Use tracing spans in rustc_trait_selection
Spans are very helpful when debugging this code. It's also hot enough to make a good benchmark.
r? `@oli-obk`
Mono collector: replace pair of ints with Range
I found the initial PR (#33171) that introduced this piece of code but I didn't find any information about why a tuple was preferred over a `Range<usize>`.
I'm hoping there are no technical reasons to not do this.
Remove unnecessary unsafe block around calls to discriminant_value
Since 63793 the discriminant_value intrinsic is safe to call. Remove
unnecessary unsafe block around calls to this intrinsic in built-in
derive macros.
Use llvm::computeLTOCacheKey to determine post-ThinLTO CGU reuse
During incremental ThinLTO compilation, we attempt to re-use the
optimized (post-ThinLTO) bitcode file for a module if it is 'safe' to do
so.
Up until now, 'safe' has meant that the set of modules that our current
modules imports from/exports to is unchanged from the previous
compilation session. See PR #67020 and PR #71131 for more details.
However, this turns out be insufficient to guarantee that it's safe
to reuse the post-LTO module (i.e. that optimizing the pre-LTO module
would produce the same result). When LLVM optimizes a module during
ThinLTO, it may look at other information from the 'module index', such
as whether a (non-imported!) global variable is used. If this
information changes between compilation runs, we may end up re-using an
optimized module that (for example) had dead-code elimination run on a
function that is now used by another module.
Fortunately, LLVM implements its own ThinLTO module cache, which is used
when ThinLTO is performed by a linker plugin (e.g. when clang is used to
compile a C proect). Using this cache directly would require extensive
refactoring of our code - but fortunately for us, LLVM provides a
function that does exactly what we need.
The function `llvm::computeLTOCacheKey` is used to compute a SHA-1 hash
from all data that might influence the result of ThinLTO on a module.
In addition to the module imports/exports that we manually track, it
also hashes information about global variables (e.g. their liveness)
which might be used during optimization. By using this function, we
shouldn't have to worry about new LLVM passes breaking our module re-use
behavior.
In LLVM, the output of this function forms part of the filename used to
store the post-ThinLTO module. To keep our current filename structure
intact, this PR just writes out the mapping 'CGU name -> Hash' to a
file. To determine if a post-LTO module should be reused, we compare
hashes from the previous session.
This should unblock PR #75199 - by sheer chance, it seems to have hit
this issue due to the particular CGU partitioning and optimization
decisions that end up getting made.
Recognize discriminant reads as no-ops in RemoveNoopLandingPads
The cleanup blocks often contain read of discriminants. Teach
RemoveNoopLandingPads to recognize them as no-ops to remove
additional no-op landing pads.
Fixes#74616
Makes progress towards #43081
Unblocks PR #76130
When pretty-printing an AST node, we may insert additional parenthesis
to ensure that precedence is properly preserved in code we output.
However, the proc macro implementation relies on comparing a
pretty-printed AST node to the captured `TokenStream`. Inserting extra
parenthesis changes the structure of the reparsed `TokenStream`, making
the comparison fail.
This PR refactors the AST pretty-printing code to allow skipping the
insertion of additional parenthesis. Several freestanding methods are
moved to trait methods on `PrintState`, which keep track of an internal
`insert_extra_parens` flag. This flag is normally `true`, but we expose
a public method which allows pretty-printing a nonterminal with
`insert_extra_parens = false`.
To avoid changing the public interface of `rustc_ast_pretty`, the
freestanding `_to_string` methods are changed to delegate to a
newly-crated `State`. The main pretty-printing code is moved to a new
`state` module to ensure that it does not accidentally call any of these
public helper functions (instead, the internal functions with the same
name should be used).
Depending on if upvar_tys inferred or not, we were returning either an
inference variable which later resolves to a tuple or else the upvar tys
themselves
Co-authored-by: Roxane Fruytier <roxane.fruytier@hotmail.com>
This commit allows us to decide the number of captures required after
completing capture ananysis, which is required as part of implementing
RFC-2229.
Co-authored-by: Aman Arora <me@aman-arora.com>
Co-authored-by: Jenny Wills <wills.jenniferg@gmail.com>
Add -Z codegen-backend dylib to deps
When the codegen-backend dylib changes, the program should be rebuilt.
---
Unfortunately I was unable to test this works locally due to running into a TLS issue when running the custom backend, `thread 'rustc' panicked at 'no ImplicitCtxt stored in tls', compiler/rustc_middle/src/ty/context.rs:1750:54`, which seems similar to https://github.com/rust-lang/rust/issues/62717 but has a completely different cause and backtrace.
`@eddyb` said to ping `@Mark-Simulacrum` about what they think about this, so, ping!
This is a solution to the file length being over 3000, something Clippy has a problem with.
The other solution to the file length is
1. to change the API of this struct by
2. encapulating certain fields of the struct into other structs.
Since 63793 the discriminant_value intrinsic is safe to call. Remove
unnecessary unsafe block around calls to this intrinsic in built-in
derive macros.
