`-Znext-solver` caching
This PR has two major changes while also fixing multiple issues found via fuzzing.
The main optimization is the ability to not discard provisional cache entries when popping the highest cycle head the entry depends on. This fixes the hang in Fuchsia with `-Znext-solver=coherence`.
It also bails if the result of a fixpoint iteration is ambiguous, even without reaching a fixpoint. This is necessary to avoid exponential blowup if a coinductive cycle results in ambiguity, e.g. due to unknowable candidates in coherence.
Updating stack entries pretty much exclusively happens lazily now, so `fn check_invariants` ended up being mostly useless and I've removed it. See https://gist.github.com/lcnr/8de338fdb2685581e17727bbfab0622a for the invariants we would be able to assert with it.
For a general overview, see the in-process update of the relevant rustc-dev-guide chapter: https://hackmd.io/1ALkSjKlSCyQG-dVb_PUHw
r? ```@compiler-errors```
Rollup of 7 pull requests
Successful merges:
- #122884 (Optimize integer `pow` by removing the exit branch)
- #127857 (Allow to customize `// TODO:` comment for deprecated safe autofix)
- #129034 (Add `#[must_use]` attribute to `Coroutine` trait)
- #129049 (compiletest: Don't panic on unknown JSON-like output lines)
- #129050 (Emit a warning instead of an error if `--generate-link-to-definition` is used with other output formats than HTML)
- #129056 (Fix one usage of target triple in bootstrap)
- #129058 (Add mw back to review rotation)
r? `@ghost`
`@rustbot` modify labels: rollup
Remove a no-longer-true assert
Fixes https://github.com/rust-lang/rust/issues/129009
The assert was simply no longer true. I thought my test suite was thorough but I had not noticed these `let`-specific diagnostics codepaths.
r? `@compiler-errors`
Shrink `TyKind::FnPtr`.
By splitting the `FnSig` within `TyKind::FnPtr` into `FnSigTys` and `FnHeader`, which can be packed more efficiently. This reduces the size of the hot `TyKind` type from 32 bytes to 24 bytes on 64-bit platforms. This reduces peak memory usage by a few percent on some benchmarks. It also reduces cache misses and page faults similarly, though this doesn't translate to clear cycles or wall-time improvements on CI.
r? `@compiler-errors`
Allow to customize `// TODO:` comment for deprecated safe autofix
Relevant for the deprecation of `CommandExt::before_exit` in #125970.
Tracking:
- #124866
bootstrap: don't use rustflags for `--rustc-args`
r? `@onur-ozkan`
This is going to require a bit of context.
https://github.com/rust-lang/rust/pull/47558 has added `--rustc-args` to `./x test` to allow passing flags when building `compiletest` tests. It was made specifically because using `RUSTFLAGS` would rebuild the compiler/stdlib, which would in turn require the flag you want to build tests with to successfully bootstrap.
#113178 made the request that it also works for other tests and doctests. This is not trivial to support across the board for `library`/`compiler` unit-tests/doctests and across stages. This issue was closed in #113948 by using `RUSTFLAGS`, seemingly incorrectly since https://github.com/rust-lang/rust/pull/123489 fixed that part to make it work.
Unfortunately #123489/#113948 have regressed the goals of `--rustc-args`:
- now we can't use rustc args that don't bootstrap, to run the UI tests: we can't test incomplete features. The new trait solver doesn't bootstrap, in-progress borrowck/polonius changes don't bootstrap, some other features are similarly incomplete, etc.
- using the flag now rebuilds everything from scratch: stage0 stdlib, stage1 compiler, stage1 stdlib. You don't need to re-do all this to compile UI tests, you only need the latter to run stdlib tests with a new flag, etc. This happens for contributors, but also on CI today. (Not to mention that in doing that it will rebuild things with flags that are not meant to be used, e.g. stdlib cfgs that don't exist in the compiler; or you could also imagine that this silently enables flags that were not meant to be enabled in this way).
Since then, bd71c71ea0 has started using it to test a stdlib feature, relying on the fact that it now rebuilds everything. So #125011 also regressed CI times more than necessary because it rebuilds everything instead of just stage 1 stdlib.
It's not easy for me to know how to properly fix#113178 in bootstrap, but #113948/#123489 are not it since they regress the initial intent. I'd think bootstrap would have to know from the list of test targets that are passed that the `library` or `compiler` paths that are passed could require rebuilding these crates with different rustflags, probably also depending on stages. Therefore I would not be able to fix it, and will just try in this PR to unregress the situation to unblock the initial use-case.
It seems miri now also uses `./x miri --rustc-args` in this incorrect meaning to rebuild the `library` paths they support to run with the new args. I've not made any bootstrap changes related to `./x miri` in this PR, so `--rustc-args` wouldn't work there anymore. I'd assume this would need to use rustflags again but I don't know how to make that work properly in bootstrap, hence opening as draft, so you can tell me how to do that. I assume we don't want to break their use-case again now that it exists, even though there are ways to use `./x test` to do exactly that.
`RUSTFLAGS_NOT_BOOTSTRAP=flag ./x test library/std` is a way to run unit tests with a new flag without rebuilding everything, while with #123489 there is no way anymore to run tests with a flag that doesn't bootstrap.
---
edit: after review, this PR:
- renames `./x test --rustc-args` to `./x test --compiletest-rustc-args` as it only applies there, and cannot use rustflags for this purpose.
- fixes the regression that using these args rebuilt everything from scratch
- speeds up some CI jobs via the above point
- removes `./x miri --rustc-args` as only library tests are supported, needs to rebuild libstd, and `./x miri --compiletest-rustc-args` wouldn't work since compiletests are not supported.
Refactor `powerpc64` call ABI handling
As the [specification](https://openpowerfoundation.org/specifications/64bitelfabi/) for the ELFv2 ABI states that returned aggregates are returned like arguments as long as they are at most two doublewords, I've merged the `classify_arg` and `classify_ret` functions to reduce code duplication. The only functional change is to fix#128579: the `classify_ret` function was incorrectly handling aggregates where `bits > 64 && bits < 128`. I've used the aggregate handling implementation from `classify_arg` which doesn't have this issue.
