Turn ProjectionElem::Subtype into CastKind::Subtype
I noticed that drop elaboration can't, in general, handle `ProjectionElem::SubType`. It creates a disjoint move path that overlaps with other move paths. (`Subslice` does too, and I'm working on a different PR to make that special case less fragile.) If its skipped and treated as the same move path as its parent then `MovePath.place` has multiple possible projections. (It would probably make sense to remove all `Subtype` projections for the canonical place but it doesn't make sense to have this special case for a problem that doesn't actually occur in real MIR.)
The only reason this doesn't break is that `Subtype` is always the sole projection of the local its applied to. For the same reason, it works fine as a `CastKind` so I figured that makes more sense than documenting and validating this hidden invariant.
cc rust-lang/rust#112651, rust-lang/rust#133258
r? Icnr (bc you've been the main person dealing with `Subtype` it looks like)
Fix some crash-test directives
- 120175 fails to crash for non-ELF targets; presumably this wasn't noticed because the CI jobs don't enable rustc assertions for non-ELF hosts.
- 34127, 125722, and 131292 have `only-x86_64`, which is overly specific.
- Unnecessary x86 directives cause friction for contributors using aarch64, especially now that many PR CI jobs also use aarch64.
r? ghost
Simplify polonius location-sensitive analysis
This PR reworks the location-sensitive analysis into what we think is a worthwhile subset of the datalog analysis. A sort of polonius alpha analysis that handles NLL problem case 3 and more, but is still using the faster "reachability as an approximation of liveness", as well as the same loans-in-scope computation as NLLs -- and thus doesn't handle full flow-sensitivity like the datalog implementation.
In the last few months, we've identified this subset as being actionable:
- we believe we can make a stabilizable version of this analysis
- it is an improvement over the status quo
- it can also be modeled in a-mir-formality, or some other formalism, for assurances about soundness, and I believe ````````@nikomatsakis```````` is interested in looking into this during H2.
- and we've identified the areas of work we wish to explore later to gradually expand the supported cases: the differences between reachability and liveness, support of kills, and considerations of time-traveling, for example.
The approach in this PR is to try less to have the graph only represent live paths, by checking whether we reach a live region during traversal and recording the loan as live there, instead of equating traversal with liveness like today because it has subtleties with the typeck edges in statements (that could forward loans to the successor point without ensuring their liveness). We can then also simplify these typeck stmt edges. And we also can simplify traversal by removing looking at kills, because that's enough to handle a bunch of NLL problem 3 cases -- and we can gradually support them more and more in traversal in the future, to reduce the approximation of liveness.
There's still some in-progress pieces of work w/r/t opaque types that I'm expecting [lcnr's opaque types rework](https://github.com/rust-lang/rust/pull/139587), and [amanda's SCCs rework](https://github.com/rust-lang/rust/pull/130227) to handle. That didn't seem to show up in tests until I rebased today (and shows lack of test coverage once again) when https://github.com/rust-lang/rust/pull/142255 introduced a couple of test failures with the new captures rules from edition 2024. It's not unexpected since we know more work is needed with member constraints (and we're not even using SCCs in this prototype yet)
I'll look into these anyways, both for future work, and checking how these other 2 PRs would change things.
---
I'm not sure the following means a lot until we have some formalism in-place, but:
- I've changed the polonius compare-mode to use this analysis: the tests pass with it, except 2 cases with minor diagnostics differences, and the 2 edition 2024 opaque types one I mentioned above and need to investigate
- things that are expected to work still do work: it bootstraps, can run our rustc-perf benchmarks (and the results are not even that bad), and a crater run didn't find any regressions (forgetting that crater currently fails to test around a quarter of all crates 👼)
- I've added tests with improvements, like the NLL problem case 3 and others, as well as some that behave the same as NLLs today and are thus worse than the datalog implementation
r? ````````@jackh726````````
(no rush I know you're deep in phd work and "implmentating" the new trait solver for r-a :p <3)
This also fixesrust-lang/rust#135646, a diagnostics ICE from the previous implementation.
uniquify root goals during HIR typeck
We need to rely on region identity to deal with hangs such as https://github.com/rust-lang/trait-system-refactor-initiative/issues/210 and to keep the current behavior of `fn try_merge_responses`.
This is a problem as borrowck starts by replacing each *occurrence* of a region with a unique inference variable. This frequently splits a single region during HIR typeck into multiple distinct regions. As we assume goals to always succeed during borrowck, relying on two occurances of a region being identical during HIR typeck causes ICE. See the now fixed examples in https://github.com/rust-lang/trait-system-refactor-initiative/issues/27 and rust-lang/rust#139409.
