Delayed bug audit
I went through all the calls to `delayed_bug` and `span_delayed_bug` and found a few places where they could be avoided.
r? `@compiler-errors`
Remove useless lifetime of ArchiveBuilder
`trait ArchiveBuilder<'a>` has a seemingly useless lifetime a, so I remove it. If this is intentional, please reject this PR.
```rust
pub trait ArchiveBuilder<'a> {
fn add_file(&mut self, path: &Path);
fn add_archive(
&mut self,
archive: &Path,
skip: Box<dyn FnMut(&str) -> bool + 'static>,
) -> io::Result<()>;
fn build(self: Box<Self>, output: &Path) -> bool;
}
```
Modified according to https://github.com/rust-lang/compiler-team/issues/505.
By running test cases, I found that modifying the attribute's only_local tag sometimes causes some unintuitive error reports, so I tend to split it into multiple PRs and edit a small number of attributes each time to prevent too many changes at once. Prevent possible subsequent difficulties in locating problems.
rename 'try' intrinsic to 'catch_unwind'
The intrinsic has nothing to do with `try` blocks, and corresponds to the stable `catch_unwind` function, so this makes a lot more sense IMO.
Also rename Miri's special function while we are at it, to reflect the level of abstraction it works on: it's an unwinding mechanism, on which Rust implements panics.
Rollup of 6 pull requests
Successful merges:
- #121389 (llvm-wrapper: fix few warnings)
- #121493 (By changing some attributes to only_local, reducing encoding attributes in the crate metadate.)
- #121615 (Move `emit_stashed_diagnostic` call in rustfmt.)
- #121617 (Actually use the right closure kind when checking async Fn goals)
- #121628 (Do not const prop unions)
- #121629 (fix some references to no-longer-existing ReprOptions.layout_seed)
r? `@ghost`
`@rustbot` modify labels: rollup
By changing some attributes to only_local, reducing encoding attributes in the crate metadate.
Thank you.
This is part of changing attributes to only_local. I hope get your opinion whether I should split into multiple PRs, or submit in one.
According to [try to not rely on attributes from extern crates](https://github.com/rust-lang/compiler-team/issues/505) and lcnr's guidance.
avoid generalization inside of aliases
The basic idea of this PR is that we don't generalize aliases when the instantiation could fail later on, either due to the *occurs check* or because of a universe error. We instead replace the whole alias with an inference variable and emit a nested `AliasRelate` goal. This `AliasRelate` then fully normalizes the alias before equating it with the inference variable, at which point the alias can be treated like any other rigid type.
We now treat aliases differently depending on whether they are *rigid* or not. To detect whether an alias is rigid we check whether `NormalizesTo` fails. While we already do so inside of `AliasRelate` anyways, also doing so when instantiating a query response would be both ugly/difficult and likely inefficient. To avoid that I change `instantiate_and_apply_query_response` to relate types completely structurally. This change generally removes a lot of annoying complexity, which is nice. It's implemented by adding a flag to `Equate` to change it to structurally handle aliases.
We currently always apply constraints from canonical queries right away. By providing all the necessary information to the canonical query, we can guarantee that instantiating the query response never fails, which further simplifies the implementation. This does add the invariant that *any information which could cause instantiating type variables to fail must also be available inside of the query*.
While it's acceptable for canonicalization to result in more ambiguity, we must not cause the solver to incompletely structurally relate aliases by erasing information. This means we have to be careful when merging universes during canonicalization. As we only generalize for type and const variables we have to make sure that anything nameable by such a type or const variable inside of the canonical query is also nameable outside of it. Because of this we both stop merging universes of existential variables when canonicalizing inputs, we put all uniquified regions into a higher universe which is not nameable by any type or const variable.
I will look into always replacing aliases with inference variables when generalizing in a later PR unless the alias references bound variables. This should both pretty much fix https://github.com/rust-lang/trait-system-refactor-initiative/issues/4. This may allow us to merge the universes of existential variables again by changing generalize to not consider their universe when deciding whether to generalize aliases. This requires some additional non-trivial changes to alias-relate, so I am leaving that as future work.
Fixes https://github.com/rust-lang/trait-system-refactor-initiative/issues/79. While it would be nice to decrement universe indices when existing a `forall`, that was surprisingly difficult and not necessary to fix this issue. I am really happy with the approach in this PR think it is the correct way forward to also fix the remaining cases of https://github.com/rust-lang/trait-system-refactor-initiative/issues/8.
only checking whether nested goals hold means that we don't consider
their inference constraints. Given that we now emit `AliasRelate` when relating
aliases and infer vars, this previously resulted in an "unconstrained" inference var
in `coerce_unsized`.
Fix more #121208 fallout (round 3)
#121208 converted lots of delayed bugs to bugs. Unsurprisingly, there were a few invalid conversion found via fuzzing.
r? `@lcnr`
Split Diagnostics for Uncommon Codepoints: Add Individual Identifier Types
This pull request further modifies the `uncommon_codepoints` lint, adding the individual identifier types of `Technical`, `Not_NFKC`, `Exclusion` and `Limited_Use` to the diagnostic message.
Example rendered diagnostic:
```
error: identifier contains a Unicode codepoint that is not used in normalized strings: 'ij'
--> $DIR/lint-uncommon-codepoints.rs:6:4
|
LL | fn dijkstra() {}
| ^^^^^^^
= note: this character is included in the Not_NFKC Unicode general security profile
```
Second step of #120228.
Add `StructurallyRelateAliases` to allow instantiating infer vars with rigid aliases.
Change `instantiate_query_response` to be infallible in the new solver. This requires canonicalization to not hide any information used by the query, so weaken
universe compression. It also modifies `term_is_fully_unconstrained` to allow
region inference variables in a higher universe.
Prevent cycle in implied predicates computation
Makes #65913 from hang -> fail. I believe fail is the correct state for this test to remain for the long term.
Add newtypes for bool fields/params/return types
Fixed all the cases of this found with some simple searches for `*/ bool` and `bool /*`; probably many more
Stabilize `cfg_target_abi`
This stabilizes the `cfg` option called `target_abi`:
```rust
#[cfg(target_abi = "eabihf")]
```
Tracking issue: #80970fixes#78791resolves#80970
Add `#[rustc_no_mir_inline]` for standard library UB checks
should help with #121110 and also with #120848
Because the MIR inliner cannot know whether the checks are enabled or not, so inlining is an unnecessary compile time pessimization when debug assertions are disabled. LLVM knows whether they are enabled or not, so it can optimize accordingly without wasting time.
r? `@saethlin`
promotion: don't promote int::MIN / -1
Looks like I entirely forgot about this case when adding the div-by-zero check, which was supposed to ensure that we never promote operations that can fail...
Cc https://github.com/rust-lang/rust/issues/80619
This is a breaking change, so needs a crater run.
r? ``@oli-obk``
match lowering: Split off `test_candidates` into several functions and improve comments
The logic of `test_candidates` has three steps: pick a test, sort the candidates, and generate code for everything. So I split it off into three methods.
