Turn ProjectionElem::Subtype into CastKind::Subtype
I noticed that drop elaboration can't, in general, handle `ProjectionElem::SubType`. It creates a disjoint move path that overlaps with other move paths. (`Subslice` does too, and I'm working on a different PR to make that special case less fragile.) If its skipped and treated as the same move path as its parent then `MovePath.place` has multiple possible projections. (It would probably make sense to remove all `Subtype` projections for the canonical place but it doesn't make sense to have this special case for a problem that doesn't actually occur in real MIR.)
The only reason this doesn't break is that `Subtype` is always the sole projection of the local its applied to. For the same reason, it works fine as a `CastKind` so I figured that makes more sense than documenting and validating this hidden invariant.
cc rust-lang/rust#112651, rust-lang/rust#133258
r? Icnr (bc you've been the main person dealing with `Subtype` it looks like)
Rollup of 8 pull requests
Successful merges:
- rust-lang/rust#146593 (Allow specifying multiple bounds for same associated item, except in trait objects)
- rust-lang/rust#147177 ([DebugInfo] Fix MSVC tuple child creation)
- rust-lang/rust#147195 (iter repeat: add tests for new count and last behavior)
- rust-lang/rust#147202 (Swap order of `resolve_coroutine_interiors` and `handle_opaque_type_uses`)
- rust-lang/rust#147204 (Refactor ArrayWindows to use a slice)
- rust-lang/rust#147219 (Add proper error handling for closure in impl)
- rust-lang/rust#147226 (include `outer_inclusive_binder` of pattern types)
- rust-lang/rust#147230 (Fix typo in 'unfulfilled_lint_expectation' to plural)
r? `@ghost`
`@rustbot` modify labels: rollup
[DebugInfo] Fix MSVC tuple child creation
This is a fix for the debugger visualizer scripts
For whatever reason, using `CreateChildAtOffset` on the child element sometimes caused issues with pointers (and maybe some other types). The resulting child's memory would be a block 4 bytes too far forward. Creating the child off of the parent `valobj` and using the type definition to get the correct offset seems to fix that.
Before:
<img width="489" height="136" alt="image" src="https://github.com/user-attachments/assets/fb4cb95c-f199-49a6-8eba-6d3ff486b69a" />
After:
<img width="518" height="145" alt="image" src="https://github.com/user-attachments/assets/3f50dbc3-19ca-4fd8-87c5-b4be295f6e7c" />
This shouldn't affect any tests as we don't run debuginfo tests for MSVC afaik
Don't create a top-level `true` directory when running UI tests
The funny thing about writing `-Cincremental=true` is that it *does* enable incremental compilation ... using an incremental compilation dir of `./true`.
And for UI tests, that ends up creating a `true` directory in the repository root, which is annoying.
Fortunately, compiletest has an existing `//@ incremental` directive that takes care of creating an empty incremental directory, and passing it to `-Cincremental`.
---
I have manually checked that reverting rust-lang/rust#146649 still causes the updated test to fail.
Remove `no-remap-src-base` from tests
Previously in the `//`-compiletest-directive times, `no-remap-src-base` was implemented as a special `no-*` directive parsing. In the migration from `//` -> `//`@`,` the `// no-remap-src-base` directive was lost, most likely because it had no effect -- the default is not remapping `src-base`.
So remove occurrences of `no-remap-src-base`, as these are not valid directives.
r? `@Zalathar` (since we discussed this on discord, or compiler)
Fix autodiff empty ret regression
closes https://github.com/rust-lang/rust/issues/147144
The two gsoc summer projects caused a bit of churn, which was to be expected, especially since we don't run autodiff in CI yet.
This adds a void return testcase that we should have had anyway, and fixes the regression.
r? `@Zalathar` (Just guessing since I've seen you in a few LLVM PRs and Oli is probably still busy. Feel free to reroll!)
Replace `rustc_span::Span` with a stripped down version for librustdoc's highlighter
While profiling rustdoc's syntax highlighter, I noticed a lot of time being spent in the `Span` interner, due to the highlighter creating a lot of (new) spans.
