summaryrefslogtreecommitdiff
path: root/backends/lean/Base/Primitives/Scalar.lean
blob: aaa4027f365b41f0ab23257d5b7444233b489285 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
import Lean
import Lean.Meta.Tactic.Simp
import Mathlib.Tactic.Linarith
import Base.Primitives.Base
import Base.Diverge.Base
import Base.Progress.Base
import Base.Arith.Int

namespace Primitives

----------------------
-- MACHINE INTEGERS --
----------------------

-- We redefine our machine integers types.

-- For Isize/Usize, we reuse `getNumBits` from `USize`. You cannot reduce `getNumBits`
-- using the simplifier, meaning that proofs do not depend on the compile-time value of
-- USize.size. (Lean assumes 32 or 64-bit platforms, and Rust doesn't really support, at
-- least officially, 16-bit microcontrollers, so this seems like a fine design decision
-- for now.)

-- Note from Chris Bailey: "If there's more than one salient property of your
-- definition then the subtyping strategy might get messy, and the property part
-- of a subtype is less discoverable by the simplifier or tactics like
-- library_search." So, we will not add refinements on the return values of the
-- operations defined on Primitives, but will rather rely on custom lemmas to
-- invert on possible return values of the primitive operations.

-- Machine integer constants, done via `ofNatCore`, which requires a proof that
-- the `Nat` fits within the desired integer type. We provide a custom tactic.

open Result Error
open System.Platform.getNumBits

-- TODO: is there a way of only importing System.Platform.getNumBits?
--
@[simp] def size_num_bits : Nat := (System.Platform.getNumBits ()).val

-- Remark: Lean seems to use < for the comparisons with the upper bounds by convention.

-- The "structured" bounds
def Isize.smin : Int := - (HPow.hPow 2 (size_num_bits - 1))
def Isize.smax : Int := (HPow.hPow 2 (size_num_bits - 1)) - 1
def I8.smin    : Int := - (HPow.hPow 2 7)
def I8.smax    : Int := HPow.hPow 2 7 - 1
def I16.smin   : Int := - (HPow.hPow 2 15)
def I16.smax   : Int := HPow.hPow 2 15 - 1
def I32.smin   : Int := -(HPow.hPow 2 31)
def I32.smax   : Int := HPow.hPow 2 31 - 1
def I64.smin   : Int := -(HPow.hPow 2 63)
def I64.smax   : Int := HPow.hPow 2 63 - 1
def I128.smin  : Int := -(HPow.hPow 2 127)
def I128.smax  : Int := HPow.hPow 2 127 - 1
def Usize.smin : Int := 0
def Usize.smax : Int := HPow.hPow 2 size_num_bits - 1
def U8.smin    : Int := 0
def U8.smax    : Int := HPow.hPow 2 8 - 1
def U16.smin   : Int := 0
def U16.smax   : Int := HPow.hPow 2 16 - 1
def U32.smin   : Int := 0
def U32.smax   : Int := HPow.hPow 2 32 - 1
def U64.smin   : Int := 0
def U64.smax   : Int := HPow.hPow 2 64 - 1
def U128.smin  : Int := 0
def U128.smax  : Int := HPow.hPow 2 128 - 1

-- The "normalized" bounds, that we use in practice
def I8.min    := -128
def I8.max   := 127
def I16.min  := -32768
def I16.max  := 32767
def I32.min  := -2147483648
def I32.max  := 2147483647
def I64.min  := -9223372036854775808
def I64.max  := 9223372036854775807
def I128.min := -170141183460469231731687303715884105728
def I128.max := 170141183460469231731687303715884105727
@[simp] def U8.min   := 0
def U8.max   := 255
@[simp] def U16.min  := 0
def U16.max  := 65535
@[simp] def U32.min  := 0
def U32.max  := 4294967295
@[simp] def U64.min  := 0
def U64.max  := 18446744073709551615
@[simp] def U128.min := 0
def U128.max := 340282366920938463463374607431768211455
@[simp] def Usize.min := 0