I also ended up reworking the comments that explain the algorithm. In particular I added detailed examples. I removed the digression about https://github.com/rust-lang/rust/issues/29740 because it's no longer relevant to how the code is structured today.
r? ``@matthewjasper``
Rollup of 7 pull requests
Successful merges:
- #121435 (Account for RPITIT in E0310 explicit lifetime constraint suggestion)
- #121490 (Rustdoc: include crate name in links for local primitives)
- #121520 (delay cloning of iterator items)
- #121522 (check that simd_insert/extract indices are in-bounds)
- #121531 (Ignore less tests in debug builds)
- #121539 (compiler/rustc_target/src/spec/base/apple/tests.rs: Avoid unnecessary large move)
- #121542 (update stdarch)
r? `@ghost`
`@rustbot` modify labels: rollup
compiler/rustc_target/src/spec/base/apple/tests.rs: Avoid unnecessary large move
Fixes:
$ MAGIC_EXTRA_RUSTFLAGS=-Zmove-size-limit=4096 ./x test compiler/rustc_target
error: moving 6216 bytes
--> compiler/rustc_target/src/spec/base/apple/tests.rs:17:19
|
17 | for target in all_sim_targets {
| ^^^^^^^^^^^^^^^ value moved from here
|
= note: The current maximum size is 4096, but it can be customized with the move_size_limit attribute: `#![move_size_limit = "..."]`
= note: `-D large-assignments` implied by `-D warnings`
= help: to override `-D warnings` add `#[allow(large_assignments)]`
Part of https://github.com/rust-lang/rust/issues/83518
The goal is to enable the lint by default at one point. There are many cases where it needs to be allowed. But here we don't need to allow it. We can simply fix it.
I don't want to enable the lint in the file, because I don't want the "default size threshold for when a value is OK to move" to be duplicated in many places. We'll want the threshold in one central place once we enable it by default.
Account for RPITIT in E0310 explicit lifetime constraint suggestion
When given
```rust
trait Original {
fn f() -> impl Fn();
}
trait Erased {
fn f(&self) -> Box<dyn Fn()>;
}
impl<T: Original> Erased for T {
fn f(&self) -> Box<dyn Fn()> {
Box::new(<T as Original>::f())
}
}
```
emit do not emit an invalid suggestion restricting the `Trait::{opaque}` type in a `where` clause:
```
error[E0310]: the associated type `<T as Original>::{opaque#0}` may not live long enough
--> $DIR/missing-static-bound-from-impl.rs:11:9
|
LL | Box::new(<T as Original>::f())
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| |
| the associated type `<T as Original>::{opaque#0}` must be valid for the static lifetime...
| ...so that the type `impl Fn()` will meet its required lifetime bounds
```
Partially address #119773. Ideally we'd suggest modifying `Erased::f` instead.
r? `@compiler-errors`
Fixes:
$ MAGIC_EXTRA_RUSTFLAGS=-Zmove-size-limit=4096 ./x test compiler/rustc_target
error: moving 6216 bytes
--> compiler/rustc_target/src/spec/base/apple/tests.rs:17:19
|
17 | for target in all_sim_targets {
| ^^^^^^^^^^^^^^^ value moved from here
|
= note: The current maximum size is 4096, but it can be customized with the move_size_limit attribute: `#![move_size_limit = "..."]`
= note: `-D large-assignments` implied by `-D warnings`
= help: to override `-D warnings` add `#[allow(large_assignments)]`
It looks like LLD will detect object files being either 32 or 64-bit
depending on any memory present. LLD will additionally reject 32-bit
objects during a 64-bit link. Previously metadata objects did not have
any memories in them which led LLD to conclude they were 32-bit objects
which broke 64-bit targets for wasm.
This commit fixes this by ensuring that for 64-bit targets there's a
memory object present to get LLD to detect it's a 64-bit target.
Additionally this commit moves away from a hand-crafted wasm encoder to
the `wasm-encoder` crate on crates.io as the complexity grows for the
generated object file.
Closes#121460
coverage: Rename `is_closure` to `is_hole`
Extracted from #121433, since I was having second thoughts about some of the other changes bundled in that PR, but these changes are still fine.
---
When refining covspans, we don't specifically care which ones represent closures; we just want to know which ones represent "holes" that should be carved out of other spans and then discarded.
(Closures are currently the only source of hole spans, but in the future we might want to also create hole spans for nested items and inactive `#[cfg(..)]` regions.)
``@rustbot`` label +A-code-coverage
Don't ICE on anonymous struct in enum variant
Fixes#121446
Computing `adt_def` for the anon struct calls `adt_def` on the parent to find its repr. If the parent is a non-item (e.g. an enum variant) we should have already emitted at least one error, so we just use the repr of the anonymous struct to avoid an ICE.
cc ``@frank-king``
mark `min_exhaustive_patterns` as complete
This is step 1 and 2 of my [proposal](https://github.com/rust-lang/rust/issues/119612#issuecomment-1918097361) to move `min_exhaustive_patterns` forward. The vast majority of in-tree use cases of `exhaustive_patterns` are covered by `min_exhaustive_patterns`. There are a few cases that still require `exhaustive_patterns` in tests and they're all behind references.
r? ``@ghost``
Thank you.
This is part of changing attributes to only_local. I hope get your opinion whether I should split into multiple PRs, or submit in one.
According to [try to not rely on attributes from extern crates](https://github.com/rust-lang/compiler-team/issues/505) and lcnr's guidance.
region unification: update universe of region vars
necessary for #119106. see inline comment for why this is necessary
r? `@compiler-errors` `@BoxyUwU`
When refining covspans, we don't specifically care which ones represent
closures; we just want to know which ones represent "holes" that should be
carved out of other spans and then discarded.
(Closures are currently the only source of hole spans, but in the future we
might want to also create hole spans for nested items and inactive `#[cfg(..)]`
regions.)
coverage: Use variable name `this` in `CoverageGraph::from_mir`
A tiny little improvement, extracted from #120013.
This makes it easier to see that we're manipulating the instance that is being constructed, and is a lot less verbose than `basic_coverage_blocks`.
Allow for a missing `adt_def` in `NamePrivacyVisitor`.
This was caused by 72b172bdf6 in #121206. That commit removed an early return from `analysis` when there are stashed errors. As a result, it's possible to reach privacy analysis when there are stashed errors, which means more code paths can be reached. One such code path was handled in that commit, where a `span_bug` was changed to a `span_delayed_bug`.
This commit handles another such code path uncovered by fuzzing, in much the same way.
Fixes#121455.
r? `@oli-obk`
Fix more #121208 fallout
#121208 converted lots of delayed bugs to bugs. Unsurprisingly, there were a few invalid conversion found via fuzzing.
r? `@lcnr`
fix generalizer unsoundness
I ended up getting confused while trying to flip the variances when flipping the order. Should be all right now.
This is only exploitable when generalizing if the `ambient_variance` of the relation is `Contravariant`. This can currently only be the case in the NLL generalizer which only rarely generalizes, causing us to miss this regression. Very much an issue with #121462 however.
When encountering `<&T as Clone>::clone(x)` because `T: Clone`, suggest `#[derive(Clone)]`
CC #40699.
```
warning: call to `.clone()` on a reference in this situation does nothing
--> $DIR/noop-method-call.rs:23:71
|
LL | let non_clone_type_ref_clone: &PlainType<u32> = non_clone_type_ref.clone();
| ^^^^^^^^
|
= note: the type `PlainType<u32>` does not implement `Clone`, so calling `clone` on `&PlainType<u32>` copies the reference, which does not do anything and can be removed
help: remove this redundant call
|
LL - let non_clone_type_ref_clone: &PlainType<u32> = non_clone_type_ref.clone();
LL + let non_clone_type_ref_clone: &PlainType<u32> = non_clone_type_ref;
|
help: if you meant to clone `PlainType<u32>`, implement `Clone` for it
|
LL + #[derive(Clone)]
LL | struct PlainType<T>(T);
|
```
Provide suggestions through `rustc_confusables` annotations
Help with common API confusion, like asking for `push` when the data structure really has `append`.
```
error[E0599]: no method named `size` found for struct `Vec<{integer}>` in the current scope
--> $DIR/rustc_confusables_std_cases.rs:17:7
|
LL | x.size();
| ^^^^
|
help: you might have meant to use `len`
|
LL | x.len();
| ~~~
help: there is a method with a similar name
|
LL | x.resize();
| ~~~~~~
```
Fix#59450 (we can open subsequent tickets for specific cases).