`@awilfox` could you test this on `powerpc64-unknown-linux-musl`? I'm only able to cross-test on `powerpc64-unknown-linux-gnu` and `powerpc64le-unknown-linux-gnu` locally at the moment, and as a tier 3 target `powerpc64-unknown-linux-musl` has zero CI coverage.
Fixes: #128579
miri: make vtable addresses not globally unique
Miri currently gives vtables a unique global address. That's not actually matching reality though. So this PR enables Miri to generate different addresses for the same type-trait pair.
To avoid generating an unbounded number of `AllocId` (and consuming unbounded amounts of memory), we use the "salt" technique that we also already use for giving constants non-unique addresses: the cache is keyed on a "salt" value n top of the actually relevant key, and Miri picks a random salt (currently in the range `0..16`) each time it needs to choose an `AllocId` for one of these globals -- that means we'll get up to 16 different addresses for each vtable. The salt scheme is integrated into the global allocation deduplication logic in `tcx`, and also used for functions and string literals. (So this also fixes the problem that casting the same function to a fn ptr over and over will consume unbounded memory.)
r? `@saethlin`
Fixes https://github.com/rust-lang/miri/issues/3737
Line numbers of multiply-inlined functions were fixed in #114643 by using a
single DISubprogram. That, however, triggered assertions because parameters
weren't deduplicated. The "solution" to that in #115417 was to insert a
DILexicalScope below the DISubprogram and parent all of the parameters to that
scope. That fixed the assertion, but debuggers (including gdb and lldb) don't
recognize variables that are not parented to the subprogram itself as parameters,
even if they are emitted with DW_TAG_formal_parameter.
Consider the program:
use std::env;
fn square(n: i32) -> i32 {
n * n
}
fn square_no_inline(n: i32) -> i32 {
n * n
}
fn main() {
let x = square(env::vars().count() as i32);
let y = square_no_inline(env::vars().count() as i32);
println!("{x} == {y}");
}
When making a release build with debug=2 and rustc 1.82.0-nightly (8b3870784 2024-08-07)
(gdb) r
Starting program: /ephemeral/tmp/target/release/tmp
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Breakpoint 1, tmp::square () at src/main.rs:5
5 n * n
(gdb) info args
No arguments.
(gdb) info locals
n = 31
(gdb) c
Continuing.
Breakpoint 2, tmp::square_no_inline (n=31) at src/main.rs:10
10 n * n
(gdb) info args
n = 31
(gdb) info locals
No locals.
This issue is particularly annoying because it removes arguments from stack traces.
The DWARF for the inlined function looks like this:
< 2><0x00002132 GOFF=0x00002132> DW_TAG_subprogram
DW_AT_linkage_name _ZN3tmp6square17hc507052ff3d2a488E
DW_AT_name square
DW_AT_decl_file 0x0000000f /ephemeral/tmp/src/main.rs
DW_AT_decl_line 0x00000004
DW_AT_type 0x00001a56<.debug_info+0x00001a56>
DW_AT_inline DW_INL_inlined
< 3><0x00002142 GOFF=0x00002142> DW_TAG_lexical_block
< 4><0x00002143 GOFF=0x00002143> DW_TAG_formal_parameter
DW_AT_name n
DW_AT_decl_file 0x0000000f /ephemeral/tmp/src/main.rs
DW_AT_decl_line 0x00000004
DW_AT_type 0x00001a56<.debug_info+0x00001a56>
< 4><0x0000214e GOFF=0x0000214e> DW_TAG_null
< 3><0x0000214f GOFF=0x0000214f> DW_TAG_null
That DW_TAG_lexical_block inhibits every debugger I've tested from recognizing
'n' as a parameter.
This patch removes the additional lexical scope. Parameters can be easily
deduplicated by a tuple of their scope and the argument index, at the trivial
cost of taking a Hash + Eq bound on DIScope.
Store `do_not_recommend`-ness in impl header
Alternative to #128674
It's less flexible, but also less invasive. Hopefully it's also performant. I'd recommend we think separately about the design for how to gate arbitrary diagnostic attributes moving forward.
Normalize struct tail properly for `dyn` ptr-to-ptr casting in new solver
Realized that the new solver didn't handle ptr-to-ptr casting correctly.
r? lcnr
Built on #128694
When encountering the following, mention the precense of conflicting crates:
```
error[E0599]: no function or associated item named `get_decoded` found for struct `HpkeConfig` in the current scope
--> src/main.rs:7:17
|
7 | HpkeConfig::get_decoded(&foo);
| ^^^^^^^^^^^ function or associated item not found in `HpkeConfig`
|
note: if you're trying to build a new `HpkeConfig`, consider using `HpkeConfig::new` which returns `HpkeConfig`
--> ~/.cargo/registry/src/index.crates.io-6f17d22bba15001f/janus_messages-0.3.1/src/lib.rs:908:5
|
908 | / pub fn new(
909 | | id: HpkeConfigId,
910 | | kem_id: HpkeKemId,
911 | | kdf_id: HpkeKdfId,
912 | | aead_id: HpkeAeadId,
913 | | public_key: HpkePublicKey,
914 | | ) -> HpkeConfig {
| |___________________^
note: there are multiple different versions of crate `prio` in the dependency graph
--> src/main.rs:1:5
|
1 | use prio::codec::Decode;
| ^^^^^^^^^^^^^^^^^^^ `prio` imported here doesn't correspond to the right crate version
|
::: ~/.cargo/registry/src/index.crates.io-6f17d22bba15001f/prio-0.9.1/src/codec.rs:35:1
|
35 | pub trait Decode: Sized {
| ----------------------- this is the trait that was imported
|
::: ~/.cargo/registry/src/index.crates.io-6f17d22bba15001f/prio-0.10.3/src/codec.rs:35:1
|
35 | pub trait Decode: Sized {
| ----------------------- this is the trait that is needed
...