def Isize.refined_min : { n:Int // n = I32.min  n = I64.min } :=
   Isize.smin, by
    simp [Isize.smin]
    cases System.Platform.numBits_eq <;>
    unfold System.Platform.numBits at * <;> simp [*] 

def Isize.refined_max : { n:Int // n = I32.max  n = I64.max } :=
   Isize.smax, by
    simp [Isize.smax]
    cases System.Platform.numBits_eq <;>
    unfold System.Platform.numBits at * <;> simp [*] 

def Usize.refined_max : { n:Int // n = U32.max  n = U64.max } :=
   Usize.smax, by
    simp [Usize.smax]
    cases System.Platform.numBits_eq <;>
    unfold System.Platform.numBits at * <;> simp [*] 

def Isize.min := Isize.refined_min.val
def Isize.max := Isize.refined_max.val
def Usize.max := Usize.refined_max.val

inductive ScalarTy :=
| Isize
| I8
| I16
| I32
| I64
| I128
| Usize
| U8
| U16
| U32
| U64
| U128

def ScalarTy.isSigned (ty : ScalarTy) : Bool :=
  match ty with
  | Isize
  | I8
  | I16
  | I32
  | I64
  | I128 => true
  | Usize
  | U8
  | U16
  | U32
  | U64
  | U128 => false


def Scalar.smin (ty : ScalarTy) : Int :=
  match ty with
  | .Isize => Isize.smin
  | .I8    => I8.smin
  | .I16   => I16.smin
  | .I32   => I32.smin
  | .I64   => I64.smin
  | .I128  => I128.smin
  | .Usize => Usize.smin
  | .U8    => U8.smin
  | .U16   => U16.smin
  | .U32   => U32.smin
  | .U64   => U64.smin
  | .U128  => U128.smin

def Scalar.smax (ty : ScalarTy) : Int :=
  match ty with
  | .Isize => Isize.smax
  | .I8    => I8.smax
  | .I16   => I16.smax
  | .I32   => I32.smax
  | .I64   => I64.smax
  | .I128  => I128.smax
  | .Usize => Usize.smax
  | .U8    => U8.smax
  | .U16   => U16.smax
  | .U32   => U32.smax
  | .U64   => U64.smax
  | .U128  => U128.smax

def Scalar.min (ty : ScalarTy) : Int :=
  match ty with
  | .Isize => Isize.min
  | .I8    => I8.min
  | .I16   => I16.min
  | .I32   => I32.min
  | .I64   => I64.min
  | .I128  => I128.min
  | .Usize => Usize.min
  | .U8    => U8.min
  | .U16   => U16.min
  | .U32   => U32.min
  | .U64   => U64.min
  | .U128  => U128.min

def Scalar.max (ty : ScalarTy) : Int :=
  match ty with
  | .Isize => Isize.max
  | .I8    => I8.max
  | .I16   => I16.max
  | .I32   => I32.max
  | .I64   => I64.max
  | .I128  => I128.max
  | .Usize => Usize.max
  | .U8    => U8.max
  | .U16   => U16.max
  | .U32   => U32.max
  | .U64   => U64.max
  | .U128  => U128.max

def Scalar.smin_eq (ty : ScalarTy) : Scalar.min ty = Scalar.smin ty := by
  cases ty <;> rfl

def Scalar.smax_eq (ty : ScalarTy) : Scalar.max ty = Scalar.smax ty := by
  cases ty <;> rfl

-- "Conservative" bounds
-- We use those because we can't compare to the isize bounds (which can't
-- reduce at compile-time). Whenever we perform an arithmetic operation like
-- addition we need to check that the result is in bounds: we first compare
-- to the conservative bounds, which reduce, then compare to the real bounds.
-- This is useful for the various #asserts that we want to reduce at
-- type-checking time.
def Scalar.cMin (ty : ScalarTy) : Int :=
  match ty with
  | .Isize => Scalar.min .I32
  | _ => Scalar.min ty

def Scalar.cMax (ty : ScalarTy) : Int :=
  match ty with
  | .Isize => Scalar.max .I32
  | .Usize => Scalar.max .U32
  | _ => Scalar.max ty