Fix#108437:
```
error[E0599]: `Option<{integer}>` is not an iterator
--> f101.rs:3:9
|
3 | opt.flat_map(|val| Some(val));
| ^^^^^^^^ `Option<{integer}>` is not an iterator
|
::: /home/gh-estebank/rust/library/core/src/option.rs:571:1
|
571 | pub enum Option<T> {
| ------------------ doesn't satisfy `Option<{integer}>: Iterator`
|
= note: the following trait bounds were not satisfied:
`Option<{integer}>: Iterator`
which is required by `&mut Option<{integer}>: Iterator`
help: you might have meant to use `and_then`
|
3 | opt.and_then(|val| Some(val));
| ~~~~~~~~
```
On type error of method call arguments, look at confusables for suggestion. Fix#87212:
```
error[E0308]: mismatched types
--> f101.rs:8:18
|
8 | stuff.append(Thing);
| ------ ^^^^^ expected `&mut Vec<Thing>`, found `Thing`
| |
| arguments to this method are incorrect
|
= note: expected mutable reference `&mut Vec<Thing>`
found struct `Thing`
note: method defined here
--> /home/gh-estebank/rust/library/alloc/src/vec/mod.rs:2025:12
|
2025 | pub fn append(&mut self, other: &mut Self) {
| ^^^^^^
help: you might have meant to use `push`
|
8 | stuff.push(Thing);
| ~~~~
```
This was caused by 72b172bdf6 in #121206. That commit removed an early
return from `analysis` when there are stashed errors. As a result, it's
possible to reach privacy analysis when there are stashed errors, which
means more code paths can be reached. One such code path was handled in
that commit, where a `span_bug` was changed to a `span_delayed_bug`.
This commit handles another such code path uncovered by fuzzing, in much
the same way.
Fixes#121455.
remove `sub_relations` from the `InferCtxt`
While doing so, I tried to remove the `delay_span_bug` in `rematch_impl` again, which lead me to discover another `freshen` bug, fixing that one in the second commit. See commit descriptions for the reasoning behind each change.
r? `@compiler-errors`
When given
```rust
trait Original {
fn f() -> impl Fn();
}
trait Erased {
fn f(&self) -> Box<dyn Fn()>;
}
impl<T: Original> Erased for T {
fn f(&self) -> Box<dyn Fn()> {
Box::new(<T as Original>::f())
}
}
```
avoid suggestion to restrict the `Trait::{opaque}` type in a `where` clause:
```
error[E0310]: the associated type `<T as Original>::{opaque#0}` may not live long enough
--> $DIR/missing-static-bound-from-impl.rs:11:9
|
LL | Box::new(<T as Original>::f())
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| |
| the associated type `<T as Original>::{opaque#0}` must be valid for the static lifetime...
| ...so that the type `impl Fn()` will meet its required lifetime bounds
```
CC #119773.
Do not provide a structured suggestion when the arguments don't match.
```
error[E0599]: no method named `test_mut` found for struct `Vec<{integer}>` in the current scope
--> $DIR/auto-ref-slice-plus-ref.rs:7:7
|
LL | a.test_mut();
| ^^^^^^^^
|
= help: items from traits can only be used if the trait is implemented and in scope
note: `MyIter` defines an item `test_mut`, perhaps you need to implement it
--> $DIR/auto-ref-slice-plus-ref.rs:14:1
|
LL | trait MyIter {
| ^^^^^^^^^^^^
help: there is a method `get_mut` with a similar name, but with different arguments
--> $SRC_DIR/core/src/slice/mod.rs:LL:COL
```
Consider methods beyond inherent ones when suggesting typos.
```
error[E0599]: no method named `owned` found for reference `&dyn Foo` in the current scope
--> $DIR/object-pointer-types.rs:11:7
|
LL | fn owned(self: Box<Self>);
| --------- the method might not be found because of this arbitrary self type
...
LL | x.owned();
| ^^^^^ help: there is a method with a similar name: `to_owned`
```
Fix#101013.
```
error[E0308]: mismatched types
--> $DIR/rustc_confusables_std_cases.rs:20:14
|
LL | x.append(42);
| ------ ^^ expected `&mut Vec<{integer}>`, found integer
| |
| arguments to this method are incorrect
|
= note: expected mutable reference `&mut Vec<{integer}>`
found type `{integer}`
note: method defined here
--> $SRC_DIR/alloc/src/vec/mod.rs:LL:COL
help: you might have meant to use `push`
|
LL | x.push(42);
| ~~~~
```
Help with common API confusion, like asking for `push` when the data structure really has `append`.
```
error[E0599]: no method named `size` found for struct `Vec<{integer}>` in the current scope
--> $DIR/rustc_confusables_std_cases.rs:17:7
|
LL | x.size();
| ^^^^
|
help: you might have meant to use `len`
|
LL | x.len();
| ~~~
help: there is a method with a similar name
|
LL | x.resize();
| ~~~~~~
```
#59450
match lowering: Introduce a `TestCase` enum to replace most matching on `PatKind`
Introduces `TestCase` that represents the specific outcome of a test. It complements `TestKind` which represents a test. In `MatchPair::new()` we select the appropriate `TestCase` for the pattern, and after that we almost never have to inspect the pattern directly during match lowering.
Together with https://github.com/rust-lang/rust/pull/120904, this makes `MatchPair` into a standalone abstraction that hides the details of `thir::Pat`. This will become even truer in the next PR where I make `TestCase` handle or patterns. This opens the door to a lot of future simplifications.
r? `@matthewjasper`
No need to `validate_alias_bound_self_from_param_env` in `assemble_alias_bound_candidates`
We already fully normalize the self type before we reach `assemble_alias_bound_candidates`, so there's no reason to double check that a projection is truly rigid by checking param-env bounds.
I think this is also blocked on us making sure to always normalize opaques: #120549.
r? lcnr
Without doing so we use the same candidate cache entry
for `?0: Trait<?1>` and `?0: Trait<?0>`. These goals are different
and we must not use the same entry for them.
we don't track them when canonicalizing or when freshening,
resulting in instable caching in the old solver, and issues when
instantiating query responses in the new one.
Make intrinsic fallback bodies cross-crate inlineable
This change was prompted by the stage1 compiler spending 4% of its time when compiling the polymorphic-recursion MIR opt test in `unlikely`.
Intrinsic fallback bodies like `unlikely` should always be inlined, it's very silly if they are not. To do this, we enable the fallback bodies to be cross-crate inlineable. Not that this matters for our workloads since the compiler never actually _uses_ the "fallback bodies", it just uses whatever was cfg(bootstrap)ped, so I've also added `#[inline]` to those.
See the comments for more information.
r? oli-obk
Improve codegen diagnostic handling
Clarify the workings of the temporary `Diagnostic` type used to send diagnostics from codegen threads to the main thread.
r? `@estebank`
intrinsics::simd: add missing functions, avoid UB-triggering fast-math
Turns out stdarch declares a bunch more SIMD intrinsics that are still missing from libcore.
I hope I got the docs and in particular the safety requirements right for these "unordered" and "nanless" intrinsics.
Many of these are unused even in stdarch, but they are implemented in the codegen backend, so we may as well list them here.
r? `@Amanieu`
Cc `@calebzulawski` `@workingjubilee`
PR #119097 made the decision to make all `IntoDiagnostic` impls generic,
because this allowed a bunch of nice cleanups. But four hand-written
impls were unintentionally overlooked. This commit makes them generic.