Provide structured suggestions when finding structs when expecting a trait
When finding an ADT in a trait object definition provide some solutions. Fix#45817.
Given `<Param as Trait>::Assoc: Ty` suggest `Param: Trait<Assoc = Ty>`. Fix#75829.
Allow generic parameters in intra-doc links
Fixes#62834.
---
The contents of the generics will be mostly ignored (except for warning
if fully-qualified syntax is used, which is currently unsupported in
intra-doc links - see issue #74563).
* Allow links like `Vec<T>`, `Result<T, E>`, and `Option<Box<T>>`
* Allow links like `Vec::<T>::new()`
* Warn on
* Unbalanced angle brackets (e.g. `Vec<T` or `Vec<T>>`)
* Missing type to apply generics to (`<T>` or `<Box<T>>`)
* Use of fully-qualified syntax (`<Vec as IntoIterator>::into_iter`)
* Invalid path separator (`Vec:<T>:new`)
* Too many angle brackets (`Vec<<T>>`)
* Empty angle brackets (`Vec<>`)
Note that this implementation *does* allow some constructs that aren't
valid in the actual Rust syntax, for example `Box::<T>new()`. That may
not be supported in rustdoc in the future; it is an implementation
detail.
They were not formatted correctly, so rustdoc was interpreting some
parts as code. Also cleaned up some other query docs that weren't
causing issues, but were formatted incorrectly.
Add TraitDef::find_map_relevant_impl
This PR adds a method to `TraitDef`. While `for_each_relevant_impl` covers the general use case, sometimes it's not necessary to scan through all the relevant implementations, so this PR introduces a new method, `find_map_relevant_impl`. I've also replaced the `for_each_relevant_impl` calls where possible.
I'm hoping for a tiny bit of efficiency gain here and there.
Cleanup of `eat_while()` in lexer
The size of a lexer Token was inflated by the largest `TokenKind` variants `LiteralKind::RawStr` and `RawByteStr`, because
* it used `usize` although `u32` is sufficient in rustc, since crates must be smaller than 4GB,
* and it stored the 20 bytes big `RawStrError` enum for error reporting.
If a raw string is invalid, it now needs to be reparsed to get the `RawStrError` data, but that is a very cold code path.
Technically this breaks other tools that depend on rustc_lexer because they are now also restricted to a max file size of 4GB. But this shouldn't matter in practice, and rustc_lexer isn't stable anyway.
Can I also get a perf run?
Edit: This makes no difference in performance. The PR now only contains a small cleanup.
Add asm! support for mips64
- [x] Updated `src/doc/unstable-book/src/library-features/asm.md`.
- [ ] No vector type support. I don't know much about those types.
cc #76839
rustc_target: Refactor away `TargetResult`
Follow-up to https://github.com/rust-lang/rust/pull/77202.
Construction of a built-in target is always infallible now, so `TargetResult` is no longer necessary.
The second commit contains some further cleanup based on built-in target construction being infallible.
The cleanup blocks often contain read of discriminants. Teach
RemoveNoopLandingPads to recognize them as no-ops to remove
additional no-op landing pads.
Implementation of RFC2867
https://github.com/rust-lang/rust/issues/74727
So I've started work on this, I think my next steps are to make use of the `instruction_set` value in the llvm codegen but this is the point where I begin to get a bit lost. I'm looking at the code but it would be nice to have some guidance on what I've currently done and what I'm doing next 😄
RunCompiler::new takes non-optional params, and optional
params can be set using set_*field_name* method.
finally `run` will forward all fields to `run_compiler`.
Upgrade to tracing-subscriber 0.2.13
The primary motivation is to get the changes from
https://github.com/tokio-rs/tracing/pull/990. Example output:
```
$ RUSTDOC_LOG=debug rustdoc +rustc2
warning: some trace filter directives would enable traces that are disabled statically
| `debug` would enable the DEBUG level for all targets
= note: the static max level is `info`
= help: to enable DEBUG logging, remove the `max_level_info` feature
```
r? `@Mark-Simulacrum`
cc `@hawkw` ❤️
Use `pretty::create_dump_file` for dumping dataflow results
The old code wasn't incorporating promoteds into the path, meaning other `dot` files could get clobbered. Use the MIR dump infrastructure to generate paths so that this doesn't occur in the future.
