Add ability to transmute (somewhat) with generic consts in arrays
Previously if the expression contained generic consts and did not have a directly equivalent type, transmuting the type in this way was forbidden, despite the two sizes being identical. Instead, we should be able to lazily tell if the two consts are identical, and if so allow them to be transmuted.
This is done by normalizing the forms of expressions into sorted order of multiplied terms, which is not generic over all expressions, but should handle most cases.
This allows for some _basic_ transmutations between types that are equivalent in size without requiring additional stack space at runtime.
I only see one other location at which `SizeSkeleton` is being used, and it checks for equality so this shouldn't affect anywhere else that I can tell.
See [this Stackoverflow post](https://stackoverflow.com/questions/73085012/transmute-nested-const-generic-array-rust) for what was previously necessary to convert between types. This PR makes converting nested `T -> [T; 1]` transmutes possible, and `[uB*2; N] -> [uB; N * 2]` possible as well.
I'm not sure whether this is something that would be wanted, and if it is it definitely should not be insta-stable, so I'd add a feature gate.
Move `doc(primitive)` future incompat warning to `invalid_doc_attributes`
Fixes#88070.
It's been a while since this was turned into a "future incompatible lint" so I think we can now turn it into a hard error without problem.
r? `@jyn514`
Initial support for return type notation (RTN)
See: https://smallcultfollowing.com/babysteps/blog/2023/02/13/return-type-notation-send-bounds-part-2/
1. Only supports `T: Trait<method(): Send>` style bounds, not `<T as Trait>::method(): Send`. Checking validity and injecting an implicit binder for all of the late-bound method generics is harder to do for the latter.
* I'd add this in a follow-up.
3. ~Doesn't support RTN in general type position, i.e. no `let x: <T as Trait>::method() = ...`~
* I don't think we actually want this.
5. Doesn't add syntax for "eliding" the function args -- i.e. for now, we write `method(): Send` instead of `method(..): Send`.
* May be a hazard if we try to add it in the future. I'll probably add it in a follow-up later, with a structured suggestion to change `method()` to `method(..)` once we add it.
7. ~I'm not in love with the feature gate name 😺~
* I renamed it to `return_type_notation` ✔️
Follow-up PRs will probably add support for `where T::method(): Send` bounds. I'm not sure if we ever want to support return-type-notation in arbitrary type positions. I may also make the bounds require `..` in the args list later.
r? `@ghost`
Insert alignment checks for pointer dereferences when debug assertions are enabled
Closes https://github.com/rust-lang/rust/issues/54915
- [x] Jake tells me this sounds like a place to use `MirPatch`, but I can't figure out how to insert a new basic block with a new terminator in the middle of an existing basic block, using `MirPatch`. (if nobody else backs up this point I'm checking this as "not actually a good idea" because the code looks pretty clean to me after rearranging it a bit)
- [x] Using `CastKind::PointerExposeAddress` is definitely wrong, we don't want to expose. Calling a function to get the pointer address seems quite excessive. ~I'll see if I can add a new `CastKind`.~ `CastKind::Transmute` to the rescue!
- [x] Implement a more helpful panic message like slice bounds checking.
r? `@oli-obk`
Rename `with_source_map` as `set_source_map`. Because `with` functions
(e.g. `with_session_globals`, `scoped_tls::ScopedKey::with`) are for
*getting* a value for the duration of a closure, and `set` functions
(e.g. `set_session_globals_then` `scoped_tls::ScopedKey::with`) are for
*setting* a value for the duration of a closure.
Also fix up the comment, which is wrong:
- The bit about `TyCtxt` is wrong.
- `span_debug1` doesn't exist any more.
- There's only one level of fallback, not two.
(This is effectively a follow-up to the changes in #93936.)
Also add a comment explaining that `SessionGlobals::source_map` should
only be used when absolutely necessary.
Rollup of 8 pull requests
Successful merges:
- #91793 (socket ancillary data implementation for FreeBSD (from 13 and above).)
- #92284 (Change advance(_back)_by to return the remainder instead of the number of processed elements)
- #102472 (stop special-casing `'static` in evaluation)
- #108480 (Use Rayon's TLV directly)
- #109321 (Erase impl regions when checking for impossible to eagerly monomorphize items)
- #109470 (Correctly substitute GAT's type used in `normalize_param_env` in `check_type_bounds`)
- #109562 (Update ar_archive_writer to 0.1.3)
- #109629 (remove obsolete `givens` from regionck)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Eagerly intern and check CrateNum/StableCrateId collisions
r? ``@cjgillot``
It seems better to check things ahead of time than checking them afterwards.
