Handle gracefully true/false in `cfg(target(..))` compact
This PR handles gracefully `true`/`false` in `cfg(target(..))` compact instead of ICE.
r? `@nnethercote`
Fixes#131759
Fix unnecessary nesting in run-make test output directories
Run-make tests were using `output_base_name` to determine their output directory, which results in a redundant subdirectory (e.g. `$build/test/run-make/<foo>/<foo>/`) because that method is intended to produce the name of an individual file.
The previous attempt to fix this double-nesting tried adding a special case in `output_base_dir`, which had the side-effect of breaking up-to-date checking for run-make tests, and had to be reverted in #131681.
The fix is simply to call `output_base_dir` directory, which gives the desired directory without any redundant part.
r? jieyouxu
Don't report bivariance error when nesting a struct with field errors into another struct
We currently have logic to avoid reporting lifetime bivariance ("lifetime parameter ... is never used") errors when a struct has field resolution errors. However, this doesn't apply transitively. This PR implements a simple visitor to do so.
This was reported [here](https://twitter.com/fasterthanlime/status/1846257921086165033) since a `derive(Deserialize, Serialize)` ends up generating helper structs which have bivariant lifetimes due to containing the offending struct (that's being derived on).
Fix most ui tests on emscripten target
To fix the linker errors, we need to set the output extension to `.js` instead of `.wasm`. Setting the output to a `.wasm` file puts Emscripten into standalone mode which is effectively a distinct target. We need to set the runner to be `node` as well.
This fixes most of the ui tests. I fixed 4 additional tests with simple problems:
- `intrinsics/intrinsic-alignment.rs` -- Two `#[cfg]` macros match for Emscripten so we got duplicate definition
- `structs-enums/rec-align-u64.rs` -- same problem
- `issues/issue-12699.rs` -- hangs so I disabled it
- `process/process-sigpipe.rs` -- Not expected to work on Emscripten so I disabled it
Resolves#131666.
There are 7 more failing tests. I'll try to investigate more and see if I can fix them or at least understand why they happen.
- abi/numbers-arithmetic/return-float.rs (problem with [wasm treatment of noncanonical floats](https://webassembly.github.io/spec/core/exec/numerics.html#nan-propagation)?)
- async-await/issue-60709.rs -- linker error related to memcpy. Possible Emscripten bug?
- backtrace/dylib-dep.rs -- Says "Not supported"
- backtrace/line-tables-only.rs -- Says "Not supported"
- no_std/no-std-unwind-binary.rs -- compiler says `error: lang item required, but not found: eh_catch_typeinfo`
- structs-enums/enum-rec/issue-17431-6.rs -- One of the two compiler errors is missing
- test-attrs/test-passed.rs
r?workingjubilee r?jieyouxu
Don't report `on_unimplemented` message for negative traits
Kinda useless change but it was affecting my ability to read error messages when experimenting with negative bounds.
stabilize `-Znext-solver=coherence` again
r? `@compiler-errors`
---
This PR stabilizes the use of the next generation trait solver in coherence checking by enabling `-Znext-solver=coherence` by default. More specifically its use in the *implicit negative overlap check*. The tracking issue for this is https://github.com/rust-lang/rust/issues/114862. Closes#114862.
This is a direct copy of #121848 which has been reverted due to a hang in `nalgebra`: #130056. This hang should have been fixed by #130617 and #130821. See the added section in the stabilization report containing user facing changes merged since the original FCP.
## Background
### The next generation trait solver
The new solver lives in [`rustc_trait_selection::solve`](https://github.com/rust-lang/rust/blob/master/compiler/rustc_trait_selection/src/solve/mod.rs) and is intended to replace the existing *evaluate*, *fulfill*, and *project* implementation. It also has a wider impact on the rest of the type system, for example by changing our approach to handling associated types.
For a more detailed explanation of the new trait solver, see the [rustc-dev-guide](https://rustc-dev-guide.rust-lang.org/solve/trait-solving.html). This does not stabilize the current behavior of the new trait solver, only the behavior impacting the implicit negative overlap check. There are many areas in the new solver which are not yet finalized. We are confident that their final design will not conflict with the user-facing behavior observable via coherence. More on that further down.
Please check out [the chapter](https://rustc-dev-guide.rust-lang.org/solve/significant-changes.html) summarizing the most significant changes between the existing and new implementations.
### Coherence and the implicit negative overlap check
Coherence checking detects any overlapping impls. Overlapping trait impls always error while overlapping inherent impls result in an error if they have methods with the same name. Coherence also results in an error if any other impls could exist, even if they are currently unknown. This affects impls which may get added to upstream crates in a backwards compatible way and impls from downstream crates.
Coherence failing to detect overlap is generally considered to be unsound, even if it is difficult to actually get runtime UB this way. It is quite easy to get ICEs due to bugs in coherence.
It currently consists of two checks:
The [orphan check] validates that impls do not overlap with other impls we do not know about: either because they may be defined in a sibling crate, or because an upstream crate is allowed to add it without being considered a breaking change.
The [overlap check] validates that impls do not overlap with other impls we know about. This is done as follows:
- Instantiate the generic parameters of both impls with inference variables
- Equate the `TraitRef`s of both impls. If it fails there is no overlap.
- [implicit negative]: Check whether any of the instantiated `where`-bounds of one of the impls definitely do not hold when using the constraints from the previous step. If a `where`-bound does not hold, there is no overlap.
- *explicit negative (still unstable, ignored going forward)*: Check whether the any negated `where`-bounds can be proven, e.g. a `&mut u32: Clone` bound definitely does not hold as an explicit `impl<T> !Clone for &mut T` exists.
The overlap check has to *prove that unifying the impls does not succeed*. This means that **incorrectly getting a type error during coherence is unsound** as it would allow impls to overlap: coherence has to be *complete*.
Completeness means that we never incorrectly error. This means that during coherence we must only add inference constraints if they are definitely necessary. During ordinary type checking [this does not hold](https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=01d93b592bd9036ac96071cbf1d624a9), so the trait solver has to behave differently, depending on whether we're in coherence or not.
The implicit negative check only considers goals to "definitely not hold" if they could not be implemented downstream, by a sibling, or upstream in a backwards compatible way. If the goal is is "unknowable" as it may get added in another crate, we add an ambiguous candidate: [source](bea5bebf3d/compiler/rustc_trait_selection/src/solve/assembly/mod.rs (L858-L883)).
[orphan check]: fd80c02c16/compiler/rustc_trait_selection/src/traits/coherence.rs (L566-L579)
[overlap check]: fd80c02c16/compiler/rustc_trait_selection/src/traits/coherence.rs (L92-L98)
[implicit negative]: fd80c02c16/compiler/rustc_trait_selection/src/traits/coherence.rs (L223-L281)
## Motivation
Replacing the existing solver in coherence fixes soundness bugs by removing sources of incompleteness in the type system. The new solver separately strengthens coherence, resulting in more impls being disjoint and passing the coherence check. The concrete changes will be elaborated further down. We believe the stabilization to reduce the likelihood of future bugs in coherence as the new implementation is easier to understand and reason about.
It allows us to remove the support for coherence and implicit-negative reasoning in the old solver, allowing us to remove some code and simplifying the old trait solver. We will only remove the old solver support once this stabilization has reached stable to make sure we're able to quickly revert in case any unexpected issues are detected before then.
Stabilizing the use of the next-generation trait solver expresses our confidence that its current behavior is intended and our work towards enabling its use everywhere will not require any breaking changes to the areas used by coherence checking. We are also confident that we will be able to replace the existing solver everywhere, as maintaining two separate systems adds a significant maintainance burden.
## User-facing impact and reasoning
### Breakage due to improved handling of associated types
The new solver fixes multiple issues related to associated types. As these issues caused coherence to consider more types distinct, fixing them results in more overlap errors. This is therefore a breaking change.
#### Structurally relating aliases containing bound vars
Fixes https://github.com/rust-lang/rust/issues/102048. In the existing solver relating ambiguous projections containing bound variables is structural. This is *incomplete* and allows overlapping impls. These was mostly not exploitable as the same issue also caused impls to not apply when trying to use them. The new solver defers alias-relating to a nested goal, fixing this issue:
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
trait Trait {}
trait Project {
type Assoc<'a>;
}
impl Project for u32 {
type Assoc<'a> = &'a u32;
}
// Eagerly normalizing `<?infer as Project>::Assoc<'a>` is ambiguous,
// so the old solver ended up structurally relating
//
// (?infer, for<'a> fn(<?infer as Project>::Assoc<'a>))
//
// with
//
// ((u32, fn(&'a u32)))
//
// Equating `&'a u32` with `<u32 as Project>::Assoc<'a>` failed, even
// though these types are equal modulo normalization.
impl<T: Project> Trait for (T, for<'a> fn(<T as Project>::Assoc<'a>)) {}
impl<'a> Trait for (u32, fn(&'a u32)) {}
//[next]~^ ERROR conflicting implementations of trait `Trait` for type `(u32, for<'a> fn(&'a u32))`
```
A crater run did not discover any breakage due to this change.
