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

use awint_internals::*;
use const_fn::const_fn;

use crate::Bits;

// TODO optimize for high leading zero and trailing zero cases

/// # Multiplication
impl Bits {
    /// Assigns `cin + (self * rhs)` to `self` and returns the overflow
    #[const_fn(cfg(feature = "const_support"))]
    pub const fn digit_cin_mul_(&mut self, cin: Digit, rhs: Digit) -> Digit {
        let mut carry = cin;
        unsafe_for_each_mut!(
            self,
            x,
            {
                let tmp = widen_mul_add(*x, rhs, carry);
                *x = tmp.0;
                carry = tmp.1;
            },
            false
        );
        let oflow = if self.extra() == 0 {
            carry
        } else {
            (self.last() >> self.extra()) | (carry << (BITS - self.extra()))
        };
        self.clear_unused_bits();
        oflow
    }

    /// Add-assigns `lhs * rhs` to `self` and returns if overflow happened
    #[const_fn(cfg(feature = "const_support"))]
    #[must_use]
    pub const fn digit_mul_add_(&mut self, lhs: &Self, rhs: Digit) -> Option<bool> {
        let mut mul_carry = 0;
        let mut add_carry = 0;
        unsafe_binop_for_each_mut!(
            self,
            lhs,
            self_x,
            lhs_x,
            {
                let tmp0 = widen_mul_add(lhs_x, rhs, mul_carry);
                mul_carry = tmp0.1;
                let tmp1 = widen_add(*self_x, tmp0.0, add_carry);
                add_carry = tmp1.1;
                *self_x = tmp1.0;
            },
            true
        );
        Some((mul_carry != 0) || (add_carry != 0))
    }

    /// Multiplies `lhs` by `rhs` and add-assigns the product to `self`. Three
    /// operands eliminates the need for an allocating temporary.
    #[const_fn(cfg(feature = "const_support"))]
    #[must_use]
    pub const fn mul_add_(&mut self, lhs: &Self, rhs: &Self) -> Option<()> {
        if self.bw() != lhs.bw() || self.bw() != rhs.bw() {
            return None
        }
        unsafe {
            const_for!(lhs_i in {0..self.total_digits()} {
                // carry from the short multiplication
                let mut carry0 = 0;
                let mut carry1 = 0;
                const_for!(rhs_i in {0..(self.total_digits() - lhs_i)} {
                    let tmp0 =
                        widen_mul_add(lhs.get_unchecked(lhs_i), rhs.get_unchecked(rhs_i), carry0);
                    carry0 = tmp0.1;
                    let tmp1 = widen_add(self.get_unchecked(lhs_i + rhs_i), tmp0.0, carry1);
                    carry1 = tmp1.1;
                    *self.get_unchecked_mut(lhs_i + rhs_i) = tmp1.0;
                });
            });
        }
        self.clear_unused_bits();
        Some(())
    }

    /// Multiply-assigns `self` by `rhs`. `pad` is a scratchpad that will be
    /// mutated arbitrarily.
    #[const_fn(cfg(feature = "const_support"))]
    #[must_use]
    pub const fn mul_(&mut self, rhs: &Self, pad: &mut Self) -> Option<()> {
        if self.bw() != rhs.bw() || self.bw() != pad.bw() {
            return None
        }
        pad.zero_();
        unsafe {
            const_for!(self_i in {0..self.total_digits()} {
                // carry from the short multiplication
                let mut carry0 = 0;
                let mut carry1 = 0;
                const_for!(rhs_i in {0..(self.total_digits() - self_i)} {
                    let tmp0 =
                        widen_mul_add(self.get_unchecked(self_i), rhs.get_unchecked(rhs_i), carry0);
                    carry0 = tmp0.1;
                    let tmp1 = widen_add(pad.get_unchecked(self_i + rhs_i), tmp0.0, carry1);
                    carry1 = tmp1.1;
                    *pad.get_unchecked_mut(self_i + rhs_i) = tmp1.0;
                });
            });
        }
        pad.clear_unused_bits();
        self.copy_(pad).unwrap();
        Some(())
    }