The [previous version](https://github.com/rust-lang/rust/pull/108390) was a bit nonsensical, so this addresses the feedback
move Option::as_slice to intrinsic
````@scottmcm```` suggested on #109095 I use a direct approach of unpacking the operation in MIR lowering, so here's the implementation.
cc ````@nikic```` as this should hopefully unblock #107224 (though perhaps other changes to the prior implementation, which I left for bootstrapping, are needed).
Ensure `ptr::read` gets all the same LLVM `load` metadata that dereferencing does
I was looking into `array::IntoIter` optimization, and noticed that it wasn't annotating the loads with `noundef` for simple things like `array::IntoIter<i32, N>`. Trying to narrow it down, it seems that was because `MaybeUninit::assume_init_read` isn't marking the load as initialized (<https://rust.godbolt.org/z/Mxd8TPTnv>), which is unfortunate since that's basically its reason to exist.
The root cause is that `ptr::read` is currently implemented via the *untyped* `copy_nonoverlapping`, and thus the `load` doesn't get any type-aware metadata: no `noundef`, no `!range`. This PR solves that by lowering `ptr::read(p)` to `copy *p` in MIR, for which the backends already do the right thing.
Fortuitiously, this also improves the IR we give to LLVM for things like `mem::replace`, and fixes a couple of long-standing bugs where `ptr::read` on `Copy` types was worse than `*`ing them.
Zulip conversation: <https://rust-lang.zulipchat.com/#narrow/stream/219381-t-libs/topic/Move.20array.3A.3AIntoIter.20to.20ManuallyDrop/near/341189936>
cc `@erikdesjardins` `@JakobDegen` `@workingjubilee` `@the8472`
Fixes#106369Fixes#73258
Remove `identity_future` indirection
This was previously needed because the indirection used to hide some unexplained lifetime errors, which it turned out were related to the `min_choice` algorithm.
Removing the indirection also solves a couple of cycle errors, large moves and makes async blocks support the `#[track_caller]`annotation.
Fixes https://github.com/rust-lang/rust/issues/104826.
Allow binary files to go through the `FileLoader`
I'd like for `include_bytes!` to go through the `FileLoader` in an out-of-tree `rustc_driver` wrapper, and I can't find a reason it's not already done. It seems like most folks providing a custom `FileLoader` would want this too, so I added it.
I can solve my problem in other ways if there's a strong reason not to do it, but it seems simple and harmless.
I was looking into `array::IntoIter` optimization, and noticed that it wasn't annotating the loads with `noundef` for simple things like `array::IntoIter<i32, N>`.
Turned out to be a more general problem as `MaybeUninit::assume_init_read` isn't marking the load as initialized (<https://rust.godbolt.org/z/Mxd8TPTnv>), which is unfortunate since that's basically its reason to exist.
This PR lowers `ptr::read(p)` to `copy *p` in MIR, which fortuitiously also improves the IR we give to LLVM for things like `mem::replace`.
Make `unused_allocation` lint against `Box::new` too
Previously it only linted against `box` syntax, which likely won't ever be stabilized, which is pretty useless. Even now I'm not sure if it's a meaningful lint, but it's at least something 🤷
This means that code like the following will be linted against:
```rust
Box::new([1, 2, 3]).len();
f(&Box::new(1)); // where f : &i32 -> ()
```
The lint works by checking if a `Box::new` (or `box`) expression has an a borrow adjustment, meaning that the code that first stores the box in a variable won't be linted against:
```rust
let boxed = Box::new([1, 2, 3]); // no lint
boxed.len();
```
This was previously needed because the indirection used to hide some unexplained lifetime errors, which it turned out were related to the `min_choice` algorithm.
Removing the indirection also solves a couple of cycle errors, large moves and makes async blocks support the `#[track_caller]` annotation.
Add `round_ties_even` to `f32` and `f64`
Tracking issue: #96710
Redux of #82273. See also #55107
Adds a new method, `round_ties_even`, to `f32` and `f64`, that rounds the float to the nearest integer , rounding halfway cases to the number with an even least significant bit. Uses the `roundeven` LLVM intrinsic to do this.
Of the five IEEE 754 rounding modes, this is the only one that doesn't already have a round-to-integer function exposed by Rust (others are `round`, `floor`, `ceil`, and `trunc`). Ties-to-even is also the rounding mode used for int-to-float and float-to-float `as` casts, as well as float arithmentic operations. So not having an explicit rounding method for it seems like an oversight.
Bikeshed: this PR currently uses `round_ties_even` for the name of the method. But maybe `round_ties_to_even` is better, or `round_even`, or `round_to_even`?
Desugaring of drop and replace at MIR build
This commit desugars the drop and replace deriving from an
assignment at MIR build, avoiding the construction of the
`DropAndReplace` terminator (which will be removed in a following PR).
In order to retain the same error messages for replaces a new
`DesugaringKind::Replace` variant is introduced.
