fixes rust-lang#128857

…lags

Fixes a regression introduced in rust-lang#128252.

As our implementation of MCP411 nears completion and we begin to
solicit testing, it's no longer reasonable to expect testers to
type or remember `BikeshedIntrinsicFrom`. The name degrades the
ease-of-reading of documentation, and the overall experience of
using compiler safe transmute.

Tentatively, we'll instead adopt `TransmuteFrom`.

This name seems to be the one most likely to be stabilized, after
discussion on Zulip [1]. We may want to revisit the ordering of
`Src` and `Dst` before stabilization, at which point we'd likely
consider `TransmuteInto` or `Transmute`.

[1] https://rust-lang.zulipchat.com/#narrow/stream/216762-project-safe-transmute/topic/What.20should.20.60BikeshedIntrinsicFrom.60.20be.20named.3F

This might have been left over when coverage regions were stored in individual
MIR statements, instead of a separate table attached to the MIR body.

LLVM uses the word "code" to refer to a particular kind of coverage mapping.
This unrelated usage of the word is confusing, and makes it harder to introduce
types whose names correspond to the LLVM classification of coverage kinds.

…tions, r=workingjubilee,compiler-errors

add repr to the allowlist for naked functions

Fixes rust-lang#129412 (combining unstable features rust-lang#90957 (`#![feature(naked_functions)]`) and rust-lang#82232 (`#![feature(fn_align)]`)

docs: correct panic conditions for rem_euclid and similar functions

fixes rust-lang#128857

also fixes the documentation for functions behind the `int_roundings` feature (rust-lang#88581)

…thlin

ub_checks intrinsics: fall back to cfg(ub_checks)

Not sure why the fallback body uses `debug_assertions`, probably a leftover from when `cfg!(ub_checks)` did not exist yet?

r? `@saethlin`

…aethlin

const-eval: do not make UbChecks behavior depend on current crate's flags

Fixes rust-lang#129552

Let's see if we can get away with just always enabling these checks.

…saethlin

interpret: do not make const-eval query result depend on tcx.sess

The check against calling functions with missing target features uses `tcx.sess` to determine which target features are available. However, this can differ between different crates in a crate graph, so the same const-eval query can come to different conclusions about whether a constant evaluates successfully or not -- which is bad, we should consistently get the same result everywhere.

…-js-resource-suffix, r=GuillaumeGomez

rustdoc: fix missing resource suffix on `crates.js`

Fixes a regression introduced in rust-lang#128252.

…-errors

Rename `BikeshedIntrinsicFrom` to `TransmuteFrom`

As our implementation of MCP411 nears completion and we begin to solicit testing, it's no longer reasonable to expect testers to type or remember `BikeshedIntrinsicFrom`. The name degrades the ease-of-reading of documentation, and the overall experience of using compiler safe transmute.

Tentatively, we'll instead adopt `TransmuteFrom`.

This name seems to be the one most likely to be stabilized, after discussion on Zulip [1]. We may want to revisit the ordering of `Src` and `Dst` before stabilization, at which point we'd likely consider `TransmuteInto` or `Transmute`.

[1] https://rust-lang.zulipchat.com/#narrow/stream/216762-project-safe-transmute/topic/What.20should.20.60BikeshedIntrinsicFrom.60.20be.20named.3F

Tracking Issue: rust-lang#99571

r​? `@compiler-errors`

…errors

interpret: add missing alignment check in raw_eq

The intrinsic requires alignment, but we forgot to check for that in Miri and const-eval.

…ichaelwoerister

Rustc driver cleanup

This adds a few comments to the driver to clarify a bit what's happening and does some cleanup.

…tolnay

Fix Pin::set bounds regression

Fixes rust-lang#129601

Fixes the regression from rust-lang#129449, where changing the bounds of the impl block containing `Pin::set` changed the method resolution behavior.

```rust
struct A;
impl A {
    fn set(&self) {}
}

let a: Pin<&A>;
a.set();
// before:
// - checks <impl<Ptr: DerefMut> Pin<Ptr>>::set(): `&A` doesn't impl `DerefMut`
// - autorefs -> &A: resolves to A::set()
// now:
// - checks <impl<Ptr: Deref> Pin<Ptr>>::set(): `&A` impls `Deref`! resolves to Pin::set()
// - check method bounds: `&A` doesn't impl DerefMut: error
```

r? `@dtolnay`

…-errors

coverage: Rename `CodeRegion` to `SourceRegion`

LLVM uses the word "code" to refer to a particular kind of coverage mapping. This unrelated usage of the word is confusing, and makes it harder to introduce types whose names correspond to the LLVM classification of coverage kinds.

No functional changes.