interpret: on a signed deref check, mention the right pointer in the error
When a negative offset (like `ptr.offset(-10)`) goes out-of-bounds, we currently show an error saying that we expect the *resulting* pointer to be inbounds for 10 bytes. That's confusing, so this PR makes it so that instead we say that we expect the *original* pointer `ptr` to have 10 bytes *to the left*.
I also realized I can simplify the pointer arithmetic logic and handling of "staying inbounds of a target `usize`" quite a bit; the second commit does that.
Add target page for riscv64gc-unknown-linux-gnu
I was reading https://github.com/rust-lang/rust/issues/113739 and realized I knew most of the information necessary to create the `riscv64gc-unknown-linux-gnu` target page.
More unsafe attr verification
This code denies unsafe on attributes such as `#[test]` and `#[ignore]`, while also changing the `MetaItem` parsing so `unsafe` in args like `#[allow(unsafe(dead_code))]` is not accidentally allowed.
Tracking:
- https://github.com/rust-lang/rust/issues/123757
When collecting tokens there are two kinds of range:
- a range relative to the parser's full token stream (which we get when
we are parsing);
- a range relative to a single AST node's token stream (which we use
within `LazyAttrTokenStreamImpl` when replacing tokens).
These are currently both represented with `Range<u32>` and it's easy to
mix them up -- until now I hadn't properly understood the difference.
This commit introduces `ParserRange` and `NodeRange` to distinguish
them. This also requires splitting `ReplaceRange` in two, giving the new
types `ParserReplacement` and `NodeReplacement`. (These latter two names
reduce the overloading of the word "range".)
The commit also rewrites some comments to be clearer.
The end result is a little more verbose, but much clearer.
docs: Fix JSON example for rust-analyzer.workspace.discoverConfig
The user does not specify `{arg}` in their JSON, and be pedantic about commas in JSON sample.
Finish blessing `coverage/mcdc` tests after LLVM 19 upgrade
Context: https://github.com/rust-lang/rust/pull/127513#issuecomment-2260887708
These changes aren't needed for Rust CI, because after the LLVM 19 upgrade we have no jobs that run these tests in `coverage-run` mode against LLVM 18. But they might help external builds.
The longer-term plan is to completely drop (unstable) MC/DC support on LLVM 18, as part of getting it working on LLVM 19 in #126733.
cc `@cuviper`
Emit an error if `#[optimize]` is applied to an incompatible item
#54882
The RFC specifies that this should emit a lint. I used the same allow logic as the `coverage` attribute (also allowing modules and impl blocks) - this should possibly be changed depending on if it's decided to allow 'propogation' of the attribute.
Create COFF archives for non-LLVM backends
`ar_archive_writer` now supports creating COFF archives, so enable them for the non-LLVM backends when requested.
r? ``@bjorn3``
improve bootstrap to allow selecting llvm tools individually
Everything works as before, + now bootstrap allows for individually selecting LLVM tools (e.g., `x dist opt llvm-dis`) to include in the dist artifact.
android: Remove libstd hacks for unsupported Android APIs
Our minimum supported API version is 21, remove hacks to support older Android APIs.
try-job: arm-android
r? tgross35
Migrate `symbol-visibility` `run-make` test to rmake
Part of #121876 and the associated [Google Summer of Code project](https://blog.rust-lang.org/2024/05/01/gsoc-2024-selected-projects.html).
Pretty scary!
- The expected number of symbols on each check has been changed slightly to reflect the differences between `llvm_readobj` and `nm`, as I think the former will print hidden symbols once and visible symbols twice, while the latter will only print visible symbols.
- The original test ran the same exact checks on `cdylib` twice, for seemingly no reason. I have removed it.
- This may be possible to optimize some more? `llvm_readobj` could get called only once for each library type, and the regex could avoid being created repeatedly. I am not sure if these kinds of considerations are important for a `run-make` test.
Demands a Windows try-job.
try-job: x86_64-mingw
Rollup of 7 pull requests
Successful merges:
- #123813 (Add `REDUNDANT_IMPORTS` lint for new redundant import detection)
- #126697 ([RFC] mbe: consider the `_` in 2024 an expression)
- #127159 (match lowering: Hide `Candidate` from outside the lowering algorithm)
- #128244 (Peel off explicit (or implicit) deref before suggesting clone on move error in borrowck, remove some hacks)
- #128431 (Add myself as VxWorks target maintainer for reference)
- #128438 (Add special-case for [T, 0] in dropck_outlives)
- #128457 (Fix docs for OnceLock::get_mut_or_init)
r? `@ghost`
`@rustbot` modify labels: rollup
Rollup of 6 pull requests
Successful merges:
- #127567 (std: implement the `once_wait` feature)
- #128162 (Cleanup sys module to match house style)
- #128296 (Update target-spec metadata for loongarch64 targets)
- #128443 (Properly mark loop as diverging if it has no breaks)
- #128449 (Temporarily switch `ambiguous_negative_literals` lint to allow)
- #128452 (derive(SmartPointer): require pointee to be maybe sized)
r? `@ghost`
`@rustbot` modify labels: rollup
derive(SmartPointer): require pointee to be maybe sized
cc ``@Darksonn``
So `#[pointee]` has to be `?Sized` in order for deriving `SmartPointer` to be meaningful.
cc ``@compiler-errors`` for suggestions in #127681