    /// Arbitrarily-unsigned-multiplies `lhs` by `rhs` and add-assigns the
    /// product to `self`. This function is equivalent to:
    /// ```
    /// use awint::awi::*;
    ///
    /// fn arb_umul_(add: &mut Bits, lhs: &Bits, rhs: &Bits) {
    ///     let mut resized_lhs = Awi::zero(add.nzbw());
    ///     // Note that this function is specified as unsigned,
    ///     // because we use `zero_resize_`
    ///     resized_lhs.zero_resize_(lhs);
    ///     let mut resized_rhs = Awi::zero(add.nzbw());
    ///     resized_rhs.zero_resize_(rhs);
    ///     add.mul_add_(&resized_lhs, &resized_rhs).unwrap();
    /// }
    /// ```
    /// except that it avoids allocation and is more efficient overall
    #[const_fn(cfg(feature = "const_support"))]
    pub const fn arb_umul_add_(&mut self, lhs: &Self, rhs: &Self) {
        // first, we swap references so that `x0.bw() <= x1.bw()`
        let (x0, x1) = if lhs.bw() <= rhs.bw() {
            (lhs, rhs)
        } else {
            (rhs, lhs)
        };
        let x0_upper_bound = if self.total_digits() < x0.total_digits() {
            self.total_digits()
        } else {
            x0.total_digits()
        };
        // Safety: all the `get_unchecked_` are in bounds, since `x0_i < x0_upper_bound
        // < x0.total_digits()` and there are independent checks every loop for the
        // `x1_i` and `self_i` cases.
        unsafe {
            const_for!(x0_i in {0..x0_upper_bound} {
                // carry from the short multiplication
                let mut carry0 = 0;
                let mut carry1 = 0;
                let mut x1_i = 0;
                let mut self_i = x0_i;
                loop {
                    if x1_i >= x1.total_digits() || self_i >= self.total_digits() {
                        break
                    }
                    let tmp0 =
                        widen_mul_add(x0.get_unchecked(x0_i), x1.get_unchecked(x1_i), carry0);
                    carry0 = tmp0.1;
                    let tmp1 = widen_add(self.get_unchecked(self_i), tmp0.0, carry1);
                    carry1 = tmp1.1;
                    *self.get_unchecked_mut(self_i) = tmp1.0;
                    x1_i += 1;
                    self_i += 1;
                }
                // handle the last short multiplication carry if `self` continues
                if self_i < self.total_digits() {
                    let tmp = widen_add(self.get_unchecked(self_i), carry0, carry1);
                    *self.get_unchecked_mut(self_i) = tmp.0;
                    carry1 = tmp.1;
                    self_i += 1;
                    // handle arbitrarily many addition carries
                    while self_i < self.total_digits() && carry1 != 0 {
                        let tmp = widen_add(self.get_unchecked(self_i), carry1, 0);
                        *self.get_unchecked_mut(self_i) = tmp.0;
                        carry1 = tmp.1;
                        self_i += 1;
                    }
                }
            });
        }
        self.clear_unused_bits();
    }

    /// Arbitrarily-signed-multiplies `lhs` by `rhs` and add-assigns the product
    /// to `self`. Has the same behavior as [Bits::arb_umul_add_] except that is
    /// interprets the arguments as signed. `lhs` and `rhs` are marked
    /// mutable but their values are not changed by this function.
    #[const_fn(cfg(feature = "const_support"))]
    pub const fn arb_imul_add_(&mut self, lhs: &mut Self, rhs: &mut Self) {
        let lhs_msb = lhs.msb();
        let rhs_msb = rhs.msb();
        lhs.neg_(lhs_msb);
        rhs.neg_(rhs_msb);
        self.neg_(lhs_msb != rhs_msb);
        self.arb_umul_add_(lhs, rhs);
        lhs.neg_(lhs_msb);
        rhs.neg_(rhs_msb);
        self.neg_(lhs_msb != rhs_msb);
    }
}