revm_bytecode/legacy/
analysis.rs

1use super::JumpTable;
2use crate::opcode;
3use bitvec::{bitvec, order::Lsb0, vec::BitVec};
4use primitives::Bytes;
5use std::vec::Vec;
6
7/// Analyzes the bytecode for use in [`LegacyAnalyzedBytecode`](crate::LegacyAnalyzedBytecode).
8///
9/// See [`LegacyAnalyzedBytecode`](crate::LegacyAnalyzedBytecode) for more details.
10///
11/// Prefer using [`LegacyAnalyzedBytecode::analyze`](crate::LegacyAnalyzedBytecode::analyze) instead.
12pub fn analyze_legacy(bytecode: Bytes) -> (JumpTable, Bytes) {
13    if bytecode.is_empty() {
14        return (JumpTable::default(), Bytes::from_static(&[opcode::STOP]));
15    }
16
17    let mut jumps: BitVec<u8> = bitvec![u8, Lsb0; 0; bytecode.len()];
18    let range = bytecode.as_ptr_range();
19    let start = range.start;
20    let mut iterator = start;
21    let end = range.end;
22    let mut opcode = 0;
23
24    while iterator < end {
25        opcode = unsafe { *iterator };
26        if opcode == opcode::JUMPDEST {
27            // SAFETY: Jumps are max length of the code
28            unsafe { jumps.set_unchecked(iterator.offset_from(start) as usize, true) }
29            iterator = unsafe { iterator.add(1) };
30        } else {
31            let push_offset = opcode.wrapping_sub(opcode::PUSH1);
32            if push_offset < 32 {
33                // SAFETY: Iterator access range is checked in the while loop
34                iterator = unsafe { iterator.add(push_offset as usize + 2) };
35            } else {
36                // SAFETY: Iterator access range is checked in the while loop
37                iterator = unsafe { iterator.add(1) };
38            }
39        }
40    }
41
42    let overflow_padding = (iterator as usize) - (end as usize);
43
44    let stop_padding = opcode::OpCode::info_by_op(opcode)
45        .map(|o| !o.is_terminating() as usize)
46        .unwrap_or(1);
47
48    let padding = overflow_padding + stop_padding;
49
50    let bytecode = if padding > 0 {
51        let mut padded = Vec::with_capacity(bytecode.len() + padding);
52        padded.extend_from_slice(&bytecode);
53        padded.resize(padded.len() + padding, 0);
54        Bytes::from(padded)
55    } else {
56        bytecode
57    };
58
59    (JumpTable::new(jumps), bytecode)
60}
61
62#[cfg(test)]
63mod tests {
64    use super::*;
65
66    #[test]
67    fn test_bytecode_ends_with_stop_no_padding_needed() {
68        let bytecode = vec![
69            opcode::PUSH1,
70            0x01,
71            opcode::PUSH1,
72            0x02,
73            opcode::ADD,
74            opcode::STOP,
75        ];
76        let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
77        assert_eq!(padded_bytecode.len(), bytecode.len());
78    }
79
80    #[test]
81    fn test_bytecode_ends_without_stop_requires_padding() {
82        let bytecode = vec![opcode::PUSH1, 0x01, opcode::PUSH1, 0x02, opcode::ADD];
83        let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
84        assert_eq!(padded_bytecode.len(), bytecode.len() + 1);
85    }
86
87    #[test]
88    fn test_bytecode_ends_with_push16_requires_17_bytes_padding() {
89        let bytecode = vec![opcode::PUSH1, 0x01, opcode::PUSH16];
90        let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
91        assert_eq!(padded_bytecode.len(), bytecode.len() + 17);
92    }
93
94    #[test]
95    fn test_bytecode_ends_with_push2_requires_2_bytes_padding() {
96        let bytecode = vec![opcode::PUSH1, 0x01, opcode::PUSH2, 0x02];
97        let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
98        assert_eq!(padded_bytecode.len(), bytecode.len() + 2);
99    }
100
101    #[test]
102    fn test_empty_bytecode_requires_stop() {
103        let bytecode = vec![];
