Pass alignments through the shim as `Alignment` (not `usize`)
We're using `Layout` on both sides, so might as well skip the transmutes back and forth to `usize`.
The mir-opt test shows that doing so allows simplifying the boxed-slice drop slightly, for example.
tests: adapt align-offset.rs for InstCombine improvements in LLVM 23
Upstream [has improved InstCombine](8d2078332c) so that it can shrink added constants using known zeroes, which caused a little bit of change in this test. As far as I can tell either output is fine, so we just accept both.
@rustbot label: +llvm-main
UnsafePinned: implement opsem effects of UnsafeUnpin
This implements the next step for https://github.com/rust-lang/rust/issues/125735: actually making `UnsafePinned` have special opsem effects by suppressing the `noalias` *even if* the type is wrapped in an `Unpin` wrapper.
For backwards compatibility we also still keep the `Unpin` hack, i.e. a type must be both `Unpin` and `UnsafeUnpin` to get `noalias`.
Optimize indexing slices and strs with inclusive ranges
Instead of separately checking for `end == usize::MAX` and `end + 1 > slice.len()`, we can check for `end >= slice.len()`. Also consolidate all the str indexing related panic functions into a single function which reports the correct error depending on the arguments, as the slice indexing code already does.
The downside of all this is that the panic message is slightly less specific when trying to index with `[..=usize::MAX]`: instead of saying "attempted to index str up to maximum usize" it just says "end byte index {end} out of bounds". But this is a rare enough case that I think it is acceptable
Make operational semantics of pattern matching independent of crate and module
The question of "when does matching an enum against a pattern of one of its variants read its discriminant" is currently an underspecified part of the language, causing weird behavior around borrowck, drop order, and UB.
Of course, in the common cases, the discriminant must be read to distinguish the variant of the enum, but currently the following exceptions are implemented:
1. If the enum has only one variant, we currently skip the discriminant read.
- This has the advantage that single-variant enums behave the same way as structs in this regard.
- However, it means that if the discriminant exists in the layout, we can't say that this discriminant being invalid is UB. This makes me particularly uneasy in its interactions with niches – consider the following example ([playground](https://play.rust-lang.org/?version=stable&mode=debug&edition=2024&gist=5904a6155cbdd39af4a2e7b1d32a9b1a)), where miri currently doesn't detect any UB (because the semantics don't specify any):
<details><summary>Example 1</summary>
```rust
#![allow(dead_code)]
use core::mem::{size_of, transmute};
#[repr(u8)]
enum Inner {
X(u8),
}
enum Outer {
A(Inner),
B(u8),
}
fn f(x: &Inner) {
match x {
Inner::X(v) => {
println!("{v}");
}
}
}
fn main() {
assert_eq!(size_of::<Inner>(), 2);
assert_eq!(size_of::<Outer>(), 2);
let x = Outer::B(42);
let y = &x;
f(unsafe { transmute(y) });
}
```
</details>
2. For the purpose of the above, enums with marked with `#[non_exhaustive]` are always considered to have multiple variants when observed from foreign crates, but the actual number of variants is considered in the current crate.
- This means that whether code has UB can depend on which crate it is in: https://github.com/rust-lang/rust/issues/147722
- In another case of `#[non_exhaustive]` affecting the runtime semantics, its presence or absence can change what gets captured by a closure, and by extension, the drop order: https://github.com/rust-lang/rust/issues/147722#issuecomment-3674554872
- Also at the above link, there is an example where removing `#[non_exhaustive]` can cause borrowck to suddenly start failing in another crate.
3. Moreover, we currently make a more specific check: we only read the discriminant if there is more than one *inhabited* variant in the enum.
- This means that the semantics can differ between `foo<!>`, and a copy of `foo` where `T` was manually replaced with `!`: rust-lang/rust#146803
- Moreover, due to the privacy rules for inhabitedness, it means that the semantics of code can depend on the *module* in which it is located.
- Additionally, this inhabitedness rule is even uglier due to the fact that closure capture analysis needs to happen before we can determine whether types are uninhabited, which means that whether the discriminant read happens has a different answer specifically for capture analysis.