The changes in the borrowck are also useful for future work in moving drop elaboration
before borrowck, as no `DropAndReplace` would be present there anymore.
Notes on test diffs:
* `tests/ui/borrowck/issue-58776-borrowck-scans-children`: the assignment deriving from the desugaring kills the borrow.
* `tests/ui/async-await/async-fn-size-uninit-locals.rs`, `tests/mir-opt/issue_41110.test.ElaborateDrops.after.mir`, `tests/mir-opt/issue_41888.main.ElaborateDrops.after.mir`: drop elaboration generates (or reads from) a useless drop flag due to an issue with the dataflow analysis. Will be fixed independently by https://github.com/rust-lang/rust/pull/106430.
See https://github.com/rust-lang/rust/pull/104488 for more context
This commit desugars the drop and replace deriving from an
assignment at MIR build, avoiding the construction of the
DropAndReplace terminator (which will be removed in a followign PR)
In order to retain the same error messages for replaces a new
DesugaringKind::Replace variant is introduced.
Remove `from` lang item
It was probably a leftover from the old `?` desugaring but anyways, it's unused now except for clippy, which can just use a diagnostics item.
Implement partial support for non-lifetime binders
This implements support for non-lifetime binders. It's pretty useless currently, but I wanted to put this up so the implementation can be discussed.
Specifically, this piggybacks off of the late-bound lifetime collection code in `rustc_hir_typeck::collect::lifetimes`. This seems like a necessary step given the fact we don't resolve late-bound regions until this point, and binders are sometimes merged.
Q: I'm not sure if I should go along this route, or try to modify the earlier nameres code to compute the right bound var indices for type and const binders eagerly... If so, I'll need to rename all these queries to something more appropriate (I've done this for `resolve_lifetime::Region` -> `resolve_lifetime::ResolvedArg`)
cc rust-lang/types-team#81
r? `@ghost`
Use associated items of `char` instead of freestanding items in `core::char`
The associated functions and constants on `char` have been stable since 1.52 and the freestanding items have soft-deprecated since 1.62 (https://github.com/rust-lang/rust/pull/95566). This PR ~~marks them as "deprecated in future", similar to the integer and floating point modules (`core::{i32, f32}` etc)~~ replaces all uses of `core::char::*` with `char::*` to prepare for future deprecation of `core::char::*`.
Enable new rlib in non stable cases
If bundled static library uses cfg (unstable) or whole-archive (wasn't supported) bundled libs are packed even without packed_bundled_libs.
r? `@petrochenkov`
Rename `PointerSized` to `PointerLike`
The old name was unnecessarily vague. This PR renames a nightly language feature that I added, so I don't think it needs any additional approval, though anyone can feel free to speak up if you dislike the rename.
It's still unsatisfying that we don't the user which of {size, alignment} is wrong, but this trait really is just a stepping stone for a more generalized mechanism to create `dyn*`, just meant for nightly testing, so I don't think it really deserves additional diagnostic machinery for now.
Fixes#107696, cc ``@RalfJung``
r? ``@eholk``
Move format_args!() into AST (and expand it during AST lowering)
Implements https://github.com/rust-lang/compiler-team/issues/541
This moves FormatArgs from rustc_builtin_macros to rustc_ast_lowering. For now, the end result is the same. But this allows for future changes to do smarter things with format_args!(). It also allows Clippy to directly access the ast::FormatArgs, making things a lot easier.
This change turns the format args types into lang items. The builtin macro used to refer to them by their path. After this change, the path is no longer relevant, making it easier to make changes in `core`.
This updates clippy to use the new language items, but this doesn't yet make clippy use the ast::FormatArgs structure that's now available. That should be done after this is merged.
Transform async `ResumeTy` in generator transform
- Eliminates all the `get_context` calls that async lowering created.
- Replace all `Local` `ResumeTy` types with `&mut Context<'_>`.
The `Local`s that have their types replaced are:
- The `resume` argument itself.
- The argument to `get_context`.
- The yielded value of a `yield`.
The `ResumeTy` hides a `&mut Context<'_>` behind an unsafe raw pointer, and the `get_context` function is being used to convert that back to a `&mut Context<'_>`.
Ideally the async lowering would not use the `ResumeTy`/`get_context` indirection, but rather directly use `&mut Context<'_>`, however that would currently lead to higher-kinded lifetime errors.
See <https://github.com/rust-lang/rust/issues/105501>.
The async lowering step and the type / lifetime inference / checking are still using the `ResumeTy` indirection for the time being, and that indirection is removed here. After this transform, the generator body only knows about `&mut Context<'_>`.
---
Fixes https://github.com/bjorn3/rustc_codegen_cranelift/issues/1330 CC `@bjorn3`
r? `@compiler-errors`
- Eliminates all the `get_context` calls that async lowering created.
