The ICE message is somewhat confusing and overly specific - the issue is
that there's no MIR available.
This should make debugging these ICEs easier since the error tells you
what's actually wrong, not what it was trying to do when it failed.
cc https://github.com/rust-lang/rust/pull/80952#issuecomment-759198841
For example, this code:
struct S(i32, f32);
let S(x) = S(0, 1.0);
will make the compiler suggest either:
let S(x, _) = S(0, 1.0);
or:
let S(x, ..) = S(0, 1.0);
Fix --pretty=expanded with --remap-path-prefix
Per https://github.com/rust-lang/rust/issues/80832, using
--pretty=expanded and --remap-path-prefix results in an ICE.
This is becasue the session source files table is stored in remapped
form, whereas --pretty-expanded looks up unremapped files. This remaps
the path prefixes before lookup.
~~There don't appear to be any existing tests for --pretty=expanded; I'll look into
adding some.~~ Never mind, found the pretty tests.
Fixes#80832
Added support for i386-unknown-linux-gnu and i486-unknown-linux-gnu
Support for both can be useful when creating new firmware, boot loaders,
or embedded operating systems.
Separate out a `hir::Impl` struct
This makes it possible to pass the `Impl` directly to functions, instead
of having to pass each of the many fields one at a time. It also
simplifies matches in many cases.
See `rustc_save_analysis::dump_visitor::process_impl` or `rustdoc::clean::clean_impl` for a good example of how this makes `impl`s easier to work with.
r? `@petrochenkov` maybe?
This makes it possible to pass the `Impl` directly to functions, instead
of having to pass each of the many fields one at a time. It also
simplifies matches in many cases.
The optimized MIR for closures is being encoded unconditionally, while
being unnecessary for cargo check. This turns out to be especially
costly with MIR inlining enabled, since it triggers computation of
optimized MIR for all callees that are being examined for inlining
purposes.
Skip encoding of optimized MIR for closures, enum constructors, struct
constructors, and trait fns when not doing codegen, like it is already
done for other items since 49433.
Turn type inhabitedness into a query to fix `exhaustive_patterns` perf
We measured in https://github.com/rust-lang/rust/pull/79394 that enabling the [`exhaustive_patterns` feature](https://github.com/rust-lang/rust/issues/51085) causes significant perf degradation. It was conjectured that the culprit is type inhabitedness checking, and [I hypothesized](https://github.com/rust-lang/rust/pull/79394#issuecomment-733861149) that turning this computation into a query would solve most of the problem.
This PR turns `tcx.is_ty_uninhabited_from` into a query, and I measured a 25% perf gain on the benchmark that stress-tests `exhaustiveness_patterns`. This more than compensates for the 30% perf hit I measured [when creating it](https://github.com/rust-lang/rustc-perf/pull/801). We'll have to measure enabling the feature again, but I suspect this fixes the perf regression entirely.
I'd like a perf run on this PR obviously.
I made small atomic commits to help reviewing. The first one is just me discovering the "revisions" feature of the testing framework.
I believe there's a push to move things out of `rustc_middle` because it's huge. I guess `inhabitedness/mod.rs` could be moved out, but it's quite small. `DefIdForest` might be movable somewhere too. I don't know what the policy is for that.
Ping `@camelid` since you were interested in following along
`@rustbot` modify labels: +A-exhaustiveness-checking
Since `DefIdForest` contains 0 or 1 elements the large majority of the
time, by allocating only in the >1 case we avoid almost all allocations,
compared to `Arc<SmallVec<[DefId;1]>>`. This shaves off 0.2% on the
benchmark that stresses uninhabitedness checking.
Remove much of the special-case handling around crate metadata
dependency tracking by replacing `DepKind::CrateMetadata` and the
pre-allocation of corresponding `DepNodes` with on-demand invocation
of the `crate_hash` query.
Make CTFE able to check for UB...
... by not doing any optimizations on the `const fn` MIR used in CTFE. This means we duplicate all `const fn`'s MIR now, once for CTFE, once for runtime. This PR is for checking the perf effect, so we have some data when talking about https://github.com/rust-lang/const-eval/blob/master/rfcs/0000-const-ub.md
To do this, we now have two queries for obtaining mir: `optimized_mir` and `mir_for_ctfe`. It is now illegal to invoke `optimized_mir` to obtain the MIR of a const/static item's initializer, an array length, an inline const expression or an enum discriminant initializer. For `const fn`, both `optimized_mir` and `mir_for_ctfe` work, the former returning the MIR that LLVM should use if the function is called at runtime. Similarly it is illegal to invoke `mir_for_ctfe` on regular functions.
This is all checked via appropriate assertions and I don't think it is easy to get wrong, as there should be no `mir_for_ctfe` calls outside the const evaluator or metadata encoding. Almost all rustc devs should keep using `optimized_mir` (or `instance_mir` for that matter).
Enhance type inference errors involving the `?` operator
This patch adds a special-cased note on type inference errors when the error span points to a `?` return. It also makes the primary label for such errors "cannot infer type of `?` error" in cases where before we would have only said "cannot infer type".
One beneficiary of this change is async blocks, where we can't explicitly annotate the return type and so may not generate any other help (#77880); this lets us at least print the error type we're converting from and anything we know about the type we can't fully infer. More generally, it signposts that an implicit conversion is happening that may have impeded type inference the user was expecting. We already do something similar for [mismatched type errors](2987785df3/src/test/ui/try-block/try-block-bad-type.stderr (L7)).
The check for a relevant `?` operator is built into the existing HIR traversal which looks for places that could be annotated to resolve the error. That means we could identify `?` uses anywhere in the function that output the type we can't infer, but this patch just sticks to adding the note if the primary span given for the error has the operator; if there are other expressions where the type occurs and one of them is selected for the error instead, it's more likely that the `?` operator's implicit conversion isn't the sole cause of the inference failure and that adding an additional diagnostic would just be noise. I added a ui test for one such case.
The data about the `?` conversion is passed around in a `UseDiagnostic` enum that in theory could be used to add more of this kind of note in the future. It was also just easier to pass around than something with a more specific name. There are some follow-up refactoring commits for the code that generates the error label, which was already pretty involved and made a bit more complicated by this change.
Rollup of 8 pull requests
Successful merges:
- #79757 (Replace tabs earlier in diagnostics)
- #80600 (Add `MaybeUninit` method `array_assume_init`)
- #80880 (Move some tests to more reasonable directories)
- #80897 (driver: Use `atty` instead of rolling our own)
- #80898 (Add another test case for #79808)
- #80917 (core/slice: remove doc comment about scoped borrow)
- #80927 (Replace a simple `if let` with the `matches` macro)
- #80930 (fix typo in trait method mutability mismatch help)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
driver: Use `atty` instead of rolling our own
Fixes#80888.
Rationale:
- `atty` is widely used in the Rust ecosystem
- We already use it (in `rustc_errors` and other places)
- We shouldn't be rolling our own TTY detector when there's a
widely-used, well-tested package that we can use
Replace tabs earlier in diagnostics
This replaces tabs earlier in the diagnostics emitting process, which allows various margin calculations to ignore the existence of tabs. It does add a string copy for the source lines that are emitted.
Fixes https://github.com/rust-lang/rust/issues/78438
r? `@estebank`
Serialize incr comp structures to file via fixed-size buffer
Reduce a large memory spike that happens during serialization by writing
the incr comp structures to file by way of a fixed-size buffer, rather
than an unbounded vector.
Effort was made to keep the instruction count close to that of the
previous implementation. However, buffered writing to a file inherently
has more overhead than writing to a vector, because each write may
result in a handleable error. To reduce this overhead, arrangements are
made so that each LEB128-encoded integer can be written to the buffer
with only one capacity and error check. Higher-level optimizations in
which entire composite structures can be written with one capacity and
error check are possible, but would require much more work.
The performance is mostly on par with the previous implementation, with
small to moderate instruction count regressions. The memory reduction is
significant, however, so it seems like a worth-while trade-off.
Rationale:
- `atty` is widely used in the Rust ecosystem
- We already use it (in `rustc_errors` and other places)
- We shouldn't be rolling our own TTY detector when there's a
widely-used, well-tested package that we can use
Suggest async {} for async || {}
Fixes#76011
This adds support for adding help diagnostics to the feature gating checks and
then uses it for the async_closure gate to add the extra bit of help
information as described in the issue.
resolve: Simplify built-in macro table
We don't use full `SyntaxExtension`s from the table, only `SyntaxExtensionKind`s, and `Ident` in `register_builtin_macro` always had dummy span. This PR removes unnecessary data from the table and related function signatures.
Noticed when reviewing #80850.
Explain method-call move errors in loops
PR #73708 added a more detailed explanation of move errors that occur
due to a call to a method that takes `self`. This PR extends that logic
to work when a move error occurs due to a method call in the previous
iteration of a loop.
Split a func into cold/hot parts, reducing binary size
I noticed that the Size::bits function is called in many places,
and is inlined into them. On x86_64-pc-windows-msvc, this function
is inlined 527 times, and compiled separately (non-inlined) 3 times.
Each of those inlined calls contains code that panics. This commit
moves the `panic!` call into a separate function and marks that
function with `#[cold]`.
This reduces binary size by 24 KB. Not much, but it's something.
Changes like this often reduce pressure on instruction-caches,
since it reduces the amount of code that is inlined into hot code
paths. Or more precisely, it removes cold code from hot cache lines.
Use correct ABI for wasm32 by default
Introduces `wasm32-unknown-bindgen` for those wishing to use the bindgen compat abi. `wasm32-*` now uses the correct abi by default.
Fixes https://github.com/rustwasm/team/issues/291
The signed LEB128 decoding function used a hardcoded constant of 64
instead of the number of bits in the type of integer being decoded,
which resulted in incorrect results for some inputs. Fix this, make the
decoding more consistent with the unsigned version, and increase the
LEB128 encoding and decoding test coverage.
Reduce a large memory spike that happens during serialization by writing
the incr comp structures to file by way of a fixed-size buffer, rather
than an unbounded vector.
Effort was made to keep the instruction count close to that of the
previous implementation. However, buffered writing to a file inherently
has more overhead than writing to a vector, because each write may
result in a handleable error. To reduce this overhead, arrangements are
made so that each LEB128-encoded integer can be written to the buffer
with only one capacity and error check. Higher-level optimizations in
which entire composite structures can be written with one capacity and
error check are possible, but would require much more work.
The performance is mostly on par with the previous implementation, with
small to moderate instruction count regressions. The memory reduction is
significant, however, so it seems like a worth-while trade-off.
log-color: Detect TTY based on stderr, not stdout
Fixes#78435 (again).
Logging goes to stderr, not stdout, so we should base our automated
detection on stderr instead of stdout.
Thanks to Ralf Jung for noticing and reporting the bug!
r? `@oli-obk`
cc `@RalfJung`
Add ABI argument to `find_mir_or_eval_fn`
Add ABI argument for called function in `find_mir_or_eval_fn` and
`call_extra_fn`. Useful for comparing with expected ABI in interpreters.
Related to [miri/1631](https://github.com/rust-lang/miri/issues/1631)
r? `@RalfJung`
Use standard formatting for "rust-call" ABI message
Nearly all error messages start with a lowercase letter and don't use
articles - instead they refer to the plural case.
resolve: Scope visiting doesn't need an `Ident`
Resolution scope visitor (`fn visit_scopes`) currently takes an `Ident` parameter, but it doesn't need a full identifier, or even its span, it only needs the `SyntaxContext` part.
The `SyntaxContext` part is necessary because scope visitor has to jump to macro definition sites, so it has to be directed by macro expansion information somehow.
I think it's clearer to pass only the necessary part.
Yes, usually visiting happens as a part of an identifier resolution, but in cases like collecting traits in scope (#80765) or collecting typo suggestions that's not the case.
r? `@matthewjasper`
Logging goes to stderr, not stdout, so we should base our automated
detection on stderr instead of stdout.
Thanks to Ralf Jung for noticing and reporting the bug!
Add ABI argument for called function in `find_mir_or_eval_fn` and
`call_extra_fn`. Useful for comparing with expected ABI in interpreters.