- For the two above points, see the following example ([playground](https://play.rust-lang.org/?version=nightly&mode=debug&edition=2024&gist=a07d8a3ec0b31953942e96e2130476d9)):
<details><summary>Example 2</summary>
```rust
#![allow(unused)]
mod foo {
enum Never {}
struct PrivatelyUninhabited(Never);
pub enum A {
V(String, String),
Y(PrivatelyUninhabited),
}
fn works(mut x: A) {
let a = match x {
A::V(ref mut a, _) => a,
_ => unreachable!(),
};
let b = match x {
A::V(_, ref mut b) => b,
_ => unreachable!(),
};
a.len(); b.len();
}
fn fails(mut x: A) {
let mut f = || match x {
A::V(ref mut a, _) => (),
_ => unreachable!(),
};
let mut g = || match x {
A::V(_, ref mut b) => (),
_ => unreachable!(),
};
f(); g();
}
}
use foo::A;
fn fails(mut x: A) {
let a = match x {
A::V(ref mut a, _) => a,
_ => unreachable!(),
};
let b = match x {
A::V(_, ref mut b) => b,
_ => unreachable!(),
};
a.len(); b.len();
}
fn fails2(mut x: A) {
let mut f = || match x {
A::V(ref mut a, _) => (),
_ => unreachable!(),
};
let mut g = || match x {
A::V(_, ref mut b) => (),
_ => unreachable!(),
};
f(); g();
}
```
</details>
In light of the above, and following the discussion at rust-lang/rust#138961 and rust-lang/rust#147722, this PR ~~makes it so that, operationally, matching on an enum *always* reads its discriminant.~~ introduces the following changes to this behavior:
- matching on a `#[non_exhaustive]` enum will always introduce a discriminant read, regardless of whether the enum is from an external crate
- uninhabited variants now count just like normal ones, and don't get skipped in the checks
As per the discussion below, the resolution for point (1) above is that it should land as part of a separate PR, so that the subtler decision can be more carefully considered.
Note that this is a breaking change, due to the aforementioned changes in borrow checking behavior, new UB (or at least UB newly detected by miri), as well as drop order around closure captures. However, it seems to me that the combination of this PR with rust-lang/rust#138961 should have smaller real-world impact than rust-lang/rust#138961 by itself.
Fixesrust-lang/rust#142394Fixesrust-lang/rust#146590Fixesrust-lang/rust#146803 (though already marked as duplicate)
Fixes parts of rust-lang/rust#147722Fixesrust-lang/miri#4778
r? @Nadrieril @RalfJung
@rustbot label +A-closures +A-patterns +T-opsem +T-lang
We're using `Layout` on both sides, so might as well skip the transmutes back and forth to `usize`.
The mir-opt test shows that doing so allows simplifying the boxed-slice drop slightly, for example.
llvm will look at both
1. the values of "target-features" and
2. the function string attributes.
this removes the redundant function string attribute because it is not needed at all.
rustc sets the `+backchain` attribute through `target_features_attr(...)`
Replace `self.end() == usize::MAX` and `self.end() + 1 > slice.len()`
with `self.end() >= slice.len()`. Same reasoning as previous commit.
Also consolidate the str panicking functions into function.
The checks for `self.end() == usize::MAX` and `self.end() + 1 > slice.len()`
can be replaced with `self.end() >= slice.len()`, since
`self.end() < slice.len()` implies both
`self.end() <= slice.len()` and
`self.end() < usize::MAX`.
Add a `codegen-llvm` test to check the number of `icmp` instrucitons
generated for each `SliceIndex` method on the various range types. This
will be updated in the next commit when `SliceIndex::get` is optimized
for `RangeInclusive`.
Upstream has improved InstCombine so that it can shrink added constants
using known zeroes, which caused a little bit of change in this test. As
far as I can tell either output is fine, so we just accept both.
Cleanup offload datatransfer
There are 3 steps to run code on a GPU: Copy data from the host to the device, launch the kernel, and move it back.
