fusevm 0.13.4

Language-agnostic bytecode VM with fused superinstructions and a 3-tier Cranelift JIT (linear/block/tracing)
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
//! Block JIT tests — verify correctness of compiled loops and branches.

#![cfg(feature = "jit")]

use fusevm::{ChunkBuilder, JitCompiler, Op};

#[test]
fn block_jit_awk_sin_float_slot_matches_libm() {
    use fusevm::SlotKind;
    // slot0 = sin(slot0) where slot0 starts as f64 0.5 → sin(0.5).
    let mut b = ChunkBuilder::new();
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::AwkSin, 1);
    b.emit(Op::Dup, 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::Pop, 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(
        jit.is_block_eligible(&chunk),
        "AwkSin must be block-eligible"
    );

    let kinds = [SlotKind::Float];
    let mut slots = vec![0.5f64.to_bits() as i64; 1];
    let _ = jit
        .try_run_block_eager_kinded(&chunk, &mut slots, &kinds)
        .expect("AwkSin float-slot chunk must compile");
    assert_eq!(
        f64::from_bits(slots[0] as u64),
        0.5f64.sin(),
        "block JIT sin libcall must match libm"
    );
}

#[test]
fn block_jit_awk_atan2_float_slots_match_libm() {
    use fusevm::SlotKind;
    // slot0 = atan2(slot0, slot1); awk pushes y then x, so GetSlot(0)=y first,
    // GetSlot(1)=x second (x on top), matching Op::AwkAtan2's pop order.
    let mut b = ChunkBuilder::new();
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::AwkAtan2, 1);
    b.emit(Op::Dup, 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::Pop, 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let kinds = [SlotKind::Float, SlotKind::Float];
    let mut slots = vec![1.0f64.to_bits() as i64, 2.0f64.to_bits() as i64];
    let _ = jit
        .try_run_block_eager_kinded(&chunk, &mut slots, &kinds)
        .expect("AwkAtan2 float-slot chunk must compile");
    assert_eq!(
        f64::from_bits(slots[0] as u64),
        1.0f64.atan2(2.0),
        "block JIT atan2 libcall must match libm (y=1, x=2)"
    );
}

#[test]
fn block_jit_typed_returns_exact_float_result() {
    use fusevm::{BlockNum, SlotKind};
    // Chunk whose RESULT (top of stack at return) is a float:
    //   slot0 (f64) * 1.5 + 2.0, left on the operand stack.
    let mut b = ChunkBuilder::new();
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::LoadFloat(1.5), 1);
    b.emit(Op::Mul, 1);
    b.emit(Op::LoadFloat(2.0), 1);
    b.emit(Op::Add, 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let kinds = [SlotKind::Float];
    let mut slots = vec![4.0f64.to_bits() as i64; 1];
    let out = jit
        .try_run_block_eager_typed_kinded(&chunk, &mut slots, &kinds)
        .expect("float-result chunk must compile");
    match out {
        BlockNum::Float(v) => assert_eq!(v, 4.0 * 1.5 + 2.0, "exact float result preserved"),
        BlockNum::Int(n) => panic!("expected Float, got Int({n})"),
    }

    // The plain i64 entry point must still truncate the same result.
    let mut slots2 = vec![4.0f64.to_bits() as i64; 1];
    let truncated = jit
        .try_run_block_eager_kinded(&chunk, &mut slots2, &kinds)
        .expect("compile");
    assert_eq!(
        truncated,
        (4.0 * 1.5 + 2.0) as i64,
        "i64 entry truncates float"
    );
}

#[test]
fn block_jit_awk_div_mod_float_slots_compute() {
    use fusevm::SlotKind;
    // slot0 = OP(slot0, slot1). awk div/mod pop divisor (top) then dividend,
    // so GetSlot(0)=dividend first, GetSlot(1)=divisor second (divisor on top).
    let run = |op: Op| -> f64 {
        let mut b = ChunkBuilder::new();
        b.emit(Op::GetSlot(0), 1);
        b.emit(Op::GetSlot(1), 1);
        b.emit(op.clone(), 1);
        b.emit(Op::Dup, 1);
        b.emit(Op::SetSlot(0), 1);
        b.emit(Op::Pop, 1);
        let chunk = b.build();

        let jit = JitCompiler::new();
        assert!(
            jit.is_block_eligible(&chunk),
            "{op:?} must be block-eligible"
        );