104        let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
105        assert_eq!(padded_bytecode.len(), 1); // Just STOP
106    }
107
108    #[test]
109    fn test_bytecode_with_jumpdest_at_start() {
110        let bytecode = vec![opcode::JUMPDEST, opcode::PUSH1, 0x01, opcode::STOP];
111        let (jump_table, _) = analyze_legacy(bytecode.clone().into());
112        assert!(jump_table.is_valid(0)); // First byte should be a valid jumpdest
113    }
114
115    #[test]
116    fn test_bytecode_with_jumpdest_after_push() {
117        let bytecode = vec![opcode::PUSH1, 0x01, opcode::JUMPDEST, opcode::STOP];
118        let (jump_table, _) = analyze_legacy(bytecode.clone().into());
119        assert!(jump_table.is_valid(2)); // JUMPDEST should be at position 2
120    }
121
122    #[test]
123    fn test_bytecode_with_multiple_jumpdests() {
124        let bytecode = vec![
125            opcode::JUMPDEST,
126            opcode::PUSH1,
127            0x01,
128            opcode::JUMPDEST,
129            opcode::STOP,
130        ];
131        let (jump_table, _) = analyze_legacy(bytecode.clone().into());
132        assert!(jump_table.is_valid(0)); // First JUMPDEST
133        assert!(jump_table.is_valid(3)); // Second JUMPDEST
134    }
135
136    #[test]
137    fn test_bytecode_with_max_push32() {
138        let bytecode = vec![opcode::PUSH32];
139        let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
140        assert_eq!(padded_bytecode.len(), bytecode.len() + 33); // PUSH32 + 32 bytes + STOP
141    }
142
143    #[test]
144    fn test_bytecode_with_invalid_opcode() {
145        let bytecode = vec![0xFF, opcode::STOP]; // 0xFF is an invalid opcode
146        let (jump_table, _) = analyze_legacy(bytecode.clone().into());
147        assert!(!jump_table.is_valid(0)); // Invalid opcode should not be a jumpdest
148    }
149
150    #[test]
151    fn test_bytecode_with_sequential_pushes() {
152        let bytecode = vec![
153            opcode::PUSH1,
154            0x01,
155            opcode::PUSH2,
156            0x02,
157            0x03,
158            opcode::PUSH4,
159            0x04,
160            0x05,
161            0x06,
162            0x07,
163            opcode::STOP,
164        ];
165        let (jump_table, padded_bytecode) = analyze_legacy(bytecode.clone().into());
166        assert_eq!(padded_bytecode.len(), bytecode.len());
167        assert!(!jump_table.is_valid(0)); // PUSH1
168        assert!(!jump_table.is_valid(2)); // PUSH2
169        assert!(!jump_table.is_valid(5)); // PUSH4
170    }
171
172    #[test]
173    fn test_bytecode_with_jumpdest_in_push_data() {
174        let bytecode = vec![
175            opcode::PUSH2,
176            opcode::JUMPDEST, // This should not be treated as a JUMPDEST
177            0x02,
178            opcode::STOP,
179        ];
180        let (jump_table, _) = analyze_legacy(bytecode.clone().into());
181        assert!(!jump_table.is_valid(1)); // JUMPDEST in push data should not be valid
182    }
183
184    #[test]
185    fn test_terminating_opcodes_behavior() {
186        // Test all known terminating opcodes
187        let terminating_opcodes = [
188            opcode::STOP,
189            opcode::RETURN,
190            opcode::REVERT,
191            opcode::INVALID,
192            opcode::SELFDESTRUCT,
193        ];
194
195        for &terminating_opcode in &terminating_opcodes {
196            let bytecode = vec![opcode::PUSH1, 0x01, terminating_opcode];
197            let (_, padded_bytecode) = analyze_legacy(bytecode.clone().into());
198            assert_eq!(padded_bytecode.len(), bytecode.len());
199        }
200    }
201}
revm_bytecode/legacy/analysis.rs

revm_bytecode/legacy/
analysis.rs