Detect blocks that could be struct expr bodies
This approach lives exclusively in the parser, so struct expr bodies
that are syntactically correct on their own but are otherwise incorrect
will still emit confusing errors, like in the following case:
```rust
fn foo() -> Foo {
bar: Vec::new()
}
```
```
error[E0425]: cannot find value `bar` in this scope
--> src/file.rs:5:5
|
5 | bar: Vec::new()
| ^^^ expecting a type here because of type ascription
error[E0214]: parenthesized type parameters may only be used with a `Fn` trait
--> src/file.rs:5:15
|
5 | bar: Vec::new()
| ^^^^^ only `Fn` traits may use parentheses
error[E0107]: wrong number of type arguments: expected 1, found 0
--> src/file.rs:5:10
|
5 | bar: Vec::new()
| ^^^^^^^^^^ expected 1 type argument
```
If that field had a trailing comma, that would be a parse error and it
would trigger the new, more targetted, error:
```
error: struct literal body without path
--> file.rs:4:17
|
4 | fn foo() -> Foo {
| _________________^
5 | | bar: Vec::new(),
6 | | }
| |_^
|
help: you might have forgotten to add the struct literal inside the block
|
4 | fn foo() -> Foo { Path {
5 | bar: Vec::new(),
6 | } }
|
```
Partially address last remaining part of #34255.
perf: UninhabitedEnumBranching avoid n^2
Avoid n² complexity. This showed up in a profile for match-stress-enum that has 8192 variants
I have only profiled locally against `match-stress-enum`, so we should have it perf tested to make sure it does not regress other crates.
The primary motivation is to get the changes from
https://github.com/tokio-rs/tracing/pull/990. Example output:
```
$ RUSTDOC_LOG=debug rustdoc +rustc2
warning: some trace filter directives would enable traces that are disabled statically
| `debug` would enable the DEBUG level for all targets
= note: the static max level is `info`
= help: to enable DEBUG logging, remove the `max_level_info` feature
```
- Remove useless test
This was testing for an ICE when passing `RUST_LOG=rustc_middle`. I
noticed it because it started giving the tracing warning (because tests
are not run with debug-logging enabled). Since this bug seems unlikely
to re-occur, I just removed it altogether.
Prevent stack overflow in deeply nested types.
Related issue #75577 (?)
Unfortunately, I am unable to test whether this actually solves the problem because apparently, 12GB RAM + 2GB swap is not enough to compile the (admittedly toy) source file.
This approach lives exclusively in the parser, so struct expr bodies
that are syntactically correct on their own but are otherwise incorrect
will still emit confusing errors, like in the following case:
```rust
fn foo() -> Foo {
bar: Vec::new()
}
```
```
error[E0425]: cannot find value `bar` in this scope
--> src/file.rs:5:5
|
5 | bar: Vec::new()
| ^^^ expecting a type here because of type ascription
error[E0214]: parenthesized type parameters may only be used with a `Fn` trait
--> src/file.rs:5:15
|
5 | bar: Vec::new()
| ^^^^^ only `Fn` traits may use parentheses
error[E0107]: wrong number of type arguments: expected 1, found 0
--> src/file.rs:5:10
|
5 | bar: Vec::new()
| ^^^^^^^^^^ expected 1 type argument
```
If that field had a trailing comma, that would be a parse error and it
would trigger the new, more targetted, error:
```
error: struct literal body without path
--> file.rs:4:17
|
4 | fn foo() -> Foo {
| _________________^
5 | | bar: Vec::new(),
6 | | }
| |_^
|
help: you might have forgotten to add the struct literal inside the block
|
4 | fn foo() -> Foo { Path {
5 | bar: Vec::new(),
6 | } }
|
```
Partially address last part of #34255.
Give `impl Trait` in a `const fn` its own feature gate
...previously it was gated under `#![feature(const_fn)]`.
I think we actually want to do this in all const-contexts? If so, this should be `#![feature(const_impl_trait)]` instead. I don't think there's any way to make use of `impl Trait` within a `const` initializer.
cc #77463
r? `@oli-obk`
resolve: improve "try using the enum's variant"
Fixes#73427.
This PR improves the "try using the enum's variant" suggestion:
- Variants in suggestions would not result in more errors (e.g. use of a struct variant is only suggested if the suggestion can trivially construct that variant). Therefore, suggestions are only emitted for variants that have no fields (since the suggestion can't know what value fields would have).
- Suggestions include the syntax for constructing the variant. If a struct or tuple variant is suggested, then it is constructed in the suggestion - unless in pattern-matching or when arguments are already provided.
- A help message is added which mentions the variants which are no longer suggested.
All of the diagnostic logic introduced by this PR is separated from the normal code path for a successful compilation.
r? `@estebank`
Rollup of 11 pull requests
Successful merges:
- #76784 (Add some docs to rustdoc::clean::inline and def_id functions)
- #76911 (fix VecDeque::iter_mut aliasing issues)
- #77400 (Fix suggestions for x.py setup)
- #77515 (Update to chalk 0.31)
- #77568 (inliner: use caller param_env)
- #77571 (Use matches! for core::char methods)
- #77582 (Move `EarlyOtherwiseBranch` to mir-opt-level 2)
- #77590 (Update RLS and Rustfmt)
- #77605 (Fix rustc_def_path to show the full path and not the trimmed one)
- #77614 (Let backends access span information)
- #77624 (Add c as a shorthand check alternative for new options #77603)
Failed merges:
r? `@ghost`
Let backends access span information
Sometimes, a backend may need to emit warnings, errors, or otherwise need to know the span of the current item in a basic block. So, add a `set_span` method to give the backend that information.
