make sure ScalarPair enums have ScalarPair variants; add some layout sanity checks
`@eddyb` suggested that it might be reasonable for `ScalarPair` enums to simply adjust the ABI of their variants accordingly, such that the layout invariant Miri expects actually holds. This PR implements that. I should note though that I don't know much about this layout computation code and what non-Miri consumers expect from it, so tread with caution!
I also added a function to sanity-check that computed layouts are internally consistent. This helped a lot in figuring out the final shape of this PR, though I am also not 100% sure that these sanity checks are the right ones.
Cc `@oli-obk`
Fixes https://github.com/rust-lang/rust/issues/96221
Optimize switch sources representation and usage
* Avoid constructing switch sources unless necessary - switch sources are used by backward analysis with a custom switch int edge effects, but are otherwise unnecessarily computed.
* Use sparse representation of switch sources to avoid quadratic space overhead.
Some subst cleanup
Two separate things here. Both changes are useful for some refactoring I'm doing to add an "EarlyBinder" newtype. (Part of chalkification).
1) Remove `subst_spanned` and just use `subst`. It wasn't used much anyways. In practice, I think we can probably get most of the info just from the actual error message. If not, outputting logs should do the trick. (The specific line probably wouldn't help much anyways).
2) Call `.subst()` before `replace_bound_vars_with_fresh_vars` and `erase_late_bound_regions` in three places that do the opposite. I think there might have been some time in the past that the order here matter for something, but this shouldn't be the case anymore. Conceptually, it makes more sense to the of the *early bound* vars on `fn`s as "outside" the late bound vars.
Remove `#[rustc_deprecated]`
This removes `#[rustc_deprecated]` and introduces diagnostics to help users to the right direction (that being `#[deprecated]`). All uses of `#[rustc_deprecated]` have been converted. CI is expected to fail initially; this requires #95958, which includes converting `stdarch`.
I plan on following up in a short while (maybe a bootstrap cycle?) removing the diagnostics, as they're only intended to be short-term.
Switch sources are used by backward analysis with a custom switch int
edge effects, but are otherwise unnecessarily computed.
Delay the computation until we know that switch sources are indeed
required and avoid the computation otherwise.
make Size and Align debug-printing a bit more compact
In particular in `{:#?}`-mode, these take up a lot of space, so I think this is the better alternative (even though it is a bit longer in `{:?}` mode, I think it is still more readable).
We could make it even smaller by deviating further from what the actual code looks like, e.g. via something like `Size(4 bytes)`. Not sure what people would think about that?
Cc `````@oli-obk`````
Begin fixing all the broken doctests in `compiler/`
Begins to fix#95994.
All of them pass now but 24 of them I've marked with `ignore HELP (<explanation>)` (asking for help) as I'm unsure how to get them to work / if we should leave them as they are.
There are also a few that I marked `ignore` that could maybe be made to work but seem less important.
Each `ignore` has a rough "reason" for ignoring after it parentheses, with
- `(pseudo-rust)` meaning "mostly rust-like but contains foreign syntax"
- `(illustrative)` a somewhat catchall for either a fragment of rust that doesn't stand on its own (like a lone type), or abbreviated rust with ellipses and undeclared types that would get too cluttered if made compile-worthy.
- `(not-rust)` stuff that isn't rust but benefits from the syntax highlighting, like MIR.
- `(internal)` uses `rustc_*` code which would be difficult to make work with the testing setup.
Those reason notes are a bit inconsistently applied and messy though. If that's important I can go through them again and try a more principled approach. When I run `rg '```ignore \(' .` on the repo, there look to be lots of different conventions other people have used for this sort of thing. I could try unifying them all if that would be helpful.
I'm not sure if there was a better existing way to do this but I wrote my own script to help me run all the doctests and wade through the output. If that would be useful to anyone else, I put it here: https://github.com/Elliot-Roberts/rust_doctest_fixing_tool
Allow inline consts to reference generic params
Tracking issue: #76001
The RFC says that inline consts cannot reference to generic parameters (for now), same as array length expressions. And expresses that it's desirable for it to reference in-scope generics, when array length expressions gain that feature as well.
However it is possible to implement this for inline consts before doing this for all anon consts, because inline consts are only used as values and they won't be used in the type system. So we can have:
```rust
fn foo<T>() {
let x = [4i32; std::mem::size_of::<T>()]; // NOT ALLOWED (for now)
let x = const { std::mem::size_of::<T>() }; // ALLOWED with this PR!
