safe transmute: support `Single` enums
Previously, the implementation of `Tree::from_enum` incorrectly treated enums with `Variants::Single` and `Variants::Multiple` identically. This is incorrect for `Variants::Single` enums, which delegate their layout to that of a variant with a particular index (or no variant at all if the enum is empty).
This flaw manifested first as an ICE. `Tree::from_enum` attempted to compute the tag of variants other than the one at `Variants::Single`'s `index`, and fell afoul of a sanity-checking assertion in `compiler/rustc_const_eval/src/interpret/discriminant.rs`. This assertion is non-load-bearing, and can be removed; the routine its in is well-behaved even without it.
With the assertion removed, the proximate issue becomes apparent: calling `Tree::from_variant` on a variant that does not exist is ill-defined. A sanity check the given variant has `FieldShapes::Arbitrary` fails, and the analysis is (correctly) aborted with `Err::NotYetSupported`.
This commit corrects this chain of failures by ensuring that `Tree::from_variant` is not called on variants that are, as far as layout is concerned, nonexistent. Specifically, the implementation of `Tree::from_enum` is now partitioned into three cases:
1. enums that are uninhabited
2. enums for which all but one variant is uninhabited
3. enums with multiple inhabited variants
`Tree::from_variant` is now only invoked in the third case. In the first case, `Tree::uninhabited()` is produced. In the second case, the layout is delegated to `Variants::Single`'s index.
Fixes#125811
Harmonize using root or leaf obligation in trait error reporting
When #121826 changed the error reporting to use root obligation and not the leafmost obligation, it didn't actually make sure that all the other diagnostics helper functions used the right obligation.
Specifically, when reporting similar impl candidates we are looking for impls of the root obligation, but trying to match them against the trait ref of the leaf obligation.
This does a few other miscellaneous changes. There's a lot more clean-up that could be done here, but working with this code is really grief-inducing due to how messy it has become over the years. Someone really needs to show it love. 😓
r? ``@estebank``
Fixes#126129
Walk into alias-eq nested goals even if normalization fails
Somewhat broken due to the fact that we don't handle aliases well, nor do we handle ambiguities well. Still want to put up this incremental piece, since it improves type errors for projections whose trait refs are not satisfied.
r? lcnr
Migrate `run-make/pgo-branch-weights` 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).
This is a scary one and I expect things to break. Set as draft, because this isn't ready.
- [x] There is this comment here, which suggests the test is excluded from the testing process due to a platform specific issue? I can't see anything here that would cause this test to not run...
> // FIXME(mati865): MinGW GCC miscompiles compiler-rt profiling library but with Clang it works
// properly. Since we only have GCC on the CI ignore the test for now."
EDIT: This is specific to Windows-gnu.
- [x] The Makefile has this line:
```
ifneq (,$(findstring x86,$(TARGET)))
COMMON_FLAGS=-Clink-args=-fuse-ld=gold
```
I honestly can't tell whether this is checking if the target IS x86, or IS NOT. EDIT: It's checking if it IS x86.
- [x] I don't know why the Makefile was trying to pass an argument directly in the Makefile instead of setting that "aaaaaaaaaaaa2bbbbbbbbbbbb2bbbbbbbbbbbbbbbbcc" input as a variable in the Rust program directly. I changed that, let me know if that was wrong.
- [x] Trying to rewrite `cat "$(TMPDIR)/interesting.ll" | "$(LLVM_FILECHECK)" filecheck-patterns.txt` resulted in some butchery. For starters, in `tools.mk`, LLVM_FILECHECK corrects its own backslashes on Windows distributions, but there is no further mention of it, so I assume this is a preset environment variable... but is it really? Then, the command itself uses a Standard Input and a passed input file as an argument simultaneously, according to the [documentation](https://llvm.org/docs/CommandGuide/FileCheck.html#synopsis).
try-job: aarch64-gnu
Previously, the implementation of `Tree::from_enum` incorrectly
treated enums with `Variants::Single` and `Variants::Multiple`
identically. This is incorrect for `Variants::Single` enums,
which delegate their layout to that of a variant with a particular
index (or no variant at all if the enum is empty).
This flaw manifested first as an ICE. `Tree::from_enum` attempted
to compute the tag of variants other than the one at
`Variants::Single`'s `index`, and fell afoul of a sanity-checking
assertion in `compiler/rustc_const_eval/src/interpret/discriminant.rs`.