        let kinds = [SlotKind::Float, SlotKind::Float];
        let mut slots = vec![17.0f64.to_bits() as i64, 5.0f64.to_bits() as i64];
        jit.try_run_block_eager_kinded(&chunk, &mut slots, &kinds)
            .unwrap_or_else(|| panic!("{op:?} float-slot chunk must compile"));
        f64::from_bits(slots[0] as u64)
    };

    assert_eq!(run(Op::AwkDivJit), 17.0 / 5.0, "17 / 5");
    assert_eq!(run(Op::AwkModJit), 17.0 % 5.0, "17 % 5");
}

#[test]
fn block_jit_awk_div_mod_nonzero_no_trap() {
    use fusevm::SlotKind;
    // A nonzero divisor must NOT set the trap channel: a subsequent VM run on a
    // div/mod chunk would observe no error (verified end-to-end via awkrs). Here
    // we only assert the compiled block returns the correct quotient without the
    // guarded early-exit firing (result is well-defined, not the sentinel).
    let mut b = ChunkBuilder::new();
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::AwkDivJit, 1);
    b.emit(Op::Dup, 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::Pop, 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    let kinds = [SlotKind::Float, SlotKind::Float];
    let mut slots = vec![10.0f64.to_bits() as i64, 4.0f64.to_bits() as i64];
    jit.try_run_block_eager_kinded(&chunk, &mut slots, &kinds)
        .expect("div chunk must compile");
    assert_eq!(f64::from_bits(slots[0] as u64), 2.5);
}

#[test]
fn block_jit_awk_and_or_xor_float_slots_match_scalar() {
    use fusevm::SlotKind;
    // slot0 = OP(slot0, slot1) with Float slots 12.0 and 10.0.
    // and(12,10)=8, or(12,10)=14, xor(12,10)=6 — pushed Int, stored as f64.
    let run = |op: Op| -> f64 {
        let mut b = ChunkBuilder::new();
        b.emit(Op::GetSlot(0), 1);
        b.emit(Op::GetSlot(1), 1);
        b.emit(op.clone(), 1);
        b.emit(Op::Dup, 1);
        b.emit(Op::SetSlot(0), 1);
        b.emit(Op::Pop, 1);
        let chunk = b.build();

        let jit = JitCompiler::new();
        assert!(
            jit.is_block_eligible(&chunk),
            "{op:?} must be block-eligible"
        );

        let kinds = [SlotKind::Float, SlotKind::Float];
        let mut slots = vec![12.0f64.to_bits() as i64, 10.0f64.to_bits() as i64];
        jit.try_run_block_eager_kinded(&chunk, &mut slots, &kinds)
            .unwrap_or_else(|| panic!("{op:?} float-slot chunk must compile"));
        f64::from_bits(slots[0] as u64)
    };

    assert_eq!(run(Op::AwkAnd(2)), 8.0, "and(12,10)");
    assert_eq!(run(Op::AwkOr(2)), 14.0, "or(12,10)");
    assert_eq!(run(Op::AwkXor(2)), 6.0, "xor(12,10)");
}

#[test]
fn block_jit_awk_and_saturates_like_awkrs() {
    use fusevm::SlotKind;
    // num_to_u64 = `n.trunc() as i64` saturates: a huge f64 → i64::MAX, and
    // and(huge, huge) = i64::MAX & i64::MAX = i64::MAX → that as f64. Verifies
    // the JIT uses `fcvt_to_sint_sat` (no trap on out-of-range).
    let mut b = ChunkBuilder::new();
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::AwkAnd(2), 1);
    b.emit(Op::Dup, 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::Pop, 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    let kinds = [SlotKind::Float];
    let mut slots = vec![1e30f64.to_bits() as i64];
    jit.try_run_block_eager_kinded(&chunk, &mut slots, &kinds)
        .expect("saturating and() must compile");
    // Rust reference: ((1e30_f64.trunc() as i64) & (1e30_f64.trunc() as i64)) as f64
    let want = ((1e30f64.trunc() as i64) & (1e30f64.trunc() as i64)) as f64;
    assert_eq!(f64::from_bits(slots[0] as u64), want);
}

