splicer 2.3.0

Plan and generate middleware splice operations for WebAssembly component composition graphs.
Documentation
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
//! Codegen: walk a [`LiftPlan`] and emit the wasm that writes one
//! cell per (param | result) into the cells slab, plus the result-
//! lift emission for Direct (sync flat) and Compound result kinds.

use wasm_encoder::{BlockType, Function, Instruction, MemArg, ValType};
use wit_bindgen_core::abi::lift_from_memory;
use wit_parser::{Resolve, SizeAlign};

use super::super::super::abi::cast;
use super::super::super::abi::emit::{
    direct_return_type, emit_bitcast, emit_cabi_realloc_call_runtime, wasm_type_to_val, BlobSlice,
    RecordLayout, I32_STORE_LOG2_ALIGN, I64_STORE_LOG2_ALIGN, I8_STORE_LOG2_ALIGN, MAX_UTF8_LEN,
    OPTION_NONE, OPTION_SOME, SLICE_LEN_OFFSET, SLICE_PTR_OFFSET, STRING_FLAT_BYTES,
};
use super::super::super::abi::flat_types;
use super::super::super::abi::WasmEncoderBindgen;
use super::super::super::indices::{FrozenLocals, LocalsBuilder};
use super::super::cells::{CellLayout, PayloadSource};
use super::super::FuncDispatch;
use super::classify::{InfoCounts, ResultSourceLayout};
use super::plan::{ArmGuard, Cell, LiftPlan, ListElementClass, ListSpec};
use super::sidetable::flags_info::FlagsRuntimeFill;
use super::sidetable::handle_info::HandleRuntimeFill;
use super::sidetable::record_info::RecordRuntimeFill;
use super::sidetable::variant_info::VariantRuntimeFill;
use super::sidetable::{CellSideData, CharScratch, TupleIdxSource};
use wit_parser::abi::WasmType;

/// Wrapper-body locals, allocated up front so all downstream emit
/// phases share the same indices. Result-lift-only locals live on
/// [`ResultEmitPlan`].
///
/// **Stage-then-consume invariant** (applies to the `Option<u32>`
/// staging slots: `tuple_slot_ptr`, `list_elem_handle_base`,
/// `list_elem_flags_base`, `list_elem_record_base`,
/// `list_elem_variant_base`, `handle_info_base`, etc.): each is a
/// single shared local set immediately before its consumer emits.
/// Interleaving any emit between stage and consume silently clobbers
/// it. Future refactors that reorder must keep stage→consume
/// contiguous, or split into one local per consumer.
pub(crate) struct WrapperLocals {
    /// Scratch for the cell write address.
    pub addr: u32,
    /// Packed status from canon-async hook calls.
    pub st: u32,
    /// Waitable-set handle for the wait loop.
    pub ws: u32,
    /// i64 widening source for IntegerSignExt/ZeroExt.
    pub(super) ext64: u32,
    /// f64 promoted source for FloatingF32.
    pub(super) ext_f64: u32,
    /// Scratch i32 locals for joined-flat widening reads. `_a` lands
    /// the bitcast for any i32-arm leaf; `_b` is reserved for the
    /// second slot of `Cell::Text` / `Cell::Bytes`. Liveness analysis
    /// drops them when unused.
    pub(super) widen_i32_a: u32,
    pub(super) widen_i32_b: u32,
    /// Scratch f32 local for joined-flat F32 widening — rare (only
    /// when an F32 leaf shares a joined slot with a wider arm).
    pub(super) widen_f32: u32,
    /// `Cell::Flags` bit-walk cursor + count.
    pub(super) flags_addr: u32,
    pub(super) flags_count: u32,
    /// `Cell::Char` utf-8 encoder locals. Both `Some` iff any
    /// `Cell::Char` (top-level or list-element).
    pub(super) char_len: Option<u32>,
    pub(super) char_scratch_addr: Option<u32>,
    /// Staging slot for list-element child cell-array indices
    /// (Option/Result payloads in `PrestagedChildIdx` element class).
    pub(super) list_elem_child_idx: Option<u32>,
    /// Staging slot for list-element TupleOf cells' per-call
    /// indices-buffer slot ptr. Stage-then-consume.
    pub(super) tuple_slot_ptr: Option<u32>,
    /// Direct-return value when the export sig has a single flat result.
    pub result: Option<u32>,
    /// Address local for async `task.return` flat loads. `None` for
    /// sync, void async, and async with retptr-passthrough.
    pub tr_addr: Option<u32>,
    /// i64 call-id local. Tier-2 always wires at least one hook.
    pub id_local: u32,
    /// Pre-built bindgen load sequence for async `task.return`.
    /// Stored here so every local the bindgen needed is in `FrozenLocals`.
    pub task_return_loads: Option<Vec<Instruction<'static>>>,
    /// Pre-built bindgen lower for async indirect_params. Same
    /// FrozenLocals rationale as `task_return_loads`.
    pub params_lower_seq: Option<Vec<Instruction<'static>>>,
    /// Bump snapshot at wrapper entry; restored at exit.
    pub saved_bump: u32,
    /// Active plan's cells slab base; rewritten per plan.
    pub cells_base: u32,
    /// Running cell-index counter; holds `total_cells` after pre-pass.
    pub next_cell_idx: u32,
    /// Per-iteration `handle_slot_base + j * handles_per_elem`. Stage-
    /// then-consume; `Some` iff [`fn_has_list_elem_handle`].
    pub list_elem_handle_base: Option<u32>,
    /// Per-iteration `flags_slot_base + j * flags_per_elem` + scratch
    /// base. Stage-then-consume; `Some` iff [`fn_has_list_elem_flags`].
    pub list_elem_flags_base: Option<u32>,
    pub list_elem_flags_scratch_base: Option<u32>,
    /// Runtime entry-address scratch (list-element flags only).
    pub flags_slot_addr: Option<u32>,
    /// Runtime cell-payload idx scratch.
    pub flags_payload_idx: Option<u32>,
    /// Runtime slot byte-address scratch (list-element handles only).
    pub handle_slot_addr: Option<u32>,
    /// Runtime cell-payload idx (list-element handles, non-zero offset).
    pub handle_payload_idx: Option<u32>,
    /// Running handle-info entry count: starts at static count, bumps
    /// by `len * handles_per_elem` per list. Sizes `cabi_realloc` +
    /// patches `handle_infos.len`. `Some` iff list-elem-handle present.
    pub next_handle_idx: Option<u32>,
    pub next_flags_idx: Option<u32>,
    pub next_record_idx: Option<u32>,
    /// Per-iteration `record_slot_base + j * records_per_elem` +
    /// tuples sub-region base. Stage-then-consume.
    pub list_elem_record_base: Option<u32>,
    pub list_elem_record_tuples_base: Option<u32>,
    pub record_slot_addr: Option<u32>,
    pub record_payload_idx: Option<u32>,
    /// Runtime per-record field-tuples sub-slice base; reused across
    /// per-tuple field writes.
    pub record_tuples_slice_addr: Option<u32>,
    /// Active plan's record-info buffer base. Stage-then-consume.
    pub record_info_base: Option<u32>,
    /// Active plan's variant-info buffer base. Stage-then-consume.
    pub variant_info_base: Option<u32>,
    pub next_variant_idx: Option<u32>,
    pub list_elem_variant_base: Option<u32>,
    pub variant_slot_addr: Option<u32>,
    pub variant_payload_idx: Option<u32>,
    /// Active plan's flags-info buffer base. Stage-then-consume.
    pub flags_info_base: Option<u32>,
    /// Active plan's handle-info buffer base. Stage-then-consume —
    /// load-bearing on (1) only `Cell::Handle` reads it and (2) per-
    /// plan alloc immediately precedes the plan's lift. A future
    /// reader or interleaved fill must restage or split per plan.
    pub handle_info_base: Option<u32>,
    /// Per-param list emit locals; `param_list_locals[i]` parallels
    /// `params[i].lift.plan.list_specs()`.
    pub param_list_locals: Vec<Vec<ListEmitLocals>>,
}

/// Result-side lift bundle. Direct carries side-data inline;
/// Compound borrows it from the layout phase.
pub(crate) enum ResultEmitPlan<'a> {
    /// Void or unsupported result: no lift fires.
    None,
    /// Sync flat return — source already in `source_local`.
    Direct {
        cell: Cell,
        source_local: u32,
        side_data: CellSideData,
    },
    /// Retptr-loaded result. `addr_local` drives the
    /// `lift_from_memory`-built `loads` sequence; the wrapper
    /// `local.set`s values into `synth_locals` (LIFO), with
    /// `local_base = synth_locals[0]`.
    Compound {
        plan: &'a LiftPlan,
        retptr_offset: i32,
        addr_local: u32,
        synth_locals: Vec<u32>,
        loads: Vec<Instruction<'static>>,
        side_refs: CellSideRefs<'a>,
        /// Per-list emit locals, parallel to `plan.list_specs()`.
        list_locals: Vec<ListEmitLocals>,
    },
}

/// Per-plan-cell side-table data borrowed off `ParamLayout` /
/// `ResultSourceLayout::Compound`. One entry per cell.
#[derive(Clone, Copy)]
pub(crate) struct CellSideRefs<'a> {
    pub cell_side: &'a [CellSideData],
}

/// Per-build context shared across every lift emit. Bundles
/// `cell_layout` + `cabi_realloc_idx` + per-call buffer geometries
/// so per-cell helpers don't pay `offset_of` per emit.
#[derive(Clone, Copy)]
pub(crate) struct LiftEmitCtx<'a> {
    pub cell_layout: &'a CellLayout,
    pub cabi_realloc_idx: u32,
    pub handle_info: HandleInfoOffsets,
    pub flags_info: FlagsInfoOffsets,
    pub record_info: RecordInfoOffsets,
    pub variant_info: VariantInfoOffsets,
}

/// Build-time-resolved geometry of one `record handle-info` entry.
#[derive(Clone, Copy)]
pub(crate) struct HandleInfoOffsets {
    pub entry_size: u32,
    pub align: u32,
    pub type_name_off: u32,
    pub id_off: u32,
}

impl HandleInfoOffsets {
    pub(crate) fn from_layout(layout: &RecordLayout) -> Self {
        use super::super::schema::{HANDLE_INFO_ID, HANDLE_INFO_TYPE_NAME};
        Self {
            entry_size: layout.size,
            align: layout.align,
            type_name_off: layout.offset_of(HANDLE_INFO_TYPE_NAME),
            id_off: layout.offset_of(HANDLE_INFO_ID),
        }
    }
}

/// Build-time-resolved geometry of one `record flags-info` entry.
/// Same shape as [`HandleInfoOffsets`] for the flags-info side.
#[derive(Clone, Copy)]
pub(crate) struct FlagsInfoOffsets {
    pub entry_size: u32,
    pub align: u32,
    pub type_name_off: u32,
    pub set_flags_off: u32,
}

impl FlagsInfoOffsets {
    pub(crate) fn from_layout(layout: &RecordLayout) -> Self {
        use super::super::schema::FLAGS_INFO_SET_FLAGS;
        use super::sidetable::INFO_TYPE_NAME;
        Self {
            entry_size: layout.size,
            align: layout.align,
            type_name_off: layout.offset_of(INFO_TYPE_NAME),
            set_flags_off: layout.offset_of(FLAGS_INFO_SET_FLAGS),
        }
    }
}

/// Build-time-resolved geometry of one `record variant-info` entry,
/// including the `option<u32>` payload's value-byte sub-offset.
/// `payload_off` lands the option-disc byte; `payload_off +
/// payload_value_off` lands the u32 value slot.
#[derive(Clone, Copy)]
pub(crate) struct VariantInfoOffsets {
    pub entry_size: u32,
    pub align: u32,
    pub type_name_off: u32,
    pub case_name_off: u32,
    pub payload_off: u32,
    pub payload_value_off: u32,
}

impl VariantInfoOffsets {
    /// `payload_value_off` is a separate arg because `payload` is
    /// `option<u32>`, not a record — `RecordLayout::offset_of` can't
    /// reach into it. Sourced from `option_payload_offset`.
    pub(crate) fn from_layout(layout: &RecordLayout, payload_value_off: u32) -> Self {
        use super::super::schema::{VARIANT_INFO_CASE_NAME, VARIANT_INFO_PAYLOAD};
        use super::sidetable::INFO_TYPE_NAME;
        Self {
            entry_size: layout.size,
            align: layout.align,
            type_name_off: layout.offset_of(INFO_TYPE_NAME),
            case_name_off: layout.offset_of(VARIANT_INFO_CASE_NAME),
            payload_off: layout.offset_of(VARIANT_INFO_PAYLOAD),
            payload_value_off,
        }
    }
}

