rustc_next_trait_solver/solve/eval_ctxt/

mod.rs

use std::mem;
use std::ops::ControlFlow;

#[cfg(feature = "nightly")]
use rustc_macros::StableHash;
use rustc_type_ir::data_structures::HashSet;
use rustc_type_ir::inherent::*;
use rustc_type_ir::region_constraint::RegionConstraint;
use rustc_type_ir::relate::Relate;
use rustc_type_ir::relate::solver_relating::RelateExt;
use rustc_type_ir::search_graph::{CandidateHeadUsages, LowerAvailableDepth, PathKind};
use rustc_type_ir::solve::{
    AccessedOpaques, ExternalRegionConstraints, FetchEligibleAssocItemResponse, MaybeInfo,
    NoSolutionOrRerunNonErased, OpaqueTypesJank, QueryResultOrRerunNonErased, RerunCondition,
    RerunNonErased, RerunReason, RerunResultExt, SmallCopyList,
};
use rustc_type_ir::{
    self as ty, CanonicalVarValues, ClauseKind, InferCtxtLike, Interner, MayBeErased,
    OpaqueTypeKey, PredicateKind, Region, TypeFoldable, TypeSuperVisitable, TypeVisitable,
    TypeVisitableExt, TypeVisitor, TypingMode, eager_resolve_vars,
};
use thin_vec::ThinVec;
use tracing::{Level, debug, instrument, trace, warn};

use super::has_only_region_constraints;
use crate::canonical::{
    canonicalize_goal, canonicalize_response, instantiate_and_apply_query_response,
    response_no_constraints_raw,
};
use crate::coherence;
use crate::delegate::SolverDelegate;
use crate::normalize::{NormalizationFolder, NormalizationWasAmbiguous};
use crate::placeholder::BoundVarReplacer;
use crate::solve::eval_ctxt::fast_path::{
    RerunStalled, compute_goal_fast_path, rerunning_stalled_goal_may_make_progress,
};
use crate::solve::fast_path::compute_goal_fast_path_cold;
use crate::solve::search_graph::SearchGraph;
use crate::solve::ty::may_use_unstable_feature;
use crate::solve::{
    CanonicalInput, CanonicalResponse, Certainty, ExternalConstraintsData, FIXPOINT_STEP_LIMIT,
    Goal, GoalEvaluation, GoalSource, GoalStalledOn, GoalStalledOnOpaques, HasChanged, MaybeCause,
    NestedNormalizationGoals, NoSolution, QueryInput, QueryResult, Response, SucceededInErased,
    VisibleForLeakCheck, inspect,
};

pub mod fast_path;
mod probe;
mod solver_region_constraints;

/// The kind of goal we're currently proving.
///
/// This has effects on cycle handling handling and on how we compute
/// query responses, see the variant descriptions for more info.
#[derive(Debug, Copy, Clone)]
enum CurrentGoalKind {
    Misc,
    /// We're proving an trait goal for a coinductive trait, either an auto trait or `Sized`.
    ///
    /// These are currently the only goals whose impl where-clauses are considered to be
    /// productive steps.
    CoinductiveTrait,
    // FIXME: Consider renaming `PredicateKind::NormalizesTo` to match with this
    /// Unlike other goals, `NormalizesTo` goals aren't independent goals but just implementation
    /// details for handling projections of associated terms. When we encounter a `Projection` goal
    /// whose `projection_term` is an associated term, we create a `NormalizesTo` goal whose
    /// expected term is fully unconstrained and evaluate it.
    ///
    /// This would weaken inference however, as the nested goals of normalizes-to never get the
    /// inference constraints from the actual expected term. We just gather candidates from the
    /// normalizes-to goal and return any ambiguous nested goals of it to the caller (`Projection
    /// goal`). The caller handle and evaluate them as if they were its own nested goals.
    ///
    /// Because of this, evaluating a normalizes-to goal is computing candidates for projection of
    /// an associated term and it never leaks out of the solver.
    ProjectionComputeAssocTermCandidate,
}

impl CurrentGoalKind {
    fn from_query_input<I: Interner>(cx: I, input: QueryInput<I, I::Predicate>) -> CurrentGoalKind {
        match input.goal.predicate.kind().skip_binder() {
            ty::PredicateKind::Clause(ty::ClauseKind::Trait(pred)) => {
                if cx.trait_is_coinductive(pred.trait_ref.def_id) {
                    CurrentGoalKind::CoinductiveTrait
                } else {
                    CurrentGoalKind::Misc
                }
            }
            ty::PredicateKind::NormalizesTo(_) => {
                CurrentGoalKind::ProjectionComputeAssocTermCandidate
            }
            _ => CurrentGoalKind::Misc,
        }
    }
}

pub struct EvalCtxt<'a, D, I = <D as SolverDelegate>::Interner>
where
    D: SolverDelegate<Interner = I>,
    I: Interner,
{
    /// The inference context that backs (mostly) inference and placeholder terms
    /// instantiated while solving goals.
    ///
    /// NOTE: The `InferCtxt` that backs the `EvalCtxt` is intentionally private,
    /// because the `InferCtxt` is much more general than `EvalCtxt`. Methods such
    /// as  `take_registered_region_obligations` can mess up query responses,
    /// using `At::normalize` is totally wrong, calling `evaluate_root_goal` can
    /// cause coinductive unsoundness, etc.
    ///
    /// Methods that are generally of use for trait solving are *intentionally*
    /// re-declared through the `EvalCtxt` below, often with cleaner signatures
    /// since we don't care about things like `ObligationCause`s and `Span`s here.
    /// If some `InferCtxt` method is missing, please first think defensively about
    /// the method's compatibility with this solver, or if an existing one does
    /// the job already.
    delegate: &'a D,

    /// The variable info for the `var_values`, only used to make an ambiguous response
    /// with no constraints.
    var_kinds: I::CanonicalVarKinds,

    /// What kind of goal we're currently computing, see the enum definition
    /// for more info.
    current_goal_kind: CurrentGoalKind,
    pub(super) var_values: CanonicalVarValues<I>,

    /// The highest universe index nameable by the caller.
    ///
    /// When we enter a new binder inside of the query we create new universes
    /// which the caller cannot name. We have to be careful with variables from
    /// these new universes when creating the query response.
    ///
    /// Both because these new universes can prevent us from reaching a fixpoint
    /// if we have a coinductive cycle and because that's the only way we can return
    /// new placeholders to the caller.
    pub(super) max_input_universe: ty::UniverseIndex,
    /// The opaque types from the canonical input. We only need to return opaque types
    /// which have been added to the storage while evaluating this goal.
    pub(super) initial_opaque_types_storage_num_entries:
        <D::Infcx as InferCtxtLike>::OpaqueTypeStorageEntries,

    pub(super) search_graph: &'a mut SearchGraph<D>,

    nested_goals: Vec<(GoalSource, Goal<I, I::Predicate>, Option<GoalStalledOn<I>>)>,

    pub(super) origin_span: I::Span,

    // Has this `EvalCtxt` errored out with `NoSolution` in `try_evaluate_added_goals`?
    //
    // If so, then it can no longer be used to make a canonical query response,
    // since subsequent calls to `try_evaluate_added_goals` have possibly dropped
    // ambiguous goals. Instead, a probe needs to be introduced somewhere in the
    // evaluation code.
    tainted: Result<(), NoSolution>,

    /// Tracks accesses of opaque types while in [`TypingMode::ErasedNotCoherence`].
    pub(super) opaque_accesses: AccessedOpaques<I>,

    pub(super) inspect: inspect::EvaluationStepBuilder<D>,
}

#[derive(PartialEq, Eq, Debug, Hash, Clone, Copy)]
#[cfg_attr(feature = "nightly", derive(StableHash))]
pub enum GenerateProofTree {
    Yes,
    No,
}

pub trait SolverDelegateEvalExt: SolverDelegate {
    /// Evaluates a goal from **outside** of the trait solver.
    ///
    /// Using this while inside of the solver is wrong as it uses a new
    /// search graph which would break cycle detection.
    fn evaluate_root_goal(
        &self,
        goal: Goal<Self::Interner, <Self::Interner as Interner>::Predicate>,
        span: <Self::Interner as Interner>::Span,
        stalled_on: Option<GoalStalledOn<Self::Interner>>,
    ) -> Result<GoalEvaluation<Self::Interner>, NoSolution>;

    /// Checks whether a stalled goal would remain stalled if re-evaluated, without consuming
    /// `stalled_on`.
    fn goal_remains_stalled(&self, stalled_on: &GoalStalledOn<Self::Interner>)
    -> Option<Certainty>;

    /// Checks whether evaluating `goal` may hold while treating not-yet-defined
    /// opaque types as being kind of rigid.
    ///
    /// See the comment on [OpaqueTypesJank] for more details.
    fn root_goal_may_hold_opaque_types_jank(
        &self,
        goal: Goal<Self::Interner, <Self::Interner as Interner>::Predicate>,
    ) -> bool;