43 | fn get_decoded(bytes: &[u8]) -> Result<Self, CodecError> {
| -------------------------------------------------------- the method is available for `HpkeConfig` here
help: there is an associated function `decode` with a similar name
|
7 | HpkeConfig::decode(&foo);
| ~~~~~~
```
When a type comes indirectly from one crate version but the imported trait comes from a separate crate version, the called method won't be found. We now show additional context:
```
error[E0599]: no method named `foo` found for struct `dep_2_reexport::Type` in the current scope
--> multiple-dep-versions.rs:8:10
|
8 | Type.foo();
| ^^^ method not found in `Type`
|
note: you have multiple different versions of crate `dependency` in your dependency graph
--> multiple-dep-versions.rs:4:32
|
4 | use dependency::{do_something, Trait};
| ^^^^^ `dependency` imported here doesn't correspond to the right crate version
|
::: ~/rust/build/x86_64-unknown-linux-gnu/test/run-make/crate-loading/rmake_out/multiple-dep-versions-1.rs:4:1
|
4 | pub trait Trait {
| --------------- this is the trait that was imported
|
::: ~/rust/build/x86_64-unknown-linux-gnu/test/run-make/crate-loading/rmake_out/multiple-dep-versions-2.rs:4:1
|
4 | pub trait Trait {
| --------------- this is the trait that is needed
5 | fn foo(&self);
| --- the method is available for `dep_2_reexport::Type` here
```
Fix bug in `Parser::look_ahead`.
The special case was failing to handle invisible delimiters on one path.
Fixes (but doesn't close until beta backported) #128895.
r? `@davidtwco`
const vector passed through to codegen
This allows constant vectors using a repr(simd) type to be propagated
through to the backend by reusing the functionality used to do a similar
thing for the simd_shuffle intrinsic
#118209
r? RalfJung
nontemporal_store: make sure that the intrinsic is truly just a hint
The `!nontemporal` flag for stores in LLVM *sounds* like it is just a hint, but actually, it is not -- at least on x86, non-temporal stores need very special treatment by the programmer or else the Rust memory model breaks down. LLVM still treats these stores as-if they were normal stores for optimizations, which is [highly dubious](https://github.com/llvm/llvm-project/issues/64521). Let's avoid all that dubiousness by making our own non-temporal stores be truly just a hint, which is possible on some targets (e.g. ARM). On all other targets, non-temporal stores become regular stores.
~~Blocked on https://github.com/rust-lang/stdarch/pull/1541 propagating to the rustc repo, to make sure the `_mm_stream` intrinsics are unaffected by this change.~~
Fixes https://github.com/rust-lang/rust/issues/114582
Cc `@Amanieu` `@workingjubilee`
this makes it easier to maintain and modify going forward.
There may be a small performance cost as we now need to
access the provisional cache *and* walk through the stack
to detect cycles. However, the provisional cache should be
mostly empty and the stack should only have a few elements
so the performance impact is likely minimal.
Given the complexity of the search graph maintainability
trumps linear performance improvements.
doing so requires overwriting global cache entries and
generally adds significant complexity to the solver. This is
also only ever done for root goals, so it feels easier to wrap
the `evaluate_canonical_goal` in an ordinary query if
necessary.
Link `std` statically in `rustc_driver`
This makes `rustc_driver` statically link to `std`. This is done by not passing `-C prefer-dynamic` when building `rustc_driver`. However building `rustc-main` won't work currently as it tries to dynamically link to both `rustc_driver` and `std` resulting in a crate graph with `std` duplicated. To fix that new command line option `-Z prefer_deps_of_dynamic` is added which prevents linking to a dylib if there's a static variant of it already statically linked into another dylib dependency.
The main motivation for this change is to enable `#[global_allocator]` to be used in `rustc_driver` allowing overriding the allocator used in rustc on all platforms.
---
Instead of adding `-Z prefer_deps_of_dynamic`, this PR is changed to crate opt-in to the linking change via the `rustc_private` feature instead, as that would be typically needed to link to `rustc_driver` anyway.
---
try-job: aarch64-apple
try-job: x86_64-msvc
try-job: i686-mingw
try-job: dist-x86_64-msvc
try-job: aarch64-gnu
rm `import.used`
By the way, `import_used_map` will only be used during `build_reduced_graph` and `finalize`, so it can be split from `Resolver` in the future.
r? ``@petrochenkov``
Use more slice patterns inside the compiler
Nothing super noteworthy. Just replacing the common 'fragile' pattern of "length check followed by indexing or unwrap" with slice patterns for legibility and 'robustness'.
r? ghost
Promote aarch64-apple-darwin to Tier 1
This promotes aarch64-apple-darwin to Tier 1 status as per rust-lang/rfcs#3671 and tracking issue #73908. Not sure what else is necessary for this to impement the aforementioned RFC, however I figured I'd try. I did read in previous issues and PRs that the necessary infrastructure was already in place for the aarch64-apple-darwin target, and the RFC mentions the same. So this should be all thats necessary in order for the target to be promoted.
This is a recreation of my previous PR because I accidentally did an incorrect git rebase which caused unnecessary changes to various commit SHAs. So this PR is a recreation of my previous PR without said stumble. My bad.
Preliminary cleanup of `WitnessPat` hoisting/printing
Follow-up to #128430.
The eventual goal is to remove `print::Pat` entirely, but in the course of working towards that I made so many small improvements that it seems wise to let those be reviewed/merged on their own first.
Best reviewed commit-by-commit, most of which should be pretty simple and straightforward.
r? ``@Nadrieril``
Fix `ElaborateBoxDerefs` on debug varinfo
Slightly simplifies the `ElaborateBoxDerefs` pass to fix cases where it was applying the wrong projections to debug var infos containing places that deref boxes.
From what I can tell[^1], we don't actually have any tests (or code anywhere, really) that exercise `debug x => *(...: Box<T>)`, and it's very difficult to trigger this in surface Rust, so I wrote a custom MIR test.
What happens is that the pass was turning `*(SOME_PLACE: Box<T>)` into `*(*((((SOME_PLACE).0: Unique<T>).0: NonNull<T>).0: *const T))` in debug var infos. In particular, notice the *double deref*, which was wrong.