#### Unknowable candidates for higher ranked trait goals
This avoids an unsoundness by attempting to normalize in `trait_ref_is_knowable`, fixing https://github.com/rust-lang/rust/issues/114061. This is a side-effect of supporting lazy normalization, as that forces us to attempt to normalize when checking whether a `TraitRef` is knowable: [source](47dd709bed/compiler/rustc_trait_selection/src/solve/assembly/mod.rs (L754-L764)).
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
trait IsUnit {}
impl IsUnit for () {}
pub trait WithAssoc<'a> {
type Assoc;
}
// We considered `for<'a> <T as WithAssoc<'a>>::Assoc: IsUnit`
// to be knowable, even though the projection is ambiguous.
pub trait Trait {}
impl<T> Trait for T
where
T: 'static,
for<'a> T: WithAssoc<'a>,
for<'a> <T as WithAssoc<'a>>::Assoc: IsUnit,
{
}
impl<T> Trait for Box<T> {}
//[next]~^ ERROR conflicting implementations of trait `Trait`
```
The two impls of `Trait` overlap given the following downstream crate:
```rust
use dep::*;
struct Local;
impl WithAssoc<'_> for Box<Local> {
type Assoc = ();
}
```
There a similar coherence unsoundness caused by our handling of aliases which is fixed separately in https://github.com/rust-lang/rust/pull/117164.
This change breaks the [`derive-visitor`](https://crates.io/crates/derive-visitor) crate. I have opened an issue in that repo: nikis05/derive-visitor#16.
### Evaluating goals to a fixpoint and applying inference constraints
In the old implementation of the implicit-negative check, each obligation is [checked separately without applying its inference constraints](bea5bebf3d/compiler/rustc_trait_selection/src/traits/coherence.rs (L323-L338)). The new solver instead [uses a `FulfillmentCtxt`](bea5bebf3d/compiler/rustc_trait_selection/src/traits/coherence.rs (L315-L321)) for this, which evaluates all obligations in a loop until there's no further inference progress.
This is necessary for backwards compatibility as we do not eagerly normalize with the new solver, resulting in constraints from normalization to only get applied by evaluating a separate obligation. This also allows more code to compile:
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
trait Mirror {
type Assoc;
}
impl<T> Mirror for T {
type Assoc = T;
}
trait Foo {}
trait Bar {}
// The self type starts out as `?0` but is constrained to `()`
// due to the where-clause below. Because `(): Bar` is known to
// not hold, we can prove the impls disjoint.
impl<T> Foo for T where (): Mirror<Assoc = T> {}
//[current]~^ ERROR conflicting implementations of trait `Foo` for type `()`
impl<T> Foo for T where T: Bar {}
fn main() {}
```
The old solver does not run nested goals to a fixpoint in evaluation. The new solver does do so, strengthening inference and improving the overlap check:
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
trait Foo {}
impl<T> Foo for (u8, T, T) {}
trait NotU8 {}
trait Bar {}
impl<T, U: NotU8> Bar for (T, T, U) {}
trait NeedsFixpoint {}
impl<T: Foo + Bar> NeedsFixpoint for T {}
impl NeedsFixpoint for (u8, u8, u8) {}
trait Overlap {}
impl<T: NeedsFixpoint> Overlap for T {}
impl<T, U: NotU8, V> Overlap for (T, U, V) {}
//[current]~^ ERROR conflicting implementations of trait `Foo`
```
### Breakage due to removal of incomplete candidate preference
Fixes#107887. In the old solver we incompletely prefer the builtin trait object impl over user defined impls. This can break inference guidance, inferring `?x` in `dyn Trait<u32>: Trait<?x>` to `u32`, even if an explicit impl of `Trait<u64>` also exists.
This caused coherence to incorrectly allow overlapping impls, resulting in ICEs and a theoretical unsoundness. See https://github.com/rust-lang/rust/issues/107887#issuecomment-1997261676. This compiles on stable but results in an overlap error with `-Znext-solver=coherence`:
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
struct W<T: ?Sized>(*const T);
trait Trait<T: ?Sized> {
type Assoc;
}
// This would trigger the check for overlap between automatic and custom impl.
// They actually don't overlap so an impl like this should remain possible
// forever.
//
// impl Trait<u64> for dyn Trait<u32> {}
trait Indirect {}
impl Indirect for dyn Trait<u32, Assoc = ()> {}
impl<T: Indirect + ?Sized> Trait<u64> for T {
type Assoc = ();
}
// Incomplete impl where `dyn Trait<u32>: Trait<_>` does not hold, but
// `dyn Trait<u32>: Trait<u64>` does.
trait EvaluateHack<U: ?Sized> {}
impl<T: ?Sized, U: ?Sized> EvaluateHack<W<U>> for T
where
T: Trait<U, Assoc = ()>, // incompletely constrains `_` to `u32`
U: IsU64,
T: Trait<U, Assoc = ()>, // incompletely constrains `_` to `u32`
{
}
trait IsU64 {}
impl IsU64 for u64 {}
trait Overlap<U: ?Sized> {
type Assoc: Default;
}
impl<T: ?Sized + EvaluateHack<W<U>>, U: ?Sized> Overlap<U> for T {
type Assoc = Box<u32>;
}
impl<U: ?Sized> Overlap<U> for dyn Trait<u32, Assoc = ()> {
//[next]~^ ERROR conflicting implementations of trait `Overlap<_>`
type Assoc = usize;
}
```
### Considering region outlives bounds in the `leak_check`
For details on the `leak_check`, see the FCP proposal #119820.[^leak_check]
[^leak_check]: which should get moved to the dev-guide :3
In both coherence and during candidate selection, the `leak_check` relies on the region constraints added in `evaluate`. It therefore currently does not register outlives obligations: [source](ccb1415eac/compiler/rustc_trait_selection/src/traits/select/mod.rs (L792-L810)). This was likely done as a performance optimization without considering its impact on the `leak_check`. This is the case as in the old solver, *evaluatation* and *fulfillment* are split, with evaluation being responsible for candidate selection and fulfillment actually registering all the constraints.
This split does not exist with the new solver. The `leak_check` can therefore eagerly detect errors caused by region outlives obligations. This improves both coherence itself and candidate selection:
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
trait LeakErr<'a, 'b> {}
// Using this impl adds an `'b: 'a` bound which results
// in a higher-ranked region error. This bound has been
// previously ignored but is now considered.
impl<'a, 'b: 'a> LeakErr<'a, 'b> for () {}
trait NoOverlapDir<'a> {}
impl<'a, T: for<'b> LeakErr<'a, 'b>> NoOverlapDir<'a> for T {}
impl<'a> NoOverlapDir<'a> for () {}
//[current]~^ ERROR conflicting implementations of trait `NoOverlapDir<'_>`
// --------------------------------------
// necessary to avoid coherence unknowable candidates
struct W<T>(T);
trait GuidesSelection<'a, U> {}
impl<'a, T: for<'b> LeakErr<'a, 'b>> GuidesSelection<'a, W<u32>> for T {}
impl<'a, T> GuidesSelection<'a, W<u8>> for T {}
trait NotImplementedByU8 {}
trait NoOverlapInd<'a, U> {}
impl<'a, T: GuidesSelection<'a, W<U>>, U> NoOverlapInd<'a, U> for T {}
impl<'a, U: NotImplementedByU8> NoOverlapInd<'a, U> for () {}
//[current]~^ conflicting implementations of trait `NoOverlapInd<'_, _>`
```
### Removal of `fn match_fresh_trait_refs`
The old solver tries to [eagerly detect unbounded recursion](b14fd2359f/compiler/rustc_trait_selection/src/traits/select/mod.rs (L1196-L1211)), forcing the affected goals to be ambiguous. This check is only an approximation and has not been added to the new solver.
The check is not necessary in the new solver and it would be problematic for caching. As it depends on all goals currently on the stack, using a global cache entry would have to always make sure that doing so does not circumvent this check.
