fixed

Pure Haskell large fixed-width integers and Montgomery arithmetic (docs.ppad.tech/fixed).
git clone git://git.ppad.tech/fixed.git
Log | Files | Refs | README | LICENSE

Curve.hs (37926B)


      1 {-# LANGUAGE BangPatterns #-}
      2 {-# LANGUAGE MagicHash #-}
      3 {-# LANGUAGE NumericUnderscores #-}
      4 {-# LANGUAGE PatternSynonyms #-}
      5 {-# LANGUAGE ViewPatterns #-}
      6 {-# LANGUAGE UnboxedSums #-}
      7 {-# LANGUAGE UnboxedTuples #-}
      8 {-# LANGUAGE UnliftedNewtypes #-}
      9 
     10 -- |
     11 -- Module: Numeric.Montgomery.Secp256k1.Curve
     12 -- Copyright: (c) 2025 Jared Tobin
     13 -- License: MIT
     14 -- Maintainer: Jared Tobin <jared@ppad.tech>
     15 --
     16 -- Montgomery form 'Wider' words, as well as arithmetic operations, with
     17 -- domain derived from the secp256k1 elliptic curve field prime.
     18 
     19 module Numeric.Montgomery.Secp256k1.Curve (
     20   -- * Montgomery form, secp256k1 field prime modulus
     21     Montgomery(..)
     22   , render
     23   , to
     24   , from
     25   , zero
     26   , one
     27 
     28   -- * Comparison
     29   , eq
     30   , eq_vartime
     31 
     32   -- * Reduction and retrieval
     33   , redc
     34   , redc#
     35   , retr
     36   , retr#
     37 
     38   -- * Constant-time selection
     39   , select
     40   , select#
     41 
     42   -- * Montgomery arithmetic
     43   , add
     44   , add#
     45   , sub
     46   , sub#
     47   , mul
     48   , mul#
     49   , sqr
     50   , sqr#
     51   , neg
     52   , neg#
     53   , inv
     54   , inv#
     55   , sqrt_vartime
     56   , sqrt#
     57   , exp
     58   , exp#
     59   , odd#
     60   , odd_vartime
     61   ) where
     62 
     63 import Control.DeepSeq
     64 import qualified Data.Choice as C
     65 import Data.Word.Limb (Limb(..))
     66 import qualified Data.Word.Limb as L
     67 import qualified Data.Word.Wide as W
     68 import Data.Word.Wider (Wider(..))
     69 import qualified Data.Word.Wider as WW
     70 import GHC.Exts (Word(..), Word#)
     71 import Prelude hiding (or, and, not, sqrt, exp)
     72 
     73 -- montgomery arithmetic, specialized to the secp256k1 field prime modulus
     74 -- 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEFFFFFC2F
     75 
     76 -- | Montgomery-form 'Wider' words, on the Montgomery domain defined by
     77 --   the secp256k1 field prime.
     78 --
     79 --   >>> let one = 1 :: Montgomery
     80 --   >>> one
     81 --   1
     82 --   >>> putStrLn (render one)
     83 --   (4294968273, 0, 0, 0)
     84 data Montgomery = Montgomery !Limb4
     85 
     86 -- | Render a 'Montgomery' value as a 'String', showing its individual
     87 --   'Limb's.
     88 --
     89 --   >>> putStrLn (render 1)
     90 --   (4294968273, 0, 0, 0)
     91 render :: Montgomery -> String
     92 render (Montgomery (L4 a b c d)) =
     93      "(" <> show (W# a) <> ", " <> show (W# b) <> ", "
     94   <> show (W# c) <> ", " <> show (W# d) <> ")"
     95 
     96 instance Show Montgomery where
     97   show = show . from
     98 
     99 -- | Note that 'fromInteger' necessarily runs in variable time due
    100 --   to conversion from the variable-size, potentially heap-allocated
    101 --   'Integer' type.
    102 instance Num Montgomery where
    103   a + b = add a b
    104   a - b = sub a b
    105   a * b = mul a b
    106   negate a = neg a
    107   abs = id
    108   fromInteger = to . WW.to_vartime
    109   signum (Montgomery (# l0, l1, l2, l3 #)) =
    110     let !(Limb l) = l0 `L.or#` l1 `L.or#` l2 `L.or#` l3
    111         !n = C.from_word_nonzero# l
    112         !(Montgomery z) = zero
    113         !(Montgomery o) = one
    114     in  Montgomery (select# z o n)
    115 
    116 instance NFData Montgomery where
    117   rnf (Montgomery a) = case a of (# _, _, _, _ #) -> ()
    118 
    119 -- utilities ------------------------------------------------------------------
    120 
    121 type Limb2 = (# Limb, Limb #)
    122 
    123 type Limb4 = (# Limb, Limb, Limb, Limb #)
    124 
    125 pattern L4 :: Word# -> Word# -> Word# -> Word# -> Limb4
    126 pattern L4 w0 w1 w2 w3 = (# Limb w0, Limb w1, Limb w2, Limb w3 #)
    127 {-# COMPLETE L4 #-}
    128 
    129 -- Wide wrapping addition, when addend is only a limb.
    130 wadd_w# :: Limb2 -> Limb -> Limb2
    131 wadd_w# (# x_lo, x_hi #) y_lo =
    132   let !(# s0, c0 #) = L.add_o# x_lo y_lo
    133       !(# s1, _ #) = L.add_o# x_hi c0
    134   in  (# s0, s1 #)
    135 {-# INLINE wadd_w# #-}
    136 
    137 -- Truncate a wide word to a 'Limb'.
    138 lo :: Limb2 -> Limb
    139 lo (# l, _ #) = l
    140 {-# INLINE lo #-}
    141 
    142 -- comparison -----------------------------------------------------------------
    143 
    144 -- | Constant-time equality comparison.
    145 eq :: Montgomery -> Montgomery -> C.Choice
    146 eq (Montgomery (L4 a0 a1 a2 a3)) (Montgomery (L4 b0 b1 b2 b3)) =
    147   C.eq_wider# (# a0, a1, a2, a3 #) (# b0, b1, b2, b3 #)
    148 {-# INLINE eq #-}
    149 
    150 -- | Variable-time equality comparison.