theorem Scalar.cMin_bound ty : Scalar.min ty  Scalar.cMin ty := by
  cases ty <;> simp [Scalar.min, Scalar.max, Scalar.cMin, Scalar.cMax] at *
  have h := Isize.refined_min.property
  cases h <;> simp [*, Isize.min]

theorem Scalar.cMax_bound ty : Scalar.cMax ty  Scalar.max ty := by
  cases ty <;> simp [Scalar.min, Scalar.max, Scalar.cMin, Scalar.cMax] at *
  . have h := Isize.refined_max.property
    cases h <;> simp [*, Isize.max]
  . have h := Usize.refined_max.property
    cases h <;> simp [*, Usize.max]

theorem Scalar.cMin_suffices ty (h : Scalar.cMin ty  x) : Scalar.min ty  x := by
  have := Scalar.cMin_bound ty
  linarith

theorem Scalar.cMax_suffices ty (h : x  Scalar.cMax ty) : x  Scalar.max ty := by
  have := Scalar.cMax_bound ty
  linarith

structure Scalar (ty : ScalarTy) where
  val : Int
  hmin : Scalar.min ty  val
  hmax : val  Scalar.max ty
deriving Repr

theorem Scalar.bound_suffices (ty : ScalarTy) (x : Int) :
  Scalar.cMin ty  x  x  Scalar.cMax ty ->
  Scalar.min ty  x  x  Scalar.max ty
  :=
  λ h => by
  apply And.intro <;> have hmin := Scalar.cMin_bound ty <;> have hmax := Scalar.cMax_bound ty <;> linarith

def Scalar.ofIntCore {ty : ScalarTy} (x : Int)
  (hmin : Scalar.min ty  x) (hmax : x  Scalar.max ty) : Scalar ty :=
  { val := x, hmin := hmin, hmax := hmax }

-- Tactic to prove that integers are in bounds
-- TODO: use this: https://leanprover.zulipchat.com/#narrow/stream/270676-lean4/topic/instance.20with.20tactic.20autoparam
syntax "intlit" : tactic
macro_rules
  | `(tactic| intlit) => `(tactic| apply Scalar.bound_suffices; decide)

def Scalar.ofInt {ty : ScalarTy} (x : Int)
  (h : Scalar.min ty  x  x  Scalar.max ty := by intlit) : Scalar ty :=
  -- Remark: we initially wrote:
  --  let ⟨ hmin, hmax ⟩ := h
  --  Scalar.ofIntCore x hmin hmax
  -- We updated to the line below because a similar pattern in `Scalar.tryMk`
  -- made reduction block. Both versions seem to work for `Scalar.ofInt`, though.
  -- TODO: investigate
  Scalar.ofIntCore x h.left h.right

@[simp] def Scalar.check_bounds (ty : ScalarTy) (x : Int) : Bool :=
  (Scalar.cMin ty  x || Scalar.min ty  x)  (x  Scalar.cMax ty || x  Scalar.max ty)

theorem Scalar.check_bounds_prop {ty : ScalarTy} {x : Int} (h: Scalar.check_bounds ty x) :
  Scalar.min ty  x  x  Scalar.max ty := by
  simp at *
  have  hmin, hmax  := h
  have hbmin := Scalar.cMin_bound ty
  have hbmax := Scalar.cMax_bound ty
  cases hmin <;> cases hmax <;> apply And.intro <;> linarith

-- Further thoughts: look at what has been done here:
-- https://github.com/leanprover-community/mathlib4/blob/master/Mathlib/Data/Fin/Basic.lean
-- and
-- https://github.com/leanprover-community/mathlib4/blob/master/Mathlib/Data/UInt.lean
-- which both contain a fair amount of reasoning already!
def Scalar.tryMk (ty : ScalarTy) (x : Int) : Result (Scalar ty) :=
  if h:Scalar.check_bounds ty x then
    -- If we do:
    -- ```
    -- let ⟨ hmin, hmax ⟩ := (Scalar.check_bounds_prop h)
    -- Scalar.ofIntCore x hmin hmax
    -- ```
    -- then normalization blocks (for instance, some proofs which use reflexivity fail).
    -- However, the version below doesn't block reduction (TODO: investigate):
    return Scalar.ofInt x (Scalar.check_bounds_prop h)
  else fail integerOverflow

def Scalar.neg {ty : ScalarTy} (x : Scalar ty) : Result (Scalar ty) := Scalar.tryMk ty (- x.val)