/// Build-time-resolved geometry of one `record record-info` entry +
/// the inner `tuple<string, u32>` field-tuple shape that
/// `record-info.fields` points at. Bundled so list-element
/// `emit_record_runtime_fill` can write per-iteration tuples without
/// re-resolving offsets. The tuple geometry is also constant across
/// every call site, so the layout pays a single `offset_of` walk in
/// `from_layout`.
#[derive(Clone, Copy)]
pub(crate) struct RecordInfoOffsets {
    pub entry_size: u32,
    pub align: u32,
    pub type_name_off: u32,
    pub fields_off: u32,
    pub tuple_size: u32,
    pub tuple_align: u32,
    pub tuple_name_off: u32,
    pub tuple_idx_off: u32,
}

impl RecordInfoOffsets {
    pub(crate) fn from_layout(layout: &RecordLayout, tuple_layout: &RecordLayout) -> Self {
        use super::super::schema::{
            RECORD_FIELD_TUPLE_IDX, RECORD_FIELD_TUPLE_NAME, RECORD_INFO_FIELDS,
        };
        use super::sidetable::INFO_TYPE_NAME;
        Self {
            entry_size: layout.size,
            align: layout.align,
            type_name_off: layout.offset_of(INFO_TYPE_NAME),
            fields_off: layout.offset_of(RECORD_INFO_FIELDS),
            tuple_size: tuple_layout.size,
            tuple_align: tuple_layout.align,
            tuple_name_off: tuple_layout.offset_of(RECORD_FIELD_TUPLE_NAME),
            tuple_idx_off: tuple_layout.offset_of(RECORD_FIELD_TUPLE_IDX),
        }
    }
}

/// Per-plan-walk cursor: the plan + the wrapper-local offset added to
/// each cell's plan-relative flat slot. `elem_cell_base` is `Some`
/// inside list-element bodies — Option/Result resolve runtime idx as
/// `elem_cell_base + relative_idx`; top-level walks use build-time-
/// known absolute idx.
#[derive(Clone, Copy)]
pub(crate) struct PlanCursor<'a> {
    pub plan: &'a LiftPlan,
    pub local_base: u32,
    pub elem_cell_base: Option<u32>,
}

/// Per-`Cell::ListOf` emit-time bundle. One entry per list-of cell;
/// parallel to [`LiftPlan::list_specs`].
pub(crate) struct ListEmitLocals {
    /// Cell-idx where this list's element cells begin.
    pub start_i: u32,
    /// Captured source `len` flat slot value.
    pub len: u32,
    /// Per-call indices buffer base (`len * 4` bytes).
    pub indices_ptr: u32,
    /// Element-loop counter (0..len).
    pub j: u32,
    /// Per-iter source element address; drives `elem_loads`.
    pub elem_addr: u32,
    /// One local per element-plan flat slot, contiguous so plan slot
    /// N maps to `elem_flat_locals[0] + N`.
    pub elem_flat_locals: Vec<u32>,
    /// Pre-built `lift_from_memory` loads — pushes element flat values
    /// for capture into `elem_flat_locals` (LIFO).
    pub elem_loads: Vec<Instruction<'static>>,
    /// Canonical-ABI byte size of one element.
    pub elem_byte_size: u32,
    /// Side-data parallel to `element_plan.cells`.
    pub elem_cell_side: Vec<CellSideData>,
    /// Per-call utf-8 scratch for `Cell::Char` element cells. The k-th
    /// char in iteration j lives at `(j * chars_per_elem + k) * MAX_UTF8_LEN`.
    pub char_scratch_base: Option<u32>,
    pub chars_per_elem: u32,
    /// Per-call buffer for `Cell::TupleOf` element cells —
    /// `len * tuple_idx_count_per_elem * 4` bytes.
    pub tuple_idx_buf_base: Option<u32>,
    pub tuple_idx_count_per_elem: u32,
    /// Per-iteration cell-array base (`start_i + j*elem_count`).
    /// Always allocated — saves 1–2 insns per cell per iter vs recomputing.
    pub elem_cell_base: u32,
    /// Handle-info buffer starting slot for this list's elements.
    pub handle_slot_base: Option<u32>,
    pub handles_per_elem: u32,
    pub flags_slot_base: Option<u32>,
    pub flags_per_elem: u32,
    /// Per-call set-flags scratch buffer base; sized at
    /// `len * flags_scratch_bytes_per_elem`.
    pub flags_scratch_buf_base: Option<u32>,
    /// `Σ info.item_names.len() * STRING_FLAT_BYTES` across element-
    /// plan `Cell::Flags` cells — variable per cell, so each carries
    /// its own `scratch_offset_in_elem`.
    pub flags_scratch_bytes_per_elem: u32,
    pub record_slot_base: Option<u32>,
    pub records_per_elem: u32,
    /// Per-call field-tuples scratch base; sized at
    /// `len * record_tuples_bytes_per_elem`.
    pub record_tuples_buf_base: Option<u32>,
    /// `Σ fields.len() * tuple_size` across element-plan `Cell::RecordOf`
    /// cells — variable per cell.
    pub record_tuples_bytes_per_elem: u32,
    pub variant_slot_base: Option<u32>,
    pub variants_per_elem: u32,
}

/// Allocate per-list emit locals + pre-build the `lift_from_memory`
/// loads for every `Cell::ListOf` in `plan`. Runs while the builder
/// is live (bindgen may allocate scratch locals).
pub(super) fn alloc_list_emit_locals(
    plan: &LiftPlan,
    resolve: &Resolve,
    size_align: &SizeAlign,
    record_tuple_size: u32,
    builder: &mut LocalsBuilder,
) -> Vec<ListEmitLocals> {
    plan.list_specs()
        .map(|spec: ListSpec<'_>| {
            build_one_list_emit_locals(spec, resolve, size_align, record_tuple_size, builder)
        })
        .collect()
}

fn build_one_list_emit_locals(
    spec: ListSpec<'_>,
    resolve: &Resolve,
    size_align: &SizeAlign,
    record_tuple_size: u32,
    builder: &mut LocalsBuilder,
) -> ListEmitLocals {
    let start_i = builder.alloc_local(ValType::I32);
    let len = builder.alloc_local(ValType::I32);
    let indices_ptr = builder.alloc_local(ValType::I32);
    let j = builder.alloc_local(ValType::I32);
    let elem_addr = builder.alloc_local(ValType::I32);
    // Contiguous flat-slot locals: plan slot N → elem_flat_locals[0] + N.
    let elem_ty = spec.element_plan.source_ty;
    let flat = flat_types(resolve, &elem_ty, None)
        .expect("list element type must flatten within MAX_FLAT_PARAMS");
    let elem_flat_locals: Vec<u32> = flat
        .iter()
        .map(|wt| builder.alloc_local(wasm_type_to_val(*wt)))
        .collect();
    debug_assert!(
        elem_flat_locals.windows(2).all(|w| w[1] == w[0] + 1),
        "elem_flat_locals must be contiguous (plan slot N = elem_flat_locals[0] + N)",
    );
    let mut bindgen = WasmEncoderBindgen::new(size_align, elem_addr, builder);
    lift_from_memory(resolve, &mut bindgen, (), &elem_ty);
    let elem_loads = bindgen.into_instructions();
    let elem_byte_size = size_align.size(&elem_ty).size_wasm32() as u32;
    let (elem_cell_side, counts) = walk_element_plan(spec.element_plan, record_tuple_size);
    let char_scratch_base = (counts.chars > 0).then(|| builder.alloc_local(ValType::I32));
    let tuple_idx_buf_base =
        (counts.tuple_idx_slots > 0).then(|| builder.alloc_local(ValType::I32));
    let chars_per_elem = counts.chars;
    let tuple_idx_count_per_elem = counts.tuple_idx_slots;
    let handle_slot_base = (counts.handles > 0).then(|| builder.alloc_local(ValType::I32));
    let handles_per_elem = counts.handles;
    let flags_slot_base = (counts.flags > 0).then(|| builder.alloc_local(ValType::I32));
    let flags_scratch_buf_base =
        (counts.flags_scratch_bytes > 0).then(|| builder.alloc_local(ValType::I32));
    let flags_per_elem = counts.flags;
    let flags_scratch_bytes_per_elem = counts.flags_scratch_bytes;
    let record_slot_base = (counts.records > 0).then(|| builder.alloc_local(ValType::I32));
    let records_per_elem = counts.records;
    let record_tuples_buf_base =
        (counts.record_tuples_bytes > 0).then(|| builder.alloc_local(ValType::I32));
    let record_tuples_bytes_per_elem = counts.record_tuples_bytes;
    let variant_slot_base = (counts.variants > 0).then(|| builder.alloc_local(ValType::I32));
    let variants_per_elem = counts.variants;
    let elem_cell_base = builder.alloc_local(ValType::I32);
    ListEmitLocals {
        start_i,
        len,
        indices_ptr,
        j,
        elem_addr,
        elem_flat_locals,
        elem_loads,
        elem_byte_size,
        elem_cell_side,
        char_scratch_base,
        chars_per_elem,
        tuple_idx_buf_base,
        tuple_idx_count_per_elem,
        elem_cell_base,
        handle_slot_base,
        handles_per_elem,
        flags_slot_base,
        flags_per_elem,
        flags_scratch_buf_base,
        flags_scratch_bytes_per_elem,
        record_slot_base,
        records_per_elem,
        record_tuples_buf_base,
        record_tuples_bytes_per_elem,
        variant_slot_base,
        variants_per_elem,
    }
}

/// Whether any list-element cell across the wrapper's plans has the
/// given [`ListElementClass`]. Gates shared wrapper locals.
fn fn_has_list_elem_class(fd: &FuncDispatch, want: ListElementClass) -> bool {
    let plan_has = |plan: &LiftPlan| {
        plan.list_specs().any(|spec| {
            spec.element_plan
                .cells
                .iter()
                .any(|c| c.list_element_class() == Some(want))
        })
    };
    if fd.params.iter().any(|p| plan_has(&p.lift.plan)) {
        return true;
    }
    match fd.result_lift.as_ref().map(|rl| &rl.source) {
        Some(ResultSourceLayout::Compound { compound, .. }) => plan_has(&compound.plan),
        _ => false,
    }
}

fn fn_has_list_elem_child_idx(fd: &FuncDispatch) -> bool {
    fn_has_list_elem_class(fd, ListElementClass::PrestagedChildIdx)
}

fn fn_has_list_elem_tuple(fd: &FuncDispatch) -> bool {
    fn_has_list_elem_class(fd, ListElementClass::PrestagedTupleIndices)
}

fn fn_has_list_elem_handle(fd: &FuncDispatch) -> bool {
    fn_has_list_elem_class(fd, ListElementClass::PrestagedHandle)
}

fn fn_has_list_elem_flags(fd: &FuncDispatch) -> bool {
    fn_has_list_elem_class(fd, ListElementClass::PrestagedFlags)
}

fn fn_has_list_elem_record(fd: &FuncDispatch) -> bool {
    fn_has_list_elem_class(fd, ListElementClass::PrestagedRecord)
}

fn fn_has_list_elem_variant(fd: &FuncDispatch) -> bool {
    fn_has_list_elem_class(fd, ListElementClass::PrestagedVariant)
}

/// Any `Cell::Char` in the wrapper. Gates `char_len` + `char_scratch_addr`.
fn fn_contains_char(fd: &FuncDispatch) -> bool {
    if fd.params.iter().any(|p| p.lift.plan.contains_char()) {
        return true;
    }
    if let Some(rl) = fd.result_lift.as_ref() {
        match &rl.source {
            ResultSourceLayout::Direct { cell, .. } => matches!(cell, Cell::Char { .. }),
            ResultSourceLayout::Compound { compound, .. } => compound.plan.contains_char(),
        }
    } else {
        false
    }
}

/// Any cell of `kind` in the wrapper. Gates the matching
/// `WrapperLocals.*_info_base`.
fn fn_has_info_cells(
    fd: &FuncDispatch,
    kind: ListElementClass,
    count: fn(&InfoCounts) -> u32,
) -> bool {
    fn_has_list_elem_class(fd, kind)
        || fd.params.iter().any(|p| count(&p.info_counts) > 0)
        || fd
            .result_lift
            .as_ref()
            .is_some_and(|rl| count(&rl.info_counts) > 0)
}

fn fn_has_handle_cells(fd: &FuncDispatch) -> bool {
    fn_has_info_cells(fd, ListElementClass::PrestagedHandle, |c| c.handle)
}
fn fn_has_flags_cells(fd: &FuncDispatch) -> bool {
    fn_has_info_cells(fd, ListElementClass::PrestagedFlags, |c| c.flags)
}
fn fn_has_record_cells(fd: &FuncDispatch) -> bool {
    fn_has_info_cells(fd, ListElementClass::PrestagedRecord, |c| c.record)
}
fn fn_has_variant_cells(fd: &FuncDispatch) -> bool {
    fn_has_info_cells(fd, ListElementClass::PrestagedVariant, |c| c.variant)
}