    151 eq_vartime :: Montgomery -> Montgomery -> Bool
    152 eq_vartime (Montgomery (Wider -> a)) (Montgomery (Wider -> b)) =
    153   WW.eq_vartime a b
    154 
    155 -- innards --------------------------------------------------------------------
    156 
    157 redc_inner#
    158   :: Limb4             -- ^ upper limbs
    159   -> Limb4             -- ^ lower limbs
    160   -> (# Limb4, Limb #) -- ^ upper limbs, meta-carry
    161 redc_inner# (# u0, u1, u2, u3 #) (# l0, l1, l2, l3 #) =
    162   let !(# m0, m1, m2, m3 #) =
    163         (# Limb 0xFFFFFFFEFFFFFC2F##, Limb 0xFFFFFFFFFFFFFFFF##
    164         ,  Limb 0xFFFFFFFFFFFFFFFF##, Limb 0xFFFFFFFFFFFFFFFF## #)
    165       !n                = Limb 0xD838091DD2253531##
    166       !w_0              = L.mul_w# l0 n
    167       !(# _, c_00 #)    = L.mac# w_0 m0 l0 (Limb 0##)
    168       !(# l0_1, c_01 #) = L.mac# w_0 m1 l1 c_00
    169       !(# l0_2, c_02 #) = L.mac# w_0 m2 l2 c_01
    170       !(# l0_3, c_03 #) = L.mac# w_0 m3 l3 c_02
    171       !(# u_0, mc_0 #)  = L.add_c# u0 c_03 (Limb 0##)
    172       !w_1              = L.mul_w# l0_1 n
    173       !(# _, c_10 #)    = L.mac# w_1 m0 l0_1 (Limb 0##)
    174       !(# l1_1, c_11 #) = L.mac# w_1 m1 l0_2 c_10
    175       !(# l1_2, c_12 #) = L.mac# w_1 m2 l0_3 c_11
    176       !(# u1_3, c_13 #) = L.mac# w_1 m3 u_0 c_12
    177       !(# u_1, mc_1 #)  = L.add_c# u1 c_13 mc_0
    178       !w_2              = L.mul_w# l1_1 n
    179       !(# _, c_20 #)    = L.mac# w_2 m0 l1_1 (Limb 0##)
    180       !(# l2_1, c_21 #) = L.mac# w_2 m1 l1_2 c_20
    181       !(# u2_2, c_22 #) = L.mac# w_2 m2 u1_3 c_21
    182       !(# u2_3, c_23 #) = L.mac# w_2 m3 u_1 c_22
    183       !(# u_2, mc_2 #)  = L.add_c# u2 c_23 mc_1
    184       !w_3              = L.mul_w# l2_1 n
    185       !(# _, c_30 #)    = L.mac# w_3 m0 l2_1 (Limb 0##)
    186       !(# u3_1, c_31 #) = L.mac# w_3 m1 u2_2 c_30
    187       !(# u3_2, c_32 #) = L.mac# w_3 m2 u2_3 c_31
    188       !(# u3_3, c_33 #) = L.mac# w_3 m3 u_2 c_32
    189       !(# u_3, mc_3 #)  = L.add_c# u3 c_33 mc_2
    190   in  (# (# u3_1, u3_2, u3_3, u_3 #), mc_3 #)
    191 {-# INLINE redc_inner# #-}
    192 
    193 -- | Montgomery reduction.
    194 redc#
    195   :: Limb4 -- ^ lower limbs
    196   -> Limb4 -- ^ upper limbs
    197   -> Limb4 -- ^ result
    198 redc# l u =
    199   let -- field prime
    200       !m = L4 0xFFFFFFFEFFFFFC2F## 0xFFFFFFFFFFFFFFFF##
    201               0xFFFFFFFFFFFFFFFF## 0xFFFFFFFFFFFFFFFF##
    202       !(# nu, mc #) = redc_inner# u l
    203   in  WW.sub_mod_c# nu mc m m
    204 {-# INLINE redc# #-}
    205 
    206 -- | Montgomery reduction.
    207 --
    208 --   The first argument represents the low words, and the second the
    209 --   high words, of an extra-large eight-limb word in Montgomery form.
    210 redc
    211   :: Montgomery -- ^ low wider-word, Montgomery form
    212   -> Montgomery -- ^ high wider-word, Montgomery form
    213   -> Montgomery -- ^ reduced value
    214 redc (Montgomery l) (Montgomery u) =
    215   let !res = redc# l u
    216   in  Montgomery res
    217 
    218 retr_inner#
    219   :: Limb4 -- ^ value in montgomery form
    220   -> Limb4 -- ^ retrieved value
    221 retr_inner# (# x0, x1, x2, x3 #) =
    222   let !(# m0, m1, m2, m3 #) =
    223         L4 0xFFFFFFFEFFFFFC2F## 0xFFFFFFFFFFFFFFFF##
    224            0xFFFFFFFFFFFFFFFF## 0xFFFFFFFFFFFFFFFF##
    225       !n                = Limb 0xD838091DD2253531##
    226       !u_0              = L.mul_w# x0 n
    227       !(# _, o0 #)      = L.mac# u_0 m0 x0 (Limb 0##)
    228       !(# o0_1, p0_1 #) = L.mac# u_0 m1 (Limb 0##) o0
    229       !(# p0_2, q0_2 #) = L.mac# u_0 m2 (Limb 0##) p0_1
    230       !(# q0_3, r0_3 #) = L.mac# u_0 m3 (Limb 0##) q0_2
    231       !u_1              = L.mul_w# (L.add_w# o0_1 x1) n
    232       !(# _, o1 #)      = L.mac# u_1 m0 x1 o0_1
    233       !(# o1_1, p1_1 #) = L.mac# u_1 m1 p0_2 o1
    234       !(# p1_2, q1_2 #) = L.mac# u_1 m2 q0_3 p1_1
    235       !(# q1_3, r1_3 #) = L.mac# u_1 m3 r0_3 q1_2
    236       !u_2              = L.mul_w# (L.add_w# o1_1 x2) n
    237       !(# _, o2 #)      = L.mac# u_2 m0 x2 o1_1
    238       !(# o2_1, p2_1 #) = L.mac# u_2 m1 p1_2 o2
    239       !(# p2_2, q2_2 #) = L.mac# u_2 m2 q1_3 p2_1
    240       !(# q2_3, r2_3 #) = L.mac# u_2 m3 r1_3 q2_2
    241       !u_3              = L.mul_w# (L.add_w# o2_1 x3) n
    242       !(# _, o3 #)      = L.mac# u_3 m0 x3 o2_1
    243       !(# o3_1, p3_1 #) = L.mac# u_3 m1 p2_2 o3
    244       !(# p3_2, q3_2 #) = L.mac# u_3 m2 q2_3 p3_1
    245       !(# q3_3, r3_3 #) = L.mac# u_3 m3 r2_3 q3_2
    246   in  (# o3_1, p3_2, q3_3, r3_3 #)
    247 {-# INLINE retr_inner# #-}
    248 
    249 retr#
    250   :: Limb4 -- montgomery form
    251   -> Limb4
    252 retr# f = retr_inner# f
    253 {-# INLINE retr# #-}
    254 
    255 -- | Retrieve a 'Montgomery' value from the Montgomery domain, producing
    256 --   a 'Wider' word.
    257 retr
    258   :: Montgomery -- ^ value in montgomery form
    259   -> Wider      -- ^ retrieved value
    260 retr (Montgomery f) =
    261   let !res = retr# f
    262   in  (Wider res)
    263 
    264 -- | Montgomery multiplication (FIOS), without conditional subtract.
