncFlags flags, wasm::BytecodeOffset off,
                                    Label* rejoin)
      DEFINED_ON(arm, arm64, x86_shared, mips_shared, loong64, riscv64, wasm32);

  // This function takes care of loading the callee's instance and pinned regs
  // but it is the caller's responsibility to save/restore instance or pinned
  // regs.
  CodeOffset wasmCallImport(const wasm::CallSiteDesc& desc,
                            const wasm::CalleeDesc& callee);

#ifdef ENABLE_WASM_TAIL_CALLS
  CodeOffset wasmReturnCallImport(const wasm::CallSiteDesc& desc,
                                  const wasm::CalleeDesc& callee,
                                  const ReturnCallAdjustmentInfo& retCallInfo);

  CodeOffset wasmReturnCall(const wasm::CallSiteDesc& desc,
                            uint32_t funcDefIndex,
                            const ReturnCallAdjustmentInfo& retCallInfo);

  void wasmCollapseFrameSlow(const ReturnCallAdjustmentInfo& retCallInfo,
                             wasm::CallSiteDesc desc);

  void wasmCollapseFrameFast(const ReturnCallAdjustmentInfo& retCallInfo);

  void wasmCheckSlowCallsite(Register ra, Label* notSlow, Register temp1,
                             Register temp2)
      DEFINED_ON(x86, x64, arm, arm64, loong64, mips64, riscv64);

  void wasmMarkSlowCall()
      DEFINED_ON(x86, x64, arm, arm64, loong64, mips64, riscv64);
#endif

  // WasmTableCallIndexReg must contain the index of the indirect call.  This is
  // for wasm calls only.
  //
  // Indirect calls use a dual-path mechanism where a run-time test determines
  // whether a context switch is needed (slow path) or not (fast path).  This
  // gives rise to two call instructions, both of which need safe points.  As
  // per normal, the call offsets are the code offsets at the end of the call
  // instructions (the return points).
  //
  // `boundsCheckFailedLabel` is non-null iff a bounds check is required.
  // `nullCheckFailedLabel` is non-null only on platforms that can't fold the
  // null check into the rest of the call instructions.
  void wasmCallIndirect(const wasm::CallSiteDesc& desc,
                        const wasm::CalleeDesc& callee,
                        Label* boundsCheckFailedLabel,
                        Label* nullCheckFailedLabel,
                        mozilla::Maybe<uint32_t> tableSize,
                        CodeOffset* fastCallOffset, CodeOffset* slowCallOffset);

#ifdef ENABLE_WASM_TAIL_CALLS
  // WasmTableCallIndexReg must contain the index of the indirect call.  This is
  // for wasm calls only.
  //
  // `boundsCheckFailedLabel` is non-null iff a bounds check is required.
  // `nullCheckFailedLabel` is non-null only on platforms that can't fold the
  // null check into the rest of the call instructions.
  void wasmReturnCallIndirect(const wasm::CallSiteDesc& desc,
                              const wasm::CalleeDesc& callee,
                              Label* boundsCheckFailedLabel,
                              Label* nullCheckFailedLabel,
                              mozilla::Maybe<uint32_t> tableSize,
                              const ReturnCallAdjustmentInfo& retCallInfo);
#endif  // ENABLE_WASM_TAIL_CALLS

  // This function takes care of loading the callee's instance and address from
  // pinned reg.
  void wasmCallRef(const wasm::CallSiteDesc& desc,
                   const wasm::CalleeDesc& callee, CodeOffset* fastCallOffset,
                   CodeOffset* slowCallOffset);

#ifdef ENABLE_WASM_TAIL_CALLS
  void wasmReturnCallRef(const wasm::CallSiteDesc& desc,
                         const wasm::CalleeDesc& callee,
                         const ReturnCallAdjustmentInfo& retCallInfo);
#endif  // ENABLE_WASM_TAIL_CALLS

  // WasmTableCallIndexReg must contain the index of the indirect call.
  // This is for asm.js calls only.
  CodeOffset asmCallIndirect(const wasm::CallSiteDesc& desc,
                             const wasm::CalleeDesc& callee);

  // This function takes care of loading the pointer to the current instance
  // as the implicit first argument. It preserves instance and pinned registers.
  // (instance & pinned regs are non-volatile registers in the system ABI).
  CodeOffset wasmCallBuiltinInstanceMethod(const wasm::CallSiteDesc& desc,
                                           const ABIArg& instanceArg,
                                           wasm::SymbolicAddress builtin,
                                           wasm::FailureMode failureMode);

  // Performs a bounds check for ranged wasm operations like memory.fill or
  // array.fill. This handles the bizarre edge case in the wasm spec where a
  // write to index N is valid as long as the length is zero - despite the index
  // itself being out of bounds.
  //
  // `length` and `limit` will be unchanged.
  void wasmBoundsCheckRange32(Register index, Register length, Register limit,
                              Register tmp,
                              wasm::BytecodeOffset bytecodeOffset);

  // Returns information about which registers are necessary for a
  // branchWasmRefIsSubtype call.
  static BranchWasmRefIsSubtypeRegisters regsForBranchWasmRefIsSubtype(
      wasm::RefType type);

  // Perform a subtype check that `ref` is a subtype of `type`, branching to
  // `label` depending on `onSuccess`.
  //
  // Will select one of the other branchWasmRefIsSubtype* functions depending on
  // destType. See each function for the register allocation requirements, as
  // well as which registers will be preserved.
  void branchWasmRefIsSubtype(Register ref, wasm::RefType sourceType,
                              wasm::RefType destType, Label* label,
                              bool onSuccess, Register superSTV,
                              Register scratch1, Register scratch2);

  // Perform a subtype check that `ref` is a subtype of `type`, branching to
  // `label` depending on `onSuccess`. `type` must be in the `any` hierarchy.
  //
  // `superSTV` is required iff the destination type is a concrete
  // type. `scratch1` is required iff the destination type is eq or lower and
  // not none. `scratch2` is required iff the destination type is a concrete
  // type and its `subTypingDepth` is >= wasm::MinSuperTypeVectorLength. See
  // regsForBranchWasmRefIsSubtype.
  //
  // `ref` and `superSTV` are preserved. Scratch registers are
  // clobbered.
  void branchWasmRefIsSubtypeAny(Register ref, wasm::RefType sourceType,
                                 wasm::RefType destType, Label* label,
                                 bool onSuccess, Register superSTV,
                                 Register scratch1, Register scratch2);

  // Perform a subtype check that `ref` is a subtype of `type`, branching to
  // `label` depending on `onSuccess`. `type` must be in the `func` hierarchy.
  //
  // `superSTV` and `scratch1` are required iff the destination type
  // is a concrete type (not func and not nofunc). `scratch2` is required iff
  // the destination type is a concrete type and its `subTypingDepth` is >=
  // wasm::MinSuperTypeVectorLength. See regsForBranchWasmRefIsSubtype.
  //
  // `ref` and `superSTV` are preserved. Scratch registers are
  // clobbered.
  void branchWasmRefIsSubtypeFunc(Register ref, wasm::RefType sourceType,
                                  wasm::RefType destType, Label* label,
                                  bool onSuccess, Register superSTV,
                                  Register scratch1, Register scratch2);

  // Perform a subtype check that `ref` is a subtype of `destType`, branching to
  // `label` depending on `onSuccess`. `type` must be in the `extern` hierarchy.
  void branchWasmRefIsSubtypeExtern(Register ref, wasm::RefType sourceType,
                                    wasm::RefType destType, Label* label,
                                    bool onSuccess);

  // Perform a subtype check that `ref` is a subtype of `destType`, branching to
  // `label` depending on `onSuccess`. `type` must be in the `exn` hierarchy.
  void branchWasmRefIsSubtypeExn(Register ref, wasm::RefType sourceType,
                                 wasm::RefType destType, Label* label,
                                 bool onSuccess);

  // Perform a subtype check that `subSTV` is a subtype of `superSTV`, branching
  // to `label` depending on `onSuccess`. This method is a specialization of the
  // general `wasm::TypeDef::isSubTypeOf` method for the case where the
  // `superSTV` is statically known, which is the case for all wasm
  // instructions.
  //
  // `scratch` is required iff the `superDepth` is >=
  // wasm::MinSuperTypeVectorLength. `subSTV` is clobbered by this method.
  // `superSTV` is preserved.
  void branchWasmSTVIsSubtype(Register subSTV, Register superSTV,
                              Register scratch, uint32_t superDepth,
                              Label* label, bool onSuccess);

  // Same as branchWasmSTVIsSubtype, but looks up a dynamic position in the
  // super type vector.
  //
  // `scratch` is always required. `subSTV` and `superDepth` are clobbered.
  // `superSTV` is preserved.
  void branchWasmSTVIsSubtypeDynamicDepth(Register subSTV, Register superSTV,
                                          Register superDepth, Register scratch,
                                          Label* label, bool onSuccess);

  // Branch if the wasm anyref `src` is or is not the null value.
  void branchWasmAnyRefIsNull(bool isNull, Register src, Label* label);
  // Branch if the wasm anyref `src` is or is not an I31.
  void branchWasmAnyRefIsI31(bool isI31, Register src, Label* label);
  // Branch if the wasm anyref `src` is or is not a JSObject*.
  void branchWasmAnyRefIsObjectOrNull(bool isObject, Register src,
                                      Label* label);
  // Branch if the wasm anyref `src` is or is not a GC thing.
  void branchWasmAnyRefIsGCThing(bool isGCThing, Register src, Label* label);
  // Branch if the wasm anyref `src` is or is not pointing to a nursery cell.
  void branchWasmAnyRefIsNurseryCell(bool isNurseryCell, Register src,
                                     Register scratch, Label* label);

  // Create a wasm i31ref by truncating the 32-bit integer.
  void truncate32ToWasmI31Ref(Register src, Register dest);
  // Convert a wasm i31ref to a signed 32-bit integer.
  void convertWasmI31RefTo32Signed(Register src, Register dest);
  // Convert a wasm i31ref to an unsigned 32-bit integer.
  void convertWasmI31RefTo32Unsigned(Register src, Register dest);

  // Branch if the JS value `src` would need to be boxed out of line to be
  // converted to a wasm anyref.
  void branchValueConvertsToWasmAnyRefInline(ValueOperand src,
                                             Register scratchInt,
                                             FloatRegister scratchFloat,
                                             Label* label);
  // Convert a JS value to a wasm anyref. If the value requires boxing, this
  // will branch to `oolConvert`.
  void convertValueToWasmAnyRef(ValueOperand src, Register dest,
                                FloatRegister scratchFloat, Label* oolConvert);
  // Convert a JS object to a wasm anyref. This cannot fail.
  void convertObjectToWasmAnyRef(Register src, Register dest);
  // Convert a JS string to a wasm anyref. This cannot fail.
  void convertStringToWasmAnyRef(Register src, Register dest);

  // Convert a wasm anyref to a JS value. This cannot fail.
  //
  // Due to spectre mitigations, these methods may clobber src.
  void convertWasmAnyRefToValue(Register instance, Register src,
                                ValueOperand dst, Register scratch);
  void convertWasmAnyRefToValue(Register instance, Register src,
                                const Address& dst, Register scratch);