Since the only data from the `Span` that we use is the `hi` and `lo` byte positions - I replaced the regular `Span` with a simple one with two fields, and in my benchmarks it seemed to make a big dent in the highlighter's perf, so thought I would see what the perf runner says.
compiletest: Pass around `DirectiveLine` instead of bare strings
This is an incremental step towards being able to clean up and centralize compiletest directive parsing.
My original plan was to add features to `DirectiveLine`, and then gradually migrate parsing code to use those features. However, that turned out to be impractical, because of how the existing directive parsers call each other. So instead this PR focuses on getting them to all take `DirectiveLine` instead of bare strings, to enable incremental work in the future.
Because this is part of an ongoing cleanup, I've prioritised clean diffs over nice code, because much of this code is going to be modified again when `DirectiveLine` is more capable.
r? jieyouxu
Previously in the `//`-compiletest-directive times, this was implemented
as a special `no-*` directive parsing. In the migration from `//` ->
`//@`, the `// no-remap-src-base` directive was lost, most likely
because it had no effect -- the default is not remapping `src-base`.
So remove occurrences of `no-remap-src-base`, as these are not valid
directives.
Rollup of 6 pull requests
Successful merges:
- rust-lang/rust#143069 (Add fast-path for accessing the current thread id)
- rust-lang/rust#146518 (Improve the documentation around `ZERO_AR_DATE`)
- rust-lang/rust#146596 (Add a dummy codegen backend)
- rust-lang/rust#146617 (Don’t suggest foreign `doc(hidden)` types in "the following other types implement trait" diagnostics)
- rust-lang/rust#146635 (cg_llvm: Stop using `as_c_char_ptr` for coverage-related bindings)
- rust-lang/rust#147184 (Fix the bevy implied bounds hack for the next solver)
r? `@ghost`
`@rustbot` modify labels: rollup
Fix the bevy implied bounds hack for the next solver
The diff is trivial, of course, and basically what you already suggested. Mostly dug around a bunch to learn. I hope this is roughly what you had in mind.
Fixes https://github.com/rust-lang/trait-system-refactor-initiative/issues/236.
r? `@lcnr`
cg_llvm: Stop using `as_c_char_ptr` for coverage-related bindings
[As explained by a note in `ffi.rs`](https://github.com/rust-lang/rust/blob/8a1b39995e5b630c5872f5de5079f1f569bd5ac2/compiler/rustc_codegen_llvm/src/llvm/ffi.rs#L4-L11), passing strings and byte slices through FFI is more convenient if we take advantage of the fact that `*const c_uchar` and `*const c_char` have the same ABI.
Doing so avoids having to rely on a special helper function, since we can just call `as_ptr` instead.
(The same logic applies to every other binding that currently uses the `as_c_char_ptr` helper; I just haven't adjusted all of them yet.)
---
As a drive-by change, this PR also marks some coverage-related FFI bindings as `safe`.
Don’t suggest foreign `doc(hidden)` types in "the following other types implement trait" diagnostics
Fixes https://github.com/rust-lang/rust/issues/132024.
``@rustbot`` label A-diagnostics T-compiler
Add a dummy codegen backend
This allows building a rustc capable of running the frontend without any backend present. While this may not seem all that useful, it allows running the frontend of rustc to report errors or running miri to interpret a program without any backend present. This is useful when you are trying to say run miri in the browser as upstream LLVM can't be compiled for wasm yet. Or to run rustc itself in miri like I did a while ago and caught some UB.
Improve the documentation around `ZERO_AR_DATE`
In particular, document why we don't use the new `-reproducible` flag.
I went through [the source for Apple's old linker](https://github.com/apple-oss-distributions/ld64), and compared the versions with [the mapping to Xcode versions on Wikipedia](https://en.wikipedia.org/wiki/Xcode) to find the relevant Xcode versions for these features.
r? compiler
Add fast-path for accessing the current thread id
Accessing the thread id is often used in profiling and debugging, as well as some approaches for sound single-threaded access to shared data.
Currently the only way to access the thread id is by first obtaining a handle to the current thread. While this is not exactly slow, it does require an atomic inc-ref and dec-ref operation, as well as the injection of `Thread`'s drop code into the caller.
This publicly exposes the existing fast-path for accessing the current thread id.
edit: ACP: https://github.com/rust-lang/libs-team/issues/650
Jonathan Brouwer