…bilee

add `simd_splat` intrinsic

Add `simd_splat` which lowers to the LLVM canonical splat sequence.

```llvm
insertelement <N x elem> poison, elem %x, i32 0
shufflevector <N x elem> v0, <N x elem> poison, <N x i32> zeroinitializer
```

Right now we try to fake it using one of

```rust
fn splat(x: u32) -> u32x8 {
    u32x8::from_array([x; 8])
}
```

or (in `stdarch`)

```rust
fn splat(value: $elem_type) -> $name {
    #[derive(Copy, Clone)]
    #[repr(simd)]
    struct JustOne([$elem_type; 1]);
    let one = JustOne([value]);
    // SAFETY: 0 is always in-bounds because we're shuffling
    // a simd type with exactly one element.
    unsafe { simd_shuffle!(one, one, [0; $len]) }
}
```

Both of these can confuse the LLVM optimizer, producing sub-par code. Some examples:

- rust-lang#60637
- rust-lang#137407
- rust-lang#122623
- rust-lang#97804

---

As far as I can tell there is no way to provide a fallback implementation for this intrinsic, because there is no `const` way of evaluating the number of elements (there might be issues beyond that, too). So, I added implementations for all 4 backends.

Both GCC and const-eval appear to have some issues with simd vectors containing pointers. I have a workaround for GCC, but haven't yet been able to make const-eval work. See the comments below.

Currently this just adds the intrinsic, it does not actually use it anywhere yet.

…bilee

add `simd_splat` intrinsic

Add `simd_splat` which lowers to the LLVM canonical splat sequence.

```llvm
insertelement <N x elem> poison, elem %x, i32 0
shufflevector <N x elem> v0, <N x elem> poison, <N x i32> zeroinitializer
```

Right now we try to fake it using one of

```rust
fn splat(x: u32) -> u32x8 {
    u32x8::from_array([x; 8])
}
```

or (in `stdarch`)

```rust
fn splat(value: $elem_type) -> $name {
    #[derive(Copy, Clone)]
    #[repr(simd)]
    struct JustOne([$elem_type; 1]);
    let one = JustOne([value]);
    // SAFETY: 0 is always in-bounds because we're shuffling
    // a simd type with exactly one element.
    unsafe { simd_shuffle!(one, one, [0; $len]) }
}
```

Both of these can confuse the LLVM optimizer, producing sub-par code. Some examples:

- rust-lang#60637
- rust-lang#137407
- rust-lang#122623
- rust-lang#97804

---

As far as I can tell there is no way to provide a fallback implementation for this intrinsic, because there is no `const` way of evaluating the number of elements (there might be issues beyond that, too). So, I added implementations for all 4 backends.

Both GCC and const-eval appear to have some issues with simd vectors containing pointers. I have a workaround for GCC, but haven't yet been able to make const-eval work. See the comments below.

Currently this just adds the intrinsic, it does not actually use it anywhere yet.

…uwer

Rollup of 3 pull requests

Successful merges:

 - #151346 (add `simd_splat` intrinsic)
 - #151353 (compiletest: Make `aux-crate` directive explicitly handle `--extern` modifiers)
 - #151571 (Fix cstring-merging test for Hexagon target)

r? @ghost

…bilee

add `simd_splat` intrinsic

Add `simd_splat` which lowers to the LLVM canonical splat sequence.

```llvm
insertelement <N x elem> poison, elem %x, i32 0
shufflevector <N x elem> v0, <N x elem> poison, <N x i32> zeroinitializer
```

Right now we try to fake it using one of

```rust
fn splat(x: u32) -> u32x8 {
    u32x8::from_array([x; 8])
}
```

or (in `stdarch`)

```rust
fn splat(value: $elem_type) -> $name {
    #[derive(Copy, Clone)]
    #[repr(simd)]
    struct JustOne([$elem_type; 1]);
    let one = JustOne([value]);
    // SAFETY: 0 is always in-bounds because we're shuffling
    // a simd type with exactly one element.
    unsafe { simd_shuffle!(one, one, [0; $len]) }
}
```

Both of these can confuse the LLVM optimizer, producing sub-par code. Some examples:

- rust-lang#60637
- rust-lang#137407
- rust-lang#122623
- rust-lang#97804

---

As far as I can tell there is no way to provide a fallback implementation for this intrinsic, because there is no `const` way of evaluating the number of elements (there might be issues beyond that, too). So, I added implementations for all 4 backends.