This is the root cause of #128554, so this PR fixes#128554 as well. The reason that async closures was affected is because of the way that we compute the [`ByMove` body](https://github.com/rust-lang/rust/blob/master/compiler/rustc_mir_transform/src/coroutine/by_move_body.rs), which resulted in `*(...: Box<T>)` in debug var info. But this really has nothing to do with async closures.
[^1]: Validated by literally replacing the `if elem == PlaceElem::Deref && base_ty.is_box() { ... }` innards with a `panic!()`, which compiled all of stage2 without panicking.
1. Decouple them.
2. Make logic around `diagnostic_outside_of_impl`'s early exits simpler.
3. Make `untranslatable_diagnostic` run one loop instead of two
and not allocate an intermediate vec.
4. Overall, reduce the amount of code executed
when the lints do not end up firing.
Differentiate between methods and associated functions in diagnostics
Accurately refer to assoc fn without receiver as assoc fn instead of methods. Add `AssocItem::descr` method to centralize where we call methods and associated functions.
Stabilize `min_exhaustive_patterns`
## Stabilisation report
I propose we stabilize the [`min_exhaustive_patterns`](https://github.com/rust-lang/rust/issues/119612) language feature.
With this feature, patterns of empty types are considered unreachable when matched by-value. This allows:
```rust
enum Void {}
fn foo() -> Result<u32, Void>;
fn main() {
let Ok(x) = foo();
// also
match foo() {
Ok(x) => ...,
}
}
```
This is a subset of the long-unstable [`exhaustive_patterns`](https://github.com/rust-lang/rust/issues/51085) feature. That feature is blocked because omitting empty patterns is tricky when *not* matched by-value. This PR stabilizes the by-value case, which is not tricky.
The not-by-value cases (behind references, pointers, and unions) stay as they are today, e.g.
```rust
enum Void {}
fn foo() -> Result<u32, &Void>;
fn main() {
let Ok(x) = foo(); // ERROR: missing `Err(_)`
}
```
The consequence on existing code is some extra "unreachable pattern" warnings. This is fully backwards-compatible.
### Comparison with today's rust
This proposal only affects match checking of empty types (i.e. types with no valid values). Non-empty types behave the same with or without this feature. Note that everything below is phrased in terms of `match` but applies equallly to `if let` and other pattern-matching expressions.
To be precise, a visibly empty type is:
- an enum with no variants;
- the never type `!`;
- a struct with a *visible* field of a visibly empty type (and no #[non_exhaustive] annotation);
- a tuple where one of the types is visibly empty;
- en enum with all variants visibly empty (and no `#[non_exhaustive]` annotation);
- a `[T; N]` with `N != 0` and `T` visibly empty;
- all other types are nonempty.
(An extra change was proposed below: that we ignore #[non_exhaustive] for structs since adding fields cannot turn an empty struct into a non-empty one)
For normal types, exhaustiveness checking requires that we list all variants (or use a wildcard). For empty types it's more subtle: in some cases we require a `_` pattern even though there are no valid values that can match it. This is where the difference lies regarding this feature.
#### Today's rust
Under today's rust, a `_` is required for all empty types, except specifically: if the matched expression is of type `!` (the never type) or `EmptyEnum` (where `EmptyEnum` is an enum with no variants), then the `_` is not required.
```rust
let foo: Result<u32, !> = ...;
match foo {
Ok(x) => ...,
Err(_) => ..., // required
}
let foo: Result<u32, &!> = ...;
match foo {
Ok(x) => ...,
Err(_) => ..., // required
}
let foo: &! = ...;
match foo {
_ => ..., // required
}
fn blah(foo: (u32, !)) {
match foo {
_ => ..., // required
}
}
unsafe {
let ptr: *const ! = ...;
match *ptr {} // allowed
let ptr: *const (u32, !) = ...;
match *ptr {
(x, _) => { ... } // required
}
let ptr: *const Result<u32, !> = ...;
match *ptr {
Ok(x) => { ... }
Err(_) => { ... } // required
}
}
```
#### After this PR
After this PR, a pattern of an empty type can be omitted if (and only if):
- the match scrutinee expression has type `!` or `EmptyEnum` (like before);
- *or* the empty type is matched by value (that's the new behavior).
In all other cases, a `_` is required to match on an empty type.
```rust
let foo: Result<u32, !> = ...;
match foo {
Ok(x) => ..., // `Err` not required
}
let foo: Result<u32, &!> = ...;
match foo {
Ok(x) => ...,
Err(_) => ..., // required because `!` is under a dereference
}
let foo: &! = ...;
match foo {
_ => ..., // required because `!` is under a dereference
}
fn blah(foo: (u32, !)) {
match foo {} // allowed
}
unsafe {
let ptr: *const ! = ...;
match *ptr {} // allowed
let ptr: *const (u32, !) = ...;
match *ptr {
(x, _) => { ... } // required because the matched place is under a (pointer) dereference
}
let ptr: *const Result<u32, !> = ...;
match *ptr {
Ok(x) => { ... }
Err(_) => { ... } // required because the matched place is under a (pointer) dereference
}
}
```
### Documentation
The reference does not say anything specific about exhaustiveness checking, hence there is nothing to update there. The nomicon does, I opened https://github.com/rust-lang/nomicon/pull/445 to reflect the changes.
### Tests
The relevant tests are in `tests/ui/pattern/usefulness/empty-types.rs`.
### Unresolved Questions
None that I know of.
try-job: dist-aarch64-apple
Cache supertrait outlives of impl header for soundness check
This caches the results of computing the transitive supertraits of an impl and filtering it to its outlives obligations. This is purely an optimization to improve https://github.com/rust-lang/rust/pull/124336.
Stop unnecessarily taking GenericPredicates by `&self`
This results in overcapturing in edition 2024, and is unnecessary since `GenericPredicates: Copy`.
WF-check struct field types at construction site
Fixes#126272.
Fixes#127299.
Rustc of course already WF-checked the field types at the definition
site, but for error tainting of consts to work properly, there needs to
be an error emitted at the use site. Previously, with no use-site error,
we proceeded with CTFE and ran into ICEs since we are running code with
type errors.