`Rustc::emit_diagnostic` reconstructs a diagnostic passed in from the
macro machinery. Currently it uses the type `DiagnosticBuilder<'_,
ErrorGuaranteed>`, which is incorrect, because the diagnostic might be a
warning. And if it is a warning, because of the `ErrorGuaranteed` we end
up calling into `emit_producing_error_guaranteed` and the assertion
within that function (correctly) fails because the level is not an error
level.
The fix is simple: change the type to `DiagnosticBuilder<'_, ()>`. Using
`()` works no matter what the diagnostic level is, and we don't need an
`ErrorGuaranteed` here.
The panic was reported in #120576.
- Make it more closely match `rustc_errors::Diagnostic`, by making the
field names match, and adding `children`, which requires adding
`rustc_codegen_ssa:🔙:write::Subdiagnostic`.
- Check that we aren't missing important info when converting
diagnostics.
- Add better comments.
- Tweak `rustc_errors::Diagnostic::replace_args` so that we don't need
to do any cloning when converting diagnostics.
First, introduce a typedef `DiagnosticArgMap`.
Second, make the `args` field public, and remove the `args` getter and
`replace_args` setter. These were necessary previously because the getter
had a `#[allow(rustc::potential_query_instability)]` attribute, but that
was removed in #120931 when the args were changed from `FxHashMap` to
`FxIndexMap`. (All the other `Diagnostic` fields are public.)
Unify dylib loading between proc macros and codegen backends
As bonus this makes the errors when failing to load a proc macro more informative to match the backend loading errors. In addition it makes it slightly easier to patch rustc to work on platforms that don't support dynamic linking like wasm.
never patterns: Fix liveness analysis in the presence of never patterns
There's a bunch of code that only looks at the first alternative of an or-pattern, under the assumption that all alternatives have the same set of bindings. This is true except for never pattern alternatives (e.g. `Ok(x) | Err(!)`), so we skip these. I expect there's other code with this problem, I'll have to check that later.
I don't have tests for this yet because mir lowering causes other issues; I'll have some in the next PR.
r? ``@compiler-errors``
coverage: Remove `pending_dups` from the span refiner
When extracting coverage spans from a function's MIR, we need to decide how to handle spans that are associated with more than one node (BCB) in the coverage control flow graph.
The existing code for managing those duplicate spans is very subtle and difficult to modify. But by eagerly deduplicating those extracted spans in a much simpler way, we can remove a massive chunk of complexity from the span refiner.
There is a tradeoff here, in that we no longer try to retain *all* nondominating BCBs that have the same span, only the last one in the (semi-arbitrary) dominance ordering. But in practice, this produces very little difference in our coverage tests, and the simplification is so significant that I think it's worthwhile.
``@rustbot`` label +A-code-coverage
Top level error handling
The interactions between the following things are surprisingly complicated:
- `emit_stashed_diagnostics`,
- `flush_delayed`,
- normal return vs `abort_if_errors`/`FatalError.raise()` unwinding in the call to the closure in `interface::run_compiler`.
This PR disentangles it all.
r? `@oli-obk`
rename ptr::invalid -> ptr::without_provenance
It has long bothered me that `ptr::invalid` returns a pointer that is actually valid for zero-sized memory accesses. In general, it doesn't even make sense to ask "is this pointer valid", you have to ask "is this pointer valid for a given memory access". We could say that a pointer is invalid if it is not valid for *any* memory access, but [the way this FCP is going](https://github.com/rust-lang/unsafe-code-guidelines/issues/472), it looks like *all* pointers will be valid for zero-sized memory accesses.
Two possible alternative names emerged as people's favorites:
1. Something involving `dangling`, in analogy to `NonNull::dangling`. To avoid inconsistency with the `NonNull` method, the address-taking method could be called `dangling_at(addr: usize) -> *const T`.
2. `without_provenance`, to be symmetric with the inverse operation `ptr.addr_without_provenance()` (currently still called `ptr.addr()` but probably going to be renamed)
I have no idea which one of these is better. I read [this comment](https://github.com/rust-lang/rust/pull/117658#issuecomment-1830934701) as expressing a slight preference for something like the second option, so I went for that. I'm happy to go with `dangling_at` as well.
Cc `@rust-lang/opsem`
Currently `emit_stashed_diagnostic` is called from four(!) different
places: `print_error_count`, `DiagCtxtInner::drop`, `abort_if_errors`,
and `compile_status`.
And `flush_delayed` is called from two different places:
`DiagCtxtInner::drop` and `Queries`.
This is pretty gross! Each one should really be called from a single
place, but there's a bunch of entanglements. This commit cleans up this
mess.
Specifically, it:
- Removes all the existing calls to `emit_stashed_diagnostic`, and adds
a single new call in `finish_diagnostics`.
- Removes the early `flush_delayed` call in `codegen_and_build_linker`,
replacing it with a simple early return if delayed bugs are present.
- Changes `DiagCtxtInner::drop` and `DiagCtxtInner::flush_delayed` so
they both assert that the stashed diagnostics are empty (i.e.
processed beforehand).
- Changes `interface::run_compiler` so that any errors emitted during
`finish_diagnostics` (i.e. late-emitted stashed diagnostics) are
counted and cannot be overlooked. This requires adding
`ErrorGuaranteed` return values to several functions.
- Removes the `stashed_err_count` call in `analysis`. This is possible
now that we don't have to worry about calling `flush_delayed` early
from `codegen_and_build_linker` when stashed diagnostics are pending.
- Changes the `span_bug` case in `handle_tuple_field_pattern_match` to a
`delayed_span_bug`, because it now can be reached due to the removal
of the `stashed_err_count` call in `analysis`.
- Slightly changes the expected output of three tests. If no errors are
emitted but there are delayed bugs, the error count is no longer
printed. This is because delayed bugs are now always printed after the
error count is printed (or not printed, if the error count is zero).
There is a lot going on in this commit. It's hard to break into smaller
pieces because the existing code is very tangled. It took me a long time
and a lot of effort to understand how the different pieces interact, and
I think the new code is a lot simpler and easier to understand.
Currently `has_errors` excludes lint errors. This commit changes it to
include lint errors.
The motivation for this is that for most places it doesn't matter
whether lint errors are included or not. But there are multiple places
where they must be includes, and only one place where they must not be
included. So it makes sense for `has_errors` to do the thing that fits
the most situations, and the new `has_errors_excluding_lint_errors`
method in the one exceptional place.
The same change is made for `err_count`. Annoyingly, this requires the
introduction of `err_count_excluding_lint_errs` for one place, to
preserve existing error printing behaviour. But I still think the change
is worthwhile overall.
make it possible for outside crates to inspect a mir::ConstValue with the interpreter
For MiniRust we need to convert MIR constant values into MiniRust constant values. However, it's not currently possible to get nice high-level access to the inerts of a `ConstValue`: we can access the raw contents (the allocation / `ScalarInt`), but if it is e.g. of enum type and we want to determine which variant is encoded, we are stuck. There's only `try_destructure_mir_constant_for_user_output` which is meant for diagnostics, so it doesn't fit.
The interpreter has all the APIs to traverse such a value, so we just need a way to get such a ConstValue into an interpreter instance. This adds the public functions necessary to make that happen.
Remove an `unchecked_error_guaranteed` call.
If we abort immediately after complaining about the obsolete `impl Trait for ..` syntax, then we avoid reaching HIR lowering. This means we can use `TyKind::Dummy` instead of `TyKind::Err`.
r? `@oli-obk`
Remove `diagnostic_builder.rs`
#120576 moved a big chunk of `DiagnosticBuilder`'s functionality out of `diagnostic_builder.rs` into `diagnostic.rs`, which left `DiagnosticBuilder` spread across the two files.