Related to [miri/1631](https://github.com/rust-lang/miri/issues/1631)
Allow #[rustc_builtin_macro = "name"]
This adds the option of specifying the name of a builtin macro in the `#[rustc_builtin_macro]` attribute: `#[rustc_builtin_macro = "name"]`.
This makes it possible to have both `std::panic!` and `core::panic!` as a builtin macro, by using different builtin macro names for each. This is needed to implement the edition-specific behaviour of the panic macros of RFC 3007.
Also removes `SyntaxExtension::is_derive_copy`, as the macro name (e.g. `sym::Copy`) is now tracked and provides that information directly.
r? ``@petrochenkov``
Use correct span for structured suggestion
On structured suggestion for `let` -> `const` and `const` -> `let`, use
a proper `Span` and update tests to check the correct application.
Follow up to #80012.
Improve core::ptr::drop_in_place debuginfo
* Use span of the dropped type as function span when possible.
* Rename symbol from `core::ptr::drop_in_place::$hash` to `{{drop}}::<$TYPE>::$hash`.
Fixes#77465
(I haven't yet updated the tests)
Rustdoc: Fix macros 2.0 and built-in derives being shown at the wrong path
Fixes#74355
- ~~waiting on author + draft PR since my code ought to be cleaned up _w.r.t._ the way I avoid the `.unwrap()`s:~~
- ~~dummy items may avoid the first `?`,~~
- ~~but within the module traversal some tests did fail (hence the second `?`), meaning the crate did not possess the exact path of the containing module (`extern` / `impl` blocks maybe? I'll look into that).~~
r? `@jyn514`
Per https://github.com/rust-lang/rust/issues/80832, using
--pretty=expanded and --remap-path-prefix results in an ICE.
This is becasue the session source files table is stored in remapped
form, whereas --pretty-expanded looks up unremapped files. This remaps
the path prefixes before lookup.
Don't mark `force_query_with_job` as `inline(always)`
It's rather large, and using `inline(always)` forces it to be recompiled
in each calling crate. Hopefully this change will help with #65031. I intentionally only removed inline from `force_query_with_job` because the other functions are tiny and I wanted to measure this change on its own.
This may conflict with https://github.com/rust-lang/rust/issues/78780. I am not sure if it will hurt or help.
cc `@cjgillot`
Before 2021, this was a breaking change, as std::panic and core::panic
are different. In edition 2021 they will be identical, making it
possible again to apply proper hygiene here.
Use an empty `TokenCursorFrame` stack when capturing tokens
We will never need to pop past our starting frame during token
capturing. Using an empty stack allows us to avoid pointless heap
allocations/deallocations.
This makes it possible to have both std::panic and core::panic as a
builtin macro, by using different builtin macro names for each.
Also removes SyntaxExtension::is_derive_copy, as the macro name (e.g.
sym::Copy) is now tracked and provides that information directly.
Make target-cpu=native detect individual features
This PR makes target-cpu=native check for and enable/disable individual features instead of detecting and targeting a CPU by name. This brings the flag's behavior more in line with clang and gcc and ensures that the host actually supports each feature that we are compiling for.
This should resolve issues with miscompilations on e.g. "Haswell" Pentiums and Celerons that lack support for AVX, and also enable support for `aes` on Broadwell processors that support it. It should also resolve issues with failing to detect feature support in newer CPUs that aren't yet known by LLVM (see: #80633).
Fixes#54688Fixes#48464Fixes#38218
We will never need to pop past our starting frame during token
capturing. Using an empty stack allows us to avoid pointless heap
allocations/deallocations.
PR #73708 added a more detailed explanation of move errors that occur
due to a call to a method that takes `self`. This PR extends that logic
to work when a move error occurs due to a method call in the previous
iteration of a loop.
Access query (DepKind) metadata through fields
This refactors the access to query definition metadata (attributes such as eval always, anon, has_params) and loading/forcing functions to generate a number of structs, instead of matching on the DepKind enum. This makes access to the fields cheaper to compile. Using a struct means that finding the metadata for a given query is just an offset away; previously the match may have been compiled to a jump table but likely not completely inlined as we expect here.
A previous attempt explored a similar strategy, but using trait objects in #78314 that proved less effective, likely due to higher overheads due to forcing dynamic calls and poorer cache utilization (all metadata is fairly densely packed with this PR).
Optimize away some `fs::metadata` calls.
This also eliminates a use of a `Path` convenience function, in support
of #80741, refactoring `std::path` to focus on pure data structures and
algorithms.
Make sure rust-call errors occur correctly for traits
Fixes#79669
Adds trait method resolution to the error, and adds UI tests to ensure it doesn't happen again. Opening as draft because I'm getting weird link errors from unrelated code on my machine, and want to see what CI thinks.
On structured suggestion for `let` -> `const` and `const` -> `let`, use
a proper `Span` and update tests to check the correct application.
Follow up to #80012.
rustc_parse: Better spans for synthesized token streams
I think using the nonterminal span for synthesizing its tokens is a better approximation than using `DUMMY_SP` or the attribute span like #79472 did in `expand.rs`.
r? `@Aaron1011`
MIR Inline is incompatible with coverage
Fixes: #80060
Fixed by disabling inlining if `-Zinstrument-coverage` is set.
The PR also adds additional use cases to the coverage test for doctests.
r? `@wesleywiser`
cc: `@tmandry`
Make `ExpnData` fields `krate` and `orig_id` private
These fields are only used by hygiene serialized, and should not be
accessed by anything outside of `rustc_span`.
Optimize DST field access
For
struct X<T: ?Sized>(T)
struct Y<T: ?Sized>(u8, T)
the offset of the unsized field is
0
mem::align_of_val(&self.1)
respectively. This patch changes the expression used to compute these
offsets so that the optimizer can perform this optimization.
Consider
```rust
fn f(x: &X<dyn Any>) -> &dyn Any {
&x.0
}
```
Before:
```asm
test:
movq %rsi, %rdx
movq 16(%rsi), %rax
leaq -1(%rax), %rcx
negq %rax
andq %rcx, %rax
addq %rdi, %rax
retq
```
After:
```asm
test:
movq %rsi, %rdx
movq %rdi, %rax
retq
```
Optimize away a `fs::metadata` call.
This also eliminates a use of a `Path` convenience function, in support
of #80741, refactoring `std::path` to focus on pure data structures and
algorithms.
* Rename `ModuleData.normal_ancestor_id` to `nearest_parent_mod`
`normal_ancestor_id` is a very confusing name if you don't already
understand what it means. Adding docs helps, but using a clearer and
more obvious name is also important.
* Rename `Resolver::nearest_mod_parent` to `nearest_parent_mod`
* Add more docs
This also eliminates a use of a `Path` convenience function, in support
of #80741, refactoring `std::path` to focus on pure data structures and
algorithms.
This also eliminates a use of a `Path` convenience function, in support
of #80741, refactoring `std::path` to focus on pure data structures and
algorithms.
I noticed that the Size::bits function is called in many places,
and is inlined into them. On x86_64-pc-windows-msvc, this function
is inlined 527 times, and compiled separately (non-inlined) 3 times.
Each of those inlined calls contains code that panics. This commit
moves the `panic!` call into a separate function and marks that
function with `#[cold]`.
This reduces binary size by 24 KB. By itself, that's not a substantial
reduction. However, changes like this often reduce pressure on
instruction-caches, since it reduces the amount of code that is inlined
into hot code paths. Or more precisely, it removes cold code from hot
cache lines. It also removes all conditionals from Size::bits(),
which is called in many places.
Move variable into the only branch where it is relevant
At the `if` branch `filter` (the `let` binding) is `None` iff `filter` (the parameter) was `None`.
We can branch on the parameter, move the binding into the `if`, and the complexity of handling
`Option<Option<_>` largely dissolves.
`@rustbot` modify labels +C-cleanup +T-compiler
Note: I have no idea how hot this code is. If this method frequently gets called with a `None` filter, there might be a small perf improvement.
Add check for `[T;N]`/`usize` mismatch in astconv
Helps clarify the issue in #80506
by adding a specific check for mismatches between [T;N] and usize.
r? `@lcnr`
use PlaceRef more consistently instead of loosely coupled local+projection
Instead of working directly with the `projections` array, use `iter_projections` and `last_projection`. This avoids having to construct new `PlaceRef` from the pieces everywhere.
I only did this for a few files, to see how people think about this. If y'all are happy with this, I'll open an E-mentor issue to complete this. I grepped for `Place::ty_from` to find the places that need adjusting -- this could miss some, but I am not sure what else to grep for.
Add note to non-exhaustive match on reference to empty
Rust prints "type `&A` is non-empty" even is A is empty.
This is the intended behavior, but can be confusing.
This commit adds a note to non-exhaustive pattern errors if they are a
reference to something uninhabited.
I did not add tests to check that the note is not shown for
non-references or inhabited references, because this is already done
in other tests.
Maybe the added test is superfluous, because
`always-inhabited-union-ref` already checks for this case.
This does not handle &&Void or &&&void etc. I could add those as special
cases as well and ignore people who need quadruple
references.
Fixes#78123
Allow references to interior mutable data behind a feature gate
supercedes #80373 by simply not checking for interior mutability on borrows of locals that have `StorageDead` and thus can never be leaked to the final value of the constant
tracking issue: https://github.com/rust-lang/rust/issues/80384
r? `@RalfJung`
Rust prints "type `&A` is non-empty" even is A is empty.
This is the intended behavior, but can be confusing.
This commit adds a note to non-exhaustive pattern errors if they are a
reference to something uninhabited.
I did not add tests to check that the note is not shown for
non-references or inhabited references, because this is already done
in other tests.
Maybe the added test is superfluous, because
`always-inhabited-union-ref` already checks for this case.
This does not handle &&Void or &&&void etc. I could add those as special
cases as well and ignore people who need quadruple
references.
Fixes#78123
Clean up in `each_child_of_item`
This PR hopes to eliminate some of the surprising elements I encountered while reading the function.
- `macros_only` is checked against inside the loop body, but if it is `true`, the loop is skipped anyway
- only query `span` when relevant
- no need to allocate attribute vector
At the `if` branch `filter` (the `let` binding) is `None` iff `filter` (the parameter) was `None`.
We can branch on the parameter, move the binding into the `if`, and the complexity of handling
`Option<Option<_>` largely dissolves.
`const_generics_defaults`: don't ICE in the unimplemented parts
The thought was that we could use `todo!`s to ensure we wouldn't forget to implement parts of the experimental gate.
However, that can also lead to a suboptimal experience for users as shown in #80589 having both the error/warning about the experimental feature, and the ICE.
Fixes#80589
r? `@varkor`
improve unconditional_panic description
The fact that the lint is triggered by the ConstProp pass is an implementation detail, I do not think that this should be mentioned in the description.
Cc `@oli-obk` `@ehuss`
rustc_serialize: specialize opaque encoding and decoding of some u8 sequences
This specializes encoding and decoding of some contiguous u8 sequences to use a more efficient implementation. The default implementations process each u8 individually, but that isn't necessary for the opaque encoder and decoder. The opaque encoding for u8s is a no-op, so we can just copy entire sequences as-is, rather than process them byte by byte.
This also changes some encode and decode implementations for contiguous sequences to forward to the slice and vector implementations, so that they can take advantage of the new specialization when applicable.
Enable ASan, TSan, UBSan for aarch64-apple-darwin.
I confirmed ASan, TSan, UBSan all work for me locally with `clang` on my new Macbook Air.
~This requires https://github.com/rust-lang/llvm-project/pull/86~
- Adds optional default values to const generic parameters in the AST
and HIR
- Parses these optional default values
- Adds a `const_generics_defaults` feature gate
Add edition 2021.
🎆 Happy new ~~year~~ Rust. 🍾
This adds --edition=2021, and updates suggestions about 2018 to say "2018 *or later*".
Related Cargo PR: https://github.com/rust-lang/cargo/pull/8922
---
Edit: This adds the new edition as *unstable*. Without `-Z unstable-options`, `--edition=2021` results in:
```
$ rustc --edition=2021
error: edition 2021 is unstable and only available with -Z unstable-options.
```
Update and improve `rustc_codegen_{llvm,ssa}` docs
Fixes#75342.
These docs were very out of date and misleading. They even said that
they codegen'd the *AST*!