/// Per-class counts driving per-list allocation decisions.
#[derive(Default, Clone, Copy, Debug, PartialEq, Eq)]
pub(super) struct ElementCounts {
    pub chars: u32,
    /// Sum of `children.len()` across `Cell::TupleOf` cells.
    pub tuple_idx_slots: u32,
    pub handles: u32,
    pub flags: u32,
    /// Total set-flags scratch bytes across `Cell::Flags` cells —
    /// variable-stride per cell.
    pub flags_scratch_bytes: u32,
    pub records: u32,
    /// Total field-tuples bytes across `Cell::RecordOf` cells —
    /// variable-stride per cell.
    pub record_tuples_bytes: u32,
    pub variants: u32,
}

/// Single walk over `element_plan.cells` producing per-cell side
/// data + per-class counts. Driven off [`Cell::list_element_class`] so
/// adding a class forces a fold arm at compile time.
pub(super) fn walk_element_plan(
    element_plan: &LiftPlan,
    record_tuple_size: u32,
) -> (Vec<CellSideData>, ElementCounts) {
    let mut counts = ElementCounts::default();
    let side: Vec<CellSideData> = element_plan
        .cells
        .iter()
        .map(|cell| {
            let class = cell.list_element_class().unwrap_or_else(|| {
                unreachable!(
                    "Cell {cell:?} reached walk_element_plan despite \
                     Cell::list_element_class() rejecting it"
                )
            });
            match class {
                ListElementClass::Scalar => CellSideData::None,
                ListElementClass::PrestagedChar => {
                    counts.chars += 1;
                    CellSideData::Char {
                        scratch: CharScratch::Prestaged,
                    }
                }
                // Option/Result resolve child idx via PlanCursor.elem_cell_base.
                ListElementClass::PrestagedChildIdx => CellSideData::None,
                ListElementClass::PrestagedTupleIndices => {
                    let Cell::TupleOf { children } = cell else {
                        unreachable!("PrestagedTupleIndices class on non-TupleOf {cell:?}")
                    };
                    let off = counts.tuple_idx_slots * 4;
                    counts.tuple_idx_slots += children.len() as u32;
                    CellSideData::Tuple {
                        source: TupleIdxSource::PerIteration {
                            offset_in_elem: off,
                        },
                    }
                }
                ListElementClass::PrestagedHandle => {
                    let Cell::Handle { type_name, .. } = cell else {
                        unreachable!("PrestagedHandle class on non-Handle {cell:?}")
                    };
                    let offset_in_elem = counts.handles;
                    counts.handles += 1;
                    CellSideData::Handle(Box::new(
                        super::sidetable::handle_info::HandleRuntimeFill {
                            slot_source:
                                super::sidetable::handle_info::HandleSlotSource::PerIteration {
                                    offset_in_elem,
                                },
                            type_name: *type_name,
                        },
                    ))
                }
                ListElementClass::PrestagedFlags => {
                    let Cell::Flags {
                        type_name,
                        flag_names,
                        ..
                    } = cell
                    else {
                        unreachable!("PrestagedFlags class on non-Flags {cell:?}")
                    };
                    let entry_offset_in_elem = counts.flags;
                    let scratch_offset_in_elem = counts.flags_scratch_bytes;
                    counts.flags += 1;
                    counts.flags_scratch_bytes += flag_names.len() as u32 * STRING_FLAT_BYTES;
                    CellSideData::Flags(Box::new(super::sidetable::flags_info::FlagsRuntimeFill {
                        slot_source: super::sidetable::flags_info::FlagsSlotSource::PerIteration {
                            entry_offset_in_elem,
                            scratch_offset_in_elem,
                        },
                        type_name: *type_name,
                        flag_names: flag_names.clone(),
                    }))
                }
                ListElementClass::PrestagedRecord => {
                    let Cell::RecordOf { type_name, fields } = cell else {
                        unreachable!("PrestagedRecord class on non-RecordOf {cell:?}")
                    };
                    let entry_offset_in_elem = counts.records;
                    let tuples_offset_in_elem = counts.record_tuples_bytes;
                    counts.records += 1;
                    counts.record_tuples_bytes += fields.len() as u32 * record_tuple_size;
                    CellSideData::Record(Box::new(
                        super::sidetable::record_info::RecordRuntimeFill {
                            slot_source:
                                super::sidetable::record_info::RecordSlotSource::PerIteration {
                                    entry_offset_in_elem,
                                    tuples_offset_in_elem,
                                },
                            type_name: *type_name,
                            fields_len: fields.len() as u32,
                        },
                    ))
                }
                ListElementClass::PrestagedVariant => {
                    let Cell::Variant {
                        type_name,
                        case_names,
                        per_case_payload,
                        ..
                    } = cell
                    else {
                        unreachable!("PrestagedVariant class on non-Variant {cell:?}")
                    };
                    let entry_offset_in_elem = counts.variants;
                    counts.variants += 1;
                    CellSideData::Variant(Box::new(
                        super::sidetable::variant_info::VariantRuntimeFill {
                            slot_source:
                                super::sidetable::variant_info::VariantSlotSource::PerIteration {
                                    entry_offset_in_elem,
                                },
                            type_name: *type_name,
                            case_names: case_names.clone(),
                            per_case_payload: per_case_payload.clone(),
                        },
                    ))
                }
            }
        })
        .collect();
    (side, counts)
}

/// Allocate every wrapper-body local + build compound-result and
/// task-return load sequences, then freeze the locals list. Taking
/// `builder` by value is the typestate hinge: post-freeze allocation
/// is a compile error rather than a runtime trap.
pub(crate) fn alloc_wrapper_locals<'a>(
    resolve: &Resolve,
    size_align: &SizeAlign,
    record_tuple_size: u32,
    mut builder: LocalsBuilder,
    fd: &'a FuncDispatch,
    func: &wit_parser::Function,
) -> (WrapperLocals, ResultEmitPlan<'a>, FrozenLocals) {
    let addr = builder.alloc_local(ValType::I32);
    let st = builder.alloc_local(ValType::I32);
    let ws = builder.alloc_local(ValType::I32);
    let ext64 = builder.alloc_local(ValType::I64);
    let ext_f64 = builder.alloc_local(ValType::F64);
    let widen_i32_a = builder.alloc_local(ValType::I32);
    let widen_i32_b = builder.alloc_local(ValType::I32);
    let widen_f32 = builder.alloc_local(ValType::F32);
    let flags_addr = builder.alloc_local(ValType::I32);
    let flags_count = builder.alloc_local(ValType::I32);
    let needs_char_locals = fn_contains_char(fd);
    let char_len = needs_char_locals.then(|| builder.alloc_local(ValType::I32));
    let char_scratch_addr = needs_char_locals.then(|| builder.alloc_local(ValType::I32));
    let list_elem_child_idx =
        fn_has_list_elem_child_idx(fd).then(|| builder.alloc_local(ValType::I32));
    let tuple_slot_ptr = fn_has_list_elem_tuple(fd).then(|| builder.alloc_local(ValType::I32));
    let needs_list_handle_locals = fn_has_list_elem_handle(fd);
    let list_elem_handle_base = needs_list_handle_locals.then(|| builder.alloc_local(ValType::I32));
    let handle_slot_addr = needs_list_handle_locals.then(|| builder.alloc_local(ValType::I32));
    let handle_payload_idx = needs_list_handle_locals.then(|| builder.alloc_local(ValType::I32));
    let needs_list_flags_locals = fn_has_list_elem_flags(fd);
    let list_elem_flags_base = needs_list_flags_locals.then(|| builder.alloc_local(ValType::I32));
    let list_elem_flags_scratch_base =
        needs_list_flags_locals.then(|| builder.alloc_local(ValType::I32));
    let flags_slot_addr = needs_list_flags_locals.then(|| builder.alloc_local(ValType::I32));
    let flags_payload_idx = needs_list_flags_locals.then(|| builder.alloc_local(ValType::I32));
    let needs_list_record_locals = fn_has_list_elem_record(fd);
    let list_elem_record_base = needs_list_record_locals.then(|| builder.alloc_local(ValType::I32));
    let list_elem_record_tuples_base =
        needs_list_record_locals.then(|| builder.alloc_local(ValType::I32));
    let record_slot_addr = needs_list_record_locals.then(|| builder.alloc_local(ValType::I32));
    let record_payload_idx = needs_list_record_locals.then(|| builder.alloc_local(ValType::I32));
    let record_tuples_slice_addr =
        needs_list_record_locals.then(|| builder.alloc_local(ValType::I32));
    let needs_list_variant_locals = fn_has_list_elem_variant(fd);
    let list_elem_variant_base =
        needs_list_variant_locals.then(|| builder.alloc_local(ValType::I32));
    let variant_slot_addr = needs_list_variant_locals.then(|| builder.alloc_local(ValType::I32));
    let variant_payload_idx = needs_list_variant_locals.then(|| builder.alloc_local(ValType::I32));
    let result = direct_return_type(&fd.export_sig).map(|t| builder.alloc_local(t));
    // Non-retptr-passthrough async task.return: i32 addr drives
    // `lift_from_memory` flat-load out of the retptr scratch.
    let tr_uses_flat_loads = fd
        .shape
        .task_return()
        .is_some_and(|tr| !tr.sig.indirect_params && fd.result_ty.is_some());
    let tr_addr = tr_uses_flat_loads.then(|| builder.alloc_local(ValType::I32));

    // Compound: extra locals + bindgen-driven `lift_from_memory` may
    // allocate scratch locals — must run before freeze.
    let result_emit = match fd.result_lift.as_ref() {
        None => ResultEmitPlan::None,
        Some(rl) => match &rl.source {
            ResultSourceLayout::Direct { cell, side_data } => ResultEmitPlan::Direct {
                cell: cell.clone(),
                source_local: result
                    .expect("ResultSourceLayout::Direct → direct-return local allocated"),
                side_data: side_data.clone(),
            },
            ResultSourceLayout::Compound {
                compound,
                retptr_offset,
                cell_side,
            } => {
                let side_refs = CellSideRefs { cell_side };
                let addr_local = builder.alloc_local(ValType::I32);
                let flat = flat_types(resolve, &compound.ty, None).unwrap_or_else(|| {
                    panic!(
                        "Compound result must flatten within MAX_FLAT_PARAMS ({}) — \
                             classify_result_lift only returns Compound for kinds that do",
                        Resolve::MAX_FLAT_PARAMS
                    )
                });
                assert_eq!(
                    flat.len(),
                    compound.plan.flat_slot_count as usize,
                    "canonical-ABI flat count (emit) must match classify-time plan"
                );
                // Contiguous synth locals: cell N's flat slot
                // resolves to `synth_locals[0] + N = synth_locals[N]`.
                let synth_locals: Vec<u32> = flat
                    .into_iter()
                    .map(|wt| builder.alloc_local(wasm_type_to_val(wt)))
                    .collect();
                debug_assert!(
                    synth_locals.windows(2).all(|w| w[1] == w[0] + 1),
                    "synth_locals must be contiguous (plan slot N = synth_locals[0] + N)",
                );
                let mut bindgen = WasmEncoderBindgen::new(size_align, addr_local, &mut builder);
                lift_from_memory(resolve, &mut bindgen, (), &compound.ty);
                let loads = bindgen.into_instructions();
                let list_locals = alloc_list_emit_locals(
                    &compound.plan,
                    resolve,
                    size_align,
                    record_tuple_size,
                    &mut builder,
                );
                ResultEmitPlan::Compound {
                    plan: &compound.plan,
                    retptr_offset: *retptr_offset,
                    addr_local,
                    synth_locals,
                    loads,
                    side_refs,
                    list_locals,
                }
            }
        },
    };

    // Must allocate before freeze; empty inner Vec for list-free params.
    let param_list_locals: Vec<Vec<ListEmitLocals>> = fd
        .params
        .iter()
        .map(|p| {
            alloc_list_emit_locals(
                &p.lift.plan,
                resolve,
                size_align,
                record_tuple_size,
                &mut builder,
            )
        })
        .collect();

    // Second `lift_from_memory` pass; must run before freeze.
    let task_return_loads: Option<Vec<Instruction<'static>>> = tr_addr.map(|addr_local| {
        let result_ty = fd
            .result_ty
            .as_ref()
            .expect("flat task.return loads → result_ty");
        let mut bindgen = WasmEncoderBindgen::new(size_align, addr_local, &mut builder);
        lift_from_memory(resolve, &mut bindgen, (), result_ty);
        bindgen.into_instructions()
    });

