Stall computing instance for drop shim until it has no unsubstituted const params
Do not inline the drop shim for types that still have unsubstituted const params.
## Why?
#127030 ICEs because it tries to inline the drop shim for a type with an unsubstituted const param. In order to generate this shim, this requires calling the drop shim builder, which invokes the trait solver to compute whether constituent types need drop (since we compute if a type is copy to disqualify any `Drop` behavior):
9c3bc805dd/compiler/rustc_mir_dataflow/src/elaborate_drops.rs (L378)
However, since we don't keep the param-env of the instance we resolved the item for, we use the wrong param-env:
9c3bc805dd/compiler/rustc_mir_transform/src/shim.rs (L278)
(which is the param-env of `std::ptr::drop_in_place`)
This param-env is notably missing `ConstParamHasTy` predicates, and since we removed the type from consts in https://github.com/rust-lang/rust/pull/125958, we literally cannot prove these predicates in this (relatively) empty param-env. This currently happens in places like the MIR inliner, but may happen elsewhere such as in lints that resolve terminators.
## What?
We force the inliner to not consider calls for `drop_in_place` for types that have unsubstituted const params.
## So what?
This may negatively affect MIR inlining, but I doubt this matters in practice, and fixes a beta regression, so let's fix it. I will look into approaches for fixing this in a more maintainable way, perhaps delaying the creation of drop shim bodies until codegen (like how intrinsics work).
Enable const casting for `f16` and `f128`
I have an open PR to the Miri repo adding tests for this behavior https://github.com/rust-lang/miri/pull/3688, but that unfortunately hits the ICE path here. The changes seem reasonably low risk that it might be okay to merge separately from the tests, and I tested the result locally against an older version of https://github.com/rust-lang/miri/pull/3688.
Cc ``````@RalfJung``````
Switch back `non_local_definitions` lint to allow-by-default
This PR switch back (again) the `non_local_definitions` lint to allow-by-default as T-lang is requesting some (major) changes in the lint inner workings in https://github.com/rust-lang/rust/issues/126768#issuecomment-2192634762.
This PR will need to be beta-backported, as the lint is currently warn-by-default in beta.
Update AST validation module docs
Drive-by doc update for AST validation pass:
- Syntax extensions are replaced by proc macros.
- Add rationale for why AST validation pass need to be run
post-expansion and why the pass is needed in the first place.
This was discussed during this week's [rustc-dev-guide reading club](https://rust-lang.zulipchat.com/#narrow/stream/196385-t-compiler.2Fwg-rustc-dev-guide), and the rationale was explained by cc ``````@bjorn3.``````
Instead of keeping a list of architectures which have native support
for 64-bit atomics, just use #[cfg(target_has_atomic = "64")] and its
inverted counterpart to determine whether we need to use portable
AtomicU64 on the target architecture.
Tighten `fn_decl_span` for async blocks
Tightens the span of `async {}` blocks in diagnostics, and subsequently async closures and async fns, by actually setting the `fn_decl_span` correctly. This is kinda a follow-up on #125078, but it fixes the problem in a more general way.
I think the diagnostics are significantly improved, since we no longer have a bunch of overlapping spans. I'll point out one caveat where I think the diagnostic may get a bit more confusing, but where I don't think it matters.
r? ````@estebank```` or ````@oli-obk```` or someone else on wg-diag or compiler i dont really care lol
Support fetching `Attribute` of items.
Fixes [https://github.com/rust-lang/project-stable-mir/issues/83](https://github.com/rust-lang/project-stable-mir/issues/83)
`rustc_ast::ast::Attribute` doesn't impl `Hash` and `Eq`. Thus it cannot be directly used as key of `IndexMap` in `rustc_smir::rustc_smir::Tables` and we cannot define stable `Attribute` as index to `rustc_ast::ast::Attribute` like `Span` and many other stable definitions.
Since an string (or tokens) and its span contain all info about an attribute, I defined a simple `Attribute` struct on stable side.
I choose to fetch attributes via `tcx::get_attrs_by_path()` due to `get_attrs()` is marked as deprecated and `get_attrs_by_name()` cannot handle name of multiple segments like `rustfmt::skip`.
r? `@celinval`
patchable-function-entry: Add unstable compiler flag and attribute
Tracking issue: #123115
Add the -Z patchable-function-entry compiler flag and the #[patchable_function_entry(prefix_nops = m, entry_nops = n)] attribute.