This PR fixes that messiness by merging what remains of `diagnostic_builder.rs` into `diagnostic.rs`.
This is part of https://github.com/rust-lang/compiler-team/issues/722.
r? `@davidtwco`
match lowering: test one or pattern at a time
This is a bit more opinionated than the previous PRs. On the face of it this is less efficient and more complex than before, but I personally found the loop that digs into `leaf_candidates` on each iteration very confusing. Instead this does "generate code for this or-pattern" then "generate further code for each branch if needed" in two steps.
Incidentally this way we don't _require_ or patterns to be sorted at the end. It's still an important optimization but I find it clearer to not rely on it for correctness.
r? `@matthewjasper`
Support async trait bounds in macros
r? fmease
This is similar to your work on const trait bounds. This theoretically regresses `impl async $ident:ident` in macros, but I doubt this is occurring in practice.
As bonus this makes the errors when failing to load a proc macro more
informative to match the backend loading errors. In addition it makes it
slightly easier to patch rustc to work on platforms that don't support
dynamic linking like wasm.
llvm-wrapper/ArchiveWrapper.cpp(70): warning C4305: 'argument': truncation from 'int' to 'bool'
while in llvm 12 signature was
static ErrorOr<std::unique_ptr<MemoryBuffer>> getFile(const Twine &Filename, int64_t FileSize = -1, bool RequiresNullTerminator = true, bool IsVolatile = false);
fed41342a8/llvm/include/llvm/Support/MemoryBuffer.h (L85-L87)
in llvm 13 and later it was changed to
static ErrorOr<std::unique_ptr<MemoryBuffer>> getFile(const Twine &Filename, bool IsText = false, bool RequiresNullTerminator = true, bool IsVolatile = false);
75e33f71c2/llvm/include/llvm/Support/MemoryBuffer.h (L86-L88)
so code was interpreted as MemoryBuffer::getFile(Path, /*IsText*/true, /*RequiresNullTerminator=*/false), but now will be MemoryBuffer::getFile(Path, /*IsText*/false, /*RequiresNullTerminator=*/false). How that worked before?
Add "algebraic" fast-math intrinsics, based on fast-math ops that cannot return poison
Setting all of LLVM's fast-math flags makes our fast-math intrinsics very dangerous, because some inputs are UB. This set of flags permits common algebraic transformations, but according to the [LangRef](https://llvm.org/docs/LangRef.html#fastmath), only the flags `nnan` (no nans) and `ninf` (no infs) can produce poison.
And this uses the algebraic float ops to fix https://github.com/rust-lang/rust/issues/120720
cc `@orlp`
Downgrade ambiguous_wide_pointer_comparisons suggestions to MaybeIncorrect
In certain cases like #121330, it is possible to have more than one suggestion from the `ambiguous_wide_pointer_comparisons` lint (which before this PR are `MachineApplicable`). When this gets passed to rustfix, rustfix makes *multiple* changes according to the suggestions which result in incorrect code.
This is a temporary workaround. The real long term solution to problems like these is to address <https://github.com/rust-lang/rust/issues/53934>.
This PR also includes a drive-by edit to the panic message emitted by compiletest because "ui" test suite now uses `//`@`` directives.
Fixes#121330.
Make --verbose imply -Z write-long-types-to-disk=no
When shortening the type it is necessary to take into account the `--verbose` flag, if it is activated, we must always show the entire type and not write it in a file.
Fixes: https://github.com/rust-lang/rust/issues/119130
Convert `delayed_bug`s to `bug`s.
I have a suspicion that quite a few delayed bug paths are impossible to reach, so I did an experiment.
I converted every `delayed_bug` to a `bug`, ran the full test suite, then converted back every `bug` that was hit. A surprising number were never hit.
This is too dangerous to merge. Increased coverage (fuzzing or a crater run) would likely hit more cases. But it might be useful for people to look at and think about which paths are genuinely unreachable.
r? `@ghost`
wasm: Store rlib metadata in wasm object files
The goal of this commit is to remove warnings using LLVM tip-of-tree `wasm-ld`. In llvm/llvm-project#78658 the `wasm-ld` LLD driver no longer looks at archive indices and instead looks at all the objects in archives. Previously `lib.rmeta` files were simply raw rustc metadata bytes, not wasm objects, meaning that `wasm-ld` would emit a warning indicating so.
WebAssembly targets previously passed `--fatal-warnings` to `wasm-ld` by default which meant that if Rust were to update to LLVM 18 then all wasm targets would not work. This immediate blocker was resolved in rust-lang/rust#120278 which removed `--fatal-warnings` which enabled a theoretical update to LLVM 18 for wasm targets. This current state is ok-enough for now because rustc squashes all linker output by default if it doesn't fail. This means, for example, that rustc squashes all the linker warnings coming out of `wasm-ld` about `lib.rmeta` files with LLVM 18. This again isn't a pressing issue because the information is all hidden, but it runs the risk of being annoying if another linker error were to happen and then the output would have all these unrelated warnings that couldn't be fixed.
Thus, this PR comes into the picture. The goal of this PR is to resolve these warnings by using the WebAssembly object file format on wasm targets instead of using raw rustc metadata. When I first implemented the rlib-in-objects scheme in #84449 I remember either concluding that `wasm-ld` would either include the metadata in the output or I thought we didn't have to do anything there at all. I think I was wrong on both counts as `wasm-ld` does not include the metadata in the final output unless the object is referenced and we do actually need to do something to resolve these warnings.
This PR updates the object file format containing rustc metadata on WebAssembly targets to be an actual WebAssembly file. To avoid bringing in any new dependencies I've opted to hand-code this encoding at this time. If the object gets more complicated though it'd probably be best to pull in `wasmparser` and `wasm-encoder`. For now though there's two adjacent functions reading/writing wasm.
The only caveat I know of with this is that if `wasm-ld` does indeed look at the object file then the metadata will be included in the final output. I believe the only thing that could cause that at this time is `--whole-archive` which I don't think is passed for rlibs. I would clarify that I'm not 100% certain about this, however.
If we abort immediately after complaining about the obsolete `impl Trait
for ..` syntax, then we avoid reaching HIR lowering. This means we can
use `TyKind::Dummy` instead of `TyKind::Err`.
match lowering: simplify empty candidate selection
In match lowering, `match_simplified_candidates` is tasked with removing candidates that are fully matched and linking them up properly. The code that does that was needlessly complicated; this PR simplifies it.
The overall change isn't big but I split it up into tiny commits to convince myself that I was correctly preserving behavior. The test changes are all due to the first commit. Let me know if you'd prefer me to split up the PR to make reviewing easier.
r? `@matthewjasper`
match lowering: eagerly simplify match pairs
This removes one important complication from match lowering. Before this, match pair simplification (which includes collecting bindings and type ascriptions) was intertwined with the whole match lowering algorithm.
I'm avoiding this by storing in each `MatchPair` the sub-`MatchPair`s that correspond to its subfields. This makes it possible to simplify everything (except or-patterns) in `Candidate::new()`.
This should open up further simplifications. It will also give us proper control over the order of bindings.
r? `@matthewjasper`
Because:
- `diagnostic_builder.rs` is small (282 lines),
- `Diagnostic` and `DiagnosticBuilder` are closely related types, and
- there's already an `impl DiagnosticBuilder` block in `diagnostic.rs`.
At the same time, reorder a few of things already in `diagnostic.rs`,
e.g. move `struct Diagnostic` just before `impl Diagnostic`.
This commit only moves code around. There are no functional changes.
I have a suspicion that quite a few delayed bug paths are impossible to
reach, so I did an experiment.