For some reason, the `rustc_codegen_ssa::base` docs were exactly
identical to the `rustc_codegen_llvm::base` docs. They didn't really
make sense, because they had LLVM-specific information even though
`rustc_codegen_ssa` is supposed to be somewhat generic. So I removed
them as they were misleading.
r? ``@pnkfelix`` maybe?
Add `#[track_caller]` to `bug!` and `register_renamed`
Before:
```
thread 'rustc' panicked at 'compiler/rustc_lint/src/context.rs:267:18: invalid lint renaming of broken_intra_doc_links to rustdoc::broken_intra_doc_links', compiler/rustc_middle/src/util/bug.rs:34:26
```
After:
```
thread 'rustc' panicked at 'src/librustdoc/core.rs:455:24: invalid lint renaming of broken_intra_doc_links to rustdoc::broken_intra_doc_links', compiler/rustc_middle/src/util/bug.rs:35:26
```
The reason I added it to `register_renamed` too is that any panic in
that function will be the caller's fault.
Rollup of 9 pull requests
Successful merges:
- #78934 (refactor: removing library/alloc/src/vec/mod.rs ignore-tidy-filelength)
- #79479 (Add `Iterator::intersperse`)
- #80128 (Edit rustc_ast::ast::FieldPat docs)
- #80424 (Don't give an error when creating a file for the first time)
- #80458 (Some Promotion Refactoring)
- #80488 (Do not create dangling &T in Weak<T>::drop)
- #80491 (Miri: make size/align_of_val work for dangling raw ptrs)
- #80495 (Rename kw::Invalid -> kw::Empty)
- #80513 (Add regression test for #80062)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Some Promotion Refactoring
Clean up promotion a bit:
* factor out some common code
* more exhaustive matches
This *should* not break anything... the only potentially-breaking change is that `BorrowKind::Shallow | BorrowKind::Unique` are now rejected for internal references.
r? ``@oli-obk``
Rollup of 7 pull requests
Successful merges:
- #80185 (Fix ICE when pointing at multi bytes character)
- #80260 (slightly more typed interface to panic implementation)
- #80311 (Improvements to NatVis support)
- #80337 (Use `desc` as a doc-comment for queries if there are no doc comments)
- #80381 (Revert "Cleanup markdown span handling")
- #80492 (remove empty wraps, don't return Results from from infallible functions)
- #80509 (where possible, pass slices instead of &Vec or &String (clippy::ptr_arg))
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
remove empty wraps, don't return Results from from infallible functions
This makes code easier to understand because it is more obvious when a function actually can't fail (return Err or None)
Make functions that only ever return Some(x), return x directly
Remove return type from functions that return Option<(), Err> but would only ever return Ok(()).
Found with `clippy::unnecessary_wraps`
remove unused return type of dropck::check_drop_obligations()
don't wrap return type in Option in get_macro_by_def_id() since we would always return Some(..)
remove redundant return type of back::write::optimize()
don't Option-wrap return type of compute_type_parameters() since we always return Some(..)
don't return empty Result in assemble_generator_candidates()
don't return empty Result in assemble_closure_candidates()
don't return empty result in assemble_fn_pointer_candidates()
don't return empty result in assemble_candidates_from_impls()
don't return empty result in assemble_candidates_from_auto_impls()
don't return emtpy result in assemble_candidates_for_trait_alias()
don't return empty result in assemble_builtin_bound_candidates()
don't return empty results in assemble_extension_candidates_for_traits_in_scope() and assemble_extension_candidates_for_trait()
remove redundant wrapping of return type of StripItem::strip() since it always returns Some(..)
remove unused return type of assemble_extension_candidates_for_all_traits()
Edit rustc_span documentation
Various changes to the `rustc_span` docs, including the following:
- Additions to top-level docs
- Edits to the source_map module docs
- Edits to documentation for `Span` and `SpanData`
- Added intra-docs links
- Documentation for Levenshtein distances
- Fixed missing punctuation
Lint on redundant trailing semicolon after item
We now lint on code like this:
```rust
fn main() {
fn foo() {};
struct Bar {};
}
```
Previously, this caused warnings in Cargo, so it was disabled.
Replace pretty-print/compare/retokenize hack with targeted workarounds
Based on https://github.com/rust-lang/rust/pull/78296
cc https://github.com/rust-lang/rust/issues/43081
The 'pretty-print/compare/retokenize' hack is used to try to avoid passing an outdated `TokenStream` to a proc-macro when the underlying AST is modified in some way (e.g. cfg-stripping before derives). Unfortunately, retokenizing throws away spans (including hygiene information), which causes issues of its own. Every improvement to the accuracy of the pretty-print/retokenize comparison has resulted in non-trivial ecosystem breakage due to hygiene changes. In extreme cases, users deliberately wrote unhygienic `macro_rules!` macros (likely because they did not realize that the compiler's behavior was a bug).
Additionaly, the comparison between the original and pretty-printed/retoknized token streams comes at a non-trivial runtime cost, as shown by https://github.com/rust-lang/rust/pull/79338
This PR removes the pretty-print/compare/retokenize logic from `nt_to_tokenstream`. We only discard the original `TokenStream` under two circumstances:
* Inner attributes are used (detected by examining the AST)
* `cfg`/`cfg_attr` processing modifies the AST. This is detected by making the visitor update a flag when it performs a modification, instead of trying to detect the modification after-the-fact. Note that a 'matching' `cfg` (e.g. `#[cfg(not(FALSE)]`) does not actually get removed from the AST, allowing us to preserve the original `TokenStream`.
In all other cases, we preserve the original `TokenStream`.
This could use a bit of refactoring/renaming - opening for a Crater run.
r? `@ghost`
Before:
```
thread 'rustc' panicked at 'compiler/rustc_lint/src/context.rs:267:18: invalid lint renaming of broken_intra_doc_links to rustdoc::broken_intra_doc_links', compiler/rustc_middle/src/util/bug.rs:34:26
```
After:
```
thread 'rustc' panicked at 'src/librustdoc/core.rs:455:24: invalid lint renaming of broken_intra_doc_links to rustdoc::broken_intra_doc_links', compiler/rustc_middle/src/util/bug.rs:35:26
```
The reason I added it to `register_renamed` too is that any panic in
that function will be the caller's fault.
- Replace {} with the stringified expr
Giant thank you to `@danielhenrymantilla` for figuring out how to make
this work ❤️
- Note that this is just an approximation and it would be better to add
a doc-comment
Document `InferTy` & co.
I finally figured out what `TyVid` means! The name is quite opaque, so I
decided to document it and related types.
I don't know that much about `InferTy` & co., but I was able to *infer*
( :) ) from the names and what I know generally about type inference to
add some basic documentation.
Rollup of 11 pull requests
Successful merges:
- #79662 (Move some more code out of CodegenBackend::{codegen_crate,link})
- #79815 (Update RELEASES.md for 1.49.0)
- #80284 (Suggest fn ptr rather than fn item and suggest to use `Fn` trait bounds rather than the unique closure type in E0121)
- #80331 (Add more comments to trait queries)
- #80344 (use matches!() macro in more places)
- #80353 (BTreeMap: test split_off (and append) more thoroughly)
- #80362 (Document rustc_macros on nightly-rustc)
- #80399 (Remove FIXME in rustc_privacy)
- #80408 (Sync rustc_codegen_cranelift)
- #80411 (rustc_span: Remove `Symbol::with`)
- #80434 (bootstrap: put the component name in the tarball temp dir path)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Sync rustc_codegen_cranelift
The highlight of this sync are two JIT mode improvements. The first is that it is now possible to use JIT mode when using `-Zcodegen-backend` instead of the custom driver using `-Cllvm-args=mode=jit`. The second one is a new JIT mode that lazily compiles functions when they are called the first time: https://github.com/bjorn3/rustc_codegen_cranelift/pull/1120
In addition this includes a few small runtime performance improvements and various fixes for rustc changes that didn't cause compilation to fail.
r? ``@ghost``
``@rustbot`` label +A-codegen +A-cranelift +T-compiler
Suggest fn ptr rather than fn item and suggest to use `Fn` trait bounds rather than the unique closure type in E0121
Previously, using `_` as a return type in a function that returned a function/closure would provide a diagnostic that would cause a papercut. For example:
```rust
fn f() -> i32 { 0 }
fn fn_ptr() -> _ { f }
fn closure() -> _ { || 0 }
```
would result in this diagnostic:
```rust
error[E0121]: the type placeholder `_` is not allowed within types on item signatures
--> <anon>:2:16
|
2 | fn fn_ptr() -> _ { f }
| ^
| |
| not allowed in type signatures
| help: replace with the correct return type: `fn() -> i32 {f}`
error[E0121]: the type placeholder `_` is not allowed within types on item signatures
--> <anon>:3:17
|
3 | fn closure() -> _ { || 0 }
| ^
| |
| not allowed in type signatures
| help: replace with the correct return type: `[closure@<anon>:3:21: 3:25]`
error: aborting due to 2 previous errors
For more information about this error, try `rustc --explain E0121`.
```
As can be seen, it was suggested to use the function definition return type `fn() -> i32 { f }` which is not valid syntax as a return type. Additionally, closures cause a papercut as unique closure types (notated in this case as `[closure@<anon>:3:21: 3:25]`) are not valid syntax either.
Instead, this PR implements this version of the diagnostic (this example is for the same code featured above):
```rust
error[E0121]: the type placeholder `_` is not allowed within types on item signatures
--> <anon>:2:16
|
2 | fn fn_ptr() -> _ { f }
| ^
| |
| not allowed in type signatures
| help: replace with the correct return type: `fn() -> i32`
error[E0121]: the type placeholder `_` is not allowed within types on item signatures
--> <anon>:3:17
|
3 | fn closure() -> _ { || 0 }
| ^ not allowed in type signatures
|
= help: consider using an `Fn`, `FnMut`, or `FnOnce` trait bound
= note: for more information on `Fn` traits and closure types, see https://doc.rust-lang.org/book/ch13-01-closures.html
error: aborting due to 2 previous errors
For more information about this error, try `rustc --explain E0121`.
```
As can be seen in this diagnostic, the papercut for returning a function item is fixed by suggesting the usage of a function pointer as the return type. As for closures, it's suggested to use an `Fn`, `FnMut`, or `FnOnce` trait bound (with further reading on closures and `Fn` traits in *The Book* for beginners). I did not implement a suggestion to use `impl Fn() -> i32` syntax as that was out-of-scope for my abilities at the moment, therefore someone in the future may want to implement that. Also, it's possible to use either `impl Trait` syntax, generics, or generics with a `where` clause, and some users may not want to use `impl Trait` syntax for their own reasons.
This PR fixes#80179.
Prevent caching normalization results with a cycle
When normalizing a projection which results in a cycle, we would cache the result of `project_type` without the nested obligations (because they're not needed for inference). This would result in the nested obligations only being handled once in fulfill, which would avoid the cycle error. `get_paranoid_cache_value_obligation` used to add an obligation that resulted in a cycle in this case previously, but was removed by #73905.
This PR makes the projection cache not cache the value of a projection if it was ever normalized in a cycle (except in a snapshot that's rolled back).
Fixes#79714.
r? `@nikomatsakis`
Various changes to the `rustc_span` docs, including the following:
- Additions to top-level docs
- Edits to the source_map module docs
- Edits to documentation for `Span` and `SpanData`
- Added intra-docs links
- Documentation for Levenshtein distances
- Fixed missing punctuation
validate promoteds
Turn on const-value validation for promoteds. This is made possible now that https://github.com/rust-lang/rust/issues/67534 is resolved.
I don't think this is a breaking change. We don't promote any unsafe operation any more (since https://github.com/rust-lang/rust/pull/77526 landed). We *do* promote `const fn` calls under some circumstances (in `const`/`static` initializers), but union field access and similar operations are not allowed in `const fn`. So now is a perfect time to add this check. :D
r? `@oli-obk`
Fixes https://github.com/rust-lang/rust/issues/67465
Exclude unnecessary info from CodegenResults
`foreign_module` and `wasm_import_module` are not needed for linking, and hence can be removed from CodegenResults.