    265 mul_inner#
    266   :: Limb4              -- ^ x
    267   -> Limb4              -- ^ y
    268   -> (# Limb4, Limb #)  -- ^ product, meta-carry
    269 mul_inner# (# x0, x1, x2, x3 #) (# y0, y1, y2, y3 #) =
    270   let !(# m0, m1, m2, m3 #) =
    271         L4 0xFFFFFFFEFFFFFC2F## 0xFFFFFFFFFFFFFFFF##
    272            0xFFFFFFFFFFFFFFFF## 0xFFFFFFFFFFFFFFFF##
    273       !n                           = Limb 0xD838091DD2253531##
    274       !axy0                        = L.mul_c# x0 y0
    275       !u0                          = L.mul_w# (lo axy0) n
    276       !(# (# _, a0 #), c0 #)       = W.add_o# (L.mul_c# u0 m0) axy0
    277       !carry0                      = (# a0, c0 #)
    278       !axy0_1                      = L.mul_c# x0 y1
    279       !umc0_1                      = W.add_w# (L.mul_c# u0 m1) carry0
    280       !(# (# o0, ab0_1 #), c0_1 #) = W.add_o# axy0_1 umc0_1
    281       !carry0_1                    = (# ab0_1, c0_1 #)
    282       !axy0_2                      = L.mul_c# x0 y2
    283       !umc0_2                      = W.add_w# (L.mul_c# u0 m2) carry0_1
    284       !(# (# p0, ab0_2 #), c0_2 #) = W.add_o# axy0_2 umc0_2
    285       !carry0_2                    = (# ab0_2, c0_2 #)
    286       !axy0_3                      = L.mul_c# x0 y3
    287       !umc0_3                      = W.add_w# (L.mul_c# u0 m3) carry0_2
    288       !(# (# q0, ab0_3 #), c0_3 #) = W.add_o# axy0_3 umc0_3
    289       !carry0_3                    = (# ab0_3, c0_3 #)
    290       !(# r0, mc0 #)               = carry0_3
    291       !axy1                        = wadd_w# (L.mul_c# x1 y0) o0
    292       !u1                          = L.mul_w# (lo axy1) n
    293       !(# (# _, a1 #), c1 #)       = W.add_o# (L.mul_c# u1 m0) axy1
    294       !carry1                      = (# a1, c1 #)
    295       !axy1_1                      = wadd_w# (L.mul_c# x1 y1) p0
    296       !umc1_1                      = W.add_w# (L.mul_c# u1 m1) carry1
    297       !(# (# o1, ab1_1 #), c1_1 #) = W.add_o# axy1_1 umc1_1
    298       !carry1_1                    = (# ab1_1, c1_1 #)
    299       !axy1_2                      = wadd_w# (L.mul_c# x1 y2) q0
    300       !umc1_2                      = W.add_w# (L.mul_c# u1 m2) carry1_1
    301       !(# (# p1, ab1_2 #), c1_2 #) = W.add_o# axy1_2 umc1_2
    302       !carry1_2                    = (# ab1_2, c1_2 #)
    303       !axy1_3                      = wadd_w# (L.mul_c# x1 y3) r0
    304       !umc1_3                      = W.add_w# (L.mul_c# u1 m3) carry1_2
    305       !(# (# q1, ab1_3 #), c1_3 #) = W.add_o# axy1_3 umc1_3
    306       !carry1_3                    = (# ab1_3, c1_3 #)
    307       !(# r1, mc1 #)               = wadd_w# carry1_3 mc0
    308       !axy2                        = wadd_w# (L.mul_c# x2 y0) o1
    309       !u2                          = L.mul_w# (lo axy2) n
    310       !(# (# _, a2 #), c2 #)       = W.add_o# (L.mul_c# u2 m0) axy2
    311       !carry2                      = (# a2, c2 #)
    312       !axy2_1                      = wadd_w# (L.mul_c# x2 y1) p1
    313       !umc2_1                      = W.add_w# (L.mul_c# u2 m1) carry2
    314       !(# (# o2, ab2_1 #), c2_1 #) = W.add_o# axy2_1 umc2_1
    315       !carry2_1                    = (# ab2_1, c2_1 #)
    316       !axy2_2                      = wadd_w# (L.mul_c# x2 y2) q1
    317       !umc2_2                      = W.add_w# (L.mul_c# u2 m2) carry2_1
    318       !(# (# p2, ab2_2 #), c2_2 #) = W.add_o# axy2_2 umc2_2
    319       !carry2_2                    = (# ab2_2, c2_2 #)
    320       !axy2_3                      = wadd_w# (L.mul_c# x2 y3) r1
    321       !umc2_3                      = W.add_w# (L.mul_c# u2 m3) carry2_2
    322       !(# (# q2, ab2_3 #), c2_3 #) = W.add_o# axy2_3 umc2_3
    323       !carry2_3                    = (# ab2_3, c2_3 #)
    324       !(# r2, mc2 #)               = wadd_w# carry2_3 mc1
    325       !axy3                        = wadd_w# (L.mul_c# x3 y0) o2
    326       !u3                          = L.mul_w# (lo axy3) n
    327       !(# (# _, a3 #), c3 #)       = W.add_o# (L.mul_c# u3 m0) axy3
    328       !carry3                      = (# a3, c3 #)
    329       !axy3_1                      = wadd_w# (L.mul_c# x3 y1) p2
    330       !umc3_1                      = W.add_w# (L.mul_c# u3 m1) carry3
    331       !(# (# o3, ab3_1 #), c3_1 #) = W.add_o# axy3_1 umc3_1
    332       !carry3_1                    = (# ab3_1, c3_1 #)
    333       !axy3_2                      = wadd_w# (L.mul_c# x3 y2) q2
    334       !umc3_2                      = W.add_w# (L.mul_c# u3 m2) carry3_1
    335       !(# (# p3, ab3_2 #), c3_2 #) = W.add_o# axy3_2 umc3_2
    336       !carry3_2                    = (# ab3_2, c3_2 #)
    337       !axy3_3                      = wadd_w# (L.mul_c# x3 y3) r2
    338       !umc3_3                      = W.add_w# (L.mul_c# u3 m3) carry3_2
    339       !(# (# q3, ab3_3 #), c3_3 #) = W.add_o# axy3_3 umc3_3
    340       !carry3_3                    = (# ab3_3, c3_3 #)
    341       !(# r3, mc3 #)               = wadd_w# carry3_3 mc2
    342   in  (# (# o3, p3, q3, r3 #), mc3 #)
    343 {-# INLINE mul_inner# #-}
    344 
    345 mul#
    346   :: Limb4
    347   -> Limb4
    348   -> Limb4
    349 mul# a b =
    350   let -- field prime
    351       !m = L4 0xFFFFFFFEFFFFFC2F## 0xFFFFFFFFFFFFFFFF##
    352               0xFFFFFFFFFFFFFFFF## 0xFFFFFFFFFFFFFFFF##
    353       !(# nu, mc #) = mul_inner# a b
    354   in  WW.sub_mod_c# nu mc m m
    355 {-# NOINLINE mul# #-} -- cannot be inlined without exploding comp time
    356 
    357 -- | Multiplication in the Montgomery domain.
    358 --
    359 --   Note that 'Montgomery' is an instance of 'Num', so you can use '*'
    360 --   to apply this function.
    361 --
    362 --   >>> 1 * 1 :: Montgomery
    363 --   1
    364 mul
    365   :: Montgomery -- ^ multiplicand in montgomery form
    366   -> Montgomery -- ^ multiplier in montgomery form
    367   -> Montgomery -- ^ montgomery product
    368 mul (Montgomery a) (Montgomery b) = Montgomery (mul# a b)
    369 
    370 to#
    371   :: Limb4 -- ^ integer
    372   -> Limb4
    373 to# x =
    374   let !r2 = L4 0x000007A2000E90A1## 0x1## 0## 0## -- r^2 mod m
    375   in  mul# x r2
    376 {-# INLINE to# #-}
    377 
    378 -- | Convert a 'Wider' word to the Montgomery domain.
    379 to :: Wider -> Montgomery
    380 to (Wider x) = Montgomery (to# x)
    381 
    382 -- | Retrieve a 'Montgomery' word from the Montgomery domain.
    383 --
    384 --   This function is a synonym for 'retr'.
    385 from :: Montgomery -> Wider
    386 from = retr
    387 
    388 add#
    389   :: Limb4 -- ^ augend
    390   -> Limb4 -- ^ addend
    391   -> Limb4 -- ^ sum
    392 add# a b =
    393   let -- field prime
    394       !m = L4 0xFFFFFFFEFFFFFC2F## 0xFFFFFFFFFFFFFFFF##
    395               0xFFFFFFFFFFFFFFFF## 0xFFFFFFFFFFFFFFFF##
    396   in  WW.add_mod# a b m
    397 {-# INLINE add# #-}
    398 
    399 -- | Addition in the Montgomery domain.
