| Age | Commit message (Collapse) | Author | Lines |
|---|
|
print raw lifetime idents with r#
This replaces rust-lang/rust#143185 and fixes rust-lang/rust#143150
cc ``@fmease``
|
|
Convert moves of references to copies in ReferencePropagation
This is a fix for https://github.com/rust-lang/rust/issues/141101.
The root cause of this miscompile is that the SsaLocals analysis that MIR transforms use is supposed to detect locals that are only written to once, in their single assignment. But that analysis is subtly wrong; it does not consider `Operand::Move` to be a write even though the meaning ascribed to `Operand::Move` (at least as a function parameter) by Miri is that the callee may have done arbitrary writes to the caller's Local that the Operand wraps (because `Move` is pass-by-pointer). So Miri conwiders `Operand::Move` to be a write but both the MIR visitor system considers it a read, and so does SsaLocals.
I have tried fixing this by changing the `PlaceContext` that is ascribed to an `Operand::Move` to a `MutatingUseContext` but that seems to have borrow checker implications, and changing SsaLocals seems to have wide-ranging regressions in MIR optimizations.
So instead of doing those, this PR adds a new kludge to ReferencePropagation, which follows the same line of thinking as the kludge in CopyProp that solves this same problem inside that pass: https://github.com/rust-lang/rust/blob/a5584a8fe16037dc01782064fa41424a6dbe9987/compiler/rustc_mir_transform/src/copy_prop.rs#L65-L98
|
|
Recover `param: Ty = EXPR`
Fixes #137310
Pretty basic recovery here, but better than giving an unexpected token error.
|
|
Split transmute check from HIR typeck
This resolves a FIXME in the implementation of `check_transmute`.
`check_transmute` needs to compute type layout, hence needing to see reveal opaques and all type aliases.
Having this inside typeck causes a cycle. For instance: `tests/ui/impl-trait/transmute/in-defining-scope.rs`.
This PR moves the transmute check outside of typeck, by putting the list of deferred transmute checks in typeck results.
|
|
Only unpack span data once to compute end_point and next_point.
Split from https://github.com/rust-lang/rust/pull/144930
|
|
|
|
This reverts commit 585a40963ea59808e74803f8610659a505b145e0.
|
|
|
|
|
|
|
|
|
|
Signed-off-by: Jens Reidel <adrian@travitia.xyz>
|
|
When encountering an unmet trait bound, point at local type that doesn't implement the trait:
```
error[E0277]: the trait bound `Bar<T>: Foo` is not satisfied
--> $DIR/issue-64855.rs:9:19
|
LL | pub struct Bar<T>(<Self as Foo>::Type) where Self: ;
| ^^^^^^^^^^^^^^^^^^^ unsatisfied trait bound
|
help: the trait `Foo` is not implemented for `Bar<T>`
--> $DIR/issue-64855.rs:9:1
|
LL | pub struct Bar<T>(<Self as Foo>::Type) where Self: ;
| ^^^^^^^^^^^^^^^^^
```
|
|
|
|
This more closely matches the RFC and what our T-lang contact has asked
for, see https://github.com/rust-lang/rust/issues/136889#issuecomment-3212715312
|
|
The difference between `Late` and `All` only matters when `finalize` is enabled.
So add a `stage` field to `Finalize` and use it instead.
|
|
|
|
It can be passed in a more usual way through `ParentScope`
|
|
|
|
|
|
Revise the extra `r: 'static` constraints added upon universe issues
to add an explanation, and use that explanation during constraint blame
search. This greatly simplifies the region inference logic, which
now does not need to reverse-engineer the event that caused a region
to outlive 'static.
|
|
Rewrite the new attribute argument parser
Fixes https://github.com/rust-lang/rust/issues/143940
This rewrites the parser, should improve performance and maintainability.
This can be reviewed commit by commit
|
|
directly
|
|
This variant doesn't appear to have ever been used.
There's a matching message in `rustc_resolve`, that used to have a FIXME
for porting it to the new diagnostic infrastructure, but that message is
using `feature_err`, which doesn't use buffered lints. Thus, even when
that does get ported, it won't use `BuiltinLintDiag`.
|
|
|
|
|
|
Lint buffering currently relies on a giant enum `BuiltinLintDiag`
containing all the lints that might potentially get buffered. In
addition to being an unwieldy enum in a central crate, this also makes
`rustc_lint_defs` a build bottleneck: it depends on various types from
various crates (with a steady pressure to add more), and many crates
depend on it.