-- Our custom remainder operation, which satisfies the semantics of Rust
-- TODO: is there a better way?
def scalar_rem (x y : Int) : Int :=
  if 0  x then x % y
  else - (|x| % |y|)

@[simp]
def scalar_rem_nonneg {x y : Int} (hx : 0  x) : scalar_rem x y = x % y := by
  intros
  simp [*, scalar_rem]

-- Our custom division operation, which satisfies the semantics of Rust
-- TODO: is there a better way?
def scalar_div (x y : Int) : Int :=
  if 0  x && 0  y then x / y
  else if 0  x && y < 0 then - (|x| / |y|)
  else if x < 0 && 0  y then - (|x| / |y|)
  else |x| / |y|

@[simp]
def scalar_div_nonneg {x y : Int} (hx : 0  x) (hy : 0  y) : scalar_div x y = x / y := by
  intros
  simp [*, scalar_div]

-- Checking that the remainder operation is correct
#assert scalar_rem 1 2 = 1
#assert scalar_rem (-1) 2 = -1
#assert scalar_rem 1 (-2) = 1
#assert scalar_rem (-1) (-2) = -1
#assert scalar_rem 7 3 = (1:Int)
#assert scalar_rem (-7) 3 = -1
#assert scalar_rem 7 (-3) = 1
#assert scalar_rem (-7) (-3) = -1

-- Checking that the division operation is correct
#assert scalar_div 3 2 = 1
#assert scalar_div (-3) 2 = -1
#assert scalar_div 3 (-2) = -1
#assert scalar_div (-3) (-2) = 1
#assert scalar_div 7 3 = 2
#assert scalar_div (-7) 3 = -2
#assert scalar_div 7 (-3) = -2
#assert scalar_div (-7) (-3) = 2

def Scalar.div {ty : ScalarTy} (x : Scalar ty) (y : Scalar ty) : Result (Scalar ty) :=
  if y.val != 0 then Scalar.tryMk ty (scalar_div x.val y.val) else fail divisionByZero

def Scalar.rem {ty : ScalarTy} (x : Scalar ty) (y : Scalar ty) : Result (Scalar ty) :=
  if y.val != 0 then Scalar.tryMk ty (scalar_rem x.val y.val) else fail divisionByZero

def Scalar.add {ty : ScalarTy} (x : Scalar ty) (y : Scalar ty) : Result (Scalar ty) :=
  Scalar.tryMk ty (x.val + y.val)

def Scalar.sub {ty : ScalarTy} (x : Scalar ty) (y : Scalar ty) : Result (Scalar ty) :=
  Scalar.tryMk ty (x.val - y.val)

def Scalar.mul {ty : ScalarTy} (x : Scalar ty) (y : Scalar ty) : Result (Scalar ty) :=
  Scalar.tryMk ty (x.val * y.val)

-- TODO: instances of +, -, * etc. for scalars

-- Cast an integer from a [src_ty] to a [tgt_ty]
-- TODO: check the semantics of casts in Rust
def Scalar.cast {src_ty : ScalarTy} (tgt_ty : ScalarTy) (x : Scalar src_ty) : Result (Scalar tgt_ty) :=
  Scalar.tryMk tgt_ty x.val

-- The scalar types
-- We declare the definitions as reducible so that Lean can unfold them (useful
-- for type class resolution for instance).
@[reducible] def Isize := Scalar .Isize
@[reducible] def I8    := Scalar .I8
@[reducible] def I16   := Scalar .I16
@[reducible] def I32   := Scalar .I32
@[reducible] def I64   := Scalar .I64
@[reducible] def I128  := Scalar .I128
@[reducible] def Usize := Scalar .Usize
@[reducible] def U8    := Scalar .U8
@[reducible] def U16   := Scalar .U16
@[reducible] def U32   := Scalar .U32
@[reducible] def U64   := Scalar .U64
@[reducible] def U128  := Scalar .U128

-- TODO: below: not sure this is the best way.
-- Should we rather overload operations like +, -, etc.?
-- Also, it is possible to automate the generation of those definitions
-- with macros (but would it be a good idea? It would be less easy to
-- read the file, which is not supposed to change a lot)

-- Negation

/--
Remark: there is no heterogeneous negation in the Lean prelude: we thus introduce
one here.