This changes some goals to error - or succeed - instead of failing with ambiguity. This allows more code to compile:
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
// Need to use this local wrapper for the impls to be fully
// knowable as unknowable candidate result in ambiguity.
struct Local<T>(T);
trait Trait<U> {}
// This impl does not hold, but is ambiguous in the old
// solver due to its overflow approximation.
impl<U> Trait<U> for Local<u32> where Local<u16>: Trait<U> {}
// This impl holds.
impl Trait<Local<()>> for Local<u8> {}
// In the old solver, `Local<?t>: Trait<Local<?u>>` is ambiguous,
// resulting in `Local<?u>: NoImpl`, also being ambiguous.
//
// In the new solver the first impl does not apply, constraining
// `?u` to `Local<()>`, causing `Local<()>: NoImpl` to error.
trait Indirect<T> {}
impl<T, U> Indirect<U> for T
where
T: Trait<U>,
U: NoImpl
{}
// Not implemented for `Local<()>`
trait NoImpl {}
impl NoImpl for Local<u8> {}
impl NoImpl for Local<u16> {}
// `Local<?t>: Indirect<Local<?u>>` cannot hold, so
// these impls do not overlap.
trait NoOverlap<U> {}
impl<T: Indirect<U>, U> NoOverlap<U> for T {}
impl<T, U> NoOverlap<Local<U>> for Local<T> {}
//~^ ERROR conflicting implementations of trait `NoOverlap<Local<_>>`
```
### Non-fatal overflow
The old solver immediately emits a fatal error when hitting the recursion limit. The new solver instead returns overflow. This both allows more code to compile and is results in performance and potential future compatability issues.
Non-fatal overflow is generally desirable. With fatal overflow, changing the order in which we evaluate nested goals easily causes breakage if we have goal which errors and one which overflows. It is also required to prevent breakage due to the removal of `fn match_fresh_trait_refs`, e.g. [in `typenum`](https://github.com/rust-lang/trait-system-refactor-initiative/issues/73).
#### Enabling more code to compile
In the below example, the old solver first tried to prove an overflowing goal, resulting in a fatal error. The new solver instead returns ambiguity due to overflow for that goal, causing the implicit negative overlap check to succeed as `Box<u32>: NotImplemented` does not hold.
```rust
// revisions: current next
//[next] compile-flags: -Znext-solver=coherence
//[current] ERROR overflow evaluating the requirement
trait Indirect<T> {}
impl<T: Overflow<()>> Indirect<T> for () {}
trait Overflow<U> {}
impl<T, U> Overflow<U> for Box<T>
where
U: Indirect<Box<Box<T>>>,
{}
trait NotImplemented {}
trait Trait<U> {}
impl<T, U> Trait<U> for T
where
// T: NotImplemented, // causes old solver to succeed
U: Indirect<T>,
T: NotImplemented,
{}
impl Trait<()> for Box<u32> {}
```
#### Avoiding hangs with non-fatal overflow
Simply returning ambiguity when reaching the recursion limit can very easily result in hangs, e.g.
```rust
trait Recur {}
impl<T, U> Recur for ((T, U), (U, T))
where
(T, U): Recur,
(U, T): Recur,
{}
trait NotImplemented {}
impl<T: NotImplemented> Recur for T {}
```
This can happen quite frequently as it's easy to have exponential blowup due to multiple nested goals at each step. As the trait solver is depth-first, this immediately caused a fatal overflow error in the old solver. In the new solver we have to handle the whole proof tree instead, which can very easily hang.
To avoid this we restrict the recursion depth after hitting the recursion limit for the first time. We also **ignore all inference constraints from goals resulting in overflow**. This is mostly backwards compatible as any overflow in the old solver resulted in a fatal error.
### sidenote about normalization
We return ambiguous nested goals of `NormalizesTo` goals to the caller and ignore their impact when computing the `Certainty` of the current goal. See the [normalization chapter](https://rustc-dev-guide.rust-lang.org/solve/normalization.html) for more details.This means we apply constraints resulting from other nested goals and from equating the impl header when normalizing, even if a nested goal results in overflow. This is necessary to avoid breaking the following example:
```rust
trait Trait {
type Assoc;
}
struct W<T: ?Sized>(*mut T);
impl<T: ?Sized> Trait for W<W<T>>
where
W<T>: Trait,
{
type Assoc = ();
}
// `W<?t>: Trait<Assoc = u32>` does not hold as
// `Assoc` gets normalized to `()`. However, proving
// the where-bounds of the impl results in overflow.
//
// For this to continue to compile we must not discard
// constraints from normalizing associated types.
trait NoOverlap {}
impl<T: Trait<Assoc = u32>> NoOverlap for T {}
impl<T: ?Sized> NoOverlap for W<T> {}
```
#### Future compatability concerns
Non-fatal overflow results in some unfortunate future compatability concerns. Changing the approach to avoid more hangs by more strongly penalizing overflow can cause breakage as we either drop constraints or ignore candidates necessary to successfully compile. Weakening the overflow penalities instead allows more code to compile and strengthens inference while potentially causing more code to hang.
While the current approach is not perfect, we believe it to be good enough. We believe it to apply the necessary inference constraints to avoid breakage and expect there to not be any desirable patterns broken by our current penalities. Similarly we believe the current constraints to avoid most accidental hangs. Ignoring constraints of overflowing goals is especially useful, as it may allow major future optimizations to our overflow handling. See [this summary](https://hackmd.io/ATf4hN0NRY-w2LIVgeFsVg) and the linked documents in case you want to know more.
### changes to performance
In general, trait solving during coherence checking is not significant for performance. Enabling the next-generation trait solver in coherence does not impact our compile time benchmarks. We are still unable to compile the benchmark suite when fully enabling the new trait solver.
There are rare cases where the new solver has significantly worse performance due to non-fatal overflow, its reliance on fixpoint algorithms and the removal of the `fn match_fresh_trait_refs` approximation. We encountered such issues in [`typenum`](https://crates.io/crates/typenum) and believe it should be [pretty much as bad as it can get](https://github.com/rust-lang/trait-system-refactor-initiative/issues/73).
Due to an improved structure and far better caching, we believe that there is a lot of room for improvement and that the new solver will outperform the existing implementation in nearly all cases, sometimes significantly. We have not yet spent any time micro-optimizing the implementation and have many unimplemented major improvements, such as fast-paths for trivial goals.
### Unstable features
#### Unsupported unstable features
The new solver currently does not support all unstable features, most notably `#![feature(generic_const_exprs)]`, `#![feature(associated_const_equality)]` and `#![feature(adt_const_params)]` are not yet fully supported in the new solver. We are confident that supporting them is possible, but did not consider this to be a priority. This stabilization introduces new ICE when using these features in impl headers.
#### fixes to `#![feature(specialization)]`
- fixes#105782
- fixes#118987
#### fixes to `#![feature(type_alias_impl_trait)]`
- fixes#119272
- https://github.com/rust-lang/rust/issues/105787#issuecomment-1750112388
- fixes#124207
### Important changes since the original FCP
https://github.com/rust-lang/rust/pull/127574 changes the coherence unknowable candidate to only apply if all the super trait bounds may hold. This allows more code to compile and fixes a regression in `pyella`
https://github.com/rust-lang/rust/pull/130617 bails with ambiguity if the query response would contain too many non-region inference variables. This should only be triggered in case the result contains a lot of ambiguous aliases in which case further constraining the goal should resolve this.
https://github.com/rust-lang/rust/pull/130821 adds caching to a lot of type folders, which is necessary to handle exponentially large types and handles the hang in `nalgebra` together with #130617.
## This does not stabilize the whole solver
While this stabilizes the use of the new solver in coherence checking, there are many parts of the solver which will remain fully unstable. We may still adapt these areas while working towards stabilizing the new solver everywhere. We are confident that we are able to do so without negatively impacting coherence.
### goals with a non-empty `ParamEnv`
Coherence always uses an empty environment. We therefore do not depend on the behavior of `AliasBound` and `ParamEnv` candidates. We only stabilizes the behavior of user-defined and builtin implementations of traits. There are still many open questions there.
### opaque types in the defining scope
The handling of opaque types - `impl Trait` - in both the new and old solver is still not fully figured out. Luckily this can be ignored for now. While opaque types are reachable during coherence checking by using `impl_trait_in_associated_types`, the behavior during coherence is separate and self-contained. The old and new solver fully agree here.
### normalization is hard
This stabilizes that we equate associated types involving bound variables using deferred-alias-equality. We also stop eagerly normalizing in coherence, which should not have any user-facing impact.