Both GCC and const-eval appear to have some issues with simd vectors containing pointers. I have a workaround for GCC, but haven't yet been able to make const-eval work. See the comments below.

Currently this just adds the intrinsic, it does not actually use it anywhere yet.

…bilee

add `simd_splat` intrinsic

Add `simd_splat` which lowers to the LLVM canonical splat sequence.

```llvm
insertelement <N x elem> poison, elem %x, i32 0
shufflevector <N x elem> v0, <N x elem> poison, <N x i32> zeroinitializer
```

Right now we try to fake it using one of

```rust
fn splat(x: u32) -> u32x8 {
    u32x8::from_array([x; 8])
}
```

or (in `stdarch`)

```rust
fn splat(value: $elem_type) -> $name {
    #[derive(Copy, Clone)]
    #[repr(simd)]
    struct JustOne([$elem_type; 1]);
    let one = JustOne([value]);
    // SAFETY: 0 is always in-bounds because we're shuffling
    // a simd type with exactly one element.
    unsafe { simd_shuffle!(one, one, [0; $len]) }
}
```

Both of these can confuse the LLVM optimizer, producing sub-par code. Some examples:

- rust-lang#60637
- rust-lang#137407
- rust-lang#122623
- rust-lang#97804

---

As far as I can tell there is no way to provide a fallback implementation for this intrinsic, because there is no `const` way of evaluating the number of elements (there might be issues beyond that, too). So, I added implementations for all 4 backends.

Both GCC and const-eval appear to have some issues with simd vectors containing pointers. I have a workaround for GCC, but haven't yet been able to make const-eval work. See the comments below.

Currently this just adds the intrinsic, it does not actually use it anywhere yet.

…uwer

Rollup of 3 pull requests

Successful merges:

 - #151346 (add `simd_splat` intrinsic)
 - #151353 (compiletest: Make `aux-crate` directive explicitly handle `--extern` modifiers)
 - #151571 (Fix cstring-merging test for Hexagon target)

r? @ghost

Rollup of 11 pull requests

Successful merges:

 - #149962 (Promote powerpc64-unknown-linux-musl to tier 2 with host tools)
 - #150138 (Add new Tier 3 targets for ARMv6)
 - #150905 (Fix(lib/win/net): Remove hostname support under Win7)
 - #151094 (remote-test-server: Fix compilation on UEFI targets)
 - #151346 (add `simd_splat` intrinsic)
 - #151353 (compiletest: Make `aux-crate` directive explicitly handle `--extern` modifiers)
 - #151538 (std: `sleep_until` on Motor and VEX)
 - #151098 (Add Korean translation to Rust By Example)
 - #151157 (Extend build-manifest local test guide)
 - #151403 (std: use 64-bit `clock_nanosleep` on GNU/Linux if available)
 - #151571 (Fix cstring-merging test for Hexagon target)

Rollup merge of #151346 - folkertdev:simd-splat, r=workingjubilee

add `simd_splat` intrinsic

Add `simd_splat` which lowers to the LLVM canonical splat sequence.

```llvm
insertelement <N x elem> poison, elem %x, i32 0
shufflevector <N x elem> v0, <N x elem> poison, <N x i32> zeroinitializer
```

Right now we try to fake it using one of

```rust
fn splat(x: u32) -> u32x8 {
    u32x8::from_array([x; 8])
}
```

or (in `stdarch`)

```rust
fn splat(value: $elem_type) -> $name {
    #[derive(Copy, Clone)]
    #[repr(simd)]
    struct JustOne([$elem_type; 1]);
    let one = JustOne([value]);
    // SAFETY: 0 is always in-bounds because we're shuffling
    // a simd type with exactly one element.
    unsafe { simd_shuffle!(one, one, [0; $len]) }
}
```

Both of these can confuse the LLVM optimizer, producing sub-par code. Some examples:

- #60637
- #137407
- #122623
- #97804

---

As far as I can tell there is no way to provide a fallback implementation for this intrinsic, because there is no `const` way of evaluating the number of elements (there might be issues beyond that, too). So, I added implementations for all 4 backends.