    /// Check whether evaluating `goal` with a depth of `root_depth` may
    /// succeed. This only returns `false` if the goal is guaranteed to
    /// not hold. In case evaluation overflows and fails with ambiguity this
    /// returns `true`.
    ///
    /// This is only intended to be used as a performance optimization
    /// in coherence checking.
    fn root_goal_may_hold_with_depth(
        &self,
        root_depth: usize,
        goal: Goal<Self::Interner, <Self::Interner as Interner>::Predicate>,
    ) -> bool;

    // FIXME: This is only exposed because we need to use it in `analyse.rs`
    // which is not yet uplifted. Once that's done, we should remove this.
    fn evaluate_root_goal_for_proof_tree(
        &self,
        goal: Goal<Self::Interner, <Self::Interner as Interner>::Predicate>,
        span: <Self::Interner as Interner>::Span,
    ) -> (
        Result<NestedNormalizationGoals<Self::Interner>, NoSolution>,
        inspect::GoalEvaluation<Self::Interner>,
    );
}

impl<D, I> SolverDelegateEvalExt for D
where
    D: SolverDelegate<Interner = I>,
    I: Interner,
{
    #[instrument(level = "debug", skip(self), ret)]
    fn evaluate_root_goal(
        &self,
        goal: Goal<I, I::Predicate>,
        span: I::Span,
        stalled_on: Option<GoalStalledOn<I>>,
    ) -> Result<GoalEvaluation<I>, NoSolution> {
        // Run fast paths *before* building an `EvalCtxt`, saving a little bit of time.
        if let RerunStalled::WontMakeProgress(stalled_certainty) =
            rerunning_stalled_goal_may_make_progress(self, stalled_on.as_ref())
        {
            return Ok(GoalEvaluation {
                goal,
                certainty: stalled_certainty,
                has_changed: HasChanged::No,
                stalled_on,
            });
        }

        // No need to try the fast path if stalled_on is `None`, since we already try the fast path
        // immediately when adding new goals. If we didn't check `stalled_on` here we'd be trying
        // the fast path twice for some goals.
        if stalled_on.is_some()
            && let Some(res) = compute_goal_fast_path_cold(self, goal, span)
        {
            return Ok(res);
        }

        let mut result = EvalCtxt::enter_root(self, self.cx().recursion_limit(), span, |ecx| {
            ecx.evaluate_goal_no_fast_paths(GoalSource::Misc, goal)
        });
        maybe_evaluate_root_goal_with_higher_recursion_limit(self, goal, span, &mut result);

        match result {
            Ok(i) => Ok(i),
            Err(NoSolutionOrRerunNonErased::NoSolution(NoSolution)) => Err(NoSolution),
            Err(NoSolutionOrRerunNonErased::RerunNonErased(_)) => {
                unreachable!("this never happens at the root, we're never in erased mode here");
            }
        }
    }

    fn goal_remains_stalled(
        &self,
        stalled_on: &GoalStalledOn<Self::Interner>,
    ) -> Option<Certainty> {
        match rerunning_stalled_goal_may_make_progress(self, Some(stalled_on)) {
            RerunStalled::WontMakeProgress(certainty) => Some(certainty),
            RerunStalled::MayMakeProgress => None,
        }
    }

    #[instrument(level = "debug", skip(self), ret)]
    fn root_goal_may_hold_opaque_types_jank(
        &self,
        goal: Goal<Self::Interner, <Self::Interner as Interner>::Predicate>,
    ) -> bool {
        self.probe(|| {
            EvalCtxt::enter_root(self, self.cx().recursion_limit(), I::Span::dummy(), |ecx| {
                ecx.evaluate_goal(GoalSource::Misc, goal, None)
            })
            .is_ok_and(|r| match r.certainty {
                Certainty::Yes => true,
                Certainty::Maybe(MaybeInfo {
                    cause: _,
                    opaque_types_jank,
                    stalled_on_coroutines: _,
                }) => match opaque_types_jank {
                    OpaqueTypesJank::AllGood => true,
                    OpaqueTypesJank::ErrorIfRigidSelfTy => false,
                },
            })
        })
    }

    fn root_goal_may_hold_with_depth(
        &self,
        root_depth: usize,
        goal: Goal<Self::Interner, <Self::Interner as Interner>::Predicate>,
    ) -> bool {
        self.probe(|| {
            EvalCtxt::enter_root(self, root_depth, I::Span::dummy(), |ecx| {
                ecx.evaluate_goal(GoalSource::Misc, goal, None)
            })
        })
        .is_ok()
    }

    #[instrument(level = "debug", skip(self))]
    fn evaluate_root_goal_for_proof_tree(
        &self,
        goal: Goal<I, I::Predicate>,
        span: I::Span,
    ) -> (Result<NestedNormalizationGoals<I>, NoSolution>, inspect::GoalEvaluation<I>) {
        let mut result =
            evaluate_root_goal_for_proof_tree(self, goal, span, self.cx().recursion_limit());
        maybe_evaluate_root_goal_for_proof_tree_with_higher_recursion_limit(
            self,
            goal,
            span,
            &mut result,
        );
        result
    }
}

/// The old solver doesn't check depth requirement when looking up cache while the next solver
/// does so. Thus the next solver is more prone to overflow.
/// To mitigate breakages, we re-evaluate the overflowed goal with doubled recursion limit
/// and emit a FCW if it succeeds.
/// See the doc comment on `RECURSION_DEPTH_EXCEEDING_LIMIT` and #159228 for more details.
fn maybe_evaluate_root_goal_with_higher_recursion_limit<D, I>(
    delegate: &D,
    goal: Goal<I, I::Predicate>,
    span: I::Span,
    initial_result: &mut Result<GoalEvaluation<I>, NoSolutionOrRerunNonErased>,
) where
    D: SolverDelegate<Interner = I>,
    I: Interner,
{
    if !delegate.enable_next_solver_overflow_fcw() {
        return;
    }

    let predicate = match initial_result {
        Err(_) => return,
        Ok(goal_evaluation) if !goal_evaluation.certainty.is_overflow() => return,
        Ok(goal_evaluation) => goal_evaluation.goal.predicate,
    };

    let rerun_result = delegate.commit_if_ok(|| {
        let rerun_result =
            EvalCtxt::enter_root(delegate, delegate.cx().recursion_limit() * 2, span, |ecx| {
                ecx.evaluate_goal_no_fast_paths(GoalSource::Misc, goal)
            });
        if let Ok(goal_evaluation) = &rerun_result
            && goal_evaluation.certainty.is_yes()
        {
            Ok(rerun_result)
        } else {
            Err(())
        }
    });
    if let Ok(rerun_result) = rerun_result {
        delegate.cx().emit_next_solver_overflow_fcw(predicate, span);
        *initial_result = rerun_result;
    }
}

/// The old solver doesn't check depth requirement when looking up cache while the next solver
/// does so. Thus the next solver is more prone to overflow.
/// To mitigate breakages, we re-evaluate the overflowed goal with doubled recursion limit
/// and emit a FCW if it succeeds.
/// See the doc comment on `RECURSION_DEPTH_EXCEEDING_LIMIT` and #159228 for more details.
fn maybe_evaluate_root_goal_for_proof_tree_with_higher_recursion_limit<D, I>(
    delegate: &D,
    goal: Goal<I, I::Predicate>,
    span: I::Span,
    initial_result: &mut (
        Result<NestedNormalizationGoals<I>, NoSolution>,
        inspect::GoalEvaluation<I>,
    ),
) where
    D: SolverDelegate<Interner = I>,
    I: Interner,
{
    if !delegate.enable_next_solver_overflow_fcw() {
        return;
    }

    let goal_evaluation = &initial_result.1;
    match goal_evaluation.result {
        Err(_) => return,
        Ok(response) if !response.value.certainty.is_overflow() => return,
        Ok(_) => {}
    }

    let rerun_result = delegate.commit_if_ok(|| {
        let (new_result, new_goal_evaluation) = evaluate_root_goal_for_proof_tree(
            delegate,
            goal,
            span,
            delegate.cx().recursion_limit() * 2,
        );
        if let Ok(response) = &new_goal_evaluation.result
            && response.value.certainty.is_yes()
        {
            Ok((new_result, new_goal_evaluation))
        } else {
            Err(())
        }
    });
    if let Ok(rerun_result) = rerun_result {
        let predicate: I::Predicate = goal_evaluation.uncanonicalized_goal.predicate;
        delegate.cx().emit_next_solver_overflow_fcw(predicate, span);
        *initial_result = rerun_result;
    }
}

impl<'a, D, I> EvalCtxt<'a, D>
where
    D: SolverDelegate<Interner = I>,
    I: Interner,
{
    pub(super) fn typing_mode(&self) -> TypingMode<I> {
        self.delegate.typing_mode_raw()
    }