Fixes#77857
Rename rustc_middle::lint::LintSource
Rename [`rustc_middle::lint::LintSource`](https://doc.rust-lang.org/nightly/nightly-rustc/rustc_middle/lint/enum.LintSource.html) to `rustc_middle::lint::LintLevelSource`.
This enum represents the source of a *lint level*, not a lint. This should improve code readability.
Update: Also documents `rustc_middle::lint::LevelSource` to clarify.
Implemented a compiler diagnostic for move async mistake
Fixes#79694
First time contributing, so I hope I'm doing everything right.
(If not, please correct me!)
This code performs a check when a move capture clause is parsed. The check is to detect if the user has reversed the async move keywords and to provide a diagnostic with a suggestion to fix it.
Checked code:
```rust
fn main() {
move async { };
}
```
Previous output:
```txt
PS C:\Repos\move_async_test> cargo build
Compiling move_async_test v0.1.0 (C:\Repos\move_async_test)
error: expected one of `|` or `||`, found keyword `async`
--> src\main.rs:2:10
|
2 | move async { };
| ^^^^^ expected one of `|` or `||`
error: aborting due to previous error
error: could not compile `move_async_test`
```
New output:
```txt
PS C:\Repos\move_async_test> cargo +dev build
Compiling move_async_test v0.1.0 (C:\Repos\move_async_test)
error: the order of `move` and `async` is incorrect
--> src\main.rs:2:13
|
2 | let _ = move async { };
| ^^^^^^^^^^
|
help: try switching the order
|
2 | let _ = async move { };
| ^^^^^^^^^^
error: aborting due to previous error
error: could not compile `move_async_test`
```
Is there a file/module where these kind of things are tested?
Would love some feedback 😄
Remap instrument-coverage line numbers in doctests
This uses the `SourceMap::doctest_offset_line` method to re-map line
numbers from doctests. Remapping columns is not yet done, and rustdoc
still does not output the correct filename when running doctests in a
workspace.
Part of #79417 although I dont consider that fixed until both filenames
and columns are mapped correctly.
r? `@richkadel`
I might jump on zulip the comming days. Still need to figure out how to properly write tests for this, and deal with other doctest issues in the meantime.
Rework beautify_doc_string so that it returns a Symbol instead of a String
This commit comes from https://github.com/rust-lang/rust/pull/80261, the goal here is to inspect the impact on performance of this change on its own.
The idea of rewriting `beautify_doc_string` is to not go through `String` if we don't need to update the doc comment to be able to keep the original `Symbol` and also to have better performance.
r? `@jyn514`
rustc_query_system: reduce dependency graph memory usage
This change implements, at a high level, two space optimizations to the dependency graph.
The first optimization is sharing graph data with the previous dependency graph. Whenever we intern a node, we know whether that node is new (not in the previous graph) or not, and if not, the color of the node in the previous graph.
Red and green nodes have their `DepNode` present in the previous graph, so for that piece of node data, we can just store the index of the node in the previous graph rather than duplicate the `DepNode`. Green nodes additionally have the the same result `Fingerprint`, so we can avoid duplicating that too. Finally, we distinguish between "light" and "dark" green nodes, where the latter are nodes that were marked green because all of their dependencies were marked green. These nodes can additionally share edges with the previous graph, because we know that their set of dependencies is the same (technically, light green and red nodes can have the same dependencies too, but we don't try to figure out whether or not that's the case).
Also, some effort is made to pack data tightly, and to avoid storing `DepNode`s as map keys more than once.
The second optimization is storing edges in a more compact representation, as in the `SerializedDepGraph`, that is, in a single vector, rather than one `EdgesVec` per node. An `EdgesVec` is a `SmallVec` with an inline buffer for 8 elements. Each `EdgesVec` is, at minimum, 40 bytes, and has a per-node overhead of up to 40 bytes. In the ideal case of exactly 8 edges, then 32 bytes are used for edges, and the overhead is 8 bytes. But most of the time, the overhead is higher.
In contrast, using a single vector to store all edges, and having each node specify its start and end elements as 4 byte indices into the vector has a constant overhead of 8 bytes--the best case scenario for the per-node `EdgesVec` approach.
The downside of this approach is that `EdgesVec`s built up during query execution have to be copied into the vector, whereas before, we could just take ownership over them. However, we mostly make up for this because the single vector representation enables a more efficient implementation of `DepGraph::serialize`.
These docs were very out of date and misleading. They even said that
they codegen'd the *AST*!
For some reason, the `rustc_codegen_ssa::base` docs were exactly
identical to the `rustc_codegen_llvm::base` docs. They didn't really
make sense, because they had LLVM-specific information even though
`rustc_codegen_ssa` is supposed to be somewhat generic. So I removed
them as they were misleading.
Turn helper method into a closure
`replace_prefix` is currently implemented as a method but has no real relation
to the struct it is implemented on. Turn it into a closure and move it into the
only method from which it is called.
`@rustbot` modify labels +C-cleanup +T-compiler
r? `@lcnr`
rustc_span: Provide a reserved identifier check for a specific edition
while keeping edition evaluation lazy because it may be expensive.
Needed for https://github.com/rust-lang/rust/pull/80226.
Add module-level docs to rustc_middle::ty
I thought it would be nice to point out `Ty` and `TyCtxt` on the module page, and link out to the [rustc-dev-guide chapter](https://rustc-dev-guide.rust-lang.org/ty.html).
docs: Edit rustc_middle::ty::query::on_disk_cache
Expand abbreviations for "incremental compliation".
Also added the word "to" to the description of CacheEncoder.
Reduce memory consumption by sharing the previous dependency graph's
edges with the current graph when it is known to be valid to do so. It
is known to be valid whenever we mark a node green because all of its
dependencies were green. It is *not* known to be valid when we mark a
node green because we re-executed its query and its result was the same
as in the previous compilation session. In that case, the dependency set
might have changed (we don't try to determine whether or not it changed
and whether or not we can share).
Reduce memory consumption by taking advantage of red/green algorithm
properties to share the previous dependency graph's node data with the
current graph instead of storing node data redundantly. Red nodes can
share the `DepNode`, and green nodes can share the `DepNode` and
`Fingerprint`. Edges will be shared when possible in a later change.
Clarify constructor splitting in exhaustiveness checking
I reworked the explanation of the algorithm completely to make it properly account for the various extensions we've added. This includes constructor splitting, which was previously not clearly included in the algorithm. This makes wildcards less magical; I added some detailed examples; and this distinguishes clearly between constructors that only make sense in patterns (like ranges) and those that make sense for values (like `Some`). This reformulation had been floating around in my mind for a while, and I'm quite happy with how it turned out. Let me know how you feel about it.
I also factored out all three cases of splitting (wildcards, ranges and slices) into dedicated structs to encapsulate the complicated bits.
I measured no perf impact but I don't trust my local measurements for refactors since https://github.com/rust-lang/rust/pull/79284.
r? `@varkor`
`@rustbot` modify labels: +A-exhaustiveness-checking
rustc_query_system: explicitly register reused dep nodes
Register nodes that we've reused from the previous session explicitly
with `OnDiskCache`. Previously, we relied on this happening as a side
effect of accessing the nodes in the `PreviousDepGraph`. For the sake of
performance and avoiding unintended side effects, register explictily.
`replace_prefix` is currently implemented as a method but has no real relation
to the struct it is implemented on. Turn it into a closure and move it into the
only method from which it is called.
Rename `overlapping_patterns` lint
As discussed in https://github.com/rust-lang/rust/issues/65477. I also tweaked a few things along the way.
r? `@varkor`
`@rustbot` modify labels: +A-exhaustiveness-checking
Reserve necessary space for params in generics_of
Always reserve space for the exact number of generic parameters we need in generics_of. As far as I can see, the default is 0/4 elements based on has_self, and the vector grows on after that.
Acknowledge that `[CONST; N]` is stable
When `const_in_array_repeat_expressions` (RFC 2203) got unstably implemented as part of https://github.com/rust-lang/rust/pull/61749, accidentally, the special case of repeating a *constant* got stabilized immediately. That is why the following code works on stable:
```rust
const EMPTY: Vec<i32> = Vec::new();
pub const fn bar() -> [Vec<i32>; 2] {
[EMPTY; 2]
}
fn main() {
let x = bar();
}
```
In contrast, if we had written `[expr; 2]` for some expression that is not *literally* a constant but could be evaluated at compile-time (e.g. `(EMPTY,).0`), this would have failed.
We could take back this stabilization as it was clearly accidental. However, I propose we instead just officially accept this and stabilize a small subset of RFC 2203, while leaving the more complex case of general expressions that could be evaluated at compile-time unstable. Making that case work well is pretty much blocked on inline `const` expressions (to avoid relying too much on [implicit promotion](https://github.com/rust-lang/const-eval/blob/master/promotion.md)), so it could take a bit until it comes to full fruition. `[CONST; N]` is an uncontroversial subset of this feature that has no semantic ambiguities, does not rely on promotion, and basically provides the full expressive power of RFC 2203 but without the convenience (people have to define constants to repeat them, possibly using associated consts if generics are involved).
Well, I said "no semantic ambiguities", that is only almost true... the one point I am not sure about is `[CONST; 0]`. There are two possible behaviors here: either this is equivalent to `let x = CONST; [x; 0]`, or it is a NOP (if we argue that the constant is never actually instantiated). The difference between the two is that if `CONST` has a destructor, it should run in the former case (but currently doesn't, due to https://github.com/rust-lang/rust/issues/74836); but should not run if it is considered a NOP. For regular `[x; 0]` there seems to be consensus on running drop (there isn't really an alternative); any opinions for the `CONST` special case? Should this instantiate the const only to immediately run its destructors? That seems somewhat silly to me. After all, the `let`-expansion does *not* work in general, for `N > 1`.
Cc `@rust-lang/lang` `@rust-lang/wg-const-eval`
Cc https://github.com/rust-lang/rust/issues/49147
Ran the tidy check
Following the diagnostic guide better
Diagnostic generation is now relegated to its own function in the diagnostics module.
Added tests
Fixed the ui test
Fix pretty printing an AST representing `&(mut ident)`
The PR fixes a misguiding help diagnostic in the parser that I reported in #80186. I discovered that the parsers recovery and reporting logic was correct but the pretty printer produced wrong code for the example. (Details in https://github.com/rust-lang/rust/issues/80186#issuecomment-748498676)
Example:
```rust
#![allow(unused_variables)]
fn main() {
let mut &x = &0;
}
```
The AST fragment
`PatKind::Ref(PatKind::Ident(BindingMode::ByValue(Mutability::Mut), ..), Mutability::Not)`
was printed to be `&mut ident`. But this wouldn't round trip through parsing again, because then it would be:
`PatKind::Ref(PatKind::Ident(BindingMode::ByValue(Mutability::Not), ..), Mutability::Mut)`
Now the pretty-printer prints `&(mut ident)`. Reparsing that code results in the AST fragment
`PatKind::Ref(PatKind::Paren(PatKind::Ident(BindingMode::ByValue(Mutability::Mut), ..)), Mutability::Not)`
which I think should behave like the original pattern.
Old diagnostic:
```
error: `mut` must be attached to each individual binding
--> src/main.rs:3:9
|
3 | let mut &x = &0;
| ^^^^^^ help: add `mut` to each binding: `&mut x`
|
= note: `mut` may be followed by `variable` and `variable @ pattern`
```
New diagnostic:
```
error: `mut` must be attached to each individual binding
--> src/main.rs:3:9
|
3 | let mut &x = &0;
| ^^^^^^ help: add `mut` to each binding: `&(mut x)`
|
= note: `mut` may be followed by `variable` and `variable @ pattern`
```
Fixes#80186
Fixes#80233
We already have logic in `evaluate_predicates` that tries to add
unimplemented predicates to our `ParamEnv`. Trying to add a predicate
that already holds can lead to errors later on, since projection
will prefer trait candidates from the `ParamEnv` to predicates from an
impl.
Minor cleanups in LateResolver
- Avoid calculating hash twice
- Avoid creating a closure in every iteration of a loop
- Reserve space for path in advance
- Some readability changes
Handle desugaring in impl trait bound suggestion
Fixes#79843.
