btree: don't leak value if destructor of key panics
This PR fixes a regression from https://github.com/rust-lang/rust/pull/84904.
The `BTreeMap` already attempts to handle panicking destructors of the key-value pairs by continuing to execute the remaining destructors after one destructor panicked. However, after #84904 the destructor of a value in a key-value pair gets skipped if the destructor of the key panics, only continuing with the next key-value pair. This PR reverts to the behavior before #84904 to also drop the corresponding value if the destructor of a key panics.
This avoids potential memory leaks and can fix the soundness of programs that rely on the destructors being executed (even though this should not be relied upon, because the std collections currently do not guarantee that the remaining elements are dropped after a panic in a destructor).
cc `@Amanieu` because you had opinions on panicking destructors
Update TRPL to add new Chapter 17: Async and Await
- Add support to `rustbook` to pass through the `-L`/`--library-path` flag to `mdbook` so that references to the `trpl` crate
- Build the `trpl` crate as part of the book tests. Make it straightforward to add other such book dependencies in the future if needed by implementing that in a fairly general way.
- Update the submodule for the book to pull in the new chapter on async and await, as well as a number of other fixes. This will happen organically/automatically in a week, too, but this lets me group this change with the next one:
- Update the compiler messages which reference the existing chapters 17–20, which are now chapters 18-21. There are only two, both previously referencing chapter 18.
- Update the UI tests which reference the compiler message outputs.
finish `Reveal` removal
After #133212 changed the `TypingMode` to be the only source of truth, this entirely rips out `Reveal`.
cc #132279
r? `@compiler-errors`
Without this change:
$ ./x test --set build.vendor=true src/doc/book
# (lots of output)
error: failed to select a version for the requirement `futures = "^0.3"` (locked to 0.3.30)
candidate versions found which didn't match: 0.3.31, 0.3.27
location searched: directory source `/Users/chris/dev/rust-lang/rust/vendor` (which is replacing registry `crates-io`)
required by package `trpl v0.2.0 (/Users/chris/dev/rust-lang/rust/src/doc/book/packages/trpl)`
perhaps a crate was updated and forgotten to be re-vendored?
Build completed unsuccessfully in 0:01:19
With this change:
$ ./x test --set build.vendor=true src/doc/book
# (lots of build output)
Testing stage1 mdbook src/doc/book (aarch64-apple-darwin)
finished in 86.949 seconds
Build completed successfully in 0:04:05
# Conflicts:
# src/bootstrap/src/core/build_steps/vendor.rs
Since TRPL now depends on a `trpl` crate, the test needs to be able to
build that crate to run mdbook against it, and also to invoke mdbook
with `--library-path` in that case. Use the support for that flag added
to `rustbook` in the previous change to invoke it with the path to the
dependencies it will need to run `rustdoc` tests which reference `trpl`.
Co-authored-by: Onur Özkan <onurozkan.dev@outlook.com>
This makes it possible to test an mdbook which has dependencies other
than the direct crate for the book itself, e.g. the `trpl` crate used in
_The Rust Programming Language_.
With the insertion of a new chapter 17 on async and await to _The Rust
Programming Language_, references in compiler output to later chapters
need to be updated to avoid confusing users. Redirects exist so that
users who click old links will end up in the right place anyway, but
this way users will be directed to the right URL in the first place.
This also incorporates a number of other changes and fixes which would
normally have been part of the automatic update, but which were blocked
from landing because of the changes required to support shipping a crate
as part of the chapter, along with those changes.
Warn if `rust.download-rustc = true` is used with
`rust.debug-assertions` as alt CI rustc is not currently built with
debug assertions (not to be confused with LLVM assertions).
Rollup of 6 pull requests
Successful merges:
- #127483 (Allow disabling ASan instrumentation for globals)
- #131505 (use `confstr(_CS_DARWIN_USER_TEMP_DIR, ...)` as a `TMPDIR` fallback on Darwin)
- #132949 (Add specific diagnostic for using macro_rules macro as attribute/derive)
- #133286 (Re-delay a resolve `bug` related to `Self`-ctor in patterns)
- #133332 (Mark `<[T; N]>::as_mut_slice` with the `const` specifier.)
- #133366 (Remove unnecessary bool from `ExpectedFound::new`)
r? `@ghost`
`@rustbot` modify labels: rollup
show abi_unsupported_vector_types lint in future breakage reports
The lint is now riding the train to 1.84. Given that crater found no case of this lint triggering at all, IMO it's fine to make it "report in deps" already for 1.85.
Part of https://github.com/rust-lang/rust/issues/116558.
Add language tests for aggregate types
This adds some tests for struct and union types, ensuring that they satisfy the rules for all structs and unions - namely that fields of structs do not overlap, fields are well-aligned, and the size of the entire.
The reference annotations used are from https://github.com/rust-lang/reference/pull/1654, though the rules tested here were FCPed in <https://github.com/rust-lang/reference/pull/1152>.