At the moment, we have a single variable describing the memory handling to do in each step, but that makes it hard for LLVM's opt pass to understand what's going on. We therefore split it into three variables, each only including the bits relevant for the corresponding stage.
cc @jdoerfert @kevinsala
r? compiler
Stabilize `core::hint::cold_path`
`cold_path` has been around unstably for a while and is a rather useful tool to have. It does what it is supposed to and there are no known remaining issues, so stabilize it here (including const).
Newly stable API:
```rust
// in core::hint
pub const fn cold_path();
```
I have opted to exclude `likely` and `unlikely` for now since they have had some concerns about ease of use that `cold_path` doesn't suffer from. `cold_path` is also significantly more flexible; in addition to working with boolean `if` conditions, it can be used in `match` arms, `if let`, closures, and other control flow blocks. `likely` and `unlikely` are also possible to implement in user code via `cold_path`, if desired.
Closes: https://github.com/rust-lang/rust/issues/136873 (tracking issue)
---
There has been some design and implementation work for making `#[cold]` function in more places, such as `if` arms, `match` arms, and closure bodies. Considering a stable `cold_path` will cover all of these usecases, it does not seem worth pursuing a more powerful `#[cold]` as an alternative way to do the same thing. If the lang team agrees, then:
Closes: https://github.com/rust-lang/rust/issues/26179
Closes: https://github.com/rust-lang/rust/pull/120193
Remove dummy loads on offload codegen
The current logic generates two dummy loads to prevent some globals from being optimized away. This blocks memtransfer loop hoisting optimizations, so it's time to remove them.
r? @ZuseZ4
skip codegen for intrinsics with big fallback bodies if backend does not need them
This hopefully fixes the perf regression from https://github.com/rust-lang/rust/pull/148478. I only added the intrinsics with big fallback bodies to the list; it doesn't seem worth the effort of going through the entire list.
Fixes https://github.com/rust-lang/rust/issues/149945
Cc @scottmcm @bjorn3
Fix autodiff codegen tests
Preparing autodiff for release on nightly. Since we haven't been running these tests in CI, they regressed over the last months. These changes fixes this and hopefully make the tests more robust for the future.
r? compiler
Add codegen test for SLP vectorization
close: rust-lang/rust#142519
This PR adds a codegen regression test for rust-lang/rust#142519. A regression in LLVM to fail to auto-vectorize, leading to significant performance loss.
The SLP vectorizer correctly groups the 4-byte operations into <4 x i8> vectors.
The loop state is maintained in SIMD registers (phi <4 x i8>).
The test remains robust across architectures (AArch64 vs x86_64) by allowing flexible store types (i32 or <4 x i8>).
The IR is a bit different (in particular wrt naming) if
debug-assertions-std is enabled. Peculiarly, the issue goes away
if overflow-check-std is also enabled, which is why CI did not
catch this.
offload: move (un)register lib into global_ctors
Right now we initialize the openmp/offload runtime before every single offload call, and tear it down directly afterwards.
What we should rather do is initialize it once in the binary startup code, and tear it down at the end of the binary execution. Here I implement these changes.
Together, our generated IR has a lot less usage of globals, which in turn simplifies the refactoring in https://github.com/rust-lang/rust/pull/150683, where I introduce a new variant of our offload intrinsic.
r? oli-obk
slice/ascii: Optimize `eq_ignore_ascii_case` with auto-vectorization
- Refactor the current functionality into a helper function
- Use `as_chunks` to encourage auto-vectorization in the optimized chunk processing function
- Add a codegen test checking for vectorization and no panicking
- Add benches for `eq_ignore_ascii_case`
---
The optimized function is initially only enabled for x86_64 which has `sse2` as part of its baseline, but none of the code is platform specific. Other platforms with SIMD instructions may also benefit from this implementation.
Performance improvements only manifest for slices of 16 bytes or longer, so the optimized path is gated behind a length check for greater than or equal to 16.
Benchmarks - Cases below 16 bytes are unaffected, cases above all show sizeable improvements.