We do not stabilize the normalization behavior outside of coherence, e.g. we currently deeply normalize all types during writeback with the new solver. This may change going forward
### how to replace `select` from the old solver
We sometimes depend on getting a single `impl` for a given trait bound, e.g. when resolving a concrete method for codegen/CTFE. We do not depend on this during coherence, so the exact approach here can still be freely changed going forward.
## Acknowledgements
This work would not have been possible without `@compiler-errors.` He implemented large chunks of the solver himself but also and did a lot of testing and experimentation, eagerly discovering multiple issues which had a significant impact on our approach. `@BoxyUwU` has also done some amazing work on the solver. Thank you for the endless hours of discussion resulting in the current approach. Especially the way aliases are handled has gone through multiple revisions to get to its current state.
There were also many contributions from - and discussions with - other members of the community and the rest of `@rust-lang/types.` This solver builds upon previous improvements to the compiler, as well as lessons learned from `chalk` and `a-mir-formality`. Getting to this point would not have been possible without that and I am incredibly thankful to everyone involved. See the [list of relevant PRs](https://github.com/rust-lang/rust/pulls?q=is%3Apr+is%3Amerged+label%3AWG-trait-system-refactor+-label%3Arollup+closed%3A%3C2024-03-22+).
To fix the linker errors, we need to set the output extension to `.js` instead
of `.wasm`. Setting the output to a `.wasm` file puts Emscripten into standalone
mode which is effectively a distinct target. We need to set the runner to be
`node` as well.
This fixes most of the ui tests. I fixed a few more tests with simple problems:
- `intrinsics/intrinsic-alignment.rs` and `structs-enums/rec-align-u64.rs` --
Two `#[cfg]` macros match for Emscripten so we got a duplicate definition of
`mod m`.
- `issues/issue-12699.rs` -- Seems to hang so I disabled it
- `process/process-sigpipe.rs` -- Not expected to work on Emscripten so I
disabled it
Rollup of 9 pull requests
Successful merges:
- #122670 (Fix bug where `option_env!` would return `None` when env var is present but not valid Unicode)
- #131095 (Use environment variables instead of command line arguments for merged doctests)
- #131339 (Expand set_ptr_value / with_metadata_of docs)
- #131652 (Move polarity into `PolyTraitRef` rather than storing it on the side)
- #131675 (Update lint message for ABI not supported)
- #131681 (Fix up-to-date checking for run-make tests)
- #131702 (Suppress import errors for traits that couldve applied for method lookup error)
- #131703 (Resolved python deprecation warning in publish_toolstate.py)
- #131710 (Remove `'apostrophes'` from `rustc_parse_format`)
r? `@ghost`
`@rustbot` modify labels: rollup
The issue was, that the disassembled label was placed one instruction
further down than expected. Therefore the test annotations check, that
the label is placed above the loop-contents (writing the one value, then
writing the other one).
This commit introduces a minimal `![no_core]`-test case running on AVR,
that contains the MCWE mentioned in [129301]. The test case itself does
not have any assertions yet, but it shows the minimal set an language
items necessary within the test case.
[129301]: https://github.com/rust-lang/rust/issues/129301#issuecomment-2301399472
Some float methods are now `const fn` under the `const_float_methods` feature gate.
In order to support `min`, `max`, `abs` and `copysign`, the implementation of some intrinsics had to be moved from Miri to rustc_const_eval.
Add `&pin (mut|const) T` type position sugar
This adds parser support for `&pin mut T` and `&pin const T` references. These are desugared to `Pin<&mut T>` and `Pin<&T>` in the AST lowering phases.
This PR currently includes #130526 since that one is in the commit queue. Only the most recent commits (bd450027eb4a94b814a7dd9c0fa29102e6361149 and following) are new.
Tracking:
- #130494
r? `@compiler-errors`
Remove `'apostrophes'` from `rustc_parse_format`
The rest of the compiler uses \`grave accents\`, while `rustc_parse_format` uses \`'apostrophes.'\`
Also makes the crate compile as a stand-alone:
```
~/rust/compiler/rustc_parse_format $ cargo check
Compiling rustc_index_macros v0.0.0 (/home/lieselotte/rust/compiler/rustc_index_macros)
error[E0277]: `syn::Lit` doesn't implement `Debug`
--> compiler/rustc_index_macros/src/newtype.rs:52:57
|
52 | panic!("Specified multiple max: {old:?}");
| ^^^^^^^ `syn::Lit` cannot be formatted using `{:?}` because it doesn't implement `Debug`
|
= help: the trait `Debug` is not implemented for `syn::Lit`
= note: this error originates in the macro `$crate::const_format_args` which comes from the expansion of the macro `panic` (in Nightly builds, run with -Z macro-backtrace for more info)
error[E0277]: `syn::Lit` doesn't implement `Debug`
--> compiler/rustc_index_macros/src/newtype.rs:64:74
|
64 | panic!("Specified multiple debug format options: {old:?}");
| ^^^^^^^ `syn::Lit` cannot be formatted using `{:?}` because it doesn't implement `Debug`
|
= help: the trait `Debug` is not implemented for `syn::Lit`
= note: this error originates in the macro `$crate::const_format_args` which comes from the expansion of the macro `panic` (in Nightly builds, run with -Z macro-backtrace for more info)
For more information about this error, try `rustc --explain E0277`.
error: could not compile `rustc_index_macros` (lib) due to 2 previous errors
```
`@rustbot` label +A-diagnostics
Suppress import errors for traits that couldve applied for method lookup error
Self-explanatory. I hit this quite often when refactoring in rustc, so even though this isn't really showing up as significant in the UI test suite, it probably will matter more for multi-module projects.
Fix up-to-date checking for run-make tests
This special case in `output_base_dir` had the unfortunate side-effect of causing all run-make tests to share the same `stamp` file. So as soon as any one of them succeeded, all of the failed tests would be incorrectly considered up-to-date and would no longer run in subsequent test invocations.
Fixes#129971.
r? jieyouxu
Update lint message for ABI not supported
Tracking issue: #130260
As requested in https://github.com/rust-lang/rust/pull/128784#pullrequestreview-2364026550 I updated the error message.
I could also change it to be the same message as if it was a hard error on a normal function:
> "`{abi}` is not a supported ABI for the current target"
Or would that get confusing when people try to google the error message?
r? compiler-errors
Fix bug where `option_env!` would return `None` when env var is present but not valid Unicode
Fixes#122669 by making `option_env!` emit an error when the value of the environment variable is not valid Unicode.
Autodiff Upstreaming - enzyme frontend
This is an upstream PR for the `autodiff` rustc_builtin_macro that is part of the autodiff feature.
For the full implementation, see: https://github.com/rust-lang/rust/pull/129175
**Content:**
It contains a new `#[autodiff(<args>)]` rustc_builtin_macro, as well as a `#[rustc_autodiff]` builtin attribute.
The autodiff macro is applied on function `f` and will expand to a second function `df` (name given by user).
It will add a dummy body to `df` to make sure it type-checks. The body will later be replaced by enzyme on llvm-ir level,
we therefore don't really care about the content. Most of the changes (700 from 1.2k) are in `compiler/rustc_builtin_macros/src/autodiff.rs`, which expand the macro. Nothing except expansion is implemented for now.
I have a fallback implementation for relevant functions in case that rustc should be build without autodiff support. The default for now will be off, although we want to flip it later (once everything landed) to on for nightly. For the sake of CI, I have flipped the defaults, I'll revert this before merging.
**Dummy function Body:**
The first line is an `inline_asm` nop to make inlining less likely (I have additional checks to prevent this in the middle end of rustc. If `f` gets inlined too early, we can't pass it to enzyme and thus can't differentiate it.
If `df` gets inlined too early, the call site will just compute this dummy code instead of the derivatives, a correctness issue. The following black_box lines make sure that none of the input arguments is getting optimized away before we replace the body.
**Motivation:**
The user facing autodiff macro can verify the user input. Then I write it as args to the rustc_attribute, so from here on I can know that these values should be sensible. A rustc_attribute also turned out to be quite nice to attach this information to the corresponding function and carry it till the backend.
This is also just an experiment, I expect to adjust the user facing autodiff macro based on user feedback, to improve usability.
As a simple example of what this will do, we can see this expansion:
From:
```
#[autodiff(df, Reverse, Duplicated, Const, Active)]
pub fn f1(x: &[f64], y: f64) -> f64 {
unimplemented!()
}
```
to
```
#[rustc_autodiff]
#[inline(never)]
pub fn f1(x: &[f64], y: f64) -> f64 {
::core::panicking::panic("not implemented")
}
#[rustc_autodiff(Reverse, Duplicated, Const, Active,)]
#[inline(never)]
pub fn df(x: &[f64], dx: &mut [f64], y: f64, dret: f64) -> f64 {
unsafe { asm!("NOP"); };
::core::hint::black_box(f1(x, y));
::core::hint::black_box((dx, dret));
::core::hint::black_box(f1(x, y))
}
```
I will add a few more tests once I figured out why rustc rebuilds every time I touch a test.