The notation typeclass for heterogeneous addition.
This enables the notation `- a : β` where `a : α`.
-/
class HNeg (α : Type u) (β : outParam (Type v)) where
  /-- `- a` computes the negation of `a`.
  The meaning of this notation is type-dependent. -/
  hNeg : α  β

prefix:75  "-"   => HNeg.hNeg

instance : HNeg Isize (Result Isize) where hNeg x := Scalar.neg x
instance : HNeg I8 (Result I8) where hNeg x := Scalar.neg x
instance : HNeg I16 (Result I16) where hNeg x := Scalar.neg x
instance : HNeg I32 (Result I32) where hNeg x := Scalar.neg x
instance : HNeg I64 (Result I64) where hNeg x := Scalar.neg x
instance : HNeg I128 (Result I128) where hNeg x := Scalar.neg x

-- Addition
instance {ty} : HAdd (Scalar ty) (Scalar ty) (Result (Scalar ty)) where
  hAdd x y := Scalar.add x y

-- Substraction
instance {ty} : HSub (Scalar ty) (Scalar ty) (Result (Scalar ty)) where
  hSub x y := Scalar.sub x y

-- Multiplication
instance {ty} : HMul (Scalar ty) (Scalar ty) (Result (Scalar ty)) where
  hMul x y := Scalar.mul x y

-- Division
instance {ty} : HDiv (Scalar ty) (Scalar ty) (Result (Scalar ty)) where
  hDiv x y := Scalar.div x y

-- Remainder
instance {ty} : HMod (Scalar ty) (Scalar ty) (Result (Scalar ty)) where
  hMod x y := Scalar.rem x y

-- TODO: make progress work at a more fine grained level (see `Scalar.add_unsigned_spec`)
@[cpspec]
theorem Scalar.add_spec {ty} {x y : Scalar ty}
  (hmin : Scalar.min ty  x.val + y.val)
  (hmax : x.val + y.val  Scalar.max ty) :
   z, x + y = ret z  z.val = x.val + y.val := by
  simp [HAdd.hAdd, add, Add.add]
  simp [tryMk]
  split
  . simp [pure]
    rfl
  . tauto

theorem Scalar.add_unsigned_spec {ty} (s: ¬ ty.isSigned) {x y : Scalar ty}
  (hmax : x.val + y.val  Scalar.max ty) :
   z, x + y = ret z  z.val = x.val + y.val := by
  have hmin : Scalar.min ty  x.val + y.val := by
    have hx := x.hmin
    have hy := y.hmin
    cases ty <;> simp [min] at * <;> linarith
  apply add_spec <;> assumption

-- TODO: make it finer grained
@[cpspec]
theorem Scalar.sub_spec {ty} {x y : Scalar ty}
  (hmin : Scalar.min ty  x.val - y.val)
  (hmax : x.val - y.val  Scalar.max ty) :
   z, x - y = ret z  z.val = x.val - y.val := by
  simp [HSub.hSub, sub, Sub.sub]
  simp [tryMk]
  split
  . simp [pure]
    rfl
  . tauto

theorem Scalar.sub_unsigned_spec {ty} (s: ¬ ty.isSigned) {x y : Scalar ty}
  (hmin : Scalar.min ty  x.val - y.val) :
   z, x - y = ret z  z.val = x.val - y.val := by
  have : x.val - y.val  Scalar.max ty := by
    have hx := x.hmin
    have hxm := x.hmax
    have hy := y.hmin
    cases ty <;> simp [min, max] at * <;> linarith
  intros
  apply sub_spec <;> assumption