Emitting use-site errors also brings struct-like constructors more in
line with fn-like constructors since they already emit use-site errors
for WF issues.
r? `@BoxyUwU`
Accurately refer to assoc fn without receiver as assoc fn instead of methods.
Add `AssocItem::descr` method to centralize where we call methods and associated functions.
Enable zstd for debug compression.
Set LLVM_ENABLE_ZSTD alongside LLVM_ENABLE_ZLIB so that --compress-debug-sections=zstd is an option.
See #120953
try-job: x86_64-gnu-tools
Subtree sync for rustc_codegen_cranelift
The main highlight this time is support for raw-dylib on Windows thanks to `@dpaoliello.` Compiling the ring crate for arm64 macOS has been fixed too.
r? `@ghost`
`@rustbot` label +A-codegen +A-cranelift +T-compiler
Ensure let stmt compound assignment removal suggestion respect codepoint boundaries
Previously we would try to issue a suggestion for `let x <op>= 1`, i.e.
a compound assignment within a `let` binding, to remove the `<op>`. The
suggestion code unfortunately incorrectly assumed that the `<op>` is an
exactly-1-byte ASCII character, but this assumption is incorrect because
we also recover Unicode-confusables like `➖=` as `-=`. In this example,
the suggestion code used a `+ BytePos(1)` to calculate the span of the
`<op>` codepoint that looks like `-` but the mult-byte Unicode
look-alike would cause the suggested removal span to be inside a
multi-byte codepoint boundary, triggering a codepoint boundary
assertion.
The fix is to use `SourceMap::start_point(token_span)` which properly accounts for codepoint boundaries.
Fixes#128845.
cc #128790
r? ````@fmease````
Use `SourceMap::end_point` instead of `- BytePos(1)` in arg removal suggestion
Previously, we tried to remove extra arg commas when providing extra arg removal suggestions. One of
the edge cases is having to account for an arg that has a closing delimiter `)` following it.
However, the previous suggestion code assumed that the delimiter is in fact exactly the 1-byte `)`
character. This assumption was proven incorrect, because we recover from Unicode-confusable
delimiters in the parser, which means that the ending delimiter could be a multi-byte codepoint
that looks *like* a `)`. Subtracing 1 byte could land us in the middle of a codepoint, triggering a
codepoint boundary assertion.
This is fixed by using `SourceMap::end_point` which properly accounts for codepoint boundaries.
Fixes#128717.
cc ````@fmease```` and #128790
use stable sort to sort multipart diagnostics
I think a stable sort should be used to sort the different parts of a multipart selection. The current unstable sort uses the text of the suggestion as a tie-breaker. That just doesn't seem right, and the order of the input is a better choice I think, because it gives the diagnostic author more control.
This came up when I was building a suggestion where
```rust
fn foo() {}
```
must be turned into an unsafe function, and an attribute must be added
```rust
#[target_feature(enable = "...")]
unsafe fn foo() {}
```
In this example, the two suggestions occur at the same position, but the order is extremely important: unsafe must come after the attribute. But the situation changes if there is a pub/pub(crate), and if the unsafe is already present. It just out that because of the suggestion text, there is no way for me to order the suggestions correctly.
This change probably should be tested, but are there tests of the diagnostics code itself in the tests?
r? ```@estebank```
Add `Steal::is_stolen()`
Writers of rustc drivers (such as myself) often encounter stealing issues. It is currently impossible to gracefully handle them. This PR adds a `Steal::is_stolen()` function for that purpose.
Set LLVM_ENABLE_ZSTD alongside LLVM_ENABLE_ZLIB so that --compress-debug-sections=zstd is an option.
Use static linking to avoid a new runtime dependency. Add an llvm.libzstd bootstrap option for LLVM
with zstd. Set it off by default except for the dist builder. Handle llvm-config --system-libs output
that contains static libraries.
Miscellaneous improvements to struct tail normalization
1. Make checks for foreign tails more accurate by normalizing the struct tail. I didn't write a test for this one.
2. Normalize when computing struct tail for `offset_of` for slice/str. This fixes the new solver only.
3. Normalizing when computing tails for disaligned reference check. This fixes both solvers.
r? lcnr
Previously we would try to issue a suggestion for `let x <op>= 1`, i.e.
a compound assignment within a `let` binding, to remove the `<op>`. The
suggestion code unfortunately incorrectly assumed that the `<op>` is an
exactly-1-byte ASCII character, but this assumption is incorrect because
we also recover Unicode-confusables like `➖=` as `-=`. In this example,
the suggestion code used a `+ BytePos(1)` to calculate the span of the
`<op>` codepoint that looks like `-` but the mult-byte Unicode
look-alike would cause the suggested removal span to be inside a
multi-byte codepoint boundary, triggering a codepoint boundary
assertion.
Issue: <https://github.com/rust-lang/rust/issues/128845>
Parser has error recovery for Unicode-confusables, which includes the
right parentheses `)`. If a multi-byte right parentheses look-alike
reaches the argument removal suggestion diagnostics, it would trigger an
assertion because the diagnostics used `- BytePos(1)` which can land
within a multi-byte codepoint.
This is fixed by using `SourceMap::end_point` to find the final right
delimiter codepoint, which correctly respects codepoint boundaries.
By splitting the `FnSig` within `TyKind::FnPtr` into `FnSigTys` and
`FnHeader`, which can be packed more efficiently. This reduces the size
of the hot `TyKind` type from 32 bytes to 24 bytes on 64-bit platforms.
This reduces peak memory usage by a few percent on some benchmarks. It
also reduces cache misses and page faults similarly, though this doesn't
translate to clear cycles or wall-time improvements on CI.
Don't inline tainted MIR bodies
Don't inline MIR bodies that are tainted, since they're not necessarily well-formed.
Fixes#128601 (I didn't add a new test, just copied one from the crashes, since they're the same root cause).
Fixes#122909.
Add comment that bors did not see pushed before it merged
In #128612, bors merged 470ada2de0 instead of 1e07c19.