Rebased and adjusted the canditate implementation to match changes in the RFC.
ast: Standardize visiting order
Order: ID, attributes, inner nodes in source order if possible, tokens, span.
Also always use exhaustive matching in visiting infra, and visit some discovered missing nodes.
Unlike https://github.com/rust-lang/rust/pull/125741 this shouldn't affect anything serious like `macro_rules` scopes.
coverage: Make `#[coverage(..)]` apply recursively to nested functions
This PR makes the (currently-unstable) `#[coverage(off)]` and `#[coverage(on)]` attributes apply recursively to all nested functions/closures, instead of just the function they are directly attached to.
Those attributes can now also be applied to modules and to impl/impl-trait blocks, where they have no direct effect, but will be inherited by all enclosed functions/closures/methods that don't override the inherited value.
---
Fixes#126625.
Fixes for 32-bit SPARC on Linux
This PR fixes a number of issues which previously prevented `rustc` from being built
successfully for 32-bit SPARC using the `sparc-unknown-linux-gnu` triplet.
In particular, it adds linking against `libatomic` where necessary, uses portable `AtomicU64`
for `rustc_data_structures` and rewrites the spec for `sparc_unknown_linux_gnu` to use
`TargetOptions` and replaces the previously used `-mv8plus` with the more portable
`-mcpu=v9 -m32`.
To make `rustc` build successfully, support for 32-bit SPARC needs to be added to the `object`
crate as well as the `nix` crate which I will be sending out later as well.
r? nagisa
Id, attributes, inner nodes in source order if possible, tokens, span.
Also always use exhaustive matching in visiting infra, and visit some missing nodes.
Remove more `PtrToPtr` casts in GVN
This addresses two things I noticed in MIR:
1. `NonNull::<T>::eq` does `(a as *mut T) == (b as *mut T)`, but it could just compare the `*const T`s, so this removes `PtrToPtr` casts that are on both sides of a pointer comparison, so long as they're not fat-to-thin casts.
2. `NonNull::<T>::addr` does `transmute::<_, usize>(p as *const ())`, but so long as `T: Thin` that cast doesn't do anything, and thus we can directly transmute the `*const T` instead.
r? mir-opt
Don't ICE during RPITIT refinement checking for resolution errors after normalization
#126670 shows a case where resolution errors after normalization can happen during RPITIT refinement checking. Our tests didn't reach this path before, and we explicitly ICEd until we had a test. We can now delay a bug since we're sure it is reachable and have the test from the isue.
The comment I added likely still needs more expert wordsmithing.
r? ``@compiler-errors`` who's making me work during vacation (j/k).
Fixes#126670
miri: make sure we can find link_section statics even for the local crate
Miri needs some way to iterate all the exported functions and "used" statics of all crates. For dependency crates, this already works fine since we can overwrite the query resonsible for computing `exported_symbols`, but it turns out for local binary crates this does not work: for binaries, `reachable_set` skips a lot of its logic and only checks `contains_extern_indicator()` and `RUSTC_STD_INTERNAL_SYMBOL`. Other flags like `CodegenFnAttrFlags::USED` are entirely ignored.
This PR proposes to use the same check, `has_custom_linkage`, in binaries that we already use to drive the main workqueue of the reachability recursive traversal. I have no idea why binaries used a slightly different check that ignores `USED` -- was that deliberate or does it just not matter most of the time?
Change E0369 to give note informations for foreign items.
Change E0369 to give note informations for foreign items.
Make it easy for developers to understand why the binop cannot be applied.
fixes#125631
Fix a span in `parse_ty_bare_fn`.
It currently goes one token too far.
Example: line 259 of `tests/ui/abi/compatibility.rs`:
```
test_abi_compatible!(fn_fn, fn(), fn(i32) -> i32);
```
This commit changes the span for the second element from `fn(),` to `fn()`, i.e. removes the extraneous comma.
This doesn't affect any tests. I found it while debugging some other code. Not a big deal but an easy fix so I figure it worth doing.
r? ``@spastorino``
Rollup of 7 pull requests
Successful merges:
- #126618 (Mark assoc tys live only if the corresponding trait is live)
- #126746 (Deny `use<>` for RPITITs)
- #126868 (not use offset when there is not ends with brace)
- #126884 (Do not ICE when suggesting dereferencing closure arg)
- #126893 (Eliminate the distinction between PREC_POSTFIX and PREC_PAREN precedence level)
- #126915 (Don't suggest awaiting in closure patterns)
- #126943 (De-duplicate all consecutive native libs regardless of their options)
r? `@ghost`
`@rustbot` modify labels: rollup
It currently goes one token too far.