let x = [4i32; const { std::mem::size_of::<T>() }]; // NOT ALLOWED (for now)
}
```
This would make inline consts super useful for compile-time checks and assertions:
```rust
fn assert_zst<T>() {
const { assert!(std::mem::size_of::<T>() == 0) };
}
```
This would create an error during monomorphization when `assert_zst` is instantiated with non-ZST `T`s. A error during mono might sound scary, but this is exactly what a "desugared" inline const would do:
```rust
fn assert_zst<T>() {
struct F<T>(T);
impl<T> F<T> {
const V: () = assert!(std::mem::size_of::<T>() == 0);
}
let _ = F::<T>::V;
}
```
It should also be noted that the current inline const implementation can already reference the type params via type inference, so this resolver-level restriction is not any useful either:
```rust
fn foo<T>() -> usize {
let (_, size): (PhantomData<T>, usize) = const {
const fn my_size_of<T>() -> (PhantomData<T>, usize) {
(PhantomData, std::mem::size_of::<T>())
}
my_size_of()
};
size
}
```
```@rustbot``` label: F-inline_const
Add a new Rust attribute to support embedding debugger visualizers
Implemented [this RFC](https://github.com/rust-lang/rfcs/pull/3191) to add support for embedding debugger visualizers into a PDB.
Added a new attribute `#[debugger_visualizer]` and updated the `CrateMetadata` to store debugger visualizers for crate dependencies.
RFC: https://github.com/rust-lang/rfcs/pull/3191
Cleanup `DebuggerVisualizerFile` type and other minor cleanup of queries.
Merge the queries for debugger visualizers into a single query.
Revert move of `resolve_path` to `rustc_builtin_macros`. Update dependencies in Cargo.toml for `rustc_passes`.
Respond to PR comments. Load visualizer files into opaque bytes `Vec<u8>`. Debugger visualizers for dynamically linked crates should not be embedded in the current crate.
Update the unstable book with the new feature. Add the tracking issue for the debugger_visualizer feature.
Respond to PR comments and minor cleanups.
Update `RValue::Discriminant` documentation
`RValue::Discriminant` returns zero for types without discriminant.
This guarantee is already documented for `discriminant_value`
intrinsics which is implemented in terms of `RValue::Discriminant`.
Only crate root def-ids don't have a parent, and in majority of cases the argument of `DefIdTree::parent` cannot be a crate root.
So we now panic by default in `parent` and introduce a new non-panicing function `opt_parent` for cases where the argument can be a crate root.
Same applies to `local_parent`/`opt_local_parent`.
`RValue::Discriminant` returns zero for types without discriminant.
This guarantee is already documented for `discriminant_value`
intrinsics which is implemented in terms of `RValue::Discriminant`.
Reduce duplication of RPO calculation of mir
Computing the RPO of mir is not a low-cost thing, but it is duplicate in many places. In particular the `iterate_to_fixpoint` method which is called multiple times when computing the data flow.
This PR reduces the number of times the RPO is recalculated as much as possible, which should save some compile time.
Implement Valtree to ConstValue conversion
Once we start to use `ValTree`s in the type system we will need to be able to convert them into `ConstValue` instances, which we want to continue to use after MIR construction.
r? `@oli-obk`
cc `@RalfJung`
Fix incremental perf regression unsafety checking
Perf regression introduced in #96294
We will simply avoid emitting the name of the unsafe function in MIR unsafeck, since we're moving to THIR unsafeck anyway.
Correct documentation for `Rvalue::ShallowInitBox`
As a part of the big MIR docs PR, I had added a comment indicating that `Rvalue::ShallowInitBox` is disallowed after drop elaboration, but this is not true (no idea why I thought it was). Codegen has support for it, and trying to enforce this rule in the validator causes compiling core to ICE on the very first `box` statement.
That being said, this `Rvalue` probably *should* be banned after drop elaboration - it doesn't seem like it's still useful for much. However, I do not have time right now to actually go investigate how difficult a change that is to make, so in the meantime fixing the docs to reflect the current situation seems like the right step.
r? rust-lang/mir-opt
Fix codegen bug in "ptx-kernel" abi related to arg passing
I found a codegen bug in the nvptx abi related to that args are passed as ptrs ([see comment](https://github.com/rust-lang/rust/issues/38788#issuecomment-1048999928)), this is not as specified in the [ptx-interoperability doc](https://docs.nvidia.com/cuda/ptx-writers-guide-to-interoperability/) or how C/C++ does it. It will also almost always fail in practice since device/host uses different memory spaces for most hardware.
This PR fixes the bug and add tests for passing structs to ptx kernels.
I observed that all nvptx assembly tests had been marked as [ignore a long time ago](https://github.com/rust-lang/rust/pull/59752#issuecomment-501713428). I'm not sure if the new one should be marked as ignore, it passed on my computer but it might fail if ptx-linker is missing on the server? I guess this is outside scope for this PR and should be looked at in a different issue/PR.
I only fixed the nvptx64-nvidia-cuda target and not the potential code paths for the non-existing 32bit target. Even though 32bit nvptx is not a supported target there are still some code under the hood supporting codegen for 32 bit ptx. I was advised to create an MCP to find out if this code should be removed or updated.
Perhaps ``@RDambrosio016`` would have interest in taking a quick look at this.
