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
234
235
236
237
238
239
use std::{
marker::PhantomData,
mem,
ops::{Deref, DerefMut},
ptr::NonNull,
};
use crate::{HashSet, HashSetCell, HashSetError};
/// A `HashSet` wrapper which can be instantiated from Solana account bytes
/// without copying them.
#[derive(Debug)]
pub struct HashSetZeroCopy<'a> {
pub hash_set: mem::ManuallyDrop<HashSet>,
_marker: PhantomData<&'a ()>,
}
impl<'a> HashSetZeroCopy<'a> {
// TODO(vadorovsky): Add a non-mut method: `from_bytes_zero_copy`.
/// Casts a byte slice into `HashSet`.
///
/// # Purpose
///
/// This method is meant to be used mostly in Solana programs, where memory
/// constraints are tight and we want to make sure no data is copied.
///
/// # Safety
///
/// This is highly unsafe. Ensuring the alignment and that the slice
/// provides actual data of the hash set is the caller's responsibility.
///
/// Calling it in async context (or anyhwere where the underlying data can
/// be moved in the memory) is certainly going to cause undefined behavior.
pub unsafe fn from_bytes_zero_copy_mut(bytes: &'a mut [u8]) -> Result<Self, HashSetError> {
if bytes.len() < HashSet::non_dyn_fields_size() {
return Err(HashSetError::BufferSize(
HashSet::non_dyn_fields_size(),
bytes.len(),
));
}
let capacity_values = usize::from_le_bytes(bytes[0..8].try_into().unwrap());
let sequence_threshold = usize::from_le_bytes(bytes[8..16].try_into().unwrap());
let offset = HashSet::non_dyn_fields_size() + mem::size_of::<usize>();
let values_size = mem::size_of::<Option<HashSetCell>>() * capacity_values;
let expected_size = HashSet::non_dyn_fields_size() + values_size;
if bytes.len() < expected_size {
return Err(HashSetError::BufferSize(expected_size, bytes.len()));
}
let buckets =
NonNull::new(bytes.as_mut_ptr().add(offset) as *mut Option<HashSetCell>).unwrap();
Ok(Self {
hash_set: mem::ManuallyDrop::new(HashSet {
capacity: capacity_values,
sequence_threshold,
buckets,
}),
_marker: PhantomData,
})
}
/// Casts a byte slice into `HashSet` and then initializes it.
///
/// * `bytes` is casted into a reference of `HashSet` and used as
/// storage for the struct.
/// * `capacity_indices` indicates the size of the indices table. It should
/// already include a desired load factor and be greater than the expected
/// number of elements to avoid filling the set too early and avoid
/// creating clusters.
/// * `capacity_values` indicates the size of the values array. It should be
/// equal to the number of expected elements, without load factor.
/// * `sequence_threshold` indicates a difference of sequence numbers which
/// make elements of the has set expired. Expiration means that they can
/// be replaced during insertion of new elements with sequence numbers
/// higher by at least a threshold.
///
/// # Purpose
///
/// This method is meant to be used mostly in Solana programs to initialize
/// a new account which is supposed to store the hash set.
///
/// # Safety
///
/// This is highly unsafe. Ensuring the alignment and that the slice has
/// a correct size, which is able to fit the hash set, is the caller's
/// responsibility.
///
/// Calling it in async context (or anywhere where the underlying data can
/// be moved in memory) is certainly going to cause undefined behavior.
pub unsafe fn from_bytes_zero_copy_init(
bytes: &'a mut [u8],
capacity_values: usize,
sequence_threshold: usize,
) -> Result<Self, HashSetError> {
if bytes.len() < HashSet::non_dyn_fields_size() {
return Err(HashSetError::BufferSize(
HashSet::non_dyn_fields_size(),
bytes.len(),
));
}
bytes[0..8].copy_from_slice(&capacity_values.to_le_bytes());
bytes[8..16].copy_from_slice(&sequence_threshold.to_le_bytes());
bytes[16..24].copy_from_slice(&0_usize.to_le_bytes());
let hash_set = Self::from_bytes_zero_copy_mut(bytes)?;
for i in 0..capacity_values {
std::ptr::write(hash_set.hash_set.buckets.as_ptr().add(i), None);
}
Ok(hash_set)
}
}
impl Drop for HashSetZeroCopy<'_> {
fn drop(&mut self) {
// SAFETY: Don't do anything here! Why?