#[test]
fn block_jit_sum_loop() {
    // sum = 0; i = 0; while (i < 100) { sum += i; i++ } → sum = 4950
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1); // sum = 0
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1); // i = 0
                               // ip=5: loop body
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::Add, 1);
    b.emit(Op::SetSlot(0), 1); // sum += i
    b.emit(Op::PreIncSlotVoid(1), 1); // i++
    b.emit(Op::SlotLtIntJumpIfFalse(1, 100, 12), 1);
    b.emit(Op::Jump(5), 1);
    // ip=12: exit
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    assert_eq!(result, 4950);
}

#[test]
fn block_jit_accum_sum_loop() {
    // AccumSumLoop(0, 1, 1000) → sum = 499500
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1);
    b.emit(Op::AccumSumLoop(0, 1, 1000), 1);
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    assert_eq!(result, 499500);
}

#[test]
fn block_jit_conditional() {
    // if (1) { slot[0] = 42 } else { slot[0] = 99 } → 42
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(1), 1); // condition = true
    b.emit(Op::JumpIfFalse(6), 1); // if false, goto else
                                   // then:
    b.emit(Op::LoadInt(42), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::Jump(8), 1); // skip else
                            // ip=6: else
    b.emit(Op::LoadInt(99), 1);
    b.emit(Op::SetSlot(0), 1);
    // ip=8: after
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    assert_eq!(result, 42);
}

#[test]
fn block_jit_conditional_false() {
    // if (0) { slot[0] = 42 } else { slot[0] = 99 } → 99
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1); // condition = false
    b.emit(Op::JumpIfFalse(6), 1);
    b.emit(Op::LoadInt(42), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::Jump(8), 1);
    // ip=6:
    b.emit(Op::LoadInt(99), 1);
    b.emit(Op::SetSlot(0), 1);
    // ip=8:
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    assert_eq!(result, 99);
}

#[test]
fn block_jit_ternary_merge_carries_stack() {
    // slot[0] = (cond ? 42 : 99); return slot[0]
    // The branch result is left on the operand stack and consumed AFTER the
    // control-flow merge — a value live across a basic-block boundary. This
    // exercises operand-stack values carried as Cranelift block params (both via
    // an explicit Jump and via fallthrough into the merge block).
    let build_chunk = |cond: i64| {
        let mut b = ChunkBuilder::new();
        b.emit(Op::PushFrame, 1);
        b.emit(Op::LoadInt(cond), 1); // ip1: condition
        b.emit(Op::JumpIfFalse(5), 1); // ip2: if false goto else (ip5)
        b.emit(Op::LoadInt(42), 1); // ip3: then value (left on stack)
        b.emit(Op::Jump(6), 1); // ip4: goto merge (ip6)
        b.emit(Op::LoadInt(99), 1); // ip5: else value (falls through to merge)
        b.emit(Op::SetSlot(0), 1); // ip6: merge — consume stack value
        b.emit(Op::GetSlot(0), 1); // ip7
        b.build()
    };

    let jit = JitCompiler::new();

    let chunk_t = build_chunk(1);
    assert!(jit.is_block_eligible(&chunk_t));
    let mut slots = vec![0i64; 4];
    let result = jit
        .try_run_block_eager(&chunk_t, &mut slots)
        .expect("ternary-merge chunk must block-JIT compile (true)");
    assert_eq!(result, 42, "cond=true must yield 42");

    let chunk_f = build_chunk(0);
    let mut slots = vec![0i64; 4];
    let result = jit
        .try_run_block_eager(&chunk_f, &mut slots)
        .expect("ternary-merge chunk must block-JIT compile (false)");
    assert_eq!(result, 99, "cond=false must yield 99");
}

#[test]
fn block_jit_ternary_as_return_jumps_to_end() {
    // return (cond ? 42 : 99) — the ternary result is the function's return value,
    // so the then-branch does a value-carrying Jump to the segment END (ops.len()),
    // and the else-branch falls through to it. Both merge at the implicit end block
    // with the value still on the operand stack. This pins cross-block stack carry
    // through a jump-to-end merge (the shape strykelang emits for ternary-bodied subs).
    let build_chunk = |cond: i64| {
        let mut b = ChunkBuilder::new();
        b.emit(Op::PushFrame, 1); // ip0
        b.emit(Op::LoadInt(cond), 1); // ip1
        b.emit(Op::JumpIfFalse(5), 1); // ip2: false -> else (ip5)
        b.emit(Op::LoadInt(42), 1); // ip3: then value
        b.emit(Op::Jump(6), 1); // ip4: -> end (ops.len() == 6)
        b.emit(Op::LoadInt(99), 1); // ip5: else value (falls through to end)
        b.build()
    };