```
before:
str::eq_ignore_ascii_case::bench_large_str_eq 4942.30ns/iter +/- 48.20
str::eq_ignore_ascii_case::bench_medium_str_eq 632.01ns/iter +/- 16.87
str::eq_ignore_ascii_case::bench_str_17_bytes_eq 16.28ns/iter +/- 0.45
str::eq_ignore_ascii_case::bench_str_31_bytes_eq 35.23ns/iter +/- 2.28
str::eq_ignore_ascii_case::bench_str_of_8_bytes_eq 7.56ns/iter +/- 0.22
str::eq_ignore_ascii_case::bench_str_under_8_bytes_eq 2.64ns/iter +/- 0.06
after:
str::eq_ignore_ascii_case::bench_large_str_eq 611.63ns/iter +/- 28.29
str::eq_ignore_ascii_case::bench_medium_str_eq 77.10ns/iter +/- 19.76
str::eq_ignore_ascii_case::bench_str_17_bytes_eq 3.49ns/iter +/- 0.39
str::eq_ignore_ascii_case::bench_str_31_bytes_eq 3.50ns/iter +/- 0.27
str::eq_ignore_ascii_case::bench_str_of_8_bytes_eq 7.27ns/iter +/- 0.09
str::eq_ignore_ascii_case::bench_str_under_8_bytes_eq 2.60ns/iter +/- 0.05
```
`cold_path` has been around unstably for a while and is a rather useful
tool to have. It does what it is supposed to and there are no known
remaining issues, so stabilize it here (including const).
Newly stable API:
// in core::hint
pub const fn cold_path();
I have opted to exclude `likely` and `unlikely` for now since they have
had some concerns about ease of use that `cold_path` doesn't suffer
from. `cold_path` is also significantly more flexible; in addition to
working with boolean `if` conditions, it can be used in `match` arms,
`if let`, closures, and other control flow blocks. `likely` and
`unlikely` are also possible to implement in user code via `cold_path`,
if desired.
The existing `aarch64-unknown-none` target assumes Armv8.0-A as a baseline. However, Arm recently released the Arm Cortex-R82 processor which is the first to implement the Armv8-R AArch64 mode architecture. This architecture is similar to Armv8-A AArch64, however it has a different set of mandatory features, and is based off of Armv8.4. It is largely unrelated to the existing Armv8-R architecture target (`armv8r-none-eabihf`), which only operates in AArch32 mode.
The second `aarch64v8r-unknown-none-softfloat` target allows for possible Armv8-R AArch64 CPUs with no FPU, or for use-cases where FPU register stacking is not desired. As with the existing `aarch64-unknown-none` target we have coupled FPU support and Neon support together - there is no 'has FPU but does not have NEON' target proposed even though the architecture technically allows for it.
This PR was developed by Ferrous Systems on behalf of Arm. Arm is the owner of these changes.
LoongArch: Fix direct-access-external-data test
On LoongArch targets, `-Cdirect-access-external-data` defaults to `no`. Since copy relocations are not supported, `dso_local` is not emitted under `-Crelocation-model=static`, unlike on other targets.
add CSE optimization tests for iterating over slice
This PR is regression test for issue rust-lang/rust#119573.
This PR introduces a new regression test to verify a critical optimization known as Common Subexpression Elimination (CSE) is correctly applied during various slice iteration patterns.
abi: add a rust-preserve-none calling convention
This is the conceptual opposite of the rust-cold calling convention and is particularly useful in combination with the new `explicit_tail_calls` feature.
For relatively tight loops implemented with tail calling (`become`) each of the function with the regular calling convention is still responsible for restoring the initial value of the preserved registers. So it is not unusual to end up with a situation where each step in the tail call loop is spilling and reloading registers, along the lines of:
foo:
push r12
; do things
pop r12
jmp next_step
This adds up quickly, especially when most of the clobberable registers are already used to pass arguments or other uses.
I was thinking of making the name of this ABI a little less LLVM-derived and more like a conceptual inverse of `rust-cold`, but could not come with a great name (`rust-cold` is itself not a great name: cold in what context? from which perspective? is it supposed to mean that the function is rarely called?)
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:
- https://github.com/rust-lang/rust/issues/60637
- https://github.com/rust-lang/rust/issues/137407
- https://github.com/rust-lang/rust/issues/122623
- https://github.com/rust-lang/rust/issues/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.