Minimally constify `Add`
* This PR removes the requirement for `impl const` to have a const stability attribute. cc ``@RalfJung`` I believe you mentioned that it would make much more sense to require `const_trait`s to have const stability instead. I agree with that sentiment but I don't think that is _required_ for a small scale experimentation like this PR. https://github.com/rust-lang/project-const-traits/issues/16 should definitely be prioritized in the future, but removing the impl check should be good for now as all callers need `const_trait_impl` enabled for any const impl to work.
* This PR is intentionally minimal as constifying other traits can become more complicated (`PartialEq`, for example, would run into requiring implementing it for `str` as that is used in matches, which runs into the implementation for slice equality which uses specialization)
Per the reasons above, anyone who is interested in making traits `const` in the standard library are **strongly encouraged** to reach out to us on the [Zulip channel](https://rust-lang.zulipchat.com/#narrow/channel/419616-t-compiler.2Fproject-const-traits) before proceeding with the work.
cc ``@rust-lang/project-const-traits``
I believe there is prior approval from libs that we can experiment, so
r? project-const-traits
[AIX] Add option -X32_64 to the "strip" command
The AIX `strip` utility requires option `-X` to specify the object mode. This patch adds the `-X32_64` option to the `strip` command so that it can handle both 32-bit and 64-bit objects. The parameter `option` of function `strip_symbols_with_external_utility`, previously a single string, has been changed to `options`, an array of string slices, to accommodate multiple `strip` options.
Mark `<[T; N]>::as_mut_slice` with the `const` specifier.
Tracking issue: #133333
`<[T; N]>::as_mut_slice` can have the `const` specifier without any changes to the function body.
Re-delay a resolve `bug` related to `Self`-ctor in patterns
For the code pattern reported in <https://github.com/rust-lang/rust/issues/133272>,
```rs
impl Foo {
fn fun() {
let S { ref Self } = todo!();
}
}
```
<https://github.com/rust-lang/rust/pull/121208> converted this to a `span_bug` from a `span_delayed_bug` because this specific self-ctor code pattern lacked test coverage. It turns out this can be hit but we just lacked test coverage, so change it back to a `span_delayed_bug` and add a targeted test case.
Follow-up to #121208, cc ``@nnethercote`` (very good exercise to expose our test coverage gaps).
Fixes#133272.
Allow disabling ASan instrumentation for globals
AddressSanitizer adds instrumentation to global variables unless the [`no_sanitize_address`](https://llvm.org/docs/LangRef.html#global-attributes) attribute is set on them.
This commit extends the existing `#[no_sanitize(address)]` attribute to set this; previously it only had the desired effect on functions.
(cc https://github.com/rust-lang/rust/issues/39699)
Rollup of 8 pull requests
Successful merges:
- #132090 (Stop being so bail-y in candidate assembly)
- #132658 (Detect const in pattern with typo)
- #132911 (Pretty print async fn sugar in opaques and trait bounds)
- #133102 (aarch64 softfloat target: always pass floats in int registers)
- #133159 (Don't allow `-Zunstable-options` to take a value )
- #133208 (generate-copyright: Now generates a library file too.)
- #133215 (Fix missing submodule in `./x vendor`)
- #133264 (implement OsString::truncate)
r? `@ghost`
`@rustbot` modify labels: rollup
Fix missing submodule in `./x vendor`
The `src/tools/rustc-perf` submodule is needed for vendoring because it is included in the vendor set.
To test this:
1. Get a fresh clone of `rust-lang/rust`
2. `./x vendor`
generate-copyright: Now generates a library file too.
We only run reuse once, so the output has to be filtered to find only the files that are relevant to the library tree.
Outputs COPYRIGHT.html and COPYRIGHT-library.html.
The license-metadata.json file is also now in the tree. We need a CI tool to check that it's correct.
r? kobzol
Remaining steps:
* [ ] Teach CI to double-check the license-metadata.json file is correct
* [ ] Add the COPYRIGHT.html and COPYRIGHT-license.html to the releases
Don't allow `-Zunstable-options` to take a value
Passing an explicit boolean value (`-Zunstable-options=on`, `off` etc.) sometimes appears to work, but actually puts the compiler into an unintended state where unstable _options_ are still forbidden, but unstable values of _some_ stable options are allowed.
This is a result of `-Zunstable-options` being checked in multiple different places, in slightly different ways. Fixing the checks in `config::nightly_options` to understand boolean values would be non-trivial, so for now it's easier to make things consistent by forbidding values in the `-Z` parser.
---
There were a few uses of this in tests, which happened to work because they were tests of unstable values.
aarch64 softfloat target: always pass floats in int registers
This is a part of https://github.com/rust-lang/rust/issues/131058: on softfloat aarch64 targets, the float registers may be unavailable. And yet, LLVM will happily use them to pass float types if the corresponding target features are enabled. That's a problem as it means enabling/disabling `neon` instructions can change the ABI.
Other targets have a `soft-float` target feature that forces the use of the soft-float ABI no matter whether float registers are enabled or not; aarch64 has nothing like that.
So we follow the aarch64 [softfloat ABI](https://github.com/rust-lang/rust/issues/131058#issuecomment-2385027423) and treat floats like integers for `extern "C"` functions. For the "Rust" ABI, we do the same for scalars, and then just do something reasonable for ScalarPair that avoids the pointer indirection.
Cc ```@workingjubilee```