The `set_source_location` method already partially does this, however, it's disabled when debug info is disabled. There needs to be a way to unconditionally provide the span.
inliner: use caller param_env
We used the callee param env instead of the caller param env by accident in #77430, this PR fixes that and caches it in the `Inliner` struct.
fixes#77564
r? @ecstatic-morse
Update to chalk 0.31
Gonna assign @nikomatsakis to the review here, but since he is pretty busy recently, if anyone else wants to review this, that would be much appreciated.
r? @nikomatsakis
Support static linking with glibc and target-feature=+crt-static
With this change, it's possible to build on a linux-gnu target and pass
RUSTFLAGS='-C target-feature=+crt-static' or the equivalent via a
`.cargo/config.toml` file, and get a statically linked executable.
Update to libc 0.2.78, which adds support for static linking with glibc.
Add `crt_static_respected` to the `linux_base` target spec.
Update `android_base` and `linux_musl_base` accordingly. Avoid enabling
crt_static_respected on Android platforms, since that hasn't been
tested.
Closes https://github.com/rust-lang/rust/issues/65447.
Sometimes, a backend may need to emit warnings, errors, or otherwise
need to know the span of the current item in a basic block. So, add a
set_span method to give the backend that information.
Only check the own predicates of associated types when confirming
projection candidates.
Also consider implied bounds when comparing trait and impl methods.
Normalizing `<dyn Iterator<Item = ()> as Iterator>::Item` no longer
requires selecting `dyn Iterator<Item = ()>: Iterator`. This was
previously worked around by using a special type-folder to normalize
things.
Bounds of the form `type Future: Future<Result=Self::Result>` exist in
some ecosystem crates. To validate these bounds for trait objects we
need to normalize `Self::Result` in a way that doesn't cause a cycle.
Record `expansion_that_defined` into crate metadata
Fixes#77523
Now that hygiene serialization is implemented, we also need to record
`expansion_that_defined` so that we properly handle a foreign
`SyntaxContext`.
Fix span for unicode escape suggestion.
If a unicode escape is missing the curly braces, the suggested fix is to add the curly braces, but the span for the fix was incorrect. It was not covering the `\u`, but the suggested text includes the `\u`, causing the resulting fix to be `"\u\u{1234}"`. This changes it so that the span includes the `\u`. An alternate fix would be to remove `\u` from the suggested fix, but I think the error message reads better if the entire escape is included.
Fix LitKind's byte buffer to use refcounted slice
While working on adding a new lint for clippy (see https://github.com/rust-lang/rust-clippy/pull/6044) for avoiding shared ownership of "mutable buffer" types (such as using `Rc<Vec<T>>` instead of `Rc<[T]>`), I noticed a type exported from rustc_ast and used by clippy gets caught by the lint. This PR fixes the exported type.
This PR includes the actual change to clippy too, but I will open a PR directly against clippy for that part (although it will currently fail to build there).
Reduce boilerplate with the matches! macro
Replaces simple bool `match`es of the form
match $expr {
$pattern => true
_ => false
}
and their inverse with invocations of the matches! macro.
Limited to rustc_middle for now to get my feet wet.
Fixes#77523
Now that hygiene serialization is implemented, we also need to record
`expansion_that_defined` so that we properly handle a foreign
`SyntaxContext`.
This is a combination of 18 commits.
Commit #2:
Additional examples and some small improvements.
Commit #3:
fixed mir-opt non-mir extensions and spanview title elements
Corrected a fairly recent assumption in runtest.rs that all MIR dump
files end in .mir. (It was appending .mir to the graphviz .dot and
spanview .html file names when generating blessed output files. That
also left outdated files in the baseline alongside the files with the
incorrect names, which I've now removed.)
Updated spanview HTML title elements to match their content, replacing a
hardcoded and incorrect name that was left in accidentally when
originally submitted.
Commit #4:
added more test examples
also improved Makefiles with support for non-zero exit status and to
force validation of tests unless a specific test overrides it with a
specific comment.
Commit #5:
Fixed rare issues after testing on real-world crate
Commit #6:
Addressed PR feedback, and removed temporary -Zexperimental-coverage
-Zinstrument-coverage once again supports the latest capabilities of
LLVM instrprof coverage instrumentation.
Also fixed a bug in spanview.
Commit #7:
Fix closure handling, add tests for closures and inner items
And cleaned up other tests for consistency, and to make it more clear
where spans start/end by breaking up lines.
Commit #8:
renamed "typical" test results "expected"
Now that the `llvm-cov show` tests are improved to normally expect
matching actuals, and to allow individual tests to override that
expectation.
Commit #9:
test coverage of inline generic struct function
Commit #10:
Addressed review feedback
* Removed unnecessary Unreachable filter.
* Replaced a match wildcard with remining variants.
* Added more comments to help clarify the role of successors() in the
CFG traversal
Commit #11:
refactoring based on feedback
* refactored `fn coverage_spans()`.