    // Indirect-params lower (async overflowed MAX_FLAT_ASYNC_PARAMS);
    // driven through the same builder so scratch lands in `frozen`.
    let params_lower_seq: Option<Vec<Instruction<'static>>> =
        fd.import_sig.indirect_params.then(|| {
            let base = fd
                .params_record_offset
                .expect("indirect_params → params_record_offset reserved");
            super::super::super::abi::emit::build_lower_params_to_memory(
                resolve,
                size_align,
                &mut builder,
                func,
                base,
            )
        });

    let id_local = builder.alloc_local(ValType::I64);
    let saved_bump = builder.alloc_local(ValType::I32);
    let cells_base = builder.alloc_local(ValType::I32);
    let next_cell_idx = builder.alloc_local(ValType::I32);
    let handle_info_base = fn_has_handle_cells(fd).then(|| builder.alloc_local(ValType::I32));
    let flags_info_base = fn_has_flags_cells(fd).then(|| builder.alloc_local(ValType::I32));
    let record_info_base = fn_has_record_cells(fd).then(|| builder.alloc_local(ValType::I32));
    let variant_info_base = fn_has_variant_cells(fd).then(|| builder.alloc_local(ValType::I32));
    let next_handle_idx = needs_list_handle_locals.then(|| builder.alloc_local(ValType::I32));
    let next_flags_idx = needs_list_flags_locals.then(|| builder.alloc_local(ValType::I32));
    let next_record_idx = needs_list_record_locals.then(|| builder.alloc_local(ValType::I32));
    let next_variant_idx = needs_list_variant_locals.then(|| builder.alloc_local(ValType::I32));

    let frozen = builder.freeze();
    (
        WrapperLocals {
            addr,
            st,
            ws,
            ext64,
            ext_f64,
            widen_i32_a,
            widen_i32_b,
            widen_f32,
            flags_addr,
            flags_count,
            char_len,
            char_scratch_addr,
            list_elem_child_idx,
            tuple_slot_ptr,
            list_elem_handle_base,
            handle_slot_addr,
            handle_payload_idx,
            list_elem_flags_base,
            list_elem_flags_scratch_base,
            flags_slot_addr,
            flags_payload_idx,
            list_elem_record_base,
            list_elem_record_tuples_base,
            record_slot_addr,
            record_payload_idx,
            record_tuples_slice_addr,
            list_elem_variant_base,
            variant_slot_addr,
            variant_payload_idx,
            result,
            tr_addr,
            id_local,
            task_return_loads,
            params_lower_seq,
            saved_bump,
            cells_base,
            next_cell_idx,
            handle_info_base,
            flags_info_base,
            record_info_base,
            variant_info_base,
            next_variant_idx,
            next_handle_idx,
            next_flags_idx,
            next_record_idx,
            param_list_locals,
        },
        result_emit,
        frozen,
    )
}

/// Emit the wasm that lifts one plan into its cells slab. Walks
/// `plan.cells` in allocation order, setting `lcl.addr` per cell and
/// dispatching on the cell variant. `local_base` resolves
/// plan-relative flat slots to absolute wrapper-local indices.
pub(crate) fn emit_lift_plan(
    f: &mut Function,
    ctx: &LiftEmitCtx<'_>,
    plan: &LiftPlan,
    side_refs: CellSideRefs<'_>,
    local_base: u32,
    lcl: &WrapperLocals,
    list_locals: &[ListEmitLocals],
) {
    assert_eq!(
        side_refs.cell_side.len(),
        plan.cells.len(),
        "side-table data (emit input) must have one entry per classify-time plan cell"
    );
    debug_assert_eq!(
        list_locals.len(),
        plan.list_specs().count(),
        "per-plan list_locals must be parallel to plan.list_specs()",
    );
    for (cell_idx, op) in plan.cells.iter().enumerate() {
        f.instructions().local_get(lcl.cells_base);
        if cell_idx > 0 {
            f.instructions()
                .i32_const((cell_idx as u32 * ctx.cell_layout.size) as i32);
            f.instructions().i32_add();
        }
        f.instructions().local_set(lcl.addr);
        let list_slot = match op {
            Cell::ListOf { list_idx, .. } => Some(&list_locals[*list_idx as usize]),
            _ => None,
        };
        emit_cell_op(
            f,
            ctx,
            PlanCursor {
                plan,
                local_base,
                elem_cell_base: None,
            },
            op,
            &side_refs.cell_side[cell_idx],
            lcl,
            list_slot,
        );
    }
}

/// Resolve a leaf-level flat-slot read, applying joined-flat widening
/// bitcast when needed. Returns the absolute wrapper-local — either
/// `local_base + flat_slot` (no widening) or a typed scratch.
fn pin_leaf_flat(
    f: &mut Function,
    plan: &LiftPlan,
    local_base: u32,
    flat_slot: u32,
    arm: WasmType,
    lcl: &WrapperLocals,
) -> u32 {
    pin_leaf_flat_with_i32_scratch(f, plan, local_base, flat_slot, arm, lcl.widen_i32_a, lcl)
}

/// Form of [`pin_leaf_flat`] taking a caller-supplied i32 scratch —
/// only Text / Bytes need this (two i32 slots can both widen).
fn pin_leaf_flat_with_i32_scratch(
    f: &mut Function,
    plan: &LiftPlan,
    local_base: u32,
    flat_slot: u32,
    arm: WasmType,
    scratch_i32: u32,
    lcl: &WrapperLocals,
) -> u32 {
    let Some(joined) = plan.widening_for(flat_slot) else {
        return local_base + flat_slot;
    };
    let bc = cast(joined, arm);
    if matches!(bc, wit_bindgen_core::abi::Bitcast::None) {
        // Another arm widened the slot, but this arm doesn't need it.
        return local_base + flat_slot;
    }
    f.instructions().local_get(local_base + flat_slot);
    emit_bitcast(f, &bc);
    let scratch = match arm {
        WasmType::I32 | WasmType::Pointer | WasmType::Length => scratch_i32,
        WasmType::I64 | WasmType::PointerOrI64 => lcl.ext64,
        WasmType::F32 => lcl.widen_f32,
        WasmType::F64 => lcl.ext_f64,
    };
    f.instructions().local_set(scratch);
    scratch
}

/// Pin both i32 slots of a `Text` / `Bytes` cell into distinct
/// scratches (`widen_i32_a` for ptr, `widen_i32_b` for len) so the
/// ptr value survives the len read. Returns `(ptr_local, len_local)`
/// for the cell-layout helper.
fn pin_text_bytes_slots(
    f: &mut Function,
    plan: &LiftPlan,
    local_base: u32,
    ptr_slot: u32,
    len_slot: u32,
    lcl: &WrapperLocals,
) -> (u32, u32) {
    let ptr = pin_leaf_flat(f, plan, local_base, ptr_slot, WasmType::I32, lcl);
    let len = pin_leaf_flat_with_i32_scratch(
        f,
        plan,
        local_base,
        len_slot,
        WasmType::I32,
        lcl.widen_i32_b,
        lcl,
    );
    (ptr, len)
}

/// `local.get` then (when widening is recorded for `flat_slot`) the
/// joined→arm bitcast — leaves the arm-typed value on the wasm stack.
/// Used for cells that do their own follow-up (extend / promote /
/// `if_`) rather than handing a local index to a helper.
fn push_widened_get(
    f: &mut Function,
    plan: &LiftPlan,
    local_base: u32,
    flat_slot: u32,
    arm: WasmType,
) {
    f.instructions().local_get(local_base + flat_slot);
    if let Some(joined) = plan.widening_for(flat_slot) {
        emit_bitcast(f, &cast(joined, arm));
    }
}

/// Open one `if disc == expected` per guard. Body lands inside the
/// innermost block; pair with [`emit_close_arm_guards`].
fn emit_open_arm_guards(f: &mut Function, plan: &LiftPlan, local_base: u32, guards: &[ArmGuard]) {
    for guard in guards {
        push_widened_get(f, plan, local_base, guard.disc_slot, WasmType::I32);
        f.instructions().i32_const(guard.expected_disc as i32);
        f.instructions().i32_eq();
        f.instructions().if_(BlockType::Empty);
    }
}

/// Close `n` `if` blocks opened by [`emit_open_arm_guards`]. `n`
/// must equal the guard count passed at open or wasm validation
/// will reject the function.
fn emit_close_arm_guards(f: &mut Function, n: usize) {
    for _ in 0..n {
        f.instructions().end();
    }
}

/// Pre-pass: init `lcl.next_cell_idx` to the plan's static cell count,
/// then bump by `len · elem_count` per list (capturing `start_i` and
/// `len`). Joined-arm lists disc-gate the bump so an inactive arm's
/// bytes can't bloat the slab — zero-init keeps `ll.len`/`ll.start_i`
/// defined on the inactive path. Parallel running counters for
/// handle/flags/record/variant-info entries follow the same shape.
pub(crate) fn emit_list_pre_pass(
    f: &mut Function,
    ctx: &LiftEmitCtx<'_>,
    plan: &LiftPlan,
    static_counts: &InfoCounts,
    list_locals: &[ListEmitLocals],
    local_base: u32,
    lcl: &WrapperLocals,
) {
    debug_assert_eq!(
        list_locals.len(),
        plan.list_specs().count(),
        "per-plan list_locals must be parallel to plan.list_specs()",
    );
    f.instructions().i32_const(plan.cell_count() as i32);
    f.instructions().local_set(lcl.next_cell_idx);
    for (next_idx, count) in [
        (lcl.next_handle_idx, static_counts.handle),
        (lcl.next_flags_idx, static_counts.flags),
        (lcl.next_record_idx, static_counts.record),
        (lcl.next_variant_idx, static_counts.variant),
    ] {
        if let Some(next_idx) = next_idx {
            f.instructions().i32_const(count as i32);
            f.instructions().local_set(next_idx);
        }
    }
    for spec in plan.list_specs() {
        let ll = &list_locals[spec.list_idx as usize];
        emit_open_arm_guards(f, plan, local_base, spec.arm_guards);
        f.instructions().local_get(lcl.next_cell_idx);
        f.instructions().local_set(ll.start_i);
        push_widened_get(f, plan, local_base, spec.len_slot, WasmType::I32);
        f.instructions().local_set(ll.len);
        let elem_count = spec.element_plan.cell_count();
        super::super::super::abi::emit::emit_trap_if_list_overflows_cell_slab(
            f,
            ll.len,
            elem_count,
            lcl.next_cell_idx,
            ctx.cell_layout.size,
        );
        f.instructions().local_get(lcl.next_cell_idx);
        f.instructions().local_get(ll.len);
        if elem_count != 1 {
            f.instructions().i32_const(elem_count as i32);
            f.instructions().i32_mul();
        }
        f.instructions().i32_add();
        f.instructions().local_set(lcl.next_cell_idx);
        // Mirror the per-list bump on each kind's running counter
        // when this list contributes element entries of that kind.
        if let (Some(next_handle_idx), Some(handle_slot_base)) =
            (lcl.next_handle_idx, ll.handle_slot_base)
        {
            f.instructions().local_get(next_handle_idx);
            f.instructions().local_set(handle_slot_base);
            f.instructions().local_get(next_handle_idx);
            f.instructions().local_get(ll.len);
            if ll.handles_per_elem != 1 {
                f.instructions().i32_const(ll.handles_per_elem as i32);
                f.instructions().i32_mul();
            }
            f.instructions().i32_add();
            f.instructions().local_set(next_handle_idx);
        }
        if let (Some(next_flags_idx), Some(flags_slot_base)) =
            (lcl.next_flags_idx, ll.flags_slot_base)
        {
            f.instructions().local_get(next_flags_idx);
            f.instructions().local_set(flags_slot_base);
            f.instructions().local_get(next_flags_idx);
            f.instructions().local_get(ll.len);
            if ll.flags_per_elem != 1 {
                f.instructions().i32_const(ll.flags_per_elem as i32);
                f.instructions().i32_mul();
            }
            f.instructions().i32_add();
            f.instructions().local_set(next_flags_idx);
        }
        if let (Some(next_record_idx), Some(record_slot_base)) =
            (lcl.next_record_idx, ll.record_slot_base)
        {
            f.instructions().local_get(next_record_idx);
            f.instructions().local_set(record_slot_base);
            f.instructions().local_get(next_record_idx);
            f.instructions().local_get(ll.len);
            if ll.records_per_elem != 1 {
                f.instructions().i32_const(ll.records_per_elem as i32);
                f.instructions().i32_mul();
            }
            f.instructions().i32_add();
            f.instructions().local_set(next_record_idx);
        }
        if let (Some(next_variant_idx), Some(variant_slot_base)) =
            (lcl.next_variant_idx, ll.variant_slot_base)
        {
            f.instructions().local_get(next_variant_idx);
            f.instructions().local_set(variant_slot_base);
            f.instructions().local_get(next_variant_idx);
            f.instructions().local_get(ll.len);
            if ll.variants_per_elem != 1 {
                f.instructions().i32_const(ll.variants_per_elem as i32);
                f.instructions().i32_mul();
            }
            f.instructions().i32_add();
            f.instructions().local_set(next_variant_idx);
        }
        emit_close_arm_guards(f, spec.arm_guards.len());
    }
}