This assertion is non-load-bearing, and can be removed; the routine
its in is well-behaved even without it.
With the assertion removed, the proximate issue becomes apparent:
calling `Tree::from_variant` on a variant that does not exist is
ill-defined. A sanity check the given variant has
`FieldShapes::Arbitrary` fails, and the analysis is (correctly)
aborted with `Err::NotYetSupported`.
This commit corrects this chain of failures by ensuring that
`Tree::from_variant` is not called on variants that are, as far as
layout is concerned, nonexistent. Specifically, the implementation
of `Tree::from_enum` is now partitioned into three cases:
1. enums that are uninhabited
2. enums for which all but one variant is uninhabited
3. enums with multiple inhabited variants
`Tree::from_variant` is now only invoked in the third case. In the
first case, `Tree::uninhabited()` is produced. In the second case,
the layout is delegated to `Variants::Single`'s index.
Fixes#125811
Detect pub structs never constructed even though they impl pub trait with assoc constants
Extend dead code analysis to impl items of pub assoc constants.
<!--
If this PR is related to an unstable feature or an otherwise tracked effort,
please link to the relevant tracking issue here. If you don't know of a related
tracking issue or there are none, feel free to ignore this.
This PR will get automatically assigned to a reviewer. In case you would like
a specific user to review your work, you can assign it to them by using
r? <reviewer name>
-->
Add `target_env = "p1"` to the `wasm32-wasip1` target
This commit sets the `target_env` key for the
`wasm32-wasi{,p1,p1-threads}` targets to the string `"p1"`. This mirrors how the `wasm32-wasip2` target has `target_env = "p2"`. The intention of this is to more easily detect each target in downstream crates to enable adding custom code per-target.
cc #125803
<!--
If this PR is related to an unstable feature or an otherwise tracked effort,
please link to the relevant tracking issue here. If you don't know of a related
tracking issue or there are none, feel free to ignore this.
This PR will get automatically assigned to a reviewer. In case you would like
a specific user to review your work, you can assign it to them by using
r? <reviewer name>
-->
Rollup of 9 pull requests
Successful merges:
- #126039 (Promote `arm64ec-pc-windows-msvc` to tier 2)
- #126075 (Remove `DebugWithInfcx` machinery)
- #126228 (Provide correct parent for nested anon const)
- #126232 (interpret: dyn trait metadata check: equate traits in a proper way)
- #126242 (Simplify provider api to improve llvm ir)
- #126294 (coverage: Replace the old span refiner with a single function)
- #126295 (No uninitalized report in a pre-returned match arm)
- #126312 (Update `rustc-perf` submodule)
- #126322 (Follow up to splitting core's PanicInfo and std's PanicInfo)
r? `@ghost`
`@rustbot` modify labels: rollup
coverage: Replace the old span refiner with a single function
As more and more of the span refiner's functionality has been pulled out into separate early passes, it has finally reached the point where we can remove the rest of the old `SpansRefiner` code, and replace it with a single modestly-sized function.
~~There should be no change to the resulting coverage mappings, as demonstrated by the lack of changes to test output.~~
There is *almost* no change to the resulting coverage mappings. There are some minor changes to `loop` that on inspection appear to be neutral in terms of accuracy, with the old behaviour being a slightly-horrifying implementation detail of the old code, so I think they're acceptable.
Previous work in this direction includes:
- #125921
- #121019
- #119208
Provide correct parent for nested anon const
Fixes#126147
99% of this PR is just comments explaining what the issue is.
`tcx.parent(` and `hir().get_parent_item(` give different results as the hir owner for all the hir of anon consts is the enclosing function. I didn't attempt to change that as being a hir owner requires a `DefId` and long term we want to stop creating anon consts' `DefId`s before hir ty lowering.
So i just opted to change `generics_of` to use `tcx.parent` to get the parent for `AnonConst`'s. I'm not entirely sure about this being what we want, it does seem weird that we have two ways of getting the parent of an `AnonConst` and they both give different results.
