//! Trait Resolution. See the [rustc dev guide] for more information on how this works. //! //! [rustc dev guide]: https://rustc-dev-guide.rust-lang.org/traits/resolution.html pub mod auto_trait; pub(crate) mod coherence; pub mod const_evaluatable; mod engine; pub mod error_reporting; mod fulfill; pub mod misc; mod object_safety; pub mod outlives_bounds; pub mod project; pub mod query; #[allow(hidden_glob_reexports)] mod select; mod specialize; mod structural_match; mod structural_normalize; #[allow(hidden_glob_reexports)] mod util; pub mod vtable; pub mod wf; use crate::infer::outlives::env::OutlivesEnvironment; use crate::infer::{InferCtxt, TyCtxtInferExt}; use crate::traits::error_reporting::TypeErrCtxtExt as _; use crate::traits::query::evaluate_obligation::InferCtxtExt as _; use rustc_errors::ErrorGuaranteed; use rustc_middle::query::Providers; use rustc_middle::ty::fold::TypeFoldable; use rustc_middle::ty::visit::{TypeVisitable, TypeVisitableExt}; use rustc_middle::ty::{self, ToPredicate, Ty, TyCtxt, TypeFolder, TypeSuperVisitable}; use rustc_middle::ty::{GenericArgs, GenericArgsRef}; use rustc_span::def_id::DefId; use rustc_span::Span; use std::fmt::Debug; use std::ops::ControlFlow; pub(crate) use self::project::{needs_normalization, BoundVarReplacer, PlaceholderReplacer}; pub use self::FulfillmentErrorCode::*; pub use self::ImplSource::*; pub use self::ObligationCauseCode::*; pub use self::SelectionError::*; pub use self::coherence::{add_placeholder_note, orphan_check, overlapping_impls}; pub use self::coherence::{OrphanCheckErr, OverlapResult}; pub use self::engine::{ObligationCtxt, TraitEngineExt}; pub use self::fulfill::{FulfillmentContext, PendingPredicateObligation}; pub use self::object_safety::astconv_object_safety_violations; pub use self::object_safety::is_vtable_safe_method; pub use self::object_safety::MethodViolationCode; pub use self::object_safety::ObjectSafetyViolation; pub use self::project::NormalizeExt; pub use self::project::{normalize_inherent_projection, normalize_projection_type}; pub use self::select::{EvaluationCache, SelectionCache, SelectionContext}; pub use self::select::{EvaluationResult, IntercrateAmbiguityCause, OverflowError}; pub use self::specialize::specialization_graph::FutureCompatOverlapError; pub use self::specialize::specialization_graph::FutureCompatOverlapErrorKind; pub use self::specialize::{ specialization_graph, translate_args, translate_args_with_cause, OverlapError, }; pub use self::structural_match::search_for_structural_match_violation; pub use self::structural_normalize::StructurallyNormalizeExt; pub use self::util::elaborate; pub use self::util::{ check_args_compatible, supertrait_def_ids, supertraits, transitive_bounds, transitive_bounds_that_define_assoc_item, SupertraitDefIds, }; pub use self::util::{expand_trait_aliases, TraitAliasExpander}; pub use self::util::{get_vtable_index_of_object_method, impl_item_is_final, upcast_choices}; pub use rustc_infer::traits::*; /// Whether to skip the leak check, as part of a future compatibility warning step. /// /// The "default" for skip-leak-check corresponds to the current /// behavior (do not skip the leak check) -- not the behavior we are /// transitioning into. #[derive(Copy, Clone, PartialEq, Eq, Debug, Default)] pub enum SkipLeakCheck { Yes, #[default] No, } impl SkipLeakCheck { fn is_yes(self) -> bool { self == SkipLeakCheck::Yes } } /// The mode that trait queries run in. #[derive(Copy, Clone, PartialEq, Eq, Debug)] pub enum TraitQueryMode { /// Standard/un-canonicalized queries get accurate /// spans etc. passed in and hence can do reasonable /// error reporting on their own. Standard, /// Canonical queries get dummy spans and hence /// must generally propagate errors to /// pre-canonicalization callsites. Canonical, } /// Creates predicate obligations from the generic bounds. #[instrument(level = "debug", skip(cause, param_env))] pub fn predicates_for_generics<'tcx>( cause: impl Fn(usize, Span) -> ObligationCause<'tcx>, param_env: ty::ParamEnv<'tcx>, generic_bounds: ty::InstantiatedPredicates<'tcx>, ) -> impl Iterator> { generic_bounds.into_iter().enumerate().map(move |(idx, (clause, span))| Obligation { cause: cause(idx, span), recursion_depth: 0, param_env, predicate: clause.as_predicate(), }) } /// Determines whether the type `ty` is known to meet `bound` and /// returns true if so. Returns false if `ty` either does not meet /// `bound` or is not known to meet bound (note that this is /// conservative towards *no impl*, which is the opposite of the /// `evaluate` methods). pub fn type_known_to_meet_bound_modulo_regions<'tcx>( infcx: &InferCtxt<'tcx>, param_env: ty::ParamEnv<'tcx>, ty: Ty<'tcx>, def_id: DefId, ) -> bool { let trait_ref = ty::TraitRef::new(infcx.tcx, def_id, [ty]); pred_known_to_hold_modulo_regions(infcx, param_env, trait_ref) } /// FIXME(@lcnr): this function doesn't seem right and shouldn't exist? /// /// Ping me on zulip if you want to use this method and need help with finding /// an appropriate replacement. #[instrument(level = "debug", skip(infcx, param_env, pred), ret)] fn pred_known_to_hold_modulo_regions<'tcx>( infcx: &InferCtxt<'tcx>, param_env: ty::ParamEnv<'tcx>, pred: impl ToPredicate<'tcx>, ) -> bool { let obligation = Obligation::new(infcx.tcx, ObligationCause::dummy(), param_env, pred); let result = infcx.evaluate_obligation_no_overflow(&obligation); debug!(?result); if result.must_apply_modulo_regions() { true } else if result.may_apply() { // Sometimes obligations are ambiguous because the recursive evaluator // is not smart enough, so we fall back to fulfillment when we're not certain // that an obligation holds or not. Even still, we must make sure that // the we do no inference in the process of checking this obligation. let goal = infcx.resolve_vars_if_possible((obligation.predicate, obligation.param_env)); infcx.probe(|_| { let ocx = ObligationCtxt::new(infcx); ocx.register_obligation(obligation); let errors = ocx.select_all_or_error(); match errors.as_slice() { // Only known to hold if we did no inference. [] => infcx.shallow_resolve(goal) == goal, errors => { debug!(?errors); false } } }) } else { false } } #[instrument(level = "debug", skip(tcx, elaborated_env))] fn do_normalize_predicates<'tcx>( tcx: TyCtxt<'tcx>, cause: ObligationCause<'tcx>, elaborated_env: ty::ParamEnv<'tcx>, predicates: Vec>, ) -> Result>, ErrorGuaranteed> { let span = cause.span; // FIXME. We should really... do something with these region // obligations. But this call just continues the older // behavior (i.e., doesn't cause any new bugs), and it would // take some further refactoring to actually solve them. In // particular, we would have to handle implied bounds // properly, and that code is currently largely confined to // regionck (though I made some efforts to extract it // out). -nmatsakis // // @arielby: In any case, these obligations are checked // by wfcheck anyway, so I'm not sure we have to check // them here too, and we will remove this function when // we move over to lazy normalization *anyway*. let infcx = tcx.infer_ctxt().ignoring_regions().build(); let predicates = match fully_normalize(&infcx, cause, elaborated_env, predicates) { Ok(predicates) => predicates, Err(errors) => { let reported = infcx.err_ctxt().report_fulfillment_errors(&errors); return Err(reported); } }; debug!("do_normalize_predicates: normalized predicates = {:?}", predicates); // We can use the `elaborated_env` here; the region code only // cares about declarations like `'a: 'b`. let outlives_env = OutlivesEnvironment::new(elaborated_env); // FIXME: It's very weird that we ignore region obligations but apparently // still need to use `resolve_regions` as we need the resolved regions in // the normalized predicates. let errors = infcx.resolve_regions(&outlives_env); if !errors.is_empty() { tcx.sess.delay_span_bug( span, format!("failed region resolution while normalizing {elaborated_env:?}: {errors:?}"), ); } match infcx.fully_resolve(predicates) { Ok(predicates) => Ok(predicates), Err(fixup_err) => { // If we encounter a fixup error, it means that some type // variable wound up unconstrained. I actually don't know // if this can happen, and I certainly don't expect it to // happen often, but if it did happen it probably // represents a legitimate failure due to some kind of // unconstrained variable. // // @lcnr: Let's still ICE here for now. I want a test case // for that. span_bug!