I converted every `delayed_bug` to a `bug`, ran the full test suite,
then converted back every `bug` that was hit. A surprising number were
never hit.
The next commit will convert some more back, based on human judgment.
Fix stray trait mismatch in `resolve_associated_item` for `AsyncFn`
Copy-paste error meant that we were calling `fn_trait_kind_from_def_id` instead of `async_fn_trait_kind_from_def_id`. But turns out we don't even need to do that, since we already matched the trait def id above.
Fixes#121306
r? oli-obk
Don't use raw parameter types in `find_builder_fn`
We shouldn't really ever be using `EarlyBinder::skip_binder` then performing type equality, since param types will never be equal to other types. When checking compatibility with the signature, we instead create some fresh args.
Fixes#121314
Don't ICE when hitting overflow limit in fulfillment loop in next solver
As the title says, let's not ICE when hitting the overflow limit in fulfill. On the other hand, we don't want to treat these as true errors, since it means that whether something is considered a true error or an ambiguity is dependent on overflow handling in the solver, which seems not worth it.
Now that we use the presence of true errors in fulfillment for implicit negative coherence, we especially don't want to tie together coherence and overflow.
I guess I could also drain these errors out of fulfillment and put them into some `ambiguities` storage so we could return them in `select_all_or_error` without having to re-process them every time we call `select_where_possible`. Let me know if that's desired.
r? lcnr
The goal of this commit is to remove warnings using LLVM tip-of-tree
`wasm-ld`. In llvm/llvm-project#78658 the `wasm-ld` LLD driver no longer
looks at archive indices and instead looks at all the objects in
archives. Previously `lib.rmeta` files were simply raw rustc metadata
bytes, not wasm objects, meaning that `wasm-ld` would emit a warning
indicating so.
WebAssembly targets previously passed `--fatal-warnings` to `wasm-ld` by
default which meant that if Rust were to update to LLVM 18 then all wasm
targets would not work. This immediate blocker was resolved in
rust-lang/rust#120278 which removed `--fatal-warnings` which enabled a
theoretical update to LLVM 18 for wasm targets. This current state is
ok-enough for now because rustc squashes all linker output by default if
it doesn't fail. This means, for example, that rustc squashes all the
linker warnings coming out of `wasm-ld` about `lib.rmeta` files with
LLVM 18. This again isn't a pressing issue because the information is
all hidden, but it runs the risk of being annoying if another linker
error were to happen and then the output would have all these unrelated
warnings that couldn't be fixed.
Thus, this PR comes into the picture. The goal of this PR is to resolve
these warnings by using the WebAssembly object file format on wasm
targets instead of using raw rustc metadata. When I first implemented
the rlib-in-objects scheme in #84449 I remember either concluding that
`wasm-ld` would either include the metadata in the output or I thought
we didn't have to do anything there at all. I think I was wrong on both
counts as `wasm-ld` does not include the metadata in the final output
unless the object is referenced and we do actually need to do something
to resolve these warnings.
This PR updates the object file format containing rustc metadata on
WebAssembly targets to be an actual WebAssembly file. This enables the
`wasm` feature of the `object` crate to be able to read the custom
section in the same manner as other platforms, but currently `object`
doesn't support writing wasm object files so a handwritten encoder is
used instead.
The only caveat I know of with this is that if `wasm-ld` does indeed
look at the object file then the metadata will be included in the final
output. I believe the only thing that could cause that at this time is
`--whole-archive` which I don't think is passed for rlibs. I would
clarify that I'm not 100% certain about this, however.
It is possible to have more than one valid suggestion, which when
applied together via rustfix causes the code to no longer compile.
This is a temporary workaround; the real long term solution to these
issues is to solve <https://github.com/rust-lang/rust/issues/53934>.
Trigger `unsafe_code` lint on invocations of `global_asm`
`unsafe_code` already warns about things that don't involve the `unsafe` keyword, e.g. `#[no_mangle]`. This makes it warn on `core::arch::global_asm` too.
Fixes#103078
Rename `ConstPropLint` to `KnownPanicsLint`
`OverflowLint` is a clearer name because it communicates what the lint does instead of the underlying mechanism it uses (const propagation) which should be of secondary concern.
`OverflowLint` isn't the most accurate name because the lint looks for other errors as well such as division by zero not just overflows, but I couldn't think of another equally succinct name.
As a part of this change. I've also added/updated some of the comments.
cc ```@RalfJung``` ```@oli-obk``` for visibility in case you go looking for the lint using the old name.
Edit:
Changed the name from `OverflowLint` to `KnownPanicsLint`
Overhaul `Diagnostic` and `DiagnosticBuilder`
Implements the first part of https://github.com/rust-lang/compiler-team/issues/722, which moves functionality and use away from `Diagnostic`, onto `DiagnosticBuilder`.
Likely follow-ups:
- Move things around, because this PR was written to minimize diff size, so some things end up in sub-optimal places. E.g. `DiagnosticBuilder` has impls in both `diagnostic.rs` and `diagnostic_builder.rs`.
- Rename `Diagnostic` as `DiagInner` and `DiagnosticBuilder` as `Diag`.
r? `@davidtwco`
Always evaluate free constants and statics, even if previous errors occurred
work towards https://github.com/rust-lang/rust/issues/79738
We will need to evaluate static items before the `definitions.freeze()` below, as we will start creating new `DefId`s (for nested allocations) within the `eval_static_initializer` query.
But even without that motivation, this is a good change. Hard errors should always be reported and not silenced if other errors happened earlier.
When shortening the type it is necessary to take into account the
`--verbose` flag, if it is activated, we must always show the entire
type and not write it in a file.
Fixes: https://github.com/rust-lang/rust/issues/119130
Rollup of 10 pull requests
Successful merges:
- #120716 (Change leak check and suspicious auto trait lint warning messages)
- #121195 (unstable-book: Separate testing and production sanitizers)
- #121205 (Merge `CompilerError::CompilationFailed` and `CompilerError::ICE`.)
- #121233 (Move the extra directives for `Mode::CoverageRun` into `iter_header`)
- #121256 (Allow AST and HIR visitors to return `ControlFlow`)
- #121307 (Drive-by `DUMMY_SP` -> `Span` and fmt changes)
- #121308 (Add regression test for #103369)
- #121310 (Remove an old hack for rustdoc)
- #121311 (Make `is_nonoverlapping` `#[inline]`)
- #121319 (return `ty::Error` when equating `ty::Error`)
r? `@ghost`
`@rustbot` modify labels: rollup
return `ty::Error` when equating `ty::Error`
This helps iron out a difference in diagnostics between `Sub` and `Equate` relations, which I'm currently trying to unify.
r? oli-obk
Drive-by `DUMMY_SP` -> `Span` and fmt changes
Noticed these while doing something else. There's no practical change, but it's preferable to use `DUMMY_SP` as little as possible, particularly when we have perfectlly useful `Span`s available.
Allow AST and HIR visitors to return `ControlFlow`
Alternative to #108598.
Since rust-lang/libs-team#187 was rejected, this implements our own version of the `Try` trait (`VisitorResult`) and the `try` macro (`try_visit`). Since this change still allows visitors to return `()`, no changes have been made to the existing ones. They can be done in a separate PR.
Merge `CompilerError::CompilationFailed` and `CompilerError::ICE`.
`CompilerError` has `CompilationFailed` and `ICE` variants, which seems reasonable at first. But the way it identifies them is flawed:
- If compilation errors out, i.e. `RunCompiler::run` returns an `Err`, it uses `CompilationFailed`, which is reasonable.