- Replace all `Local` `ResumeTy` types with `&mut Context<'_>`.
The `Local`s that have their types replaced are:
- The `resume` argument itself.
- The argument to `get_context`.
- The yielded value of a `yield`.
The `ResumeTy` hides a `&mut Context<'_>` behind an unsafe raw pointer, and the
`get_context` function is being used to convert that back to a `&mut Context<'_>`.
Ideally the async lowering would not use the `ResumeTy`/`get_context` indirection,
but rather directly use `&mut Context<'_>`, however that would currently
lead to higher-kinded lifetime errors.
See <https://github.com/rust-lang/rust/issues/105501>.
The async lowering step and the type / lifetime inference / checking are
still using the `ResumeTy` indirection for the time being, and that indirection
is removed here. After this transform, the generator body only knows about `&mut Context<'_>`.
Various cleanups around pre-TyCtxt queries and functions
part of #105462
based on https://github.com/rust-lang/rust/pull/106776 (everything starting at [0e2b39f](0e2b39fd1f) is new in this PR)
r? `@petrochenkov`
I think this should be most of the uncontroversial part of #105462.
Remove duplicate sha-1 dependency
[`sha-1`](https://crates.io/crates/sha-1) is more or less a duplicate of [`sha1`](https://crates.io/crates/sha1). The `sha-1` is deprecated and no longer updated. This updates the dependencies to use the new name.
Some other dependencies that got updated as a consequence:
* The updated pest dependencies are currently only used by mdbook, and shouldn't have any issues.
* ucd-trie 0.1.3 to 0.1.5: No changelog, but looks like some tables were updated for new unicode versions: https://github.com/BurntSushi/ucd-generate/commits/master/ucd-trie. This is only used by pest (and thus mdbook).
* thiserror 1.33 to 1.38: Nothing significant in the notes at https://github.com/dtolnay/thiserror/releases.
Convert all the crates that have had their diagnostic migration
completed (except save_analysis because that will be deleted soon and
apfloat because of the licensing problem).
Encode spans relative to the enclosing item -- enable on nightly
Follow-up to #84373 with the flag `-Zincremental-relative-spans` set by default.
This PR seeks to remove one of the main shortcomings of incremental: the handling of spans.
Changing the contents of a function may require redoing part of the compilation process for another function in another file because of span information is changed.
Within one file: all the spans in HIR change, so typechecking had to be re-done.
Between files: spans of associated types/consts/functions change, so type-based resolution needs to be re-done (hygiene information is stored in the span).
The flag `-Zincremental-relative-spans` encodes local spans relative to the span of an item, stored inside the `source_span` query.
Trap: stashed diagnostics are referenced by the "raw" span, so stealing them requires to remove the span's parent.
In order to avoid too much traffic in the span interner, span encoding uses the `ctxt_or_tag` field to encode:
- the parent when the `SyntaxContext` is 0;
- the `SyntaxContext` when the parent is `None`.
Even with this, the PR creates a lot of traffic to the Span interner, when a Span has both a LocalDefId parent and a non-root SyntaxContext. They appear in lowering, when we add a parent to all spans, including those which come from macros, and during inlining when we mark inlined spans.
The last commit changes how queries of `LocalDefId` manage their cache. I can put this in a separate PR if required.
Possible future directions:
- validate that all spans are marked in HIR validation;
- mark macro-expanded spans relative to the def-site and not the use-site.
Revert "Implement allow-by-default `multiple_supertrait_upcastable` lint"
This is a clean revert of #105484.
I confirmed that reverting that PR fixes the regression reported in #106247. ~~I can't say I understand what this code is doing, but maybe it can be re-landed with a different implementation.~~ **Edit:** https://github.com/rust-lang/rust/issues/106247#issuecomment-1367174384 has an explanation of why #105484 ends up surfacing spurious `where_clause_object_safety` errors. The implementation of `where_clause_object_safety` assumes we only check whether a trait is object safe when somebody actually uses that trait with `dyn`. However the implementation of `multiple_supertrait_upcastable` added in the problematic PR involves checking *every* trait for whether it is object-safe.
FYI `@nbdd0121` `@compiler-errors`
Improve heuristics whether `format_args` string is a source literal
Previously, it only checked whether there was _a_ literal at the span of the first argument, not whether the literal actually matched up. This caused issues when a proc macro was generating a different literal with the same span.
This requires an annoying special case for literals ending in `\n` because otherwise `println` wouldn't give detailed diagnostics anymore which would be bad.
Fixes#106191
Implement allow-by-default `multiple_supertrait_upcastable` lint
The lint detects when an object-safe trait has multiple supertraits.
Enabled in libcore and liballoc as they are low-level enough that many embedded programs will use them.
r? `@nikomatsakis`
