libnyx_fuzz_runner 0.1.3

libnyx fork only for publishing
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

use std::sync::atomic::compiler_fence;
use std::sync::atomic::Ordering;

// we expect this to be a nop.
// but in some extreme cases, this
/*
use std::sync::atomic::compiler_fence;
use std::sync::atomic::Ordering;

fn barrier() {
compiler_fence(Ordering::SeqCst);
}

pub fn read2(data: &mut u32) -> u32{
    let a = *data;
    barrier();
    let b = *data;
    return a.wrapping_add(b);
}
*/

//compiles to
/*
        mov     eax, dword ptr [rdi]
        add     eax, dword ptr [rdi]
        ret
*/
//while the second access gets optimized out without the barrier.
//To ensure that reads/writes to the shared memory buffer actually are executed, we use mem_barrier to lightweight synchronize the values.

pub fn mem_barrier() {
    compiler_fence(Ordering::SeqCst);
}