Kill conflicting borrows of places with projections.
Resolves#62007.
Due to a bug, the previous version of this check did not actually kill all conflicting borrows unless the borrowed place had no projections. Specifically, `sets.on_entry` will always be empty when `statement_effect` is called. It does not contain the set of borrows which are live at this point in the program.
@pnkfelix describes why this was not caught before in #62007, and created an example where the current borrow checker failed unnecessarily. This PR adds their example as a test, but they will likely want to add some additional ones.
r? @pnkfelix
Changed the error message to more clearly explain what is allowed
This is in regard to #61634. I changed the language to make it more clear what is allowed.
macos tlv workaround
fixes: #60141
Includes:
* remove dead code: `requires_move_before_drop`. This hasn't been needed for a while now (oops I should have removed it in #57655)
* redox had a copy of `fast::Key` (not sure why?). That has been removed.
* Perform a `read_volatile` on OSX to reduce `tlv_get_addr` calls per `__getit` from (4-2 depending on context) to 1.
`tlv_get_addr` is relatively expensive (~1.5ns on my machine).
Previously, in contexts where `__getit` was inlined, 4 calls to `tlv_get_addr` were performed per lookup. For some reason when `__getit` is not inlined this is reduced to 2x - and performance improves to match.
After this PR, I have only ever seen 1x call to `tlv_get_addr` per `__getit`, and macos now benefits from situations where `__getit` is inlined.
I'm not sure if the `read_volatile(&&__KEY)` trick is working around an LLVM bug, or a rustc bug, or neither.
r? @alexcrichton
Support `cfg` and `cfg_attr` on generic parameters
`cfg` attributes are supported in all other positions where attributes are accepted at all.
They were previously prohibited in https://github.com/rust-lang/rust/pull/51283 because they weren't implemented correctly before that and were simply ignored.
Improve the explicit_outlives_requirements lint
* Don't use Strings to compare parameters
* Extend the lint to lifetime bounds
* Extend the lint to enums and unions
* Use the correct span for where clauses in tuple structs
* Try to early-out where possible
* Remove unnecessary bounds in rustc crates
Refactor interning to properly mark memory as mutable or immutable
r? @RalfJung
This implementation is incomplete out of multiple reasons
* [ ] add `-Zunleash_the_miri_inside_of_you` tests
* [ ] report an error if there's an `UnsafeCell` behind a reference in a constant
* [ ] make validity checks actually test whether the mutability of their allocations match what they see in the type
Rollup of 11 pull requests
Successful merges:
- #61505 (Only show methods that appear in `impl` blocks in the Implementors sections of trait doc pages)
- #61701 (move stray run-pass const tests into const/ folder)
- #61748 (Tweak transparent enums and unions diagnostic spans)
- #61802 (Make MaybeUninit #[repr(transparent)])
- #61839 (ci: Add a script for generating CPU usage graphs)
- #61842 (Remove unnecessary lift calls)
- #61843 (Turn down the myriad-closures test)
- #61896 (rustc_typeck: correctly compute `Substs` for `Res::SelfCtor`.)
- #61898 (syntax: Factor out common fields from `SyntaxExtension` variants)
- #61938 (create an issue for miri even in status test-fail)
- #61941 (Preserve generator and yield source for error messages)
Failed merges:
r? @ghost
Preserve generator and yield source for error messages
Previously, error messages after HIR lowering all referred
to generators and yield, regardless of whether the original
source was a generator or an async/await body. This change
tracks the kind of each generator and yield source in order
to provide appropriately tailored error messages.
Fixes#60615.
Previously, error messages after HIR lowering all referred
to generators and yield, regardless of whether the original
source was a generator or an async/await body. This change
tracks the kind of each generator and yield source in order
to provide appropriately tailored error messages.