    400 --
    401 --   Note that 'Montgomery' is an instance of 'Num', so you can use '+'
    402 --   to apply this function.
    403 --
    404 --   >>> 1 + 1 :: Montgomery
    405 --   2
    406 add :: Montgomery -> Montgomery -> Montgomery
    407 add (Montgomery a) (Montgomery b) = Montgomery (add# a b)
    408 
    409 sub#
    410   :: Limb4 -- ^ minuend
    411   -> Limb4 -- ^ subtrahend
    412   -> Limb4 -- ^ difference
    413 sub# a b =
    414   let -- field prime
    415       !m = L4 0xFFFFFFFEFFFFFC2F## 0xFFFFFFFFFFFFFFFF##
    416               0xFFFFFFFFFFFFFFFF## 0xFFFFFFFFFFFFFFFF##
    417   in  WW.sub_mod# a b m
    418 {-# INLINE sub# #-}
    419 
    420 -- | Subtraction in the Montgomery domain.
    421 --
    422 --   Note that 'Montgomery' is an instance of 'Num', so you can use '-'
    423 --   to apply this function.
    424 --
    425 --   >>> 1 - 1 :: Montgomery
    426 --   0
    427 sub :: Montgomery -> Montgomery -> Montgomery
    428 sub (Montgomery a) (Montgomery b) = Montgomery (sub# a b)
    429 
    430 neg#
    431   :: Limb4 -- ^ argument
    432   -> Limb4 -- ^ modular negation
    433 neg# a = sub# (L4 0## 0## 0## 0##) a
    434 {-# INLINE neg# #-}
    435 
    436 -- | Additive inverse in the Montgomery domain.
    437 --
    438 --   Note that 'Montgomery' is an instance of 'Num', so you can use 'negate'
    439 --   to apply this function.
    440 --
    441 --   >>> negate 1 :: Montgomery
    442 --   115792089237316195423570985008687907853269984665640564039457584007908834671662
    443 --   >>> (negate 1 :: Montgomery) + 1
    444 --   0
    445 neg :: Montgomery -> Montgomery
    446 neg (Montgomery a) = Montgomery (neg# a)
    447 
    448 sqr# :: Limb4 -> Limb4
    449 sqr# a =
    450   let !(# l, h #) = WW.sqr# a
    451   in  redc# l h
    452 {-# NOINLINE sqr# #-} -- cannot be inlined without exploding comp time
    453 
    454 -- | Squaring in the Montgomery domain.
    455 --
    456 --   >>> sqr 1
    457 --   1
    458 --   >>> sqr 2
    459 --   4
    460 --   >>> sqr (negate 2)
    461 --   4
    462 sqr :: Montgomery -> Montgomery
    463 sqr (Montgomery a) = Montgomery (mul# a a)
    464 
    465 -- | Zero (the additive unit) in the Montgomery domain.
    466 zero :: Montgomery
    467 zero = Montgomery (L4 0## 0## 0## 0##)
    468 
    469 -- | One (the multiplicative unit) in the Montgomery domain.
    470 one :: Montgomery
    471 one = Montgomery (L4 0x1000003D1## 0## 0## 0##)
    472 
    473 -- generated by etc/generate_inv.sh
    474 inv#
    475   :: Limb4
    476   -> Limb4
    477 inv# a =
    478   let
    479       !t1 = sqr# a
    480       !t2 = mul# t1 a
    481       !t3 = sqr# t2
    482       !t4 = sqr# t3
    483       !t5 = mul# t4 t2
    484       !t6 = sqr# t5
    485       !t7 = sqr# t6
    486       !t8 = sqr# t7
    487       !t9 = sqr# t8
    488       !t10 = mul# t9 t5
    489       !t11 = sqr# t10
    490       !t12 = sqr# t11
    491       !t13 = sqr# t12
    492       !t14 = sqr# t13
    493       !t15 = sqr# t14
    494       !t16 = sqr# t15
    495       !t17 = sqr# t16
    496       !t18 = sqr# t17
    497       !t19 = mul# t18 t10
    498       !t20 = sqr# t19
    499       !t21 = sqr# t20
    500       !t22 = sqr# t21
    501       !t23 = sqr# t22
    502       !t24 = sqr# t23
    503       !t25 = sqr# t24
    504       !t26 = sqr# t25
    505       !t27 = sqr# t26
    506       !t28 = sqr# t27
    507       !t29 = sqr# t28
    508       !t30 = sqr# t29
    509       !t31 = sqr# t30
    510       !t32 = sqr# t31
    511       !t33 = sqr# t32
    512       !t34 = sqr# t33
    513       !t35 = sqr# t34
    514       !t36 = mul# t35 t19
    515       !t37 = sqr# t36
    516       !t38 = sqr# t37
    517       !t39 = sqr# t38
    518       !t40 = sqr# t39
    519       !t41 = sqr# t40
    520       !t42 = sqr# t41
    521       !t43 = sqr# t42
    522       !t44 = sqr# t43
    523       !t45 = sqr# t44
    524       !t46 = sqr# t45
    525       !t47 = sqr# t46
    526       !t48 = sqr# t47
    527       !t49 = sqr# t48
    528       !t50 = sqr# t49
    529       !t51 = sqr# t50
    530       !t52 = sqr# t51
    531       !t53 = sqr# t52
    532       !t54 = sqr# t53
    533       !t55 = sqr# t54
    534       !t56 = sqr# t55
    535       !t57 = sqr# t56
    536       !t58 = sqr# t57
    537       !t59 = sqr# t58
    538       !t60 = sqr# t59
    539       !t61 = sqr# t60
    540       !t62 = sqr# t61
    541       !t63 = sqr# t62
    542       !t64 = sqr# t63
    543       !t65 = sqr# t64
    544       !t66 = sqr# t65
    545       !t67 = sqr# t66
    546       !t68 = sqr# t67
    547       !t69 = mul# t68 t36
    548       !t70 = sqr# t69
    549       !t71 = sqr# t70
    550       !t72 = sqr# t71
    551       !t73 = sqr# t72
    552       !t74 = sqr# t73
    553       !t75 = sqr# t74
    554       !t76 = sqr# t75
    555       !t77 = sqr# t76
    556       !t78 = sqr# t77
    557       !t79 = sqr# t78
    558       !t80 = sqr# t79
    559       !t81 = sqr# t80
    560       !t82 = sqr# t81
    561       !t83 = sqr# t82
    562       !t84 = sqr# t83
    563       !t85 = sqr# t84
    564       !t86 = sqr# t85
    565       !t87 = sqr# t86
    566       !t88 = sqr# t87
    567       !t89 = sqr# t88
    568       !t90 = sqr# t89
    569       !t91 = sqr# t90
    570       !t92 = sqr# t91
    571       !t93 = sqr# t92
    572       !t94 = sqr# t93
    573       !t95 = sqr# t94
    574       !t96 = sqr# t95
    575       !t97 = sqr# t96
    576       !t98 = sqr# t97
    577       !t99 = sqr# t98
    578       !t100 = sqr# t99
    579       !t101 = sqr# t100
    580       !t102 = sqr# t101
    581       !t103 = sqr# t102
    582       !t104 = sqr# t103
    583       !t105 = sqr# t104
    584       !t106 = sqr# t105
    585       !t107 = sqr# t106
    586       !t108 = sqr# t107
    587       !t109 = sqr# t108
    588       !t110 = sqr# t109
    589       !t111 = sqr# t110
    590       !t112 = sqr# t111
    591       !t113 = sqr# t112
    592       !t114 = sqr# t113
    593       !t115 = sqr# t114
    594       !t116 = sqr# t115