    /// Computes the `PathKind` for the step from the current goal to the
    /// nested goal required due to `source`.
    ///
    /// See #136824 for a more detailed reasoning for this behavior. We
    /// consider cycles to be coinductive if they 'step into' a where-clause
    /// of a coinductive trait. We will likely extend this function in the future
    /// and will need to clearly document it in the rustc-dev-guide before
    /// stabilization.
    pub(super) fn step_kind_for_source(&self, source: GoalSource) -> PathKind {
        match source {
            // We treat these goals as unknown for now. It is likely that most miscellaneous
            // nested goals will be converted to an inductive variant in the future.
            //
            // Having unknown cycles is always the safer option, as changing that to either
            // succeed or hard error is backwards compatible. If we incorrectly treat a cycle
            // as inductive even though it should not be, it may be unsound during coherence and
            // fixing it may cause inference breakage or introduce ambiguity.
            GoalSource::Misc => PathKind::Unknown,
            GoalSource::NormalizeGoal(path_kind) => path_kind,
            GoalSource::ImplWhereBound => match self.current_goal_kind {
                // We currently only consider a cycle coinductive if it steps
                // into a where-clause of a coinductive trait.
                CurrentGoalKind::CoinductiveTrait => PathKind::Coinductive,
                // We probably want to make all traits coinductive in the future,
                // so we treat cycles involving where-clauses of not-yet coinductive
                // traits as ambiguous for now.
                CurrentGoalKind::Misc | CurrentGoalKind::ProjectionComputeAssocTermCandidate => {
                    PathKind::Unknown
                }
            },
            // Relating types is always unproductive. If we were to map proof trees to
            // corecursive functions as explained in #136824, relating types never
            // introduces a constructor which could cause the recursion to be guarded.
            GoalSource::TypeRelating => PathKind::Inductive,
            // These goal sources are likely unproductive and can be changed to
            // `PathKind::Inductive`. Keeping them as unknown until we're confident
            // about this and have an example where it is necessary.
            GoalSource::AliasBoundConstCondition | GoalSource::AliasWellFormed => PathKind::Unknown,
        }
    }

    /// Creates a root evaluation context and search graph. This should only be
    /// used from outside of any evaluation, and other methods should be preferred
    /// over using this manually (such as [`SolverDelegateEvalExt::evaluate_root_goal`]).
    pub(super) fn enter_root<R>(
        delegate: &D,
        root_depth: usize,
        origin_span: I::Span,
        f: impl FnOnce(&mut EvalCtxt<'_, D>) -> R,
    ) -> R {
        let mut search_graph = SearchGraph::new(root_depth);

        let mut ecx = EvalCtxt {
            delegate,
            search_graph: &mut search_graph,
            nested_goals: Default::default(),
            inspect: inspect::EvaluationStepBuilder::new_noop(),

            // Only relevant when canonicalizing the response,
            // which we don't do within this evaluation context.
            max_input_universe: ty::UniverseIndex::ROOT,
            initial_opaque_types_storage_num_entries: Default::default(),
            var_kinds: Default::default(),
            var_values: CanonicalVarValues::dummy(),
            current_goal_kind: CurrentGoalKind::Misc,
            origin_span,
            tainted: Ok(()),
            opaque_accesses: AccessedOpaques::default(),
        };
        let result = f(&mut ecx);
        assert!(
            ecx.nested_goals.is_empty(),
            "root `EvalCtxt` should not have any goals added to it"
        );
        assert!(!ecx.opaque_accesses.might_rerun());
        assert!(search_graph.is_empty());
        result
    }

    /// Creates a nested evaluation context that shares the same search graph as the
    /// one passed in. This is suitable for evaluation, granted that the search graph
    /// has had the nested goal recorded on its stack. This method only be used by
    /// `search_graph::Delegate::compute_goal`.
    ///
    /// This function takes care of setting up the inference context, setting the anchor,
    /// and registering opaques from the canonicalized input.
    pub(super) fn enter_canonical<T>(
        cx: I,
        search_graph: &'a mut SearchGraph<D>,
        canonical_input: CanonicalInput<I>,
        proof_tree_builder: &mut inspect::ProofTreeBuilder<D>,
        f: impl FnOnce(
            &mut EvalCtxt<'_, D>,
            Goal<I, I::Predicate>,
        ) -> Result<T, NoSolutionOrRerunNonErased>,
    ) -> (Result<T, NoSolution>, AccessedOpaques<I>) {
        let (ref delegate, input, var_values) = D::build_with_canonical(cx, &canonical_input);
        for (key, ty) in input.predefined_opaques_in_body.iter() {
            let prev = delegate.register_hidden_type_in_storage(key, ty, I::Span::dummy());
            // It may be possible that two entries in the opaque type storage end up
            // with the same key after resolving contained inference variables.
            //
            // We could put them in the duplicate list but don't have to. The opaques we
            // encounter here are already tracked in the caller, so there's no need to
            // also store them here. We'd take them out when computing the query response
            // and then discard them, as they're already present in the input.
            //
            // Ideally we'd drop duplicate opaque type definitions when computing
            // the canonical input. This is more annoying to implement and may cause a
            // perf regression, so we do it inside of the query for now.
            if let Some(prev) = prev {
                debug!(?key, ?ty, ?prev, "ignore duplicate in `opaque_types_storage`");
            }
        }

        let initial_opaque_types_storage_num_entries = delegate.opaque_types_storage_num_entries();
        if cfg!(debug_assertions) && delegate.typing_mode_raw().is_erased_not_coherence() {
            assert!(delegate.clone_opaque_types_lookup_table().is_empty());
        }

        let mut ecx = EvalCtxt {
            delegate,
            var_kinds: canonical_input.canonical.var_kinds,
            var_values,
            current_goal_kind: CurrentGoalKind::from_query_input(cx, input),
            max_input_universe: canonical_input.canonical.max_universe,
            initial_opaque_types_storage_num_entries,
            search_graph,
            nested_goals: Default::default(),
            origin_span: I::Span::dummy(),
            tainted: Ok(()),
            inspect: proof_tree_builder.new_evaluation_step(var_values),
            opaque_accesses: AccessedOpaques::default(),
        };

        let result = f(&mut ecx, input.goal);
        ecx.inspect.probe_final_state(ecx.delegate, ecx.max_input_universe);
        proof_tree_builder.finish_evaluation_step(ecx.inspect);

        if canonical_input.typing_mode.0.is_erased_not_coherence() {
            debug_assert!(delegate.clone_opaque_types_lookup_table().is_empty());
        }

        // When creating a query response we clone the opaque type constraints
        // instead of taking them. This would cause an ICE here, since we have
        // assertions against dropping an `InferCtxt` without taking opaques.
        // FIXME: Once we remove support for the old impl we can remove this.
        // FIXME: Could we make `build_with_canonical` into `enter_with_canonical` and call this at the end?
        delegate.reset_opaque_types();

        let opaque_accesses = ecx.opaque_accesses;
        (
            match result {
                Ok(i) => Ok(i),
                Err(NoSolutionOrRerunNonErased::NoSolution(NoSolution)) => Err(NoSolution),
                Err(NoSolutionOrRerunNonErased::RerunNonErased(_)) => {
                    // Check that the opaque_accesses state mirrors the result we got.
                    assert!(opaque_accesses.should_bail().is_err());
                    Err(NoSolution)
                }
            },
            opaque_accesses,
        )
    }

    pub(super) fn ignore_candidate_head_usages(&mut self, usages: CandidateHeadUsages) {
        self.search_graph.ignore_candidate_head_usages(usages);
    }

    /// Recursively evaluates `goal`, returning whether any inference vars have
    /// been constrained and the certainty of the result.
    fn evaluate_goal(
        &mut self,
        source: GoalSource,
        goal: Goal<I, I::Predicate>,
        stalled_on: Option<GoalStalledOn<I>>,
    ) -> Result<GoalEvaluation<I>, NoSolutionOrRerunNonErased> {
        if let RerunStalled::WontMakeProgress(stalled_certainty) =
            rerunning_stalled_goal_may_make_progress(self.delegate, stalled_on.as_ref())
        {
            return Ok(GoalEvaluation {
                goal,
                certainty: stalled_certainty,
                has_changed: HasChanged::No,
                stalled_on,
            });
        }

        // No need to try the fast path if stalled_on is `None`, since we already try the fast path
        // immediately when adding new goals. If we didn't check `stalled_on` here we'd be trying
        // the fast path twice for some goals.
        if stalled_on.is_some()
            && let Some(res) = compute_goal_fast_path_cold(self.delegate, goal, self.origin_span)
        {
            return Ok(res);
        }

        self.evaluate_goal_no_fast_paths(source, goal)
    }

    // Outlining and `#[cold]` matter here because fast paths make it less likely to get here.
    #[cold]
    #[inline(never)]
    fn evaluate_goal_no_fast_paths(
        &mut self,
        source: GoalSource,
        goal: Goal<I, I::Predicate>,
    ) -> Result<GoalEvaluation<I>, NoSolutionOrRerunNonErased> {
        let (normalization_nested_goals, goal_evaluation) =
            self.evaluate_goal_raw(source, goal, LowerAvailableDepth::Yes)?;
        assert!(normalization_nested_goals.is_empty());
        Ok(goal_evaluation)
    }