Alternatively we could just go ahead and make `const_evaluatable_unchecked` a hard error and stop providing generics to repeat exprs. Then this isn't an issue. (The FCW has been around for almost 4 years now)
r? ````@compiler-errors````
Remove `DebugWithInfcx` machinery
This PR removes `DebugWithInfcx` after having a lot of second thoughts about it due to recent type system uplifting work. We could add it back later if we want, but I don't think the amount of boilerplate in the complier and the existence of (kindof) hacks like `NoInfcx` currently justify the existence of `DebugWithInfcx`, especially since it's not even being used anywhere in the compiler currently.
The motivation for `DebugWithInfcx` is that we want to be able to print infcx-aware information, such as universe information[^1] (though if there are other usages that I'm overlooking, please let me know). I think there are probably more tailored solutions that can specifically be employed in places where this infcx-aware printing is necessary. For example, one way of achieving this is by implementing a custom `FmtPrinter` which overloads `ty_infer_name` (perhaps also extending it to have overrideable stubs for printing placeholders too) to print the `?u.i` name for an infer var. This will necessitate uplifting `Print` from `rustc_middle::ty::print`, but this seems a bit more extensible and reusable than `DebugWithInfcx`.
One of the problems w/ `DebugWithInfcx` is its opt-in-ness. Even if a compiler dev adds a new `debug!(ty)` in a context where there is an `infcx` we can access, they have to *opt-in* to using `DebugWithInfcx` with something like `debug!(infcx.with(ty))`. This feels to me like it risks a lot of boilerplate, and very easy to just forget adding it at all, especially in cases like `#[instrument]`.
A second problem is the `NoInfcx` type itself. It's necessary to have this dummy infcx implementation since we often want to print types outside of the scope of a valid `Infcx`. Right now, `NoInfcx` is only *partially* a valid implementation of `InferCtxtLike`, except for the methods that we specifically need for `DebugWithInfcx`. As I work on uplifting the trait solver, I actually want to add a lot more methods to `InferCtxtLike` and having to add `unreachable!("this should never be called")` stubs for uplifted methods like `next_ty_var` is quite annoying.
In reality, I actually only *really* care about the second problem -- we could, perhaps, instead just try to get rid of `NoInfcx` and just just duplicate `Debug` and `DebugWithInfcx` for most types. If we're okay with duplicating all these implementations (though most of them would just be trivial `#[derive(Debug, DebugWithInfcx)]`), I'd be okay with that too 🤔
r? `@BoxyUwU` `@lcnr` would like to know your thoughts -- happy to discuss this further, mainly trying to bring this problem up
[^1]: Which in my experience is only really necessary when we're debugging things like generalizer bugs.
Add no_std Xtensa targets support
Adds no_std Xtensa targets. This enables using Rust on ESP32, ESP32-S2 and ESP32-S3 chips.
Tier 3 policy:
> A tier 3 target must have a designated developer or developers (the "target maintainers") on record to be CCed when issues arise regarding the target. (The mechanism to track and CC such developers may evolve over time.)
`@MabezDev` and I (`@SergioGasquez)` will maintain the targets.
> Targets must use naming consistent with any existing targets; for instance, a target for the same CPU or OS as an existing Rust target should use the same name for that CPU or OS. Targets should normally use the same names and naming conventions as used elsewhere in the broader ecosystem beyond Rust (such as in other toolchains), unless they have a very good reason to diverge. Changing the name of a target can be highly disruptive, especially once the target reaches a higher tier, so getting the name right is important even for a tier 3 target.
The target triple is consistent with other targets.
> Target names should not introduce undue confusion or ambiguity unless absolutely necessary to maintain ecosystem compatibility. For example, if the name of the target makes people extremely likely to form incorrect beliefs about what it targets, the name should be changed or augmented to disambiguate it.
> If possible, use only letters, numbers, dashes and underscores for the name. Periods (.) are known to cause issues in Cargo.
We follow the same naming convention as other targets.
> Tier 3 targets may have unusual requirements to build or use, but must not create legal issues or impose onerous legal terms for the Rust project or for Rust developers or users.
The target does not introduce any legal issues.
> The target must not introduce license incompatibilities.
There are no license incompatibilities
> Anything added to the Rust repository must be under the standard Rust license (MIT OR Apache-2.0).
Everything added is under that licenses
> The target must not cause the Rust tools or libraries built for any other host (even when supporting cross-compilation to the target) to depend on any new dependency less permissive than the Rust licensing policy. This applies whether the dependency is a Rust crate that would require adding new license exceptions (as specified by the tidy tool in the rust-lang/rust repository), or whether the dependency is a native library or binary. In other words, the introduction of the target must not cause a user installing or running a version of Rust or the Rust tools to be subject to any new license requirements.
