bjoernager/rust
bjoernager/rust
1
Fork 0
Code Issues Pull requests Activity

Rollup merge of #139762 - compiler-errors:non-env, r=lcnr

Browse source 
Don't assemble non-env/bound candidates if projection is rigid

Putting this up for an initial review, it's still missing comments, clean-up, and possibly a tweak to deal with ambiguities in the `BestObligation` folder.

This PR fixes https://github.com/rust-lang/trait-system-refactor-initiative/issues/173. Specifically, we're creating an unnecessary query cycle in normalization by assembling an *impl candidate* even if we know later on during `merge_candidates` that we'll be filtering out that impl candidate.

This PR adjusts the `merge_candidates` to assemble *only* env/bound candidates if we have `TraitGoalProvenVia::ParamEnv | TraitGoalProvenVia::AliasBound`.

I'll leave some thoughts/comments in the code.

r? lcnr
This commit is contained in:
Chris Denton 2025-04-19 15:09:34 +00:00 • committed by GitHub
commit 688478fe45
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 107 additions and 56 deletions
Download patch file Download diff file Expand all files Collapse all files

85
compiler/rustc_next_trait_solver/src/solve/assembly/mod.rs
View file

@ -288,6 +288,21 @@ where
) -> Vec<Candidate<I>>;
}
/// Allows callers of `assemble_and_evaluate_candidates` to choose whether to limit
/// candidate assembly to param-env and alias-bound candidates.
///
/// On top of being a micro-optimization, as it avoids doing unnecessary work when
/// a param-env trait bound candidate shadows impls for normalization, this is also
/// required to prevent query cycles due to RPITIT inference. See the issue at:
/// <https://github.com/rust-lang/trait-system-refactor-initiative/issues/173>.
pub(super) enum AssembleCandidatesFrom {
All,
/// Only assemble candidates from the environment and alias bounds, ignoring
/// user-written and built-in impls. We only expect `ParamEnv` and `AliasBound`
/// candidates to be assembled.
EnvAndBounds,
}
impl<D, I> EvalCtxt<'_, D>
where
D: SolverDelegate<Interner = I>,
@ -296,6 +311,7 @@ where
pub(super) fn assemble_and_evaluate_candidates<G: GoalKind<D>>(
&mut self,
goal: Goal<I, G>,
assemble_from: AssembleCandidatesFrom,
) -> Vec<Candidate<I>> {
let Ok(normalized_self_ty) =
self.structurally_normalize_ty(goal.param_env, goal.predicate.self_ty())
@ -322,16 +338,18 @@ where
}
}
self.assemble_impl_candidates(goal, &mut candidates);
self.assemble_builtin_impl_candidates(goal, &mut candidates);
self.assemble_alias_bound_candidates(goal, &mut candidates);
self.assemble_object_bound_candidates(goal, &mut candidates);
self.assemble_param_env_candidates(goal, &mut candidates);
match assemble_from {
AssembleCandidatesFrom::All => {
self.assemble_impl_candidates(goal, &mut candidates);
self.assemble_builtin_impl_candidates(goal, &mut candidates);
self.assemble_object_bound_candidates(goal, &mut candidates);
}
AssembleCandidatesFrom::EnvAndBounds => {}
}
candidates
}
@ -754,6 +772,9 @@ where
})
}
/// Assemble and merge candidates for goals which are related to an underlying trait
/// goal. Right now, this is normalizes-to and host effect goals.
///
/// We sadly can't simply take all possible candidates for normalization goals
/// and check whether they result in the same constraints. We want to make sure
/// that trying to normalize an alias doesn't result in constraints which aren't
@ -782,47 +803,44 @@ where
///
/// See trait-system-refactor-initiative#124 for more details.
#[instrument(level = "debug", skip(self, inject_normalize_to_rigid_candidate), ret)]
pub(super) fn merge_candidates(
pub(super) fn assemble_and_merge_candidates<G: GoalKind<D>>(
&mut self,
proven_via: Option<TraitGoalProvenVia>,
candidates: Vec<Candidate<I>>,
goal: Goal<I, G>,
inject_normalize_to_rigid_candidate: impl FnOnce(&mut EvalCtxt<'_, D>) -> QueryResult<I>,
) -> QueryResult<I> {
let Some(proven_via) = proven_via else {
// We don't care about overflow. If proving the trait goal overflowed, then
// it's enough to report an overflow error for that, we don't also have to
// overflow during normalization.
return Ok(self.make_ambiguous_response_no_constraints(MaybeCause::Ambiguity));
//
// We use `forced_ambiguity` here over `make_ambiguous_response_no_constraints`
// because the former will also record a built-in candidate in the inspector.
return self.forced_ambiguity(MaybeCause::Ambiguity).map(|cand| cand.result);
};
match proven_via {
TraitGoalProvenVia::ParamEnv | TraitGoalProvenVia::AliasBound => {
let mut considered_candidates = Vec::new();
considered_candidates.extend(
candidates
.iter()
.filter(|c| matches!(c.source, CandidateSource::ParamEnv(_)))
.map(|c| c.result),
);
// Even when a trait bound has been proven using a where-bound, we
// still need to consider alias-bounds for normalization, see
// tests/ui/next-solver/alias-bound-shadowed-by-env.rs.
//
// `tests/ui/next-solver/alias-bound-shadowed-by-env.rs`.
let candidates_from_env_and_bounds: Vec<_> = self
.assemble_and_evaluate_candidates(goal, AssembleCandidatesFrom::EnvAndBounds);
// We still need to prefer where-bounds over alias-bounds however.
// See tests/ui/winnowing/norm-where-bound-gt-alias-bound.rs.
//
// FIXME(const_trait_impl): should this behavior also be used by
// constness checking. Doing so is *at least theoretically* breaking,
// see github.com/rust-lang/rust/issues/133044#issuecomment-2500709754
if considered_candidates.is_empty() {
considered_candidates.extend(
candidates
// See `tests/ui/winnowing/norm-where-bound-gt-alias-bound.rs`.
let mut considered_candidates: Vec<_> = if candidates_from_env_and_bounds
.iter()
.filter(|c| matches!(c.source, CandidateSource::AliasBound))
.map(|c| c.result),
);
}
.any(|c| matches!(c.source, CandidateSource::ParamEnv(_)))
{
candidates_from_env_and_bounds
.into_iter()
.filter(|c| matches!(c.source, CandidateSource::ParamEnv(_)))
.map(|c| c.result)
.collect()
} else {
candidates_from_env_and_bounds.into_iter().map(|c| c.result).collect()
};
// If the trait goal has been proven by using the environment, we want to treat
// aliases as rigid if there are no applicable projection bounds in the environment.
@ -839,6 +857,9 @@ where
}
}
TraitGoalProvenVia::Misc => {
let candidates =
self.assemble_and_evaluate_candidates(goal, AssembleCandidatesFrom::All);
// Prefer "orphaned" param-env normalization predicates, which are used
// (for example, and ideally only) when proving item bounds for an impl.
let candidates_from_env: Vec<_> = candidates