  // Branch if the object `src` is or is not a WasmGcObject.
  void branchObjectIsWasmGcObject(bool isGcObject, Register src,
                                  Register scratch, Label* label);

  // `typeDefData` will be preserved. `instance` and `result` may be the same
  // register, in which case `instance` will be clobbered.
  void wasmNewStructObject(Register instance, Register result,
                           Register typeDefData, Register temp1, Register temp2,
                           Label* fail, gc::AllocKind allocKind,
                           bool zeroFields);
  // Allocates a wasm array with a dynamic number of elements.
  //
  // `numElements` and `typeDefData` will be preserved. `instance` and `result`
  // may be the same register, in which case `instance` will be clobbered.
  void wasmNewArrayObject(Register instance, Register result,
                          Register numElements, Register typeDefData,
                          Register temp, Label* fail, uint32_t elemSize,
                          bool zeroFields);
  // Allocates a wasm array with a fixed number of elements.
  //
  // `typeDefData` will be preserved. `instance` and `result` may be the same
  // register, in which case `instance` will be clobbered.
  void wasmNewArrayObjectFixed(Register instance, Register result,
                               Register typeDefData, Register temp1,
                               Register temp2, Label* fail,
                               uint32_t numElements, uint32_t storageBytes,
                               bool zeroFields);

  // This function handles nursery allocations for wasm. For JS, see
  // MacroAssembler::bumpPointerAllocate.
  //
  // `typeDefData` will be preserved. `instance` and `result` may be the same
  // register, in which case `instance` will be clobbered.
  //
  // See also the dynamically-sized version,
  // MacroAssembler::wasmBumpPointerAllocateDynamic.
  void wasmBumpPointerAllocate(Register instance, Register result,
                               Register typeDefData, Register temp1,
                               Register temp2, Label* fail, uint32_t size);
  // This function handles nursery allocations for wasm of dynamic size. For
  // fixed-size allocations, see MacroAssembler::wasmBumpPointerAllocate.
  //
  // `typeDefData` and `size` will be preserved. `instance` and `result` may be
  // the same register, in which case `instance` will be clobbered.
  void wasmBumpPointerAllocateDynamic(Register instance, Register result,
                                      Register typeDefData, Register size,
                                      Register temp1, Label* fail);

  // Compute ptr += (indexTemp32 << shift) where shift can be any value < 32.
  // May destroy indexTemp32.  The value of indexTemp32 must be positive, and it
  // is implementation-defined what happens if bits are lost or the value
  // becomes negative through the shift.  On 64-bit systems, the high 32 bits of
  // indexTemp32 must be zero, not garbage.
  void shiftIndex32AndAdd(Register indexTemp32, int shift,
                          Register pointer) PER_SHARED_ARCH;

  // The System ABI frequently states that the high bits of a 64-bit register
  // that holds a 32-bit return value are unpredictable, and C++ compilers will
  // indeed generate code that leaves garbage in the upper bits.
  //
  // Adjust the contents of the 64-bit register `r` to conform to our internal
  // convention, which requires predictable high bits.  In practice, this means
  // that the 32-bit value will be zero-extended or sign-extended to 64 bits as
  // appropriate for the platform.
  void widenInt32(Register r) DEFINED_ON(arm64, x64, mips64, loong64, riscv64);

  // As enterFakeExitFrame(), but using register conventions appropriate for
  // wasm stubs.
  void enterFakeExitFrameForWasm(Register cxreg, Register scratch,
                                 ExitFrameType type) PER_SHARED_ARCH;

 public:
  // ========================================================================
  // Barrier functions.

  void emitPreBarrierFastPath(JSRuntime* rt, MIRType type, Register temp1,
                              Register temp2, Register temp3, Label* noBarrier);

 public:
  // ========================================================================
  // Clamping functions.

  inline void clampIntToUint8(Register reg) PER_SHARED_ARCH;

 public:
  // ========================================================================
  // Primitive atomic operations.
  //
  // If the access is from JS and the eventual destination of the result is a
  // js::Value, it's probably best to use the JS-specific versions of these,
  // see further below.
  //
  // Temp registers must be defined unless otherwise noted in the per-function
  // constraints.

  // 8-bit, 16-bit, and 32-bit wide operations.
  //
  // The 8-bit and 16-bit operations zero-extend or sign-extend the result to
  // 32 bits, according to `type`. On 64-bit systems, the upper 32 bits of the
  // result will be zero on some platforms (eg, on x64) and will be the sign
  // extension of the lower bits on other platforms (eg, MIPS).

  // CompareExchange with memory.  Return the value that was in memory,
  // whether we wrote or not.
  //
  // x86-shared: `output` must be eax.
  // MIPS: `valueTemp`, `offsetTemp` and `maskTemp` must be defined for 8-bit
  // and 16-bit wide operations.

  void compareExchange(Scalar::Type type, Synchronization sync,
                       const Address& mem, Register expected,
                       Register replacement, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void compareExchange(Scalar::Type type, Synchronization sync,
                       const BaseIndex& mem, Register expected,
                       Register replacement, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void compareExchange(Scalar::Type type, Synchronization sync,
                       const Address& mem, Register expected,
                       Register replacement, Register valueTemp,
                       Register offsetTemp, Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void compareExchange(Scalar::Type type, Synchronization sync,
                       const BaseIndex& mem, Register expected,
                       Register replacement, Register valueTemp,
                       Register offsetTemp, Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  // x86: `expected` and `output` must be edx:eax; `replacement` is ecx:ebx.
  // x64: `output` must be rax.
  // ARM: Registers must be distinct; `replacement` and `output` must be
  // (even,odd) pairs.

  void compareExchange64(Synchronization sync, const Address& mem,
                         Register64 expected, Register64 replacement,
                         Register64 output)
      DEFINED_ON(arm, arm64, x64, x86, mips64, loong64, riscv64);

  void compareExchange64(Synchronization sync, const BaseIndex& mem,
                         Register64 expected, Register64 replacement,
                         Register64 output)
      DEFINED_ON(arm, arm64, x64, x86, mips64, loong64, riscv64);

  // Exchange with memory.  Return the value initially in memory.
  // MIPS: `valueTemp`, `offsetTemp` and `maskTemp` must be defined for 8-bit
  // and 16-bit wide operations.

  void atomicExchange(Scalar::Type type, Synchronization sync,
                      const Address& mem, Register value, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void atomicExchange(Scalar::Type type, Synchronization sync,
                      const BaseIndex& mem, Register value, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void atomicExchange(Scalar::Type type, Synchronization sync,
                      const Address& mem, Register value, Register valueTemp,
                      Register offsetTemp, Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicExchange(Scalar::Type type, Synchronization sync,
                      const BaseIndex& mem, Register value, Register valueTemp,
                      Register offsetTemp, Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  // x86: `value` must be ecx:ebx; `output` must be edx:eax.
  // ARM: `value` and `output` must be distinct and (even,odd) pairs.
  // ARM64: `value` and `output` must be distinct.

  void atomicExchange64(Synchronization sync, const Address& mem,
                        Register64 value, Register64 output)
      DEFINED_ON(arm, arm64, x64, x86, mips64, loong64, riscv64);

  void atomicExchange64(Synchronization sync, const BaseIndex& mem,
                        Register64 value, Register64 output)
      DEFINED_ON(arm, arm64, x64, x86, mips64, loong64, riscv64);

  // Read-modify-write with memory.  Return the value in memory before the
  // operation.
  //
  // x86-shared:
  //   For 8-bit operations, `value` and `output` must have a byte subregister.
  //   For Add and Sub, `temp` must be invalid.
  //   For And, Or, and Xor, `output` must be eax and `temp` must have a byte
  //   subregister.
  //
  // ARM: Registers `value` and `output` must differ.
  // MIPS: `valueTemp`, `offsetTemp` and `maskTemp` must be defined for 8-bit
  // and 16-bit wide operations; `value` and `output` must differ.

  void atomicFetchOp(Scalar::Type type, Synchronization sync, AtomicOp op,
                     Register value, const Address& mem, Register temp,
                     Register output) DEFINED_ON(arm, arm64, x86_shared);

  void atomicFetchOp(Scalar::Type type, Synchronization sync, AtomicOp op,
                     Imm32 value, const Address& mem, Register temp,
                     Register output) DEFINED_ON(x86_shared);

  void atomicFetchOp(Scalar::Type type, Synchronization sync, AtomicOp op,
                     Register value, const BaseIndex& mem, Register temp,
                     Register output) DEFINED_ON(arm, arm64, x86_shared);

  void atomicFetchOp(Scalar::Type type, Synchronization sync, AtomicOp op,
                     Imm32 value, const BaseIndex& mem, Register temp,
                     Register output) DEFINED_ON(x86_shared);

  void atomicFetchOp(Scalar::Type type, Synchronization sync, AtomicOp op,
                     Register value, const Address& mem, Register valueTemp,
                     Register offsetTemp, Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicFetchOp(Scalar::Type type, Synchronization sync, AtomicOp op,
                     Register value, const BaseIndex& mem, Register valueTemp,
                     Register offsetTemp, Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  // x86:
  //   `temp` must be ecx:ebx; `output` must be edx:eax.
  // x64:
  //   For Add and Sub, `temp` is ignored.
  //   For And, Or, and Xor, `output` must be rax.
  // ARM:
  //   `temp` and `output` must be (even,odd) pairs and distinct from `value`.
  // ARM64:
  //   Registers `value`, `temp`, and `output` must all differ.

  void atomicFetchOp64(Synchronization sync, AtomicOp op, Register64 value,
                       const Address& mem, Register64 temp, Register64 output)
      DEFINED_ON(arm, arm64, x64, mips64, loong64, riscv64);

  void atomicFetchOp64(Synchronization sync, AtomicOp op, const Address& value,
                       const Address& mem, Register64 temp, Register64 output)
      DEFINED_ON(x86);

  void atomicFetchOp64(Synchronization sync, AtomicOp op, Register64 value,
                       const BaseIndex& mem, Register64 temp, Register64 output)
      DEFINED_ON(arm, arm64, x64, mips64, loong64, riscv64);

  void atomicFetchOp64(Synchronization sync, AtomicOp op, const Address& value,
                       const BaseIndex& mem, Register64 temp, Register64 output)
      DEFINED_ON(x86);

  // x64:
  //   `value` can be any register.
  // ARM:
  //   `temp` must be an (even,odd) pair and distinct from `value`.
  // ARM64:
  //   Registers `value` and `temp` must differ.

  void atomicEffectOp64(Synchronization sync, AtomicOp op, Register64 value,
                        const Address& mem) DEFINED_ON(x64);

  void atomicEffectOp64(Synchronization sync, AtomicOp op, Register64 value,
                        const Address& mem, Register64 temp)
      DEFINED_ON(arm, arm64, mips64, loong64, riscv64);

  void atomicEffectOp64(Synchronization sync, AtomicOp op, Register64 value,
                        const BaseIndex& mem) DEFINED_ON(x64);

  void atomicEffectOp64(Synchronization sync, AtomicOp op, Register64 value,
                        const BaseIndex& mem, Register64 temp)
      DEFINED_ON(arm, arm64, mips64, loong64, riscv64);