Tracking:
- https://github.com/rust-lang/rust/issues/124509
try-job: dist-x86_64-msvc
Special treatment empty tuple when suggest adding a string literal in format macro.
For example:
```rust
let s = "123";
println!({}, "sss", s);
```
Suggest:
`println!("{:?} {} {}", {}, "sss", s);`
fixes#130170
Update precondition tests (especially for zero-size access to null)
I don't much like the current way I've updated the precondition check helpers, but I couldn't come up with anything better. Ideas welcome.
I've organized `tests/ui/precondition-checks` mostly with one file per function that has `assert_unsafe_precondition` in it, with revisions that check each precondition. The important new test is `tests/ui/precondition-checks/zero-size-null.rs`.
Fix clobber_abi and disallow SVE-related registers in Arm64EC inline assembly
Currently `clobber_abi` in Arm64EC inline assembly is implemented using `InlineAsmClobberAbi::AArch64NoX18`, but broken since it attempts to clobber registers that cannot be used in Arm64EC: https://godbolt.org/z/r3PTrGz5r
```
error: cannot use register `x13`: x13, x14, x23, x24, x28, v16-v31 cannot be used for Arm64EC
--> <source>:6:14
|
6 | asm!("", clobber_abi("C"), options(nostack, nomem, preserves_flags));
| ^^^^^^^^^^^^^^^^
error: cannot use register `x14`: x13, x14, x23, x24, x28, v16-v31 cannot be used for Arm64EC
--> <source>:6:14
|
6 | asm!("", clobber_abi("C"), options(nostack, nomem, preserves_flags));
| ^^^^^^^^^^^^^^^^
<omitted the same errors for v16-v31>
```
Additionally, this disallows SVE-related registers per https://github.com/rust-lang/rust/pull/131332#issuecomment-2401189142.
cc ``@dpaoliello``
r? ``@Amanieu``
``@rustbot`` label O-windows O-AArch64 +A-inline-assembly
This special case in `output_base_dir` had the unfortunate side-effect of
causing all run-make tests to share the same `stamp` file. So as soon as any
one of them succeeded, all of the failed tests would be considered up-to-date
and would no longer run in subsequent test invocations.
Remap path prefix in the panic message of `tests/ui/meta/revision-bad.rs`
Otherwise `error-pattern` on the test run stderr can incorrectly match if the paths in panic backtrace has a matching substring (like if we look for `bar` in the error pattern, but the username is `baron`).
Tested locally by checking run output `./x test .\tests\ui\meta\revision-bad.rs -- -- --nocapture`:
```
--- stderr -------------------------------
thread 'main' panicked at remapped\meta\revision-bad.rs:14:5:
foo
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
------------------------------------------
```
Fixes#130996.
Check ABI target compatibility for function pointers
Tracking issue: https://github.com/rust-lang/rust/issues/130260
Related tracking issue: #87678
Compatibility of an ABI for a target was previously only performed on function definitions and `extern` blocks. This PR adds it also to function pointers to be consistent.
This might have broken some of the `tests/ui/` depending on the platform, so a try run seems like a good idea.
Also this might break existing code, because we now emit extra errors. Does this require a crater run?
# Example
```rust
// build with: --target=x86_64-unknown-linux-gnu
// These raise E0570
extern "thiscall" fn foo() {}
extern "thiscall" { fn bar() }
// This did not raise any error
fn baz(f: extern "thiscall" fn()) { f() }
```
# Open Questions
* [x] Should this report a future incompatibility warning like #87678 ?
* [ ] Is this the best place to perform the check?
std: fix stdout-before-main
Fixes#130210.
Since #124881, `ReentrantLock` uses `ThreadId` to identify threads. This has the unfortunate consequence of breaking uses of `Stdout` before main: Locking the `ReentrantLock` that synchronizes the output will initialize the thread ID before the handle for the main thread is set in `rt::init`. But since that would overwrite the current thread ID, `thread::set_current` triggers an abort.
This PR fixes the problem by using the already initialized thread ID for constructing the main thread handle and allowing `set_current` calls that do not change the thread's ID.
Stabilize const `ptr::write*` and `mem::replace`
Since `const_mut_refs` and `const_refs_to_cell` have been stabilized, we may now also stabilize the ability to write to places during const evaluation inside our library API. So, we now propose the `const fn` version of `ptr::write` and its variants. This allows us to also stabilize `mem::replace` and `ptr::replace`.
- const `mem::replace`: https://github.com/rust-lang/rust/issues/83164#issuecomment-2338660862
- const `ptr::write{,_bytes,_unaligned}`: https://github.com/rust-lang/rust/issues/86302#issuecomment-2330275266
Their implementation requires an additional internal stabilization of `const_intrinsic_forget`, which is required for `*::write*` and thus `*::replace`. Thus we const-stabilize the internal intrinsics `forget`, `write_bytes`, and `write_via_move`.
Add suggestion for removing invalid path sep `::` in fn def
Add suggestion for removing invalid path separator `::` in function definition.
for example: `fn invalid_path_separator::<T>() {}`
fixes#130791
Fixes#130210.
Since #124881, `ReentrantLock` uses `ThreadId` to identify threads. This has the unfortunate consequence of breaking uses of `Stdout` before main: Locking the `ReentrantLock` that synchronizes the output will initialize the thread ID before the handle for the main thread is set in `rt::init`. But since that would overwrite the current thread ID, `thread::set_current` triggers an abort.
This PR fixes the problem by using the already initialized thread ID for constructing the main thread handle and allowing `set_current` calls that do not change the thread's ID.
Remove deprecation note in the `non_local_definitions` lint
This PR removes the edition deprecation note emitted by the `non_local_definitions` lint.
Specifically this part:
```
= note: this lint may become deny-by-default in the edition 2024 and higher, see the tracking issue <https://github.com/rust-lang/rust/issues/120363>
```
because it [didn't make the cut](https://github.com/rust-lang/rust/issues/120363#issuecomment-2407833300) for the 2024 edition.
`@rustbot` label +L-non_local_definitions
Make unused_parens's suggestion considering expr's attributes.
For the expr with attributes,
like `let _ = (#[inline] || println!("Hello!"));`,
the suggestion's span should contains the attributes, or the suggestion will remove them.
fixes#129833
Support clobber_abi in MSP430 inline assembly
This supports `clobber_abi` which is one of the requirements of stabilization mentioned in #93335.
Refs: Section 3.2 "Register Conventions" in [MSP430 Embedded Application Binary Interface](https://www.ti.com/lit/an/slaa534a/slaa534a.pdf)
cc ``@cr1901``
r? ``@Amanieu``
``@rustbot`` label +O-msp430
Otherwise `error-pattern` on the test run stderr can incorrectly match
if e.g. the paths in panic backtrace has a matching substring (like if
we look for `bar` in the error pattern but username is `baron`).
For the expr with attributes, like `let _ = (#[inline] || println!("Hello!"));`, the suggestion's span should contains the attributes, or the suggestion will remove them.
fixes#129833
`IndexRange::len` is justified as an overall invariant, and
`take_prefix` and `take_suffix` are justified by local branch
conditions. A few more UB-checked calls remain in cases that are only
supported locally by `debug_assert!`, which won't do anything in
distributed builds, so those UB checks may still be useful.
We generally expect core's `#![rustc_preserve_ub_checks]` to optimize
away in user's release builds, but the mere presence of that extra code
can sometimes inhibit optimization, as seen in #131563.
Add intrinsics `fmuladd{f16,f32,f64,f128}`. This computes `(a * b) +
c`, to be fused if the code generator determines that (i) the target
instruction set has support for a fused operation, and (ii) that the
fused operation is more efficient than the equivalent, separate pair
of `mul` and `add` instructions.
https://llvm.org/docs/LangRef.html#llvm-fmuladd-intrinsic
MIRI support is included for f32 and f64.
The codegen_cranelift uses the `fma` function from libc, which is a
correct implementation, but without the desired performance semantic. I
think this requires an update to cranelift to expose a suitable
instruction in its IR.
I have not tested with codegen_gcc, but it should behave the same
way (using `fma` from libc).