/// Emit one cell at `lcl.addr`. `list_slot` is `Some` exactly for
/// `Cell::ListOf`. New `Cell` variants add an arm (no `_` catchall).
fn emit_cell_op(
    f: &mut Function,
    ctx: &LiftEmitCtx<'_>,
    cur: PlanCursor<'_>,
    op: &Cell,
    side_data: &CellSideData,
    lcl: &WrapperLocals,
    list_slot: Option<&ListEmitLocals>,
) {
    let PlanCursor {
        plan,
        local_base,
        elem_cell_base,
    } = cur;
    let addr = lcl.addr;
    let cell_layout = ctx.cell_layout;
    match op {
        Cell::Bool { flat_slot }
        | Cell::IntegerSignExt { flat_slot }
        | Cell::IntegerZeroExt { flat_slot }
        | Cell::EnumCase { flat_slot, .. }
        | Cell::Flags { flat_slot, .. }
        | Cell::Char { flat_slot }
        | Cell::Handle { flat_slot, .. } => {
            let src = pin_leaf_flat(f, plan, local_base, *flat_slot, WasmType::I32, lcl);
            emit_single_slot_cell(f, ctx, op, side_data, src, lcl, elem_cell_base);
        }
        Cell::Integer64 { flat_slot } => {
            let src = pin_leaf_flat(f, plan, local_base, *flat_slot, WasmType::I64, lcl);
            emit_single_slot_cell(f, ctx, op, side_data, src, lcl, elem_cell_base);
        }
        Cell::FloatingF32 { flat_slot } => {
            let src = pin_leaf_flat(f, plan, local_base, *flat_slot, WasmType::F32, lcl);
            emit_single_slot_cell(f, ctx, op, side_data, src, lcl, elem_cell_base);
        }
        Cell::FloatingF64 { flat_slot } => {
            let src = pin_leaf_flat(f, plan, local_base, *flat_slot, WasmType::F64, lcl);
            emit_single_slot_cell(f, ctx, op, side_data, src, lcl, elem_cell_base);
        }
        Cell::Option { disc_slot, .. }
        | Cell::Result { disc_slot, .. }
        | Cell::Variant { disc_slot, .. } => {
            let src = pin_leaf_flat(f, plan, local_base, *disc_slot, WasmType::I32, lcl);
            emit_single_slot_cell(f, ctx, op, side_data, src, lcl, elem_cell_base);
        }
        Cell::Text { ptr_slot, len_slot } => {
            let (ptr, len) = pin_text_bytes_slots(f, plan, local_base, *ptr_slot, *len_slot, lcl);
            cell_layout.emit_text(f, addr, ptr, len);
        }
        Cell::Bytes { ptr_slot, len_slot } => {
            let (ptr, len) = pin_text_bytes_slots(f, plan, local_base, *ptr_slot, *len_slot, lcl);
            cell_layout.emit_bytes(f, addr, ptr, len);
        }
        Cell::RecordOf { fields, .. } => {
            let CellSideData::Record(fill) = side_data else {
                panic!("RecordOf cell paired with non-Record side data {side_data:?}");
            };
            emit_record_runtime_fill(f, fill, fields, elem_cell_base, lcl, ctx.record_info);
            let payload = stage_record_cell_payload(f, lcl, fill);
            cell_layout.emit_record_of(f, addr, payload);
        }
        Cell::TupleOf { children } => {
            let CellSideData::Tuple { source: src_kind } = side_data else {
                panic!("TupleOf cell paired with non-Tuple side data {side_data:?}");
            };
            emit_tuple_of_cell(f, cell_layout, addr, children, src_kind, lcl);
        }
        Cell::ListOf {
            ptr_slot,
            element_plan,
            arm_guards,
            ..
        } => {
            let ll =
                list_slot.expect("ListOf cell must arrive with a matching ListEmitLocals slot");
            // Disc-gate cabi_realloc + element loop so an inactive
            // sibling arm's bytes can't surface as `len`.
            emit_open_arm_guards(f, plan, local_base, arm_guards);
            let ptr = pin_leaf_flat(f, plan, local_base, *ptr_slot, WasmType::I32, lcl);
            emit_list_of_arm(f, ctx, ll, ptr, element_plan, lcl);
            emit_close_arm_guards(f, arm_guards.len());
        }
    }
}

/// Emit one single-source cell at `lcl.addr`, reading from `source`.
/// Shared by `emit_cell_op` and `emit_lift_result`'s Direct branch.
/// `elem_cell_base = Some` inside list-element bodies.
fn emit_single_slot_cell(
    f: &mut Function,
    ctx: &LiftEmitCtx<'_>,
    cell: &Cell,
    side_data: &CellSideData,
    source: u32,
    lcl: &WrapperLocals,
    elem_cell_base: Option<u32>,
) {
    let addr = lcl.addr;
    let cell_layout = ctx.cell_layout;
    match cell {
        Cell::Bool { .. } => cell_layout.emit_bool(f, addr, source),
        Cell::IntegerSignExt { .. } => {
            f.instructions().local_get(source);
            f.instructions().i64_extend_i32_s();
            f.instructions().local_set(lcl.ext64);
            cell_layout.emit_integer(f, addr, lcl.ext64);
        }
        Cell::IntegerZeroExt { .. } => {
            f.instructions().local_get(source);
            f.instructions().i64_extend_i32_u();
            f.instructions().local_set(lcl.ext64);
            cell_layout.emit_integer(f, addr, lcl.ext64);
        }
        Cell::Integer64 { .. } => cell_layout.emit_integer(f, addr, source),
        Cell::FloatingF32 { .. } => {
            f.instructions().local_get(source);
            f.instructions().f64_promote_f32();
            f.instructions().local_set(lcl.ext_f64);
            cell_layout.emit_floating(f, addr, lcl.ext_f64);
        }
        Cell::FloatingF64 { .. } => cell_layout.emit_floating(f, addr, source),
        Cell::EnumCase { .. } => cell_layout.emit_enum_case(f, addr, source),
        Cell::Flags { .. } => {
            let CellSideData::Flags(fill) = side_data else {
                panic!("Flags cell paired with non-Flags side data {side_data:?}");
            };
            emit_flags_runtime_fill(f, source, fill, lcl, ctx.flags_info);
            let payload = stage_flags_cell_payload(f, lcl, fill);
            cell_layout.emit_flags_set(f, addr, payload);
        }
        Cell::Char { .. } => {
            let CellSideData::Char { scratch } = side_data else {
                panic!("Char cell paired with non-Char side data {side_data:?}");
            };
            let scratch_addr_local = lcl
                .char_scratch_addr
                .expect("fn_contains_char must agree with cells reaching here");
            let len_local = lcl.char_len.expect("same gate as char_scratch_addr");
            match scratch {
                CharScratch::Static { scratch_addr } => {
                    f.instructions().i32_const(*scratch_addr);
                    f.instructions().local_set(scratch_addr_local);
                }
                // Caller (emit_list_of_arm) wrote scratch_addr_local.
                CharScratch::Prestaged => {}
            }
            cell_layout.emit_char(f, addr, source, scratch_addr_local, len_local);
        }
        Cell::Handle { kind, .. } => {
            let CellSideData::Handle(fill) = side_data else {
                panic!("Handle cell paired with non-Handle side data {side_data:?}");
            };
            emit_handle_runtime_fill(f, source, fill, lcl, ctx.handle_info);
            let payload = stage_handle_cell_payload(f, lcl, fill);
            cell_layout.emit_handle_cell(f, addr, kind.cell_disc_case(), payload);
        }
        Cell::Option { child_idx, .. } => {
            // Stage inside the `some` arm; none skips it entirely.
            f.instructions().local_get(source);
            f.instructions().if_(BlockType::Empty);
            let child_idx_source = stage_child_idx_source(f, lcl, elem_cell_base, *child_idx);
            cell_layout.emit_option_some(f, addr, child_idx_source);
            f.instructions().else_();
            cell_layout.emit_option_none(f, addr);
            f.instructions().end();
        }
        Cell::Result {
            ok_idx, err_idx, ..
        } => {
            // Stage per arm; unit arms skip (has_payload == false).
            f.instructions().local_get(source);
            f.instructions().if_(BlockType::Empty);
            // wasm `if` fires on non-zero, so err goes in the if block.
            let err_source = match err_idx {
                Some(rel) => stage_child_idx_source(f, lcl, elem_cell_base, *rel),
                None => PayloadSource::ConstI32(0),
            };
            cell_layout.emit_result_err(f, addr, err_idx.is_some(), err_source);
            f.instructions().else_();
            let ok_source = match ok_idx {
                Some(rel) => stage_child_idx_source(f, lcl, elem_cell_base, *rel),
                None => PayloadSource::ConstI32(0),
            };
            cell_layout.emit_result_ok(f, addr, ok_idx.is_some(), ok_source);
            f.instructions().end();
        }
        Cell::Variant { .. } => {
            let CellSideData::Variant(fill) = side_data else {
                panic!("Variant cell paired with non-Variant side data {side_data:?}");
            };
            emit_variant_runtime_fill(f, source, fill, elem_cell_base, lcl, ctx.variant_info);
            let payload = stage_variant_cell_payload(f, lcl, fill);
            cell_layout.emit_variant_case(f, addr, payload);
        }
        Cell::Text { .. }
        | Cell::Bytes { .. }
        | Cell::RecordOf { .. }
        | Cell::TupleOf { .. }
        | Cell::ListOf { .. } => {
            unreachable!("emit_single_slot_cell reached non-single-source Cell {cell:?}")
        }
    }
}

/// Emit one `Cell::ListOf` arm: write the list-of payload, allocate
/// per-call buffers, loop `j ∈ 0..len` lifting each element. Each
/// element cell `k` lands at `start_i + j*elem_count + k`.
fn emit_list_of_arm(
    f: &mut Function,
    ctx: &LiftEmitCtx<'_>,
    ll: &ListEmitLocals,
    list_ptr_local: u32,
    element_plan: &LiftPlan,
    lcl: &WrapperLocals,
) {
    let cell_layout = ctx.cell_layout;
    let elem_count = element_plan.cell_count();
    emit_cabi_realloc_call_runtime(f, ctx.cabi_realloc_idx, 4, ll.len, 4, ll.indices_ptr);
    // Per-element utf-8 scratch: chars_per_elem * MAX_UTF8_LEN bytes.
    if let Some(scratch_base) = ll.char_scratch_base {
        emit_cabi_realloc_call_runtime(
            f,
            ctx.cabi_realloc_idx,
            1,
            ll.len,
            ll.chars_per_elem * MAX_UTF8_LEN,
            scratch_base,
        );
    }
    // Per-element tuple-indices buffer: tuple_idx_count_per_elem u32 slots.
    if let Some(buf_base) = ll.tuple_idx_buf_base {
        let elem_bytes = ll
            .tuple_idx_count_per_elem
            .checked_mul(4)
            .expect("tuple_idx_count_per_elem * 4 overflowed u32");
        emit_cabi_realloc_call_runtime(f, ctx.cabi_realloc_idx, 4, ll.len, elem_bytes, buf_base);
    }
    // Per-element flags set-flags scratch.
    if let Some(buf_base) = ll.flags_scratch_buf_base {
        emit_cabi_realloc_call_runtime(
            f,
            ctx.cabi_realloc_idx,
            4,
            ll.len,
            ll.flags_scratch_bytes_per_elem,
            buf_base,
        );
    }
    // Per-element record field-tuples scratch.
    if let Some(buf_base) = ll.record_tuples_buf_base {
        emit_cabi_realloc_call_runtime(
            f,
            ctx.cabi_realloc_idx,
            ctx.record_info.tuple_align,
            ll.len,
            ll.record_tuples_bytes_per_elem,
            buf_base,
        );
    }
    cell_layout.emit_list_of(f, lcl.addr, ll.indices_ptr, ll.len);

    // for (j = 0; j < len; j++) { ... }
    f.instructions().i32_const(0);
    f.instructions().local_set(ll.j);
    f.instructions().block(BlockType::Empty);
    f.instructions().loop_(BlockType::Empty);
    f.instructions().local_get(ll.j);
    f.instructions().local_get(ll.len);
    f.instructions().i32_ge_u();
    f.instructions().br_if(1);