This means it dropped a useful comment I added, and a stage rename that is more descriptive.
Don't implement `AsyncFn` for `FnDef`/`FnPtr` that wouldnt implement `Fn`
Due to unsafety, ABI, or the presence of target features, some `FnDef`/`FnPtr` types don't implement `Fn*`. Do the same for `AsyncFn*`.
Noticed this due to #128764, but this isn't really related to that ICE, which is fixed in #128792.
The existing code check for `where_bounds.is_empty()` twice when
it can be combined into one. Moreover, the refactored code reads
better and feels straightforward.
```
error[E0277]: the size for values of type `str` cannot be known at compilation time
--> $DIR/unsized-str-in-return-expr-arg-and-local.rs:15:9
|
LL | let x = *"";
| ^ doesn't have a size known at compile-time
|
= help: the trait `Sized` is not implemented for `str`
= note: all local variables must have a statically known size
= help: unsized locals are gated as an unstable feature
help: references are always `Sized`, even if they point to unsized data; consider not dereferencing the expression
|
LL - let x = *"";
LL + let x = "";
|
```
Update E0517 message to reflect RFC 2195.
E0517 occurs when a `#[repr(..)]` attribute is placed on an unsupported item. Currently, the explanation of the error implies that `#[repr(u*/i*)]` cannot be placed on fieldful enums, which is no longer the case since [RFC 2195](https://github.com/rust-lang/rfcs/pull/2195) was [stabilized](https://github.com/rust-lang/rust/issues/60553), which allows placing `#[repr(u*/i*)]` and/or `#[repr(C)]` on fieldful enums to produce a defined layout.
This PR doesn't (currently) add a description of the semantics of placing `#[repr(u*/i*)]` on a fieldful enum to the error explanation, it just removes the claims/implications that it is not allowed.
Make `validate_mir` ensure the final MIR for all bodies
A lot of the crashes tests use `-Zpolymorphize` or `-Zdump-mir` for their side effect of computing the `optimized_mir` for all bodies, which will uncover bugs with late MIR passes like the inliner. I don't like having all these tests depend on `-Zpolymorphize` (or other hacky ways) for no reason, so this PR extends the `-Zvalidate-mir` flag to ensure `optimized_mir`/`mir_for_ctfe` for all body owners during the analysis phase.
Two thoughts:
1. This could be moved later in the compilation pipeline I guess? I don't really think it matters, though.
1. This could alternatively be expressed using a new flag, though I don't necessarily see much value in separating these.
For example, #128171 could have used this flag, in the `tests/ui/polymorphization/inline-incorrect-early-bound.rs`.
r? mir
Some `const { }` asserts for #128200
The correctness of code in #128200 relies on an array being sorted (so that it can be used in binary search later), which is currently enforced with `// tidy-alphabetical` (and characters being written in `\u{XXXX}` form), as well as lack of duplicate entries with conflicting keys, which is not currently enforced.
This PR changes it to using a `const{ }` assertion (and also checks for duplicate entries). Sadly, we cannot use the recently-stabilized `is_sorted_by_key` here, because it is not const (but it would not allow us to check for uniqueness anyways). Instead, let's write a manual loop.
Alternative approach (perfect hash function): #128463
r? `@ghost`
Rollup of 7 pull requests
Successful merges:
- #128520 (Skip over args when determining if async-closure's inner coroutine consumes its upvars)
- #128552 (Emit an error for invalid use of the `#[no_sanitize]` attribute)
- #128691 (Update `compiler-builtins` to 0.1.117)
- #128702 (Add -Zmetrics-dir=PATH to save diagnostic metadata to disk)
- #128797 (Fuchsia Test Runner: enable ffx repository server)
- #128798 (refactor(rustc_expand::mbe): Don't require full ExtCtxt when not necessary)
- #128800 (Add tracking issue to core-pattern-type)
r? `@ghost`
`@rustbot` modify labels: rollup
refactor(rustc_expand::mbe): Don't require full ExtCtxt when not necessary
Refactor `mbe::diagnostics::failed_to_match_macro()` to not require a full `ExtCtxt`, but only a `&ParseSess`. It hard-required the `ExtCtxt` only for a call to `cx.trace_macros_diag()`, which we move instead to the only call-site of the function.
Note: This could be a potential change in observed behavior, because a call to `cx.trace_macros_diag()` now always happens after `failed_to_match_macro()` was called, where before it was only called at the end of the main return path of the function. But since `trace_macros_diag()` "flushes" out any not-yet-reported errors, it should be ok to call it for all paths, since there shouldn't be any on the non-main paths I think. However, I don't know the rest of the codebase well enough to say that with 100% confidence, but `tests/ui` still pass, which gives at least some confidence in the change.
Also concretize the return type from `Box<dyn MacResult>` to `(Span, ErrorGuaranteed)`, because this function will _always_ return an error, and never any other kind of result.
Was part of #128605 and #128747, but is a standalone refactoring.
r? ``@petrochenkov``
Emit an error for invalid use of the `#[no_sanitize]` attribute
fixes#128487.
Currently, the use of the `#[no_sanitize]` attribute for Mod, Impl,... is incorrectly permitted. This PR will correct this issue by generating errors, and I've also added some UI test cases for it.
Referenced #128458. As far as I know, the `#[no_sanitize]` attribute can only be used with functions, so I changed that part to `Fn` and `Method` using `check_applied_to_fn_or_method`. However, I couldn't find explicit documentation on this, so I could be mistaken...
