translation: eager translation
Part of #100717. See [Zulip thread](https://rust-lang.zulipchat.com/#narrow/stream/336883-i18n/topic/.23100717.20lists!/near/295010720) for additional context.
- **Store diagnostic arguments in a `HashMap`**: Eager translation will enable subdiagnostics to be translated multiple times with different arguments - this requires the ability to replace the value of one argument with a new value, which is better suited to a `HashMap` than the previous storage, a `Vec`.
- **Add `AddToDiagnostic::add_to_diagnostic_with`**: `AddToDiagnostic::add_to_diagnostic_with` is similar to the previous `AddToDiagnostic::add_to_diagnostic` but takes a function that can be used by the caller to modify diagnostic messages originating from the subdiagnostic (such as performing translation eagerly). `add_to_diagnostic` now just calls `add_to_diagnostic_with` with an empty closure.
- **Add `DiagnosticMessage::Eager`**: Add variant of `DiagnosticMessage` for eagerly translated messages
(messages in the target language which don't need translated by the emitter during emission). Also adds `eager_subdiagnostic` function which is intended to be invoked by the diagnostic derive for subdiagnostic fields which are marked as needing eager translation.
- **Support `#[subdiagnostic(eager)]`**: Add support for `eager` argument to the `subdiagnostic` attribute which generates a call to `eager_subdiagnostic`.
- **Finish migrating `rustc_query_system`**: Using eager translation, migrate the remaining repeated cycle stack diagnostic.
- **Split formatting initialization and use in diagnostic derives**: Diagnostic derives have previously had to take special care when ordering the generated code so that fields were not used after a move.
This is unlikely for most fields because a field is either annotated with a subdiagnostic attribute and is thus likely a `Span` and copiable, or is a argument, in which case it is only used once by `set_arg`
anyway.
However, format strings for code in suggestions can result in fields being used after being moved if not ordered carefully. As a result, the derive currently puts `set_arg` calls last (just before emission), such as:
let diag = { /* create diagnostic */ };
diag.span_suggestion_with_style(
span,
fluent::crate::slug,
format!("{}", __binding_0),
Applicability::Unknown,
SuggestionStyle::ShowAlways
);
/* + other subdiagnostic additions */
diag.set_arg("foo", __binding_0);
/* + other `set_arg` calls */
diag.emit();
For eager translation, this doesn't work, as the message being translated eagerly can assume that all arguments are available - so arguments _must_ be set first.
Format strings for suggestion code are now separated into two parts - an initialization line that performs the formatting into a variable, and a usage in the subdiagnostic addition.
By separating these parts, the initialization can happen before arguments are set, preserving the desired order so that code compiles, while still enabling arguments to be set before subdiagnostics are added.
let diag = { /* create diagnostic */ };
let __code_0 = format!("{}", __binding_0);
/* + other formatting */
diag.set_arg("foo", __binding_0);
/* + other `set_arg` calls */
diag.span_suggestion_with_style(
span,
fluent::crate::slug,
__code_0,
Applicability::Unknown,
SuggestionStyle::ShowAlways
);
/* + other subdiagnostic additions */
diag.emit();
- **Remove field ordering logic in diagnostic derive:** Following the approach taken in earlier commits to separate formatting initialization from use in the subdiagnostic derive, simplify the diagnostic derive by removing the field-ordering logic that previously solved this problem.
r? ```@compiler-errors```
Prevent UB in child process after calling libc::fork
After calling libc::fork, the child process tried to access a TLS variable when processing a panic. This caused a memory allocation which is UB in the child.
To prevent this from happening, the panic handler will not access the TLS variable in case `panic::always_abort` was called before.
Fixes#85261 (not only on Android systems, but also on Linux/QNX with TLS disabled, see issue for more details)
Main drawbacks of this fix:
* Panic messages can incorrectly omit `core::panic::PanicInfo` struct in case several panics (of multiple threads) occur at the same time. The handler cannot distinguish between multiple panics in different threads or recursive ones in the same thread, but the message will contain a hint about the uncertainty.
* `panic_count::increase()` will be a bit slower as it has an additional `if`, but this should be irrelevant as it is only called in case of a panic.
Remove `TokenStreamBuilder`
`TokenStreamBuilder` is used to combine multiple token streams. It can be removed, leaving the code a little simpler and a little faster.
r? `@Aaron1011`
Interpret EH actions properly
The EH actions stored in the LSDA follows the format of GCC except table (even for LLVM-generated code). An missing action in the table is the encoding for `Terminate`, see https://github.com/gcc-mirror/gcc/blob/master/libstdc%2B%2B-v3/libsupc%2B%2B/eh_personality.cc#L522-L526.
The currently code interprets it as `None`, as a workaround for #35011, an issue that seems to occur in LLVM 3.7 and not after 3.9. These are very old versions of LLVM and we don't support them anymore, so remove this workaround and interpret them properly.
Note that LLVM currently does not emit any `Terminate` actions, but GCC does. Although GCC backend currently doesn't do unwinding, removing it preemptively would prevent future developers from wasting time to figure out what's wrong.