* changed the way I expand an empty coverage span to improve performance
* fixed a typo that I had accidently left in, in visit.rs
Commit #12:
Optimized use of SourceMap and SourceFile
Commit #13:
Fixed a regression, and synched with upstream
Some generated test file names changed due to some new change upstream.
Commit #14:
Stripping out crate disambiguators from demangled names
These can vary depending on the test platform.
Commit #15:
Ignore llvm-cov show diff on test with generics, expand IO error message
Tests with generics produce llvm-cov show results with demangled names
that can include an unstable "crate disambiguator" (hex value). The
value changes when run in the Rust CI Windows environment. I added a sed
filter to strip them out (in a prior commit), but sed also appears to
fail in the same environment. Until I can figure out a workaround, I'm
just going to ignore this specific test result. I added a FIXME to
follow up later, but it's not that critical.
I also saw an error with Windows GNU, but the IO error did not
specify a path for the directory or file that triggered the error. I
updated the error messages to provide more info for next, time but also
noticed some other tests with similar steps did not fail. Looks
spurious.
Commit #16:
Modify rust-demangler to strip disambiguators by default
Commit #17:
Remove std::process::exit from coverage tests
Due to Issue #77553, programs that call std::process::exit() do not
generate coverage results on Windows MSVC.
Commit #18:
fix: test file paths exceeding Windows max path len
Fix miscompile in SimplifyBranchSame
Cherry-picked from #77486, but with a different test case that used to be compiled incorrectly on both master & beta branches.
Replace `(Body, DefId)` with `Body` where possible
Follow-up to #77430.
I `grep`-ed for parameter lists in which a `Body` appeared within a few lines of a `DefId`, so it's possible that I missed some cases, but this should be pretty complete. Most of these changes were mechanical, but there's a few places where I started calling things "caller" and "callee" when multiple `DefId`s were in-scope at once. Also, we should probably have a helper function on `Body` that returns a `LocalDefId`. I can do that in this PR or in a follow-up.
With this change, it's possible to build on a linux-gnu target and pass
RUSTFLAGS='-C target-feature=+crt-static' or the equivalent via a
`.cargo/config.toml` file, and get a statically linked executable.
This requires libc 0.2.79, which adds support for static linking with
glibc.
Add `crt_static_respected` to the `linux_base` target spec.
Update `android_base` and `linux_musl_base` accordingly. Avoid enabling
crt_static_respected on Android platforms, since that hasn't been
tested.
Rollup of 11 pull requests
Successful merges:
- #75853 (Use more intra-doc-links in `core::fmt`)
- #75928 (Remove trait_selection error message in specific case)
- #76329 (Add check for doc alias attribute at crate level)
- #77219 (core::global_allocator docs link to std::alloc::GlobalAlloc)
- #77395 (BTreeMap: admit the existence of leaf edges in comments)
- #77407 (Improve build-manifest to work with the improved promote-release)
- #77426 (Include scope id in SocketAddrV6::Display)
- #77439 (Fix missing diagnostic span for `impl Trait` with const generics, and add various tests for `min_const_generics` and `const_generics`)
- #77471 (BTreeMap: refactoring around edges, missed spots)
- #77512 (Allow `Abort` terminators in all const-contexts)
- #77514 (Replace some once(x).chain(once(y)) with [x, y] IntoIter)
Failed merges:
r? `@ghost`
Replace some once(x).chain(once(y)) with [x, y] IntoIter
Now that we have by-value array iterators that are [already used](25c8c53dd9/compiler/rustc_hir/src/def.rs (L305-L307))...
For example,
```diff
- once(self.type_ns).chain(once(self.value_ns)).chain(once(self.macro_ns)).filter_map(|it| it)
+ IntoIter::new([self.type_ns, self.value_ns, self.macro_ns]).filter_map(|it| it)
```
Allow `Abort` terminators in all const-contexts
We never unwind during const-eval, so we basically have these semantics already. Also I just figured out that these only appear along the cleanup path, which doesn't get const-checked. In other words, this doesn't actually change behavior: the `check-pass` test I added compiles just fine on nightly.
r? @RalfJung
cc @rust-lang/wg-const-eval
Remove trait_selection error message in specific case
In the case that a trait is not implemented for an ADT with type errors, cancel the error.
Fixes#75627
There's a cleaner way of doing this, but it involves passing
`WithOptConstParam` around in more places. We're going to try to explore
different approaches before committing to that.
Enable RenameReturnPlace MIR optimization on mir-opt-level >= 2
The destination propagation as currently implemented does not supersede the NRVO, e.g., the destination propagation never applies if either local has an address taken, while NRVO might.
Additionally, the issue with failing assertions had been already resolved.
Continue running both optimizations at mir-opt-level >= 2.
Move target feature whitelist from cg_llvm to cg_ssa
These target features have to be supported or at least emulated by alternative codegen backends anyway as they are used by common crates. By moving this list to cg_ssa, other codegen backends don't have to copy
this code.
