| Age | Commit message (Collapse) | Author | Lines |
|---|
|
Rollup of 3 pull requests
Successful merges:
- #130466 (tests: add repr/transparent test for aarch64)
- #130468 (Make sure that def id <=> lang item map is bidirectional)
- #130499 (Add myself to the libs review rotation)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Take more advantage of the `isize::MAX` limit in `Layout`
Things like `padding_needed_for` are current implemented being super careful to handle things like `Layout::size` potentially being `usize::MAX`.
But now that #95295 has happened, that's no longer a concern. It's possible to add two `Layout::size`s together without risking overflow now.
So take advantage of that to remove a bunch of checked math that's not actually needed. For example, the round-up-and-add-next-size in `extend` doesn't need any overflow checks at all, just the final check for compatibility with the alignment.
(And while I was doing that I made it all unstably const, because there's nothing in `Layout` that's fundamentally runtime-only.)
|
|
Add myself to the libs review rotation
|
|
Make sure that def id <=> lang item map is bidirectional
Self-explanatory from assertion. Just makes sure of an invariant that I forgot to enforce when I added `LanguageItems::from_def_id`.
|
|
tests: add repr/transparent test for aarch64
Fixes #74396.
Moves `transparent-struct-ptr.rs` to `transparent-byval-struct-ptr.rs` and then adds a new `transparent-opaque-ptr.rs` for aarch64.
|
|
|
|
|
|
Rollup of 4 pull requests
Successful merges:
- #130116 (Implement a Method to Seal `DiagInner`'s Suggestions)
- #130489 (Ensure that `keyword_ident` lint doesn't trigger on `'r#kw` lifetime)
- #130491 (more crash tests)
- #130496 (Fix circular fn_sig queries to correct number of args for methods)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Things like `padding_needed_for` are current implemented being super careful to handle things like `Layout::size` potentially being `usize::MAX`.
But now that 95295 has happened, that's no longer a concern. It's possible to add two `Layout::size`s together without risking overflow now.
So take advantage of that to remove a bunch of checked math that's not actually needed. For example, the round-up-and-add-next-size in `extend` doesn't need any overflow checks at all, just the final check for compatibility with the alignment.
(And while I was doing that I made it all unstably const, because there's nothing in `Layout` that's fundamentally runtime-only.)
|
|
Fix circular fn_sig queries to correct number of args for methods
Fixes #130400. This was a [debug assert](https://github.com/rust-lang/rust/blob/28e8f01c2a2f33fb4214925a704e3223b372cad5/compiler/rustc_hir_typeck/src/fn_ctxt/checks.rs#L2557) added to some argument error reporting code in #129320 which verified that the number of params (from the HIR) matched the `matched_inputs` which ultimately come from ty::FnSig. In the reduced test case:
```
fn foo(&mut self) -> _ {
foo()
}
```
There is a circular dependency computing the ty::FnSig -- when trying to compute it, we try to figure out the return value, which again depends on this ty::FnSig. In #105162, this was supported by short-circuiting the cycle by synthesizing a FnSig with error types for parameters. The [code in question](https://github.com/rust-lang/rust/pull/105162/files#diff-a65feec6bfffb19fbdc60a80becd1030c82a56c16b177182cd277478fdb04592R44) computes the number of parameters by taking the number of parameters from the hir::FnDecl and adding 1 if there is an implicit self parameter.
I might be missing a subtlety here, but AFAICT the adjustment for implicit self args is unnecessary and results in one too many args. For example, for this non-errorful code:
```
trait Foo {
fn bar(&self) {}
}
```
The resulting hir::FnDecl and ty::FnSig both have the same number of inputs -- 1. So, this PR removes that adjustment and adds a test for the debug ICE.
r? `@compiler-errors`
|
|
more crash tests
r? `@compiler-errors`
|
|
Ensure that `keyword_ident` lint doesn't trigger on `'r#kw` lifetime
Fixes #130486
|
|
Implement a Method to Seal `DiagInner`'s Suggestions
This PR adds a method on `DiagInner` called `.seal_suggestions()` to prevent new suggestions from being added while preserving existing suggestions.
This is useful because currently there is no way to prevent new suggestions from being added to a diagnostic. `.disable_suggestions()` is the closest but it gets rid of all suggestions before and after the call.
Therefore, `.seal_suggestions()` can be used when, for example, misspelled keyword is detected and reported. In such cases, we may want to prevent other suggestions from being added to the diagnostic, as they would likely be meaningless once the misspelled keyword is identified. For context: https://github.com/rust-lang/rust/pull/129899#discussion_r1741307132
To store an additional state, the type of the `suggestions` field in `DiagInner` was changed into a three variant enum. While this change affects files across different crates, care was taken to preserve the existing code's semantics. This is validated by the fact that all UI tests pass without any modifications.
r? chenyukang
|
|
tests: allow trunc/select instructions to be missing
On LLVM 20, these instructions already get eliminated, which at least partially satisfies a TODO. I'm not talented enough at using FileCheck to try and constrain this further, but if we really want to we could copy an LLVM 20 specific version of this test that would restore it to being CHECK-NEXT: insertvalue ...