Rollup of 6 pull requests
Successful merges:
- #129079 (Create `_imp__` symbols also when doing ThinLTO)
- #131208 (ABI: Pass aggregates by value on AIX)
- #131394 (fix(rustdoc): add space between struct fields and their descriptions)
- #131519 (Use Default visibility for rustc-generated C symbol declarations)
- #131541 (compiletest: Extract auxiliary-crate properties to their own module/struct)
- #131542 (next-solver: remove outdated FIXMEs)
r? `@ghost`
`@rustbot` modify labels: rollup
Use Default visibility for rustc-generated C symbol declarations
Non-default visibilities should only be used for definitions, not declarations, otherwise linking can fail.
This is based on https://github.com/rust-lang/rust/pull/123994.
Issue https://github.com/rust-lang/rust/issues/123427
When I changed `default-hidden-visibility` to `default-visibility` in https://github.com/rust-lang/rust/pull/130005, I updated all places in the code that used `default-hidden-visibility`, replicating the hidden-visibility bug to also happen for protected visibility.
Without this change, trying to build rustc with `-Z default-visibility=protected` fails with a link error.
ABI: Pass aggregates by value on AIX
On AIX we pass aggregates byval. Adds new ABI for AIX for powerpc64.
313ad85dfa/clang/lib/CodeGen/Targets/PPC.cpp (L216)
Fixes the following 2 testcases on AIX:
```
tests/ui/abi/extern/extern-pass-TwoU16s.rs
tests/ui/abi/extern/extern-pass-TwoU8s.rs
```
Create `_imp__` symbols also when doing ThinLTO
When generating a rlib crate on Windows we create `dllimport` / `_imp__` symbols for each global. This effectively makes the rlib contain an import library for itself and allows them to both be dynamically and statically linked. However when doing ThinLTO we do not generate these and thus we end up with missing symbols. Microsoft's `link` can fix these up (and emits warnings), but `lld` seems to currently be unable to.
This PR also does this generation for ThinLTO avoiding those issues with `lld` and also avoids the warnings on `link`.
This is an workaround for https://github.com/rust-lang/rust/issues/81408.
cc `@lqd`
Retire the `unnamed_fields` feature for now
`#![feature(unnamed_fields)]` was implemented in part in #115131 and #115367, however work on that feature has (afaict) stalled and in the mean time there have been some concerns raised (e.g.[^1][^2]) about whether `unnamed_fields` is worthwhile to have in the language, especially in its current desugaring. Because it represents a compiler implementation burden including a new kind of anonymous ADT and additional complication to field selection, and is quite prone to bugs today, I'm choosing to remove the feature.
However, since I'm not one to really write a bunch of words, I'm specifically *not* going to de-RFC this feature. This PR essentially *rolls back* the state of this feature to "RFC accepted but not yet implemented"; however if anyone wants to formally unapprove the RFC from the t-lang side, then please be my guest. I'm just not totally willing to summarize the various language-facing reasons for why this feature is or is not worthwhile, since I'm coming from the compiler side mostly.
Fixes#117942Fixes#121161Fixes#121263Fixes#121299Fixes#121722Fixes#121799Fixes#126969Fixes#131041
Tracking:
* https://github.com/rust-lang/rust/issues/49804
[^1]: https://rust-lang.zulipchat.com/#narrow/stream/213817-t-lang/topic/Unnamed.20struct.2Funion.20fields
[^2]: https://github.com/rust-lang/rust/issues/49804#issuecomment-1972619108
Consider outermost const-anon in `non_local_def` lint
This PR change the logic for finding the parent of the `impl` definition in the `non_local_definitions` lint to consider multiple level of const-anon items, instead of only one currently.
I also took the opportunity to cleanup the related code.
cc ``@traviscross``
Fixes https://github.com/rust-lang/rust/issues/131474
When making changes that have a large impact on coverage counter creation, this
makes it easier to see whether the number of physical counters has changed.
(The highest counter ID seen in coverage maps is not necessarily the same as
the number of physical counters actually used by the instrumented code, but
it's the best approximation we can get from looking only at the coverage maps,
and it should be reasonably accurate in most cases.)
compiletest: Remove the magic hacks for finding output with `lto=thin`
This hack was intended to handle the case where `-Clto=thin` causes the compiler to emit multiple output files (when producing LLVM-IR or assembly).
The hack only affects 4 tests, of which 3 are just meta-tests for the hack itself. The one remaining test that motivated the hack currently doesn't even need it!
(`tests/codegen/issues/issue-81408-dllimport-thinlto-windows.rs`)
This hack was intended to handle the case where `-Clto=thin` causes the
compiler to emit multiple output files (when producing LLVM-IR or assembly).
The hack only affects 4 tests, of which 3 are just meta-tests for the hack
itself. The one remaining test that motivated the hack currently doesn't even
need it!
(`tests/codegen/issues/issue-81408-dllimport-thinlto-windows.rs`)
Non-default visibilities should only be used for definitions, not
declarations, otherwise linking can fail.
Co-authored-by: Collin Baker <collinbaker@chromium.org>
Compiler & its UI tests: Rename remaining occurrences of "object safe" to "dyn compatible"
Follow-up to #130826.
Part of #130852.
1. 1st commit: Fix stupid oversights. Should've been part of #130826.
2. 2nd commit: Rename the unstable feature `object_safe_for_dispatch` to `dyn_compatible_for_dispatch`. Might not be worth the churn, you decide.
3. 3rd commit: Apply the renaming to all UI tests (contents and paths).
Precise capturing in traits
This PR begins to implement `feature(precise_capturing_in_traits)`, which enables using the `impl Trait + use<..>` syntax for RPITITs. It implements this by giving the desugared GATs variance, and representing the uncaptured lifetimes as bivariant, like how opaque captures work.
Right now, I've left out implementing a necessary extension to the `refining_impl_trait` lint, and also I've made it so that all RPITITs always capture the parameters that come from the trait, because I'm not totally yet convinced that it's sound to not capture these args. It's certainly required to capture the type and const parameters from the trait (e.g. Self), or else users could bivariantly relate two RPITIT args that come from different impls, but region parameters don't affect trait selection in the same way, so it *may* be possible to relax this in the future. Let's stay conservative for now, though.
I'm not totally sure what tests could be added on top of the ones I already added, since we really don't need to exercise the `precise_capturing` feature but simply what makes it special for RPITITs.
r? types
Tracking issue:
* #130044
rustc_target: Add sme-b16b16 as an explicit aarch64 target feature
LLVM 20 split out what used to be called b16b16 and correspond to aarch64
FEAT_SVE_B16B16 into sve-b16b16 and sme-b16b16.
Add sme-b16b16 as an explicit feature and update the codegen accordingly.
Resolves https://github.com/rust-lang/rust/pull/129894.
codegen_ssa: consolidate tied target checks
Fixes#105110.
Fixes#105111.
`rustc_codegen_llvm` and `rustc_codegen_gcc` duplicated logic for checking if tied target features were partially enabled. This PR consolidates these checks into `rustc_codegen_ssa` in the `codegen_fn_attrs` query, which also is run pre-monomorphisation for each function, which ensures that this check is run for unused functions, as would be expected.
Also adds a test confirming that enabling one tied feature doesn't imply another - the appropriate error for this was already being emitted. I did a bisect and narrowed it down to two patches it was likely to be - something in #128796, probably #128221 or #128679.
Introduce SolverRelating type relation to the new solver
Redux of #128744.
Splits out relate for the new solver so that implementors don't need to implement it themselves.
r? lcnr
LLVM and Cranelift disagree about how to return values that don't fit
in the registers designated for return values. LLVM will force the
entire return value to be passed by return area pointer, while
Cranelift will look at each IR level return value independently and
decide to pass it in a register or not, which would result in the
return value being passed partially in registers and partially through
a return area pointer.
While Cranelift may need to be fixed as the LLVM behavior is generally
more correct with respect to the surface language, forcing this
behavior in rustc itself makes it easier for other backends to conform
to the Rust ABI and for the C ABI rustc already handles this behavior
anyway.
In addition LLVM's decision to pass the return value in registers or
using a return area pointer depends on how exactly the return type is
lowered to an LLVM IR type. For example `Option<u128>` can be lowered
as `{ i128, i128 }` in which case the x86_64 backend would use a return
area pointer, or it could be passed as `{ i32, i128 }` in which case
the x86_64 backend would pass it in registers by taking advantage of an
LLVM ABI extension that allows using 3 registers for the x86_64 sysv
call conv rather than the officially specified 2 registers.