  // 64-bit atomic load. On 64-bit systems, use regular load with
  // Synchronization::Load, not this method.
  //
  // x86: `temp` must be ecx:ebx; `output` must be edx:eax.
  // ARM: `output` must be (even,odd) pair.

  void atomicLoad64(Synchronization sync, const Address& mem, Register64 temp,
                    Register64 output) DEFINED_ON(x86);

  void atomicLoad64(Synchronization sync, const BaseIndex& mem, Register64 temp,
                    Register64 output) DEFINED_ON(x86);

  void atomicLoad64(Synchronization sync, const Address& mem, Register64 output)
      DEFINED_ON(arm);

  void atomicLoad64(Synchronization sync, const BaseIndex& mem,
                    Register64 output) DEFINED_ON(arm);

  // 64-bit atomic store. On 64-bit systems, use regular store with
  // Synchronization::Store, not this method.
  //
  // x86: `value` must be ecx:ebx; `temp` must be edx:eax.
  // ARM: `value` and `temp` must be (even,odd) pairs.

  void atomicStore64(Synchronization sync, const Address& mem, Register64 value,
                     Register64 temp) DEFINED_ON(x86, arm);

  void atomicStore64(Synchronization sync, const BaseIndex& mem,
                     Register64 value, Register64 temp) DEFINED_ON(x86, arm);

  // ========================================================================
  // Wasm atomic operations.
  //
  // Constraints, when omitted, are exactly as for the primitive operations
  // above.

  void wasmCompareExchange(const wasm::MemoryAccessDesc& access,
                           const Address& mem, Register expected,
                           Register replacement, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void wasmCompareExchange(const wasm::MemoryAccessDesc& access,
                           const BaseIndex& mem, Register expected,
                           Register replacement, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void wasmCompareExchange(const wasm::MemoryAccessDesc& access,
                           const Address& mem, Register expected,
                           Register replacement, Register valueTemp,
                           Register offsetTemp, Register maskTemp,
                           Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void wasmCompareExchange(const wasm::MemoryAccessDesc& access,
                           const BaseIndex& mem, Register expected,
                           Register replacement, Register valueTemp,
                           Register offsetTemp, Register maskTemp,
                           Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void wasmAtomicExchange(const wasm::MemoryAccessDesc& access,
                          const Address& mem, Register value, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void wasmAtomicExchange(const wasm::MemoryAccessDesc& access,
                          const BaseIndex& mem, Register value, Register output)
      DEFINED_ON(arm, arm64, x86_shared);

  void wasmAtomicExchange(const wasm::MemoryAccessDesc& access,
                          const Address& mem, Register value,
                          Register valueTemp, Register offsetTemp,
                          Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void wasmAtomicExchange(const wasm::MemoryAccessDesc& access,
                          const BaseIndex& mem, Register value,
                          Register valueTemp, Register offsetTemp,
                          Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void wasmAtomicFetchOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                         Register value, const Address& mem, Register temp,
                         Register output) DEFINED_ON(arm, arm64, x86_shared);

  void wasmAtomicFetchOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                         Imm32 value, const Address& mem, Register temp,
                         Register output) DEFINED_ON(x86_shared);

  void wasmAtomicFetchOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                         Register value, const BaseIndex& mem, Register temp,
                         Register output) DEFINED_ON(arm, arm64, x86_shared);

  void wasmAtomicFetchOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                         Imm32 value, const BaseIndex& mem, Register temp,
                         Register output) DEFINED_ON(x86_shared);

  void wasmAtomicFetchOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                         Register value, const Address& mem, Register valueTemp,
                         Register offsetTemp, Register maskTemp,
                         Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void wasmAtomicFetchOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                         Register value, const BaseIndex& mem,
                         Register valueTemp, Register offsetTemp,
                         Register maskTemp, Register output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  // Read-modify-write with memory.  Return no value.
  //
  // MIPS: `valueTemp`, `offsetTemp` and `maskTemp` must be defined for 8-bit
  // and 16-bit wide operations.

  void wasmAtomicEffectOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                          Register value, const Address& mem, Register temp)
      DEFINED_ON(arm, arm64, x86_shared);

  void wasmAtomicEffectOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                          Imm32 value, const Address& mem, Register temp)
      DEFINED_ON(x86_shared);

  void wasmAtomicEffectOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                          Register value, const BaseIndex& mem, Register temp)
      DEFINED_ON(arm, arm64, x86_shared);

  void wasmAtomicEffectOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                          Imm32 value, const BaseIndex& mem, Register temp)
      DEFINED_ON(x86_shared);

  void wasmAtomicEffectOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                          Register value, const Address& mem,
                          Register valueTemp, Register offsetTemp,
                          Register maskTemp)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void wasmAtomicEffectOp(const wasm::MemoryAccessDesc& access, AtomicOp op,
                          Register value, const BaseIndex& mem,
                          Register valueTemp, Register offsetTemp,
                          Register maskTemp)
      DEFINED_ON(mips_shared, loong64, riscv64);

  // 64-bit wide operations.

  // 64-bit atomic load.  On 64-bit systems, use regular wasm load with
  // Synchronization::Load, not this method.
  //
  // x86: `temp` must be ecx:ebx; `output` must be edx:eax.
  // ARM: `temp` should be invalid; `output` must be (even,odd) pair.
  // MIPS32: `temp` should be invalid.

  void wasmAtomicLoad64(const wasm::MemoryAccessDesc& access,
                        const Address& mem, Register64 temp, Register64 output)
      DEFINED_ON(arm, mips32, x86, wasm32);

  void wasmAtomicLoad64(const wasm::MemoryAccessDesc& access,
                        const BaseIndex& mem, Register64 temp,
                        Register64 output) DEFINED_ON(arm, mips32, x86, wasm32);

  // x86: `expected` must be the same as `output`, and must be edx:eax.
  // x86: `replacement` must be ecx:ebx.
  // x64: `output` must be rax.
  // ARM: Registers must be distinct; `replacement` and `output` must be
  // (even,odd) pairs.
  // ARM64: The base register in `mem` must not overlap `output`.
  // MIPS: Registers must be distinct.

  void wasmCompareExchange64(const wasm::MemoryAccessDesc& access,
                             const Address& mem, Register64 expected,
                             Register64 replacement,
                             Register64 output) PER_ARCH;

  void wasmCompareExchange64(const wasm::MemoryAccessDesc& access,
                             const BaseIndex& mem, Register64 expected,
                             Register64 replacement,
                             Register64 output) PER_ARCH;

  // x86: `value` must be ecx:ebx; `output` must be edx:eax.
  // ARM: Registers must be distinct; `value` and `output` must be (even,odd)
  // pairs.
  // MIPS: Registers must be distinct.

  void wasmAtomicExchange64(const wasm::MemoryAccessDesc& access,
                            const Address& mem, Register64 value,
                            Register64 output) PER_ARCH;

  void wasmAtomicExchange64(const wasm::MemoryAccessDesc& access,
                            const BaseIndex& mem, Register64 value,
                            Register64 output) PER_ARCH;

  // x86: `output` must be edx:eax, `temp` must be ecx:ebx.
  // x64: For And, Or, and Xor `output` must be rax.
  // ARM: Registers must be distinct; `temp` and `output` must be (even,odd)
  // pairs.
  // MIPS: Registers must be distinct.
  // MIPS32: `temp` should be invalid.

  void wasmAtomicFetchOp64(const wasm::MemoryAccessDesc& access, AtomicOp op,
                           Register64 value, const Address& mem,
                           Register64 temp, Register64 output)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, x64);

  void wasmAtomicFetchOp64(const wasm::MemoryAccessDesc& access, AtomicOp op,
                           Register64 value, const BaseIndex& mem,
                           Register64 temp, Register64 output)
      DEFINED_ON(arm, arm64, mips32, mips64, loong64, riscv64, x64);

  void wasmAtomicFetchOp64(const wasm::MemoryAccessDesc& access, AtomicOp op,
                           const Address& value, const Address& mem,
                           Register64 temp, Register64 output) DEFINED_ON(x86);

  void wasmAtomicFetchOp64(const wasm::MemoryAccessDesc& access, AtomicOp op,
                           const Address& value, const BaseIndex& mem,
                           Register64 temp, Register64 output) DEFINED_ON(x86);

  // Here `value` can be any register.

  void wasmAtomicEffectOp64(const wasm::MemoryAccessDesc& access, AtomicOp op,
                            Register64 value, const BaseIndex& mem)
      DEFINED_ON(x64);

  void wasmAtomicEffectOp64(const wasm::MemoryAccessDesc& access, AtomicOp op,
                            Register64 value, const BaseIndex& mem,
                            Register64 temp) DEFINED_ON(arm64);

  // ========================================================================
  // JS atomic operations.
  //
  // Here the arrayType must be a type that is valid for JS.  As of 2017 that
  // is an 8-bit, 16-bit, or 32-bit integer type.
  //
  // If arrayType is Scalar::Uint32 then:
  //
  //   - `output` must be a float register
  //   - if the operation takes one temp register then `temp` must be defined
  //   - if the operation takes two temp registers then `temp2` must be defined.
  //
  // Otherwise `output` must be a GPR and `temp`/`temp2` should be InvalidReg.
  // (`temp1` must always be valid.)
  //
  // For additional register constraints, see the primitive 32-bit operations
  // and/or wasm operations above.