    // elem_addr = list_ptr + j * elem_byte_size
    f.instructions().local_get(list_ptr_local);
    f.instructions().local_get(ll.j);
    if ll.elem_byte_size != 1 {
        f.instructions().i32_const(ll.elem_byte_size as i32);
        f.instructions().i32_mul();
    }
    f.instructions().i32_add();
    f.instructions().local_set(ll.elem_addr);

    // Lift element flat values from memory into elem_flat_locals (LIFO capture).
    for inst in &ll.elem_loads {
        f.instruction(inst);
    }
    for &local in ll.elem_flat_locals.iter().rev() {
        f.instructions().local_set(local);
    }

    // elem_cell_base = start_i + j*elem_count — staged once per iter.
    f.instructions().local_get(ll.start_i);
    f.instructions().local_get(ll.j);
    if elem_count != 1 {
        f.instructions().i32_const(elem_count as i32);
        f.instructions().i32_mul();
    }
    f.instructions().i32_add();
    f.instructions().local_set(ll.elem_cell_base);

    // list_elem_handle_base = handle_slot_base + j * handles_per_elem
    if let Some(slot_base) = ll.handle_slot_base {
        let dest = lcl
            .list_elem_handle_base
            .expect("fn_has_list_elem_handle disagrees with handle_slot_base");
        f.instructions().local_get(slot_base);
        f.instructions().local_get(ll.j);
        if ll.handles_per_elem != 1 {
            f.instructions().i32_const(ll.handles_per_elem as i32);
            f.instructions().i32_mul();
        }
        f.instructions().i32_add();
        f.instructions().local_set(dest);
    }

    // list_elem_flags_base = flags_slot_base + j * flags_per_elem,
    // list_elem_flags_scratch_base = flags_scratch_buf_base + j * scratch_bytes.
    if let Some(slot_base) = ll.flags_slot_base {
        let dest = lcl
            .list_elem_flags_base
            .expect("fn_has_list_elem_flags disagrees with flags_slot_base");
        f.instructions().local_get(slot_base);
        f.instructions().local_get(ll.j);
        if ll.flags_per_elem != 1 {
            f.instructions().i32_const(ll.flags_per_elem as i32);
            f.instructions().i32_mul();
        }
        f.instructions().i32_add();
        f.instructions().local_set(dest);
    }
    if let Some(buf_base) = ll.flags_scratch_buf_base {
        let dest = lcl
            .list_elem_flags_scratch_base
            .expect("fn_has_list_elem_flags disagrees with flags_scratch_buf_base");
        // Stride min is 1 cell's `STRING_FLAT_BYTES` (gated by Some).
        debug_assert!(
            ll.flags_scratch_bytes_per_elem >= STRING_FLAT_BYTES,
            "flags_scratch_bytes_per_elem ({}) below 1 cell's worth of \
             pair-bytes ({STRING_FLAT_BYTES})",
            ll.flags_scratch_bytes_per_elem,
        );
        f.instructions().local_get(buf_base);
        f.instructions().local_get(ll.j);
        f.instructions()
            .i32_const(ll.flags_scratch_bytes_per_elem as i32);
        f.instructions().i32_mul();
        f.instructions().i32_add();
        f.instructions().local_set(dest);
    }

    // list_elem_record_base = record_slot_base + j * records_per_elem,
    // list_elem_record_tuples_base = record_tuples_buf_base + j * ...
    if let Some(slot_base) = ll.record_slot_base {
        let dest = lcl
            .list_elem_record_base
            .expect("fn_has_list_elem_record disagrees with record_slot_base");
        f.instructions().local_get(slot_base);
        f.instructions().local_get(ll.j);
        if ll.records_per_elem != 1 {
            f.instructions().i32_const(ll.records_per_elem as i32);
            f.instructions().i32_mul();
        }
        f.instructions().i32_add();
        f.instructions().local_set(dest);
    }
    if let Some(buf_base) = ll.record_tuples_buf_base {
        let dest = lcl
            .list_elem_record_tuples_base
            .expect("fn_has_list_elem_record disagrees with record_tuples_buf_base");
        f.instructions().local_get(buf_base);
        f.instructions().local_get(ll.j);
        f.instructions()
            .i32_const(ll.record_tuples_bytes_per_elem as i32);
        f.instructions().i32_mul();
        f.instructions().i32_add();
        f.instructions().local_set(dest);
    }

    // list_elem_variant_base = variant_slot_base + j * variants_per_elem
    if let Some(slot_base) = ll.variant_slot_base {
        let dest = lcl
            .list_elem_variant_base
            .expect("fn_has_list_elem_variant disagrees with variant_slot_base");
        f.instructions().local_get(slot_base);
        f.instructions().local_get(ll.j);
        if ll.variants_per_elem != 1 {
            f.instructions().i32_const(ll.variants_per_elem as i32);
            f.instructions().i32_mul();
        }
        f.instructions().i32_add();
        f.instructions().local_set(dest);
    }

    // Per element-plan cell: stage addr (and any per-iter scratch),
    // then dispatch. `char_idx` walks char-cells so multi-char
    // elements get distinct slots in the per-call buffer.
    let mut char_idx: u32 = 0;
    for (cell_pos, elem_cell) in element_plan.cells.iter().enumerate() {
        emit_set_elem_cell_addr(f, lcl, ll, cell_pos as u32, cell_layout.size);
        match elem_cell {
            Cell::Char { .. } => {
                emit_stage_char_scratch_addr(f, lcl, ll, char_idx);
                char_idx += 1;
            }
            Cell::TupleOf { children } => {
                let CellSideData::Tuple {
                    source: TupleIdxSource::PerIteration { offset_in_elem },
                } = ll.elem_cell_side[cell_pos]
                else {
                    unreachable!(
                        "list-element TupleOf at {cell_pos} must carry PerIteration side data, \
                         got {:?}",
                        ll.elem_cell_side[cell_pos]
                    );
                };
                emit_stage_tuple_slot(f, lcl, ll, offset_in_elem, children);
            }
            _ => {}
        }
        emit_cell_op(
            f,
            ctx,
            PlanCursor {
                plan: element_plan,
                local_base: ll.elem_flat_locals[0],
                elem_cell_base: Some(ll.elem_cell_base),
            },
            elem_cell,
            &ll.elem_cell_side[cell_pos],
            lcl,
            None,
        );
    }
    debug_assert_eq!(
        char_idx, ll.chars_per_elem,
        "emit walk visited {char_idx} Cell::Char element cells; \
         build_one_list_emit_locals counted {}",
        ll.chars_per_elem,
    );

    // indices_ptr[j*4] = elem_cell_base + root.
    f.instructions().local_get(ll.indices_ptr);
    f.instructions().local_get(ll.j);
    f.instructions().i32_const(4);
    f.instructions().i32_mul();
    f.instructions().i32_add();
    f.instructions().local_get(ll.elem_cell_base);
    if element_plan.root() != 0 {
        f.instructions().i32_const(element_plan.root() as i32);
        f.instructions().i32_add();
    }
    f.instructions().i32_store(MemArg {
        offset: 0,
        align: I32_STORE_LOG2_ALIGN,
        memory_index: 0,
    });

    f.instructions().local_get(ll.j);
    f.instructions().i32_const(1);
    f.instructions().i32_add();
    f.instructions().local_set(ll.j);
    f.instructions().br(0);
    f.instructions().end(); // loop
    f.instructions().end(); // block
}

/// Stage the absolute cell-array address of element-plan position
/// `cell_pos` into `lcl.addr`:
/// `cells_base + (elem_cell_base + cell_pos) * cell_size`.
fn emit_set_elem_cell_addr(
    f: &mut Function,
    lcl: &WrapperLocals,
    ll: &ListEmitLocals,
    cell_pos: u32,
    cell_size: u32,
) {
    f.instructions().local_get(lcl.cells_base);
    f.instructions().local_get(ll.elem_cell_base);
    if cell_pos != 0 {
        f.instructions().i32_const(cell_pos as i32);
        f.instructions().i32_add();
    }
    f.instructions().i32_const(cell_size as i32);
    f.instructions().i32_mul();
    f.instructions().i32_add();
    f.instructions().local_set(lcl.addr);
}

/// Resolve a cell-payload child cell-array index. Static → ConstI32;
/// list-element → stage `elem_cell_base + relative_idx` into
/// `lcl.list_elem_child_idx`. `relative_idx == 0` reuses `base` directly.
fn stage_child_idx_source(
    f: &mut Function,
    lcl: &WrapperLocals,
    elem_cell_base: Option<u32>,
    relative_idx: u32,
) -> PayloadSource {
    let Some(base) = elem_cell_base else {
        return PayloadSource::ConstI32(relative_idx as i32);
    };
    if relative_idx == 0 {
        return PayloadSource::Local(base);
    }
    let dest = lcl
        .list_elem_child_idx
        .expect("fn_has_list_elem_child_idx disagrees with cells reaching here");
    f.instructions().local_get(base);
    f.instructions().i32_const(relative_idx as i32);
    f.instructions().i32_add();
    f.instructions().local_set(dest);
    PayloadSource::Local(dest)
}

/// Emit one `Cell::TupleOf`. Static cells point at the build-time
/// blob; list-element cells consume `lcl.tuple_slot_ptr`.
fn emit_tuple_of_cell(
    f: &mut Function,
    cell_layout: &CellLayout,
    addr: u32,
    children: &[u32],
    src_kind: &TupleIdxSource,
    lcl: &WrapperLocals,
) {
    let len = children.len() as u32;
    match src_kind {
        TupleIdxSource::Static(slice) => {
            debug_assert_eq!(slice.len, len);
            cell_layout.emit_tuple_of(f, addr, PayloadSource::ConstI32(slice.off as i32), len);
        }
        TupleIdxSource::PerIteration { .. } => {
            let slot_ptr_local = lcl.tuple_slot_ptr.expect(
                "tuple_slot_ptr unset — fn_has_list_elem_tuple must agree with \
                 PerIteration cells reaching emit_tuple_of_cell",
            );
            cell_layout.emit_tuple_of(f, addr, PayloadSource::Local(slot_ptr_local), len);
        }
    }
}

/// Stage `lcl.tuple_slot_ptr = ll.tuple_idx_buf_base + j *
/// tuple_idx_count_per_elem * 4 + offset_in_elem`, then write each
/// child's runtime cell-array index (`elem_cell_base + relative`)
/// into `mem[slot_ptr + i*4]`. Called once per iteration before the
/// matching `Cell::TupleOf` element-plan cell's emit fires.
fn emit_stage_tuple_slot(
    f: &mut Function,
    lcl: &WrapperLocals,
    ll: &ListEmitLocals,
    offset_in_elem: u32,
    children: &[u32],
) {
    let buf_base = ll.tuple_idx_buf_base.expect(
        "Cell::TupleOf element requires tuple_idx_buf_base — \
         build_one_list_emit_locals must have allocated it",
    );
    let dest = lcl.tuple_slot_ptr.expect(
        "tuple_slot_ptr unset for list with TupleOf elements — \
         fn_has_list_elem_tuple must include element-plan TupleOf cells",
    );
    // slot_ptr = buf_base + j*stride + offset_in_elem. Stride is
    // always ≥4 (one u32 per child × ≥1 child per TupleOf).
    let stride_bytes = ll
        .tuple_idx_count_per_elem
        .checked_mul(4)
        .expect("tuple_idx_count_per_elem * 4 overflowed u32");
    f.instructions().local_get(buf_base);
    f.instructions().local_get(ll.j);
    f.instructions().i32_const(stride_bytes as i32);
    f.instructions().i32_mul();
    f.instructions().i32_add();
    if offset_in_elem != 0 {
        f.instructions().i32_const(offset_in_elem as i32);
        f.instructions().i32_add();
    }
    f.instructions().local_set(dest);
    // mem[slot_ptr + i*4] = elem_cell_base + child[i]
    for (i, child) in children.iter().enumerate() {
        f.instructions().local_get(dest);
        f.instructions().local_get(ll.elem_cell_base);
        if *child != 0 {
            f.instructions().i32_const(*child as i32);
            f.instructions().i32_add();
        }
        f.instructions().i32_store(MemArg {
            offset: ((i as u32) * 4) as u64,
            align: I32_STORE_LOG2_ALIGN,
            memory_index: 0,
        });
    }
}

