novakv 1.0.1

An embedded, ordered key-value store for Rust.
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

const POLY: u32 = 0x82f63b78;
const MASK_DELTA: u32 = 0xa282ead8;
const TABLE0: [u32; 256] = make_table0();
const TABLE1: [u32; 256] = make_table_next(TABLE0, TABLE0);
const TABLE2: [u32; 256] = make_table_next(TABLE1, TABLE0);
const TABLE3: [u32; 256] = make_table_next(TABLE2, TABLE0);

/// CRC32C of `data`, continued from `init_crc`. Uses a hardware
/// CRC32C instruction when the running CPU has one (ARM `crc`,
/// x86-64 SSE4.2), else a portable slicing-by-4 table. Every
/// path is bit-identical: the hardware `crc32c` instruction
/// computes the same Castagnoli CRC32C as the table.
pub fn extend(init_crc: u32, data: &[u8]) -> u32 {
    #[cfg(target_arch = "aarch64")]
    {
        if std::arch::is_aarch64_feature_detected!("crc") {
            // SAFETY: gated by runtime detection of the `crc`
            // feature, which `extend_hw_aarch64` requires.
            return unsafe { extend_hw_aarch64(init_crc, data) };
        }
    }
    #[cfg(target_arch = "x86_64")]
    {
        if std::is_x86_feature_detected!("sse4.2") {
            // SAFETY: gated by runtime detection of SSE4.2,
            // which `extend_hw_x86_64` requires.
            return unsafe { extend_hw_x86_64(init_crc, data) };
        }
    }
    extend_software(init_crc, data)
}

/// Portable slicing-by-4 table fallback. Used when no hardware
/// CRC32C instruction is available.
fn extend_software(init_crc: u32, data: &[u8]) -> u32 {
    let mut crc = init_crc ^ 0xffff_ffff;
    let mut offset = 0usize;
    while offset + 16 <= data.len() {
        crc = step4(crc, data, offset);
        crc = step4(crc, data, offset + 4);
        crc = step4(crc, data, offset + 8);
        crc = step4(crc, data, offset + 12);
        offset += 16;
    }
    while offset + 4 <= data.len() {
        crc = step4(crc, data, offset);
        offset += 4;
    }
    while offset < data.len() {
        let index = ((crc as u8) ^ data[offset]) as usize;
        crc = TABLE0[index] ^ (crc >> 8);
        offset += 1;
    }
    crc ^ 0xffff_ffff
}

/// ARM CRC32C path. The `crc` target feature provides the
/// `crc32c{b,h,w,d}` instructions used here.
#[cfg(target_arch = "aarch64")]
#[target_feature(enable = "crc")]
unsafe fn extend_hw_aarch64(init_crc: u32, data: &[u8]) -> u32 {
    use std::arch::aarch64::{__crc32cb, __crc32cd, __crc32ch, __crc32cw};
    let mut crc = init_crc ^ 0xffff_ffff;
    let mut chunks = data.chunks_exact(8);
    for chunk in &mut chunks {
        crc = __crc32cd(crc, u64::from_le_bytes(chunk.try_into().unwrap()));
    }
    let mut rem = chunks.remainder();
    if rem.len() >= 4 {
        crc = __crc32cw(crc, u32::from_le_bytes(rem[..4].try_into().unwrap()));
        rem = &rem[4..];
    }
    if rem.len() >= 2 {
        crc = __crc32ch(crc, u16::from_le_bytes(rem[..2].try_into().unwrap()));
        rem = &rem[2..];
    }
    if let Some(&byte) = rem.first() {
        crc = __crc32cb(crc, byte);
    }
    crc ^ 0xffff_ffff
}

/// x86-64 CRC32C path (SSE4.2 `crc32` instruction).
#[cfg(target_arch = "x86_64")]
#[target_feature(enable = "sse4.2")]
unsafe fn extend_hw_x86_64(init_crc: u32, data: &[u8]) -> u32 {
    use std::arch::x86_64::{_mm_crc32_u16, _mm_crc32_u32, _mm_crc32_u64, _mm_crc32_u8};
    let mut crc64 = u64::from(init_crc ^ 0xffff_ffff);
    let mut chunks = data.chunks_exact(8);
    for chunk in &mut chunks {
        crc64 = _mm_crc32_u64(crc64, u64::from_le_bytes(chunk.try_into().unwrap()));
    }
    let mut crc = crc64 as u32;
    let mut rem = chunks.remainder();
    if rem.len() >= 4 {
        crc = _mm_crc32_u32(crc, u32::from_le_bytes(rem[..4].try_into().unwrap()));
        rem = &rem[4..];
    }
    if rem.len() >= 2 {
        crc = _mm_crc32_u16(crc, u16::from_le_bytes(rem[..2].try_into().unwrap()));
        rem = &rem[2..];
    }
    if let Some(&byte) = rem.first() {
        crc = _mm_crc32_u8(crc, byte);
    }
    crc ^ 0xffff_ffff
}

#[inline]
fn step4(crc: u32, data: &[u8], offset: usize) -> u32 {
    let word = u32::from_le_bytes([
        data[offset],
        data[offset + 1],
        data[offset + 2],
        data[offset + 3],
    ]);
    let c = crc ^ word;
    TABLE3[(c & 0xff) as usize]
        ^ TABLE2[((c >> 8) & 0xff) as usize]
        ^ TABLE1[((c >> 16) & 0xff) as usize]
        ^ TABLE0[(c >> 24) as usize]
}

pub fn value(data: &[u8]) -> u32 {
    extend(0, data)
}

pub fn mask(crc: u32) -> u32 {
    crc.rotate_right(15).wrapping_add(MASK_DELTA)
}

pub fn unmask(masked_crc: u32) -> u32 {
    masked_crc.wrapping_sub(MASK_DELTA).rotate_right(17)
}

const fn make_table0() -> [u32; 256] {
    let mut table = [0u32; 256];
    let mut i = 0;
    while i < 256 {
        let mut crc = i as u32;
        let mut j = 0;
        while j < 8 {
            if crc & 1 != 0 {
                crc = (crc >> 1) ^ POLY;
            } else {
                crc >>= 1;
            }
            j += 1;
        }
        table[i] = crc;
        i += 1;
    }
    table
}

const fn make_table_next(previous: [u32; 256], base: [u32; 256]) -> [u32; 256] {
    let mut table = [0u32; 256];
    let mut i = 0;
    while i < 256 {
        let crc = previous[i];
        table[i] = (crc >> 8) ^ base[(crc & 0xff) as usize];
        i += 1;
    }
    table
}