    /// Recursively evaluates `goal`, returning the nested goals in case
    /// the nested goal is a `NormalizesTo` goal.
    ///
    /// As all other goal kinds do not return any nested goals and
    /// `NormalizesTo` is only used by `Projection`, all other callsites
    /// should use [`EvalCtxt::evaluate_goal`] which discards that empty
    /// storage.
    pub(super) fn evaluate_goal_raw(
        &mut self,
        source: GoalSource,
        goal: Goal<I, I::Predicate>,
        increase_depth_for_nested: LowerAvailableDepth,
    ) -> Result<(NestedNormalizationGoals<I>, GoalEvaluation<I>), NoSolutionOrRerunNonErased> {
        // We only care about one entry per `OpaqueTypeKey` here,
        // so we only canonicalize the lookup table and ignore
        // duplicate entries.
        let opaque_types = self.delegate.clone_opaque_types_lookup_table();
        let (goal, opaque_types) = eager_resolve_vars(&**self.delegate, (goal, opaque_types));
        let typing_mode = self.typing_mode();
        let step_kind = self.step_kind_for_source(source);

        let tracing_span = tracing::span!(
            Level::DEBUG,
            "evaluate_goal_raw in typing mode",
            "{:?} opaques={:?}",
            typing_mode,
            opaque_types
        )
        .entered();

        let (result, orig_values, canonical_goal, succeeded_in_erased) = 'retry_canonicalize: {
            let skip_erased_attempt = match typing_mode {
                TypingMode::Reflection | TypingMode::Coherence => true,
                TypingMode::Typeck { .. }
                | TypingMode::PostTypeckUntilBorrowck { .. }
                | TypingMode::PostBorrowck { .. }
                | TypingMode::Codegen
                | TypingMode::PostAnalysis
                | TypingMode::ErasedNotCoherence(_) => {
                    let mut skip = false;
                    if opaque_types.iter().any(|(_, ty)| ty.is_ty_var())
                        && let PredicateKind::Clause(ClauseKind::Trait(..)) =
                            goal.predicate.kind().skip_binder()
                    {
                        skip = true;
                    }

                    if let PredicateKind::Clause(ClauseKind::Trait(tr)) =
                        goal.predicate.kind().skip_binder()
                        && tr.self_ty().has_coroutines()
                        && self.cx().trait_is_auto(tr.trait_ref.def_id)
                    {
                        // FIXME(#155443): this doesn't make a difference now, but with eager normalization
                        // it likely will.
                        // skip_erased_attempt = true;
                    }

                    skip
                }
            };

            if skip_erased_attempt {
                if typing_mode.is_erased_not_coherence() {
                    match self.opaque_accesses.rerun_always(RerunReason::SkipErasedAttempt)? {}
                } else {
                    debug!("running in original typing mode");
                }
            } else {
                debug!("trying without opaques: {goal:?}");

                let (orig_values, canonical_goal) = canonicalize_goal(
                    self.delegate,
                    goal,
                    &[],
                    TypingMode::ErasedNotCoherence(MayBeErased),
                );

                let (canonical_result, accessed_opaques) = self.search_graph.evaluate_goal(
                    self.cx(),
                    canonical_goal,
                    step_kind,
                    increase_depth_for_nested,
                    &mut inspect::ProofTreeBuilder::new_noop(),
                );

                let should_rerun = should_rerun_after_erased_canonicalization(
                    accessed_opaques,
                    self.typing_mode(),
                    &opaque_types,
                );
                match should_rerun {
                    RerunDecision::Yes => debug!("rerunning in original typing mode"),
                    RerunDecision::No => {
                        break 'retry_canonicalize (
                            canonical_result,
                            orig_values,
                            canonical_goal,
                            SucceededInErased::Yes { accessed_opaques },
                        );
                    }
                    RerunDecision::EagerlyPropagateToParent => {
                        self.opaque_accesses.update(accessed_opaques)?;
                        break 'retry_canonicalize (
                            canonical_result,
                            orig_values,
                            canonical_goal,
                            // If we're propagating up, we should never retry the goal.
                            // That means `No` is fine to return, it doesn't really matter.
                            SucceededInErased::No,
                        );
                    }
                }
            }

            let (orig_values, canonical_goal) =
                canonicalize_goal(self.delegate, goal, &opaque_types, typing_mode);

            let (canonical_result, accessed_opaques) = self.search_graph.evaluate_goal(
                self.cx(),
                canonical_goal,
                step_kind,
                increase_depth_for_nested,
                &mut inspect::ProofTreeBuilder::new_noop(),
            );
            assert!(
                !accessed_opaques.might_rerun(),
                "we run without TypingMode::ErasedNotCoherence, so opaques are available, and we don't retry if the outer typing mode is ErasedNotCoherence: {accessed_opaques:?} after {goal:?}"
            );

            (canonical_result, orig_values, canonical_goal, SucceededInErased::No)
        };

        debug!(?result);
        let response = match result {
            Ok(response) => {
                debug!("success");
                response
            }
            Err(NoSolution) => {
                debug!("normal failure");
                return Err(NoSolution.into());
            }
        };

        drop(tracing_span);

        let has_changed =
            if !has_only_region_constraints(response) { HasChanged::Yes } else { HasChanged::No };

        let (normalization_nested_goals, certainty) = instantiate_and_apply_query_response(
            self.delegate,
            goal.param_env,
            &orig_values,
            response,
            self.origin_span,
        );

        // FIXME: We previously had an assert here that checked that recomputing
        // a goal after applying its constraints did not change its response.
        //
        // This assert was removed as it did not hold for goals constraining
        // an inference variable to a recursive alias, e.g. in
        // tests/ui/traits/next-solver/overflow/recursive-self-normalization.rs.
        //
        // Once we have decided on how to handle trait-system-refactor-initiative#75,
        // we should re-add an assert here.

        let stalled_on = match certainty {
            Certainty::Yes => None,
            Certainty::Maybe { .. } => match has_changed {
                // FIXME: We could recompute a *new* set of stalled variables by walking
                // through the orig values, resolving, and computing the root vars of anything
                // that is not resolved. Only when *these* have changed is it meaningful
                // to recompute this goal.
                HasChanged::Yes => None,
                HasChanged::No => Some(self.build_stalled_on(
                    canonical_goal,
                    certainty,
                    orig_values,
                    succeeded_in_erased,
                )),
            },
        };

        Ok((
            normalization_nested_goals,
            GoalEvaluation { goal, certainty, has_changed, stalled_on },
        ))
    }

    fn build_stalled_on(
        &self,
        canonical_goal: CanonicalInput<I>,
        certainty: Certainty,
        mut stalled_vars: ThinVec<I::GenericArg>,
        previously_succeeded_in_erased: SucceededInErased<I>,
    ) -> GoalStalledOn<I> {
        // Remove the canonicalized universal vars, since we only care about stalled existentials.
        let mut sub_roots = ThinVec::new();
        stalled_vars.retain(|arg| match arg.kind() {
            // Lifetimes can never stall goals.
            ty::GenericArgKind::Lifetime(_) => false,
            ty::GenericArgKind::Type(ty) => match ty.kind() {
                ty::Infer(ty::TyVar(vid)) => {
                    sub_roots.push(self.delegate.sub_unification_table_root_var(vid));
                    true
                }
                ty::Infer(_) => true,
                ty::Param(_) | ty::Placeholder(_) => false,
                _ => unreachable!("unexpected orig_value: {ty:?}"),
            },
            ty::GenericArgKind::Const(ct) => match ct.kind() {
                ty::ConstKind::Infer(_) => true,
                ty::ConstKind::Param(_) | ty::ConstKind::Placeholder(_) => false,
                _ => unreachable!("unexpected orig_value: {ct:?}"),
            },
        });

        GoalStalledOn {
            stalled_vars,
            sub_roots,
            stalled_certainty: certainty,
            opaques: GoalStalledOnOpaques::Yes {
                num_opaques_in_storage: canonical_goal
                    .canonical
                    .value
                    .predefined_opaques_in_body
                    .len(),
                previously_succeeded_in_erased,
            },
        }
    }

    pub(super) fn compute_goal(
        &mut self,
        goal: Goal<I, I::Predicate>,
    ) -> QueryResultOrRerunNonErased<I> {
        let Goal { param_env, predicate } = goal;
        let kind = predicate.kind();
        self.enter_forall_with_assumptions(kind, param_env, |ecx, kind| {
            Ok(match kind {
                ty::PredicateKind::Clause(ty::ClauseKind::Trait(predicate)) => {
                    ecx.compute_trait_goal(Goal { param_env, predicate }).map(|(r, _via)| r)?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::HostEffect(predicate)) => {
                    ecx.compute_host_effect_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::Projection(predicate)) => {
                    ecx.compute_projection_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::TypeOutlives(predicate)) => {
                    ecx.compute_type_outlives_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::RegionOutlives(predicate)) => {
                    ecx.compute_region_outlives_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::ConstArgHasType(ct, ty)) => {
                    ecx.compute_const_arg_has_type_goal(Goal { param_env, predicate: (ct, ty) })?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::UnstableFeature(symbol)) => {
                    ecx.compute_unstable_feature_goal(param_env, symbol)?
                }
                ty::PredicateKind::Subtype(predicate) => {
                    ecx.compute_subtype_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::Coerce(predicate) => {
                    ecx.compute_coerce_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::DynCompatible(trait_def_id) => {
                    ecx.compute_dyn_compatible_goal(trait_def_id)?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::WellFormed(term)) => {
                    ecx.compute_well_formed_goal(Goal { param_env, predicate: term })?
                }
                ty::PredicateKind::Clause(ty::ClauseKind::ConstEvaluatable(ct)) => {
                    ecx.compute_const_evaluatable_goal(Goal { param_env, predicate: ct })?
                }
                ty::PredicateKind::ConstEquate(_, _) => {
                    panic!("ConstEquate should not be emitted when `-Znext-solver` is active")
                }
                ty::PredicateKind::NormalizesTo(predicate) => {
                    ecx.compute_normalizes_to_goal(Goal { param_env, predicate })?
                }
                ty::PredicateKind::Ambiguous => {
                    ecx.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS)?
                }
            })
        })
    }