/// Stage utf-8 scratch addr for the `char_idx`-th `Cell::Char` of
/// element_plan: `base + (j * chars_per_elem + char_idx) * MAX_UTF8_LEN`.
/// Pairs with `CharScratch::Prestaged`.
fn emit_stage_char_scratch_addr(
    f: &mut Function,
    lcl: &WrapperLocals,
    ll: &ListEmitLocals,
    char_idx: u32,
) {
    let scratch_base = ll
        .char_scratch_base
        .expect("Cell::Char element requires char_scratch_base");
    let scratch_addr_local = lcl
        .char_scratch_addr
        .expect("fn_contains_char must include element-plan chars");
    debug_assert!(char_idx < ll.chars_per_elem);
    // base + (j * chars_per_elem + char_idx) * MAX_UTF8_LEN
    f.instructions().local_get(scratch_base);
    f.instructions().local_get(ll.j);
    if ll.chars_per_elem != 1 {
        f.instructions().i32_const(ll.chars_per_elem as i32);
        f.instructions().i32_mul();
    }
    if char_idx != 0 {
        f.instructions().i32_const(char_idx as i32);
        f.instructions().i32_add();
    }
    if MAX_UTF8_LEN != 1 {
        f.instructions().i32_const(MAX_UTF8_LEN as i32);
        f.instructions().i32_mul();
    }
    f.instructions().i32_add();
    f.instructions().local_set(scratch_addr_local);
}

/// Fill one `Cell::Handle`'s slot: const `type-name` + runtime
/// zero-extended `id`. Static folds the offset into memargs;
/// PerIteration stages slot_addr once and reuses it.
fn emit_handle_runtime_fill(
    f: &mut Function,
    handle_local: u32,
    fill: &HandleRuntimeFill,
    lcl: &WrapperLocals,
    info: HandleInfoOffsets,
) {
    use super::sidetable::handle_info::HandleSlotSource;
    let base_local = lcl
        .handle_info_base
        .expect("fn_has_handle_cells disagrees with cells reaching here");
    let (slot_local, type_name_off, id_off) = match fill.slot_source {
        HandleSlotSource::Static(idx) => {
            let entry_off = idx * info.entry_size;
            (
                base_local,
                entry_off + info.type_name_off,
                entry_off + info.id_off,
            )
        }
        HandleSlotSource::PerIteration { offset_in_elem } => {
            // slot_addr = base + (iter_base + offset_in_elem) * entry_size
            let iter_base = lcl
                .list_elem_handle_base
                .expect("fn_has_list_elem_handle disagrees with walk_element_plan");
            let scratch = lcl
                .handle_slot_addr
                .expect("fn_has_list_elem_handle disagrees with cells reaching here");
            f.instructions().local_get(iter_base);
            if offset_in_elem != 0 {
                f.instructions().i32_const(offset_in_elem as i32);
                f.instructions().i32_add();
            }
            f.instructions().i32_const(info.entry_size as i32);
            f.instructions().i32_mul();
            f.instructions().local_get(base_local);
            f.instructions().i32_add();
            f.instructions().local_set(scratch);
            (scratch, info.type_name_off, info.id_off)
        }
    };

    let store_i32 = |f: &mut Function, off: u32, value: i32| {
        f.instructions().local_get(slot_local);
        f.instructions().i32_const(value);
        f.instructions().i32_store(MemArg {
            offset: off as u64,
            align: I32_STORE_LOG2_ALIGN,
            memory_index: 0,
        });
    };
    store_i32(
        f,
        type_name_off + SLICE_PTR_OFFSET,
        fill.type_name.off as i32,
    );
    store_i32(
        f,
        type_name_off + SLICE_LEN_OFFSET,
        fill.type_name.len as i32,
    );

    f.instructions().local_get(slot_local);
    f.instructions().local_get(handle_local);
    f.instructions().i64_extend_i32_u();
    f.instructions().i64_store(MemArg {
        offset: id_off as u64,
        align: I64_STORE_LOG2_ALIGN,
        memory_index: 0,
    });
}

/// Fill one `Cell::RecordOf`'s slot: write `type-name` + `fields`
/// slice. Static folds offsets into memargs; PerIteration stages
/// entry + tuples addrs from the iter bases and writes each field's
/// `(name, child-cell-idx)` tuple with idx = elem_cell_base + child_pos.
fn emit_record_runtime_fill(
    f: &mut Function,
    fill: &RecordRuntimeFill,
    fields: &[(BlobSlice, u32)],
    elem_cell_base: Option<u32>,
    lcl: &WrapperLocals,
    info: RecordInfoOffsets,
) {
    use super::sidetable::record_info::RecordSlotSource;
    debug_assert_eq!(
        matches!(fill.slot_source, RecordSlotSource::PerIteration { .. }),
        elem_cell_base.is_some(),
        "slot_source / elem_cell_base must agree (PerIteration ↔ Some)",
    );
    let base_local = lcl
        .record_info_base
        .expect("fn_has_record_cells disagrees with cells reaching here");
    let store_i32 = |off: u32| MemArg {
        offset: off as u64,
        align: I32_STORE_LOG2_ALIGN,
        memory_index: 0,
    };

    match fill.slot_source {
        RecordSlotSource::Static {
            entry_idx,
            fields_ptr,
        } => {
            let entry_off = entry_idx * info.entry_size;
            let type_name_off = entry_off + info.type_name_off;
            let fields_off = entry_off + info.fields_off;
            let store_const = |f: &mut Function, off: u32, value: i32| {
                f.instructions().local_get(base_local);
                f.instructions().i32_const(value);
                f.instructions().i32_store(store_i32(off));
            };
            store_const(
                f,
                type_name_off + SLICE_PTR_OFFSET,
                fill.type_name.off as i32,
            );
            store_const(
                f,
                type_name_off + SLICE_LEN_OFFSET,
                fill.type_name.len as i32,
            );
            store_const(f, fields_off + SLICE_PTR_OFFSET, fields_ptr);
            store_const(f, fields_off + SLICE_LEN_OFFSET, fill.fields_len as i32);
        }
        RecordSlotSource::PerIteration {
            entry_offset_in_elem,
            tuples_offset_in_elem,
        } => {
            let iter_entry_base = lcl
                .list_elem_record_base
                .expect("fn_has_list_elem_record disagrees");
            let iter_tuples_base = lcl
                .list_elem_record_tuples_base
                .expect("fn_has_list_elem_record disagrees");
            let entry_dest = lcl
                .record_slot_addr
                .expect("fn_has_list_elem_record disagrees");
            let tuples_dest = lcl
                .record_tuples_slice_addr
                .expect("fn_has_list_elem_record disagrees");
            // entry_addr = base + (iter_entry_base + entry_offset_in_elem) * entry_size
            f.instructions().local_get(iter_entry_base);
            if entry_offset_in_elem != 0 {
                f.instructions().i32_const(entry_offset_in_elem as i32);
                f.instructions().i32_add();
            }
            f.instructions().i32_const(info.entry_size as i32);
            f.instructions().i32_mul();
            f.instructions().local_get(base_local);
            f.instructions().i32_add();
            f.instructions().local_set(entry_dest);
            // tuples_slice_addr = iter_tuples_base + tuples_offset_in_elem
            f.instructions().local_get(iter_tuples_base);
            if tuples_offset_in_elem != 0 {
                f.instructions().i32_const(tuples_offset_in_elem as i32);
                f.instructions().i32_add();
            }
            f.instructions().local_set(tuples_dest);

            // type-name + fields.len const; fields.ptr = tuples_slice_addr.
            let store_const = |f: &mut Function, off: u32, value: i32| {
                f.instructions().local_get(entry_dest);
                f.instructions().i32_const(value);
                f.instructions().i32_store(store_i32(off));
            };
            store_const(
                f,
                info.type_name_off + SLICE_PTR_OFFSET,
                fill.type_name.off as i32,
            );
            store_const(
                f,
                info.type_name_off + SLICE_LEN_OFFSET,
                fill.type_name.len as i32,
            );
            f.instructions().local_get(entry_dest);
            f.instructions().local_get(tuples_dest);
            f.instructions()
                .i32_store(store_i32(info.fields_off + SLICE_PTR_OFFSET));
            store_const(
                f,
                info.fields_off + SLICE_LEN_OFFSET,
                fill.fields_len as i32,
            );

            // tuple at tuples_slice_addr + i*tuple_size:
            // name = const, idx = elem_cell_base + child_pos.
            let elem_base = elem_cell_base.expect(
                "PerIteration record fill needs elem_cell_base — emit_cell_op must thread it",
            );
            debug_assert_eq!(
                fields.len() as u32,
                fill.fields_len,
                "fill.fields_len must match the cell's fields slice",
            );
            for (i, (name, child_pos_in_elem)) in fields.iter().enumerate() {
                let i = i as u32;
                let tuple_off = i * info.tuple_size;
                let store_name = |f: &mut Function, off: u32, value: i32| {
                    f.instructions().local_get(tuples_dest);
                    f.instructions().i32_const(value);
                    f.instructions().i32_store(store_i32(off));
                };
                store_name(
                    f,
                    tuple_off + info.tuple_name_off + SLICE_PTR_OFFSET,
                    name.off as i32,
                );
                store_name(
                    f,
                    tuple_off + info.tuple_name_off + SLICE_LEN_OFFSET,
                    name.len as i32,
                );
                f.instructions().local_get(tuples_dest);
                f.instructions().local_get(elem_base);
                if *child_pos_in_elem != 0 {
                    f.instructions().i32_const(*child_pos_in_elem as i32);
                    f.instructions().i32_add();
                }
                f.instructions()
                    .i32_store(store_i32(tuple_off + info.tuple_idx_off));
            }
        }
    }
}

/// Resolve a `Cell::RecordOf`'s `cell::record-of(idx)` payload.
fn stage_record_cell_payload(
    f: &mut Function,
    lcl: &WrapperLocals,
    fill: &RecordRuntimeFill,
) -> PayloadSource {
    use super::sidetable::record_info::RecordSlotSource;
    match fill.slot_source {
        RecordSlotSource::Static { entry_idx, .. } => PayloadSource::ConstI32(entry_idx as i32),
        RecordSlotSource::PerIteration {
            entry_offset_in_elem,
            ..
        } => {
            let iter_base = lcl
                .list_elem_record_base
                .expect("fn_has_list_elem_record disagrees");
            if entry_offset_in_elem == 0 {
                return PayloadSource::Local(iter_base);
            }
            let dest = lcl
                .record_payload_idx
                .expect("fn_has_list_elem_record disagrees");
            f.instructions().local_get(iter_base);
            f.instructions().i32_const(entry_offset_in_elem as i32);
            f.instructions().i32_add();
            f.instructions().local_set(dest);
            PayloadSource::Local(dest)
        }
    }
}

/// Resolve a `Cell::Flags`'s `cell::flags-set(idx)` payload.
fn stage_flags_cell_payload(
    f: &mut Function,
    lcl: &WrapperLocals,
    fill: &FlagsRuntimeFill,
) -> PayloadSource {
    use super::sidetable::flags_info::FlagsSlotSource;
    match fill.slot_source {
        FlagsSlotSource::Static { entry_idx, .. } => PayloadSource::ConstI32(entry_idx as i32),
        FlagsSlotSource::PerIteration {
            entry_offset_in_elem,
            ..
        } => {
            let iter_base = lcl
                .list_elem_flags_base
                .expect("fn_has_list_elem_flags disagrees");
            if entry_offset_in_elem == 0 {
                return PayloadSource::Local(iter_base);
            }
            let dest = lcl
                .flags_payload_idx
                .expect("fn_has_list_elem_flags disagrees");
            f.instructions().local_get(iter_base);
            f.instructions().i32_const(entry_offset_in_elem as i32);
            f.instructions().i32_add();
            f.instructions().local_set(dest);
            PayloadSource::Local(dest)
        }
    }
}

/// Resolve a `Cell::Handle`'s `cell::*-handle(idx)` payload.
fn stage_handle_cell_payload(
    f: &mut Function,
    lcl: &WrapperLocals,
    fill: &HandleRuntimeFill,
) -> PayloadSource {
    use super::sidetable::handle_info::HandleSlotSource;
    match fill.slot_source {
        HandleSlotSource::Static(idx) => PayloadSource::ConstI32(idx as i32),
        HandleSlotSource::PerIteration { offset_in_elem } => {
            let iter_base = lcl.list_elem_handle_base.expect(
                "list_elem_handle_base unset — fn_has_list_elem_handle gate \
                 disagrees with the cells reaching stage_handle_cell_payload",
            );
            if offset_in_elem == 0 {
                return PayloadSource::Local(iter_base);
            }
            let dest = lcl.handle_payload_idx.expect(
                "handle_payload_idx unset — fn_has_list_elem_handle gate \
                 disagrees with the cells reaching stage_handle_cell_payload",
            );
            f.instructions().local_get(iter_base);
            f.instructions().i32_const(offset_in_elem as i32);
            f.instructions().i32_add();
            f.instructions().local_set(dest);
            PayloadSource::Local(dest)
        }
    }
}

