{i686,x86_64}-pc-windows-gnu Appveyor builds failing

[gnzlbg]

It appears that incorrect results are produced on windows when the gnu toolchain is used and NaNs are involved.

failures:
---- v64::f32x2_math_cos::cos stdout ----
thread 'v64::f32x2_math_cos::cos' panicked at 'assertion failed: `(left == right)`
  left: `f32x2(-0.00000004371139, -0.00000004371139)`,
 right: `f32x2(NaN, -0.00000004371139)`', src\v64.rs:43:1
---- v64::f32x2_math_fma::fma stdout ----
thread 'v64::f32x2_math_fma::fma' panicked at 'assertion failed: `(left == right)`
  left: `f32x2(1.0, 1.0)`,
 right: `f32x2(NaN, NaN)`', src\v64.rs:43:1
---- v64::f32x2_math_sin::sin stdout ----
thread 'v64::f32x2_math_sin::sin' panicked at 'assertion failed: `(left == right)`
  left: `f32x2(1.0, 1.0)`,
 right: `f32x2(NaN, 1.0)`', src\v64.rs:43:1
---- v64::f32x2_ops_scalar_arith::ops_scalar_arithmetic stdout ----
thread 'v64::f32x2_ops_scalar_arith::ops_scalar_arithmetic' panicked at 'assertion failed: `(left == right)`
  left: `f32x2(NaN, 0.0)`,
 right: `f32x2(0.0, 0.0)`', src\v64.rs:43:1
---- v64::f32x2_ops_vector_arith::ops_vector_arithmetic stdout ----
thread 'v64::f32x2_ops_vector_arith::ops_vector_arithmetic' panicked at 'assertion failed: `(left == right)`
  left: `f32x2(NaN, 0.0)`,
 right: `f32x2(0.0, 0.0)`', src\v64.rs:43:1
failures:
    v64::f32x2_math_cos::cos
    v64::f32x2_math_fma::fma
    v64::f32x2_math_sin::sin
    v64::f32x2_ops_scalar_arith::ops_scalar_arithmetic
    v64::f32x2_ops_vector_arith::ops_vector_arithmetic
test result: FAILED. 1436 passed; 5 failed; 0 ignored; 0 measured; 0 filtered out

cc @retep007

[gnzlbg]

The following operations of the f32x2 vector type produce incorrect results on gnu windows targets ({i686, x86_64}-pc-windows-gnu) in debug builds: sin, cos, fma, scalar (+,-,*,/, unary -).

This playground example might be used to reproduce this though (I can't know since I don't have access to the platform - it would be good to know if that reproduces the issue or not to narrow down the root cause):

#![feature(
    repr_simd,
    simd_ffi,
    link_llvm_intrinsics,
    platform_intrinsics
)]

#[repr(simd)]
#[derive(Copy, Clone, PartialEq, Debug)]
#[allow(non_camel_case_types)]
struct f32x2(f32, f32);

extern "platform-intrinsic" {
    fn simd_fma<T>(a: T, b: T, c: T) -> T;
}

#[allow(improper_ctypes)]
extern "C" {
    #[link_name = "llvm.cos.v2f32"]
    fn cos_v2f32(x: f32x2) -> f32x2;
    #[link_name = "llvm.sin.v2f32"]
    fn sin_v2f32(x: f32x2) -> f32x2;
}

fn main() {
    let a = f32x2(1.0, 1.0);
    let b = f32x2(0.0, 0.0);
    let c = f32x2(2.0, 2.0);

    unsafe {
        assert_eq!(simd_fma(a, c, b), c);
        assert_eq!(sin_v2f32(b), b);
        assert_eq!(cos_v2f32(b), a);
    }
}

[gnzlbg]

Filled issue upstream: rust-lang/rust#53254

[gnzlbg]

This might be related to: #25 which shows a similar failure on apple darwin.

[gnzlbg]

Maybe @CAD97 can help set this up?

[CAD97]

^{Why me? 😛 Is it just because I jumped ship from windows-msvc to -gnu? Sure, I'm a CI guy but you already have that set up!
Happy to help though!}

The given minimal case passes on the Windows 10 machine I'm writing this from under windows-gnu.

%projects%\untitled>rustup show

active toolchain
----------------

nightly-x86_64-pc-windows-gnu (directory override for '%projects%\untitled')
rustc 1.30.0-nightly (f49f6e73a 2018-09-23)

%projects%\untitled>type src\main.rs
#![feature(
repr_simd,
simd_ffi,
link_llvm_intrinsics,
platform_intrinsics
)]

#[repr(simd)]
#[derive(Copy, Clone, PartialEq, Debug)]
#[allow(non_camel_case_types)]
struct f32x2(f32, f32);

extern "platform-intrinsic" {
    fn simd_fma<T>(a: T, b: T, c: T) -> T;
}

#[allow(improper_ctypes)]
extern "C" {
    #[link_name = "llvm.cos.v2f32"]
    fn cos_v2f32(x: f32x2) -> f32x2;
    #[link_name = "llvm.sin.v2f32"]
    fn sin_v2f32(x: f32x2) -> f32x2;
}

fn main() {
    let a = f32x2(1.0, 1.0);
    let b = f32x2(0.0, 0.0);
    let c = f32x2(2.0, 2.0);

    unsafe {
        assert_eq!(simd_fma(a, c, b), c);
        assert_eq!(sin_v2f32(b), b);
        assert_eq!(cos_v2f32(b), a);
    }

    println!("ok");
}

%projects%\untitled>cargo run
    Finished dev [unoptimized + debuginfo] target(s) in 0.01s
     Running `target\debug\untitled.exe`
ok

Running --tests f32x2 on this repo I get the following failures (and it does indeed pass in --release:

failures:
    v64::f32x2_math_cos::cos
    v64::f32x2_math_exp::exp
    v64::f32x2_math_ln::ln
    v64::f32x2_math_powf::powf
    v64::f32x2_math_sin::sin
    v64::f32x2_ops_scalar_arith::ops_scalar_arithmetic
    v64::f32x2_ops_vector_arith::ops_vector_arithmetic

[gnzlbg]

Thank you for confirming that the repro fails!

[CAD97]

I see that the upstream LLVM patch has landed. Is there documentation of how to build a local version of the compiler against a development LLVM so I could potentially test to see if the patch effected this issue?

[gnzlbg]

@CAD97 see this comment: rust-lang/stdarch#513 (comment)

[mati865]

Rust's llvm already has this fix. I'll see if I can reproduce it later on today.