-- TODO: make it finer grained
@[cpspec]
theorem Scalar.mul_spec {ty} {x y : Scalar ty}
  (hmin : Scalar.min ty  x.val * y.val)
  (hmax : x.val * y.val  Scalar.max ty) :
   z, x * y = ret z  z.val = x.val * y.val := by
  simp [HMul.hMul, mul, Mul.mul]
  simp [tryMk]
  split
  . simp [pure]
    rfl
  . tauto

theorem Scalar.mul_unsigned_spec {ty} (s: ¬ ty.isSigned) {x y : Scalar ty}
  (hmax : x.val * y.val  Scalar.max ty) :
   z, x * y = ret z  z.val = x.val * y.val := by
  have : Scalar.min ty  x.val * y.val := by
    have hx := x.hmin
    have hy := y.hmin
    cases ty <;> simp at * <;> apply mul_nonneg hx hy
  apply mul_spec <;> assumption

-- TODO: make it finer grained
@[cpspec]
theorem Scalar.div_spec {ty} {x y : Scalar ty}
  (hnz : y.val  0)
  (hmin : Scalar.min ty  scalar_div x.val y.val)
  (hmax : scalar_div x.val y.val  Scalar.max ty) :
   z, x / y = ret z  z.val = scalar_div x.val y.val := by
  simp [HDiv.hDiv, div, Div.div]
  simp [tryMk, *]
  simp [pure]
  rfl

theorem Scalar.div_unsigned_spec {ty} (s: ¬ ty.isSigned) (x : Scalar ty) {y : Scalar ty}
  (hnz : y.val  0) :
   z, x / y = ret z  z.val = x.val / y.val := by
  have h : Scalar.min ty = 0 := by cases ty <;> simp at *
  have hx := x.hmin
  have hy := y.hmin
  simp [h] at hx hy
  have hmin : 0  x.val / y.val := Int.ediv_nonneg hx hy
  have hmax : x.val / y.val  Scalar.max ty := by
    have := Int.ediv_le_self y.val hx
    have := x.hmax
    linarith
  have hs := @div_spec ty x y hnz
  simp [*] at hs
  apply hs

-- TODO: make it finer grained
@[cpspec]
theorem Scalar.rem_spec {ty} {x y : Scalar ty}
  (hnz : y.val  0)
  (hmin : Scalar.min ty  scalar_rem x.val y.val)
  (hmax : scalar_rem x.val y.val  Scalar.max ty) :
   z, x % y = ret z  z.val = scalar_rem x.val y.val := by
  simp [HMod.hMod, rem]
  simp [tryMk, *]
  simp [pure]
  rfl

theorem Scalar.rem_unsigned_spec {ty} (s: ¬ ty.isSigned) (x : Scalar ty) {y : Scalar ty}
  (hnz : y.val  0) :
   z, x % y = ret z  z.val = scalar_rem x.val y.val := by
  have h : Scalar.min ty = 0 := by cases ty <;> simp at *
  have hx := x.hmin
  have hy := y.hmin
  simp [h] at hx hy
  have hmin : 0  x.val % y.val := Int.emod_nonneg x.val hnz
  have hmax : x.val % y.val  Scalar.max ty := by
    have h := @Int.ediv_emod_unique x.val y.val (x.val % y.val) (x.val / y.val)
    simp at h
    have : 0 < y.val := by int_tac
    simp [*] at h
    have := y.hmax
    linarith
  have hs := @rem_spec ty x y hnz
  simp [*] at hs
  simp [*]

-- ofIntCore
-- TODO: typeclass?
def Isize.ofIntCore := @Scalar.ofIntCore .Isize
def I8.ofIntCore    := @Scalar.ofIntCore .I8
def I16.ofIntCore   := @Scalar.ofIntCore .I16
def I32.ofIntCore   := @Scalar.ofIntCore .I32
def I64.ofIntCore   := @Scalar.ofIntCore .I64
def I128.ofIntCore  := @Scalar.ofIntCore .I128
def Usize.ofIntCore := @Scalar.ofIntCore .Usize
def U8.ofIntCore    := @Scalar.ofIntCore .U8
def U16.ofIntCore   := @Scalar.ofIntCore .U16
def U32.ofIntCore   := @Scalar.ofIntCore .U32
def U64.ofIntCore   := @Scalar.ofIntCore .U64
def U128.ofIntCore  := @Scalar.ofIntCore .U128