  void compareExchangeJS(Scalar::Type arrayType, Synchronization sync,
                         const Address& mem, Register expected,
                         Register replacement, Register temp,
                         AnyRegister output) DEFINED_ON(arm, arm64, x86_shared);

  void compareExchangeJS(Scalar::Type arrayType, Synchronization sync,
                         const BaseIndex& mem, Register expected,
                         Register replacement, Register temp,
                         AnyRegister output) DEFINED_ON(arm, arm64, x86_shared);

  void compareExchangeJS(Scalar::Type arrayType, Synchronization sync,
                         const Address& mem, Register expected,
                         Register replacement, Register valueTemp,
                         Register offsetTemp, Register maskTemp, Register temp,
                         AnyRegister output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void compareExchangeJS(Scalar::Type arrayType, Synchronization sync,
                         const BaseIndex& mem, Register expected,
                         Register replacement, Register valueTemp,
                         Register offsetTemp, Register maskTemp, Register temp,
                         AnyRegister output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicExchangeJS(Scalar::Type arrayType, Synchronization sync,
                        const Address& mem, Register value, Register temp,
                        AnyRegister output) DEFINED_ON(arm, arm64, x86_shared);

  void atomicExchangeJS(Scalar::Type arrayType, Synchronization sync,
                        const BaseIndex& mem, Register value, Register temp,
                        AnyRegister output) DEFINED_ON(arm, arm64, x86_shared);

  void atomicExchangeJS(Scalar::Type arrayType, Synchronization sync,
                        const Address& mem, Register value, Register valueTemp,
                        Register offsetTemp, Register maskTemp, Register temp,
                        AnyRegister output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicExchangeJS(Scalar::Type arrayType, Synchronization sync,
                        const BaseIndex& mem, Register value,
                        Register valueTemp, Register offsetTemp,
                        Register maskTemp, Register temp, AnyRegister output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicFetchOpJS(Scalar::Type arrayType, Synchronization sync,
                       AtomicOp op, Register value, const Address& mem,
                       Register temp1, Register temp2, AnyRegister output)
      DEFINED_ON(arm, arm64, x86_shared);

  void atomicFetchOpJS(Scalar::Type arrayType, Synchronization sync,
                       AtomicOp op, Register value, const BaseIndex& mem,
                       Register temp1, Register temp2, AnyRegister output)
      DEFINED_ON(arm, arm64, x86_shared);

  void atomicFetchOpJS(Scalar::Type arrayType, Synchronization sync,
                       AtomicOp op, Imm32 value, const Address& mem,
                       Register temp1, Register temp2, AnyRegister output)
      DEFINED_ON(x86_shared);

  void atomicFetchOpJS(Scalar::Type arrayType, Synchronization sync,
                       AtomicOp op, Imm32 value, const BaseIndex& mem,
                       Register temp1, Register temp2, AnyRegister output)
      DEFINED_ON(x86_shared);

  void atomicFetchOpJS(Scalar::Type arrayType, Synchronization sync,
                       AtomicOp op, Register value, const Address& mem,
                       Register valueTemp, Register offsetTemp,
                       Register maskTemp, Register temp, AnyRegister output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicFetchOpJS(Scalar::Type arrayType, Synchronization sync,
                       AtomicOp op, Register value, const BaseIndex& mem,
                       Register valueTemp, Register offsetTemp,
                       Register maskTemp, Register temp, AnyRegister output)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicEffectOpJS(Scalar::Type arrayType, Synchronization sync,
                        AtomicOp op, Register value, const Address& mem,
                        Register temp) DEFINED_ON(arm, arm64, x86_shared);

  void atomicEffectOpJS(Scalar::Type arrayType, Synchronization sync,
                        AtomicOp op, Register value, const BaseIndex& mem,
                        Register temp) DEFINED_ON(arm, arm64, x86_shared);

  void atomicEffectOpJS(Scalar::Type arrayType, Synchronization sync,
                        AtomicOp op, Imm32 value, const Address& mem,
                        Register temp) DEFINED_ON(x86_shared);

  void atomicEffectOpJS(Scalar::Type arrayType, Synchronization sync,
                        AtomicOp op, Imm32 value, const BaseIndex& mem,
                        Register temp) DEFINED_ON(x86_shared);

  void atomicEffectOpJS(Scalar::Type arrayType, Synchronization sync,
                        AtomicOp op, Register value, const Address& mem,
                        Register valueTemp, Register offsetTemp,
                        Register maskTemp)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicEffectOpJS(Scalar::Type arrayType, Synchronization sync,
                        AtomicOp op, Register value, const BaseIndex& mem,
                        Register valueTemp, Register offsetTemp,
                        Register maskTemp)
      DEFINED_ON(mips_shared, loong64, riscv64);

  void atomicIsLockFreeJS(Register value, Register output);

  // ========================================================================
  // Spectre Mitigations.
  //
  // Spectre attacks are side-channel attacks based on cache pollution or
  // slow-execution of some instructions. We have multiple spectre mitigations
  // possible:
  //
  //   - Stop speculative executions, with memory barriers. Memory barriers
  //     force all branches depending on loads to be resolved, and thus
  //     resolve all miss-speculated paths.
  //
  //   - Use conditional move instructions. Some CPUs have a branch predictor,
  //     and not a flag predictor. In such cases, using a conditional move
  //     instruction to zero some pointer/index is enough to add a
  //     data-dependency which prevents any futher executions until the load is
  //     resolved.

  void spectreMaskIndex32(Register index, Register length, Register output);
  void spectreMaskIndex32(Register index, const Address& length,
                          Register output);
  void spectreMaskIndexPtr(Register index, Register length, Register output);
  void spectreMaskIndexPtr(Register index, const Address& length,
                           Register output);

  // The length must be a power of two. Performs a bounds check and Spectre
  // index masking.
  void boundsCheck32PowerOfTwo(Register index, uint32_t length, Label* failure);

  void speculationBarrier() PER_SHARED_ARCH;

  //}}} check_macroassembler_decl_style
 public:
  // Unsafe here means the caller is responsible for Spectre mitigations if
  // needed. Prefer branchTestObjClass or one of the other masm helpers!
  inline void loadObjClassUnsafe(Register obj, Register dest);

  template <typename EmitPreBarrier>
  inline void storeObjShape(Register shape, Register obj,
                            EmitPreBarrier emitPreBarrier);
  template <typename EmitPreBarrier>
  inline void storeObjShape(Shape* shape, Register obj,
                            EmitPreBarrier emitPreBarrier);

  inline void loadObjProto(Register obj, Register dest);

  inline void loadStringLength(Register str, Register dest);

  void loadStringChars(Register str, Register dest, CharEncoding encoding);

  void loadNonInlineStringChars(Register str, Register dest,
                                CharEncoding encoding);
  void loadNonInlineStringCharsForStore(Register str, Register dest);
  void storeNonInlineStringChars(Register chars, Register str);

  void loadInlineStringChars(Register str, Register dest,
                             CharEncoding encoding);
  void loadInlineStringCharsForStore(Register str, Register dest);

 private:
  enum class CharKind { CharCode, CodePoint };

  void branchIfMaybeSplitSurrogatePair(Register leftChild, Register index,
                                       Register scratch, Label* maybeSplit,
                                       Label* notSplit);

  void loadRopeChild(CharKind kind, Register str, Register index,
                     Register output, Register maybeScratch, Label* isLinear,
                     Label* splitSurrogate);

  void branchIfCanLoadStringChar(CharKind kind, Register str, Register index,
                                 Register scratch, Register maybeScratch,
                                 Label* label);
  void branchIfNotCanLoadStringChar(CharKind kind, Register str, Register index,
                                    Register scratch, Register maybeScratch,
                                    Label* label);

  void loadStringChar(CharKind kind, Register str, Register index,
                      Register output, Register scratch1, Register scratch2,
                      Label* fail);

 public:
  void branchIfCanLoadStringChar(Register str, Register index, Register scratch,
                                 Label* label) {
    branchIfCanLoadStringChar(CharKind::CharCode, str, index, scratch,
                              InvalidReg, label);
  }
  void branchIfNotCanLoadStringChar(Register str, Register index,
                                    Register scratch, Label* label) {
    branchIfNotCanLoadStringChar(CharKind::CharCode, str, index, scratch,
                                 InvalidReg, label);
  }

  void branchIfCanLoadStringCodePoint(Register str, Register index,
                                      Register scratch1, Register scratch2,
                                      Label* label) {
    branchIfCanLoadStringChar(CharKind::CodePoint, str, index, scratch1,
                              scratch2, label);
  }
  void branchIfNotCanLoadStringCodePoint(Register str, Register index,
                                         Register scratch1, Register scratch2,
                                         Label* label) {
    branchIfNotCanLoadStringChar(CharKind::CodePoint, str, index, scratch1,
                                 scratch2, label);
  }

  void loadStringChar(Register str, Register index, Register output,
                      Register scratch1, Register scratch2, Label* fail) {
    loadStringChar(CharKind::CharCode, str, index, output, scratch1, scratch2,
                   fail);
  }

  void loadStringChar(Register str, int32_t index, Register output,
                      Register scratch1, Register scratch2, Label* fail);

  void loadStringCodePoint(Register str, Register index, Register output,
                           Register scratch1, Register scratch2, Label* fail) {
    loadStringChar(CharKind::CodePoint, str, index, output, scratch1, scratch2,
                   fail);
  }

  void loadRopeLeftChild(Register str, Register dest);
  void loadRopeRightChild(Register str, Register dest);
  void storeRopeChildren(Register left, Register right, Register str);

  void loadDependentStringBase(Register str, Register dest);
  void storeDependentStringBase(Register base, Register str);

  void loadStringIndexValue(Register str, Register dest, Label* fail);

  /**
   * Store the character in |src| to |dest|.
   */
  template <typename T>
  void storeChar(const T& src, Address dest, CharEncoding encoding) {
    if (encoding == CharEncoding::Latin1) {
      store8(src, dest);
    } else {
      store16(src, dest);
    }
  }

  /**
   * Load the character at |src| into |dest|.
   */
  template <typename T>
  void loadChar(const T& src, Register dest, CharEncoding encoding) {
    if (encoding == CharEncoding::Latin1) {
      load8ZeroExtend(src, dest);
    } else {
      load16ZeroExtend(src, dest);
    }
  }

  /**
   * Load the character at |chars[index + offset]| into |dest|. The optional
   * offset argument is not scaled to the character encoding.
   */
  void loadChar(Register chars, Register index, Register dest,
                CharEncoding encoding, int32_t offset = 0);

  /**
   * Add |index| to |chars| so that |chars| now points at |chars[index]|.
   */
  void addToCharPtr(Register chars, Register index, CharEncoding encoding);

  /**
   * Branch if |src| is not a lead surrogate character.
   */
  void branchIfNotLeadSurrogate(Register src, Label* label);

 private:
  enum class SurrogateChar { Lead, Trail };
  void branchSurrogate(Assembler::Condition cond, Register src,
                       Register scratch, Label* label,
                       SurrogateChar surrogateChar);

 public:
  /**
   * Branch if |src| is a lead surrogate character.
   */
  void branchIfLeadSurrogate(Register src, Register scratch, Label* label) {
    branchSurrogate(Assembler::Equal, src, scratch, label, SurrogateChar::Lead);
  }

  /**
   * Branch if |src| is not a lead surrogate character.
   */
  void branchIfNotLeadSurrogate(Register src, Register scratch, Label* label) {
    branchSurrogate(Assembler::NotEqual, src, scratch, label,
                    SurrogateChar::Lead);
  }

  /**
   * Branch if |src| is not a trail surrogate character.
   */
  void branchIfNotTrailSurrogate(Register src, Register scratch, Label* label) {
    branchSurrogate(Assembler::NotEqual, src, scratch, label,
                    SurrogateChar::Trail);
  }

 private:
  void loadStringFromUnit(Register unit, Register dest,
                          const StaticStrings& staticStrings);
  void loadLengthTwoString(Register c1, Register c2, Register dest,
                           const StaticStrings& staticStrings);

 public:
  /**
   * Lookup the length-one string from the static strings cache.
   */
  void lookupStaticString(Register ch, Register dest,
                          const StaticStrings& staticStrings);

  /**
   * Lookup the length-one string from the static strings cache. Jumps to |fail|
   * when the string wasn't found in the strings cache.
   */
  void lookupStaticString(Register ch, Register dest,
                          const StaticStrings& staticStrings, Label* fail);