``@rustbot`` label: +T-compiler
Migrate `codegen_ssa` to diagnostics structs - [Part 1]
Initial migration of `codegen_ssa`. Going to split this crate migration in at least two PRs in order to avoid a huge PR and to quick off some questions around:
1. Translating messages from "external" crates.
2. Interfacing with OS messages.
3. Adding UI tests while migrating diagnostics.
_See comments below._
fs::get_path solarish version.
similar to linux, albeit there is no /proc/self notion on solaris
based system thus flattening the difference for simplification sake.
Warn about safety of `fetch_update`
Specifically as it relates to the ABA problem.
`fetch_update` is a useful function, and one that isn't provided by, say, C++. However, this does not mean the function is magic. It is implemented in terms of `compare_exchange_weak`, and in particular, suffers from the ABA problem. See the following code, which is a naive implementation of `pop` in a lock-free queue:
```rust
fn pop(&self) -> Option<i32> {
self.front.fetch_update(Ordering::Relaxed, Ordering::Acquire, |front| {
if front == ptr::null_mut() {
None
}
else {
Some(unsafe { (*front).next })
}
}.ok()
}
```
This code is unsound if called from multiple threads because of the ABA problem. Specifically, suppose nodes are allocated with `Box`. Suppose the following sequence happens:
```
Initial: Queue is X -> Y.
Thread A: Starts popping, is pre-empted.
Thread B: Pops successfully, twice, leaving the queue empty.
Thread C: Pushes, and `Box` returns X (very common for allocators)
Thread A: Wakes up, sees the head is still X, and stores Y as the new head.
```
But `Y` is deallocated. This is undefined behaviour.
Adding a note about this problem to `fetch_update` should hopefully prevent users from being misled, and also, a link to this common problem is, in my opinion, an improvement to our docs on atomics.
Check uniqueness of impl items by trait item when applicable.
When checking uniqueness of item names in impl blocks, we currently use the same definition of hygiene as for toplevel items. This means that a plain item and one generated by a macro 2.0 do not collide.
This hygiene rule does not match with how impl items resolve to associated trait items. As a consequence, we misdiagnose the trait impls.
This PR proposes to consider that trait impl items are uses of the corresponding trait items during resolution, instead of checking for duplicates later. An error is emitted when a trait impl item is used twice.
There should be no stable breakage, since macros 2.0 are still unstable.
r? ``@petrochenkov``
cc ``@RalfJung``
Fixes https://github.com/rust-lang/rust/issues/71614.
tools/remote-test-{server,client}: Use /data/local/tmp on Android
The /data/tmp directory does not exist, at least not on recent versions of Android, which currently leads to test failures on that platform. I checked a virtual device running AOSP master and a Nexus 5 running Android Marshmallow and on both devices the /data/tmp directory does not exist and /data/local/tmp does, so let's switch to /data/local/tmp.
Fix the sanitizer_scs_attr_check.rs test
The test is failing when targeting aarch64 Android. The intent appears to have been to look for a function attributes comment (or the absence of one) on the line preceding the function declaration. But this isn't quite possible with FileCheck and the test as written was looking for a line with `no_scs` after a line with `scs`, which doesn't appear in the output. Instead, match on the function attributes comment on the line following the demangled function name comment.
Improve rustdoc-gui search-color test
Thanks to the add of "functions" in `browser-ui-test`, we can start to reduce the size of the scripts. It'll be very useful for all color checks.
r? `@notriddle`
suggest candidates for unresolved import
Currently we prompt suggestion of candidates(help notes of `use xxx::yyy`) for names which cannot be resolved, but we don't do that for import statements themselves that couldn't be resolved. It seems reasonable to add candidate help information for these statements as well.
Fixes#102711
rustdoc: clean up overly complex `.trait-impl` CSS selectors
When added in 45964368f4a2e31c94e9bcf1cef933c087d21544, these multi-class selectors were present in the initial commit, but no reason was given why the shorter selector wouldn't work.
update to syn-1.0.102
This update removes the only `.gitignore` found in `rustc-src`:
vendor/syn/tests/.gitignore
vendor/syn-1.0.91/tests/.gitignore
vendor/syn-1.0.95/tests/.gitignore
To check-in `rustc-src` for hermetic builds in environments with
restrictive `.gitignore` policies, one has to remove these
`tests/.gitignore` and patch the respective
`.cargo-checksum.json`.`syn` >1.0.101 includes dtolnay/syn@3c49303bed,
which removes its `tests/.gitignore`. Now the `syn` crates.io package
has no `.gitignore`.
[`rustc-src`'s `vendor`][] is produced from the root `Cargo.toml`,
`src/tools/rust-analyzer/Cargo.toml`,
`compiler/rustc_codegen_cranelift/Cargo.toml`, and
`src/bootstrap/Cargo.toml`. `rustc_codegen_cranelift` does not use
`syn`.