These target features have to be supported or at least emulated by
alternative codegen backends anyway as they are used by common crates.
By moving this list to cg_ssa, other codegen backends don't have to copy
this code.
Implement Make `handle_alloc_error` default to panic (for no_std + liballoc)
Related: https://github.com/rust-lang/rust/issues/66741
Guarded with `#![feature(default_alloc_error_handler)]` a default
`alloc_error_handler` is called, if a custom allocator is used and no
other custom `#[alloc_error_handler]` is defined.
Use `tracing` spans to trace the entire MIR interp stack
r? @RalfJung
While being very verbose, this allows really good tracking of what's going on. While I considered schemes like the previous indenter that we had (which we could get by using the `tracing-tree` crate), this will break down horribly with things like multithreaded rustc. Instead, we can now use `RUSTC_LOG` to restrict the things being traced. You could specify a filter in a way that only shows the logging of a specific frame.
If we lower the span's level to `debug`, then in `info` level logging we'd not see the frames, but in `debug` level we would see them. The filtering rules in `tracing` are super powerful, but I'm not sure if we can specify a filter so we do see `debug` level events, but *not* the `frame` spans. The documentation at https://docs.rs/tracing-subscriber/0.2.10/tracing_subscriber/struct.EnvFilter.html makes me think that we can only turn on things, not turn off things at a more precise level.
cc @hawkw
Backports LLVM commit:
[APFloat] convert SNaN to QNaN in convert() and raise Invalid signal
149f5b573c
SNaN to QNaN conversion also matches what my Intel x86_64 hardware does.
The destination propagation as currently implemented does not supersede
the NRVO, e.g., the destination propagation never applies if either
local has an address taken, while NRVO might.
Additionally, the issue with failing assertions had been already
resolved.
Continue running both optimizations at mir-opt-level >= 2.
These appear along the cleanup path inside functions with
`#[unwind(aborts)]`. We don't const-check the cleanup path anyways,
since const-eval already has "abort-on-panic" semantics and there's
often drops that would otherwise be forbidden, so the check wasn't
really preventing anything anyways.
Permit ty::Bool in const generics for v0 mangling
This should unbreak using new-symbol-mangling = true in config.toml (once it lands in beta anyway).
Fixes#76365 (well, it will, but seems fine to close as soon as we have support)
r? @eddyb (for mangling) but I'm okay with some other reviewer too :)
Bypass const_item_mutation if const's type has Drop impl
Follow-up to #75573. This PR disables the const_item_mutation lint in cases that the const has a Drop impl which observes the mutation.
```rust
struct Log { msg: &'static str }
const LOG: Log = Log { msg: "" };
impl Drop for Log {
fn drop(&mut self) { println!("{}", self.msg); }
}
LOG.msg = "wow"; // prints "wow"
```
r? @Aaron1011
Add `-Zprecise-enum-drop-elaboration`
Its purpose is to assist in debugging #77382 and #74551. Passing `-Zprecise-enum-drop-elaboration=no` will turn off the added precision that seems to be causing issues on some platforms. This assumes that we can reproduce #77382 on the latest master. I should have done this earlier. Oh well.
cc @cuviper
r? @pnkfelix
resolve: prohibit anon const non-static lifetimes
Fixes#75323, fixes#74447 and fixes#73375.
This PR prohibits non-static lifetimes in anonymous constants when only the `min_const_generics` feature is enabled. ~~To do so, `to_region_vid`'s `bug!` had to be changed into a delayed bug, which unfortunately required providing it a `TyCtxt`.~~
---
~~While I am happy with how the implementation of the error turned out in `rustc_passes::check_const`, emitting an error wasn't sufficient to avoid hitting the ICE later. I also tried implementing the error in `rustc_mir::transform::check_consts::validation` and that worked, but it didn't silence the ICE either. To silence the ICE, I changed it to a delayed bug which worked but was more invasive that I would have liked, and required I return an incorrect lifetime. It's possible that this check should be implemented earlier in the compiler to make the invasive changes unnecessary, but I wasn't sure where that would be and wanted to get some feedback first.~~
The approach taken by this PR has been changed to implement the error in name resolution, which ended up being much simpler.
cc @rust-lang/wg-const-eval
r? @lcnr
This commit modifies name resolution to emit an error when non-static
lifetimes are used in anonymous constants when the `min_const_generics`
feature is enabled.
Signed-off-by: David Wood <david@davidtw.co>
Disable the SimplifyArmIdentity mir-opt
The optimization still has some bugs that need to be worked out
such as #77359.
We can try re-enabling this again after the known issues are resolved.
r? `@oli-obk`
Related: https://github.com/rust-lang/rust/issues/66741
Guarded with `#![feature(default_alloc_error_handler)]` a default
`alloc_error_handler` is called, if a custom allocator is used and no
other custom `#[alloc_error_handler]` is defined.