When an associated type of a generic function parameter needs extra bounds, the diagnostics may suggest replacing an `impl Trait` with a named type parameter so that it can be referenced in the where clause. On stable and nightly, the suggestion can be malformed, for instance transforming:
```rust
async fn run(_: &(), foo: impl Foo) -> std::io::Result<()>
```
Into:
```rust
async fn run(_: &, F: Foo(), foo: F) -> std::io::Result<()> where <F as Foo>::Bar: Send
^^^^^^^^ ^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^
```
Where we want something like:
```rust
async fn run<F: Foo>(_: &(), foo: F) -> std::io::Result<()> where <F as Foo>::Bar: Send
^^^^^^^^ ^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^
```
The problem is that the elided lifetime of `&()` is added as a generic parameter when desugaring the async fn; the suggestion code sees this as an existing generic parameter and tries to use its span as an anchor to inject `F` into the parameter list. There doesn't seem to be an entirely principled way to check which generic parameters in the HIR were explicitly named in the source, so this commit changes the heuristics when generating the suggestion to only consider type parameters whose spans are contained within the span of the `Generics` when determining how to insert an additional type parameter into the declaration. (And to be safe it also excludes parameters whose spans are marked as originating from desugaring, although that doesn't seem to handle this elided lifetime.)
Edit rustc_middle docs
Re-word doc comment for rustc_middle::hir::place::Projection.
Also adds:
- Missing end stop punctuation, and
- Documentation links to `rustc_middle::mir::Place`.
When normalizing a projection which results in a cycle, we would
cache the result of `project_type` without the nested obligations
(because they're not needed for inference). This would result in
the nested obligations only being handled once in fulfill, which
would avoid the cycle error.
Fixes#79714, a regresion from #79305 caused by the removal of
`get_paranoid_cache_value_obligation`.
Turn quadratic time on number of impl blocks into linear time
Previously, if you had a lot of inherent impl blocks on a type like:
```Rust
struct Foo;
impl Foo { fn foo_1() {} }
// ...
impl Foo { fn foo_100_000() {} }
```
The compiler would be very slow at processing it, because
an internal algorithm would run in O(n^2), where n is the number
of impl blocks. Now, we add a new algorithm that allocates but
is faster asymptotically.
Comparing rustc nightly with a local build of rustc as of this PR (results in seconds):
| N | real time before | real time after |
| - | - | - |
| 4_000 | 0.57 | 0.46 |
| 8_000 | 1.31 | 0.84 |
| 16_000 | 3.56 | 1.69 |
| 32_000 | 10.60 | 3.73 |
I've tuned up the numbers to make the effect larger than the startup noise of rustc, but the asymptotic difference should hold for smaller n as well.
Note: current state of the PR omits error messages if there are other errors present already. For now, I'm mainly interested in a perf run to study whether this issue is present at all. Please queue one for this PR. Thanks!
Re-word doc comment for rustc_middle::hir::place::Projection.
Also adds:
- Missing end stop punctuation, and
- Documentation links to `rustc_middle::mir::Place`.
`PatKind::Ref(PatKind::Ident(BindingMode::ByValue(Mutability::Mut), ..), ..)`
is an AST representing `&(mut ident)`. It was errorneously printed as
`&mut ident` which reparsed into a syntactically different AST.
This affected help diagnostics in the parser.
Make BoundRegion have a kind of BoungRegionKind
Split from #76814
Also includes making `replace_escaping_bound_vars` only return `T`
Going to r? `@lcnr`
Feel free to reassign
or_patterns: implement :pat edition-specific behavior
cc #54883 `@joshtriplett`
This PR implements the edition-specific behavior of `:pat` wrt or-patterns, as determined by the crater runs and T-lang consensus in https://github.com/rust-lang/rust/issues/54883#issuecomment-745509090.
I believe this can unblock stabilization of or_patterns.
r? `@petrochenkov`
const_evaluatable_checked: fix occurs check
fixes#79615
this is kind of a hack because we use `TypeRelation` for both the `Generalizer` and the `ConstInferUnifier` but i am not sure if there is a useful way to disentangle this without unnecessarily duplicating some code.
The error in the added test is kind of unavoidable until we erase the unused substs of `ConstKind::Unevaluated`. We talked a bit about this in the cg lazy norm meeting (https://rust-lang.zulipchat.com/#narrow/stream/260443-project-const-generics/topic/lazy_normalization_consts)
For
struct X<T: ?Sized>(T)
struct Y<T: ?Sized>(u8, T)
the offset of the unsized field is
0
mem::align_of_val(&self.1)
respectively. This patch changes the expression used to compute these
offsets so that the optimizer can perform this optimization.
Consider
```rust
fn test(x: &X<dyn Any>) -> &dyn Any {
&x.0
}
```
Before:
```asm
test:
movq %rsi, %rdx
movq 16(%rsi), %rax
leaq -1(%rax), %rcx
negq %rax
andq %rcx, %rax
addq %rdi, %rax
retq
```
After:
```asm
test:
movq %rsi, %rdx
movq %rdi, %rax
retq
```
Improve and fix diagnostics of exhaustiveness checking
Primarily, this fixes https://github.com/rust-lang/rust/issues/56379. This also fixes incorrect interactions between or-patterns and slice patterns that I discovered while working on #56379. Those two examples show the incorrect diagnostics:
```rust
match &[][..] {
[true] => {}
[true // detected as unreachable but that's not true
| false, ..] => {}
_ => {}
}
match (true, None) {
(true, Some(_)) => {}
(false, Some(true)) => {}
(true | false, None | Some(true // should be detected as unreachable
| false)) => {}
}
```
I did not measure any perf impact. However, I suspect that [`616ba9f`](616ba9f9f7) should have a negative impact on large or-patterns. I'll see what the perf run says; I have optimization ideas up my sleeve if needed.
EDIT: I initially had a noticeable perf impact that I thought unavoidable. I then proceeded to avoid it x)
r? `@varkor`
`@rustbot` label +A-exhaustiveness-checking
Stabilize or_insert_with_key
Stabilizes the `or_insert_with_key` feature from https://github.com/rust-lang/rust/issues/71024. This allows inserting key-derived values when a `HashMap`/`BTreeMap` entry is vacant.
The difference between this and `.or_insert_with(|| ... )` is that this provides a reference to the key to the closure after it is moved with `.entry(key_being_moved)`, avoiding the need to copy or clone the key.
passes: prohibit invalid attrs on generic params
Fixes#78957.
This PR modifies the `check_attr` pass so that attribute placement on generic parameters is checked for validity.
r? `@lcnr`
Register nodes that we've reused from the previous session explicitly
with `OnDiskCache`. Previously, we relied on this happening as a side
effect of accessing the nodes in the `PreviousDepGraph`. For the sake of
performance and avoiding unintended side effects, register explictily.
This is elegant but a bit of a perf gamble. That said, or-patterns
rarely have many branches and it's easy to optimize or revert if we ever
need to. In the meantime simpler code is worth it.
Stop using intermediate macros in definition of symbols
Currently, the rustc_macros::symbols macro generates two
`macro_rules!` macros as its output. These two macros are
used in rustc_span/src/symbol.rs.
This means that each Symbol that we define is represented
in the AST of rustc_symbols twice: once in the definition
of the `define_symbols!` macro (similarly for the
`keywords! macro), and once in the rustc_span::symbols
definition.
That would be OK if there were only a handful of symbols,
but currently we define over 1100 symbols. The definition
of the `define_symbols!` macro contains the expanded definition
of each symbol, so that's a lot of AST storage wasted on a
macro that is used exactly once.
This commit removes the `define_symbols` macro, and simply
allows the proc macro to directly generate the
`rustc_symbols::symbol::sym` module.
The benefit is mainly in reducing memory wasted during
compilation of rustc itself. It should also reduce memory used
by Rust Analyzer.
This commit also reduces the size of the AST for symbol
definitions, by moving two `#[allow(...)]` attributes from
the symbol constants to the `sym` module. This eliminates 2200+
attribute nodes.
This commit also eliminates the need for the `digits_array`
constant. There's no need to store an array of Symbol values
for digits. We can simply define a constant of the base value,
and add to that base value.
I left the `sym::integer` function in rustc_span/src/symbol.rs
instead of moving it into rustc_macros/src/symbols.rs for two
reasons. First, because it's human-written code; it doesn't need
to be generated by the proc-macro. Second, because I didn't want
the `#[allow(...)]` attributes that I moved to the `sym` module
scope to apply to this function. The `sym` module re-exports the
`integer` function from its parent module.
Rollup of 5 pull requests
Successful merges:
- #78164 (Prefer regions with an `external_name` in `approx_universal_upper_bound`)
- #80003 (Fix overflow when converting ZST Vec to VecDeque)
- #80023 (Enhance error message when misspelled label to value in break expression)
- #80046 (Add more documentation to `Diagnostic` and `DiagnosticBuilder`)
- #80109 (Remove redundant and unreliable coverage test results)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Prefer regions with an `external_name` in `approx_universal_upper_bound`
Fixes#75785
When displaying a MIR borrowcheck error, we may need to find an upper
bound for a region, which gives us a region to point to in the error
message. However, a region might outlive multiple distinct universal
regions, in which case the only upper bound is 'static
To try to display a meaningful error message, we compute an
'approximate' upper bound by picking one of the universal regions.
Currently, we pick the region with the lowest index - however, this
caused us to produce a suboptimal error message in issue #75785
This PR `approx_universal_upper_bound` to prefer regions with an
`external_name`. This causes us to prefer regions from function
arguments/upvars, which seems to lead to a nicer error message in some
cases.
Currently, the rustc_macros::symbols macro generates two
`macro_rules!` macros as its output. These two macros are
used in rustc_span/src/symbol.rs.
This means that each Symbol that we define is represented
in the AST of rustc_symbols twice: once in the definition
of the `define_symbols!` macro (similarly for the
`keywords! macro), and once in the rustc_span::symbols
definition.
That would be OK if there were only a handful of symbols,
but currently we define over 1100 symbols. The definition
of the `define_symbols!` macro contains the expanded definition
of each symbol, so that's a lot of AST storage wasted on a
macro that is used exactly once.
This commit removes the `define_symbols` macro, and simply
allows the proc macro to directly generate the
`rustc_symbols::symbol::sym` module.
The benefit is mainly in reducing memory wasted during
compilation of rustc itself. It should also reduce memory used
by Rust Analyzer.
This commit also reduces the size of the AST for symbol
definitions, by moving two `#[allow(...)]` attributes from
the symbol constants to the `sym` module. This eliminates 2200+
attribute nodes.
This commit also eliminates the need for the `digits_array`
constant. There's no need to store an array of Symbol values
for digits. We can simply define a constant of the base value,
and add to that base value.
Fixes#75785
When displaying a MIR borrowcheck error, we may need to find an upper
bound for a region, which gives us a region to point to in the error
message. However, a region might outlive multiple distinct universal
regions, in which case the only upper bound is 'static
To try to display a meaningful error message, we compute an
'approximate' upper bound by picking one of the universal regions.
Currently, we pick the region with the lowest index - however, this
caused us to produce a suboptimal error message in issue #75785
This PR `approx_universal_upper_bound` to prefer regions with an
`external_name`. This causes us to prefer regions from function
arguments/upvars, which seems to lead to a nicer error message in some
cases.
Move binder for dyn to each list item
This essentially changes `ty::Binder<&'tcx List<ExistentialTraitRef>>` to `&'tcx List<ty::Binder<ExistentialTraitRef>>`.
This is a first step in moving the `dyn Trait` representation closer to Chalk, which we've talked about in `@rust-lang/wg-traits.`
r? `@nikomatsakis`
Always run intrinsics lowering pass
Move intrinsics lowering pass from the optimization phase (where it
would not run if -Zmir-opt-level=0), to the drop lowering phase where it
runs unconditionally.
The implementation of those intrinsics in code generation and
interpreter is unnecessary. Remove it.
Fixed conflict with drop elaboration and coverage
See
https://github.com/rust-lang/rust/issues/80045#issuecomment-745733339
Coverage statements are moved to the beginning of the BCB. This does
also affect what's counted before a panic, changing some results, but I
think these results may even be preferred? In any case, there are no
guarantees about what's counted when a panic occurs (by design).
r? `@tmandry`
FYI `@wesleywiser` `@ecstatic-morse`
Fix issue #78496
EarlyOtherwiseBranch finds MIR structures like:
```
bb0: {
...