--  ofInt
-- TODO: typeclass?
def Isize.ofInt := @Scalar.ofInt .Isize
def I8.ofInt    := @Scalar.ofInt .I8
def I16.ofInt   := @Scalar.ofInt .I16
def I32.ofInt   := @Scalar.ofInt .I32
def I64.ofInt   := @Scalar.ofInt .I64
def I128.ofInt  := @Scalar.ofInt .I128
def Usize.ofInt := @Scalar.ofInt .Usize
def U8.ofInt    := @Scalar.ofInt .U8
def U16.ofInt   := @Scalar.ofInt .U16
def U32.ofInt   := @Scalar.ofInt .U32
def U64.ofInt   := @Scalar.ofInt .U64
def U128.ofInt  := @Scalar.ofInt .U128

-- Comparisons
instance {ty} : LT (Scalar ty) where
  lt a b := LT.lt a.val b.val

instance {ty} : LE (Scalar ty) where le a b := LE.le a.val b.val

instance Scalar.decLt {ty} (a b : Scalar ty) : Decidable (LT.lt a b) := Int.decLt ..
instance Scalar.decLe {ty} (a b : Scalar ty) : Decidable (LE.le a b) := Int.decLe ..

theorem Scalar.eq_of_val_eq {ty} :  {i j : Scalar ty}, Eq i.val j.val  Eq i j
  | _, _, _⟩, _, _, _⟩, rfl => rfl

theorem Scalar.val_eq_of_eq {ty} {i j : Scalar ty} (h : Eq i j) : Eq i.val j.val :=
  h  rfl

theorem Scalar.ne_of_val_ne {ty} {i j : Scalar ty} (h : Not (Eq i.val j.val)) : Not (Eq i j) :=
  fun h' => absurd (val_eq_of_eq h') h

instance (ty : ScalarTy) : DecidableEq (Scalar ty) :=
  fun i j =>
    match decEq i.val j.val with
    | isTrue h  => isTrue (Scalar.eq_of_val_eq h)
    | isFalse h => isFalse (Scalar.ne_of_val_ne h)

/- Remark: we can't write the following instance because of restrictions about
   the type class parameters (`ty` doesn't appear in the return type, which is
   forbidden):

   ```
   instance Scalar.cast (ty : ScalarTy) : Coe (Scalar ty) Int where coe := λ v => v.val
   ```
 -/
def Scalar.toInt {ty} (n : Scalar ty) : Int := n.val

-- -- We now define a type class that subsumes the various machine integer types, so
-- -- as to write a concise definition for scalar_cast, rather than exhaustively
-- -- enumerating all of the possible pairs. We remark that Rust has sane semantics
-- -- and fails if a cast operation would involve a truncation or modulo.

-- class MachineInteger (t: Type) where
--   size: Nat
--   val: t -> Fin size
--   ofNatCore: (n:Nat) -> LT.lt n size -> t

-- set_option hygiene false in
-- run_cmd
--   for typeName in [`UInt8, `UInt16, `UInt32, `UInt64, `USize].map Lean.mkIdent do
--   Lean.Elab.Command.elabCommand (← `(
--     namespace $typeName
--     instance: MachineInteger $typeName where
--       size := size
--       val := val
--       ofNatCore := ofNatCore
--     end $typeName
--   ))

-- -- Aeneas only instantiates the destination type (`src` is implicit). We rely on
-- -- Lean to infer `src`.

-- def scalar_cast { src: Type } (dst: Type) [ MachineInteger src ] [ MachineInteger dst ] (x: src): Result dst :=
--   if h: MachineInteger.val x < MachineInteger.size dst then
--     .ret (MachineInteger.ofNatCore (MachineInteger.val x).val h)
--   else
--     .fail integerOverflow

end Primitives