bitptr 0.2.1

`core::ptr` for bit offsets
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

use crate::BitPtrMut;
use core::ptr;


/// Swaps `bit_count` bits between the two regions of memory beginning at `x` and `y`. The two regions must *not* overlap.
///
/// `bitptr` does not provide a `swap` function for memory which might overlap.
///
/// The copy is "untyped" in the sense that data may be uninitialized. The initialization state is preserved exactly.
///
///
/// # Safety
///
/// Behavior is undefined if any of the following conditions are violated:
/// - `x.floor_byte()` must be [valid](core::ptr#safety) for reads of `((x.subbyte_bit().get() as usize) + bit_count).div_ceil(8)` bytes.
/// - `y.floor_byte()` must be [valid](core::ptr#safety) for writes of `((y.subbyte_bit().get() as usize) + bit_count).div_ceil(8)` bytes.
/// - The region of memory beginning at `x` with a size of `bit_count` bits must *not* overlap with the region of memory beginning at `y` with the same size.
///   The byte region may overlap. The relevant bits themselves may not.
///
///
/// # Footguns
///
/// Make sure to account for endianness.
/// ```rust should_panic
/// #![cfg(not(target_endian = "big"))]
///
/// use bitptr::{ BitPtr, BitPtrMut };
///
/// let mut x = 0b_0101101110010110_u16;
/// //                    ^^^^^^^ This is the region that is 'supposed to' be read/written.
/// let mut y = 0b_1110100011010010_u16;
/// //                    ^^^^^^^ This is the region that is 'supposed to' be read/written.
///
/// let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 7) };
/// let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 3) };
///
/// unsafe { bitptr::swap_nonoverlapping(xptr, yptr, 7); }
/// assert_eq!(x, 0b_0101101010001110_u16);
/// assert_eq!(y, 0b_1111100101010010_u16);
/// ```
/// On a little-endian system, the `assert_eq!` in the code above will panic.
///
/// Explicitely converting from native-endian to big-endian before, and big-endian to native-endian after, can solve this issue.
/// ```rust
/// use bitptr::{ BitPtr, BitPtrMut };
///
/// let mut x = 0b_0101101110010110_u16.to_be();
/// //                    ^^^^^^^ This is the region that is 'supposed to' be read/written.
/// let mut y = 0b_1110100011010010_u16.to_be();
/// //                    ^^^^^^^ This is the region that is 'supposed to' be read/written.
///
/// let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 7) };
/// let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 3) };
///
/// unsafe { bitptr::swap_nonoverlapping(xptr, yptr, 7); }
/// assert_eq!(u16::from_be(x), 0b_0101101010001110_u16);
/// assert_eq!(u16::from_be(y), 0b_1111100101010010_u16);
/// ```
///
///
/// ---
/// Analagous to [`ptr::swap_nonoverlapping`](core::ptr::swap_nonoverlapping).
pub unsafe fn swap_nonoverlapping(x : BitPtrMut, y : BitPtrMut, bit_count : usize) {
    if (bit_count == 0) { return; }

    let (x_byte, x_bit,) = x.as_inner();
    let x_bit_l = x_bit.get() as usize;
    let x_bit_r = (8isize - ((x_bit_l + bit_count) as isize)).rem_euclid(8);
    let (y_byte, y_bit,) = y.as_inner();
    let y_bit_l = y_bit.get() as usize;
    let y_bit_r = (8isize - ((y_bit_l + bit_count) as isize)).rem_euclid(8);

    let mut x_rolling = u32::from_be(unsafe { ptr::read(x_byte.byte_sub(2) as *const u32) });
    let mut y_rolling = u32::from_be(unsafe { ptr::read(y_byte.byte_sub(2) as *const u32) });

    let x_byte_count = (x_bit_l + bit_count).div_ceil(8);
    let y_byte_count = (y_bit_l + bit_count).div_ceil(8);
    for offset in 0..(x_byte_count.max(y_byte_count)) {
        let x_byte = unsafe { x_byte.byte_add(offset) };
        let y_byte = unsafe { y_byte.byte_add(offset) };

