| Age | Commit message (Collapse) | Author | Lines |
|---|
|
It was inconsistently done (sometimes even within a single function) and
most of the rest of the compiler uses fatal errors instead, which need
to be caught using catch_with_exit_code anyway. Using fatal errors
instead of ErrorGuaranteed everywhere in the driver simplifies things a
bit.
|
|
And pass this to the individual emitters when necessary.
|
|
|
|
|
|
Sync cg clif 2024 12 06
The main highlights this time are a Cranelift update disabling the clif ir verifier by default for better performance.
r? `@ghost`
`@rustbot` label +A-codegen +A-cranelift +T-compiler
|
|
Cranelift being disabled
|
|
sync_cg_clif-2024-12-06
|
|
|
|
Rollup of 5 pull requests
Successful merges:
- #130777 (rust_for_linux: -Zreg-struct-return commandline flag for X86 (#116973))
- #133211 (Extend Miri to correctly pass mutable pointers through FFI)
- #133790 (Improve documentation for Vec::extend_from_within)
- #133930 (rustbook: update to use new mdbook-trpl package from The Book)
- #133931 (Only allow PassMode::Direct for aggregates on wasm when using the C ABI)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
r=nnethercote
Only allow PassMode::Direct for aggregates on wasm when using the C ABI
For the Rust ABI we don't have any ABI compat reasons to allow PassMode::Direct for aggregates.
|
|
rustbook: update to use new mdbook-trpl package from The Book
Updates to the latest merge from `rust-lang/book` and simplifies the dependency chain there. There are now three preprocessors, but only one package, so everything is a lot nicer to deal with from the consuming POV (i.e. here).
|
|
Improve documentation for Vec::extend_from_within
This closes #104762.
It rephrases some of the explanations, and greatly improves the clarity of the example.
Based on this PR and its discussions:
https://github.com/rust-lang/rust/pull/105030/files#r1059808792
|
|
Extend Miri to correctly pass mutable pointers through FFI
Based off of https://github.com/rust-lang/rust/pull/129684, this PR further extends Miri to execute native calls that make use of pointers to *mutable* memory.
We adapt Miri's bookkeeping of internal state upon any FFI call that gives external code permission to mutate memory.
Native code may now possibly write and therefore initialize and change the pointer provenance of bytes it has access to: Such memory is assumed to be *initialized* afterwards and bytes are given *arbitrary (wildcard) provenance*. This enables programs that correctly use mutating FFI calls to run Miri without errors, at the cost of possibly missing Undefined Behaviour caused by incorrect usage of mutating FFI.
> <details>
>
> <summary> Simple example </summary>
>
> ```rust
> extern "C" {
> fn init_int(ptr: *mut i32);
> }
>
> fn main() {
> let mut x = std::mem::MaybeUninit::<i32>::uninit();
> let x = unsafe {
> init_int(x.as_mut_ptr());
> x.assume_init()
> };
>
> println!("C initialized my memory to: {x}");
> }
> ```
> ```c
> void init_int(int *ptr) {
> *ptr = 42;
> }
> ```
> should now show `C initialized my memory to: 42`.
>
> </details>
r? ``@RalfJung``
|
|
rust_for_linux: -Zreg-struct-return commandline flag for X86 (#116973)
Command line flag `-Zreg-struct-return` for X86 (32-bit) for rust-for-linux.
This flag enables the same behavior as the `abi_return_struct_as_int` target spec key.
- Tracking issue: https://github.com/rust-lang/rust/issues/116973
|
|
For the Rust ABI we don't have any ABI compat reasons to allow
PassMode::Direct for aggregates.
|
|
compiler-errors:structurally-resolve-adjust-for-branch, r=lcnr
Structurally resolve in `adjust_for_branches`
r? lcnr
|
|
Rollup of 8 pull requests
Successful merges:
- #132155 (Always display first line of impl blocks even when collapsed)
- #133256 (CI: use free runners for i686-gnu jobs)
- #133607 (implement checks for tail calls)
- #133821 (Replace black with ruff in `tidy`)
- #133827 (CI: rfl: move job forward to Linux v6.13-rc1)
- #133910 (Normalize target-cpus.rs stdout test for LLVM changes)
- #133921 (Adapt codegen tests for NUW inference)
- #133936 (Avoid fetching the anon const hir node that is already available)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Avoid fetching the anon const hir node that is already available
|
|
Adapt codegen tests for NUW inference
These were broken by https://github.com/llvm/llvm-project/commit/462cb3cd6cecd0511ecaf0e3ebcaba455ece587d
Let me know if you think we should have a FIXME / tracking issue to update the tests after the LLVM 20 upgrade to make these required.
`@rustbot` label: +llvm-main
r? `@nikic`
|
|
Normalize target-cpus.rs stdout test for LLVM changes
LLVM has recently added support for the `lime1` CPU in https://github.com/llvm/llvm-project/commit/35cce408eef1a253df12c0023c993d78b180b1f3, so the `target-cpus.rs` test currently produces different output depending on the LLVM version.
This CL adds a normalization directive, to remove the new CPU from the output list.
