//! The next-generation trait solver, currently still WIP. //! //! As a user of rust, you can use `-Znext-solver` to enable the new trait solver. //! //! As a developer of rustc, you shouldn't be using the new trait //! solver without asking the trait-system-refactor-initiative, but it can //! be enabled with `InferCtxtBuilder::with_next_trait_solver`. This will //! ensure that trait solving using that inference context will be routed //! to the new trait solver. //! //! For a high-level overview of how this solver works, check out the relevant //! section of the rustc-dev-guide. //! //! FIXME(@lcnr): Write that section. If you read this before then ask me //! about it on zulip. use rustc_hir::def_id::DefId; use rustc_infer::infer::canonical::{Canonical, CanonicalVarValues}; use rustc_infer::traits::query::NoSolution; use rustc_macros::extension; use rustc_middle::infer::canonical::CanonicalVarInfos; use rustc_middle::traits::solve::{ CanonicalResponse, Certainty, ExternalConstraintsData, Goal, GoalSource, QueryResult, Response, }; use rustc_middle::ty::{self, AliasRelationDirection, Ty, TyCtxt, UniverseIndex}; use rustc_middle::ty::{ CoercePredicate, RegionOutlivesPredicate, SubtypePredicate, TypeOutlivesPredicate, }; mod alias_relate; mod assembly; mod eval_ctxt; mod fulfill; pub mod inspect; mod normalize; mod normalizes_to; mod project_goals; mod search_graph; mod trait_goals; pub use eval_ctxt::{EvalCtxt, GenerateProofTree, InferCtxtEvalExt, InferCtxtSelectExt}; pub use fulfill::FulfillmentCtxt; pub(crate) use normalize::deeply_normalize_for_diagnostics; pub use normalize::{deeply_normalize, deeply_normalize_with_skipped_universes}; /// How many fixpoint iterations we should attempt inside of the solver before bailing /// with overflow. /// /// We previously used `tcx.recursion_limit().0.checked_ilog2().unwrap_or(0)` for this. /// However, it feels unlikely that uncreasing the recursion limit by a power of two /// to get one more itereation is every useful or desirable. We now instead used a constant /// here. If there ever ends up some use-cases where a bigger number of fixpoint iterations /// is required, we can add a new attribute for that or revert this to be dependant on the /// recursion limit again. However, this feels very unlikely. const FIXPOINT_STEP_LIMIT: usize = 8; #[derive(Debug, Clone, Copy)] enum SolverMode { /// Ordinary trait solving, using everywhere except for coherence. Normal, /// Trait solving during coherence. There are a few notable differences /// between coherence and ordinary trait solving. /// /// Most importantly, trait solving during coherence must not be incomplete, /// i.e. return `Err(NoSolution)` for goals for which a solution exists. /// This means that we must not make any guesses or arbitrary choices. Coherence, } #[derive(Debug, Copy, Clone, PartialEq, Eq)] enum GoalEvaluationKind { Root, Nested, } #[extension(trait CanonicalResponseExt)] impl<'tcx> Canonical<'tcx, Response<'tcx>> { fn has_no_inference_or_external_constraints(&self) -> bool { self.value.external_constraints.region_constraints.is_empty() && self.value.var_values.is_identity() && self.value.external_constraints.opaque_types.is_empty() } } impl<'a, 'tcx> EvalCtxt<'a, 'tcx> { #[instrument(level = "debug", skip(self))] fn compute_type_outlives_goal( &mut self, goal: Goal<'tcx, TypeOutlivesPredicate<'tcx>>, ) -> QueryResult<'tcx> { let ty::OutlivesPredicate(ty, lt) = goal.predicate; self.register_ty_outlives(ty, lt); self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } #[instrument(level = "debug", skip(self))] fn compute_region_outlives_goal( &mut self, goal: Goal<'tcx, RegionOutlivesPredicate<'tcx>>, ) -> QueryResult<'tcx> { let ty::OutlivesPredicate(a, b) = goal.predicate; self.register_region_outlives(a, b); self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } #[instrument(level = "debug", skip(self))] fn compute_coerce_goal( &mut self, goal: Goal<'tcx, CoercePredicate<'tcx>>, ) -> QueryResult<'tcx> { self.compute_subtype_goal(Goal { param_env: goal.param_env, predicate: SubtypePredicate { a_is_expected: false, a: goal.predicate.a, b: goal.predicate.b, }, }) } #[instrument(level = "debug", skip(self))] fn compute_subtype_goal( &mut self, goal: Goal<'tcx, SubtypePredicate<'tcx>>, ) -> QueryResult<'tcx> { if goal.predicate.a.is_ty_var() && goal.predicate.b.is_ty_var() { self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS) } else { self.sub(goal.param_env, goal.predicate.a, goal.predicate.b)?; self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } } fn compute_object_safe_goal(&mut self, trait_def_id: DefId) -> QueryResult<'tcx> { if self.tcx().check_is_object_safe(trait_def_id) { self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } else { Err(NoSolution) } } #[instrument(level = "debug", skip(self))] fn compute_well_formed_goal( &mut self, goal: Goal<'tcx, ty::GenericArg<'tcx>>, ) -> QueryResult<'tcx> { match