The panic message does not contain the size anymore, because it would
pull in the fmt machinery, which would blow up the code size
significantly.
rustc_metadata: Do not forget to encode inherent impls for foreign types
So I tried to move FFI interface for LLVM from `rustc_codegen_llvm` to `rustc_llvm` and immediately encountered this fascinating issue.
Fixes https://github.com/rust-lang/rust/issues/46665.
Overhaul const-checking diagnostics
The primary purpose of this PR was to remove `NonConstOp::STOPS_CONST_CHECKING`, which causes any additional errors found by the const-checker to be silenced. I used this flag to preserve diagnostic parity with `qualify_min_const_fn.rs`, which has since been removed.
However, simply removing the flag caused a deluge of errors in some cases, since an error would be emitted any time a local or temporary had a wrong type. To remedy this, I added an alternative system (`DiagnosticImportance`) to silence additional error messages that were likely to distract the user from the underlying issue. When an error of the highest importance occurs, all less important errors are silenced. When no error of the highest importance occurs, all less important errors are emitted after checking is complete. Following the suggestions from the important error is usually enough to fix the less important errors, so this should lead to better UX most of the time.
There's also some unrelated diagnostics improvements in this PR isolated in their own commits. Splitting them out would be possible, but a bit of a pain. This isn't as tidy as some of my other PRs, but it should *only* affect diagnostics, never whether or not something passes const-checking. Note that there are a few trivial exceptions to this, like banning `Yield` in all const-contexts, not just `const fn`.
As always, meant to be reviewed commit-by-commit.
r? `@oli-obk`
adt_destructor by default also validates the Drop impl using
dropck::check_drop_impl, which contains an expect_local(). This
leads to ICE in check_const_item_mutation if the const's type is
not a local type.
thread 'rustc' panicked at 'DefId::expect_local: `DefId(5:4805 ~ alloc[d7e9]::vec::{impl#50})` isn't local', compiler/rustc_span/src/def_id.rs:174:43
stack backtrace:
0: rust_begin_unwind
1: rustc_span::def_id::DefId::expect_local::{{closure}}
2: rustc_typeck::check::dropck::check_drop_impl
3: rustc_middle::ty::util::<impl rustc_middle::ty::context::TyCtxt>::calculate_dtor::{{closure}}
4: rustc_middle::ty::trait_def::<impl rustc_middle::ty::context::TyCtxt>::for_each_relevant_impl
5: rustc_middle::ty::util::<impl rustc_middle::ty::context::TyCtxt>::calculate_dtor
6: rustc_typeck::check::adt_destructor
7: rustc_middle::ty::query::<impl rustc_query_system::query::config::QueryAccessors<rustc_middle::ty::context::TyCtxt> for rustc_middle::ty::query::queries::adt_destructor>::compute
8: rustc_query_system::dep_graph::graph::DepGraph<K>::with_task_impl
9: rustc_query_system::query::plumbing::get_query_impl
10: rustc_mir::transform::check_const_item_mutation::ConstMutationChecker::is_const_item_without_destructor
Stable hashing: add comments and tests concerning platform-independence
SipHasher128 implements short_write in an endian-independent way, yet
its write_xxx Hasher trait methods undo this endian-independence by byte
swapping the integer inputs on big-endian hardware. StableHasher then
adds endian-independence back by also byte-swapping on big-endian
hardware prior to invoking SipHasher128.
This double swap may have the appearance of being a no-op, but is in
fact by design. In particular, we really do want SipHasher128 to be
platform-dependent, in order to be consistent with the libstd SipHasher.
Try to clarify this intent. Also, add and update a couple of unit tests.
---
Previous commit text:
~SipHasher128: fix platform-independence confusion~
~StableHasher is supposed to ensure platform independence by converting
integers to little-endian and extending isize and usize to 64 bits as
necessary, but in fact, much of that work is already handled by
SipHasher128.~
~In particular, SipHasher128 implements short_write in an
endian-independent way, yet both StableHasher and SipHasher128
additionally attempt to achieve endian-independence by byte swapping on
BE hardware before invoking short writes. This double swap has no
effect, so let's remove it.~
~Because short_write is endian-independent, SipHasher128 is already
handling part of the platform-independence, and it would be somewhat
difficult to make it *not* handle that part with the current
implementation. As splitting platform-independence responsibilities
between StableHasher and SipHasher128 would be confusing, let's make
SipHasher128 handle all of it.~
~Finally, update some incorrect comments and increase test coverage.
Unit tests pass on both LE and BE systems.~
const evaluatable: improve `TooGeneric` handling
Instead of emitting an error in `fulfill`, we now correctly stall on inference variables.
As `const_eval_resolve` returns `ErrorHandled::TooGeneric` when encountering generic parameters on which
we actually do want to error, we check for inference variables and eagerly emit an error if they don't exist, returning `ErrorHandled::Reported` instead.
Also contains a small bugfix for `ConstEquate` where we previously only stalled on type variables. This is probably a leftover from
when we did not yet support stalling on const inference variables.
r? @oli-obk cc @varkor @eddyb
Defer Apple SDKROOT detection to link time.