Example: line 259 of `tests/ui/abi/compatibility.rs`:
```
test_abi_compatible!(fn_fn, fn(), fn(i32) -> i32);
```
This commit changes the span for the second element from `fn(),` to
`fn()`, i.e. removes the extraneous comma.
Tweak `FlatPat::new` to avoid a temporarily-invalid state
It was somewhat confusing that the old constructor would create a `FlatPat` in a (possibly) non-simplified state, and then simplify its contents in-place.
So instead we now create its fields as local variables, perform simplification, and then create the struct afterwards.
This doesn't affect correctness, but is less confusing.
---
I've also included some semi-related comments that I made while trying to navigate this code.
Tweak a confusing comment in `create_match_candidates`
This comment was accurate at the time it was written, but various later changes reshuffled things in ways that caused the existing comment to become confusing.
I've therefore tried to clarify that *these* candidates are 1:1 with match arms, while also warning that that isn't the case in general.
Detect unused structs which derived Default
<!--
If this PR is related to an unstable feature or an otherwise tracked effort,
please link to the relevant tracking issue here. If you don't know of a related
tracking issue or there are none, feel free to ignore this.
This PR will get automatically assigned to a reviewer. In case you would like
a specific user to review your work, you can assign it to them by using
r? <reviewer name>
-->
Fixes#98871
`-Z patchable-function-entry` works like `-fpatchable-function-entry`
on clang/gcc. The arguments are total nop count and function offset.
See MCP rust-lang/compiler-team#704
De-duplicate all consecutive native libs regardless of their options
Address https://github.com/rust-lang/rust/pull/126913#issuecomment-2188184011 by no longer de-duplicating based on the "options" but by only looking at the generated link args, as to avoid consecutive libs that originated from different native-lib with different options (like `raw-dylib` on Windows) but isn't relevant for `--print=native-static-libs`.
r? ``@petrochenkov``
Don't suggest awaiting in closure patterns
Fixes#126903.
For
```rust
async fn do_async() {}
fn main() {
Some(do_async()).map(|()| {});
}
```
the error is now
```rust
error[E0308]: mismatched types
--> src/main.rs:4:27
|
4 | Some(do_async()).map(|()| {});
| ^^
| |
| expected future, found `()`
| expected due to this
|
= note: expected opaque type `impl Future<Output = ()>`
found unit type `()`
```
Ideally, if `main` were to be `async`, it should be
```rs
error[E0308]: mismatched types
--> src/main.rs:4:27
|
4 | Some(do_async()).map(|()| {});
| ^^
| |
| expected future, found `()`
| expected due to this
|
= note: expected opaque type `impl Future<Output = ()>`
found unit type `()`
help: consider `await`ing on the `Future`
|
4 | Some(do_async().await).map(|()| {});
| ++++++
```
However, this would mean `FnCtx::check_pat_top` would have to be called with an `origin_expr` in `rustc_hir_typeck::check::check_fn`, and that expr would have to be somehow plumbed through `FnCtxt::check_expr_closure` and closure signature deduction. I'm willing to work on the plumbing but unsure how to start.
Eliminate the distinction between PREC_POSTFIX and PREC_PAREN precedence level
I have been tangling with precedence as part of porting some pretty-printer improvements from syn back to rustc (related to parenthesization of closures, returns, and breaks by the AST pretty-printer).
As far as I have been able to tell, there is no difference between the 2 different precedence levels that rustc identifies as `PREC_POSTFIX` (field access, square bracket index, question mark, method call) and `PREC_PAREN` (loops, if, paths, literals).
There are a bunch of places that look at either `prec < PREC_POSTFIX` or `prec >= PREC_POSTFIX`. But there is nothing that needs to distinguish PREC_POSTFIX and PREC_PAREN from one another.
d49994b060/compiler/rustc_ast/src/util/parser.rs (L236-L237)d49994b060/compiler/rustc_hir_typeck/src/fn_ctxt/suggestions.rs (L2829)d49994b060/compiler/rustc_hir_typeck/src/fn_ctxt/suggestions.rs (L1290)
In the interest of eliminating a distinction without a difference, this PR collapses these 2 levels down to 1.