- If compilation panics with `FatalError`, it catches the panic and uses `ICE`. This is sometimes right, because ICEs do cause `FatalError` panics, but sometimes wrong, because certain compiler errors also cause `FatalError` panics. (The compiler/rustdoc/clippy/whatever just catches the `FatalError` with `catch_with_exit_code` in `main`.)
In other words, certain non-ICE compilation failures get miscategorized as ICEs. It's not possible to reliably distinguish the two cases, so this commit merges them. It also renames the combined variant as just `Failed`, to better match the existing `Interrupted` and `Skipped` variants.
Here is an example of a non-ICE failure that causes a `FatalError` panic, from `tests/ui/recursion_limit/issue-105700.rs`:
```
#![recursion_limit="4"]
#![invalid_attribute]
#![invalid_attribute]
#![invalid_attribute]
#![invalid_attribute]
#![invalid_attribute]
//~^ERROR recursion limit reached while expanding
fn main() {{}}
```
r? ``@spastorino``
Change leak check and suspicious auto trait lint warning messages
The leak check lint message "this was previously accepted by the compiler but is being phased out; it will become a hard error in a future release!" is misleading as some cases may not be phased out and could end being accepted. This is under discussion still.
The suspicious auto trait lint the change in behavior already happened, so the new message is probably more accurate.
r? `@lcnr`
Closes#93367
pattern_analysis: Move constructor selection logic to `PlaceInfo`
This is a small refactor PR. There was a dense bit of constructor-related logic in `compute_exhaustiveness_and_usefulness`. I'm moving it out into a `PlaceInfo` method to make it easier to follow both separately. I also have plans that will complicate it further so it's good that it's somewhat encapsulated.
r? `@compiler-errors`
Currently many diagnostic modifier methods are available on both
`Diagnostic` and `DiagnosticBuilder`. This commit removes most of them
from `Diagnostic`. To minimize the diff size, it keeps them within
`diagnostic.rs` but changes the surrounding `impl Diagnostic` block to
`impl DiagnosticBuilder`. (I intend to move things around later, to give
a more sensible code layout.)
`Diagnostic` keeps a few methods that it still needs, like `sub`,
`arg`, and `replace_args`.
The `forward!` macro, which defined two additional methods per call
(e.g. `note` and `with_note`), is replaced by the `with_fn!` macro,
which defines one additional method per call (e.g. `with_note`). It's
now also only used when necessary -- not all modifier methods currently
need a `with_*` form. (New ones can be easily added as necessary.)
All this also requires changing `trait AddToDiagnostic` so its methods
take `DiagnosticBuilder` instead of `Diagnostic`, which leads to many
mechanical changes. `SubdiagnosticMessageOp` gains a type parameter `G`.
There are three subdiagnostics -- `DelayedAtWithoutNewline`,
`DelayedAtWithNewline`, and `InvalidFlushedDelayedDiagnosticLevel` --
that are created within the diagnostics machinery and appended to
external diagnostics. These are handled at the `Diagnostic` level, which
means it's now hard to construct them via `derive(Diagnostic)`, so
instead we construct them by hand. This has no effect on what they look
like when printed.
There are lots of new `allow` markers for `untranslatable_diagnostics`
and `diagnostics_outside_of_impl`. This is because
`#[rustc_lint_diagnostics]` annotations were present on the `Diagnostic`
modifier methods, but missing from the `DiagnosticBuilder` modifier
methods. They're now present.
Fix `IPHONEOS_DEPLOYMENT_TARGET` on Mac Catalyst
Some of the target code invalidly assumed that the deployment target variable on Mac Catalyst is `MACOSX_DEPLOYMENT_TARGET`, which is wrong, Mac Catalyst uses the same environment variable as iOS.
Additionally, the deployment target was hardcoded to `14.0`, I've lowered this to `13.1` ([same default as Clang](d022f32c73/clang/lib/Driver/ToolChains/Darwin.cpp (L2038))), and made it properly load from the environment.
This shouldn't require any changes to the `cc` crate, as that uses `rustc --print=deployment-target` to get this information automatically.
CC `@BlackHoleFox`
r? `@rust-lang/macos`
deduplicate infer var instantiation
Having 3 separate implementations of one of the most subtle parts of our type system is not a good strategy if we want to maintain a sound type system ✨ while working on this I already found some subtle bugs in the existing code, so that's awesome 🎉 cc #121159
This was necessary as I am not confident in my nll changes in #119106, so I am first cleaning this up in a separate PR.
r? `@BoxyUwU`
This change was prompted by the stage1 compiler spending 4% of its time
when compiling the polymorphic-recursion MIR opt test in `unlikely`.
Intrinsic fallback bodies like `unlikely` should always be inlined, it's
very silly if they are not. To do this, we enable the fallback bodies to
be cross-crate inlineable. Not that this matters for our workloads since
the compiler never actually _uses_ the "fallback bodies", it just uses
whatever was cfg(bootstrap)ped, so I've also added `#[inline]` to those.
Noticed these while doing something else. There's no practical change, but it's preferable to use `DUMMY_SP` as little as possible, particularly when we have perfectlly useful `Span`s available.
some trait system cleanups
Always eagerly replace projections with infer vars if normalization is ambig. Unsure why we previously didn't do so, wasn't able to find an explanation in #90887. This adds some complexity to the trait system and is afaict unnecessary.
The second commit simplifies `pred_known_to_hold_modulo_regions`, afaict the optional `fulfill` isn't necessary anymore.
r? types cc `@jackh726`
Extend Level API
I need this API for https://github.com/rust-lang/rust-clippy/pull/12303: I have a nested `cfg` attribute (so a `MetaItem`) and I'd like to still be able to match against all possible kind of `Level`s.
There are lots of functions that modify a diagnostic. This can be via a
`&mut Diagnostic` or a `&mut DiagnosticBuilder`, because the latter type
wraps the former and impls `DerefMut`.
This commit converts all the `&mut Diagnostic` occurrences to `&mut
DiagnosticBuilder`. This is a step towards greatly simplifying
`Diagnostic`. Some of the relevant function are made generic, because
they deal with both errors and warnings. No function bodies are changed,
because all the modifier methods are available on both `Diagnostic` and
`DiagnosticBuilder`.
macro_rules: Preserve all metavariable spans in a global side table
This PR preserves spans of `tt` metavariables used to pass tokens to declarative macros.
Such metavariable spans can then be used in span combination operations like `Span::to` to improve all kinds of diagnostics.
Spans of non-`tt` metavariables are currently kept in nonterminal tokens, but the long term plan is remove all nonterminal tokens from rustc parser and rely on the proc macro model with invisible delimiters (#114647, #67062).
In particular, `NtIdent` nonterminal (corresponding to `ident` metavariables) becomes easy to remove when this PR lands (#119412 does it).
The metavariable spans are kept in a global side table keyed by `Span`s of original tokens.
The alternative to the side table is keeping them in `SpanData` instead, but the performance regressions would be large because any spans from tokens passed to declarative macros would stop being inline and would work through span interner instead, and the penalty would be paid even if we never use the metavar span for the given original span.
(But also see the comment on `fn maybe_use_metavar_location` describing the map collision issues with the side table approach.)
There are also other alternatives - keeping the metavar span in `Token` or `TokenTree`, but associating it with `Span` itsel is the most natural choice because metavar spans are used in span combining operations, and those operations are not necessarily tied to tokens.
Remove const_prop.rs
Removed const_prop.rs and moved its contents to const_prop_lint.rs and dataflow_const_prop.rs where they are used.
const_prop.rs does not actually do any const propagation any more. After #116012 all it contains is code that is used by const_prop_lint.rs and one macro that is used by dataflow_const_prop.rs. So it made sense to just move it to those two places and remove this file.