`@rustbot` label: +llvm-main
r? `@DianQK`
|
|
methods
|
|
Rollup of 2 pull requests
Successful merges:
- #130481 (Remove uneeded PartialOrd bound in cmp::Ord::clamp)
- #130482 (Remove redundant test typeid equality by subtyping)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Remove redundant test typeid equality by subtyping
This known-bug label was a left over on #118247
r? `@jackh726`
This doesn't address #110395, I didn't investigate about it yet.
|
|
Remove uneeded PartialOrd bound in cmp::Ord::clamp
There is a `Self: PartialOrd` bound in `Ord::clamp`, but it is already required by the trait itself. Likely a left-over from the const trait deletion in 76dbe2910465072f85e74d6f7115ec9e6803e8bf.
Reported-by: `@noeensarguet`
|
|
|
|
|
|
Rollup of 5 pull requests
Successful merges:
- #129477 (Fix fluent diagnostics)
- #129674 (Add new_cyclic_in for Rc and Arc)
- #130452 (Update Trusty target maintainers)
- #130467 (Miri subtree update)
- #130477 (Revert #129749 to fix segfault in LLVM)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
There is a Self: PartialOrd bound in Ord::clamp, but it is already
required by the trait itself. Likely a left-over from the const trait
deletion in 76dbe2910465072f85e74d6f7115ec9e6803e8bf.
Reported-by: @noeensarguet
|
|
Revert #129749 to fix segfault in LLVM
This reverts commit 8c7a7e346be4cdf13e77ab4acbfb5ade819a4e60, reversing changes made to a00bd75b6c5c96d0a35afa2dc07ce3155112d278.
Reported in https://github.com/rust-lang/rust/pull/129749#issuecomment-2354417960. `@nikic's` theory is that the LLVM API changed in a way that makes it impossible to use concurrently from multiple threads (https://github.com/llvm/llvm-project/pull/106427#issuecomment-2354783802). I pinged `@krasimirgg` who was fine with reverting.
r? `@rust-lang/wg-llvm`
|
|
Miri subtree update
r? `@ghost`
|
|
Update Trusty target maintainers
Remove Stephen Crane from the list of Trusty target maintainers and add Andrei Homescu (`@ahomescu)` and Chris Wailes.
|
|
Add new_cyclic_in for Rc and Arc
Currently, new_cyclic_in does not exist for Rc and Arc. This is an oversight according to https://github.com/rust-lang/wg-allocators/issues/132.
This PR adds new_cyclic_in for Rc and Arc. The implementation is almost the exact same as new_cyclic with some small differences to make it allocator-specific. new_cyclic's implementation has been replaced with a call to `new_cyclic_in(data_fn, Global)`.
Remaining questions:
* ~~Is requiring Allocator to be Clone OK? According to https://github.com/rust-lang/wg-allocators/issues/88, Allocators should be cheap to clone. I'm just hesitant to add unnecessary constraints, though I don't see an obvious workaround for this function since many called functions in new_cyclic_in expect an owned Allocator. I see Allocator.by_ref() as an option, but that doesn't work on when creating Weak { ptr: init_ptr, alloc: alloc.clone() }, because the type of Weak then becomes Weak<T, &A> which is incompatible.~~ Fixed, thank you `@zakarumych!` This PR no longer requires the allocator to be Clone.
* Currently, new_cyclic_in's documentation is almost entirely copy-pasted from new_cyclic, with minor tweaks to make it more accurate (e.g. Rc<T> -> Rc<T, A>). The example section is removed to mitigate redundancy and instead redirects to cyclic_in. Is this appropriate?
* ~~The comments in new_cyclic_in (and much of the implementation) are also copy-pasted from new_cyclic. Would it be better to make a helper method new_cyclic_in_internal that both functions call, with either Global or the custom allocator? I'm not sure if that's even possible, since the internal method would have to return Arc<T, Global> and I don't know if it's possible to "downcast" that to an Arc<T>. Maybe transmute would work here?~~ Done, thanks `@zakarumych`
* Arc::new_cyclic is #[inline], but Rc::new_cyclic is not. Which is preferred?