_2 = discriminant(X)
...
switchInt(_2) -> [1_isize: bb1, otherwise: bb3]
}
bb1: {
...
_3 = discriminant(Y)
...
switchInt(_3) -> [1_isize: bb2, otherwise: bb3]
}
bb2: {...}
bb3: {...}
```
And transforms them into something like:
```
bb0: {
...
_2 = discriminant(X)
_3 = discriminant(Y)
_4 = Eq(_2, _3)
switchInt(_4) -> [true: bb4, otherwise: bb3]
}
bb2: {...} // unchanged
bb3: {...} // unchanged
bb4: {
switchInt(_2) -> [1_isize: bb2, otherwise: bb3]
}
```
But that is not always a safe thing to do -- sometimes the early `otherwise` branch is necessary so the later block could assume the value of `discriminant(X)`.
I am not totally sure what's the best way to detect that, but fixing #78496 should be easy -- we just check if `X` is a sub-expression of `Y`. A more precise test might be to check if `Y` contains a `Downcast(1)` of `X`, but I think this might be good enough.
Fix#78496
Allow `since="TBD"` for rustc_deprecated
Closes#78381.
This PR only affects `#[rustc_deprecated]`, not `#[deprecated]`, so there is no effect on any stable language feature.
Likewise this PR only implements `since="TBD"`, it does not actually tag any library functions with it, so there is no effect on any stable API.
Overview of changes:
* `rustc_middle/stability.rs`:
* change `deprecation_in_effect` function to return `false` when `since="TBD"`
* tidy up the compiler output when a deprecated item has `since="TBD"`
* `rustc_passes/stability.rs`:
* allow `since="TBD"` to pass the sanity check for stable_version < deprecated_version
* refactor the "invalid stability version" and "invalid deprecation version" error into separate errors
* rustdoc: make `since="TBD"` message on a deprecated item's page match the command-line deprecation output
* tests:
* test rustdoc output
* test that the `deprecated_in_future` lint fires when `since="TBD"`
* test the new "invalid deprecation version" error message
Implement if-let match guards
Implements rust-lang/rfcs#2294 (tracking issue: #51114).
I probably should do a few more things before this can be merged:
- [x] Add tests (added basic tests, more advanced tests could be done in the future?)
- [x] Add lint for exhaustive if-let guard (comparable to normal if-let statements)
- [x] Fix clippy
However since this is a nightly feature maybe it's fine to land this and do those steps in follow-up PRs.
Thanks a lot `@matthewjasper` ❤️ for helping me with lowering to MIR! Would you be interested in reviewing this?
r? `@ghost` for now
Take into account negative impls in "trait item not found" suggestions
This removes the suggestion to implement a trait for a type when that type already has a negative implementation for the trait, and replaces it with a note to point out that the trait is explicitely unimplemented, as suggested by `@scottmcm.`
Helps with #79683.
r? `@scottmcm` do you want to review this?
llvm-dwp concatenates `DW_AT_comp_dir` with `DW_AT_GNU_dwo_name` (only
when `DW_AT_comp_dir` exists), which can result in it failing to find
the DWARF object files.
In earlier testing, `DW_AT_comp_dir` wasn't present in the final
object and the current directory was the output directory.
When running tests through compiletest, the working directory of the
compilation is different from output directory and that resulted in
`DW_AT_comp_dir` being in the object file (and set to the current
working directory, rather than the output directory), and
`DW_AT_GNU_dwo_name` being set to the full path (rather than just
the filename), so llvm-dwp was failing.
This commit changes the compilation directory provided to LLVM to match
the output directory, where DWARF objects are output; and ensures that
only the filename is used for `DW_AT_GNU_dwo_name`.
Signed-off-by: David Wood <david@davidtw.co>
This commit adds a Split DWARF compare mode to compiletest so that
debuginfo tests are also tested using Split DWARF in split mode (and
manually in single mode).
Signed-off-by: David Wood <david@davidtw.co>
This commit makes minor changes to the cranelift backend so that it can
build given changes in cg_ssa for Split DWARF.
Signed-off-by: David Wood <david@davidtw.co>
This commit implements Split DWARF support, wiring up the flag (added in
earlier commits) to the modified FFI wrapper (also from earlier
commits).
Signed-off-by: David Wood <david@davidtw.co>
This commit removes the `TargetMachineFactory` struct and adds a
`TargetMachineFactoryFn` type alias which is used everywhere that the
previous, long type was used.
Signed-off-by: David Wood <david@davidtw.co>
This commit changes some comments to documentation comments so that
they can be read on the generated rustdoc.
Signed-off-by: David Wood <david@davidtw.co>
This commit modifies the FFI bindings to LLVM required for Split DWARF
support in rustc. In particular:
- `addPassesToEmitFile`'s wrapper, `LLVMRustWriteOutputFile` now takes
a `DwoPath` `const char*`. When disabled, `nullptr` should be provided
which will preserve existing behaviour. When enabled, the path to the
`.dwo` file should be provided.
- `createCompileUnit`'s wrapper, `LLVMRustDIBuilderCreateCompileUnit`
now has two additional arguments, for the `DWOId` and to enable
`SplitDebugInlining`. `DWOId` should always be zero.
- `createTargetMachine`'s wrapper, `LLVMRustCreateTargetMachine` has an
additional argument which should be provided the path to the `.dwo`
when enabled.
Signed-off-by: David Wood <david@davidtw.co>
See
https://github.com/rust-lang/rust/issues/80045#issuecomment-745733339
Coverage statements are moved to the beginning of the BCB. This does
also affect what's counted before a panic, changing some results, but I
think these results may even be preferred? In any case, there are no
guarantees about what's counted when a panic occurs (by design).
make MIR graphviz generation use gsgdt
gsgdt [https://crates.io/crates/gsgdt] is a crate which provides an
interface for stringly typed graphs. It also provides generation of
graphviz dot format from said graph.
This is the first in a series of PRs on moving graphviz code out of rustc into normal crates and then implementating graph diffing on top of these crates.
r? `@oli-obk`
[rustdoc] Switch to Symbol for item.name
This decreases the size of `Item` from 680 to 616 bytes. It also does a
lot less work since it no longer has to copy as much.
Helps with #79103.
r? `@GuillaumeGomez`
Fixes reported bugs in Rust Coverage
Fixes: #79569Fixes: #79566Fixes: #79565
For the first issue (#79569), I got hit a `debug_assert!()` before
encountering the reported error message (because I have `debug = true`
enabled in my config.toml).
The assertion showed me that some `SwitchInt`s can have more than one
target pointing to the same `BasicBlock`.
I had thought that was invalid, but since it seems to be possible, I'm
allowing this now.
I added a new test for this.
----
In the last two cases above, both tests (intentionally) fail to compile,
but the `InstrumentCoverage` pass is invoked anyway.
The MIR starts with an `Unreachable` `BasicBlock`, which I hadn't
encountered before. (I had assumed the `InstrumentCoverage` pass
would only be invoked with MIRs from successful compilations.)
I don't have test infrastructure set up to test coverage on files that
fail to compile, so I didn't add a new test.
r? `@tmandry`
FYI: `@wesleywiser`
consider assignments of union field of ManuallyDrop type safe
Assigning to `Copy` union fields is safe because that assignment will never drop anything. However, with https://github.com/rust-lang/rust/pull/77547, unions may also have `ManuallyDrop` fields, and their assignments are currently still unsafe. That seems unnecessary though, as assigning `ManuallyDrop` does not drop anything either, and is thus safe even for union fields.
I assume this will at least require FCP.
[mir-opt] Allow debuginfo to be generated for a constant or a Place
Prior to this commit, debuginfo was always generated by mapping a name
to a Place. This has the side-effect that `SimplifyLocals` cannot remove
locals that are only used for debuginfo because their other uses have
been const-propagated.
To allow these locals to be removed, we now allow debuginfo to point to
a constant value. The `ConstProp` pass detects when debuginfo points to
a local with a known constant value and replaces it with the value. This
allows the later `SimplifyLocals` pass to remove the local.
Move intrinsics lowering pass from the optimization phase (where it
would not run if -Zmir-opt-level=0), to the drop lowering phase where it
runs unconditionally.
The implementation of those intrinsics in code generation and
interpreter is unnecessary. Remove it.
Adds checks for:
* `no_core` attribute
* explicitly-enabled `legacy` symbol mangling
* mir_opt_level > 1 (which enables inlining)
I removed code from the `Inline` MIR pass that forcibly disabled
inlining if `-Zinstrument-coverage` was set. The default `mir_opt_level`
does not enable inlining anyway. But if the level is explicitly set and
is greater than 1, I issue a warning.
The new warnings show up in tests, which is much better for diagnosing
potential option conflicts in these cases.
Improve error handling in `symbols` proc-macro
This improves how the `symbols` proc-macro handles errors.
If it finds an error in its input, the macro does not panic.
Instead, it still produces an output token stream. That token
stream will contain `compile_error!(...)` macro invocations.
This will still cause compilation to fail (which is what we want),
but it will prevent meaningless errors caused by the output not
containing symbols that the macro normally generates.
This solves a small (but annoying) problem. When you're editing
rustc_span/src/symbol.rs, and you get something wrong (dup
symbol name, misordered symbol), you want to get only the errors
that are relevant, not a burst of errors that are irrelevant.
This change also uses the correct Span when reporting errors,
so you get errors that point to the correct place in
rustc_span/src/symbol.rs where something is wrong.
This also adds several unit tests which test the `symbols` proc-macro.
This commit also makes it easy to run the `symbols` proc-macro
as an ordinary Cargo test. Just run `cargo test`. This makes it
easier to do development on the macro itself, such as running it
under a debugger.
This commit also uses the `Punctuated` type in `syn` for parsing
comma-separated lists, rather than doing it manually.
The output of the macro is not changed at all by this commit,
so rustc should be completely unchanged. This just improves
quality of life during development.
Properly capture trailing 'unglued' token
If we try to capture the `Vec<u8>` in `Option<Vec<u8>>`, we'll
need to capture a `>` token which was 'unglued' from a `>>` token.
The processing of unglueing a token for parsing purposes bypasses the
usual capturing infrastructure, so we currently lose the trailing `>`.
As a result, we fall back to the reparsed `TokenStream`, causing us to
lose spans.
This commit makes token capturing keep track of a trailing 'unglued'
token. Note that we don't need to care about unglueing except at the end
of the captured tokens - if we capture both the first and second unglued
tokens, then we'll end up capturing the full 'glued' token, which
already works correctly.
Recover on `const impl<> X for Y`
`@leonardo-m` mentioned that `const impl Foo for Bar` could be recovered from in #79287.
I'm not sure about the error strings as they are, I think it should probably be something like the error that `expected_one_of_not_found` makes + the suggestion to flip the keywords, but I'm not sure how exactly to do that. Also, I decided not to try to handle `const unsafe impl` or `unsafe const impl` cause I figured that `unsafe impl const` would be pretty rare anyway (if it's even valid?), and it wouldn't be worth making the code more messy.
Resolve enum field visibility correctly
Fixes#79593. 🎉
Previously, this code treated enum fields' visibility as if they were
struct fields. However, that's not correct because the visibility of a
struct field with `ast::VisibilityKind::Inherited` is private to the
module it's defined in, whereas the visibility of an *enum* field with
`ast::VisibilityKind::Inherited` is the visibility of the enum it
belongs to.
Remove an unused dependency that made `rustdoc` crash
Whilst struggling with https://github.com/rust-lang/rust/issues/79980 I discovered that this dependency was unused, and that made rustdoc crash. This PR removes it.
fix more clippy::complexity findings
fix clippy::unnecessary_filter_map
use if let Some(x) = .. instead of ...map(|x|) to conditionally run fns that return () (clippy::option_map_unit_fn)
fix clippy::{needless_bool, manual_unwrap_or}
don't clone types that are copy (clippy::clone_on_copy)
don't convert types into identical types with .into() (clippy::useless_conversion)
use strip_prefix over slicing (clippy::manual_strip)
r? ``@Dylan-DPC``
This improves how the `symbols` proc-macro handles errors.