Add help to `hir_analysis_unrecognized_intrinsic_function`
To help remind forgetful people like me what step they forgot.
(If this just ICE'd, https://github.com/rust-lang/compiler-team/issues/620 style, the stack trace would point me here, but since there's a "nice" error that information is lost.)
Tracking import use types for more accurate redundant import checking
fixes#117448
By tracking import use types to check whether it is scope uses or the other situations like module-relative uses, we can do more accurate redundant import checking.
For example unnecessary imports in std::prelude that can be eliminated:
```rust
use std::option::Option::Some;//~ WARNING the item `Some` is imported redundantly
use std::option::Option::None; //~ WARNING the item `None` is imported redundantly
```
fixes#117448
For example unnecessary imports in std::prelude that can be eliminated:
```rust
use std::option::Option::Some;//~ WARNING the item `Some` is imported redundantly
use std::option::Option::None; //~ WARNING the item `None` is imported redundantly
```
Rollup of 7 pull requests
Successful merges:
- #120526 (rustdoc: Correctly handle long crate names on mobile)
- #121100 (Detect when method call on argument could be removed to fulfill failed trait bound)
- #121160 (rustdoc: fix and refactor HTML rendering a bit)
- #121198 (Add more checks for `unnamed_fields` during HIR analysis)
- #121218 (Fix missing trait impls for type in rustc docs)
- #121221 (AstConv: Refactor lowering of associated item bindings a bit)
- #121237 (Use better heuristic for printing Cargo specific diagnostics)
r? `@ghost`
`@rustbot` modify labels: rollup
AstConv: Refactor lowering of associated item bindings a bit
Split off from #119385 as discussed, namely the first two commits (modulo one `FIXME` getting turned into a `NOTE`).
The second commit removes `astconv::ConvertedBinding{,Kind}` in favor of `hir::TypeBinding{,Kind}`. The former was a — in my opinion — super useless intermediary. As you can tell from the diff, its removal shaves off some code. Furthermore, yeeting it will make it easier to implement the type resolution fixes in #119385.
Nothing in this PR should have any semantic effect.
r? `@compiler-errors`
<sub>**Addendum** as in #118668: What I call “associated item bindings” are commonly referred to as “type bindings” for historical reasons. Nowadays, “type bindings” include assoc type bindings, assoc const bindings and RTN (return type notation) which is why I prefer not to use this outdated term.</sub>
Add more checks for `unnamed_fields` during HIR analysis
Fixes#121151
I also found that we don't prevent enums here so
```rs
#[repr(C)]
#[derive(Debug)]
enum A {
#[default]
B,
C,
}
#[repr(C)]
#[derive(Debug)]
struct D {
_: A,
}
```
leads to an ICE on an `self.is_struct() || self.is_union()` assertion, so fixed that too.
Properly deal with weak alias types as self types of impls
Fixes#114216.
Fixes#116100.
Not super happy about the two ad hoc “normalization” implementations for weak alias types:
1. In `inherent_impls`: The “peeling”, normalization to [“WHNF”][whnf]: Semantically that's exactly what we want (neither proper normalization nor shallow normalization would be correct here). Basically a weak alias type is “nominal” (well...^^) if the WHNF is nominal. [#97974](https://github.com/rust-lang/rust/pull/97974) followed the same approach.
2. In `constrained_generic_params`: Generic parameters are constrained by a weak alias type if the corresp. “normalized” type constrains them (where we only normalize *weak* alias types not arbitrary ones). Weak alias types are injective if the corresp. “normalized” type is injective.
Both have ad hoc overflow detection mechanisms.
**Coherence** is handled in #117164.
r? `@oli-obk` or types
[whnf]: https://en.wikipedia.org/wiki/Lambda_calculus_definition#Weak_head_normal_form
The moment we get a candidate without guard, the return block becomes a
fresh block linked to nothing. So we can keep assigning a fresh block
every iteration to reuse the `next_prebinding` logic.
Fix `cfg(target_abi = "sim")` on `i386-apple-ios`
Since https://github.com/rust-lang/rust/issues/80970 is stabilizing, I went and had a look, and found that the result was wrong on `i386-apple-ios`.
r? rust-lang/macos
Use fulfillment in next trait solver coherence
Use fulfillment in the new trait solver's `impl_intersection_has_impossible_obligation` routine. This means that inference that falls out of processing other obligations can influence whether we can determine if an obligation is impossible to satisfy. See the committed test.
This should be completely sound, since evaluation and fulfillment both respect intercrate mode equally.
We run the risk of breaking coherence later if we were to change the rules of fulfillment and/or inference during coherence, but this is a problem which affects evaluation, as nested obligations from a trait goal are processed together and can influence each other in the same way.
r? lcnr
cc #114862
Also changed obligationctxt -> fulfillmentctxt because it feels kind of redundant to use an ocx here. I don't really care enough and can change it back if it really matters much.
errors: only eagerly translate subdiagnostics
Subdiagnostics don't need to be lazily translated, they can always be eagerly translated. Eager translation is slightly more complex as we need to have a `DiagCtxt` available to perform the translation, which involves slightly more threading of that context.
This slight increase in complexity should enable later simplifications - like passing `DiagCtxt` into `AddToDiagnostic` and moving Fluent messages into the diagnostic structs rather than having them in separate files (working on that was what led to this change).
r? ```@nnethercote```
coverage: Discard spans that fill the entire function body
While debugging some other coverage changes, I discovered a frustrating inconsistency that occurs in functions containing closures, if they end with an implicit `()` return instead of an explicit trailing-expression.
This turns out to have been caused by the corresponding node in MIR having a span that covers the entire function body. When preparing coverage spans, any span that fills the whole body tends to cause more harm than good, so this PR detects and discards those spans.
(This isn't the first time whole-body spans have caused problems; we also eliminated some of them in #118525.)
Add and use a simple extension trait derive macro in the compiler
Adds `#[extension]` to `rustc_macros` for implementing an extension trait. This expands an impl (with an optional visibility) into two parallel trait + impl definitions.
before:
```rust
pub trait Extension {
fn a();
}
impl Extension for () {
fn a() {}
}
```
to:
```rust
#[extension]
pub impl Extension for () {
fn a() {}
}
```
Opted to just implement it by hand because I couldn't figure if there was a "canonical" choice of extension trait macro in the ecosystem. It's really lightweight anyways, and can always be changed.
I'm interested in adding this because I'd like to later split up the large `TypeErrCtxtExt` traits into several different files. This should make it one step easier.
const_mut_refs: allow mutable pointers to statics
Fixes https://github.com/rust-lang/rust/issues/118447
Writing this PR became a bit messy, see [Zulip](https://rust-lang.zulipchat.com/#narrow/stream/146212-t-compiler.2Fconst-eval/topic/Statics.20pointing.20to.20interior.20mutable.20statics) for some of my journey.^^ Turns out there was a long-standing bug in our qualif logic that led to us incorrectly classifying certain places as "no interior mut" when they actually had interior mut. Due to that the `const_refs_to_cell` feature gate was required far less often than it otherwise would, e.g. in [this code](https://play.rust-lang.org/?version=nightly&mode=debug&edition=2021&gist=9e0c042c451b3d11d64dd6263679a164). Fixing this however would be a massive breaking change all over libcore and likely the wider ecosystem. So I also changed the const-checking logic to just not require the feature gate for the affected cases. While doing so I discovered a bunch of logic that is not explained and that I could not explain. However I think stabilizing some const-eval feature will make most of that logic inconsequential so I just added some FIXMEs and left it at that.
r? `@oli-obk`
This fixes the issue wherein the lint didn't fire for promoteds
in the case of SHL/SHR operators in non-optimized builds
and all arithmetic operators in optimized builds