Alternatives fixes I can think of:
* Add two revisions of the test (one per LLVM version)
* Ignore the test on one of the LLVM versions
* I dislike this, because it's possible that the test won't get updated for the next LLVM version.
I don't think the exact list of target CPUs is relevant for this test, so it shouldn't be too bad if the normalization sticks around longer than necessary.
`@rustbot` label: +llvm-main
|
|
CI: rfl: move job forward to Linux v6.13-rc1
Linux v6.13-rc1 contains commit 28e848386b92 ("rust: block: fix formatting of `kernel::block::mq::request` module"), which in turn contains commit c95bbb59a9b2 ("rust: enable arbitrary_self_types and remove `Receiver`"), which is why we had a hash rather than a tag.
r? ```@Kobzol``` ```@lqd```
try-job: x86_64-rust-for-linux
```@rustbot``` label A-rust-for-linux
```@bors``` try
|
|
Replace black with ruff in `tidy`
`ruff` can both lint and format Python code (in fact, it should be a mostly drop-in replacement for `black` in terms of formatting), so it's not needed to use `black` anymore. This PR removes `black` and replaces it with `ruff`, to get rid of one Python dependency, and also to make Python formatting faster (although that's a small thing).
If we decide to merge this, we'll need to "reformat the world" - `ruff` is not perfectly compatible with `black`, and it also looks like `black` was actually ignoring some files before. I tried it locally (`./x test tidy --extra-checks=py:fmt --bless`) and it also reformatted some code in subtrees (e.g. `clippy` or `rustc_codegen_gcc`) - I'm not sure how to handle that.
|
|
implement checks for tail calls
Quoting the [RFC draft](https://github.com/phi-go/rfcs/blob/guaranteed-tco/text/0000-explicit-tail-calls.md):
> The argument to become is a function (or method) call, that exactly matches the function signature and calling convention of the callee. The intent is to ensure a matching ABI. Note that lifetimes may differ as long as they pass borrow checking, see [below](https://github.com/phi-go/rfcs/blob/guaranteed-tco/text/0000-explicit-tail-calls.md#return-type-coercion) for specifics on the return type.
> Tail calling closures and tail calling from closures is not allowed. This is due to the high implementation effort, see below, this restriction can be lifted by a future RFC.
> Invocations of operators were considered as valid targets but were rejected on grounds of being too error-prone. In any case, these can still be called as methods.
> Tail calling [variadic functions](https://doc.rust-lang.org/beta/unstable-book/language-features/c-variadic.html) and tail calling from variadic functions is not allowed. As support for variadic function is stabilized on a per target level, support for tail-calls regarding variadic functions would need to follow a similar approach. To avoid this complexity and to minimize implementation effort for backends, this interaction is currently not allowed but support can be added with a future RFC.
-----
The checks are implemented as a query, similarly to `check_unsafety`.
The code is cherry-picked straight out of #112657 which was written more than a year ago, so I expect we might need to change some things ^^"
|
|
CI: use free runners for i686-gnu jobs
try-job: i686-gnu-1
try-job: i686-gnu-2
try-job: i686-gnu-nopt-1
try-job: i686-gnu-nopt-2
|
|
Always display first line of impl blocks even when collapsed
Fixes https://github.com/rust-lang/rust/issues/130612.
It the line is too long, only the beginning will be visible:
Otherwise, it looks like this:
Can be tested [here](https://rustdoc.crud.net/imperio/impl-block-doc/foo/struct.ImplDoc.html).
r? `@notriddle`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Co-authored-by: Ralf Jung <post@ralfj.de>
|
|
|
|
Rollup of 6 pull requests
Successful merges:
- #127565 (Teach rustc about the Xtensa VaListImpl)
- #133844 (clarify simd_relaxed_fma non-determinism)
- #133867 (Fix "std" support status of some tier 3 targets)
- #133882 (Improve comments for the default backtrace printer)
- #133888 (Improve bootstrap job objects)
- #133898 (skip `setup::Hook` on non-git sources)
r? `@ghost`
`@rustbot` modify labels: rollup
|
|
Fix missing newlines that rustfmt removed.
fix trailing whitespace
Fix duplicate word.
Reformat panic reasons into a list
remove trailing whitespace 2 electric boogaloo
Change verbe tense.
Integrate suggestions
|
|
Fix the ABI for libcalls
|
|
|
|
It is ill defined and for this specific case it may become impossible
to call using Cranelift in the future.
|
|
|
|
|
|
This reverts commit 65c5c7f8cf366868b47483c94e41440143c73eae.
It unfortunately regresses the size of the target dir by a non-trivial
amount. It could be re-enabled again once each subsection doesn't get a
unique name anymore.
|
|
|
|
With lld the perf difference is within noise while producing slightly
(or in exceptional cases significantly) smaller binaries.
|
|
|
|
|
|
|
|
|
|
skip `setup::Hook` on non-git sources
Running `setup::Hook` (with `x setup`) leads tarball sources to panic and this PR resolves that problem by skipping `Hook` step on non-git sources.
|