    595       !t117 = sqr# t116
    596       !t118 = sqr# t117
    597       !t119 = sqr# t118
    598       !t120 = sqr# t119
    599       !t121 = sqr# t120
    600       !t122 = sqr# t121
    601       !t123 = sqr# t122
    602       !t124 = sqr# t123
    603       !t125 = sqr# t124
    604       !t126 = sqr# t125
    605       !t127 = sqr# t126
    606       !t128 = sqr# t127
    607       !t129 = sqr# t128
    608       !t130 = sqr# t129
    609       !t131 = sqr# t130
    610       !t132 = sqr# t131
    611       !t133 = sqr# t132
    612       !t134 = mul# t133 t69
    613       !t135 = sqr# t134
    614       !t136 = sqr# t135
    615       !t137 = sqr# t136
    616       !t138 = sqr# t137
    617       !t139 = sqr# t138
    618       !t140 = sqr# t139
    619       !t141 = sqr# t140
    620       !t142 = sqr# t141
    621       !t143 = sqr# t142
    622       !t144 = sqr# t143
    623       !t145 = sqr# t144
    624       !t146 = sqr# t145
    625       !t147 = sqr# t146
    626       !t148 = sqr# t147
    627       !t149 = sqr# t148
    628       !t150 = sqr# t149
    629       !t151 = sqr# t150
    630       !t152 = sqr# t151
    631       !t153 = sqr# t152
    632       !t154 = sqr# t153
    633       !t155 = sqr# t154
    634       !t156 = sqr# t155
    635       !t157 = sqr# t156
    636       !t158 = sqr# t157
    637       !t159 = sqr# t158
    638       !t160 = sqr# t159
    639       !t161 = sqr# t160
    640       !t162 = sqr# t161
    641       !t163 = sqr# t162
    642       !t164 = sqr# t163
    643       !t165 = sqr# t164
    644       !t166 = sqr# t165
    645       !t167 = sqr# t166
    646       !t168 = sqr# t167
    647       !t169 = sqr# t168
    648       !t170 = sqr# t169
    649       !t171 = sqr# t170
    650       !t172 = sqr# t171
    651       !t173 = sqr# t172
    652       !t174 = sqr# t173
    653       !t175 = sqr# t174
    654       !t176 = sqr# t175
    655       !t177 = sqr# t176
    656       !t178 = sqr# t177
    657       !t179 = sqr# t178
    658       !t180 = sqr# t179
    659       !t181 = sqr# t180
    660       !t182 = sqr# t181
    661       !t183 = sqr# t182
    662       !t184 = sqr# t183
    663       !t185 = sqr# t184
    664       !t186 = sqr# t185
    665       !t187 = sqr# t186
    666       !t188 = sqr# t187
    667       !t189 = sqr# t188
    668       !t190 = sqr# t189
    669       !t191 = sqr# t190
    670       !t192 = sqr# t191
    671       !t193 = sqr# t192
    672       !t194 = sqr# t193
    673       !t195 = sqr# t194
    674       !t196 = sqr# t195
    675       !t197 = sqr# t196
    676       !t198 = sqr# t197
    677       !t199 = mul# t198 t69
    678       !t200 = sqr# t199
    679       !t201 = sqr# t200
    680       !t202 = sqr# t201
    681       !t203 = sqr# t202
    682       !t204 = sqr# t203
    683       !t205 = sqr# t204
    684       !t206 = sqr# t205
    685       !t207 = sqr# t206
    686       !t208 = sqr# t207
    687       !t209 = sqr# t208
    688       !t210 = sqr# t209
    689       !t211 = sqr# t210
    690       !t212 = sqr# t211
    691       !t213 = sqr# t212
    692       !t214 = sqr# t213
    693       !t215 = sqr# t214
    694       !t216 = mul# t215 t19
    695       !t217 = sqr# t216
    696       !t218 = sqr# t217
    697       !t219 = sqr# t218
    698       !t220 = sqr# t219
    699       !t221 = sqr# t220
    700       !t222 = sqr# t221
    701       !t223 = sqr# t222
    702       !t224 = sqr# t223
    703       !t225 = mul# t224 t10
    704       !t226 = sqr# t225
    705       !t227 = sqr# t226
    706       !t228 = sqr# t227
    707       !t229 = sqr# t228
    708       !t230 = mul# t229 t5
    709       !t231 = sqr# t230
    710       !t232 = sqr# t231
    711       !t233 = mul# t232 t2
    712       !t234 = sqr# t233
    713       !t235 = mul# t234 a
    714       !t236 = sqr# t19
    715       !t237 = sqr# t236
    716       !t238 = sqr# t237
    717       !t239 = sqr# t238
    718       !t240 = mul# t239 t5
    719       !t241 = sqr# t240
    720       !t242 = sqr# t241
    721       !t243 = mul# t242 t2
    722       !t244 = sqr# t235
    723       !t245 = sqr# t244
    724       !t246 = sqr# t245
    725       !t247 = sqr# t246
    726       !t248 = sqr# t247
    727       !t249 = sqr# t248
    728       !t250 = sqr# t249
    729       !t251 = sqr# t250
    730       !t252 = sqr# t251
    731       !t253 = sqr# t252
    732       !t254 = sqr# t253
    733       !t255 = sqr# t254
    734       !t256 = sqr# t255
    735       !t257 = sqr# t256
    736       !t258 = sqr# t257
    737       !t259 = sqr# t258
    738       !t260 = sqr# t259
    739       !t261 = sqr# t260
    740       !t262 = sqr# t261
    741       !t263 = sqr# t262
    742       !t264 = sqr# t263
    743       !t265 = sqr# t264
    744       !t266 = sqr# t265
    745       !t267 = mul# t266 t243
    746       !t268 = sqr# t267
    747       !t269 = sqr# t268
    748       !t270 = sqr# t269
    749       !t271 = sqr# t270
    750       !t272 = sqr# t271
    751       !t273 = mul# t272 a
    752       !t274 = sqr# t273
    753       !t275 = sqr# t274
    754       !t276 = sqr# t275
    755       !t277 = mul# t276 t2
    756       !t278 = sqr# t277
    757       !t279 = sqr# t278
    758       !t280 = mul# t279 a
    759       !r = t280
    760   in  r
    761 {-# INLINE inv# #-}
    762 
    763 -- | Multiplicative inverse in the Montgomery domain.
    764 --
    765 --   Note that 'zero' has no multiplicative inverse; 'inv' returns
    766 --   'zero' when applied to it.
    767 --
    768 --   >> inv 2
    769 --   57896044618658097711785492504343953926634992332820282019728792003954417335832
    770 --   >> inv 2 * 2
    771 --   1
    772 inv :: Montgomery -> Montgomery
    773 inv (Montgomery w) = Montgomery (inv# w)
    774 
    775 -- | Square root (Tonelli-Shanks) in the Montgomery domain.
    776 --
    777 --   Returns 'Nothing' if the square root doesn't exist.
    778 --
    779 --   Note that the square root calculation itself is performed in
    780 --   constant time; we branch only when casting to 'Maybe' at the end.
