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
#![no_std]
pub use generic_array;
#[cfg(feature = "block-padding")]
pub use block_padding;

use core::{slice, convert::TryInto};
use generic_array::{GenericArray, ArrayLength};
#[cfg(feature = "block-padding")]
use block_padding::{Padding, PadError};

/// Buffer for block processing of data
#[derive(Clone, Default)]
pub struct BlockBuffer<BlockSize: ArrayLength<u8>>  {
    buffer: GenericArray<u8, BlockSize>,
    pos: usize,
}

impl<BlockSize: ArrayLength<u8>> BlockBuffer<BlockSize> {
    /// Process data in `input` in blocks of size `BlockSize` using function `f`.
    #[inline]
    pub fn input_block(
        &mut self, mut input: &[u8], mut f: impl FnMut(&GenericArray<u8, BlockSize>),
    ) {
        let r = self.remaining();
        if input.len() < r {
            let n = input.len();
            self.buffer[self.pos..self.pos + n].copy_from_slice(input);
            self.pos += n;
            return;
        }
        if self.pos != 0 && input.len() >= r {
            let (l, r) = input.split_at(r);
            input = r;
            self.buffer[self.pos..].copy_from_slice(l);
            f(&self.buffer);
        }

        let mut chunks_iter = input.chunks_exact(self.size());
        for chunk in &mut chunks_iter {
            f(chunk.try_into().unwrap());
        }
        let rem = chunks_iter.remainder();

        // Copy any remaining data into the buffer.
        self.buffer[..rem.len()].copy_from_slice(rem);
        self.pos = rem.len();
    }

    /// Process data in `input` in blocks of size `BlockSize` using function `f`, which accepts
    /// slice of blocks.
    #[inline]
    pub fn input_blocks(
        &mut self, mut input: &[u8], mut f: impl FnMut(&[GenericArray<u8, BlockSize>]),
    ) {
        let r = self.remaining();
        if input.len() < r {
            let n = input.len();
            self.buffer[self.pos..self.pos + n].copy_from_slice(input);
            self.pos += n;
            return;
        }
        if self.pos != 0 && input.len() >= r {
            let (l, r) = input.split_at(r);
            input = r;
            self.buffer[self.pos..].copy_from_slice(l);
            self.pos = 0;
            f(slice::from_ref(&self.buffer));
        }

        // While we have at least a full buffer size chunks's worth of data,
        // process its data without copying into the buffer
        let n_blocks = input.len()/self.size();
        let (left, right) = input.split_at(n_blocks*self.size());
        // SAFETY: we guarantee that `blocks` does not point outside of `input` 
        let blocks = unsafe {
            slice::from_raw_parts(
                left.as_ptr() as *const GenericArray<u8, BlockSize>,
                n_blocks,
            )
        };
        f(blocks);

        // Copy remaining data into the buffer.
        let n = right.len();
        self.buffer[..n].copy_from_slice(right);
        self.pos = n;
    }

    /// Variant that doesn't flush the buffer until there's additional
    /// data to be processed. Suitable for tweakable block ciphers
    /// like Threefish that need to know whether a block is the *last*
    /// data block before processing it.
    #[inline]
    pub fn input_lazy(
        &mut self, mut input: &[u8], mut f: impl FnMut(&GenericArray<u8, BlockSize>),
    ) {
        let r = self.remaining();
        if input.len() <= r {
            let n = input.len();
            self.buffer[self.pos..self.pos + n].copy_from_slice(input);
            self.pos += n;
            return;
        }
        if self.pos != 0 && input.len() > r {
            let (l, r) = input.split_at(r);
            input = r;
            self.buffer[self.pos..].copy_from_slice(l);
            f(&self.buffer);
        }

        while input.len() > self.size() {
            let (block, r) = input.split_at(self.size());
            input = r;
            f(block.try_into().unwrap());
        }

        self.buffer[..input.len()].copy_from_slice(input);
        self.pos = input.len();
    }

    /// Pad buffer with `prefix` and make sure that internall buffer
    /// has at least `up_to` free bytes. All remaining bytes get
    /// zeroed-out.
    #[inline]
    fn digest_pad(
        &mut self, up_to: usize, mut f: impl FnMut(&GenericArray<u8, BlockSize>),
    ) {
        if self.pos == self.size() {
            f(&self.buffer);
            self.pos = 0;
        }
        self.buffer[self.pos] = 0x80;
        self.pos += 1;

        set_zero(&mut self.buffer[self.pos..]);

        if self.remaining() < up_to {
            f(&self.buffer);
            set_zero(&mut self.buffer[..self.pos]);
        }
    }

    /// Pad message with 0x80, zeros and 64-bit message length
    /// using big-endian byte order
    #[inline]
    pub fn len64_padding_be(
        &mut self, data_len: u64, mut f: impl FnMut(&GenericArray<u8, BlockSize>),
    ) {
        self.digest_pad(8, &mut f);
        let b = data_len.to_be_bytes();
        let n = self.buffer.len() - b.len();
        self.buffer[n..].copy_from_slice(&b);
        f(&self.buffer);
        self.pos = 0;
    }

    /// Pad message with 0x80, zeros and 64-bit message length
    /// using little-endian byte order
    #[inline]
    pub fn len64_padding_le(
        &mut self, data_len: u64, mut f: impl FnMut(&GenericArray<u8, BlockSize>),
    ) {
        self.digest_pad(8, &mut f);
        let b = data_len.to_le_bytes();
        let n = self.buffer.len() - b.len();
        self.buffer[n..].copy_from_slice(&b);
        f(&self.buffer);
        self.pos = 0;
    }

    /// Pad message with 0x80, zeros and 128-bit message length
    /// using big-endian byte order
    #[inline]
    pub fn len128_padding_be(
        &mut self, data_len: u128, mut f: impl FnMut(&GenericArray<u8, BlockSize>),
    ) {
        self.digest_pad(16, &mut f);
        let b = data_len.to_be_bytes();
        let n = self.buffer.len() - b.len();
        self.buffer[n..].copy_from_slice(&b);
        f(&self.buffer);
        self.pos = 0;
    }

    /// Pad message with a given padding `P`
    ///
    /// Returns `PadError` if internall buffer is full, which can only happen if
    /// `input_lazy` was used.
    #[cfg(feature = "block-padding")]
    #[inline]
    pub fn pad_with<P: Padding>(&mut self)
        -> Result<&mut GenericArray<u8, BlockSize>, PadError>
    {
        P::pad_block(&mut self.buffer[..], self.pos)?;
        self.pos = 0;
        Ok(&mut self.buffer)
    }

    /// Return size of the internall buffer in bytes
    #[inline]
    pub fn size(&self) -> usize {
        BlockSize::to_usize()
    }

    /// Return current cursor position
    #[inline]
    pub fn position(&self) -> usize {
        self.pos
    }

    /// Return number of remaining bytes in the internall buffer
    #[inline]
    pub fn remaining(&self) -> usize {
        self.size() - self.pos
    }

    /// Reset buffer by setting cursor position to zero
    #[inline]
    pub fn reset(&mut self)  {
        self.pos = 0
    }
}

/// Sets all bytes in `dst` to zero
#[inline(always)]
fn set_zero(dst: &mut [u8]) {
    // SAFETY: we overwrite valid memory behind `dst`
    // note: loop is not used here because it produces
    // unnecessary branch which tests for zero-length slices
    unsafe {
        core::ptr::write_bytes(dst.as_mut_ptr(), 0, dst.len());
    }
}