async-utils 0.1.1

nothing to see here yet :)
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

use std::sync::atomic::{self, AtomicUsize, Ordering};

/// Equivalent to `atomic::fence(Ordering::SeqCst)`, but in some cases faster.
#[inline(always)]
pub fn full_fence() {
    if cfg!(all(
        any(target_arch = "x86", target_arch = "x86_64"),
        not(miri)
    )) {
        // HACK(stjepang): On x86 architectures there are two different ways of executing
        // a `SeqCst` fence.
        //
        // 1. `atomic::fence(SeqCst)`, which compiles into a `mfence` instruction.
        // 2. `_.compare_exchange(_, _, SeqCst, SeqCst)`, which compiles into a `lock cmpxchg` instruction.
        //
        // Both instructions have the effect of a full barrier, but empirical benchmarks have shown
        // that the second one is sometimes a bit faster.
        //
        // The ideal solution here would be to use inline assembly, but we're instead creating a
        // temporary atomic variable and compare-and-exchanging its value. No sane compiler to
        // x86 platforms is going to optimize this away.
        atomic::compiler_fence(Ordering::SeqCst);
        let a = AtomicUsize::new(0);
        let _ = a.compare_exchange(0, 1, Ordering::SeqCst, Ordering::SeqCst);
        atomic::compiler_fence(Ordering::SeqCst);
    } else {
        atomic::fence(Ordering::SeqCst);
    }
}