    781 --
    782 --   >>> sqrt_vartime 4
    783 --   Just 2
    784 --   >>> sqrt_vartime 15
    785 --   Just 69211104694897500952317515077652022726490027694212560352756646854116994689233
    786 --   >>> (*) <$> sqrt_vartime 15 <*> sqrt_vartime 15
    787 --   Just 15
    788 sqrt_vartime :: Montgomery -> Maybe Montgomery
    789 sqrt_vartime (Montgomery n) = case sqrt# n of
    790   (# a, c #)
    791     | C.decide c -> Just $! Montgomery a
    792     | otherwise  -> Nothing
    793 
    794 -- generated by etc/generate_sqrt.sh
    795 sqrt#
    796   :: Limb4
    797   -> (# Limb4, C.Choice #)
    798 sqrt# a =
    799   let !t0 = L4 0x1000003D1## 0## 0## 0##
    800       !t1 = sqr# t0
    801       !t2 = sqr# t1
    802       !t3 = sqr# t2
    803       !t4 = mul# a t3
    804       !t5 = sqr# t4
    805       !t6 = mul# a t5
    806       !t7 = sqr# t6
    807       !t8 = mul# a t7
    808       !t9 = sqr# t8
    809       !t10 = mul# a t9
    810       !t11 = sqr# t10
    811       !t12 = mul# a t11
    812       !t13 = sqr# t12
    813       !t14 = mul# a t13
    814       !t15 = sqr# t14
    815       !t16 = mul# a t15
    816       !t17 = sqr# t16
    817       !t18 = mul# a t17
    818       !t19 = sqr# t18
    819       !t20 = mul# a t19
    820       !t21 = sqr# t20
    821       !t22 = mul# a t21
    822       !t23 = sqr# t22
    823       !t24 = mul# a t23
    824       !t25 = sqr# t24
    825       !t26 = mul# a t25
    826       !t27 = sqr# t26
    827       !t28 = mul# a t27
    828       !t29 = sqr# t28
    829       !t30 = mul# a t29
    830       !t31 = sqr# t30
    831       !t32 = mul# a t31
    832       !t33 = sqr# t32
    833       !t34 = mul# a t33
    834       !t35 = sqr# t34
    835       !t36 = mul# a t35
    836       !t37 = sqr# t36
    837       !t38 = mul# a t37
    838       !t39 = sqr# t38
    839       !t40 = mul# a t39
    840       !t41 = sqr# t40
    841       !t42 = mul# a t41
    842       !t43 = sqr# t42
    843       !t44 = mul# a t43
    844       !t45 = sqr# t44
    845       !t46 = mul# a t45
    846       !t47 = sqr# t46
    847       !t48 = mul# a t47
    848       !t49 = sqr# t48
    849       !t50 = mul# a t49
    850       !t51 = sqr# t50
    851       !t52 = mul# a t51
    852       !t53 = sqr# t52
    853       !t54 = mul# a t53
    854       !t55 = sqr# t54
    855       !t56 = mul# a t55
    856       !t57 = sqr# t56
    857       !t58 = mul# a t57
    858       !t59 = sqr# t58
    859       !t60 = mul# a t59
    860       !t61 = sqr# t60
    861       !t62 = mul# a t61
    862       !t63 = sqr# t62
    863       !t64 = mul# a t63
    864       !t65 = sqr# t64
    865       !t66 = mul# a t65
    866       !t67 = sqr# t66
    867       !t68 = mul# a t67
    868       !t69 = sqr# t68
    869       !t70 = mul# a t69
    870       !t71 = sqr# t70
    871       !t72 = mul# a t71
    872       !t73 = sqr# t72
    873       !t74 = mul# a t73
    874       !t75 = sqr# t74
    875       !t76 = mul# a t75
    876       !t77 = sqr# t76
    877       !t78 = mul# a t77
    878       !t79 = sqr# t78
    879       !t80 = mul# a t79
    880       !t81 = sqr# t80
    881       !t82 = mul# a t81
    882       !t83 = sqr# t82
    883       !t84 = mul# a t83
    884       !t85 = sqr# t84
    885       !t86 = mul# a t85
    886       !t87 = sqr# t86
    887       !t88 = mul# a t87
    888       !t89 = sqr# t88
    889       !t90 = mul# a t89
    890       !t91 = sqr# t90
    891       !t92 = mul# a t91
    892       !t93 = sqr# t92
    893       !t94 = mul# a t93
    894       !t95 = sqr# t94
    895       !t96 = mul# a t95
    896       !t97 = sqr# t96
    897       !t98 = mul# a t97
    898       !t99 = sqr# t98
    899       !t100 = mul# a t99
    900       !t101 = sqr# t100
    901       !t102 = mul# a t101
    902       !t103 = sqr# t102
    903       !t104 = mul# a t103
    904       !t105 = sqr# t104
    905       !t106 = mul# a t105
    906       !t107 = sqr# t106
    907       !t108 = mul# a t107
    908       !t109 = sqr# t108
    909       !t110 = mul# a t109
    910       !t111 = sqr# t110
    911       !t112 = mul# a t111
    912       !t113 = sqr# t112
    913       !t114 = mul# a t113
    914       !t115 = sqr# t114
    915       !t116 = mul# a t115
    916       !t117 = sqr# t116
    917       !t118 = mul# a t117
    918       !t119 = sqr# t118
    919       !t120 = mul# a t119
    920       !t121 = sqr# t120
    921       !t122 = mul# a t121
    922       !t123 = sqr# t122
    923       !t124 = mul# a t123
    924       !t125 = sqr# t124
    925       !t126 = mul# a t125
    926       !t127 = sqr# t126
    927       !t128 = mul# a t127
    928       !t129 = sqr# t128
    929       !t130 = mul# a t129
    930       !t131 = sqr# t130
    931       !t132 = mul# a t131
    932       !t133 = sqr# t132
    933       !t134 = mul# a t133
    934       !t135 = sqr# t134
    935       !t136 = mul# a t135
    936       !t137 = sqr# t136
    937       !t138 = mul# a t137
    938       !t139 = sqr# t138
    939       !t140 = mul# a t139
    940       !t141 = sqr# t140
    941       !t142 = mul# a t141
    942       !t143 = sqr# t142
    943       !t144 = mul# a t143
    944       !t145 = sqr# t144
    945       !t146 = mul# a t145
    946       !t147 = sqr# t146
    947       !t148 = mul# a t147
    948       !t149 = sqr# t148
    949       !t150 = mul# a t149
    950       !t151 = sqr# t150
    951       !t152 = mul# a t151
    952       !t153 = sqr# t152
    953       !t154 = mul# a t153
    954       !t155 = sqr# t154
    955       !t156 = mul# a t155
    956       !t157 = sqr# t156
    957       !t158 = mul# a t157
    958       !t159 = sqr# t158
    959       !t160 = mul# a t159
    960       !t161 = sqr# t160
    961       !t162 = mul# a t161
    962       !t163 = sqr# t162
    963       !t164 = mul# a t163
    964       !t165 = sqr# t164
    965       !t166 = mul# a t165
    966       !t167 = sqr# t166
    967       !t168 = mul# a t167
    968       !t169 = sqr# t168
    969       !t170 = mul# a t169
    970       !t171 = sqr# t170
    971       !t172 = mul# a t171
    972       !t173 = sqr# t172
    973       !t174 = mul# a t173
    974       !t175 = sqr# t174
    975       !t176 = mul# a t175
    976       !t177 = sqr# t176
    977       !t178 = mul# a t177
    978       !t179 = sqr# t178
    979       !t180 = mul# a t179