* nit: does it matter where in the impl block new_cyclic_in is defined?
|
|
Fix fluent diagnostics
This line number calculation was both wrong and unnecessary.
|
|
|
|
Rollup of 9 pull requests
Successful merges:
- #128535 (Format `std::env::consts` docstrings with markdown backticks)
- #128961 (Fix #128930: Print documentation of CLI options missing their arg)
- #129988 (Use `Vec` in `rustc_interface::Config::locale_resources`)
- #130201 (Encode `coroutine_by_move_body_def_id` in crate metadata)
- #130275 (Don't call `extern_crate` when local crate name is the same as a dependency and we have a trait error)
- #130314 (Use the same precedence for all macro-like exprs)
- #130440 (Don't ICE in `opaque_hidden_inferred_bound` lint for RPITIT in trait with no default method body)
- #130458 (`rustc_codegen_ssa` cleanups)
- #130469 (Mark `where_clauses_object_safety` as removed)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Mark `where_clauses_object_safety` as removed
r? lcnr
|
|
`rustc_codegen_ssa` cleanups
Just some minor improvements I found while reading through this code.
r? ``@jieyouxu``
|
|
Don't ICE in `opaque_hidden_inferred_bound` lint for RPITIT in trait with no default method body
Inline comment should explain the fix.
Fixes #130422
|
|
Use the same precedence for all macro-like exprs
No need to make these have a different precedence since they're all written like `whatever!(expr)`, and it makes it simpler when adding new macro-based built-in operators in the future.
|
|
Don't call `extern_crate` when local crate name is the same as a dependency and we have a trait error
#124944 implemented logic to point out when a trait bound failure involves a *trait* and *type* who come from identically named but different crates. This logic calls the `extern_crate` query which is not valid on `LOCAL_CRATE` cnum, so let's filter that out eagerly.
Fixes #130272
Fixes #129184
|
|
Encode `coroutine_by_move_body_def_id` in crate metadata
We synthesize the MIR for a by-move body for the `FnOnce` implementation of async closures. It can be accessed with the `coroutine_by_move_body_def_id` query. We weren't encoding this query in the metadata though, nor were we properly recording that synthetic MIR in `mir_keys`, so the `optimized_mir` wasn't getting encoded either!
Stacked on top is a fix to consider `DefKind::SyntheticCoroutineBody` to return true in several places I missed. Specifically, we should consider the def-kind in `fn DefKind::is_fn_like()`, since that's what we were using to make sure we ensure `query mir_inliner_callees` before the MIR gets stolen for the body. This led to some CI failures that were caught by miri but which I added a test for.
|
|
Use `Vec` in `rustc_interface::Config::locale_resources`
This allows a third-party tool to injects its own resources, when receiving the config via `rustc_driver::Callbacks::config`.
|
|
Fix #128930: Print documentation of CLI options missing their arg
Fix #128930. Failing to give an argument to CLI options which require it now prints something like:
```
$ rustc --print
error: Argument to option 'print' missing
Usage:
--print [crate-name|file-names|sysroot|target-libdir|cfg|check-cfg|calling-conventions|target-list|target-cpus|target-features|relocation-models|code-models|tls-models|target-spec-json|all-target-specs-json|native-static-libs|stack-protector-strategies|link-args|deployment-target]
Compiler information to print on stdout
```
|
|
Format `std::env::consts` docstrings with markdown backticks
This clarifies possible outputs the constants might be.
**Before:**
--
<img src="https://github.com/user-attachments/assets/8ee8772a-7562-42a2-89be-f8772b76dbd5" width="500px">
**After:**
--
<img src="https://github.com/user-attachments/assets/4632e5e2-db3e-4372-b13e-006cc1701eb1" width="500px">
|
|
This line number calculation was both wrong and unnecessary.
|
|
|
|
|
|
|
|
|
|
On LLVM 20, these instructions already get eliminated, which at least
partially satisfies a TODO. I'm not talented enough at using FileCheck
to try and constrain this further, but if we really want to we could
copy an LLVM 20 specific version of this test that would restore it to
being CHECK-NEXT: insertvalue ...
@rustbot label: +llvm-main
|
|
Relate receiver invariantly in method probe for `Mode::Path`
Effectively reverts part of #126128
Fixes #126227
This PR changes method probing to use equality for fully path-based method lookup, and subtyping for receiver `.` method lookup.
r? lcnr
|
|
Moves `transparent-struct-ptr.rs` to `transparent-byval-struct-ptr.rs`
and then adds a new `transparent-opaque-ptr.rs` for aarch64.
|
|
Fix tokio test ICE
Fixes #3858
It turned out that the issue mentioned [here](https://github.com/rust-lang/miri/issues/3858#issuecomment-2336726299) is the exact cause of ICE.
So in this PR, I changed the type of ``EpollEventInterest::epfd`` from ``i32`` to ``WeakFileDescriptionRef``.
|
|
WeakFileDescriptionRef
|
|
Remove semi-nondeterminism of `DefPathHash` ordering from inliner
Déjà vu or something because I kinda thought I had put this PR up before. I recall a discussion somewhere where I think it was `@saethlin` mentioning that this check was no longer needed since we have "proper" cycle detection. Putting that up as a PR now.
This may slighlty negatively affect inlining, since the cycle breaking here means that we still inlined some cycles when the def path hashes were ordered in certain ways, this leads to really bad nondeterminism that makes minimizing ICEs and putting up inliner bugfixes difficult.
r? `@cjgillot` or `@saethlin` or someone else idk
|
|
Automatic Rustup
|