Skip over args when determining if async-closure's inner coroutine consumes its upvars
#125306 implements a strategy for when we have an `async move ||` async-closure that is inferred to be `async FnOnce`, it will force the inner coroutine to also be `move`, since we cannot borrow any upvars from the parent async-closure (since `FnOnce` is not lending):
8e86c95671/compiler/rustc_hir_typeck/src/upvar.rs (L211-L229)
However, when this strategy was implemented, it reused the `ExprUseVisitor` data from visiting the whole coroutine, which includes additional statements due to `async`-specific argument desugaring:
8e86c95671/compiler/rustc_ast_lowering/src/item.rs (L1197-L1228)
Well, it turns out that we don't care about these argument desugaring parameters, because arguments to the async-closure are not the *async-closure*'s captures -- they exist for only one invocation of the closure, and they're always consumed by construction (see the argument desugaring above), so they will force the coroutine's inferred kind to `FnOnce`. (Unless they're `Copy`, because we never consider `Copy` types to be consumed):
8e86c95671/compiler/rustc_hir_typeck/src/expr_use_visitor.rs (L60-L66)
However, since we *were* visiting these arg exprs, this resulted in us too-aggressively applying `move` to the inner coroutine, resulting in regressions. For example, this PR fixes#128516. Fascinatingly, the note above about how we never consume `Copy` types is why this only regressed when the argument types weren't all `Copy`.
I tried to leave some comments inline to make this more clear :)
Only walk ribs to collect possibly shadowed params if we are adding params in our new rib
No need to collect params from parent ribs if we literally have no params to declare in this new rib.
Attempt to win back some of the perf in https://github.com/rust-lang/rust/pull/128357#issuecomment-2262677031.
Please review with whitespace *off*, the diff should be like 2 lines.
r? petrochenkov
Rollup of 8 pull requests
Successful merges:
- #128221 (Add implied target features to target_feature attribute)
- #128261 (impl `Default` for collection iterators that don't already have it)
- #128353 (Change generate-copyright to generate HTML, with cargo dependencies included)
- #128679 (codegen: better centralize function declaration attribute computation)
- #128732 (make `import.vis` is immutable)
- #128755 (Integrate crlf directly into related test file instead via of .gitattributes)
- #128772 (rustc_codegen_ssa: Set architecture for object crate for 32-bit SPARC)
- #128782 (unused_parens: do not lint against parens around &raw)
r? `@ghost`
`@rustbot` modify labels: rollup
unused_parens: do not lint against parens around &raw
Requested by `@tmandry` in https://github.com/rust-lang/rust/pull/127679: with `&raw` one somewhat regularly has to write code like `(&raw const (*myptr).field).method()`, so parentheses around the expression are often required. To avoid churn between adding and removing parentheses as method calls appear and disappear, the proposal was made to silence the lint for unnecessary parentheses around `&raw` expressions. This PR implements that.
rustc_codegen_ssa: Set architecture for object crate for 32-bit SPARC
The `object` crate was recently updated to recognize the 32-bit SPARC ELF targets `EM_SPARC` and `EM_SPARC32PLUS`, so the proper architecture for 32-bit SPARC can now be set in `rustc_codegen_ssa`.
r? nagisa
codegen: better centralize function declaration attribute computation
For some reason, the codegen backend has two functions that compute which attributes a function declaration gets: `apply_attrs_llfn` and `attributes::from_fn_attrs`. They are called in different places, on entirely different layers of abstraction.
To me the code seems cleaner if we centralize this entirely in `apply_attrs_llfn`, so that's what this PR does.
Disallow setting some built-in cfg via set the command-line
This PR disallow users from setting some built-in cfg via set the command-line in order to prevent incoherent state, eg. `windows` cfg active but target is Linux based.
This implements MCP https://github.com/rust-lang/compiler-team/issues/610, with the caveat that we disallow cfgs no matter if they make sense or not, since I don't think it's useful to allow users to set a cfg that will be set anyway. It also complicates the implementation.
------
The `explicit_builtin_cfgs_in_flags` lint detects builtin cfgs set via the `--cfg` flag.
*(deny-by-default)*
### Example
```text
rustc --cfg unix
```
```rust,ignore (needs command line option)
fn main() {}
```
This will produce:
```text
error: unexpected `--cfg unix` flag
|
= note: config `unix` is only supposed to be controlled by `--target`
= note: manually setting a built-in cfg can and does create incoherent behaviours
= note: `#[deny(explicit_builtin_cfgs_in_flags)]` on by default
```
### Explanation
Setting builtin cfgs can and does produce incoherent behaviour, it's better to the use the appropriate `rustc` flag that controls the config. For example setting the `windows` cfg but on Linux based target.
-----
r? `@petrochenkov`
cc `@jyn514`
try-job: aarch64-apple
try-job: test-various
try-job: armhf-gnu
try-job: x86_64-msvc
try-job: x86_64-mingw
try-job: i686-msvc
try-job: i686-mingw
try-job: x86_64-gnu-llvm-17
try-job: dist-various-1
Don't arbitrarily choose one upper bound for hidden captured region error message
You could argue that the error message is objectively worse, even though it's more accurate. I guess we could also add a note explaining like "cannot capture the intersection of two regions" or something, though I'm not sure if that is confusing due to being totally technical jargon.
This addresses the fact that #128752 says "add `+ 'b`" even though it does nothing to fix the issue. It doesn't fix the issue's root cause, though.
r? `@spastorino`
More information for fully-qualified suggestion when there are multiple impls
```
error[E0790]: cannot call associated function on trait without specifying the corresponding `impl` type
--> $DIR/E0283.rs:30:21
|
LL | fn create() -> u32;
| ------------------- `Coroutine::create` defined here
...
LL | let cont: u32 = Coroutine::create();
| ^^^^^^^^^^^^^^^^^^^ cannot call associated function of trait
|
help: use a fully-qualified path to a specific available implementation
|
LL | let cont: u32 = <Impl as Coroutine>::create();
| ++++++++ +
LL | let cont: u32 = <AnotherImpl as Coroutine>::create();
| +++++++++++++++ +
```
minor `effects` cleanups
* remove the fixme comment about not needing defaults because it turns out we do need defaults (if I made it None instead it would ice a bunch of tests)
* remove the part that special cased trait args when lowering them. This is now historical because effects doesn't add host args to traits anymore (we use associated types now)
Make create_dll_import_lib easier to implement
This will make it easier to implement raw-dylib support in cg_clif and cg_gcc. This PR doesn't yet include an create_dll_import_lib implementation for cg_clif as I need to correctly implement dllimport in cg_clif first before raw-dylib can work at all with cg_clif.