    980       !t181 = sqr# t180
    981       !t182 = mul# a t181
    982       !t183 = sqr# t182
    983       !t184 = mul# a t183
    984       !t185 = sqr# t184
    985       !t186 = mul# a t185
    986       !t187 = sqr# t186
    987       !t188 = mul# a t187
    988       !t189 = sqr# t188
    989       !t190 = mul# a t189
    990       !t191 = sqr# t190
    991       !t192 = mul# a t191
    992       !t193 = sqr# t192
    993       !t194 = mul# a t193
    994       !t195 = sqr# t194
    995       !t196 = mul# a t195
    996       !t197 = sqr# t196
    997       !t198 = mul# a t197
    998       !t199 = sqr# t198
    999       !t200 = mul# a t199
   1000       !t201 = sqr# t200
   1001       !t202 = mul# a t201
   1002       !t203 = sqr# t202
   1003       !t204 = mul# a t203
   1004       !t205 = sqr# t204
   1005       !t206 = mul# a t205
   1006       !t207 = sqr# t206
   1007       !t208 = mul# a t207
   1008       !t209 = sqr# t208
   1009       !t210 = mul# a t209
   1010       !t211 = sqr# t210
   1011       !t212 = mul# a t211
   1012       !t213 = sqr# t212
   1013       !t214 = mul# a t213
   1014       !t215 = sqr# t214
   1015       !t216 = mul# a t215
   1016       !t217 = sqr# t216
   1017       !t218 = mul# a t217
   1018       !t219 = sqr# t218
   1019       !t220 = mul# a t219
   1020       !t221 = sqr# t220
   1021       !t222 = mul# a t221
   1022       !t223 = sqr# t222
   1023       !t224 = mul# a t223
   1024       !t225 = sqr# t224
   1025       !t226 = mul# a t225
   1026       !t227 = sqr# t226
   1027       !t228 = mul# a t227
   1028       !t229 = sqr# t228
   1029       !t230 = mul# a t229
   1030       !t231 = sqr# t230
   1031       !t232 = mul# a t231
   1032       !t233 = sqr# t232
   1033       !t234 = mul# a t233
   1034       !t235 = sqr# t234
   1035       !t236 = mul# a t235
   1036       !t237 = sqr# t236
   1037       !t238 = mul# a t237
   1038       !t239 = sqr# t238
   1039       !t240 = mul# a t239
   1040       !t241 = sqr# t240
   1041       !t242 = mul# a t241
   1042       !t243 = sqr# t242
   1043       !t244 = mul# a t243
   1044       !t245 = sqr# t244
   1045       !t246 = mul# a t245
   1046       !t247 = sqr# t246
   1047       !t248 = mul# a t247
   1048       !t249 = sqr# t248
   1049       !t250 = mul# a t249
   1050       !t251 = sqr# t250
   1051       !t252 = mul# a t251
   1052       !t253 = sqr# t252
   1053       !t254 = mul# a t253
   1054       !t255 = sqr# t254
   1055       !t256 = mul# a t255
   1056       !t257 = sqr# t256
   1057       !t258 = mul# a t257
   1058       !t259 = sqr# t258
   1059       !t260 = mul# a t259
   1060       !t261 = sqr# t260
   1061       !t262 = mul# a t261
   1062       !t263 = sqr# t262
   1063       !t264 = mul# a t263
   1064       !t265 = sqr# t264
   1065       !t266 = mul# a t265
   1066       !t267 = sqr# t266
   1067       !t268 = mul# a t267
   1068       !t269 = sqr# t268
   1069       !t270 = mul# a t269
   1070       !t271 = sqr# t270
   1071       !t272 = mul# a t271
   1072       !t273 = sqr# t272
   1073       !t274 = mul# a t273
   1074       !t275 = sqr# t274
   1075       !t276 = mul# a t275
   1076       !t277 = sqr# t276
   1077       !t278 = mul# a t277
   1078       !t279 = sqr# t278
   1079       !t280 = mul# a t279
   1080       !t281 = sqr# t280
   1081       !t282 = mul# a t281
   1082       !t283 = sqr# t282
   1083       !t284 = mul# a t283
   1084       !t285 = sqr# t284
   1085       !t286 = mul# a t285
   1086       !t287 = sqr# t286
   1087       !t288 = mul# a t287
   1088       !t289 = sqr# t288
   1089       !t290 = mul# a t289
   1090       !t291 = sqr# t290
   1091       !t292 = mul# a t291
   1092       !t293 = sqr# t292
   1093       !t294 = mul# a t293
   1094       !t295 = sqr# t294
   1095       !t296 = mul# a t295
   1096       !t297 = sqr# t296
   1097       !t298 = mul# a t297
   1098       !t299 = sqr# t298
   1099       !t300 = mul# a t299
   1100       !t301 = sqr# t300
   1101       !t302 = mul# a t301
   1102       !t303 = sqr# t302
   1103       !t304 = mul# a t303
   1104       !t305 = sqr# t304
   1105       !t306 = mul# a t305
   1106       !t307 = sqr# t306
   1107       !t308 = mul# a t307
   1108       !t309 = sqr# t308
   1109       !t310 = mul# a t309
   1110       !t311 = sqr# t310
   1111       !t312 = mul# a t311
   1112       !t313 = sqr# t312
   1113       !t314 = mul# a t313
   1114       !t315 = sqr# t314
   1115       !t316 = mul# a t315
   1116       !t317 = sqr# t316
   1117       !t318 = mul# a t317
   1118       !t319 = sqr# t318
   1119       !t320 = mul# a t319
   1120       !t321 = sqr# t320
   1121       !t322 = mul# a t321
   1122       !t323 = sqr# t322
   1123       !t324 = mul# a t323
   1124       !t325 = sqr# t324
   1125       !t326 = mul# a t325
   1126       !t327 = sqr# t326
   1127       !t328 = mul# a t327
   1128       !t329 = sqr# t328
   1129       !t330 = mul# a t329
   1130       !t331 = sqr# t330
   1131       !t332 = mul# a t331
   1132       !t333 = sqr# t332
   1133       !t334 = mul# a t333
   1134       !t335 = sqr# t334
   1135       !t336 = mul# a t335
   1136       !t337 = sqr# t336
   1137       !t338 = mul# a t337
   1138       !t339 = sqr# t338
   1139       !t340 = mul# a t339
   1140       !t341 = sqr# t340
   1141       !t342 = mul# a t341
   1142       !t343 = sqr# t342
   1143       !t344 = mul# a t343
   1144       !t345 = sqr# t344
   1145       !t346 = mul# a t345
   1146       !t347 = sqr# t346
   1147       !t348 = mul# a t347
   1148       !t349 = sqr# t348
   1149       !t350 = mul# a t349
   1150       !t351 = sqr# t350
   1151       !t352 = mul# a t351
   1152       !t353 = sqr# t352
   1153       !t354 = mul# a t353
   1154       !t355 = sqr# t354
   1155       !t356 = mul# a t355
   1156       !t357 = sqr# t356
   1157       !t358 = mul# a t357
   1158       !t359 = sqr# t358
   1159       !t360 = mul# a t359
   1160       !t361 = sqr# t360
   1161       !t362 = mul# a t361
   1162       !t363 = sqr# t362
   1163       !t364 = mul# a t363
   1164       !t365 = sqr# t364
   1165       !t366 = mul# a t365
   1166       !t367 = sqr# t366
   1167       !t368 = mul# a t367
   1168       !t369 = sqr# t368
   1169       !t370 = mul# a t369
   1170       !t371 = sqr# t370
   1171       !t372 = mul# a t371
   1172       !t373 = sqr# t372
   1173       !t374 = mul# a t373
   1174       !t375 = sqr# t374