  /**
   * Lookup the length-two string from the static strings cache. Jumps to |fail|
   * when the string wasn't found in the strings cache.
   *
   * Clobbers |ch1| and |ch2|.
   */
  void lookupStaticString(Register ch1, Register ch2, Register dest,
                          const StaticStrings& staticStrings, Label* fail);

  /**
   * Lookup the integer string from the static integer strings cache. Jumps to
   * |fail| when the string wasn't found in the strings cache.
   */
  void lookupStaticIntString(Register integer, Register dest, Register scratch,
                             const StaticStrings& staticStrings, Label* fail);
  void lookupStaticIntString(Register integer, Register dest,
                             const StaticStrings& staticStrings, Label* fail) {
    lookupStaticIntString(integer, dest, dest, staticStrings, fail);
  }

  /**
   * Load the string representation of |input| in base |base|. Jumps to |fail|
   * when the string representation needs to be allocated dynamically.
   */
  void loadInt32ToStringWithBase(Register input, Register base, Register dest,
                                 Register scratch1, Register scratch2,
                                 const StaticStrings& staticStrings,
                                 const LiveRegisterSet& volatileRegs,
                                 bool lowerCase, Label* fail);
  void loadInt32ToStringWithBase(Register input, int32_t base, Register dest,
                                 Register scratch1, Register scratch2,
                                 const StaticStrings& staticStrings,
                                 bool lowerCase, Label* fail);

  /**
   * Load the BigInt digits from |bigInt| into |digits|.
   */
  void loadBigIntDigits(Register bigInt, Register digits);

  /**
   * Load the first [u]int64 value from |bigInt| into |dest|.
   */
  void loadBigInt64(Register bigInt, Register64 dest);

  /**
   * Load the first digit from |bigInt| into |dest|. Handles the case when the
   * BigInt digits length is zero.
   *
   * Note: A BigInt digit is a pointer-sized value.
   */
  void loadFirstBigIntDigitOrZero(Register bigInt, Register dest);

  /**
   * Load the number stored in |bigInt| into |dest|. Handles the case when the
   * BigInt digits length is zero. Jumps to |fail| when the number can't be
   * saved into a single pointer-sized register.
   */
  void loadBigInt(Register bigInt, Register dest, Label* fail);

  /**
   * Load the number stored in |bigInt| into |dest|. Doesn't handle the case
   * when the BigInt digits length is zero. Jumps to |fail| when the number
   * can't be saved into a single pointer-sized register.
   */
  void loadBigIntNonZero(Register bigInt, Register dest, Label* fail);

  /**
   * Load the absolute number stored in |bigInt| into |dest|. Handles the case
   * when the BigInt digits length is zero. Jumps to |fail| when the number
   * can't be saved into a single pointer-sized register.
   */
  void loadBigIntAbsolute(Register bigInt, Register dest, Label* fail);

  /**
   * In-place modifies the BigInt digit to a signed pointer-sized value. Jumps
   * to |fail| when the digit exceeds the representable range.
   */
  void bigIntDigitToSignedPtr(Register bigInt, Register digit, Label* fail);

  /**
   * Initialize a BigInt from |val|. Clobbers |val|!
   */
  void initializeBigInt64(Scalar::Type type, Register bigInt, Register64 val);

  /**
   * Initialize a BigInt from the signed, pointer-sized register |val|.
   * Clobbers |val|!
   */
  void initializeBigInt(Register bigInt, Register val);

  /**
   * Initialize a BigInt from the pointer-sized register |val|.
   */
  void initializeBigIntAbsolute(Register bigInt, Register val);

  /**
   * Copy a BigInt. Jumps to |fail| on allocation failure or when the BigInt
   * digits need to be heap allocated.
   */
  void copyBigIntWithInlineDigits(Register src, Register dest, Register temp,
                                  gc::Heap initialHeap, Label* fail);

  /**
   * Compare a BigInt and an Int32 value. Falls through to the false case.
   */
  void compareBigIntAndInt32(JSOp op, Register bigInt, Register int32,
                             Register scratch1, Register scratch2,
                             Label* ifTrue, Label* ifFalse);

  /**
   * Compare two BigInts for equality. Falls through if both BigInts are equal
   * to each other.
   *
   * - When we jump to |notSameLength|, |temp1| holds the length of the right
   *   operand.
   * - When we jump to |notSameDigit|, |temp2| points to the current digit of
   *   the left operand and |temp4| holds the current digit of the right
   *   operand.
   */
  void equalBigInts(Register left, Register right, Register temp1,
                    Register temp2, Register temp3, Register temp4,
                    Label* notSameSign, Label* notSameLength,
                    Label* notSameDigit);

  void loadJSContext(Register dest);

  void loadGlobalObjectData(Register dest);

  void loadRealmFuse(RealmFuses::FuseIndex index, Register dest);

  void switchToRealm(Register realm);
  void switchToRealm(const void* realm, Register scratch);
  void switchToObjectRealm(Register obj, Register scratch);
  void switchToBaselineFrameRealm(Register scratch);
  void switchToWasmInstanceRealm(Register scratch1, Register scratch2);
  void debugAssertContextRealm(const void* realm, Register scratch);

  void loadJitActivation(Register dest);

  void guardSpecificAtom(Register str, JSAtom* atom, Register scratch,
                         const LiveRegisterSet& volatileRegs, Label* fail);

  void guardStringToInt32(Register str, Register output, Register scratch,
                          LiveRegisterSet volatileRegs, Label* fail);

  template <typename T>
  void loadTypedOrValue(const T& src, TypedOrValueRegister dest) {
    if (dest.hasValue()) {
      loadValue(src, dest.valueReg());
    } else {
      loadUnboxedValue(src, dest.type(), dest.typedReg());
    }
  }

  template <typename T>
  void storeTypedOrValue(TypedOrValueRegister src, const T& dest) {
    if (src.hasValue()) {
      storeValue(src.valueReg(), dest);
    } else if (IsFloatingPointType(src.type())) {
      FloatRegister reg = src.typedReg().fpu();
      if (src.type() == MIRType::Float32) {
        ScratchDoubleScope fpscratch(*this);
        convertFloat32ToDouble(reg, fpscratch);
        boxDouble(fpscratch, dest);
      } else {
        boxDouble(reg, dest);
      }
    } else {
      storeValue(ValueTypeFromMIRType(src.type()), src.typedReg().gpr(), dest);
    }
  }

  template <typename T>
  void storeConstantOrRegister(const ConstantOrRegister& src, const T& dest) {
    if (src.constant()) {
      storeValue(src.value(), dest);
    } else {
      storeTypedOrValue(src.reg(), dest);
    }
  }

  void storeCallPointerResult(Register reg) {
    if (reg != ReturnReg) {
      mov(ReturnReg, reg);
    }
  }

  inline void storeCallBoolResult(Register reg);
  inline void storeCallInt32Result(Register reg);

  void storeCallFloatResult(FloatRegister reg) {
    if (reg != ReturnDoubleReg) {
      moveDouble(ReturnDoubleReg, reg);
    }
  }

  inline void storeCallResultValue(AnyRegister dest, JSValueType type);

  void storeCallResultValue(ValueOperand dest) {
#if defined(JS_NUNBOX32)
    // reshuffle the return registers used for a call result to store into
    // dest, using ReturnReg as a scratch register if necessary. This must
    // only be called after returning from a call, at a point when the
    // return register is not live. XXX would be better to allow wrappers
    // to store the return value to different places.
    if (dest.typeReg() == JSReturnReg_Data) {
      if (dest.payloadReg() == JSReturnReg_Type) {
        // swap the two registers.
        mov(JSReturnReg_Type, ReturnReg);
        mov(JSReturnReg_Data, JSReturnReg_Type);
        mov(ReturnReg, JSReturnReg_Data);
      } else {
        mov(JSReturnReg_Data, dest.payloadReg());
        mov(JSReturnReg_Type, dest.typeReg());
      }
    } else {
      mov(JSReturnReg_Type, dest.typeReg());
      mov(JSReturnReg_Data, dest.payloadReg());
    }
#elif defined(JS_PUNBOX64)
    if (dest.valueReg() != JSReturnReg) {
      mov(JSReturnReg, dest.valueReg());
    }
#else
#  error "Bad architecture"
#endif
  }

  inline void storeCallResultValue(TypedOrValueRegister dest);

 private:
  TrampolinePtr preBarrierTrampoline(MIRType type);

  template <typename T>
  void unguardedCallPreBarrier(const T& address, MIRType type) {
    Label done;
    if (type == MIRType::Value) {
      branchTestGCThing(Assembler::NotEqual, address, &done);
    } else if (type == MIRType::Object || type == MIRType::String) {
      branchPtr(Assembler::Equal, address, ImmWord(0), &done);
    }

    Push(PreBarrierReg);
    computeEffectiveAddress(address, PreBarrierReg);

    TrampolinePtr preBarrier = preBarrierTrampoline(type);

    call(preBarrier);
    Pop(PreBarrierReg);
    // On arm64, SP may be < PSP now (that's OK).
    // eg testcase: tests/auto-regress/bug702915.js
    bind(&done);
  }

 public:
  template <typename T>
  void guardedCallPreBarrier(const T& address, MIRType type) {
    Label done;
    branchTestNeedsIncrementalBarrier(Assembler::Zero, &done);
    unguardedCallPreBarrier(address, type);
    bind(&done);
  }

  // Like guardedCallPreBarrier, but unlike guardedCallPreBarrier this can be
  // called from runtime-wide trampolines because it loads cx->zone (instead of
  // baking in the current Zone) if JitContext::realm is nullptr.
  template <typename T>
  void guardedCallPreBarrierAnyZone(const T& address, MIRType type,
                                    Register scratch) {
    Label done;
    branchTestNeedsIncrementalBarrierAnyZone(Assembler::Zero, &done, scratch);
    unguardedCallPreBarrier(address, type);
    bind(&done);
  }

  enum class Uint32Mode { FailOnDouble, ForceDouble };

  void boxUint32(Register source, ValueOperand dest, Uint32Mode uint32Mode,
                 Label* fail);

  template <typename T>
  void loadFromTypedArray(Scalar::Type arrayType, const T& src,
                          AnyRegister dest, Register temp, Label* fail);

  template <typename T>
  void loadFromTypedArray(Scalar::Type arrayType, const T& src,
                          const ValueOperand& dest, Uint32Mode uint32Mode,
                          Register temp, Label* fail);

  template <typename T>
  void loadFromTypedBigIntArray(Scalar::Type arrayType, const T& src,
                                Register bigInt, Register64 temp);

  template <typename S, typename T>
  void storeToTypedIntArray(Scalar::Type arrayType, const S& value,
                            const T& dest) {
    switch (arrayType) {
      case Scalar::Int8:
      case Scalar::Uint8:
      case Scalar::Uint8Clamped:
        store8(value, dest);
        break;
      case Scalar::Int16:
      case Scalar::Uint16:
        store16(value, dest);
        break;
      case Scalar::Int32:
      case Scalar::Uint32:
        store32(value, dest);
        break;
      default:
        MOZ_CRASH("Invalid typed array type");
    }
  }