[`rustc-src`'s `vendor`]:
https://github.com/rust-lang/rust/blob/c0784109daa0/src/bootstrap/dist.rs#L934-L940
This was produced with:
cargo update --package syn --precise 1.0.102 \
cargo update --package syn --precise 1.0.102 \
--manifest-path src/tools/rust-analyzer/Cargo.toml
cargo update --package syn --precise 1.0.102 \
--manifest-path src/bootstrap/Cargo.toml
Use BOLT in CI to optimize LLVM
This PR adds an optimization step in the Linux `dist` CI pipeline that uses [BOLT](https://github.com/llvm/llvm-project/tree/main/bolt) to optimize the `libLLVM.so` library built by boostrap.
Steps:
- [x] Use LLVM 15 as a bootstrap compiler and use it to build BOLT
- [x] Compile LLVM with support for relocations (`-DCMAKE_SHARED_LINKER_FLAGS="-Wl,-q"`)
- [x] Gather profile data using instrumented LLVM
- [x] Apply profile to LLVM that has already been PGOfied
- [x] Run with BOLT profiling on more benchmarks
- [x] Decide on the order of optimization (PGO -> BOLT?)
- [x] Decide how we should get `bolt` (currently we use the host `bolt`)
- [x] Clean up
The latest perf results can be found [here](https://github.com/rust-lang/rust/pull/94381#issuecomment-1258269440). The current CI build time with BOLT applied is around 1h 55 minutes.
add Vec::push_within_capacity - fallible, does not allocate
This method can serve several purposes. It
* is fallible
* guarantees that items in Vec aren't moved
* allows loops that do `reserve` and `push` separately to avoid pulling in the allocation machinery a second time in the `push` part which should make things easier on the optimizer
* eases the path towards `ArrayVec` a bit since - compared to `push()` - there are fewer questions around how it should be implemented
I haven't named it `try_push` because that should probably occupy a middle ground that will still try to reserve and only return an error in the unlikely OOM case.
resolves#84649
internal: ⬆️ xflags
The main change here should be that flags are not inhereted, so
$ rust-analyzer analysis-stats . -v -v
would do what it should do
We also no longer Don\'t
The main change here should be that flags are not inhereted, so
$ rust-analyzer analysis-stats . -v -v
would do what it should do
We also no longer Don\'t
std: use futex in `Once`
Now that we have efficient locks, let's optimize the rest of `sync` as well. This PR adds a futex-based implementation for `Once`, which drastically simplifies the implementation compared to the generic version, which is provided as fallback for platforms without futex (Windows only supports them on newer versions, so it uses the fallback for now).
Instead of storing a linked list of waiters, the new implementation adds another state (`QUEUED`), which is set when there are waiting threads. These now use `futex_wait` on that state and are woken by the running thread when it finishes and notices the `QUEUED` state, thereby avoiding unnecessary calls to `futex_wake_all`.
fix: Make go-to-def work for `#[doc = include_str!("path")]`
See the added test, go-to-def on `#[doc = include_str!("path$0")]` should navigate to `path`.
minor: Fix go-to-def for shadowed `include*!`
Add a check in go-to-def feature, so that we don't assume any macro named `include`/`include_str`/`include_bytes` is the builtin one.
Miri sync
This is a Miri sync created with my experimental fork of josh. We should probably not merge this yet, but we can use this to check if the sync looks the way it should.
r? `@oli-obk`
Fix unwind drop glue for if-then scopes
cc `@est31`
Fix#102317Fix#99852
This PR fixes the drop glue for unwinding from a panic originated in a drop while breaking out for the else block in an `if-then` scope.
MIR validation does not fail for the synchronous versions of the test program, because `StorageDead` statements are skipped over in the unwinding process. It is only becoming a problem when it is inside a generator where `StorageDead` must be kept around.
Adding target_rustcflags to `compiletest` TargetCfg creation
Adjustment to https://github.com/rust-lang/rust/pull/102134, ensures config returned by `rustc --target foo --print cfg` accurately reflects rustflags passed via `target_rustcflags`.
Fixes breaking change of not correctly handling `x.py test ... --test-args "--target-rustcflags -Cpanic=abort --target-rustcflags -Zpanic_abort_tests"`
cc `@djkoloski`
Add `AsFd` implementations for stdio lock types on WASI.
This mirrors the implementations on Unix platforms, and also mirrors the existing `AsRawFd` impls.
This is similar to #100892, but is for the `*Lock` types.
fix: use `BoundVar`s from current generic scope
Fixup for #13335, addresses https://github.com/rust-lang/rust-analyzer/pull/13339#issuecomment-1266654607
Before the change in generic parameter order, `BoundVar`s for trait reference didn't change whether you are in an impl's scope or in an associated item's scope. Now that item's generic params come before its parent's, we need to shift their indices when we are in an associated item's scope.
Move layout_of and friends from rustc_middle to rustc_ty_utils
Breaks up the very large module that is `rustc_middle::ty::layout` by fork-lifting some queries into `rustc_ty_utils::{abi, layout}`.
This does set back `rustc_ty_utils` to having untranslatable diagnostics. I'd like to leave this as a separate task.
fix: treat enum variants as generic item on their own
Fixup for #13335
It turns out I tried to merge two procedures into one utility function without noticing the incompatibility.
This time I *did* run analysis-stats on the four crates and confirmed it doesn't crash and this patch doesn't cause regression.