    // Recursively evaluates all the goals added to this `EvalCtxt` to completion, returning
    // the certainty of all the goals.
    #[instrument(level = "trace", skip(self))]
    pub(super) fn try_evaluate_added_goals(
        &mut self,
    ) -> Result<Certainty, NoSolutionOrRerunNonErased> {
        for _ in 0..FIXPOINT_STEP_LIMIT {
            match self.evaluate_added_goals_step().map_err_to_rerun()? {
                Ok(None) => {}
                Ok(Some(cert)) => return Ok(cert),
                Err(NoSolution) => {
                    self.tainted = Err(NoSolution);
                    return Err(NoSolution.into());
                }
            }
        }

        debug!("try_evaluate_added_goals: encountered overflow");
        Ok(Certainty::overflow(false))
    }

    /// Iterate over all added goals: returning `Ok(Some(_))` in case we can stop rerunning.
    ///
    /// Goals for the next step get directly added to the nested goals of the `EvalCtxt`.
    fn evaluate_added_goals_step(
        &mut self,
    ) -> Result<Option<Certainty>, NoSolutionOrRerunNonErased> {
        // If this loop did not result in any progress, what's our final certainty.
        let mut unchanged_certainty = Some(Certainty::Yes);
        // This mem::take seems super inefficient, given that we push to it again later.
        // Despite that, replacing it has no effect on performance. We tried.
        // (https://github.com/rust-lang/rust/pull/158126)
        for (source, goal, stalled_on) in mem::take(&mut self.nested_goals) {
            // We never handle `NormalizesTo` as a nested goal
            debug_assert!(!matches!(
                goal.predicate.kind().skip_binder(),
                PredicateKind::NormalizesTo(_)
            ));

            let GoalEvaluation { goal, certainty, has_changed, stalled_on } =
                self.evaluate_goal(source, goal, stalled_on)?;
            if has_changed == HasChanged::Yes {
                unchanged_certainty = None;
            }

            match certainty {
                Certainty::Yes => {}
                Certainty::Maybe { .. } => {
                    self.nested_goals.push((source, goal, stalled_on));
                    unchanged_certainty = unchanged_certainty.map(|c| c.and(certainty));
                }
            }
        }

        Ok(unchanged_certainty)
    }

    /// Record impl args in the proof tree for later access by `InspectCandidate`.
    pub(crate) fn record_impl_args(&mut self, impl_args: I::GenericArgs) {
        self.inspect.record_impl_args(self.delegate, self.max_input_universe, impl_args)
    }

    pub(super) fn cx(&self) -> I {
        self.delegate.cx()
    }

    #[instrument(level = "debug", skip(self))]
    pub(super) fn add_goal(
        &mut self,
        source: GoalSource,
        mut goal: Goal<I, I::Predicate>,
    ) -> Result<(), NoSolutionOrRerunNonErased> {
        goal.predicate = self.normalize(
            GoalSource::NormalizeGoal(self.step_kind_for_source(source)),
            goal.param_env,
            ty::Unnormalized::new_wip(goal.predicate),
        )?;
        self.inspect.add_goal(self.delegate, self.max_input_universe, source, goal);

        if let Some(GoalEvaluation { goal, certainty, has_changed: _, stalled_on }) =
            compute_goal_fast_path(self.delegate, goal, self.origin_span)
        {
            match certainty {
                // We're done here
                Certainty::Yes => {}
                Certainty::Maybe(_) => {
                    self.nested_goals.push((source, goal, stalled_on));
                }
            }
        } else {
            self.nested_goals.push((source, goal, None));
        }
        Ok(())
    }

    #[instrument(level = "trace", skip(self, goals))]
    pub(super) fn add_goals(
        &mut self,
        source: GoalSource,
        goals: impl IntoIterator<Item = Goal<I, I::Predicate>>,
    ) -> Result<(), NoSolutionOrRerunNonErased> {
        for goal in goals {
            self.add_goal(source, goal)?;
        }
        Ok(())
    }

    pub(super) fn next_region_var(&mut self) -> Region<I> {
        let region = self.delegate.next_region_infer();
        self.inspect.add_var_value(region);
        region
    }

    pub(super) fn next_ty_infer(&mut self) -> I::Ty {
        let ty = self.delegate.next_ty_infer();
        self.inspect.add_var_value(ty);
        ty
    }

    pub(super) fn next_const_infer(&mut self) -> I::Const {
        let ct = self.delegate.next_const_infer();
        self.inspect.add_var_value(ct);
        ct
    }

    /// Returns a ty infer or a const infer depending on whether `kind` is a `Ty` or `Const`.
    /// If `kind` is an integer inference variable this will still return a ty infer var.
    pub(super) fn next_term_infer_of_alias_kind(
        &mut self,
        alias_term: ty::AliasTerm<I>,
    ) -> I::Term {
        match alias_term.kind {
            ty::AliasTermKind::ProjectionTy { .. }
            | ty::AliasTermKind::InherentTy { .. }
            | ty::AliasTermKind::OpaqueTy { .. }
            | ty::AliasTermKind::FreeTy { .. } => self.next_ty_infer().into(),
            ty::AliasTermKind::FreeConst { .. }
            | ty::AliasTermKind::InherentConst { .. }
            | ty::AliasTermKind::AnonConst { .. }
            | ty::AliasTermKind::ProjectionConst { .. } => self.next_const_infer().into(),
        }
    }

    /// Is the projection predicate is of the form `exists<T> <Ty as Trait>::Assoc = T`.
    ///
    /// This is the case if the `term` does not occur in any other part of the predicate
    /// and is able to name all other placeholder and inference variables.
    #[instrument(level = "trace", skip(self), ret)]
    pub(super) fn term_is_fully_unconstrained(&self, goal: Goal<I, ty::NormalizesTo<I>>) -> bool {
        let universe_of_term = match goal.predicate.term.kind() {
            ty::TermKind::Ty(ty) => {
                if let ty::Infer(ty::TyVar(vid)) = ty.kind() {
                    self.delegate.universe_of_ty(vid).unwrap()
                } else {
                    return false;
                }
            }
            ty::TermKind::Const(ct) => {
                if let ty::ConstKind::Infer(ty::InferConst::Var(vid)) = ct.kind() {
                    self.delegate.universe_of_ct(vid).unwrap()
                } else {
                    return false;
                }
            }
        };

        struct ContainsTermOrNotNameable<'a, D: SolverDelegate<Interner = I>, I: Interner> {
            term: I::Term,
            universe_of_term: ty::UniverseIndex,
            delegate: &'a D,
            cache: HashSet<I::Ty>,
        }

        impl<D: SolverDelegate<Interner = I>, I: Interner> ContainsTermOrNotNameable<'_, D, I> {
            fn check_nameable(&self, universe: ty::UniverseIndex) -> ControlFlow<()> {
                if self.universe_of_term.can_name(universe) {
                    ControlFlow::Continue(())
                } else {
                    ControlFlow::Break(())
                }
            }
        }

        impl<D: SolverDelegate<Interner = I>, I: Interner> TypeVisitor<I>
            for ContainsTermOrNotNameable<'_, D, I>
        {
            type Result = ControlFlow<()>;
            fn visit_ty(&mut self, t: I::Ty) -> Self::Result {
                if self.cache.contains(&t) {
                    return ControlFlow::Continue(());
                }

                match t.kind() {
                    ty::Infer(ty::TyVar(vid)) => {
                        if let ty::TermKind::Ty(term) = self.term.kind()
                            && let ty::Infer(ty::TyVar(term_vid)) = term.kind()
                            && self.delegate.root_ty_var(vid) == self.delegate.root_ty_var(term_vid)
                        {
                            return ControlFlow::Break(());
                        }

                        self.check_nameable(self.delegate.universe_of_ty(vid).unwrap())?;
                    }
                    ty::Placeholder(p) => self.check_nameable(p.universe())?,
                    _ => {
                        if t.has_non_region_infer() || t.has_placeholders() {
                            t.super_visit_with(self)?
                        }
                    }
                }

                assert!(self.cache.insert(t));
                ControlFlow::Continue(())
            }

            fn visit_const(&mut self, c: I::Const) -> Self::Result {
                match c.kind() {
                    ty::ConstKind::Infer(ty::InferConst::Var(vid)) => {
                        if let ty::TermKind::Const(term) = self.term.kind()
                            && let ty::ConstKind::Infer(ty::InferConst::Var(term_vid)) = term.kind()
                            && self.delegate.root_const_var(vid)
                                == self.delegate.root_const_var(term_vid)
                        {
                            return ControlFlow::Break(());
                        }