  void storeToTypedFloatArray(Scalar::Type arrayType, FloatRegister value,
                              const BaseIndex& dest);
  void storeToTypedFloatArray(Scalar::Type arrayType, FloatRegister value,
                              const Address& dest);

  void storeToTypedBigIntArray(Scalar::Type arrayType, Register64 value,
                               const BaseIndex& dest);
  void storeToTypedBigIntArray(Scalar::Type arrayType, Register64 value,
                               const Address& dest);

  void memoryBarrierBefore(Synchronization sync);
  void memoryBarrierAfter(Synchronization sync);

  void debugAssertIsObject(const ValueOperand& val);
  void debugAssertObjHasFixedSlots(Register obj, Register scratch);

  void debugAssertObjectHasClass(Register obj, Register scratch,
                                 const JSClass* clasp);

  void debugAssertGCThingIsTenured(Register ptr, Register temp);

  void branchArrayIsNotPacked(Register array, Register temp1, Register temp2,
                              Label* label);

  void setIsPackedArray(Register obj, Register output, Register temp);

  void packedArrayPop(Register array, ValueOperand output, Register temp1,
                      Register temp2, Label* fail);
  void packedArrayShift(Register array, ValueOperand output, Register temp1,
                        Register temp2, LiveRegisterSet volatileRegs,
                        Label* fail);

  void loadArgumentsObjectElement(Register obj, Register index,
                                  ValueOperand output, Register temp,
                                  Label* fail);
  void loadArgumentsObjectElementHole(Register obj, Register index,
                                      ValueOperand output, Register temp,
                                      Label* fail);
  void loadArgumentsObjectElementExists(Register obj, Register index,
                                        Register output, Register temp,
                                        Label* fail);

  void loadArgumentsObjectLength(Register obj, Register output, Label* fail);

  void branchTestArgumentsObjectFlags(Register obj, Register temp,
                                      uint32_t flags, Condition cond,
                                      Label* label);

  void typedArrayElementSize(Register obj, Register output);

 private:
  // Shift |output| by the element shift of the ResizableTypedArray in |obj|.
  void resizableTypedArrayElementShiftBy(Register obj, Register output,
                                         Register scratch);

 public:
  void branchIfClassIsNotTypedArray(Register clasp, Label* notTypedArray);
  void branchIfClassIsNotFixedLengthTypedArray(Register clasp,
                                               Label* notTypedArray);
  void branchIfClassIsNotResizableTypedArray(Register clasp,
                                             Label* notTypedArray);

 private:
  enum class BranchIfDetached { No, Yes };

  void branchIfHasDetachedArrayBuffer(BranchIfDetached branchIf, Register obj,
                                      Register temp, Label* label);

 public:
  void branchIfHasDetachedArrayBuffer(Register obj, Register temp,
                                      Label* label) {
    branchIfHasDetachedArrayBuffer(BranchIfDetached::Yes, obj, temp, label);
  }

  void branchIfHasAttachedArrayBuffer(Register obj, Register temp,
                                      Label* label) {
    branchIfHasDetachedArrayBuffer(BranchIfDetached::No, obj, temp, label);
  }

  void branchIfResizableArrayBufferViewOutOfBounds(Register obj, Register temp,
                                                   Label* label);

  void branchIfResizableArrayBufferViewInBounds(Register obj, Register temp,
                                                Label* label);

  void branchIfNativeIteratorNotReusable(Register ni, Label* notReusable);
  void branchNativeIteratorIndices(Condition cond, Register ni, Register temp,
                                   NativeIteratorIndices kind, Label* label);

  void maybeLoadIteratorFromShape(Register obj, Register dest, Register temp,
                                  Register temp2, Register temp3,
                                  Label* failure);

  void iteratorMore(Register obj, ValueOperand output, Register temp);
  void iteratorClose(Register obj, Register temp1, Register temp2,
                     Register temp3);
  void registerIterator(Register enumeratorsList, Register iter, Register temp);

  void toHashableNonGCThing(ValueOperand value, ValueOperand result,
                            FloatRegister tempFloat);

  void toHashableValue(ValueOperand value, ValueOperand result,
                       FloatRegister tempFloat, Label* atomizeString,
                       Label* tagString);

 private:
  void scrambleHashCode(Register result);

 public:
  void prepareHashNonGCThing(ValueOperand value, Register result,
                             Register temp);
  void prepareHashString(Register str, Register result, Register temp);
  void prepareHashSymbol(Register sym, Register result);
  void prepareHashBigInt(Register bigInt, Register result, Register temp1,
                         Register temp2, Register temp3);
  void prepareHashObject(Register setObj, ValueOperand value, Register result,
                         Register temp1, Register temp2, Register temp3,
                         Register temp4);
  void prepareHashValue(Register setObj, ValueOperand value, Register result,
                        Register temp1, Register temp2, Register temp3,
                        Register temp4);

 private:
  enum class IsBigInt { No, Yes, Maybe };

  /**
   * Search for a value in a OrderedHashTable.
   *
   * When we jump to |found|, |entryTemp| holds the found hashtable entry.
   */
  template <typename OrderedHashTable>
  void orderedHashTableLookup(Register setOrMapObj, ValueOperand value,
                              Register hash, Register entryTemp, Register temp1,
                              Register temp3, Register temp4, Register temp5,
                              Label* found, IsBigInt isBigInt);

  void setObjectHas(Register setObj, ValueOperand value, Register hash,
                    Register result, Register temp1, Register temp2,
                    Register temp3, Register temp4, IsBigInt isBigInt);

  void mapObjectHas(Register mapObj, ValueOperand value, Register hash,
                    Register result, Register temp1, Register temp2,
                    Register temp3, Register temp4, IsBigInt isBigInt);

  void mapObjectGet(Register mapObj, ValueOperand value, Register hash,
                    ValueOperand result, Register temp1, Register temp2,
                    Register temp3, Register temp4, Register temp5,
                    IsBigInt isBigInt);

 public:
  void setObjectHasNonBigInt(Register setObj, ValueOperand value, Register hash,
                             Register result, Register temp1, Register temp2) {
    return setObjectHas(setObj, value, hash, result, temp1, temp2, InvalidReg,
                        InvalidReg, IsBigInt::No);
  }
  void setObjectHasBigInt(Register setObj, ValueOperand value, Register hash,
                          Register result, Register temp1, Register temp2,
                          Register temp3, Register temp4) {
    return setObjectHas(setObj, value, hash, result, temp1, temp2, temp3, temp4,
                        IsBigInt::Yes);
  }
  void setObjectHasValue(Register setObj, ValueOperand value, Register hash,
                         Register result, Register temp1, Register temp2,
                         Register temp3, Register temp4) {
    return setObjectHas(setObj, value, hash, result, temp1, temp2, temp3, temp4,
                        IsBigInt::Maybe);
  }

  void mapObjectHasNonBigInt(Register mapObj, ValueOperand value, Register hash,
                             Register result, Register temp1, Register temp2) {
    return mapObjectHas(mapObj, value, hash, result, temp1, temp2, InvalidReg,
                        InvalidReg, IsBigInt::No);
  }
  void mapObjectHasBigInt(Register mapObj, ValueOperand value, Register hash,
                          Register result, Register temp1, Register temp2,
                          Register temp3, Register temp4) {
    return mapObjectHas(mapObj, value, hash, result, temp1, temp2, temp3, temp4,
                        IsBigInt::Yes);
  }
  void mapObjectHasValue(Register mapObj, ValueOperand value, Register hash,
                         Register result, Register temp1, Register temp2,
                         Register temp3, Register temp4) {
    return mapObjectHas(mapObj, value, hash, result, temp1, temp2, temp3, temp4,
                        IsBigInt::Maybe);
  }

  void mapObjectGetNonBigInt(Register mapObj, ValueOperand value, Register hash,
                             ValueOperand result, Register temp1,
                             Register temp2, Register temp3) {
    return mapObjectGet(mapObj, value, hash, result, temp1, temp2, temp3,
                        InvalidReg, InvalidReg, IsBigInt::No);
  }
  void mapObjectGetBigInt(Register mapObj, ValueOperand value, Register hash,
                          ValueOperand result, Register temp1, Register temp2,
                          Register temp3, Register temp4, Register temp5) {
    return mapObjectGet(mapObj, value, hash, result, temp1, temp2, temp3, temp4,
                        temp5, IsBigInt::Yes);
  }
  void mapObjectGetValue(Register mapObj, ValueOperand value, Register hash,
                         ValueOperand result, Register temp1, Register temp2,
                         Register temp3, Register temp4, Register temp5) {
    return mapObjectGet(mapObj, value, hash, result, temp1, temp2, temp3, temp4,
                        temp5, IsBigInt::Maybe);
  }

 private:
  template <typename OrderedHashTable>
  void loadOrderedHashTableCount(Register setOrMapObj, Register result);

 public:
  void loadSetObjectSize(Register setObj, Register result);
  void loadMapObjectSize(Register mapObj, Register result);

  // Inline version of js_TypedArray_uint8_clamp_double.
  // This function clobbers the input register.
  void clampDoubleToUint8(FloatRegister input, Register output) PER_ARCH;

  using MacroAssemblerSpecific::ensureDouble;

  template <typename S>
  void ensureDouble(const S& source, FloatRegister dest, Label* failure) {
    Label isDouble, done;
    branchTestDouble(Assembler::Equal, source, &isDouble);
    branchTestInt32(Assembler::NotEqual, source, failure);

    convertInt32ToDouble(source, dest);
    jump(&done);

    bind(&isDouble);
    unboxDouble(source, dest);

    bind(&done);
  }

  // Inline allocation.
 private:
  void checkAllocatorState(Register temp, gc::AllocKind allocKind, Label* fail);
  bool shouldNurseryAllocate(gc::AllocKind allocKind, gc::Heap initialHeap);
  void nurseryAllocateObject(
      Register result, Register temp, gc::AllocKind allocKind,
      size_t nDynamicSlots, Label* fail,
      const AllocSiteInput& allocSite = AllocSiteInput());
  void bumpPointerAllocate(Register result, Register temp, Label* fail,
                           CompileZone* zone, JS::TraceKind traceKind,
                           uint32_t size,
                           const AllocSiteInput& allocSite = AllocSiteInput());
  void updateAllocSite(Register temp, Register result, CompileZone* zone,
                       Register site);

  void freeListAllocate(Register result, Register temp, gc::AllocKind allocKind,
                        Label* fail);
  void allocateObject(Register result, Register temp, gc::AllocKind allocKind,
                      uint32_t nDynamicSlots, gc::Heap initialHeap, Label* fail,
                      const AllocSiteInput& allocSite = AllocSiteInput());
  void nurseryAllocateString(Register result, Register temp,
                             gc::AllocKind allocKind, Label* fail);
  void allocateString(Register result, Register temp, gc::AllocKind allocKind,
                      gc::Heap initialHeap, Label* fail);
  void nurseryAllocateBigInt(Register result, Register temp, Label* fail);
  void copySlotsFromTemplate(Register obj,
                             const TemplateNativeObject& templateObj,
                             uint32_t start, uint32_t end);
  void fillSlotsWithConstantValue(Address addr, Register temp, uint32_t start,
                                  uint32_t end, const Value& v);
  void fillSlotsWithUndefined(Address addr, Register temp, uint32_t start,
                              uint32_t end);
  void fillSlotsWithUninitialized(Address addr, Register temp, uint32_t start,
                                  uint32_t end);

  void initGCSlots(Register obj, Register temp,
                   const TemplateNativeObject& templateObj);

 public:
  void callFreeStub(Register slots);
  void createGCObject(Register result, Register temp,
                      const TemplateObject& templateObj, gc::Heap initialHeap,
                      Label* fail, bool initContents = true);

  void createPlainGCObject(Register result, Register shape, Register temp,
                           Register temp2, uint32_t numFixedSlots,
                           uint32_t numDynamicSlots, gc::AllocKind allocKind,
                           gc::Heap initialHeap, Label* fail,
                           const AllocSiteInput& allocSite,
                           bool initContents = true);