This defers the detection of the SDKROOT for Apple iOS/tvOS targets to link time, instead of when the `Target` is defined. This allows commands that don't need to link to work (like `rustdoc` or `rustc --print=target-list`). This also makes `--print=target-list` a bit faster.
This also removes the note in the platform support documentation about these targets being missing. When I wrote it, I misunderstood how the SDKROOT stuff worked.
Notes:
* This means that JSON spec targets can't explicitly override these flags. I think that is probably fine, as I believe the value is generally required, and can be set with the SDKROOT environment variable.
* This changes `x86_64-apple-tvos` to use `appletvsimulator`. I think the original code was wrong (it was using `iphonesimulator`). Also, `x86_64-apple-tvos` seems broken in general, and I cannot build it locally. The `data_layout` does not appear to be correct (it is a copy of the arm64 layout instead of the x86_64 layout). I have not tried building Apple's LLVM to see if that helps, but I suspect it is just wrong (I'm uncertain since I don't know how the tvOS simulator works with its bitcode-only requirements).
* I'm tempted to remove the use of `Result` for built-in target definitions, since I don't think they should be fallible. This was added in https://github.com/rust-lang/rust/pull/34980, but that only relates to JSON definitions. I think the built-in targets shouldn't fail. I can do this now, or not.
Fixes#36156Fixes#76584
Fix recursive nonterminal expansion during pretty-print/reparse check
Makes progress towards #43081
In PR #73084, we started recursively expanded nonterminals during the
pretty-print/reparse check, allowing them to be properly compared
against the reparsed tokenstream.
Unfortunately, the recursive logic in that PR only handles the case
where a nonterminal appears inside a `TokenTree::Delimited`. If a
nonterminal appears directly in the expanded tokens of another
nonterminal, the inner nonterminal will not be expanded.
This PR fixes the recursive expansion of nonterminals, ensuring that
they are expanded wherever they occur.
Rollup of 12 pull requests
Successful merges:
- #77037 (more tiny clippy cleanups)
- #77233 (BTreeMap: keep an eye out on the size of the main components)
- #77280 (Ensure that all LLVM components requested by tests are available on CI)
- #77284 (library: Forward compiler-builtins "mem" feature)
- #77296 (liveness: Use Option::None to represent absent live nodes)
- #77322 (Add unstable book docs for `-Zunsound-mir-opts`)
- #77328 (Use `rtassert!` instead of `assert!` from the child process after fork() in std::sys::unix::process::Command::spawn())
- #77331 (Add test for async/await combined with const-generics.)
- #77338 (Fix typo in alloc vec comment)
- #77340 (Alloc vec use imported path)
- #77345 (Add test for issue #74761)
- #77348 (Update books)
Failed merges:
r? `@ghost`
Add support for cmse_nonsecure_entry attribute
This pull request adds the `cmse_nonsecure_entry` attribute under an unstable feature.
I was not sure if it was fine for me to send directly the pull-request or if I should submit a RFC first. I was told on Zulip that it was fine to do so but please close it if I need first submit a RFC or follow another process instead.
The `cmse_nonsecure_entry` attribute is a LLVM attribute that will be available in LLVM 11. I plan to rebase on the [upgrade PR](https://github.com/rust-lang/rust/pull/73526) once merged to make this one compile.
This attribute modifies code generation of the function as explained [here](https://developer.arm.com/documentation/ecm0359818/latest/) to make it work with the TrustZone-M hardware feature. This feature is only available on `thumbv8m` targets so I created an error for that if one tries to use this attribute for another target.
I added this attribute in Rust as any other LLVM attribute are added but since this one is target-dependent I am not sure if it was the best thing to do. Please indicate me if you think of other ways, like isolating target-dependent attributes together.
----------------
Tracking issue: https://github.com/rust-lang/rust/issues/75835
This patch adds support for the LLVM cmse_nonsecure_entry attribute.
This is a target-dependent attribute that only has sense for the
thumbv8m Rust targets.
You can find more information about this attribute here:
https://developer.arm.com/documentation/ecm0359818/latest/
Signed-off-by: Hugues de Valon <hugues.devalon@arm.com>
Secure entry functions do not support if arguments are passed on the
stack. An "unsupported" diagnostic will be emitted by LLVM if that is
the case.
This commits adds support in Rust for that diagnostic so that an error
will be output if that is the case!
Signed-off-by: Hugues de Valon <hugues.devalon@arm.com>
SipHasher128 implements short_write in an endian-independent way, yet
its write_xxx Hasher trait methods undo this endian-independence by byte
swapping the integer inputs on big-endian hardware. StableHasher then
adds endian-independence back by also byte-swapping on big-endian
hardware prior to invoking SipHasher128.
This double swap may have the appearance of being a no-op, but is in
fact by design. In particular, we really do want SipHasher128 to be
platform-dependent, in order to be consistent with the libstd SipHasher.
Try to clarify this intent. Also, add and update a couple of unit tests.