Both GCC and const-eval appear to have some issues with simd vectors containing pointers. I have a workaround for GCC, but haven't yet been able to make const-eval work. See the comments below.

Currently this just adds the intrinsic, it does not actually use it anywhere yet.

Rollup of 11 pull requests

Successful merges:

 - rust-lang/rust#149962 (Promote powerpc64-unknown-linux-musl to tier 2 with host tools)
 - rust-lang/rust#150138 (Add new Tier 3 targets for ARMv6)
 - rust-lang/rust#150905 (Fix(lib/win/net): Remove hostname support under Win7)
 - rust-lang/rust#151094 (remote-test-server: Fix compilation on UEFI targets)
 - rust-lang/rust#151346 (add `simd_splat` intrinsic)
 - rust-lang/rust#151353 (compiletest: Make `aux-crate` directive explicitly handle `--extern` modifiers)
 - rust-lang/rust#151538 (std: `sleep_until` on Motor and VEX)
 - rust-lang/rust#151098 (Add Korean translation to Rust By Example)
 - rust-lang/rust#151157 (Extend build-manifest local test guide)
 - rust-lang/rust#151403 (std: use 64-bit `clock_nanosleep` on GNU/Linux if available)
 - rust-lang/rust#151571 (Fix cstring-merging test for Hexagon target)

Rollup of 11 pull requests

Successful merges:

 - rust-lang/rust#149962 (Promote powerpc64-unknown-linux-musl to tier 2 with host tools)
 - rust-lang/rust#150138 (Add new Tier 3 targets for ARMv6)
 - rust-lang/rust#150905 (Fix(lib/win/net): Remove hostname support under Win7)
 - rust-lang/rust#151094 (remote-test-server: Fix compilation on UEFI targets)
 - rust-lang/rust#151346 (add `simd_splat` intrinsic)
 - rust-lang/rust#151353 (compiletest: Make `aux-crate` directive explicitly handle `--extern` modifiers)
 - rust-lang/rust#151538 (std: `sleep_until` on Motor and VEX)
 - rust-lang/rust#151098 (Add Korean translation to Rust By Example)
 - rust-lang/rust#151157 (Extend build-manifest local test guide)
 - rust-lang/rust#151403 (std: use 64-bit `clock_nanosleep` on GNU/Linux if available)
 - rust-lang/rust#151571 (Fix cstring-merging test for Hexagon target)

Rollup of 11 pull requests

Successful merges:

 - rust-lang/rust#149962 (Promote powerpc64-unknown-linux-musl to tier 2 with host tools)
 - rust-lang/rust#150138 (Add new Tier 3 targets for ARMv6)
 - rust-lang/rust#150905 (Fix(lib/win/net): Remove hostname support under Win7)
 - rust-lang/rust#151094 (remote-test-server: Fix compilation on UEFI targets)
 - rust-lang/rust#151346 (add `simd_splat` intrinsic)
 - rust-lang/rust#151353 (compiletest: Make `aux-crate` directive explicitly handle `--extern` modifiers)
 - rust-lang/rust#151538 (std: `sleep_until` on Motor and VEX)
 - rust-lang/rust#151098 (Add Korean translation to Rust By Example)
 - rust-lang/rust#151157 (Extend build-manifest local test guide)
 - rust-lang/rust#151403 (std: use 64-bit `clock_nanosleep` on GNU/Linux if available)
 - rust-lang/rust#151571 (Fix cstring-merging test for Hexagon target)