  // dynamicSlotsTemp is used to initialize the dynamic slots after allocating
  // the object. If numUsedDynamicSlots == 0, it may be InvalidReg.
  void createArrayWithFixedElements(
      Register result, Register shape, Register temp, Register dynamicSlotsTemp,
      uint32_t arrayLength, uint32_t arrayCapacity,
      uint32_t numUsedDynamicSlots, uint32_t numDynamicSlots,
      gc::AllocKind allocKind, gc::Heap initialHeap, Label* fail,
      const AllocSiteInput& allocSite = AllocSiteInput());

  void initGCThing(Register obj, Register temp,
                   const TemplateObject& templateObj, bool initContents = true);

  enum class TypedArrayLength { Fixed, Dynamic };

  void initTypedArraySlots(Register obj, Register temp, Register lengthReg,
                           LiveRegisterSet liveRegs, Label* fail,
                           FixedLengthTypedArrayObject* templateObj,
                           TypedArrayLength lengthKind);

  void newGCString(Register result, Register temp, gc::Heap initialHeap,
                   Label* fail);
  void newGCFatInlineString(Register result, Register temp,
                            gc::Heap initialHeap, Label* fail);

  void newGCBigInt(Register result, Register temp, gc::Heap initialHeap,
                   Label* fail);

 private:
  void branchIfNotStringCharsEquals(Register stringChars,
                                    const JSLinearString* linear, Label* label);

 public:
  // Returns true if |linear| is a (non-empty) string which can be compared
  // using |compareStringChars|.
  static bool canCompareStringCharsInline(const JSLinearString* linear);

  // Load the string characters in preparation for |compareStringChars|.
  void loadStringCharsForCompare(Register input, const JSLinearString* linear,
                                 Register stringChars, Label* fail);

  // Compare string characters based on the equality operator. The string
  // characters must be at least as long as the length of |linear|.
  void compareStringChars(JSOp op, Register stringChars,
                          const JSLinearString* linear, Register result);

  // Compares two strings for equality based on the JSOP.
  // This checks for identical pointers, atoms and length and fails for
  // everything else.
  void compareStrings(JSOp op, Register left, Register right, Register result,
                      Label* fail);

  // Result of the typeof operation. Falls back to slow-path for proxies.
  void typeOfObject(Register objReg, Register scratch, Label* slow,
                    Label* isObject, Label* isCallable, Label* isUndefined);

  // Implementation of IsCallable. Doesn't handle proxies.
  void isCallable(Register obj, Register output, Label* isProxy) {
    isCallableOrConstructor(true, obj, output, isProxy);
  }
  void isConstructor(Register obj, Register output, Label* isProxy) {
    isCallableOrConstructor(false, obj, output, isProxy);
  }

  void setIsCrossRealmArrayConstructor(Register obj, Register output);

  void setIsDefinitelyTypedArrayConstructor(Register obj, Register output);

  void loadMegamorphicCache(Register dest);
  void lookupStringInAtomCacheLastLookups(Register str, Register scratch,
                                          Register output, Label* fail);
  void loadMegamorphicSetPropCache(Register dest);

  void loadAtomOrSymbolAndHash(ValueOperand value, Register outId,
                               Register outHash, Label* cacheMiss);

  void loadAtomHash(Register id, Register hash, Label* done);

  void emitExtractValueFromMegamorphicCacheEntry(
      Register obj, Register entry, Register scratch1, Register scratch2,
      ValueOperand output, Label* cacheHit, Label* cacheMiss);

  template <typename IdOperandType>
  void emitMegamorphicCacheLookupByValueCommon(
      IdOperandType id, Register obj, Register scratch1, Register scratch2,
      Register outEntryPtr, Label* cacheMiss, Label* cacheMissWithEntry);

  void emitMegamorphicCacheLookup(PropertyKey id, Register obj,
                                  Register scratch1, Register scratch2,
                                  Register outEntryPtr, ValueOperand output,
                                  Label* cacheHit);

  // NOTE: |id| must either be a ValueOperand or a Register. If it is a
  // Register, we assume that it is an atom.
  template <typename IdOperandType>
  void emitMegamorphicCacheLookupByValue(IdOperandType id, Register obj,
                                         Register scratch1, Register scratch2,
                                         Register outEntryPtr,
                                         ValueOperand output, Label* cacheHit);

  void emitMegamorphicCacheLookupExists(ValueOperand id, Register obj,
                                        Register scratch1, Register scratch2,
                                        Register outEntryPtr, Register output,
                                        Label* cacheHit, bool hasOwn);

  // Given a PropertyIteratorObject with valid indices, extract the current
  // PropertyIndex, storing the index in |outIndex| and the kind in |outKind|
  void extractCurrentIndexAndKindFromIterator(Register iterator,
                                              Register outIndex,
                                              Register outKind);

  template <typename IdType>
#ifdef JS_CODEGEN_X86
  // See MegamorphicSetElement in LIROps.yaml
  void emitMegamorphicCachedSetSlot(IdType id, Register obj, Register scratch1,
                                    ValueOperand value, Label* cacheHit,
                                    void (*emitPreBarrier)(MacroAssembler&,
                                                           const Address&,
                                                           MIRType));
#else
  void emitMegamorphicCachedSetSlot(
      IdType id, Register obj, Register scratch1, Register scratch2,
      Register scratch3, ValueOperand value, Label* cacheHit,
      void (*emitPreBarrier)(MacroAssembler&, const Address&, MIRType));
#endif

  void loadDOMExpandoValueGuardGeneration(
      Register obj, ValueOperand output,
      JS::ExpandoAndGeneration* expandoAndGeneration, uint64_t generation,
      Label* fail);

  void guardNonNegativeIntPtrToInt32(Register reg, Label* fail);

  void loadArrayBufferByteLengthIntPtr(Register obj, Register output);
  void loadArrayBufferViewByteOffsetIntPtr(Register obj, Register output);
  void loadArrayBufferViewLengthIntPtr(Register obj, Register output);

  void loadGrowableSharedArrayBufferByteLengthIntPtr(Synchronization sync,
                                                     Register obj,
                                                     Register output);

 private:
  enum class ResizableArrayBufferView { TypedArray, DataView };

  void loadResizableArrayBufferViewLengthIntPtr(ResizableArrayBufferView view,
                                                Synchronization sync,
                                                Register obj, Register output,
                                                Register scratch);

 public:
  void loadResizableTypedArrayLengthIntPtr(Synchronization sync, Register obj,
                                           Register output, Register scratch) {
    loadResizableArrayBufferViewLengthIntPtr(
        ResizableArrayBufferView::TypedArray, sync, obj, output, scratch);
  }

  void loadResizableDataViewByteLengthIntPtr(Synchronization sync, Register obj,
                                             Register output,
                                             Register scratch) {
    loadResizableArrayBufferViewLengthIntPtr(ResizableArrayBufferView::DataView,
                                             sync, obj, output, scratch);
  }

  void loadResizableTypedArrayByteOffsetMaybeOutOfBoundsIntPtr(
      Register obj, Register output, Register scratch);

 private:
  void isCallableOrConstructor(bool isCallable, Register obj, Register output,
                               Label* isProxy);

 public:
  // Generates code used to complete a bailout.
  void generateBailoutTail(Register scratch, Register bailoutInfo);

 public:
#ifndef JS_CODEGEN_ARM64
  // StackPointer manipulation functions.
  // On ARM64, the StackPointer is implemented as two synchronized registers.
  // Code shared across platforms must use these functions to be valid.
  template <typename T>
  inline void addToStackPtr(T t);
  template <typename T>
  inline void addStackPtrTo(T t);

  void subFromStackPtr(Imm32 imm32)
      DEFINED_ON(mips32, mips64, loong64, riscv64, wasm32, arm, x86, x64);
  void subFromStackPtr(Register reg);

  template <typename T>
  void subStackPtrFrom(T t) {
    subPtr(getStackPointer(), t);
  }

  template <typename T>
  void andToStackPtr(T t) {
    andPtr(t, getStackPointer());
  }

  template <typename T>
  void moveToStackPtr(T t) {
    movePtr(t, getStackPointer());
  }
  template <typename T>
  void moveStackPtrTo(T t) {
    movePtr(getStackPointer(), t);
  }

  template <typename T>
  void loadStackPtr(T t) {
    loadPtr(t, getStackPointer());
  }
  template <typename T>
  void storeStackPtr(T t) {
    storePtr(getStackPointer(), t);
  }

  // StackPointer testing functions.
  // On ARM64, sp can function as the zero register depending on context.
  // Code shared across platforms must use these functions to be valid.
  template <typename T>
  inline void branchTestStackPtr(Condition cond, T t, Label* label);
  template <typename T>
  inline void branchStackPtr(Condition cond, T rhs, Label* label);
  template <typename T>
  inline void branchStackPtrRhs(Condition cond, T lhs, Label* label);

  // Move the stack pointer based on the requested amount.
  inline void reserveStack(uint32_t amount);
#else  // !JS_CODEGEN_ARM64
  void reserveStack(uint32_t amount);
#endif

 public:
  void enableProfilingInstrumentation() {
    emitProfilingInstrumentation_ = true;
  }

 private:
  // This class is used to surround call sites throughout the assembler. This
  // is used by callWithABI, and callJit functions, except if suffixed by
  // NoProfiler.
  class MOZ_RAII AutoProfilerCallInstrumentation {
   public:
    explicit AutoProfilerCallInstrumentation(MacroAssembler& masm);
    ~AutoProfilerCallInstrumentation() = default;
  };
  friend class AutoProfilerCallInstrumentation;

  void appendProfilerCallSite(CodeOffset label) {
    propagateOOM(profilerCallSites_.append(label));
  }

  // Fix up the code pointers to be written for locations where profilerCallSite
  // emitted moves of RIP to a register.
  void linkProfilerCallSites(JitCode* code);

  // This field is used to manage profiling instrumentation output. If
  // provided and enabled, then instrumentation will be emitted around call
  // sites.
  bool emitProfilingInstrumentation_;