    let jit = JitCompiler::new();

    let chunk_t = build_chunk(1);
    assert!(jit.is_block_eligible(&chunk_t));
    let mut slots = vec![0i64; 4];
    let result = jit
        .try_run_block_eager(&chunk_t, &mut slots)
        .expect("ternary-as-return chunk must block-JIT compile (true)");
    assert_eq!(result, 42);

    let chunk_f = build_chunk(0);
    let mut slots = vec![0i64; 4];
    let result = jit
        .try_run_block_eager(&chunk_f, &mut slots)
        .expect("ternary-as-return chunk must block-JIT compile (false)");
    assert_eq!(result, 99);
}

#[test]
fn block_jit_fused_backedge() {
    // i = 0; sum = 0; loop { sum += i; if (++i >= 50) break } → sum = 1225
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1); // sum = 0
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1); // i = 0
                               // ip=5: body
    b.emit(Op::AddAssignSlotVoid(0, 1), 1); // sum += i
    b.emit(Op::SlotIncLtIntJumpBack(1, 50, 5), 1); // i++; if i < 50 goto 5
                                                   // exit
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    assert_eq!(result, 1225);
}

#[test]
fn block_jit_ineligible_with_print() {
    let mut b = ChunkBuilder::new();
    b.emit(Op::LoadInt(42), 1);
    b.emit(Op::Print(1), 1); // Print is not block-eligible
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(!jit.is_block_eligible(&chunk));
}

#[test]
fn partial_jit_finds_eligible_region() {
    // A chunk with mixed eligible/ineligible ops:
    // [PushFrame, LoadInt(0), SetSlot(0), LoadInt(0), SetSlot(1),  // eligible: ip 0..5
    //  AccumSumLoop, GetSlot(0),                                   // eligible: continues
    //  Print(1),                                                    // INELIGIBLE: ip 7
    //  GetSlot(0)]                                                  // eligible (size 1, too small)
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1);
    b.emit(Op::AccumSumLoop(0, 1, 100), 1);
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::Print(1), 1); // ineligible
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(!jit.is_block_eligible(&chunk));
    let region = jit
        .find_jit_region(&chunk)
        .expect("should find eligible region");
    assert_eq!(region, (0, 7));
}

#[test]
fn partial_jit_compiles_extracted_region() {
    // Same as above — extract the eligible region and JIT-compile it.
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1);
    b.emit(Op::AccumSumLoop(0, 1, 100), 1);
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::Print(1), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    let (start, end) = jit.find_jit_region(&chunk).unwrap();
    let sub_chunk = jit.extract_region(&chunk, start, end);

    assert!(jit.is_block_eligible(&sub_chunk));
    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&sub_chunk, &mut slots).unwrap();
    assert_eq!(result, 4950); // sum 0..100
}

#[test]
fn partial_jit_rebases_jumps() {
    // Region with internal jumps — verify they're rebased to local indices.
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadInt(1), 1); // condition
    b.emit(Op::JumpIfFalse(7), 1); // ip=4, target ip=7
    b.emit(Op::LoadInt(42), 1);
    b.emit(Op::SetSlot(0), 1);
    // ip=7
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    let (start, end) = jit.find_jit_region(&chunk).unwrap();
    let sub_chunk = jit.extract_region(&chunk, start, end);

    // Find the JumpIfFalse in sub_chunk and verify target was rebased
    for op in &sub_chunk.ops {
        if let Op::JumpIfFalse(t) = op {
            assert_eq!(*t, 7 - start);
        }
    }
    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&sub_chunk, &mut slots).unwrap();
    assert_eq!(result, 42);
}

#[test]
fn block_jit_slots_written_back() {
    // Verify slots are modified in-place after the JIT runs
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1);
    b.emit(Op::AccumSumLoop(0, 1, 10), 1);
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    let mut slots = vec![0i64; 4];
    let _ = jit.try_run_block_eager(&chunk, &mut slots);
    assert_eq!(slots[0], 45); // sum 0..10
    assert_eq!(slots[1], 10); // i after loop
}