This adjustment is only necessary for the Rust ABI as for other ABI's
the calling convention implementations in rustc_target already ensure
any return value which doesn't fit in the available amount of return
registers is passed in the right way for the current target.
The `Box<T: Default>` impl currently calls `T::default()` before allocating
the `Box`.
Most `Default` impls are trivial, which should in theory allow
LLVM to construct `T: Default` directly in the `Box` allocation when calling
`<Box<T>>::default()`.
However, the allocation may fail, which necessitates calling `T's` destructor if it has one.
If the destructor is non-trivial, then LLVM has a hard time proving that it's
sound to elide, which makes it construct `T` on the stack first, and then copy it into the allocation.
Create an uninit `Box` first, and then write `T::default` into it, so that LLVM now only needs to prove
that the `T::default` can't panic, which should be trivial for most `Default` impls.
fix/update teach_note from 'escaping mutable ref/ptr' const-check
The old note was quite confusing since it talked about statics, but the message is also shown for consts. So let's reword to something that is true for both of them.
LLVM 20 split out what used to be called b16b16 and correspond to aarch64
FEAT_SVE_B16B16 into sve-b16b16 and sme-b16b16.
Add sme-b16b16 as an explicit feature and update the codegen accordingly.
Reserve guarded string literals (RFC 3593)
Implementation for RFC 3593, including:
- lexer / parser changes
- diagnostics
- migration lint
- tests
We reserve `#"`, `##"`, `###"`, `####`, and any other string of four or more repeated `#`. This avoids infinite lookahead in the lexer, though we still use infinite lookahead in the parser to provide better forward compatibility diagnostics.
This PR does not implement any special lexing of the string internals:
- strings preceded by one or more `#` are denied
- regardless of the number of trailing `#`
- string contents are lexed as if it was just a bare `"string"`
Tracking issue: #123735
RFC: rust-lang/rfcs#3593
Ignore broken-pipe-no-ice on apple (specifically macOS) for now
This test fails for me locally (initially reported by Zalathar) because apparently on macOS it doesn't say "internal compiler error" but it does report the std I/O panic, and it doesn't exit with a code of 101 but instead terminates with a wait signal of SIGPIPE.
Ignore this test on apple for now, until we try to actually address the underlying issue.
See https://github.com/rust-lang/rust/pull/131155 and https://github.com/rust-lang/rust/issues/131436 for more context.
Dont ICE when encountering post-mono layout cycle error
It's possible to encounter post-mono layout cycle errors in `fn_abi_of_instance`. Don't ICE in those cases.
This was originally discovered in an async fn, but that's not the only way to encounter such an error (which the other test I added should demonstrate).
Error messsages suck, but this fix is purely about suppressing the ICE.
Fixes#131409
Prevent building cargo from invalidating build cache of other tools due to conditionally applied `-Zon-broken-pipe=kill` via tracked `RUSTFLAGS`
This PR fixes#130980 where building cargo invalidated the tool build caches of other tools (such as rustdoc) because `-Zon-broken-pipe=kill` was conditionally passed via `RUSTFLAGS` for other tools *except* for cargo. The differing `RUSTFLAGS` triggered tool build cache invalidation as `RUSTFLAGS` is a tracked env var -- any changes in `RUSTFLAGS` requires a rebuild.
`-Zon-broken-pipe=kill` is load-bearing for rustc and rustdoc to not ICE on broken pipes due to usages of raw std `println!` that panics without the flag being set, which manifests in ICEs.
I can't say I like the changes here, but it is what it is...
See detailed discussions and history of `-Zon-broken-pipe=kill` usage in https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Internal.20lint.20for.20raw.20.60print!.60.20and.20.60println!.60.3F/near/474593815.
## Approach
This PR fixes the tool build cache invalidation by informing the `rustc` binary shim when to apply `-Zon-broken-pipe=kill` (i.e. when the rustc binary shim is not used to build cargo). This information is not communicated by `RUSTFLAGS`, which is an env var tracked by cargo, and instead uses an untracked env var `UNTRACKED_BROKEN_PIPE_FLAG` so we won't trigger tool build cache invalidation. We preserve bootstrap's behavior of not setting that flag for cargo by conditionally omitting setting `UNTRACKED_BROKEN_PIPE_FLAG` when building cargo.
Notably, the `-Zon-broken-pipe=kill` instance in 1e5719bdc4/src/bootstrap/src/core/build_steps/compile.rs (L1058) is not modified because that is used to build rustc only and not cargo itself.
Thanks to `@cuviper` for the idea!
## Testing
### Integration testing
This PR introduces a run-make test for rustc and rustdoc that checks that when they do not ICE/panic when they encounter a broken pipe of the stdout stream.
I checked this test will catch the broken pipe ICE regression for rustc on Linux (at least) by commenting out 1e5719bdc4/src/bootstrap/src/core/build_steps/compile.rs (L1058), and the test failed because rustc ICE'd.
### Manual testing
I have manually tried:
1. `./x clean && `./x test build --stage 1` -> `rustc +stage1 --print=sysroot | false`: no ICE.
2. `./x clean` -> `./x test run-make` twice: no stage 1 cargo rebuilds.
3. `./x clean` -> `./x build rustdoc` -> `rustdoc +stage1 --version | false`: no panics.
4. `./x test src/tools/cargo`: tests pass, notably `build::close_output` and `cargo_command::closed_output_ok` do not fail which would fail if cargo was built with `-Zon-broken-pipe=kill`.
## Related discussions
Thanks to everyone who helped!
- https://rust-lang.zulipchat.com/#narrow/stream/246057-t-cargo/topic/Applying.20.60-Zon-broken-pipe.3Dkill.60.20flags.20in.20bootstrap.3F
- https://rust-lang.zulipchat.com/#narrow/stream/326414-t-infra.2Fbootstrap/topic/Modifying.20run-make.20tests.20unnecessarily.20rebuild.20stage.201.20.2E.2E.2E
- https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Internal.20lint.20for.20raw.20.60print!.60.20and.20.60println!.60.3F
Fixes https://github.com/rust-lang/rust/issues/130980
Closes https://github.com/rust-lang/rust/issues/131059
---
try-job: aarch64-apple
try-job: x86_64-msvc
try-job: x86_64-mingw
[Coverage][MCDC] Adapt mcdc to llvm 19
Related issue: #126672
Also finish task 4 at #124144
[llvm #82448](https://github.com/llvm/llvm-project/pull/82448) has introduced some break changes into mcdc, causing incompatibility between llvm 18 and 19. This draft adapts to that change and gives up supporting for llvm-18.
Fix utf8-bom test
The BOM was accidentally removed in https://github.com/rust-lang/rust/pull/57108
I had to move the run-pass line down, because compiletest doesn't seem to know about BOMs, so it does not parse the header if it is the first line.
Add tests for some old fixed issues
Closes#30867Closes#30472Closes#28994Closes#26719 (and migrates the relevant test to the new run-make)
Closes#23600
cc `@jieyouxu` for the run-make-support changes
try-job: x86_64-msvc
Add missing module flags for `-Zfunction-return=thunk-extern`
This fixes a bug in the `-Zfunction-return=thunk-extern` flag. The flag needs to be passed onto LLVM to ensure that functions such as `asan.module_ctor` and `asan.module_dtor` that are created internally in LLVM have the mitigation applied to them.
This was originally discovered [in the Linux kernel](https://lore.kernel.org/all/CANiq72myZL4_poCMuNFevtpYYc0V0embjSuKb7y=C+m3vVA_8g@mail.gmail.com/).
Original flag PR: #116892
PR for similar issue: #129373
Tracking issue: #116853
cc ``@ojeda``
r? ``@wesleywiser``
Add precondition checks to ptr::offset, ptr::add, ptr::sub
All of `offset`, `add`, and `sub` (currently) have the trivial preconditions that the offset in bytes must be <= isize::MAX, and the computation of the new address must not wrap. This adds precondition checks for these, and like in slice indexing, we use intrinsics directly to implement unsafe APIs that have explicit checks, because we get a clearer error message that mentions the misused API not an implementation detail.
Experimentation indicates these checks have 1-2% compile time overhead, due primarily to adding the checks for `add`.
A crater run (https://github.com/rust-lang/rust/pull/130251#issuecomment-2395824565) indicates some people currently have buggy calls to `ptr::offset` that apply a negative offset to a null pointer, but the crater run does not hit the `ptr::add` or `ptr::sub` checks, which seems like an argument for cfg'ing out those checks on account of their overhead.
Don't allow the `#[pointee]` attribute where it doesn't belong
Error if the `#[pointee]` attribute is applied to anything but generic type parameters.
