Tracking Issue: fetch_nand

[alexcrichton]

Tracking issue for fetch_nand on integer atomics (it's stable for bool).

On android, it appears that these functions give the wrong answer. See #12964 (comment) for some analysis, and the relevant logs are:

    #[test]
    fn uint_nand() {
        let x = AtomicUint::new(0xf731);
        assert_eq!(x.fetch_nand(0x137f, SeqCst), 0xf731);
        assert_eq!(x.load(SeqCst), !(0xf731 & 0x137f));
    }

    #[test]
    fn int_nand() {
        let x = AtomicInt::new(0xf731);
        assert_eq!(x.fetch_nand(0x137f, SeqCst), 0xf731);
        assert_eq!(x.load(SeqCst), !(0xf731 & 0x137f));
    }

---- sync::atomics::test::int_nand stdout ----
    task 'sync::atomics::test::int_nand' failed at 'assertion failed: `(left == right) && (right == left)` (left: `78`, right: `-4914`)', /home/rustbuild/src/rust-buildbot/slave/auto-linux-64-x-android-t/build/src/libstd/sync/atomics.rs:1005

---- sync::atomics::test::uint_nand stdout ----
    task 'sync::atomics::test::uint_nand' failed at 'assertion failed: `(left == right) && (right == left)` (left: `78`, right: `4294962382`)', /home/rustbuild/src/rust-buildbot/slave/auto-linux-64-x-android-t/build/src/libstd/sync/atomics.rs:977  

[steveklabnik]

Triage: i don't have android to test.

[steveklabnik]

Triage: same as in 2015.

[Mark-Simulacrum]

I unfortunately do not have an android either, and after spending multiple hours trying to get an emulator setup I have decided to not try any more... the code below I believe is the test for this problem today. If someone has a working android version, and can test it, that would be great.

#![feature(core_intrinsics)]

use std::sync::atomic::{AtomicUsize, AtomicIsize};
use std::sync::atomic::Ordering::*;
use std::intrinsics::atomic_nand;

#[test]
fn uint_nand() {
    let mut x = AtomicUsize::new(0xf731);
    unsafe {
        assert_eq!(atomic_nand(x.get_mut() as *mut usize, 0x137f), 0xf731);
    }
    assert_eq!(x.load(SeqCst), !(0xf731 & 0x137f));
}

#[test]
fn int_nand() {
    let mut x = AtomicIsize::new(0xf731);
    unsafe {
        assert_eq!(atomic_nand(x.get_mut() as *mut isize, 0x137f), 0xf731);
    }
    assert_eq!(x.load(SeqCst), !(0xf731 & 0x137f));
}

[Dushistov]

@Mark-Simulacrum

I unfortunately do not have an android either, and after spending multiple hours trying to > get an emulator setup I have decided to not try any more... the code below I believe is
the test for this problem today. If someone has a working android version, and can test it, > that would be great.

I've run it on android phone (cargo test --target=arm-linux-androideabi/rustc 1.25.0-nightly (bd98fe0c0 2018-02-06)),
all looks ok:

$ cargo test --target=arm-linux-androideabi
   Compiling android-atomic-13226 v0.1.0 (file:///home/evgeniy/projects/rust-infra/bugs/android-atomic-13226)
    Finished dev [unoptimized + debuginfo] target(s) in 0.88 secs
     Running target/arm-linux-androideabi/debug/deps/android_atomic_13226-a6d2087ae9be837c
Current directory: /home/evgeniy/projects/rust-infra/bugs/android-atomic-13226
/home/evgeniy/projects/rust-infra/bugs/android-atomic-13226/target/arm...26-a6d2087ae9be837c: 1 file pushed. 2.4 MB/s (5396036 bytes in 2.134s

running 2 tests
test uint_nand ... ok
test int_nand ... ok

test result: ok. 2 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

ADB_SUCCESS!!!!!!!!!!!!!!
ADB_SUCCESS!!!!!!!!!!!!!!
[evgeniy@15inch android-atomic-13226]$ adb shell
shell@hammerhead:/ $ uname -a
Linux localhost 3.4.0-gcf10b7e #1 SMP PREEMPT Mon Sep 19 22:14:08 UTC 2016 armv7l

[Mark-Simulacrum]

@alexcrichton looks like we may be able to enable these on Android.

[alexcrichton]

Hurray!

[SimonSapin]

It looks like the only issue (incorrect behavior on Android) has been resolved.

@rfcbot fcp merge

[rfcbot]

Team member @SimonSapin has proposed to merge this. The next step is review by the rest of the tagged teams:

• @Kimundi
• @SimonSapin
• @alexcrichton
• @aturon
• @dtolnay
• @sfackler
• @withoutboats

No concerns currently listed.

Once a majority of reviewers approve (and none object), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[rfcbot]

🔔 This is now entering its final comment period, as per the review above. 🔔

[rfcbot]

The final comment period is now complete.