* Don't use Strings to compare parameters
* Extend the lint to lifetime bounds
* Extend the lint to enums and unions
* Use the correct span for where clauses in tuple structs
* Try to early-out where possible
Refactor C FFI variadics to more closely match their C counterparts, and add Clone implementation
We had to make some changes to expose `va_copy` and `va_end` directly to users (mainly for C2Rust, but not exclusively):
- redefine the Rust variadic structures to more closely correspond to C: `VaList` now matches `va_list`, and `VaListImpl` matches `__va_list_tag`
- add `Clone` for `VaListImpl`
- add explicit `as_va_list()` conversion function from `VaListImpl` to `VaList`
- add deref coercion from `VaList` to `VaListImpl`
- add support for the `asmjs` target
All these changes were needed for use cases like:
```Rust
let mut ap2 = va_copy(ap);
vprintf(fmt, ap2);
va_end(&mut ap2);
```
Optimize matches
Attempt to fix or improve #60571
This is breaking some diagnostics because the MIR for match arms isn't in source order any more.
cc @centril
Rollup of 6 pull requests
Successful merges:
- #61785 (note some safety concerns of raw-ptr-to-ref casts)
- #61805 (typeck: Fix ICE for blocks in repeat expr count.)
- #61813 (Remove some unnecessary symbol interner ops)
- #61824 (in which we decline to lint single-use lifetimes in `derive`d impls)
- #61844 (Change `...` to `..=` where applicable)
- #61854 (Minor cosmetic improvements to accompany PR 61825)
Failed merges:
r? @ghost
typeck: Fix ICE for blocks in repeat expr count.
Fixes#61336 (again). This PR fixes an ICE that occured when a block expression resolving to a const generic was used for the count of an array repeat expression.
r? @varkor
type_alias_enum_variants: fix#61801; allow a path pattern to infer
Fix#61801.
Given a type-relative path pattern referring to an enum variant through a type alias, allow inferring the generic argument applied in the expectation set by the scrutinee of a `match` expression.
Similar issues may exist for `let` statements but I don't know how to test for that since `PhantomData<T>` is necessary...)
The gist of the problem here was that `resolve_ty_and_res_ufcs` was called twice which is apparently no good... It is possible that this PR is papering over some deeper problem, but that is beyond my knowledge of the compiler.
r? @petrochenkov
cc @eddyb @alexreg
cc https://github.com/rust-lang/rust/pull/61682
cc https://github.com/rust-lang/rust/issues/49683
in which we decline to suggest the anonymous lifetime in declarations
The elided-lifetimes-in-path lint (part of our suite of Rust 2018 idiom lints which we are hoping to promote to Warn status) was firing with an illegal suggestion to write an anonymous lifetime in a
struct/item declaration (where we don't allow it). The linting code was already deciding whether to act on the basis of a `ParamMode` enum, indicating whether the present path-segment was part of an
expression, or anywhere else. The present case seemed to be part of the "anywhere else", and yet meriting different rules as far as the lint was concerned, so it seemed expedient to introduce a new enum member. We yank out `TyKind::Path` arm into its own method so that we can call it with our new `ParamMode` specifically when lowering struct fields—one would have hoped to think of something more elegant than this, but it definitely beats changing the signature of `lower_ty` to take a `ParamMode`!
Resolves#61124.
cc @memoryruins
r? @oli-obk
The elided-lifetimes-in-path lint (part of our suite of Rust 2018
idiom lints which we are hoping to promote to Warn status) was firing
with an illegal suggestion to write an anonymous lifetime in a
struct/item declaration (where we don't allow it). The linting code
was already deciding whether to act on the basis of a `ParamMode`
enum, indicating whether the present path-segment was part of an
expression, or anywhere else. The present case seemed to be part of
the "anywhere else", and yet meriting different rules as far as the
lint was concerned, so it seemed expedient to introduce a new enum
member. We yank out a `TyKind::Path` arm into its own method so that
we can call it with our new `ParamMode` specifically when lowering
struct fields. (The alternative strategy of changing the signature of
`lower_ty` to take a `ParamMode` would be inelegant given that most of
the `TyKind` match arm bodies therein don't concern themselves with
`ParamMode`.)
Resolves#61124.