Required for https://github.com/rust-lang/rustc_codegen_cranelift/issues/1345
We can replace some tricky iterator-mutation code with a much simpler version
that uses `while let` to shrink a slice.
We also check whether a subpattern would be a wildcard _before_ hoisting it,
which will be very useful when trying to get rid of `print::PatKind` later.
Version 0.3.1 has added support for writing import libraries. Version
0.3.2 fixed creating archives containing members of import libraries.
Version 0.3.3 fixed building on big-endian systems.
The object crate was recently updated to recognize the 32-bit SPARC
ELF targets EM_SPARC and EM_SPARC32PLUS, so the proper architecture
for 32-bit SPARC can now be set in rustc_codegen_ssa.
Add `f16` and `f128` math functions
This adds intrinsics and math functions for `f16` and `f128` floating point types. Support is quite limited and some things are broken so tests don't run on many platforms, but this provides a starting point.
The two altered expectation messages both seem like improvements:
- `coerce-expect-unsized-ascribed.stderr` says you can go
`Box<char> -> Box<dyn Debug>`, which you can.
- `upcast_soundness_bug.stderr` used to say that you could go
`Box<dyn Trait<u8, u8>> -> Box<dyn Trait>`, which you can't,
because the type parameters are missing in the destination
and the only ones that work aren't what's needed.
Don't ICE when getting an input file name's stem fails
Fixes#128681
The file stem is only used as a user-friendly prefix on intermediary files. While nice to have, it's not the end of the world if it fails so there's no real reason to emit an error here. We can continue with a fixed name as we do when an anonymous string is used.
interpret: refactor function call handling to be better-abstracted
Add a new function `init_stack_frame` that pushes a stack frame and passes the arguments, and use that basically everywhere that the raw underlying `push_stack_frame` used to be called. This splits the previous monster function `eval_fn_call` into two parts: figuring out the MIR to call and the arguments to pass, and then actually setting up the stack frame.
Also re-organize the files a bit:
- The previous `terminator.rs` is split into a new `call.rs` with all the argument-passing logic, and the rest goes into `step.rs` where the other main dispatcher functions already live (in particular, `eval_statement`).
- All the stack frame handling from `eval_context.rs` is moved to a new `stack.rs`.
Pass the right `ParamEnv` to `might_permit_raw_init_strict`
Fixes#119620
`might_permit_raw_init_strict` currently passes an empty `ParamEnv` to the `InterpCx`, instead of the actual `ParamEnv` that was passed in to `check_validity_requirement` at callsite.
This leads to ICEs such as the linked issue where for `UnsafeCell<*mut T>` we initially get the layout with the right `ParamEnv` (which suceeds because it can prove that `T: Sized` and therefore `UnsafeCell<*mut T>` has a known layout) but then do the rest with an empty `ParamEnv` where `T: Sized` is not known to hold so getting the layout for `*mut T` later fails.
This runs into an assertion in other layout code where it's making the (valid) assumption that, when we already have a layout for a struct (`UnsafeCell<*mut T>`), getting the layout of one of its fields (`*mut T`) should also succeed, which wasn't the case here due to using the wrong `ParamEnv`.
So, this PR changes it to just use the same `ParamEnv` all the way throughout.
Fix ICE Caused by Incorrectly Delaying E0107
Fixes #128249
For the following code:
```rust
trait Foo<T> {}
impl Foo<T: Default> for u8 {}
```
#126054 added some logic to delay emitting E0107 as the names of associated type `T` in the impl header and generic parameter `T` in `trait Foo` match.
But it failed to ensure whether such unexpected associated type bounds are coming from a impl block header. This caused an ICE as the compiler was delaying E0107 for code like:
```rust
trait Trait<Type> {
type Type;
fn method(&self) -> impl Trait<Type: '_>;
}
```
because it assumed the associated type bound `Type: '_` is for the generic parameter `Type` in `trait Trait` since the names are same.
This PR adds a check to ensure that E0107 is delayed only in the context of impl block header.
When encountering an E0277, if the type and the trait both come from a crate with the same name but different crate number, we explain that there are multiple crate versions in the dependency tree.
If there's a type that fulfills the bound, and it has the same name as the passed in type and has the same crate name, we explain that the same type in two different versions of the same crate *are different*.
```
error[E0277]: the trait bound `Type: dependency::Trait` is not satisfied
--> src/main.rs:4:18
|
4 | do_something(Type);
| ------------ ^^^^ the trait `dependency::Trait` is not implemented for `Type`
| |
| required by a bound introduced by this call
|
help: you have multiple different versions of crate `dependency` in your dependency graph
--> src/main.rs:1:5
|
1 | use bar::do_something;
| ^^^ one version of crate `dependency` is used here, as a dependency of crate `bar`
2 | use dependency::Type;
| ^^^^^^^^^^ one version of crate `dependency` is used here, as a direct dependency of the current crate
note: two types coming from two different versions of the same crate are different types even if they look the same
--> /home/gh-estebank/crate_versions/baz-2/src/lib.rs:1:1
|
1 | pub struct Type;
| ^^^^^^^^^^^^^^^ this type doesn't implement the required trait
|
::: /home/gh-estebank/crate_versions/baz/src/lib.rs:1:1
|
1 | pub struct Type;
| ^^^^^^^^^^^^^^^ this type implements the required trait
2 | pub trait Trait {}
| --------------- this is the required trait
note: required by a bound in `bar::do_something`
--> /home/gh-estebank/crate_versions/baz/src/lib.rs:4:24
|
4 | pub fn do_something<X: Trait>(_: X) {}
| ^^^^^ required by this bound in `do_something`
```
Address #22750.
Instead of saying to "consider adding a `#[repr(C)]` or
`#[repr(transparent)]` attribute to this struct", we now tell users to
"Use `*const ffi::c_char` instead, and pass the value from
`CStr::as_ptr()`" when the type involved is a `CStr` or a `CString`.
Co-authored-by: Jieyou Xu <jieyouxu@outlook.com>
On short error format, append primary span label to message
The `error-format=short` output only displays the path, error code and main error message all in the same line. We now add the primary span label as well after the error message, to provide more context.