If it finds an error in its input, the macro does not panic.
Instead, it still produces an output token stream. That token
stream will contain `compile_error!(...)` macro invocations.
This will still cause compilation to fail (which is what we want),
but it will prevent meaningless errors caused by the output not
containing symbols that the macro normally generates.
This solves a small (but annoying) problem. When you're editing
rustc_span/src/symbol.rs, and you get something wrong (dup
symbol name, misordered symbol), you want to get only the errors
that are relevant, not a burst of errors that are irrelevant.
This change also uses the correct Span when reporting errors,
so you get errors that point to the correct place in
rustc_span/src/symbol.rs where something is wrong.
This also adds several unit tests which test the `symbols` proc-macro.
This commit also makes it easy to run the `symbols` proc-macro
as an ordinary Cargo test. Just run `cargo test`. This makes it
easier to do development on the macro itself, such as running it
under a debugger.
This commit also uses the `Punctuated` type in `syn` for parsing
comma-separated lists, rather than doing it manually.
The output of the macro is not changed at all by this commit,
so rustc should be completely unchanged. This just improves
quality of life during development.
Previously, this code treated enum fields' visibility as if they were
struct fields. However, that's not correct because the visibility of a
struct field with `ast::VisibilityKind::Inherited` is private to the
module it's defined in, whereas the visibility of an *enum* field with
`ast::VisibilityKind::Inherited` is the visibility of the enum it
belongs to.
If we try to capture the `Vec<u8>` in `Option<Vec<u8>>`, we'll
need to capture a `>` token which was 'unglued' from a `>>` token.
The processing of unglueing a token for parsing purposes bypasses the
usual capturing infrastructure, so we currently lose the trailing `>`.
As a result, we fall back to the reparsed `TokenStream`, causing us to
lose spans.
This commit makes token capturing keep track of a trailing 'unglued'
token. Note that we don't need to care about unglueing except at the end
of the captured tokens - if we capture both the first and second unglued
tokens, then we'll end up capturing the full 'glued' token, which
already works correctly.
Create `rustc_type_ir`
Decided to start small 😄
This PR creates a `rustc_type_ir` crate as part of the WG-Traits plan to create a shared type library.
~~There already exists a `rustc_ty` crate, so I named the new crate `rustc_ty_library`. However I think it would make sense to rename the current `rustc_ty` to something else (e.g. `rustc_ty_passes`) to free the name for this new crate.~~
r? `@jackh726`
Fixes: #79569Fixes: #79566Fixes: #79565
For the first issue (#79569), I got hit a `debug_assert!()` before
encountering the reported error message (because I have `debug = true`
enabled in my config.toml).
The assertion showed me that some `SwitchInt`s can have more than one
target pointing to the same `BasicBlock`.
I had thought that was invalid, but since it seems to be possible, I'm
allowing this now.
I added a new test for this.
----
In the last two cases above, both tests (intentionally) fail to compile,
but the `InstrumentCoverage` pass is invoked anyway.
The MIR starts with an `Unreachable` `BasicBlock`, which I hadn't
encountered before. (I had assumed the `InstrumentCoverage` pass
would only be invoked with MIRs from successful compilations.)
I don't have test infrastructure set up to test coverage on files that
fail to compile, so I didn't add a new test.
Capture precise paths in THIR and MIR
This PR allows THIR and MIR to use the result of the new capture analysis to actually capture precise paths
To achieve we:
- Writeback min capture results to TypeckResults
- Move handling upvars to PlaceBuilder in mir_build
- Lower precise paths in THIR build by reading min_captures
- Search for ancestors in min_capture when trying to build a MIR place which starts off of an upvar
Closes: https://github.com/rust-lang/project-rfc-2229/issues/10
Partly implements: rust-lang/project-rfc-2229#18
Work that remains (not in this PR):
- [ ] [Known bugs when feature gate is enabled](https://github.com/rust-lang/project-rfc-2229/projects/1?card_filter_query=label%3Abug)
- [ ] Use min_capure_map for
- [ ] Liveness analysis
- [ ] rustc_mir/interpret/validity.rs
- [ ] regionck
- [ ] rust-lang/project-rfc-2229#8
- [ ] remove closure_captures and upvar_capture_map
r? `@ghost`
CTFE: tweak abort-on-uninhabited message
Having an "aborted execution:" makes it more consistent with the `Abort` terminator saying "the program aborted execution". Right now, at least one of the two errors will look weird in Miri.
r? `@oli-obk`
Use `def_path_hash_to_def_id` when re-using a `RawDefId`
Fixes#79890
Previously, we just copied a `RawDefId` from the 'old' map to the 'new'
map. However, the `RawDefId` for a given `DefPathHash` may be different
in the current compilation session. Using `def_path_hash_to_def_id`
ensures that the `RawDefId` we use is valid in the current session.
Use Symbol for inline asm register class names
This takes care of one "FIXME":
// FIXME: use direct symbol comparison for register class names
Instead of using string literals, this uses Symbol for register
class names.
This is part of work I am doing to improve how Symbol interning works.
Clarify the 'default is only allowed on...' error
Code like
impl Foo {
default fn foo() {}
}
will trigger the error
error: `default` is only allowed on items in `impl` definitions
--> src/lib.rs:5:5
|
5 | default fn foo() {}
| -------^^^^^^^^^
| |
| `default` because of this
but that's very confusing! I *did* put it on an item in an impl!
So this commit changes the message to
error: `default` is only allowed on items in trait impls
--> src/lib.rs:5:5
|
5 | default fn foo() {}
| -------^^^^^^^^^
| |
| `default` because of this
Dogfood `str_split_once()`
Part of https://github.com/rust-lang/rust/issues/74773.
Beyond increased clarity, this fixes some instances of a common confusion with how `splitn(2)` behaves: the first element will always be `Some()`, regardless of the delimiter, and even if the value is empty.
Given this code:
```rust
fn main() {
let val = "...";
let mut iter = val.splitn(2, '=');
println!("Input: {:?}, first: {:?}, second: {:?}", val, iter.next(), iter.next());
}
```
We get:
```
Input: "no_delimiter", first: Some("no_delimiter"), second: None
Input: "k=v", first: Some("k"), second: Some("v")
Input: "=", first: Some(""), second: Some("")
```
Using `str_split_once()` makes more clear what happens when the delimiter is not found.
This takes care of one "FIXME":
// FIXME: use direct symbol comparison for register class names
Instead of using string literals, this uses Symbol for register
class names.
Fixes#79890
Previously, we just copied a `RawDefId` from the 'old' map to the 'new'
map. However, the `RawDefId` for a given `DefPathHash` may be different
in the current compilation session. Using `def_path_hash_to_def_id`
ensures that the `RawDefId` we use is valid in the current session.
rustc_codegen_ssa: use bitcasts instead of type punning for scalar transmutes.
This specifically helps with `f32` <-> `u32` (`from_bits`, `to_bits`) in Rust-GPU (`rustc_codegen_spirv`), where (AFAIK) we don't yet have enough infrastructure to turn type punning memory accesses into SSA bitcasts.
(There may be more instances, but the one I've seen myself is `f32::signum` from `num-traits` inspecting e.g. the sign bit)
Sadly I've had to make an exception for `transmute`s between pointers and non-pointers, as LLVM disallows using `bitcast` for them.
r? `@nagisa` cc `@khyperia`
Constier maybe uninit
I was playing around trying to make `[T; N]::zip()` in #79451 be `const fn`. One of the things I bumped into was `MaybeUninit::assume_init`. Is there any reason for the intrinsic `assert_inhabited<T>()` and therefore `MaybeUninit::assume_init` not being `const`?
---
I have as best as I could tried to follow the instruction in [library/core/src/intrinsics.rs](https://github.com/rust-lang/rust/blob/master/library/core/src/intrinsics.rs#L11). I have no idea what I am doing but it seems to compile after some slight changes after the copy paste. Is this anywhere near how this should be done?
Also any ideas for name of the feature gate? I guess `const_maybe_assume_init` is quite misleading since I have added some more methods. Should I add test? If so what should be tested?
- Closures now use closure_min_captures to figure out captured paths
- Build upvar_mutbls using closure_min_captures
- Change logic in limit_capture_mutability to differentiate b/w
capturing parent's local variable or capturing a variable that is
captured by the parent (in case of nested closure) using PlaceBase.
Co-authored-by: Roxane Fruytier <roxane.fruytier@hotmail.com>
- Use closure_min_capture maps to capture precise paths
- PlaceBuilder now searches for ancestors in min_capture list
- Add API to `Ty` to allow access to the n-th element in a
tuple in O(1) time.
Co-authored-by: Roxane Fruytier <roxane.fruytier@hotmail.com>
Rollup of 12 pull requests
Successful merges:
- #79732 (minor stylistic clippy cleanups)
- #79750 (Fix trimming of lint docs)
- #79777 (Remove `first_merge` from liveness debug logs)
- #79795 (Privatize some of libcore unicode_internals)
- #79803 (Update xsv to prevent random CI failures)
- #79810 (Account for gaps in def path table during decoding)
- #79818 (Fixes to Rust coverage)
- #79824 (Strip prefix instead of replacing it with empty string)
- #79826 (Simplify visit_{foreign,trait}_item)
- #79844 (Move RWUTable to a separate module)
- #79861 (Update LLVM submodule)
- #79862 (Remove tab-lock and replace it with ctrl+up/down arrows to switch between search result tabs)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Simplify visit_{foreign,trait}_item
Using an `if` seems like a better semantic fit and saves a few lines.
Noticed while looking at https://github.com/rust-lang/rust/pull/79752, but that's already merged.
r? `@lcnr,` cc `@cjgillot`
`@rustbot` modify labels +C-cleanup +T-compiler
Strip prefix instead of replacing it with empty string
r? `@lcnr,` since you reviewed my other PR in the area.
`@rustbot` modify labels +C-cleanup +T-compiler
Fixes to Rust coverage
Fixes: #79725
Some macros can create a situation where `fn_sig_span` and `body_span`
map to different files.
New documentation on coverage tests incorrectly assumed multiple test
binaries could just be listed at the end of the `llvm-cov` command,
but it turns out each binary needs a `--object` prefix.
This PR fixes the bug and updates the documentation to correct that
issue. It also fixes a few other minor issues in internal implementation
comments, and adds documentation on getting coverage results for doc
tests.
Account for gaps in def path table during decoding
When encoding a proc-macro crate, there may be gaps in the table (since
we only encode the crate root and proc-macro items). Account for this by
checking if the entry is present, rather than using `unwrap()`
minor stylistic clippy cleanups
simplify if let Some(_) = x to if x.is_some() (clippy::redundant_pattern_matching)
don't create owned values for comparison (clippy::cmp_owned)
use .contains() or .any() instead of find(x).is_some() (clippy::search_is_some)
don't wrap code block in Ok() (clipppy::unit_arg)
Properly re-use def path hash in incremental mode
Fixes#79661
In incremental compilation mode, we update a `DefPathHash -> DefId`
mapping every time we create a `DepNode` for a foreign `DefId`.
This mapping is written out to the on-disk incremental cache, and is
read by the next compilation session to allow us to lazily decode
`DefId`s.
When we decode a `DepNode` from the current incremental cache, we need
to ensure that any previously-recorded `DefPathHash -> DefId` mapping
gets recorded in the new mapping that we write out. However, PR #74967
didn't do this in all cases, leading to us being unable to decode a
`DefPathHash` in certain circumstances.
This PR refactors some of the code around `DepNode` deserialization to
prevent this kind of mistake from happening again.
Also generate `StorageDead` in constants
r? `@eddyb`
None of this special casing is actually necessary since we started promoting within constants and statics.
We may want to keep some of it around out of perf reasons, but it's not required for user visible behaviour
somewhat related: #68622
This replaces tabs earlier in the diagnostics emitting process, which allows
various margin calculations to ignore the existence of tabs. It does add a
string copy for the source lines that are emitted.
remove this weird special case from promotion
Promotion has a special case to ignore interior mutability under some specific circumstances. The purpose of this PR is to figure out what changes if we remove that. Since `Cell::new` and friends only get promoted inside `const`/`static` initializers these days, it actually is not easy to exploit this case: you need something like
```rust
const TEST_INTERIOR_MUT: () = {
// The "0." case is already ruled out by not permitting any interior mutability in `const`.
let _val: &'static _ = &(Cell::new(1), 2).1;
};
```
I assume something like `&Some(&(Cell::new(1), 2).1)` would hit the nested case inside `validate_rvalue`... though I am not sure why that would not just trigger nested promotion, first promoting the inner reference and then the outer one?