/// Fill one `Cell::Flags`'s slot + scratch: type-name + set-flags.ptr
/// const, bit-walk writes `(name_ptr, name_len)` pairs + count.
/// Unrolled bit-walk — at ≤ 8 bits a loop's overhead dominates.
fn emit_flags_runtime_fill(
    f: &mut Function,
    bitmask_local: u32,
    fill: &FlagsRuntimeFill,
    lcl: &WrapperLocals,
    info: FlagsInfoOffsets,
) {
    use super::sidetable::flags_info::FlagsSlotSource;
    let store_i32 = |off: u32| MemArg {
        offset: off as u64,
        align: I32_STORE_LOG2_ALIGN,
        memory_index: 0,
    };

    let base_local = lcl
        .flags_info_base
        .expect("fn_has_flags_cells disagrees with cells reaching here");
    // Stage entry_addr + flags_addr cursor for the bit-walker.
    let (entry_addr_local, entry_off) = match fill.slot_source {
        FlagsSlotSource::Static {
            entry_idx,
            scratch_addr,
        } => {
            f.instructions().i32_const(scratch_addr);
            f.instructions().local_set(lcl.flags_addr);
            (base_local, entry_idx * info.entry_size)
        }
        FlagsSlotSource::PerIteration {
            entry_offset_in_elem,
            scratch_offset_in_elem,
        } => {
            let entry_dest = lcl
                .flags_slot_addr
                .expect("fn_has_list_elem_flags disagrees");
            let iter_entry_base = lcl
                .list_elem_flags_base
                .expect("fn_has_list_elem_flags disagrees");
            let iter_scratch_base = lcl
                .list_elem_flags_scratch_base
                .expect("fn_has_list_elem_flags disagrees");
            // entry_addr = base + (iter_entry_base + entry_offset) * entry_size
            f.instructions().local_get(iter_entry_base);
            if entry_offset_in_elem != 0 {
                f.instructions().i32_const(entry_offset_in_elem as i32);
                f.instructions().i32_add();
            }
            f.instructions().i32_const(info.entry_size as i32);
            f.instructions().i32_mul();
            f.instructions().local_get(base_local);
            f.instructions().i32_add();
            f.instructions().local_set(entry_dest);
            // flags_addr = iter_scratch_base + scratch_offset_in_elem
            f.instructions().local_get(iter_scratch_base);
            if scratch_offset_in_elem != 0 {
                f.instructions().i32_const(scratch_offset_in_elem as i32);
                f.instructions().i32_add();
            }
            f.instructions().local_set(lcl.flags_addr);
            (entry_dest, 0u32)
        }
    };

    let type_name_off = entry_off + info.type_name_off;
    let set_flags_off = entry_off + info.set_flags_off;

    let store_const = |f: &mut Function, off: u32, value: i32| {
        f.instructions().local_get(entry_addr_local);
        f.instructions().i32_const(value);
        f.instructions().i32_store(store_i32(off));
    };
    store_const(
        f,
        type_name_off + SLICE_PTR_OFFSET,
        fill.type_name.off as i32,
    );
    store_const(
        f,
        type_name_off + SLICE_LEN_OFFSET,
        fill.type_name.len as i32,
    );
    // set-flags.ptr captures flags_addr before the bit-walk advances it.
    f.instructions().local_get(entry_addr_local);
    f.instructions().local_get(lcl.flags_addr);
    f.instructions()
        .i32_store(store_i32(set_flags_off + SLICE_PTR_OFFSET));

    f.instructions().i32_const(0);
    f.instructions().local_set(lcl.flags_count);

    for (i, name) in fill.flag_names.iter().enumerate() {
        // (bitmask >> i) & 1
        f.instructions().local_get(bitmask_local);
        f.instructions().i32_const(i as i32);
        f.instructions().i32_shr_u();
        f.instructions().i32_const(1);
        f.instructions().i32_and();
        f.instructions().if_(BlockType::Empty);
        // *flags_addr = name.off; *(flags_addr + SLICE_LEN_OFFSET) = name.len
        f.instructions().local_get(lcl.flags_addr);
        f.instructions().i32_const(name.off as i32);
        f.instructions().i32_store(store_i32(SLICE_PTR_OFFSET));
        f.instructions().local_get(lcl.flags_addr);
        f.instructions().i32_const(name.len as i32);
        f.instructions().i32_store(store_i32(SLICE_LEN_OFFSET));
        // flags_addr += sizeof(string); flags_count += 1
        f.instructions().local_get(lcl.flags_addr);
        f.instructions().i32_const(STRING_FLAT_BYTES as i32);
        f.instructions().i32_add();
        f.instructions().local_set(lcl.flags_addr);
        f.instructions().local_get(lcl.flags_count);
        f.instructions().i32_const(1);
        f.instructions().i32_add();
        f.instructions().local_set(lcl.flags_count);
        f.instructions().end();
    }

    // Write set-flags.len = flags_count (runtime).
    f.instructions().local_get(entry_addr_local);
    f.instructions().local_get(lcl.flags_count);
    f.instructions()
        .i32_store(store_i32(set_flags_off + SLICE_LEN_OFFSET));
}

/// Fill one `Cell::Variant`'s slot: const `type-name` + disc-dispatched
/// `case-name` + `payload` (option<u32>). N-way disc dispatch is nested
/// if/else; `br_table` is a future optimization.
fn emit_variant_runtime_fill(
    f: &mut Function,
    disc_local: u32,
    fill: &VariantRuntimeFill,
    elem_cell_base: Option<u32>,
    lcl: &WrapperLocals,
    info: VariantInfoOffsets,
) {
    use super::sidetable::variant_info::VariantSlotSource;
    // PerIteration ↔ Some couple: walk_element_plan only assigns
    // PerIteration to list-element cells; drift would silently emit
    // child_idx as a const ignoring `elem_cell_base`.
    debug_assert_eq!(
        matches!(fill.slot_source, VariantSlotSource::PerIteration { .. }),
        elem_cell_base.is_some(),
        "slot_source / elem_cell_base must agree (PerIteration ↔ Some)",
    );
    let base_local = lcl
        .variant_info_base
        .expect("fn_has_variant_cells disagrees with cells reaching here");
    let store_i32 = |off: u32| MemArg {
        offset: off as u64,
        align: I32_STORE_LOG2_ALIGN,
        memory_index: 0,
    };
    let store_i8 = |off: u32| MemArg {
        offset: off as u64,
        align: I8_STORE_LOG2_ALIGN,
        memory_index: 0,
    };

    // Static folds offsets into memargs; PerIteration stages slot_addr.
    let (slot_local, slot_off) = match fill.slot_source {
        VariantSlotSource::Static { entry_idx } => (base_local, entry_idx * info.entry_size),
        VariantSlotSource::PerIteration {
            entry_offset_in_elem,
        } => {
            let iter_base = lcl
                .list_elem_variant_base
                .expect("fn_has_list_elem_variant disagrees");
            let dest = lcl
                .variant_slot_addr
                .expect("fn_has_list_elem_variant disagrees");
            // slot_addr = base + (iter_base + entry_offset_in_elem) * entry_size
            f.instructions().local_get(iter_base);
            if entry_offset_in_elem != 0 {
                f.instructions().i32_const(entry_offset_in_elem as i32);
                f.instructions().i32_add();
            }
            f.instructions().i32_const(info.entry_size as i32);
            f.instructions().i32_mul();
            f.instructions().local_get(base_local);
            f.instructions().i32_add();
            f.instructions().local_set(dest);
            (dest, 0u32)
        }
    };

    let type_name_off = slot_off + info.type_name_off;
    let store_const = |f: &mut Function, off: u32, value: i32| {
        f.instructions().local_get(slot_local);
        f.instructions().i32_const(value);
        f.instructions().i32_store(store_i32(off));
    };
    store_const(
        f,
        type_name_off + SLICE_PTR_OFFSET,
        fill.type_name.off as i32,
    );
    store_const(
        f,
        type_name_off + SLICE_LEN_OFFSET,
        fill.type_name.len as i32,
    );

    let case_name_off = slot_off + info.case_name_off;
    let payload_off = slot_off + info.payload_off;
    let payload_value_off = payload_off + info.payload_value_off;

    debug_assert_eq!(fill.case_names.len(), fill.per_case_payload.len());

    // Nested if/else per disc; last arm has no else (canonical-ABI
    // disc out of range is unreachable).
    for (i, name) in fill.case_names.iter().enumerate() {
        let is_last = i + 1 == fill.case_names.len();
        if !is_last {
            f.instructions().local_get(disc_local);
            f.instructions().i32_const(i as i32);
            f.instructions().i32_eq();
            f.instructions().if_(BlockType::Empty);
        }
        store_const(f, case_name_off + SLICE_PTR_OFFSET, name.off as i32);
        store_const(f, case_name_off + SLICE_LEN_OFFSET, name.len as i32);
        // child_idx is plan-relative; list-element resolves to
        // elem_cell_base + child_pos at runtime.
        match fill.per_case_payload[i] {
            Some(child_idx) => {
                f.instructions().local_get(slot_local);
                f.instructions().i32_const(OPTION_SOME as i32);
                f.instructions().i32_store8(store_i8(payload_off));
                f.instructions().local_get(slot_local);
                match elem_cell_base {
                    None => {
                        f.instructions().i32_const(child_idx as i32);
                    }
                    Some(base) => {
                        f.instructions().local_get(base);
                        if child_idx != 0 {
                            f.instructions().i32_const(child_idx as i32);
                            f.instructions().i32_add();
                        }
                    }
                }
                f.instructions().i32_store(store_i32(payload_value_off));
            }
            None => {
                // value slot untouched (irrelevant when disc=0)
                f.instructions().local_get(slot_local);
                f.instructions().i32_const(OPTION_NONE as i32);
                f.instructions().i32_store8(store_i8(payload_off));
            }
        }
        if !is_last {
            f.instructions().else_();
        }
    }
    for _ in 0..fill.case_names.len().saturating_sub(1) {
        f.instructions().end();
    }
}

/// Resolve a `Cell::Variant`'s `cell::variant-case(idx)` payload.
fn stage_variant_cell_payload(
    f: &mut Function,
    lcl: &WrapperLocals,
    fill: &VariantRuntimeFill,
) -> PayloadSource {
    use super::sidetable::variant_info::VariantSlotSource;
    match fill.slot_source {
        VariantSlotSource::Static { entry_idx } => PayloadSource::ConstI32(entry_idx as i32),
        VariantSlotSource::PerIteration {
            entry_offset_in_elem,
        } => {
            let iter_base = lcl
                .list_elem_variant_base
                .expect("fn_has_list_elem_variant disagrees");
            if entry_offset_in_elem == 0 {
                return PayloadSource::Local(iter_base);
            }
            let dest = lcl
                .variant_payload_idx
                .expect("fn_has_list_elem_variant disagrees");
            f.instructions().local_get(iter_base);
            f.instructions().i32_const(entry_offset_in_elem as i32);
            f.instructions().i32_add();
            f.instructions().local_set(dest);
            PayloadSource::Local(dest)
        }
    }
}

/// Emit lift for a single-cell Direct result at `lcl.addr`. Compound
/// results go through `emit_lift_compound_prefix` + `emit_lift_plan`.
pub(crate) fn emit_lift_result(
    f: &mut Function,
    ctx: &LiftEmitCtx<'_>,
    plan: &ResultEmitPlan<'_>,
    lcl: &WrapperLocals,
) {
    match plan {
        ResultEmitPlan::Direct {
            cell,
            source_local,
            side_data,
        } => {
            // classify_result_lift routes Compound-only kinds away
            // from Direct.
            debug_assert!(
                !matches!(
                    cell,
                    Cell::Option { .. } | Cell::Result { .. } | Cell::Variant { .. }
                ),
                "Direct result reached emit_lift_result with Compound-only cell {cell:?}",
            );
            emit_single_slot_cell(f, ctx, cell, side_data, *source_local, lcl, None);
        }
        ResultEmitPlan::Compound { .. } | ResultEmitPlan::None => unreachable!(
            "Compound is emitted via emit_lift_compound_prefix + emit_lift_plan; \
             emit_lift_result handles only Direct sources"
        ),
    }
}

/// Emit compound-result prefix: load bytes from `retptr_offset` via
/// `loads`, then capture each flat value into `synth_locals` in
/// REVERSE order (wasm stack is LIFO).
pub(crate) fn emit_lift_compound_prefix(
    f: &mut Function,
    plan_flat_slot_count: u32,
    retptr_offset: i32,
    loads: &[Instruction<'static>],
    addr_local: u32,
    synth_locals: &[u32],
) {
    assert_eq!(
        synth_locals.len(),
        plan_flat_slot_count as usize,
        "synthetic-local count (emit) must match classify-time plan flat slot count"
    );
    f.instructions().i32_const(retptr_offset);
    f.instructions().local_set(addr_local);
    for inst in loads {
        f.instruction(inst);
    }
    // Reverse order: stack top is the last pushed (highest slot).
    for &local in synth_locals.iter().rev() {
        f.instructions().local_set(local);
    }
}