We've previously tried to avoid this issue by always *uniquifying* regions when canonicalizing goals. This prevents caching subtrees during canonicalization which resulted in hangs for very large types. People rely on such types in practice, which caused us to revert our attempt to reinstate `#[type_length_limit]` in https://github.com/rust-lang/rust/pull/127670. The complete list of changes here:
- rust-lang/rust#107981
- rust-lang/rust#110180
- rust-lang/rust#114117
- rust-lang/rust#130821
After more consideration, all occurrences of such large types need to happen outside of typeck/borrowck. We know this as we already walk over all types in the MIR body when replacing their regions with nll vars.
This PR therefore enables us to rely on region identity inside of the trait solver by exclusively **uniquifying root goals during HIR typeck**. These are the only goals we assume to hold during borrowck. This is insufficient as type inference variables may "hide" regions we later uniquify. Because of this, we now stash proven goals which depend on inference variables in HIR typeck and reprove them after writeback. This closes https://github.com/rust-lang/trait-system-refactor-initiative/issues/127.
This was originally part of rust-lang/rust#144258 but I've moved it into a separate PR. While I believe we need to rely on region identity to fix the performance issues in some way, I don't know whether rust-lang/rust#144258 is the best approach to actually do so. Regardless of how we deal with the hangs however, this change is necessary and desirable regardless.
r? `@compiler-errors` or `@BoxyUwU`
MIR-build: No longer emit assumes in enum-as casting
This just uses the `valid_range` from the backend, so it's duplicating the range metadata that now we include on parameters and loads, and thus no longer seems to be useful -- notably there's no codegen test failures from removing it.
(Because it's using data from the same source as the backend annotations, it doesn't do anything to mitigate things like rust-lang/rust#144388 where the range in the layout is more permissive than the actual possible discriminants. A variant of this that actually checked the discriminants more specifically might be useful, so could potentially be added in future, but I don't think the *current* checks are actually providing value.)
r? mir
Randomly turns out that this
Fixes https://github.com/rust-lang/rust/issues/121097
Store the type of each GVN value
MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities.
This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code.
Fixesrust-lang/rust#128094Fixesrust-lang/rust#135128
r? ``````@ghost`````` for perf
Dont collect assoc ty item bounds from trait where clause for host effect predicates
For background, we uplift `where Self::Assoc: Trait` bounds in a trait's where clauses into *item bounds* on `type Assoc;`. This is because before we *had* syntactical item bounds, users would express their item bounds like so.
Let's opt out of doing this same behavior for `HostEffect` predicates like `where Self::Assoc: [const] Trait`. I left a comment in the code:
```rust
// FIXME(const_trait_impl): We *could* uplift the
// `where Self::Assoc: [const] Trait` bounds from the parent trait
// here too, but we'd need to split `const_conditions` into two
// queries (like we do for `trait_explicit_predicates_and_bounds`)
// since we need to also filter the predicates *out* of the const
// conditions or they lead to cycles in the trait solver when
// utilizing these bounds. For now, let's do nothing.
```
As an aside, this was an ICE that was only triggerable when building libraries and not binaries because we never were calling `tcx.ensure_ok().explicit_implied_const_bounds(def_id);` on associated types like we should have been. I adjusted the calls to `ensure_ok` to make sure this happens, so we catch bugs like this in the future more easily.
As another aside, I fixed the bound uplifting logic for *always const* predicates, since those act like normal clauses and have no notion of conditional constness.
r? ```@oli-obk``` ```@fee1-dead``` or anyone really
Fixesrust-lang/rust#133275
[perf] Compute hard errors without diagnostics in impl_intersection_has_impossible_obligation
First compute hard errors without diagnostics, then ambiguities with diagnostics since we need to know if any of them overflowed.
Don't give APITs names with macro expansion placeholder fragments in it
The `DefCollector` previously called `pprust::ty_to_string` to construct a name for APITs (arg-position impl traits). The `ast::Ty` that was being formatted however has already had its macro calls replaced with "placeholder fragments", which end up rendering like `!()` (or ICEing, in the case of rust-lang/rust#140333, since it led to a placeholder struct field with no name).
Instead, collect the name of the APIT *before* we visit its macros and replace them with placeholders in the macro expander. This makes the implementation a bit more involved, but AFAICT there's no better way to do this since we can't do a reverse mapping from placeholder fragment -> original macro call AST.
Fixesrust-lang/rust#140333