There is exactly 1 case where an expression with PREC_POSTFIX precedence needs to be parenthesized in a location that an expression with PREC_PAREN would not, and that's when the receiver of ExprKind::MethodCall is ExprKind::Field. `x.f()` means a different thing than `(x.f)()`. But this does not justify having separate precedence levels because this special case in the grammar is not governed by precedence. Field access does not have "lower precedence than" method call syntax — you can tell because if it did, then `x.f[0].f()` wouldn't be able to have its unparenthesized field access in the receiver of a method call. Because this Field/MethodCall special case is not governed by precedence, it already requires special handling and is not affected by eliminating the PREC_POSTFIX precedence level.
d49994b060/compiler/rustc_ast_pretty/src/pprust/state/expr.rs (L217-L221)
Do not ICE when suggesting dereferencing closure arg
Account for `for` lifetimes when constructing closure to see if dereferencing the return value would be valid.
Fix#125634, fix#124563.
Deny `use<>` for RPITITs
Precise capturing `use<>` syntax is currently a no-op on RPITITs, since GATs have no variance, so all captured lifetimes are captured invariantly.
We don't currently *need* to support `use<>` on RPITITs, since `use<>` is initially intended for migrating RPIT *overcaptures* from edition 2021->2024, but since RPITITs currently capture all in-scope lifetimes, we'll never need to write `use<>` on an RPITIT.
Eventually, though, it would be desirable to support precise capturing on RPITITs, since RPITITs overcapturing by default can be annoying to some folks. But let's separate that (which will likely require some delicate types team work for adding variances to GATs and adjusting the refinement rules) from the stabilization of the feature for edition 2024.
r? oli-obk cc ``@traviscross``
Tracking:
- https://github.com/rust-lang/rust/issues/123432
ast: Standardize visiting order for attributes and node IDs
This should only affect `macro_rules` scopes and order of diagnostics.
Also add a deprecation lint for `macro_rules` called outside of their scope, like in https://github.com/rust-lang/rust/issues/124535.
Various refactorings to rustc_interface
This should make it easier to move the driver interface away from queries in the future. Many custom drivers call queries like `queries.global_ctxt()` before they are supposed to be called, breaking some things like certain `--print` and `-Zunpretty` options, `-Zparse-only` and emitting the dep info at the wrong point in time. They are also not actually necessary at all. Passing around the query output manually would avoid recomputation too and would be just as easy. Removing driver queries would also reduce the amount of global mutable state of the compiler. I'm not removing driver queries in this PR to avoid breaking the aforementioned custom drivers.
It was somewhat confusing that the old constructor would create a `FlatPat` in
a (possibly) non-simplified state, and then simplify its contents in-place.
So instead we now create its fields as local variables, perform simplification,
and then create the struct afterwards.
This doesn't affect correctness, but is less confusing.
transmute size check: properly account for alignment
Fixes another place where ZST alignment was ignored when checking whether something is a newtype. I wonder how many more of these there are...
Fixes https://github.com/rust-lang/rust/issues/101084
Allow constraining opaque types during various unsizing casts
allows unsizing of tuples, arrays and Adts to constraint opaque types in their generic parameters to concrete types on either side of the unsizing cast.
Also allows constraining opaque types during trait object casts that only differ in auto traits or lifetimes.
cc #116652
Just some extra sanity checking, making explicit some values not
possible in code working with token trees -- we shouldn't be seeing
explicit delimiter tokens, because they should be represented as
`TokenTree::Delimited`.
Add `SliceLike` to `rustc_type_ir`, use it in the generic solver code (+ some other changes)
First, we split out `TraitRef::new_from_args` which takes *just* `ty::GenericArgsRef` from `TraitRef::new` which takes `impl IntoIterator<Item: Into<GenericArg>>`. I will explain in a minute why.
Second, we introduce `SliceLike`, which allows us to be generic over `List<T>` and `[T]`. This trait has an `as_slice()` and `into_iter()` method, and some other convenience functions. However, importantly, since types like `I::GenericArgs` now implement `SliceLike` rather than `IntoIter<Item = I::GenericArg>`, we can't use `TraitRef::new` on this directly. That's where `new_from_args` comes in.
Finally, we adjust all the code to use these slice operators. Some things get simpler, some things get a bit more annoying since we need to use `as_slice()` in a few places. 🤷
r? lcnr