/// A unique block-eligible sum loop (limit picks the op_hash so the per-thread
/// block cache entry doesn't collide with other tests on the same thread).
fn unique_sum_loop(limit: i32) -> fusevm::Chunk {
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadInt(0), 1);
    b.emit(Op::SetSlot(1), 1);
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::Add, 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::PreIncSlotVoid(1), 1);
    b.emit(Op::SlotLtIntJumpIfFalse(1, limit, 12), 1);
    b.emit(Op::Jump(5), 1);
    b.emit(Op::GetSlot(0), 1);
    b.build()
}

#[test]
fn block_threshold_is_configurable() {
    use fusevm::TraceJitConfig;
    let jit = JitCompiler::new();

    // Lower the block warmup to 1: the chunk must stay interpreted (None) on
    // the first call and compile (Some) on the second.
    jit.set_config(TraceJitConfig {
        block_threshold: 1,
        ..TraceJitConfig::defaults()
    });
    let chunk = unique_sum_loop(37);
    let mut slots = vec![0i64; 4];
    assert_eq!(
        jit.try_run_block(&chunk, &mut slots),
        None,
        "first call is below threshold 1"
    );
    assert_eq!(
        jit.try_run_block(&chunk, &mut slots),
        Some(666),
        "second call should compile and run with block_threshold=1"
    );

    // With an explicitly higher threshold, a different chunk must still be None
    // on its second call — proving the knob (not the compiled default) drives
    // tier selection. Uses an explicit value so the test is independent of
    // whatever the shipped default `block_threshold` happens to be.
    jit.set_config(TraceJitConfig {
        block_threshold: 5,
        ..TraceJitConfig::defaults()
    });
    let chunk2 = unique_sum_loop(38);
    let mut slots2 = vec![0i64; 4];
    assert_eq!(jit.try_run_block(&chunk2, &mut slots2), None);
    assert_eq!(jit.try_run_block(&chunk2, &mut slots2), None);
}

#[test]
fn block_jit_awk_int_truncates_float() {
    // slot0 = int(3.7) → 3.0 ; slot1 = int(-2.9) → -2.0 ; return slot0
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadFloat(3.7), 1);
    b.emit(Op::AwkInt, 1);
    b.emit(Op::SetSlot(0), 1);
    b.emit(Op::LoadFloat(-2.9), 1);
    b.emit(Op::AwkInt, 1);
    b.emit(Op::SetSlot(1), 1);
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(
        jit.is_block_eligible(&chunk),
        "AwkInt must be block-eligible"
    );

    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    // int() yields an integral value; the block JIT returns it as a plain i64.
    assert_eq!(result, 3, "int(3.7) == 3");
    assert_eq!(slots[1], -2, "int(-2.9) == -2 (toward zero)");
}

#[test]
fn block_jit_awk_int_in_loop_matches_scalar() {
    // s = 0; for (i = 0; i < 10; i++) s += int(i + 0.9); → s = 0+1+..+9 = 45
    let mut b = ChunkBuilder::new();
    b.emit(Op::PushFrame, 1);
    b.emit(Op::LoadFloat(0.0), 1);
    b.emit(Op::SetSlot(0), 1); // s = 0.0
    b.emit(Op::LoadFloat(0.0), 1);
    b.emit(Op::SetSlot(1), 1); // i = 0.0
                               // ip=5: body
    b.emit(Op::GetSlot(0), 1);
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::LoadFloat(0.9), 1);
    b.emit(Op::Add, 1);
    b.emit(Op::AwkInt, 1); // int(i + 0.9) == i
    b.emit(Op::Add, 1);
    b.emit(Op::SetSlot(0), 1); // s += int(i + 0.9)
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::LoadFloat(1.0), 1);
    b.emit(Op::Add, 1);
    b.emit(Op::SetSlot(1), 1); // i += 1
    b.emit(Op::GetSlot(1), 1);
    b.emit(Op::LoadFloat(10.0), 1);
    b.emit(Op::NumLt, 1);
    b.emit(Op::JumpIfTrue(5), 1);
    b.emit(Op::GetSlot(0), 1);
    let chunk = b.build();

    let jit = JitCompiler::new();
    assert!(jit.is_block_eligible(&chunk));

    let mut slots = vec![0i64; 4];
    let result = jit.try_run_block_eager(&chunk, &mut slots).unwrap();
    assert_eq!(result, 45);
}