Requirements are not changed for any other target.
> Compiling, linking, and emitting functional binaries, libraries, or other code for the target (whether hosted on the target itself or cross-compiling from another target) must not depend on proprietary (non-FOSS) libraries. Host tools built for the target itself may depend on the ordinary runtime libraries supplied by the platform and commonly used by other applications built for the target, but those libraries must not be required for code generation for the target; cross-compilation to the target must not require such libraries at all. For instance, rustc built for the target may depend on a common proprietary C runtime library or console output library, but must not depend on a proprietary code generation library or code optimization library. Rust's license permits such combinations, but the Rust project has no interest in maintaining such combinations within the scope of Rust itself, even at tier 3.
The linker used by the targets is the GCC linker from the GCC toolchain cross-compiled for Xtensa. GNU GPL.
> "onerous" here is an intentionally subjective term. At a minimum, "onerous" legal/licensing terms include but are not limited to: non-disclosure requirements, non-compete requirements, contributor license agreements (CLAs) or equivalent, "non-commercial"/"research-only"/etc terms, requirements conditional on the employer or employment of any particular Rust developers, revocable terms, any requirements that create liability for the Rust project or its developers or users, or any requirements that adversely affect the livelihood or prospects of the Rust project or its developers or users.
No such terms exist for this target
> Neither this policy nor any decisions made regarding targets shall create any binding agreement or estoppel by any party. If any member of an approving Rust team serves as one of the maintainers of a target, or has any legal or employment requirement (explicit or implicit) that might affect their decisions regarding a target, they must recuse themselves from any approval decisions regarding the target's tier status, though they may otherwise participate in discussions.
> This requirement does not prevent part or all of this policy from being cited in an explicit contract or work agreement (e.g. to implement or maintain support for a target). This requirement exists to ensure that a developer or team responsible for reviewing and approving a target does not face any legal threats or obligations that would prevent them from freely exercising their judgment in such approval, even if such judgment involves subjective matters or goes beyond the letter of these requirements.
Understood
> Tier 3 targets should attempt to implement as much of the standard libraries as possible and appropriate (core for most targets, alloc for targets that can support dynamic memory allocation, std for targets with an operating system or equivalent layer of system-provided functionality), but may leave some code unimplemented (either unavailable or stubbed out as appropriate), whether because the target makes it impossible to implement or challenging to implement. The authors of pull requests are not obligated to avoid calling any portions of the standard library on the basis of a tier 3 target not implementing those portions.
The target already implements core.
> The target must provide documentation for the Rust community explaining how to build for the target, using cross-compilation if possible. If the target supports running binaries, or running tests (even if they do not pass), the documentation must explain how to run such binaries or tests for the target, using emulation if possible or dedicated hardware if necessary.
Here is how to build for the target https://docs.esp-rs.org/book/installation/riscv-and-xtensa.html and it also covers how to run binaries on the target.
> Tier 3 targets must not impose burden on the authors of pull requests, or other developers in the community, to maintain the target. In particular, do not post comments (automated or manual) on a PR that derail or suggest a block on the PR based on a tier 3 target. Do not send automated messages or notifications (via any medium, including via `@)` to a PR author or others involved with a PR regarding a tier 3 target, unless they have opted into such messages.
> Backlinks such as those generated by the issue/PR tracker when linking to an issue or PR are not considered a violation of this policy, within reason. However, such messages (even on a separate repository) must not generate notifications to anyone involved with a PR who has not requested such notifications.
Understood
> Patches adding or updating tier 3 targets must not break any existing tier 2 or tier 1 target, and must not knowingly break another tier 3 target without approval of either the compiler team or the maintainers of the other tier 3 target.
> In particular, this may come up when working on closely related targets, such as variations of the same architecture with different features. Avoid introducing unconditional uses of features that another variation of the target may not have; use conditional compilation or runtime detection, as appropriate, to let each target run code supported by that target.
No other targets should be affected
> Tier 3 targets must be able to produce assembly using at least one of rustc's supported backends from any host target.