Having all of these variants in a separate crate also prevents detecting
when a variant becomes unused, which we can do with a dedicated type
defined and used in the same crate.
Refactor this to use a dyn trait, to allow using `LintDiagnostic` types
directly.
This requires boxing, but all of this is already on the slow path
(emitting an error).
Because the existing `BuiltinLintDiag` requires some additional types in
order to decorate some variants, which are only available later in
`rustc_lint`, use an enum `DecorateDiagCompat` to handle both the `dyn
LintDiagnostic` case and the `BuiltinLintDiag` case.
|
|
|
|
Sort mono items by symbol name
Trying to claw back cycles/branch/cache miss losses from https://github.com/rust-lang/rust/pull/144722.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rustc_expand: ensure stack in `InvocationCollector::visit_expr`
In Fedora, when we built rustc with PGO on ppc64le, we started failing
the test `issue-74564-if-expr-stack-overflow.rs`. This could also be
reproduced on other arches by setting a smaller `RUST_MIN_STACK`, so
it's probably just unlucky that ppc64le PGO created a large stack frame
somewhere in this recursion path. Adding an `ensure_sufficient_stack`
solves the stack overflow.
Historically, that test and its fix were added in rust-lang/rust#74708,
which was also an `ensure_sufficient_stack` in this area of code at the
time. However, the refactor in rust-lang/rust#92573 basically left that
to the general `MutVisitor`, and then rust-lang/rust#142240 removed even
that ensure call. It may be luck that our tier-1 tested targets did not
regress the original issue across those refactors.
|
|
rustc_data_structures.
|
|
|
|
This uses the feature gate for
https://github.com/rust-lang/rust/issues/143352, but is described in
https://github.com/rust-lang/rfcs/pull/3820 which is strongly tied to
the experiment.
|
|
Update outdated link in bound region comments
While reading the implementation code for bound regions, I found that a link in the comments was outdated. I've updated it with a link to the corresponding documentation in the rustc dev guide that covers the same content.
prev link: https://rustc-dev-guide.rust-lang.org/early-late-bound-params/early-late-bound-summary.html (404 error)
updated: https://rustc-dev-guide.rust-lang.org/early_late_parameters.html
|
|
remove an `as` cast in prefetch codegen
r? `@RalfJung`
|
|
Handle `ReEarlyParam` in `type_name`.
Fixes rust-lang/rust#145696.
r? `@lcnr`
|
|
Instantiate higher-ranked binder with erased when checking `IntoIterator` predicate for query instability
Fixes https://github.com/rust-lang/rust/issues/145652 which was introduced by https://github.com/rust-lang/rust/pull/139345 because we were skipping a binder before calling `Instance::try_resolve`.
r? lcnr
|
|
Gate static closures behind a parser feature
I'd like to gate `static ||` closures behind a feature gate, since we shouldn't allow people to take advantage of this syntax if it's currently unstable. Right now, since it's only rejected after ast lowering, it's accessible to macros.
Let's crater this to see if we can claw it back without breaking anyone's code.
|
|
Refactor attribute parsing to improve ergonomics and some diagnostics
|
|
Add a method to dump MIR in the middle of MIR building
This makes it easier to debug issues with MIR building by inserting dump_for_debugging calls around the suspected code responsible for the bad MIR.
|
|
aarch64-nintendo-switch-freestanding - Enable CPU features that are always available in a live system (crypto instructions, plus explicit NEON).
~~While some NEON and crypto features may not be supported on the Nintendo Switch at boot (e.g. on the a53 cores) and this has not been tested, the features will _always_ be available if running as a sysmodule or homebrew application under Horizon/Atmosphere.~~ EDIT: the a53 cores are fused out, these features are always available.
This has been tested with local tools personally, as well as building [emuiibo](https://github.com/XorTroll/emuiibo) as it uses both `sha` and `aes` primitives. This was tested using inline assembly in previous versions, and in current versions by using the `aes`, `ctr`, `hmac`, and `sha2` crates.
r? `@jam1garner`
This ended up being much delayed from our discussions about updating this. I tested a number of individual features such as the `aes` and `sha2` target-features directly to avoid a warning message with the `crypto` feature, but that appears to be caused by https://github.com/rust-lang/rust/issues/96472 and is not actually an issue.
There is also a decision to make here about explicitly enabling the `neon` feature. I am in favor of it to be explicit, but it is not necessary as it is already enabled by the `v8a` and `crypto` features. I will defer to your decision as it does not change the actual instructions available for codegen.
|
|
|
|
|
|
|