3
compiler/rustc_next_trait_solver/src/solve/effect_goals.rs
View file

@ -399,12 +399,11 @@ where
&mut self,
goal: Goal<I, ty::HostEffectPredicate<I>>,
) -> QueryResult<I> {
let candidates = self.assemble_and_evaluate_candidates(goal);
let (_, proven_via) = self.probe(|_| ProbeKind::ShadowedEnvProbing).enter(|ecx| {
let trait_goal: Goal<I, ty::TraitPredicate<I>> =
goal.with(ecx.cx(), goal.predicate.trait_ref);
ecx.compute_trait_goal(trait_goal)
})?;
self.merge_candidates(proven_via, candidates, |_ecx| Err(NoSolution))
self.assemble_and_merge_candidates(proven_via, goal, |_ecx| Err(NoSolution))
}
}

3
compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs
View file

@ -32,14 +32,13 @@ where
let cx = self.cx();
match goal.predicate.alias.kind(cx) {
ty::AliasTermKind::ProjectionTy | ty::AliasTermKind::ProjectionConst => {
let candidates = self.assemble_and_evaluate_candidates(goal);
let trait_ref = goal.predicate.alias.trait_ref(cx);
let (_, proven_via) =
self.probe(|_| ProbeKind::ShadowedEnvProbing).enter(|ecx| {
let trait_goal: Goal<I, ty::TraitPredicate<I>> = goal.with(cx, trait_ref);
ecx.compute_trait_goal(trait_goal)
})?;
self.merge_candidates(proven_via, candidates, |ecx| {
self.assemble_and_merge_candidates(proven_via, goal, |ecx| {
ecx.probe(|&result| ProbeKind::RigidAlias { result }).enter(|this| {
this.structurally_instantiate_normalizes_to_term(
goal,