It can produce assembly, but it requires a custom LLVM with Xtensa support (https://github.com/espressif/llvm-project/). The patches are trying to be upstreamed (https://github.com/espressif/llvm-project/issues/4)
As more and more of the span refiner's functionality has been pulled out into
separate early passes, it has finally reached the point where we can remove the
rest of the old `SpansRefiner` code, and replace it with a single
modestly-sized function.
Rollup of 16 pull requests
Successful merges:
- #123374 (DOC: Add FFI example for slice::from_raw_parts())
- #124514 (Recommend to never display zero disambiguators when demangling v0 symbols)
- #125978 (Cleanup: HIR ty lowering: Consolidate the places that do assoc item probing & access checking)
- #125980 (Nvptx remove direct passmode)
- #126187 (For E0277 suggest adding `Result` return type for function when using QuestionMark `?` in the body.)
- #126210 (docs(core): make more const_ptr doctests assert instead of printing)
- #126249 (Simplify `[T; N]::try_map` signature)
- #126256 (Add {{target}} substitution to compiletest)
- #126263 (Make issue-122805.rs big endian compatible)
- #126281 (set_env: State the conclusion upfront)
- #126286 (Make `storage-live.rs` robust against rustc internal changes.)
- #126287 (Update a cranelift patch file for formatting changes.)
- #126301 (Use `tidy` to sort crate attributes for all compiler crates.)
- #126305 (Make PathBuf less Ok with adding UTF-16 then `into_string`)
- #126310 (Migrate run make prefer rlib)
- #126314 (fix RELEASES: we do not support upcasting to auto traits)
r? `@ghost`
`@rustbot` modify labels: rollup
Make `storage-live.rs` robust against rustc internal changes.
Currently it can be made to fail by rearranging code within `compiler/rustc_mir_transform/src/lint.rs`.
This is a precursor to #125443.
r? ```@lqd```
Make issue-122805.rs big endian compatible
Instead of not generating the function at all on big endian (which makes the CHECK lines fail), instead use to_le() on big endian, so that we essentially perform a bswap for both endiannesses.
For E0277 suggest adding `Result` return type for function when using QuestionMark `?` in the body.
Adding suggestions for following function in E0277.
```rust
fn main() {
let mut _file = File::create("foo.txt")?;
}
```
to
```rust
fn main() -> Result<(), Box<dyn std::error::Error>> {
let mut _file = File::create("foo.txt")?;
return Ok(());
}
```
According to the issue #125997, only the code examples in the issue are targeted, but the issue covers a wider range of situations.
<!--
If this PR is related to an unstable feature or an otherwise tracked effort,
please link to the relevant tracking issue here. If you don't know of a related
tracking issue or there are none, feel free to ignore this.
This PR will get automatically assigned to a reviewer. In case you would like
a specific user to review your work, you can assign it to them by using
r? <reviewer name>
-->
Nvptx remove direct passmode
This PR does what should have been done in #117671. That is fully avoid using the `PassMode::Direct` for `extern "C" fn` for `nvptx64-nvidia-cuda` and enable the compatibility test. `@RalfJung` [pointed me in the right direction](https://github.com/rust-lang/rust/issues/117480#issuecomment-2137712501) for solving this issue.
There are still some ABI bugs after this PR is merged. These ABI tests are created based on what is actually correct, and since they continue passing with even more of them enabled things are improving. I don't have the time to tackle all the remaining issues right now, but I think getting these improvements merged is very valuable in themselves and plan to tackle more of them long term.
This also doesn't remove the use of `PassMode::Direct` for `extern "ptx-kernel" fn`. This was also not trivial to make work. And since the ABI is hidden behind an unstable feature it's less urgent.
I don't know if it's correct to request `@RalfJung` as a reviewer (due to team structures), but he helped me a lot to figure out this stuff. If that's not appropriate then `@davidtwco` would be a good candidate since he know about this topic from #117671
r? `@RalfJung`
Cleanup: HIR ty lowering: Consolidate the places that do assoc item probing & access checking
Use `probe_assoc_item` (for hygienically probing an assoc item and checking if it's accessible wrt. visibility and stability) for assoc item constraints, too, not just for assoc type paths and make the privacy error translatable.
Edition 2024: Make `!` fall back to `!`
This PR changes never type fallback to be `!` (the never type itself) in the next, 2024, edition.
This makes the never type's behavior more intuitive (in 2024 edition) and is the first step of the path to stabilize it.
r? `@compiler-errors`