self.well_formed_goals(goal.param_env, goal.predicate) { Some(goals) => { self.add_goals(GoalSource::Misc, goals); self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } None => self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS), } } #[instrument(level = "debug", skip(self))] fn compute_const_evaluatable_goal( &mut self, Goal { param_env, predicate: ct }: Goal<'tcx, ty::Const<'tcx>>, ) -> QueryResult<'tcx> { match ct.kind() { ty::ConstKind::Unevaluated(uv) => { // We never return `NoSolution` here as `try_const_eval_resolve` emits an // error itself when failing to evaluate, so emitting an additional fulfillment // error in that case is unnecessary noise. This may change in the future once // evaluation failures are allowed to impact selection, e.g. generic const // expressions in impl headers or `where`-clauses. // FIXME(generic_const_exprs): Implement handling for generic // const expressions here. if let Some(_normalized) = self.try_const_eval_resolve(param_env, uv, ct.ty()) { self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } else { self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS) } } ty::ConstKind::Infer(_) => { self.evaluate_added_goals_and_make_canonical_response(Certainty::AMBIGUOUS) } ty::ConstKind::Placeholder(_) | ty::ConstKind::Value(_) | ty::ConstKind::Error(_) => { self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } // We can freely ICE here as: // - `Param` gets replaced with a placeholder during canonicalization // - `Bound` cannot exist as we don't have a binder around the self Type // - `Expr` is part of `feature(generic_const_exprs)` and is not implemented yet ty::ConstKind::Param(_) | ty::ConstKind::Bound(_, _) | ty::ConstKind::Expr(_) => { bug!("unexpect const kind: {:?}", ct) } } } #[instrument(level = "debug", skip(self), ret)] fn compute_const_arg_has_type_goal( &mut self, goal: Goal<'tcx, (ty::Const<'tcx>, Ty<'tcx>)>, ) -> QueryResult<'tcx> { let (ct, ty) = goal.predicate; self.eq(goal.param_env, ct.ty(), ty)?; self.evaluate_added_goals_and_make_canonical_response(Certainty::Yes) } } impl<'tcx> EvalCtxt<'_, 'tcx> { #[instrument(level = "debug", skip(self, goals))] fn add_goals( &mut self, source: GoalSource, goals: impl IntoIterator>>, ) { for goal in goals { self.add_goal(source, goal); } } /// Try to merge multiple possible ways to prove a goal, if that is not possible returns `None`. /// /// In this case we tend to flounder and return ambiguity by calling `[EvalCtxt::flounder]`. #[instrument(level = "debug", skip(self), ret)] fn try_merge_responses( &mut self, responses: &[CanonicalResponse<'tcx>], ) -> Option> { if responses.is_empty() { return None; } // FIXME(-Znext-solver): We should instead try to find a `Certainty::Yes` response with // a subset of the constraints that all the other responses have. let one = responses[0]; if responses[1..].iter().all(|&resp| resp == one) { return Some(one); } responses .iter() .find(|response| { response.value.certainty == Certainty::Yes && response.has_no_inference_or_external_constraints() }) .copied() } /// If we fail to merge responses we flounder and return overflow or ambiguity. #[instrument(level = "debug", skip(self), ret)] fn flounder(&mut self, responses: &[CanonicalResponse<'tcx>]) -> QueryResult<'tcx> { if responses.is_empty() { return Err(NoSolution); } let Certainty::Maybe(maybe_cause) = responses.iter().fold(Certainty::AMBIGUOUS, |certainty, response| { certainty.unify_with(response.value.certainty) }) else { bug!("expected flounder response to be ambiguous") }; Ok(self.make_ambiguous_response_no_constraints(maybe_cause)) } /// Normalize a type for when it is structurally matched on. /// /// This function is necessary in nearly all cases before matching on a type. /// Not doing so is likely to be incomplete and therefore unsound during /// coherence. #[instrument(level = "debug", skip(self, param_env), ret)] fn structurally_normalize_ty( &mut self, param_env: ty::ParamEnv<'tcx>, ty: Ty<'tcx>, ) -> Result, NoSolution> { if let ty::Alias(..) = ty.kind() { let normalized_ty = self.next_ty_infer(); let alias_relate_goal = Goal::new( self.tcx(), param_env, ty::PredicateKind::AliasRelate( ty.into(), normalized_ty.into(), AliasRelationDirection::Equate, ), ); self.add_goal(GoalSource::Misc, alias_relate_goal); self.try_evaluate_added_goals()?; Ok(self.resolve_vars_if_possible(normalized_ty)) } else { Ok(ty) } } } fn response_no_constraints_raw<'tcx>( tcx: TyCtxt<'tcx>, max_universe: UniverseIndex, variables: CanonicalVarInfos<'tcx>, certainty: Certainty, ) -> CanonicalResponse<'tcx> { Canonical { max_universe, variables, value: Response { var_values: CanonicalVarValues::make_identity(tcx, variables), // FIXME: maybe we should store the "no response" version in tcx, like // we do for tcx.types and stuff. external_constraints: tcx.mk_external_constraints(ExternalConstraintsData::default()), certainty, }, defining_opaque_types: Default::default(), } }