   1175       !t376 = mul# a t375
   1176       !t377 = sqr# t376
   1177       !t378 = mul# a t377
   1178       !t379 = sqr# t378
   1179       !t380 = mul# a t379
   1180       !t381 = sqr# t380
   1181       !t382 = mul# a t381
   1182       !t383 = sqr# t382
   1183       !t384 = mul# a t383
   1184       !t385 = sqr# t384
   1185       !t386 = mul# a t385
   1186       !t387 = sqr# t386
   1187       !t388 = mul# a t387
   1188       !t389 = sqr# t388
   1189       !t390 = mul# a t389
   1190       !t391 = sqr# t390
   1191       !t392 = mul# a t391
   1192       !t393 = sqr# t392
   1193       !t394 = mul# a t393
   1194       !t395 = sqr# t394
   1195       !t396 = mul# a t395
   1196       !t397 = sqr# t396
   1197       !t398 = mul# a t397
   1198       !t399 = sqr# t398
   1199       !t400 = mul# a t399
   1200       !t401 = sqr# t400
   1201       !t402 = mul# a t401
   1202       !t403 = sqr# t402
   1203       !t404 = mul# a t403
   1204       !t405 = sqr# t404
   1205       !t406 = mul# a t405
   1206       !t407 = sqr# t406
   1207       !t408 = mul# a t407
   1208       !t409 = sqr# t408
   1209       !t410 = mul# a t409
   1210       !t411 = sqr# t410
   1211       !t412 = mul# a t411
   1212       !t413 = sqr# t412
   1213       !t414 = mul# a t413
   1214       !t415 = sqr# t414
   1215       !t416 = mul# a t415
   1216       !t417 = sqr# t416
   1217       !t418 = mul# a t417
   1218       !t419 = sqr# t418
   1219       !t420 = mul# a t419
   1220       !t421 = sqr# t420
   1221       !t422 = mul# a t421
   1222       !t423 = sqr# t422
   1223       !t424 = mul# a t423
   1224       !t425 = sqr# t424
   1225       !t426 = mul# a t425
   1226       !t427 = sqr# t426
   1227       !t428 = mul# a t427
   1228       !t429 = sqr# t428
   1229       !t430 = mul# a t429
   1230       !t431 = sqr# t430
   1231       !t432 = mul# a t431
   1232       !t433 = sqr# t432
   1233       !t434 = mul# a t433
   1234       !t435 = sqr# t434
   1235       !t436 = mul# a t435
   1236       !t437 = sqr# t436
   1237       !t438 = mul# a t437
   1238       !t439 = sqr# t438
   1239       !t440 = mul# a t439
   1240       !t441 = sqr# t440
   1241       !t442 = mul# a t441
   1242       !t443 = sqr# t442
   1243       !t444 = mul# a t443
   1244       !t445 = sqr# t444
   1245       !t446 = mul# a t445
   1246       !t447 = sqr# t446
   1247       !t448 = mul# a t447
   1248       !t449 = sqr# t448
   1249       !t450 = sqr# t449
   1250       !t451 = mul# a t450
   1251       !t452 = sqr# t451
   1252       !t453 = mul# a t452
   1253       !t454 = sqr# t453
   1254       !t455 = mul# a t454
   1255       !t456 = sqr# t455
   1256       !t457 = mul# a t456
   1257       !t458 = sqr# t457
   1258       !t459 = mul# a t458
   1259       !t460 = sqr# t459
   1260       !t461 = mul# a t460
   1261       !t462 = sqr# t461
   1262       !t463 = mul# a t462
   1263       !t464 = sqr# t463
   1264       !t465 = mul# a t464
   1265       !t466 = sqr# t465
   1266       !t467 = mul# a t466
   1267       !t468 = sqr# t467
   1268       !t469 = mul# a t468
   1269       !t470 = sqr# t469
   1270       !t471 = mul# a t470
   1271       !t472 = sqr# t471
   1272       !t473 = mul# a t472
   1273       !t474 = sqr# t473
   1274       !t475 = mul# a t474
   1275       !t476 = sqr# t475
   1276       !t477 = mul# a t476
   1277       !t478 = sqr# t477
   1278       !t479 = mul# a t478
   1279       !t480 = sqr# t479
   1280       !t481 = mul# a t480
   1281       !t482 = sqr# t481
   1282       !t483 = mul# a t482
   1283       !t484 = sqr# t483
   1284       !t485 = mul# a t484
   1285       !t486 = sqr# t485
   1286       !t487 = mul# a t486
   1287       !t488 = sqr# t487
   1288       !t489 = mul# a t488
   1289       !t490 = sqr# t489
   1290       !t491 = mul# a t490
   1291       !t492 = sqr# t491
   1292       !t493 = mul# a t492
   1293       !t494 = sqr# t493
   1294       !t495 = sqr# t494
   1295       !t496 = sqr# t495
   1296       !t497 = sqr# t496
   1297       !t498 = sqr# t497
   1298       !t499 = mul# a t498
   1299       !t500 = sqr# t499
   1300       !t501 = mul# a t500
   1301       !t502 = sqr# t501
   1302       !t503 = sqr# t502
   1303       !r = t503
   1304   in  (# r, WW.eq# (sqr# r) a #)
   1305 {-# INLINE sqrt# #-}
   1306 
   1307 -- | Exponentiation in the Montgomery domain.
   1308 --
   1309 --   >>> exp 2 3
   1310 --   8
   1311 --   >>> exp 2 10
   1312 --   1024
   1313 exp :: Montgomery -> Wider -> Montgomery
   1314 exp (Montgomery b) (Wider e) = Montgomery (exp# b e)
   1315 
   1316 exp#
   1317   :: Limb4
   1318   -> Limb4
   1319   -> Limb4
   1320 exp# b e =
   1321   let !o = L4 0x1000003D1## 0## 0## 0##
   1322       loop !r !m !ex n = case n of
   1323         0 -> r
   1324         _ ->
   1325           let !(# ne, bit #) = WW.shr1_c# ex
   1326               !candidate = mul# r m
   1327               !nr = select# r candidate bit
   1328               !nm = sqr# m
   1329           in  loop nr nm ne (n - 1)
   1330   in  loop o b e (256 :: Word)
   1331 {-# INLINE exp# #-}
   1332 
   1333 odd# :: Limb4 -> C.Choice
   1334 odd# = WW.odd#
   1335 {-# INLINE odd# #-}
   1336 
   1337 -- | Check if a 'Montgomery' value is odd.
   1338 --
   1339 --   Note that the comparison is performed in constant time, but we
   1340 --   branch when converting to 'Bool'.
   1341 --
   1342 --   >>> odd 1
   1343 --   True
   1344 --   >>> odd 2
   1345 --   False
   1346 --   >>> Data.Word.Wider.odd (retr 3) -- parity is preserved
   1347 --   True
   1348 odd_vartime :: Montgomery -> Bool
   1349 odd_vartime (Montgomery m) = C.decide (odd# m)
   1350 
   1351 -- constant-time selection ----------------------------------------------------
   1352 
   1353 select#
   1354   :: Limb4    -- ^ a
   1355   -> Limb4    -- ^ b
   1356   -> C.Choice -- ^ c
   1357   -> Limb4    -- ^ result
   1358 select# = WW.select#
   1359 {-# INLINE select# #-}
   1360 
   1361 -- | Return b if c is truthy, otherwise return a.
   1362 --
   1363 --   >>> import qualified Data.Choice as C
   1364 --   >>> select 0 1 (C.true# ())
   1365 --   1
   1366 select
   1367   :: Montgomery    -- ^ a
   1368   -> Montgomery    -- ^ b
   1369   -> C.Choice      -- ^ c
   1370   -> Montgomery    -- ^ result
   1371 select (Montgomery a) (Montgomery b) c = Montgomery (select# a b c)
   1372