4
compiler/rustc_next_trait_solver/src/solve/trait_goals.rs
View file

@ -13,7 +13,7 @@ use tracing::{instrument, trace};
use crate::delegate::SolverDelegate;
use crate::solve::assembly::structural_traits::{self, AsyncCallableRelevantTypes};
use crate::solve::assembly::{self, Candidate};
use crate::solve::assembly::{self, AssembleCandidatesFrom, Candidate};
use crate::solve::inspect::ProbeKind;
use crate::solve::{
BuiltinImplSource, CandidateSource, Certainty, EvalCtxt, Goal, GoalSource, MaybeCause,
@ -1365,7 +1365,7 @@ where
&mut self,
goal: Goal<I, TraitPredicate<I>>,
) -> Result<(CanonicalResponse<I>, Option<TraitGoalProvenVia>), NoSolution> {
let candidates = self.assemble_and_evaluate_candidates(goal);
let candidates = self.assemble_and_evaluate_candidates(goal, AssembleCandidatesFrom::All);
self.merge_trait_candidates(goal, candidates)
}
}

34
tests/ui/impl-unused-tps.stderr
View file

@ -7,23 +7,6 @@ LL | impl<T> Foo<T> for [isize; 0] {
LL | impl<T, U> Foo<T> for U {
| ^^^^^^^^^^^^^^^^^^^^^^^ conflicting implementation for `[isize; 0]`
error[E0207]: the type parameter `U` is not constrained by the impl trait, self type, or predicates
--> $DIR/impl-unused-tps.rs:32:9
|
LL | impl<T, U> Bar for T {
| ^ unconstrained type parameter
error[E0119]: conflicting implementations of trait `Bar`
--> $DIR/impl-unused-tps.rs:40:1
|
LL | impl<T, U> Bar for T {
| -------------------- first implementation here
...
LL | / impl<T, U> Bar for T
LL | | where
LL | | T: Bar<Out = U>,
| |____________________^ conflicting implementation
error[E0119]: conflicting implementations of trait `Foo<[isize; 0]>` for type `[isize; 0]`
--> $DIR/impl-unused-tps.rs:49:1
|
@ -52,6 +35,12 @@ error[E0207]: the type parameter `U` is not constrained by the impl trait, self
LL | impl<T, U> Foo<T> for [isize; 1] {
| ^ unconstrained type parameter
error[E0207]: the type parameter `U` is not constrained by the impl trait, self type, or predicates
--> $DIR/impl-unused-tps.rs:32:9
|
LL | impl<T, U> Bar for T {
| ^ unconstrained type parameter
error[E0207]: the type parameter `U` is not constrained by the impl trait, self type, or predicates
--> $DIR/impl-unused-tps.rs:40:9
|
@ -70,6 +59,17 @@ error[E0207]: the type parameter `V` is not constrained by the impl trait, self
LL | impl<T, U, V> Foo<T> for T
| ^ unconstrained type parameter
error[E0119]: conflicting implementations of trait `Bar`
--> $DIR/impl-unused-tps.rs:40:1
|
LL | impl<T, U> Bar for T {
| -------------------- first implementation here
...
LL | / impl<T, U> Bar for T
LL | | where
LL | | T: Bar<Out = U>,
| |____________________^ conflicting implementation
error: aborting due to 9 previous errors
Some errors have detailed explanations: E0119, E0207.

32
tests/ui/traits/next-solver/rpitit-cycle-due-to-rigid.rs Normal file
View file

@ -0,0 +1,32 @@
//@ compile-flags: -Znext-solver
//@ check-pass
//@ edition: 2024
// Ensure we don't end up in a query cycle due to trying to assemble an impl candidate
// for an RPITIT normalizes-to goal, even though that impl candidate would *necessarily*
// be made rigid by a where clause. This query cycle is thus avoidable by not assembling
// that impl candidate which we *know* we are going to throw away anyways.
use std::future::Future;
pub trait ReactiveFunction: Send {
type Output;
fn invoke(self) -> Self::Output;
}
trait AttributeValue {
fn resolve(self) -> impl Future<Output = ()> + Send;
}
impl<F, V> AttributeValue for F
where
F: ReactiveFunction<Output = V>,
V: AttributeValue,
{
async fn resolve(self) {
self.invoke().resolve().await
}
}
fn main() {}