                        self.check_nameable(self.delegate.universe_of_ct(vid).unwrap())
                    }
                    ty::ConstKind::Placeholder(p) => self.check_nameable(p.universe()),
                    _ => {
                        if c.has_non_region_infer() || c.has_placeholders() {
                            c.super_visit_with(self)
                        } else {
                            ControlFlow::Continue(())
                        }
                    }
                }
            }

            fn visit_predicate(&mut self, p: I::Predicate) -> Self::Result {
                if p.has_non_region_infer() || p.has_placeholders() {
                    p.super_visit_with(self)
                } else {
                    ControlFlow::Continue(())
                }
            }

            fn visit_clauses(&mut self, c: I::Clauses) -> Self::Result {
                if c.has_non_region_infer() || c.has_placeholders() {
                    c.super_visit_with(self)
                } else {
                    ControlFlow::Continue(())
                }
            }
        }

        let mut visitor = ContainsTermOrNotNameable {
            delegate: self.delegate,
            universe_of_term,
            term: goal.predicate.term,
            cache: Default::default(),
        };
        goal.predicate.alias.visit_with(&mut visitor).is_continue()
            && goal.param_env.visit_with(&mut visitor).is_continue()
    }

    pub(super) fn sub_unify_ty_vids_raw(&self, a: ty::TyVid, b: ty::TyVid) {
        self.delegate.sub_unify_ty_vids_raw(a, b)
    }

    #[instrument(level = "trace", skip(self, param_env), ret)]
    pub(super) fn eq<T: Relate<I>>(
        &mut self,
        param_env: I::ParamEnv,
        lhs: T,
        rhs: T,
    ) -> Result<(), NoSolutionOrRerunNonErased> {
        self.relate(param_env, lhs, ty::Variance::Invariant, rhs)
    }

    #[instrument(level = "trace", skip(self, param_env), ret)]
    pub(super) fn sub<T: Relate<I>>(
        &mut self,
        param_env: I::ParamEnv,
        sub: T,
        sup: T,
    ) -> Result<(), NoSolutionOrRerunNonErased> {
        self.relate(param_env, sub, ty::Variance::Covariant, sup)
    }

    #[instrument(level = "trace", skip(self, param_env), ret)]
    pub(super) fn relate<T: Relate<I>>(
        &mut self,
        param_env: I::ParamEnv,
        lhs: T,
        variance: ty::Variance,
        rhs: T,
    ) -> Result<(), NoSolutionOrRerunNonErased> {
        let goals = self.delegate.relate(param_env, lhs, variance, rhs, self.origin_span)?;
        for &goal in goals.iter() {
            let source = match goal.predicate.kind().skip_binder() {
                ty::PredicateKind::Subtype { .. }
                | ty::PredicateKind::Clause(ty::ClauseKind::Projection(..)) => {
                    GoalSource::TypeRelating
                }
                // FIXME(-Znext-solver=coinductive): should these WF goals also be unproductive?
                ty::PredicateKind::Clause(ty::ClauseKind::WellFormed(_)) => GoalSource::Misc,
                p => unreachable!("unexpected nested goal in `relate`: {p:?}"),
            };
            self.add_goal(source, goal)?;
        }
        Ok(())
    }

    /// Equates two values returning the nested goals without adding them
    /// to the nested goals of the `EvalCtxt`.
    ///
    /// If possible, try using `eq` instead which automatically handles nested
    /// goals correctly.
    #[instrument(level = "trace", skip(self, param_env), ret)]
    pub(super) fn eq_and_get_goals<T: Relate<I>>(
        &self,
        param_env: I::ParamEnv,
        lhs: T,
        rhs: T,
    ) -> Result<Vec<Goal<I, I::Predicate>>, NoSolution> {
        Ok(self.delegate.relate(param_env, lhs, ty::Variance::Invariant, rhs, self.origin_span)?)
    }

    pub(super) fn instantiate_binder_with_infer<T: TypeFoldable<I> + Copy>(
        &self,
        value: ty::Binder<I, T>,
    ) -> T {
        self.delegate.instantiate_binder_with_infer(value)
    }

    /// `enter_forall_with_assumptions`, but takes `&mut self` and passes it back through
    /// the callback since it can't be aliased during the call.
    ///
    /// The `param_env` is used to *compute* the assumptions of the binder, not *as* the
    /// assumptions associated with the binder.
    ///
    /// FIXME(inherent_associated_types): fix this?
    pub(super) fn enter_forall_with_assumptions<T: TypeFoldable<I>, U>(
        &mut self,
        value: ty::Binder<I, T>,
        param_env: I::ParamEnv,
        f: impl FnOnce(&mut Self, T) -> U,
    ) -> U {
        self.delegate.enter_forall_without_assumptions(value, |value| {
            let u = self.delegate.universe();
            let assumptions = if self.cx().assumptions_on_binders() {
                self.region_assumptions_for_placeholders_in_universe(value.clone(), u, param_env)
            } else {
                None
            };
            self.delegate.insert_placeholder_assumptions(u, assumptions);
            f(self, value)
        })
    }

    pub(super) fn resolve_vars_if_possible<T>(&self, value: T) -> T
    where
        T: TypeFoldable<I>,
    {
        self.delegate.resolve_vars_if_possible(value)
    }

    pub(super) fn shallow_resolve(&self, ty: I::Ty) -> I::Ty {
        self.delegate.shallow_resolve(ty)
    }

    pub(super) fn eager_resolve_region(&self, r: Region<I>) -> Region<I> {
        if let ty::ReVar(vid) = r.kind() {
            self.delegate.opportunistic_resolve_lt_var(vid)
        } else {
            r
        }
    }

    pub(super) fn fresh_args_for_item(&mut self, def_id: I::DefId) -> I::GenericArgs {
        let args = self.delegate.fresh_args_for_item(def_id);
        for arg in args.iter() {
            self.inspect.add_var_value(arg);
        }
        args
    }

    pub(super) fn register_solver_region_constraint(&self, c: RegionConstraint<I>) {
        self.delegate.register_solver_region_constraint(c);
    }

    pub(super) fn register_ty_outlives(&self, ty: I::Ty, lt: Region<I>) {
        self.delegate.register_ty_outlives(ty, lt, self.origin_span);
    }

    pub(super) fn register_region_outlives(
        &self,
        a: Region<I>,
        b: Region<I>,
        vis: VisibleForLeakCheck,
    ) {
        // `'a: 'b` ==> `'b <= 'a`
        self.delegate.sub_regions(b, a, vis, self.origin_span);
    }

    /// Computes the list of goals required for `arg` to be well-formed
    pub(super) fn well_formed_goals(
        &self,
        param_env: I::ParamEnv,
        term: I::Term,
    ) -> Option<Vec<Goal<I, I::Predicate>>> {
        self.delegate.well_formed_goals(param_env, term)
    }

    pub(super) fn trait_ref_is_knowable(
        &mut self,
        param_env: I::ParamEnv,
        trait_ref: ty::TraitRef<I>,
    ) -> Result<bool, NoSolutionOrRerunNonErased> {
        let delegate = self.delegate;
        let lazily_normalize_ty = |ty| self.structurally_normalize_ty(param_env, ty);
        coherence::trait_ref_is_knowable(&**delegate, trait_ref, lazily_normalize_ty)
            .map(|is_knowable| is_knowable.is_ok())
    }

    pub(super) fn fetch_eligible_assoc_item(
        &self,
        goal_trait_ref: ty::TraitRef<I>,
        trait_assoc_def_id: I::TraitAssocTermId,
        impl_def_id: I::ImplId,
    ) -> FetchEligibleAssocItemResponse<I> {
        self.delegate.fetch_eligible_assoc_item(goal_trait_ref, trait_assoc_def_id, impl_def_id)
    }

    #[instrument(level = "debug", skip(self), ret)]
    pub(super) fn register_hidden_type_in_storage(
        &mut self,
        opaque_type_key: ty::OpaqueTypeKey<I>,
        hidden_ty: I::Ty,
    ) -> Option<I::Ty> {
        self.delegate.register_hidden_type_in_storage(opaque_type_key, hidden_ty, self.origin_span)
    }

    pub(super) fn add_item_bounds_for_hidden_type(
        &mut self,
        opaque_def_id: I::OpaqueTyId,
        opaque_args: I::GenericArgs,
        param_env: I::ParamEnv,
        hidden_ty: I::Ty,
    ) -> Result<(), NoSolutionOrRerunNonErased> {
        let mut goals = Vec::new();
        self.delegate.add_item_bounds_for_hidden_type(
            opaque_def_id,
            opaque_args,
            param_env,
            hidden_ty,
            &mut goals,
        );
        self.add_goals(GoalSource::AliasWellFormed, goals)?;
        Ok(())
    }