Closes#128485
Related to #123430
liballoc: introduce String, Vec const-slicing
This change `const`-qualifies many methods on `Vec` and `String`, notably `as_slice`, `as_str`, `len`. These changes are made behind the unstable feature flag `const_vec_string_slice`.
## Motivation
This is to support simultaneous variance over ownership and constness. I have an enum type that may contain either `String` or `&str`, and I want to produce a `&str` from it in a possibly-`const` context.
```rust
enum StrOrString<'s> {
Str(&'s str),
String(String),
}
impl<'s> StrOrString<'s> {
const fn as_str(&self) -> &str {
match self {
// In a const-context, I really only expect to see this variant, but I can't switch the implementation
// in some mode like #[cfg(const)] -- there has to be a single body
Self::Str(s) => s,
// so this is a problem, since it's not `const`
Self::String(s) => s.as_str(),
}
}
}
```
Currently `String` and `Vec` don't support this, but can without functional changes. Similar logic applies for `len`, `capacity`, `is_empty`.
## Changes
The essential thing enabling this change is that `Unique::as_ptr` is `const`. This lets us convert `RawVec::ptr` -> `Vec::as_ptr` -> `Vec::as_slice` -> `String::as_str`.
I had to move the `Deref` implementations into `as_{str,slice}` because `Deref` isn't `#[const_trait]`, but I would expect this change to be invisible up to inlining. I moved the `DerefMut` implementations as well for uniformity.
Don't use Immediate::offset to transmute pointers to integers
This applies the relatively new `assert_matches_abi` check in the `offset` operation on immediates, which makes sure that if offsets are used to alter the layout (which is possible because the field layout is arbitrarily picked by the caller), this is not done in a way that breaks the invariant of the `Immediate` type.
This leads to ICEs in a GVN mir-opt test, so the second commit fixes GVN.
Fixes https://github.com/rust-lang/rust/issues/131064.
This change `const`-qualifies many methods on Vec and String, notably
`as_slice`, `as_str`, `len`. These changes are made behind the unstable
feature flag `const_vec_string_slice` with the following tracking issue:
https://github.com/rust-lang/rust/issues/129041
add `naked_asm!` macro for use in `#[naked]` functions
tracking issue: https://github.com/rust-lang/rust/issues/90957
Adds the `core::arch::naked_asm` macro, to be used in `#[naked]` functions, but providing better error messages and a place to explain the restrictions on assembly in naked functions.
This PR does not yet require that the `naked_asm!` macro is used inside of `#[naked]` functions:
- the `asm!` macro can still be used in `#[naked]` functions currently, with the same restrictions and error messages as before.
- the `naked_asm!` macro can be used outside of `#[naked]` functions. It has not yet been decided whether that should be allowed long-term.
In this PR, the parsing code of `naked_asm!` now enforces the restrictions on assembly in naked functions, with the exception of checking that the `noreturn` option is specified. It also has not currently been decided if `noreturn` should be implicit or not.
This PR looks large because it touches a bunch of tests. The code changes are mostly straightforward I think: we now have 3 flavors of assembly macro, and that information must be propagated through the parsing code and error messages.
cc `@Lokathor`
r? `@Amanieu`
- fix for divergence
- fix error message
- fix another cranelift test
- fix some cranelift things
- don't set the NORETURN option for naked asm
- fix use of naked_asm! in doc comment
- fix use of naked_asm! in run-make test
- use `span_bug` in unreachable branch
Make deprecated_cfg_attr_crate_type_name a hard error
Turns the forward compatibility lint added by #83744 into a hard error, so now, while the `#![crate_name]` and `#![crate_type]` attributes are still allowed in raw form, they are now forbidden to be nested inside a `#![cfg_attr()]` attribute.
The following will now be an error:
```Rust
#![cfg_attr(foo, crate_name = "foobar")]
#![cfg_attr(foo, crate_type = "bin")]
```
This code will continue working and is not deprecated:
```Rust
#![crate_name = "foobar"]
#![crate_type = "lib"]
```
The reasoning for this is explained in #83744: it allows us to not have to cfg-expand in order to determine the crate's type and name.
As of filing the PR, exactly two years have passed since #99784 has been merged, which has turned the lint's default warning level into an error, so there has been ample time to move off the now-forbidden syntax.
cc #91632 - tracking issue for the lint
On function and method calls in patterns, link to the book
```
error: expected a pattern, found an expression
--> f889.rs:3:13
|
3 | let (x, y.drop()) = (1, 2);
| ^^^^^^^^ not a pattern
|
= note: arbitrary expressions are not allowed in patterns: <https://doc.rust-lang.org/book/ch18-00-patterns.html>
error[E0532]: expected a pattern, found a function call
--> f889.rs:2:13
|
2 | let (x, drop(y)) = (1, 2);
| ^^^^ not a tuple struct or tuple variant
|
= note: function calls are not allowed in patterns: <https://doc.rust-lang.org/book/ch18-00-patterns.html>
```
Fix#97200.
Do not consider match/let/ref of place that evaluates to `!` to diverge, disallow coercions from them too
Fixes#117288.
This PR implements a heuristic which disables two things that are currently being performed on the HIR when we have **expressions that involve place-like expressions that point to `!`**. Specifically, it will (in certain cases explained below):
### (1.) Disable the `NeverToAny` coercion we implicitly insert for `!`.
Which fixes this inadvertent, sneaky unsoundness:
```
unsafe {
let x: *const ! = &0 as *const u8 as *const !;
let _: () = *x;
}
```
which is UB because currently rust emits an *implicit* NeverToAny coercion even though we really shouldn't be, since there's no read of the value pointed by `x`.
### (2.) Disable the logic which considers expression which evaluate to `!` to diverge, which affects the type returned by the containing block.
Which fixes this unsoundness:
```
fn make_up_a_value<T>() -> T {
unsafe {
let x: *const ! = &0 as *const u8 as *const !;
let _ = *x;
}
}
```
We disable these two operations **if** the expression is a place-like expression (locals, statics, field projections, index operations, and deref operations), and if the parent expression is either:
(1.) the LHS of an assignment
(2.) AddrOf
(3.) A match or let **unless** all of the *patterns consitute a read*, which is explained below:
And finally, a pattern currently is considered to constitute a read **unless** it is a wildcard, or an OR pattern. An OR pattern is considered to constitute a read if all of its subpatterns constitute a read, to remain as conservative as possible in cases like `_ | subpat` or `subpat | _`.
All other patterns are considered currently to constitute a read. Specifically, because `NeverToAny` is a coercion performed on a *value* and not a *place*, `Struct { .. }` on a `!` type must be a coercion currently, and we currently rely on this behavior to allow us to perform coercions like `let _: i32 = x;` where `x: !`.
This is already considered UB by [miri](https://play.rust-lang.org/?version=nightly&mode=debug&edition=2021&gist=daf3a2246433fe43fdc07d1389c276c9), but also means it does not affect the preexisting UB in this case:
```
let Struct { .. } = *never_ptr;
```
Even though it's likely up for debate since we're not actually reading any data out of the struct, it almost certainly causes inference changes which I do *NOT* want to fix in this PR.
```
error: expected a pattern, found an expression
--> f889.rs:3:13
|
3 | let (x, y.drop()) = (1, 2); //~ ERROR
| ^^^^^^^^ not a pattern
|
= note: arbitrary expressions are not allowed in patterns: <https://doc.rust-lang.org/book/ch18-00-patterns.html>
error[E0532]: expected a pattern, found a function call
--> f889.rs:2:13
|
2 | let (x, drop(y)) = (1, 2); //~ ERROR
| ^^^^ not a tuple struct or tuple variant
|
= note: function calls are not allowed in patterns: <https://doc.rust-lang.org/book/ch18-00-patterns.html>
```
Fix#97200.
Initial support for riscv32{e|em|emc}_unknown_none_elf
We have a research prototype of an RV32EMC target and have been successfully running the e, em, emc programs on it. I'm hoping upstreaming this configuration would make the target maintenance slightly easier.
Configuration is based on the respective {i, im, imc} variants. As defined in RISC-V Unprivileged Spec. 20191213, the only change in RVE wrt. RVI is to reduce the number of integer registers to 16 (x0-x15), which also implies
- 2 callee saved registers instead of 12
- 32-bit / 4-byte stack alignment instead of 128 bits / 16 bytes
My initial presumption is that this will not impact how the target is defined for the compiler but only becomes relevant at the runtime level. I am willing to investigate, though.
EDIT: LLVM is now told about the presumed 32-bit stack alignment.
`@Disasm` `@romancardenas`
