Intrinsic functions declared in `std::intrinsics` are an implementation
detail and should not be called directly by the user. The compiler
explicitly warns against their use in user code:
```
warning: the feature `core_intrinsics` is internal to the compiler or standard library
--> src/lib.rs:1:12
|
1 | #![feature(core_intrinsics)]
| ^^^^^^^^^^^^^^^
|
= note: using it is strongly discouraged
= note: `#[warn(internal_features)]` on by default
```
[**Playground link**]
This PR documents what the compiler warning says: these intrinsics should
not be called outside the standard library.
[**Playground link**]: https://play.rust-lang.org/?version=nightly&mode=debug&edition=2024&gist=1c893b0698291f550bbdde0151fd221b
So this is funny, the query `tcx.module_children` was top 3 in most
time consuming functions in Clippy, it was being called 24384 times in
tokio. "Unacceptable!" I thought. Digging a bit around, turns out that
`clippy::strlen_on_c_strings` was calling for `get_def_path` via
`match_libc_symbol`. This query pretty-prints things and performs some
analysis.
Yes, we were running early lint checks to see if symbols were from
`libc`.
I don't really trust callgrind when it says I've turn 81 billion
instructions
into like 10 million. So I benchmarked this the good ol' "compiling 20
times
without incr" method and it went from 0.31s-0.45s to 0.25s
constistently.
(Profiled, and "benchmarked") on tokio.
What I can get behind is via `strlen_on_c_strings` changing from 31
million instructions into 76k. 🎉🥳
changelog: [`strlen_on_c_strings`]: Optimize it by 99.75%
Rollup of 6 pull requests
Successful merges:
- rust-lang/rust#138016 (Added `Clone` implementation for `ChunkBy`)
- rust-lang/rust#141162 (refactor `AttributeGate` and `rustc_attr!` to emit notes during feature checking)
- rust-lang/rust#141474 (Add `ParseMode::Diagnostic` and fix multiline spans in diagnostic attribute lints)
- rust-lang/rust#141947 (Specify that "option-like" enums must be `#[repr(Rust)]` to be ABI-compatible with their non-1ZST field.)
- rust-lang/rust#142252 (Improve clarity of `core::sync::atomic` docs about "Considerations" in regards to CAS operations)
- rust-lang/rust#142337 (miri: add flag to suppress float non-determinism)
r? `@ghost`
`@rustbot` modify labels: rollup
Upgrade `object`, `addr2line`, and `unwinding` in the standard library
Object:
0.37.0 is a semver-breaking release but the only breakage is in `elf::R_RISCV_GNU_*` and `pe::IMAGE_WEAK_EXTERN_*` constants, as well as Mach-O dyld. This API is not used by `std`, so we should be fine to upgrade.
This new version also includes functionality for parsing Wasm object files that we may eventually like to make use of.
Changelog: https://github.com/gimli-rs/object/blob/master/CHANGELOG.md#0370
Addr2line:
0.25.0 is a breaking change only because it upgrades the `gimli` version. It also includes a change to the `compiler-builtins` dependency that helps with [1].
Changelog: https://github.com/gimli-rs/addr2line/blob/master/CHANGELOG.md#0250-20250611
[1]: https://github.com/rust-lang/rust/issues/142265
Update dependencies in `library/Cargo.lock`
This removes the `compiler_builtins` dependency from a handful of library dependencies, which is progress toward [1].
[1]: https://github.com/rust-lang/rust/issues/142265
assert more in release in `rustc_ast_lowering`
My understanding of the compiler's architecture is that in the `ast_lowering` crate, we are constructing the HIR as a one-time thing per crate. This is after tokenizing, parsing, resolution, expansion, possible reparsing, reresolution, reexpansion, and so on. In other words, there are many reasons that perf-focused PRs spend a lot of time touching `rustc_parse`, `rustc_expand`, `rustc_ast`, and then `rustc_hir` and "onwards", but `ast_lowering` is a little bit of an odd duck.
In this crate, we have a number of debug assertions. Some are clearly expensive checks that seem like they are prohibitive to run in actual optimized compiler builds, but then there are a number that are simple asserts on integer equalities, `is_empty`, or the like. I believe we should do some of them even in release builds, because the correctness gain is worth the performance cost: almost zero.
tests: Split dont-shuffle-bswaps along opt-levels and arches
This duplicates dont-shuffle-bswaps in order to make each opt level its own test. Then -opt3.rs gets split into a revision per arch we want to test, with certain architectures gaining new target-cpu minimums.
add `extern "custom"` functions
tracking issue: rust-lang/rust#140829
previous discussion: https://github.com/rust-lang/rust/issues/140566
In short, an `extern "custom"` function is a function with a custom ABI, that rust does not know about. Therefore, such functions can only be defined with `#[unsafe(naked)]` and `naked_asm!`, or via an `extern "C" { /* ... */ }` block. These functions cannot be called using normal rust syntax: calling them can only be done from inline assembly.
The motivation is low-level scenarios where a custom calling convention is used. Currently, we often pick `extern "C"`, but that is a lie because the function does not actually respect the C calling convention.
At the moment `"custom"` seems to be the name with the most support. That name is not final, but we need to pick something to actually implement this.
r? `@traviscross`
cc `@tgross35`
try-job: x86_64-apple-2
add `extern "custom"` functions
tracking issue: rust-lang/rust#140829
previous discussion: https://github.com/rust-lang/rust/issues/140566
In short, an `extern "custom"` function is a function with a custom ABI, that rust does not know about. Therefore, such functions can only be defined with `#[unsafe(naked)]` and `naked_asm!`, or via an `extern "C" { /* ... */ }` block. These functions cannot be called using normal rust syntax: calling them can only be done from inline assembly.
The motivation is low-level scenarios where a custom calling convention is used. Currently, we often pick `extern "C"`, but that is a lie because the function does not actually respect the C calling convention.
At the moment `"custom"` seems to be the name with the most support. That name is not final, but we need to pick something to actually implement this.
r? `@traviscross`
cc `@tgross35`
try-job: x86_64-apple-2