Fixes https://github.com/rust-lang/rust/issues/67534 (by simply rejecting that code^^)
r? `@oli-obk` (but for now this is not meant to be merged!)
Cc `@rust-lang/wg-const-eval`
Code like
impl Foo {
default fn foo() {}
}
will trigger the error
error: `default` is only allowed on items in `impl` definitions
--> src/lib.rs:5:5
|
5 | default fn foo() {}
| -------^^^^^^^^^
| |
| `default` because of this
but that's very confusing! I *did* put it on an item in an impl!
So this commit changes the message to
error: `default` is only allowed on items in trait impls
--> src/lib.rs:5:5
|
5 | default fn foo() {}
| -------^^^^^^^^^
| |
| `default` because of this
Compress RWU from at least 32 bits to 4 bits
The liveness uses a mixed representation of RWUs based on the
observation that most of them have invalid reader and invalid
writer. The packed variant uses 32 bits and unpacked 96 bits.
Unpacked data contains reader live node and writer live node.
Since live nodes are used only to determine their validity,
RWUs can always be stored in a packed form with four bits for
each: reader bit, writer bit, used bit, and one extra padding
bit to simplify packing and unpacking operations.
When encoding a proc-macro crate, there may be gaps in the table (since
we only encode the crate root and proc-macro items). Account for this by
checking if the entry is present, rather than using `unwrap()`
Replace simple `if let` constructs with Option::map
Replaces a few constructs of the form
```
if let Some(x) = var {
Some(...)
} else {
None
}
```
with calls to `Option::map`.
`@rustbot` modify labels +C-cleanup +T-compiler
Fixes: #79725
Some macros can create a situation where `fn_sig_span` and `body_span`
map to different files.
New documentation on coverage tests incorrectly assumed multiple test
binaries could just be listed at the end of the `llvm-cov` command,
but it turns out each binary needs a `--object` prefix.
This PR fixes the bug and updates the documentation to correct that
issue. It also fixes a few other minor issues in internal implementation
comments, and adds documentation on getting coverage results for doc
tests.
Validate naked functions definitions
Validate that naked functions are defined in terms of a single inline assembly
block that uses only `const` and `sym` operands and has `noreturn` option.
Implemented as future incompatibility lint with intention to migrate it into
hard error. When it becomes a hard error it will ensure that naked functions are
either unsafe or contain an unsafe block around the inline assembly. It will
guarantee that naked functions do not reference functions parameters (obsoleting
part of existing checks from #79411). It will limit the definitions of naked
functions to what can be reliably supported. It will also reject naked functions
implemented using legacy LLVM style assembly since it cannot satisfy those
conditions.
https://github.com/rust-lang/rfcs/pull/2774https://github.com/rust-lang/rfcs/pull/2972
Prior to this commit, debuginfo was always generated by mapping a name
to a Place. This has the side-effect that `SimplifyLocals` cannot remove
locals that are only used for debuginfo because their other uses have
been const-propagated.
To allow these locals to be removed, we now allow debuginfo to point to
a constant value. The `ConstProp` pass detects when debuginfo points to
a local with a known constant value and replaces it with the value. This
allows the later `SimplifyLocals` pass to remove the local.
Previously, if you had a lot of inherent impl blocks on a type like:
struct Foo;
impl Foo { fn foo_1() {} }
...
impl Foo { fn foo_100_000() {} }
The compiler would be very slow at processing it, because
an internal algorithm would run in O(n^2), where n is the number
of impl blocks. Now, we add a new algorithm that allocates but
is faster asymptotically.
If there is an overlap between multiple impl blocks in terms of
identifiers, we still run a O(m^2) algorithm on groups of impl
blocks that have overlaps, but that m refers to the size of the
connected component, which is hopefully smaller than the n
that refers to the sum of all connected components.
- This allows us to delay figuring out the index of a capture
in the closure structure when all projections to atleast form
a capture have been applied to the builder
Co-authored-by: Roxane Fruytier <roxane.fruytier@hotmail.com>
- Derive TypeFoldable on `hir::place::Place` and associated
structs, to them to be written into typeck results.
Co-authored-by: Jennifer Wills <wills.jenniferg@gmail.com>
Co-authored-by: Logan Mosier <logmosier@gmail.com>
extend `WithOptConstParam` docs, move rustdoc test
This should hopefully make things a bit clearer, feel free to comment on anything which can still be improved.
cc `@ecstatic-morse` `@nikomatsakis` `@RalfJung`
Const parameters can not be inferred with `_` help note
This should close: #79557
# Example output
```
error[E0747]: type provided when a constant was expected
--> inferred_const_note.rs:6:19
|
6 | let a = foo::<_, 2>([0, 1, 2]);
| ^
|
= help: Const parameters can not be inferred with `_`
error: aborting due to previous error
For more information about this error, try `rustc --explain E0747`.
```
r? `@lcnr`
The liveness uses a mixed representation of RWUs based on the
observation that most of them have invalid reader and invalid
writer. The packed variant uses 32 bits and unpacked 96 bits.
Unpacked data contains reader live node and writer live node.
Since live nodes are used only to determine their validity,
RWUs can always be stored in a packed form with four bits for
each: reader bit, writer bit, used bit, and one extra padding
bit to simplify packing and unpacking operations.
A slightly clearer diagnostic when misusing const
Fixes#79598
This produces the following diagnostic:
"expected one of `>`, a const expression, lifetime, or type, found keyword `const`"
Instead of the previous, more confusing:
"expected one of `>`, const, lifetime, or type, found keyword `const`"
This might not be completely clear as some users might not understand what a const expression is, but I do believe this is an improvement.
check the recursion limit when finding a struct's tail
fixes#79437
This does a `delay_span_bug` (via `ty_error_with_message`) rather than emit a new error message, under the assumption that there will be an error elsewhere (even if the type isn't infinitely recursive, just deeper than the recursion limit, this appears to be the case).
Fixes#79661
In incremental compilation mode, we update a `DefPathHash -> DefId`
mapping every time we create a `DepNode` for a foreign `DefId`.
This mapping is written out to the on-disk incremental cache, and is
read by the next compilation session to allow us to lazily decode
`DefId`s.
When we decode a `DepNode` from the current incremental cache, we need
to ensure that any previously-recorded `DefPathHash -> DefId` mapping
gets recorded in the new mapping that we write out. However, PR #74967
didn't do this in all cases, leading to us being unable to decode a
`DefPathHash` in certain circumstances.
This PR refactors some of the code around `DepNode` deserialization to
prevent this kind of mistake from happening again.
Fix perf regression caused by #79284https://github.com/rust-lang/rust/pull/79284 only moved code around but this changed inlining and caused a large perf regression. This fixes it for me, though I'm less confident than usual because the regression was not observable with my usual (i.e. incremental) compilation settings.
r? `@Mark-Simulacrum`
Gives a performance increase over calling byte_pos_to_line_and_col
twice, partially because it decreases the function calling overhead,
potentially because it doesn't populate the line cache with lines that
turn out to belong to invalid spans, and likely because of some other
incidental improvements made possible by having more context available.
Coverage tests for remaining TerminatorKinds and async, improve Assert
Tested and validate results for panic unwind, panic abort, assert!()
macro, TerminatorKind::Assert (for example, numeric overflow), and
async/await.
Implemented a previous documented idea to change Assert handling to be
the same as FalseUnwind and Goto, so it doesn't get its own
BasicCoverageBlock anymore. This changed a couple of coverage regions,
but I validated those changes are not any worse than the prior results,
and probably help assure some consistency (even if some people might
disagree with how the code region is consistently computed).
Fixed issue with async/await. AggregateKind::Generator needs to be
handled like AggregateKind::Closure; coverage span for the outer async
function should not "cover" the async body, which is actually executed
in a separate "closure" MIR.
Fix some clippy lints
Happy to revert these if you think they're less readable, but personally I like them better now (especially the `else { if { ... } }` to `else if { ... }` change).
Added one more test (two files) showing coverage of generics and unused
functions across crates.
Created and referenced new Issues, as requested.
Added comments.
Added a note about the possible effects of compiler options on LLVM
coverage maps.
Fixes multiple issue with counters, with simplification
Includes a change to the implicit else span in ast_lowering, so coverage
of the implicit else no longer spans the `then` block.
Adds coverage for unused closures and async function bodies.
Fixes: #78542
Adding unreachable regions for known MIR missing from coverage map
Cleaned up PR commits, and removed link-dead-code requirement and tests
Coverage no longer depends on Issue #76038 (`-C link-dead-code` is
no longer needed or enforced, so MSVC can use the same tests as
Linux and MacOS now)
Restrict adding unreachable regions to covered files
Improved the code that adds coverage for uncalled functions (with MIR
but not-codegenned) to avoid generating coverage in files not already
included in the files with covered functions.
Resolved last known issue requiring --emit llvm-ir workaround
Fixed bugs in how unreachable code spans were added.
Tested and validate results for panic unwind, panic abort, assert!()
macro, TerminatorKind::Assert (for example, numeric overflow), and
async/await.
Implemented a previous documented idea to change Assert handling to be
the same as FalseUnwind and Goto, so it doesn't get its own
BasicCoverageBlock anymore. This changed a couple of coverage regions,
but I validated those changes are not any worse than the prior results,
and probably help assure some consistency (even if some people might
disagree with how the code region is consistently computed).
Fixed issue with async/await. AggregateKind::Generator needs to be
handled like AggregateKind::Closure; coverage span for the outer async
function should not "cover" the async body, which is actually executed
in a separate "closure" MIR.
Fix `unknown-crate` when using -Z self-profile with rustdoc
... by removing a duplicate `crate_name` field in `interface::Config`,
making it clear that rustdoc should be passing it to `config::Options` instead.
Unblocks https://github.com/rust-lang/rustc-perf/issues/797.
Revert "Auto merge of #79209
r? `@nikomatsakis`
This has caused some issues (#79560) so better to revert and try to come up with a proper fix without rush.
Do not show negative polarity trait implementations in diagnostic messages for similar implementations
This fixes#79458.
Previously, this code:
```rust
#[derive(Clone)]
struct Foo<'a, T> {
x: &'a mut T,
}
```
would have suggested that `<&mut T as Clone>` was an implementation that was found. This is due to the fact that the standard library now has `impl<'_, T> !Clone for &'_ mut T`, and explicit negative polarity implementations were not filtered out in diagnostic output when suggesting similar implementations.
This PR fixes this issue by filtering out negative polarity trait implementations in `find_similar_impl_candidates` within `rustc_trait_selection::traits::error_reporting::InferCtxtPrivExt<'tcx>`. It also adds a UI regression test for this issue and fixes UI tests that had incorrectly been modified to expect the invalid output.
r? `@scottmcm`
Use true previous lint level when detecting overriden forbids
Previously, cap-lints was ignored when checking the previous forbid level, which
meant that it was a hard error to do so. This is different from the normal
behavior of lints, which are silenced by cap-lints; if the forbid would not take
effect regardless, there is not much point in complaining about the fact that we
are reducing its level.
It might be considered a bug that even `--cap-lints deny` would suffice to
silence the error on overriding forbid, depending on if one cares about failing
the build or precisely forbid being set. But setting cap-lints to deny is quite
odd and not really done in practice, so we don't try to handle it specially.
This also unifies the code paths for nested and same-level scopes. However, the
special case for CLI lint flags is left in place (introduced by #70918) to fix
the regression noted in #70819. That means that CLI flags do not lint on forbid
being overridden by a non-forbid level. It is unclear whether this is a bug or a
desirable feature, but it is certainly inconsistent. CLI flags are a
sufficiently different "type" of place though that this is deemed out of scope
for this commit.
r? `@pnkfelix` perhaps?
cc #77713 -- not marking as "Fixes" because of the lack of proper unused attribute handling in this PR