  // Record locations of the call sites.
  Vector<CodeOffset, 0, SystemAllocPolicy> profilerCallSites_;

 public:
  void loadJitCodeRaw(Register func, Register dest);
  void loadBaselineJitCodeRaw(Register func, Register dest,
                              Label* failure = nullptr);
  void storeICScriptInJSContext(Register icScript);

  void loadBaselineFramePtr(Register framePtr, Register dest);

  void pushBaselineFramePtr(Register framePtr, Register scratch) {
    loadBaselineFramePtr(framePtr, scratch);
    push(scratch);
  }

  void PushBaselineFramePtr(Register framePtr, Register scratch) {
    loadBaselineFramePtr(framePtr, scratch);
    Push(scratch);
  }

  using MacroAssemblerSpecific::movePtr;

  void movePtr(TrampolinePtr ptr, Register dest) {
    movePtr(ImmPtr(ptr.value), dest);
  }

 private:
  void handleFailure();

 public:
  Label* exceptionLabel() {
    // Exceptions are currently handled the same way as sequential failures.
    return &failureLabel_;
  }

  Label* failureLabel() { return &failureLabel_; }

  void finish();
  void link(JitCode* code);

  void assumeUnreachable(const char* output);

  void printf(const char* output);
  void printf(const char* output, Register value);

#define DISPATCH_FLOATING_POINT_OP(method, type, arg1d, arg1f, arg2) \
  MOZ_ASSERT(IsFloatingPointType(type));                             \
  if (type == MIRType::Double)                                       \
    method##Double(arg1d, arg2);                                     \
  else                                                               \
    method##Float32(arg1f, arg2);

  void loadConstantFloatingPoint(double d, float f, FloatRegister dest,
                                 MIRType destType) {
    DISPATCH_FLOATING_POINT_OP(loadConstant, destType, d, f, dest);
  }
  void boolValueToFloatingPoint(ValueOperand value, FloatRegister dest,
                                MIRType destType) {
    DISPATCH_FLOATING_POINT_OP(boolValueTo, destType, value, value, dest);
  }
  void int32ValueToFloatingPoint(ValueOperand value, FloatRegister dest,
                                 MIRType destType) {
    DISPATCH_FLOATING_POINT_OP(int32ValueTo, destType, value, value, dest);
  }
  void convertInt32ToFloatingPoint(Register src, FloatRegister dest,
                                   MIRType destType) {
    DISPATCH_FLOATING_POINT_OP(convertInt32To, destType, src, src, dest);
  }

#undef DISPATCH_FLOATING_POINT_OP

  void convertValueToFloatingPoint(ValueOperand value, FloatRegister output,
                                   Label* fail, MIRType outputType);

  void outOfLineTruncateSlow(FloatRegister src, Register dest,
                             bool widenFloatToDouble, bool compilingWasm,
                             wasm::BytecodeOffset callOffset);

  void convertInt32ValueToDouble(ValueOperand val);

  void convertValueToDouble(ValueOperand value, FloatRegister output,
                            Label* fail) {
    convertValueToFloatingPoint(value, output, fail, MIRType::Double);
  }

  void convertValueToFloat(ValueOperand value, FloatRegister output,
                           Label* fail) {
    convertValueToFloatingPoint(value, output, fail, MIRType::Float32);
  }

  //
  // Functions for converting values to int.
  //
  void convertDoubleToInt(FloatRegister src, Register output,
                          FloatRegister temp, Label* truncateFail, Label* fail,
                          IntConversionBehavior behavior);

  // Strings may be handled by providing labels to jump to when the behavior
  // is truncation or clamping. The subroutine, usually an OOL call, is
  // passed the unboxed string in |stringReg| and should convert it to a
  // double store into |temp|.
  void convertValueToInt(
      ValueOperand value, Label* handleStringEntry, Label* handleStringRejoin,
      Label* truncateDoubleSlow, Register stringReg, FloatRegister temp,
      Register output, Label* fail, IntConversionBehavior behavior,
      IntConversionInputKind conversion = IntConversionInputKind::Any);

  // This carries over the MToNumberInt32 operation on the ValueOperand
  // input; see comment at the top of this class.
  void convertValueToInt32(
      ValueOperand value, FloatRegister temp, Register output, Label* fail,
      bool negativeZeroCheck,
      IntConversionInputKind conversion = IntConversionInputKind::Any) {
    convertValueToInt(
        value, nullptr, nullptr, nullptr, InvalidReg, temp, output, fail,
        negativeZeroCheck ? IntConversionBehavior::NegativeZeroCheck
                          : IntConversionBehavior::Normal,
        conversion);
  }

  // This carries over the MTruncateToInt32 operation on the ValueOperand
  // input; see the comment at the top of this class.
  void truncateValueToInt32(ValueOperand value, Label* handleStringEntry,
                            Label* handleStringRejoin,
                            Label* truncateDoubleSlow, Register stringReg,
                            FloatRegister temp, Register output, Label* fail) {
    convertValueToInt(value, handleStringEntry, handleStringRejoin,
                      truncateDoubleSlow, stringReg, temp, output, fail,
                      IntConversionBehavior::Truncate);
  }

  void truncateValueToInt32(ValueOperand value, FloatRegister temp,
                            Register output, Label* fail) {
    truncateValueToInt32(value, nullptr, nullptr, nullptr, InvalidReg, temp,
                         output, fail);
  }

  // Convenience functions for clamping values to uint8.
  void clampValueToUint8(ValueOperand value, Label* handleStringEntry,
                         Label* handleStringRejoin, Register stringReg,
                         FloatRegister temp, Register output, Label* fail) {
    convertValueToInt(value, handleStringEntry, handleStringRejoin, nullptr,
                      stringReg, temp, output, fail,
                      IntConversionBehavior::ClampToUint8);
  }

  [[nodiscard]] bool icBuildOOLFakeExitFrame(void* fakeReturnAddr,
                                             AutoSaveLiveRegisters& save);

  // Align the stack pointer based on the number of arguments which are pushed
  // on the stack, such that the JitFrameLayout would be correctly aligned on
  // the JitStackAlignment.
  void alignJitStackBasedOnNArgs(Register nargs, bool countIncludesThis);
  void alignJitStackBasedOnNArgs(uint32_t argc, bool countIncludesThis);

  inline void assertStackAlignment(uint32_t alignment, int32_t offset = 0);

  void touchFrameValues(Register numStackValues, Register scratch1,
                        Register scratch2);

#ifdef JS_64BIT
  // See comment block "64-bit GPRs carrying 32-bit values" above.  This asserts
  // that the high bits of the register are appropriate for the architecture and
  // the value in the low bits.
  void debugAssertCanonicalInt32(Register r);
#endif
};

// StackMacroAssembler checks no GC will happen while it's on the stack.
class MOZ_RAII StackMacroAssembler : public MacroAssembler {
  JS::AutoCheckCannotGC nogc;

 public:
  StackMacroAssembler(JSContext* cx, TempAllocator& alloc);
};

// WasmMacroAssembler does not contain GC pointers, so it doesn't need the no-GC
// checking StackMacroAssembler has.
class MOZ_RAII WasmMacroAssembler : public MacroAssembler {
 public:
  explicit WasmMacroAssembler(TempAllocator& alloc, bool limitedSize = true);
  explicit WasmMacroAssembler(TempAllocator& alloc,
                              const wasm::ModuleEnvironment& env,
                              bool limitedSize = true);
  ~WasmMacroAssembler() { assertNoGCThings(); }
};

// Heap-allocated MacroAssembler used for Ion off-thread code generation.
// GC cancels off-thread compilations.
class IonHeapMacroAssembler : public MacroAssembler {
 public:
  IonHeapMacroAssembler(TempAllocator& alloc, CompileRealm* realm);
};

//{{{ check_macroassembler_style
inline uint32_t MacroAssembler::framePushed() const { return framePushed_; }

inline void MacroAssembler::setFramePushed(uint32_t framePushed) {
  framePushed_ = framePushed;
}

inline void MacroAssembler::adjustFrame(int32_t value) {
  MOZ_ASSERT_IF(value < 0, framePushed_ >= uint32_t(-value));
  setFramePushed(framePushed_ + value);
}

inline void MacroAssembler::implicitPop(uint32_t bytes) {
  MOZ_ASSERT(bytes % sizeof(intptr_t) == 0);
  MOZ_ASSERT(bytes <= INT32_MAX);
  adjustFrame(-int32_t(bytes));
}
//}}} check_macroassembler_style

static inline Assembler::DoubleCondition JSOpToDoubleCondition(JSOp op) {
  switch (op) {
    case JSOp::Eq:
    case JSOp::StrictEq:
      return Assembler::DoubleEqual;
    case JSOp::Ne:
    case JSOp::StrictNe:
      return Assembler::DoubleNotEqualOrUnordered;
    case JSOp::Lt:
      return Assembler::DoubleLessThan;
    case JSOp::Le:
      return Assembler::DoubleLessThanOrEqual;
    case JSOp::Gt:
      return Assembler::DoubleGreaterThan;
    case JSOp::Ge:
      return Assembler::DoubleGreaterThanOrEqual;
    default:
      MOZ_CRASH("Unexpected comparison operation");
  }
}

// Note: the op may have been inverted during lowering (to put constants in a
// position where they can be immediates), so it is important to use the
// lir->jsop() instead of the mir->jsop() when it is present.
static inline Assembler::Condition JSOpToCondition(JSOp op, bool isSigned) {
  if (isSigned) {
    switch (op) {
      case JSOp::Eq:
      case JSOp::StrictEq:
        return Assembler::Equal;
      case JSOp::Ne:
      case JSOp::StrictNe:
        return Assembler::NotEqual;
      case JSOp::Lt:
        return Assembler::LessThan;
      case JSOp::Le:
        return Assembler::LessThanOrEqual;
      case JSOp::Gt:
        return Assembler::GreaterThan;
      case JSOp::Ge:
        return Assembler::GreaterThanOrEqual;
      default:
        MOZ_CRASH("Unrecognized comparison operation");
    }
  } else {
    switch (op) {
      case JSOp::Eq:
      case JSOp::StrictEq:
        return Assembler::Equal;
      case JSOp::Ne:
      case JSOp::StrictNe:
        return Assembler::NotEqual;
      case JSOp::Lt:
        return Assembler::Below;
      case JSOp::Le:
        return Assembler::BelowOrEqual;
      case JSOp::Gt:
        return Assembler::Above;
      case JSOp::Ge:
        return Assembler::AboveOrEqual;
      default:
        MOZ_CRASH("Unrecognized comparison operation");
    }
  }
}

static inline size_t StackDecrementForCall(uint32_t alignment,
                                           size_t bytesAlreadyPushed,
                                           size_t bytesToPush) {
  return bytesToPush +
         ComputeByteAlignment(bytesAlreadyPushed + bytesToPush, alignment);
}

// Helper for generatePreBarrier.
inline DynFn JitPreWriteBarrier(MIRType type);
}  // namespace jit

}  // namespace js

#endif /* jit_MacroAssembler_h */