        // Get masks over the bits to write.
        let mut x_mask = u8::MAX;
        let mut y_mask = u8::MAX;
        if (offset == 0) {
            x_mask = x_mask << x_bit_l >> x_bit_l;
            y_mask = y_mask << y_bit_l >> y_bit_l;
        }
        match ((offset + 2).saturating_sub(x_byte_count)) {
            0 => { }
            1 => { x_mask = x_mask >> x_bit_r << x_bit_r; },
            _ => { x_mask = 0b00000000; }
        }
        match ((offset + 2).saturating_sub(y_byte_count)) {
            0 => { }
            1 => { y_mask = y_mask >> y_bit_r << y_bit_r; },
            _ => { y_mask = 0b00000000; }
        }

        // Build the bytes that will be written.
        let x_src_b = ((y_rolling << y_bit_l >> (8 + x_bit_l)) & 0b11111111) as u8;
        let y_src_b = ((x_rolling << x_bit_l >> (8 + y_bit_l)) & 0b11111111) as u8;

        // Get the bytes to edit.
        let mut x_dst_b = unsafe { *x_byte };
        let mut y_dst_b = unsafe { *y_byte };

        // Wipe the bits that will be overwritten.
        x_dst_b &= ! x_mask;
        y_dst_b &= ! y_mask;

        // Write the relevant bits.
        x_dst_b |= x_src_b & x_mask;
        y_dst_b |= y_src_b & y_mask;

        // Overwrite the bytes.
        unsafe { *x_byte = x_dst_b; }
        unsafe { *y_byte = y_dst_b; }

        // Roll the source values.
        x_rolling = (x_rolling << 8) | ((unsafe { x_byte.byte_add(1).read() }) as u32);
        y_rolling = (y_rolling << 8) | ((unsafe { y_byte.byte_add(1).read() }) as u32);

    }
}


#[cfg(test)]
mod tests {
    use super::*;


    #[test]
    fn swap_within_byte_boundary() {
        let mut x = 0b0101101110010110u16.to_be();
        let mut y = 0b1110100011010010u16.to_be();

        let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 9) };
        let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 3) };

        // Swap and check final values.
        unsafe { swap_nonoverlapping(xptr, yptr, 4); }
        assert_eq!(u16::from_be(x), 0b0101101110100110u16);
        assert_eq!(u16::from_be(y), 0b1110010011010010u16);
    }


    #[test]
    fn swap_across_byte_boundary() {
        let mut x = 0b0101101110010110u16.to_be();
        let mut y = 0b1110100011010010u16.to_be();

        let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 7) };
        let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 3) };

        // Swap and check final values.
        unsafe { swap_nonoverlapping(xptr, yptr, 7); }
        assert_eq!(u16::from_be(x), 0b0101101010001110u16);
        assert_eq!(u16::from_be(y), 0b1111100101010010u16);
    }


    #[test]
    fn swap_aligned_start() {
        let mut x = 0b0101101110010110u16.to_be();
        let mut y = 0b1110100011010010u16.to_be();

        let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 8) };
        let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 0) };

        // Swap and check final values.
        unsafe { swap_nonoverlapping(xptr, yptr, 5); }
        assert_eq!(u16::from_be(x), 0b0101101111101110u16);
        assert_eq!(u16::from_be(y), 0b1001000011010010u16);
    }


    #[test]
    fn swap_aligned_end() {
        let mut x = 0b0101101110010110u16.to_be();
        let mut y = 0b1110100011010010u16.to_be();

        let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 5) };
        let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 13) };

        // Swap and check final values.
        unsafe { swap_nonoverlapping(xptr, yptr, 3); }
        assert_eq!(u16::from_be(x), 0b0101101010010110u16);
        assert_eq!(u16::from_be(y), 0b1110100011010011u16);
    }


    #[test]
    fn swap_different_byte_width() {
        let mut x = 0b0101101110010110u16.to_be();
        let mut y = 0b1110100011010010u16.to_be();

        let xptr = unsafe { BitPtrMut::new_with_offset(&mut x as *mut _ as *mut _, 2) };
        let yptr = unsafe { BitPtrMut::new_with_offset(&mut y as *mut _ as *mut _, 6) };

        // Swap and check final values.
        unsafe { swap_nonoverlapping(xptr, yptr, 5); }
        assert_eq!(u16::from_be(x), 0b0100110110010110u16);
        assert_eq!(u16::from_be(y), 0b1110100110110010u16);
    }


}