//
// * Primitive fields of `HashSet` implement `Copy`, therefore `drop()`
// has no effect on them - Rust drops them when they go out of scope.
// * Don't drop the dynamic fields (`indices` and `values`). In
// `HashSetZeroCopy`, they are backed by buffers provided by the
// caller. These buffers are going to be eventually deallocated.
// Performing an another `drop()` here would result double `free()`
// which would result in aborting the program (either with `SIGABRT`
// or `SIGSEGV`).
}
}
impl Deref for HashSetZeroCopy<'_> {
type Target = HashSet;
fn deref(&self) -> &Self::Target {
&self.hash_set
}
}
impl DerefMut for HashSetZeroCopy<'_> {
fn deref_mut(&mut self) -> &mut Self::Target {
&mut self.hash_set
}
}
#[cfg(test)]
mod test {
use ark_bn254::Fr;
use ark_ff::UniformRand;
use num_bigint::BigUint;
use rand::{thread_rng, Rng};
use super::*;
#[test]
fn test_load_from_bytes() {
const VALUES: usize = 4800;
const SEQUENCE_THRESHOLD: usize = 2400;
// Create a buffer with random bytes.
let mut bytes = vec![0u8; HashSet::size_in_account(VALUES)];
thread_rng().fill(bytes.as_mut_slice());
// Create random nullifiers.
let mut rng = thread_rng();
let nullifiers: [BigUint; 2400] =
std::array::from_fn(|_| BigUint::from(Fr::rand(&mut rng)));
// Initialize a hash set on top of a byte slice.
{
let mut hs = unsafe {
HashSetZeroCopy::from_bytes_zero_copy_init(
bytes.as_mut_slice(),
VALUES,
SEQUENCE_THRESHOLD,
)
.unwrap()
};
// Ensure that the underlying data were properly initialized.
assert_eq!(hs.hash_set.get_capacity(), VALUES);
assert_eq!(hs.hash_set.sequence_threshold, SEQUENCE_THRESHOLD);
for i in 0..VALUES {
assert!(unsafe { &*hs.hash_set.buckets.as_ptr().add(i) }.is_none());
}
for (seq, nullifier) in nullifiers.iter().enumerate() {
let index = hs.insert(nullifier, seq).unwrap();
hs.mark_with_sequence_number(index, seq).unwrap();
}
}
// Read the hash set from buffers again.
{
let mut hs =
unsafe { HashSetZeroCopy::from_bytes_zero_copy_mut(bytes.as_mut_slice()).unwrap() };
for (seq, nullifier) in nullifiers.iter().enumerate() {
assert!(hs.contains(nullifier, Some(seq)).unwrap());
}
for (seq, nullifier) in nullifiers.iter().enumerate() {
hs.insert(nullifier, 2400 + seq).unwrap();
}
drop(hs);
}
// Make a copy of hash set from the same buffers.
{
let hs = unsafe { HashSet::from_bytes_copy(bytes.as_mut_slice()).unwrap() };
for (seq, nullifier) in nullifiers.iter().enumerate() {
assert!(hs.contains(nullifier, Some(2400 + seq)).unwrap());
}
}
}
#[test]
fn test_buffer_size_error() {
const VALUES: usize = 4800;
const SEQUENCE_THRESHOLD: usize = 2400;
let mut invalid_bytes = vec![0_u8; 256];
let res = unsafe {
HashSetZeroCopy::from_bytes_zero_copy_init(
invalid_bytes.as_mut_slice(),
VALUES,
SEQUENCE_THRESHOLD,
)
};
assert!(matches!(res, Err(HashSetError::BufferSize(_, _))));
}
}