    // Try to evaluate a const, or return `None` if the const is too generic.
    // This doesn't mean the const isn't evaluatable, though, and should be treated
    // as an ambiguity rather than no-solution.
    pub(super) fn evaluate_const(
        &mut self,
        param_env: I::ParamEnv,
        alias_const: ty::AliasConst<I>,
    ) -> Result<Option<I::Const>, RerunNonErased> {
        if self.typing_mode().is_erased_not_coherence() {
            match self.opaque_accesses.rerun_always(RerunReason::EvaluateConst)? {}
        }

        Ok(self.delegate.evaluate_const(param_env, alias_const))
    }

    pub(super) fn evaluate_const_and_instantiate_projection_term(
        &mut self,
        param_env: I::ParamEnv,
        projection_term: ty::AliasTerm<I>,
        expected_term: I::Term,
        alias_const: ty::AliasConst<I>,
    ) -> QueryResultOrRerunNonErased<I> {
        match self.evaluate_const(param_env, alias_const)? {
            Some(evaluated) => {
                self.eq(param_env, expected_term, evaluated.into())?;
                self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes)
            }
            None if self.cx().features().generic_const_args() => {
                // HACK(khyperia): calling `resolve_vars_if_possible` here shouldn't be necessary,
                // `try_evaluate_const` calls `resolve_vars_if_possible` already. However, we want
                // to check `has_non_region_infer` against the type with vars resolved (i.e. check
                // if there are vars we failed to resolve), so we need to call it again here.
                // Perhaps we could split EvaluateConstErr::HasGenericsOrInfers into HasGenerics and
                // HasInfers or something, make evaluate_const return that, and make this branch be
                // based on that, rather than checking `has_non_region_infer`.
                if self.resolve_vars_if_possible(alias_const).has_non_region_infer() {
                    self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS)
                } else {
                    // We do not instantiate to the `alias_const` passed in, but rather
                    // `goal.predicate.alias`. The `alias_const` passed in might correspond to the `impl`
                    // form of a constant (with generic arguments corresponding to the impl block),
                    // however, we want to structurally instantiate to the original, non-rebased,
                    // trait `Self` form of the constant (with generic arguments being the trait
                    // `Self` type).
                    self.eq(
                        param_env,
                        projection_term.to_term(self.cx(), ty::IsRigid::Yes),
                        expected_term,
                    )?;
                    self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes)
                }
            }
            None => {
                // Legacy behavior: always treat as ambiguous
                self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS)
            }
        }
    }

    pub(super) fn is_transmutable(
        &mut self,
        src: I::Ty,
        dst: I::Ty,
        assume: I::Const,
    ) -> Result<Certainty, NoSolution> {
        self.delegate.is_transmutable(dst, src, assume)
    }

    pub(super) fn replace_bound_vars<T: TypeFoldable<I>>(
        &self,
        t: T,
        universes: &mut Vec<Option<ty::UniverseIndex>>,
    ) -> T {
        BoundVarReplacer::replace_bound_vars(&**self.delegate, universes, t).0
    }

    pub(super) fn may_use_unstable_feature(
        &mut self,
        param_env: I::ParamEnv,
        symbol: I::Symbol,
    ) -> Result<bool, RerunNonErased> {
        if self.typing_mode().is_erased_not_coherence() {
            match self.opaque_accesses.rerun_always(RerunReason::MayUseUnstableFeature)? {}
        }

        Ok(may_use_unstable_feature(&**self.delegate, param_env, symbol))
    }

    pub(crate) fn opaques_with_sub_unified_hidden_type(
        &self,
        self_ty: I::Ty,
    ) -> Vec<ty::OpaqueAliasTy<I>> {
        if let ty::Infer(ty::TyVar(vid)) = self_ty.kind() {
            self.delegate.opaques_with_sub_unified_hidden_type(vid)
        } else {
            vec![]
        }
    }

    /// To return the constraints of a canonical query to the caller, we canonicalize:
    ///
    /// - `var_values`: a map from bound variables in the canonical goal to
    ///   the values inferred while solving the instantiated goal.
    /// - `external_constraints`: additional constraints which aren't expressible
    ///   using simple unification of inference variables.
    ///
    /// This takes the `shallow_certainty` which represents whether we're confident
    /// that the final result of the current goal only depends on the nested goals.
    ///
    /// In case this is `Certainty::Maybe`, there may still be additional nested goals
    /// or inference constraints required for this candidate to be hold. The candidate
    /// always requires all already added constraints and nested goals.
    #[instrument(level = "trace", skip(self), ret)]
    pub(in crate::solve) fn evaluate_added_goals_and_make_canonical_response(
        &mut self,
        shallow_certainty: Certainty,
    ) -> QueryResultOrRerunNonErased<I> {
        self.inspect.make_canonical_response(shallow_certainty);

        let goals_certainty = self.try_evaluate_added_goals()?;
        assert_eq!(
            self.tainted,
            Ok(()),
            "EvalCtxt is tainted -- nested goals may have been dropped in a \
            previous call to `try_evaluate_added_goals!`"
        );

        let goals_certainty = match self.delegate.cx().assumptions_on_binders() {
            true => {
                let certainty = self.eagerly_handle_placeholders()?;
                certainty.and(goals_certainty)
            }
            false => {
                // We only check for leaks from universes which were entered inside
                // of the query.
                self.delegate.leak_check(self.max_input_universe).map_err(|NoSolution| {
                    trace!("failed the leak check");
                    NoSolution
                })?;

                goals_certainty
            }
        };

        let (certainty, normalization_nested_goals) =
            match (self.current_goal_kind, shallow_certainty) {
                // When normalizing, we've replaced the expected term with an unconstrained
                // inference variable. This means that we dropped information which could
                // have been important. We handle this by instead returning the nested goals
                // to the caller, where they are then handled. We only do so if we do not
                // need to recompute the `NormalizesTo` goal afterwards to avoid repeatedly
                // uplifting its nested goals. This is the case if the `shallow_certainty` is
                // `Certainty::Yes`.
                (CurrentGoalKind::ProjectionComputeAssocTermCandidate, Certainty::Yes) => {
                    let goals = std::mem::take(&mut self.nested_goals);
                    // As we return all ambiguous nested goals, we can ignore the certainty
                    // returned by `self.try_evaluate_added_goals()`.
                    if goals.is_empty() {
                        assert!(matches!(goals_certainty, Certainty::Yes));
                    }
                    (
                        Certainty::Yes,
                        NestedNormalizationGoals(
                            goals.into_iter().map(|(s, g, _)| (s, g)).collect(),
                        ),
                    )
                }
                _ => {
                    let certainty = shallow_certainty.and(goals_certainty);
                    (certainty, NestedNormalizationGoals::empty())
                }
            };

        if let Certainty::Maybe(
            maybe_info @ MaybeInfo {
                cause: MaybeCause::Overflow { keep_constraints: false, .. },
                opaque_types_jank: _,
                stalled_on_coroutines: _,
            },
        ) = certainty
        {
            // If we have overflow, it's probable that we're substituting a type
            // into itself infinitely and any partial substitutions in the query
            // response are probably not useful anyways, so just return an empty
            // query response.
            //
            // This may prevent us from potentially useful inference, e.g.
            // 2 candidates, one ambiguous and one overflow, which both
            // have the same inference constraints.
            //
            // Changing this to retain some constraints in the future
            // won't be a breaking change, so this is good enough for now.
            return Ok(self.make_ambiguous_response_no_constraints(maybe_info));
        }

        let external_constraints =
            self.compute_external_query_constraints(certainty, normalization_nested_goals);
        let (var_values, mut external_constraints) =
            eager_resolve_vars(&**self.delegate, (self.var_values, external_constraints));

        // Remove any trivial or duplicated region constraints once we've resolved regions
        let mut unique = HashSet::default();
        if let ExternalRegionConstraints::Old(r) = &mut external_constraints.region_constraints {
            r.retain(|(outlives, _)| !outlives.is_trivial() && unique.insert(*outlives));
        }

        let canonical = canonicalize_response(
            self.delegate,
            self.max_input_universe,
            Response {
                var_values,
                certainty,
                external_constraints: self.cx().mk_external_constraints(external_constraints),
            },
        );

        Ok(canonical)
    }

    /// Constructs a totally unconstrained, ambiguous response to a goal.
    ///
    /// Take care when using this, since often it's useful to respond with
    /// ambiguity but return constrained variables to guide inference.
    pub(in crate::solve) fn make_ambiguous_response_no_constraints(
        &self,
        maybe: MaybeInfo,
    ) -> CanonicalResponse<I> {
        response_no_constraints_raw(
            self.cx(),
            self.max_input_universe,
            self.var_kinds,
            Certainty::Maybe(maybe),
        )
    }