( span, "inference variables in normalized parameter environment: {}", fixup_err ); } } } // FIXME: this is gonna need to be removed ... /// Normalizes the parameter environment, reporting errors if they occur. #[instrument(level = "debug", skip(tcx))] pub fn normalize_param_env_or_error<'tcx>( tcx: TyCtxt<'tcx>, unnormalized_env: ty::ParamEnv<'tcx>, cause: ObligationCause<'tcx>, ) -> ty::ParamEnv<'tcx> { // I'm not wild about reporting errors here; I'd prefer to // have the errors get reported at a defined place (e.g., // during typeck). Instead I have all parameter // environments, in effect, going through this function // and hence potentially reporting errors. This ensures of // course that we never forget to normalize (the // alternative seemed like it would involve a lot of // manual invocations of this fn -- and then we'd have to // deal with the errors at each of those sites). // // In any case, in practice, typeck constructs all the // parameter environments once for every fn as it goes, // and errors will get reported then; so outside of type inference we // can be sure that no errors should occur. let mut predicates: Vec<_> = util::elaborate( tcx, unnormalized_env.caller_bounds().into_iter().map(|predicate| { if tcx.features().generic_const_exprs { return predicate; } struct ConstNormalizer<'tcx>(TyCtxt<'tcx>); impl<'tcx> TypeFolder> for ConstNormalizer<'tcx> { fn interner(&self) -> TyCtxt<'tcx> { self.0 } fn fold_const(&mut self, c: ty::Const<'tcx>) -> ty::Const<'tcx> { // While it is pretty sus to be evaluating things with an empty param env, it // should actually be okay since without `feature(generic_const_exprs)` the only // const arguments that have a non-empty param env are array repeat counts. These // do not appear in the type system though. c.eval(self.0, ty::ParamEnv::empty()) } } // This whole normalization step is a hack to work around the fact that // `normalize_param_env_or_error` is fundamentally broken from using an // unnormalized param env with a trait solver that expects the param env // to be normalized. // // When normalizing the param env we can end up evaluating obligations // that have been normalized but can only be proven via a where clause // which is still in its unnormalized form. example: // // Attempting to prove `T: Trait<::Assoc>` in a param env // with a `T: Trait<::Assoc>` where clause will fail because // we first normalize obligations before proving them so we end up proving // `T: Trait`. Since lazy normalization is not implemented equating `u8` // with `::Assoc` fails outright so we incorrectly believe that // we cannot prove `T: Trait`. // // The same thing is true for const generics- attempting to prove // `T: Trait` with the same thing as a where clauses // will fail. After normalization we may be attempting to prove `T: Trait<4>` with // the unnormalized where clause `T: Trait`. In order // for the obligation to hold `4` must be equal to `ConstKind::Unevaluated(...)` // but as we do not have lazy norm implemented, equating the two consts fails outright. // // Ideally we would not normalize consts here at all but it is required for backwards // compatibility. Eventually when lazy norm is implemented this can just be removed. // We do not normalize types here as there is no backwards compatibility requirement // for us to do so. // // FIXME(-Ztrait-solver=next): remove this hack since we have deferred projection equality predicate.fold_with(&mut ConstNormalizer(tcx)) }), ) .collect(); debug!("normalize_param_env_or_error: elaborated-predicates={:?}", predicates); let elaborated_env = ty::ParamEnv::new(tcx.mk_clauses(&predicates), unnormalized_env.reveal()); // HACK: we are trying to normalize the param-env inside *itself*. The problem is that // normalization expects its param-env to be already normalized, which means we have // a circularity. // // The way we handle this is by normalizing the param-env inside an unnormalized version // of the param-env, which means that if the param-env contains unnormalized projections, // we'll have some normalization failures. This is unfortunate. // // Lazy normalization would basically handle this by treating just the // normalizing-a-trait-ref-requires-itself cycles as evaluation failures. // // Inferred outlives bounds can create a lot of `TypeOutlives` predicates for associated // types, so to make the situation less bad, we normalize all the predicates *but* // the `TypeOutlives` predicates first inside the unnormalized parameter environment, and // then we normalize the `TypeOutlives` bounds inside the normalized parameter environment. // // This works fairly well because trait matching does not actually care about param-env // TypeOutlives predicates - these are normally used by regionck. let outlives_predicates: Vec<_> = predicates .extract_if(|predicate| { matches!(predicate.kind().skip_binder(), ty::ClauseKind::TypeOutlives(..)) }) .collect(); debug!( "normalize_param_env_or_error: predicates=(non-outlives={:?}, outlives={:?})", predicates, outlives_predicates ); let Ok(non_outlives_predicates) = do_normalize_predicates(tcx, cause.clone(), elaborated_env, predicates) else { // An unnormalized env is better than nothing. debug!