    /// Computes the region constraints and *new* opaque types registered when
    /// proving a goal.
    ///
    /// If an opaque was already constrained before proving this goal, then the
    /// external constraints do not need to record that opaque, since if it is
    /// further constrained by inference, that will be passed back in the var
    /// values.
    #[instrument(level = "trace", skip(self), ret)]
    fn compute_external_query_constraints(
        &self,
        certainty: Certainty,
        normalization_nested_goals: NestedNormalizationGoals<I>,
    ) -> ExternalConstraintsData<I> {
        // We only return region constraints once the certainty is `Yes`. This
        // is necessary as we may drop nested goals on ambiguity, which may result
        // in unconstrained inference variables in the region constraints. It also
        // prevents us from emitting duplicate region constraints, avoiding some
        // unnecessary work. This slightly weakens the leak check in case it uses
        // region constraints from an ambiguous nested goal. This is tested in both
        // `tests/ui/higher-ranked/leak-check/leak-check-in-selection-5-ambig.rs` and
        // `tests/ui/higher-ranked/leak-check/leak-check-in-selection-6-ambig-unify.rs`.
        let region_constraints = if self.cx().assumptions_on_binders() {
            ExternalRegionConstraints::NextGen(if let Certainty::Yes = certainty {
                self.delegate.get_solver_region_constraint()
            } else {
                RegionConstraint::new_true()
            })
        } else {
            ExternalRegionConstraints::Old(if let Certainty::Yes = certainty {
                self.delegate.make_deduplicated_region_constraints()
            } else {
                vec![]
            })
        };

        // We only return *newly defined* opaque types from canonical queries.
        //
        // Constraints for any existing opaque types are already tracked by changes
        // to the `var_values`.
        let opaque_types = self
            .delegate
            .clone_opaque_types_added_since(self.initial_opaque_types_storage_num_entries);

        if self.typing_mode().is_erased_not_coherence() {
            assert!(opaque_types.is_empty());
        }

        ExternalConstraintsData { region_constraints, opaque_types, normalization_nested_goals }
    }

    pub(super) fn normalize<T: TypeFoldable<I>>(
        &mut self,
        source: GoalSource,
        param_env: I::ParamEnv,
        value: ty::Unnormalized<I, T>,
    ) -> Result<T, NoSolutionOrRerunNonErased> {
        let value = self.delegate.resolve_vars_if_possible(value.skip_normalization());

        if !self.cx().renormalize_rigid_aliases() && !value.has_non_rigid_aliases() {
            return Ok(value);
        }

        // To drop the mutable borrow of self early.
        let infcx = self.delegate.deref();
        let mut folder = NormalizationFolder::new(infcx, vec![], |alias_term| {
            let infer_term = self.next_term_infer_of_alias_kind(alias_term);
            let pred = ty::ProjectionPredicate { projection_term: alias_term, term: infer_term };
            let goal = Goal::new(self.cx(), param_env, pred);
            self.inspect.add_goal(self.delegate, self.max_input_universe, source, goal);
            let GoalEvaluation { goal, certainty, has_changed: _, stalled_on } =
                self.evaluate_goal(source, goal, None)?;
            let normalization_was_ambiguous = match certainty {
                Certainty::Yes => NormalizationWasAmbiguous::No,
                Certainty::Maybe(_) => {
                    self.nested_goals.push((source, goal, stalled_on));
                    NormalizationWasAmbiguous::Yes
                }
            };

            Ok((self.resolve_vars_if_possible(infer_term), normalization_was_ambiguous))
        });
        value.try_fold_with(&mut folder)
    }
}

#[derive(Debug)]
enum RerunDecision {
    Yes,
    No,
    EagerlyPropagateToParent,
}

#[tracing::instrument(ret)]
fn should_rerun_after_erased_canonicalization<I: Interner>(
    AccessedOpaques { reason: _, rerun }: AccessedOpaques<I>,
    original_typing_mode: TypingMode<I>,
    parent_opaque_types: &[(OpaqueTypeKey<I>, I::Ty)],
) -> RerunDecision {
    let parent_opaque_def_ids = parent_opaque_types.iter().map(|(key, _)| key.def_id.into());
    let opaque_in_storage = |opaques: I::LocalDefIds, def_ids: SmallCopyList<_>| {
        if def_ids.as_ref().is_empty() {
            RerunDecision::No
        } else if opaques
            .iter()
            .chain(parent_opaque_def_ids)
            .any(|opaque| def_ids.as_ref().contains(&opaque))
        {
            RerunDecision::Yes
        } else {
            RerunDecision::No
        }
    };
    let any_opaque_has_infer_as_hidden = || {
        if parent_opaque_types.iter().any(|(_, ty)| ty.is_ty_var()) {
            RerunDecision::Yes
        } else {
            RerunDecision::No
        }
    };

    match (rerun, original_typing_mode) {
        // =============================
        (RerunCondition::Never, _) => RerunDecision::No,
        // =============================
        (_, TypingMode::ErasedNotCoherence(MayBeErased)) => RerunDecision::EagerlyPropagateToParent,
        // =============================
        // In coherence, we never switch to erased mode, so we will never register anything
        // in the rerun state, so we should've taken the first branch of this match
        (_, TypingMode::Coherence) => unreachable!(),
        // =============================
        (RerunCondition::Always, _) => RerunDecision::Yes,
        // =============================
        (
            RerunCondition::OpaqueInStorage(..),
            TypingMode::PostAnalysis | TypingMode::Codegen | TypingMode::Reflection,
        ) => RerunDecision::Yes,
        (
            RerunCondition::OpaqueInStorage(defids),
            TypingMode::PostBorrowck { defined_opaque_types: opaques }
            | TypingMode::Typeck { defining_opaque_types_and_generators: opaques }
            | TypingMode::PostTypeckUntilBorrowck { defining_opaque_types: opaques },
        ) => opaque_in_storage(opaques, defids),
        // =============================
        (RerunCondition::AnyOpaqueHasInferAsHidden, TypingMode::Typeck { .. }) => {
            any_opaque_has_infer_as_hidden()
        }
        (
            RerunCondition::AnyOpaqueHasInferAsHidden,
            TypingMode::PostBorrowck { .. }
            | TypingMode::PostAnalysis
            | TypingMode::Codegen
            | TypingMode::Reflection
            | TypingMode::PostTypeckUntilBorrowck { .. },
        ) => RerunDecision::No,
        // =============================
        (
            RerunCondition::OpaqueInStorageOrAnyOpaqueHasInferAsHidden(_),
            TypingMode::PostAnalysis | TypingMode::Codegen | TypingMode::Reflection,
        ) => RerunDecision::Yes,
        (
            RerunCondition::OpaqueInStorageOrAnyOpaqueHasInferAsHidden(defids),
            TypingMode::Typeck { defining_opaque_types_and_generators: opaques },
        ) => {
            if let RerunDecision::Yes = any_opaque_has_infer_as_hidden() {
                RerunDecision::Yes
            } else if let RerunDecision::Yes = opaque_in_storage(opaques, defids) {
                RerunDecision::Yes
            } else {
                RerunDecision::No
            }
        }
        (
            RerunCondition::OpaqueInStorageOrAnyOpaqueHasInferAsHidden(defids),
            TypingMode::PostBorrowck { defined_opaque_types: opaques }
            | TypingMode::PostTypeckUntilBorrowck { defining_opaque_types: opaques },
        ) => opaque_in_storage(opaques, defids),
    }
}

/// Do not call this directly, use the `tcx` query instead.
pub fn evaluate_root_goal_for_proof_tree_raw_provider<
    D: SolverDelegate<Interner = I>,
    I: Interner,
>(
    cx: I,
    canonical_goal: CanonicalInput<I>,
    root_depth: usize,
) -> (QueryResult<I>, I::Probe) {
    let mut inspect = inspect::ProofTreeBuilder::new();
    let (canonical_result, accessed_opaques) = SearchGraph::<D>::evaluate_root_goal_for_proof_tree(
        cx,
        root_depth,
        canonical_goal,
        &mut inspect,
    );
    let final_revision = inspect.unwrap();

    assert!(!accessed_opaques.might_rerun());
    (canonical_result, cx.mk_probe(final_revision))
}

/// Evaluate a goal to build a proof tree.
///
/// This is a copy of [EvalCtxt::evaluate_goal_raw] which avoids relying on the
/// [EvalCtxt] and uses a separate cache.
pub(super) fn evaluate_root_goal_for_proof_tree<D: SolverDelegate<Interner = I>, I: Interner>(
    delegate: &D,
    goal: Goal<I, I::Predicate>,
    origin_span: I::Span,
    root_depth: usize,
) -> (Result<NestedNormalizationGoals<I>, NoSolution>, inspect::GoalEvaluation<I>) {
    let opaque_types = delegate.clone_opaque_types_lookup_table();
    let (goal, opaque_types) = eager_resolve_vars(&**delegate, (goal, opaque_types));
    let typing_mode = delegate.typing_mode_raw().assert_not_erased();

    let (orig_values, canonical_goal) =
        canonicalize_goal(delegate, goal, &opaque_types, typing_mode.into());

    let (canonical_result, final_revision) =
        delegate.cx().evaluate_root_goal_for_proof_tree_raw(canonical_goal, root_depth);

    let proof_tree = inspect::GoalEvaluation {
        uncanonicalized_goal: goal,
        orig_values,
        final_revision,
        result: canonical_result,
    };

    let response = match canonical_result {
        Err(e) => return (Err(e), proof_tree),
        Ok(response) => response,
    };

    let (normalization_nested_goals, _certainty) = instantiate_and_apply_query_response(
        delegate,
        goal.param_env,
        &proof_tree.orig_values,
        response,
        origin_span,
    );

    (Ok(normalization_nested_goals), proof_tree)
}