("normalize_param_env_or_error: errored resolving non-outlives predicates"); return elaborated_env; }; debug!("normalize_param_env_or_error: non-outlives predicates={:?}", non_outlives_predicates); // Not sure whether it is better to include the unnormalized TypeOutlives predicates // here. I believe they should not matter, because we are ignoring TypeOutlives param-env // predicates here anyway. Keeping them here anyway because it seems safer. let outlives_env = non_outlives_predicates.iter().chain(&outlives_predicates).cloned(); let outlives_env = ty::ParamEnv::new(tcx.mk_clauses_from_iter(outlives_env), unnormalized_env.reveal()); let Ok(outlives_predicates) = do_normalize_predicates(tcx, cause, outlives_env, outlives_predicates) else { // An unnormalized env is better than nothing. debug!("normalize_param_env_or_error: errored resolving outlives predicates"); return elaborated_env; }; debug!("normalize_param_env_or_error: outlives predicates={:?}", outlives_predicates); let mut predicates = non_outlives_predicates; predicates.extend(outlives_predicates); debug!("normalize_param_env_or_error: final predicates={:?}", predicates); ty::ParamEnv::new(tcx.mk_clauses(&predicates), unnormalized_env.reveal()) } /// Normalize a type and process all resulting obligations, returning any errors. /// /// FIXME(-Ztrait-solver=next): This should be replaced by `At::deeply_normalize` /// which has the same behavior with the new solver. Because using a separate /// fulfillment context worsens caching in the old solver, `At::deeply_normalize` /// is still lazy with the old solver as it otherwise negatively impacts perf. #[instrument(skip_all)] pub fn fully_normalize<'tcx, T>( infcx: &InferCtxt<'tcx>, cause: ObligationCause<'tcx>, param_env: ty::ParamEnv<'tcx>, value: T, ) -> Result>> where T: TypeFoldable>, { let ocx = ObligationCtxt::new(infcx); debug!(?value); let normalized_value = ocx.normalize(&cause, param_env, value); debug!(?normalized_value); debug!("select_all_or_error start"); let errors = ocx.select_all_or_error(); if !errors.is_empty() { return Err(errors); } debug!("select_all_or_error complete"); let resolved_value = infcx.resolve_vars_if_possible(normalized_value); debug!(?resolved_value); Ok(resolved_value) } /// Normalizes the predicates and checks whether they hold in an empty environment. If this /// returns true, then either normalize encountered an error or one of the predicates did not /// hold. Used when creating vtables to check for unsatisfiable methods. pub fn impossible_predicates<'tcx>(tcx: TyCtxt<'tcx>, predicates: Vec>) -> bool { debug!("impossible_predicates(predicates={:?})", predicates); let infcx = tcx.infer_ctxt().build(); let param_env = ty::ParamEnv::reveal_all(); let ocx = ObligationCtxt::new(&infcx); let predicates = ocx.normalize(&ObligationCause::dummy(), param_env, predicates); for predicate in predicates { let obligation = Obligation::new(tcx, ObligationCause::dummy(), param_env, predicate); ocx.register_obligation(obligation); } let errors = ocx.select_all_or_error(); let result = !errors.is_empty(); debug!("impossible_predicates = {:?}", result); result } fn subst_and_check_impossible_predicates<'tcx>( tcx: TyCtxt<'tcx>, key: (DefId, GenericArgsRef<'tcx>), ) -> bool { debug!("subst_and_check_impossible_predicates(key={:?})", key); let mut predicates = tcx.predicates_of(key.0).instantiate(tcx, key.1).predicates; // Specifically check trait fulfillment to avoid an error when trying to resolve // associated items. if let Some(trait_def_id) = tcx.trait_of_item(key.0) { let trait_ref = ty::TraitRef::from_method(tcx, trait_def_id, key.1); predicates.push(ty::Binder::dummy(trait_ref).to_predicate(tcx)); } predicates.retain(|predicate| !predicate.has_param()); let result = impossible_predicates(tcx, predicates); debug!("subst_and_check_impossible_predicates(key={:?}) = {:?}", key, result); result } /// Checks whether a trait's associated item is impossible to reference on a given impl. /// /// This only considers predicates that reference the impl's generics, and not /// those that reference the method's generics. fn is_impossible_associated_item( tcx: TyCtxt<'_>, (impl_def_id, trait_item_def_id): (DefId, DefId), ) -> bool { struct ReferencesOnlyParentGenerics<'tcx> { tcx: TyCtxt<'tcx>, generics: &'tcx ty::Generics, trait_item_def_id: DefId, } impl<'tcx> ty::TypeVisitor> for ReferencesOnlyParentGenerics<'tcx> { type BreakTy = (); fn visit_ty(&mut self, t: Ty<'tcx>) -> ControlFlow { // If this is a parameter from the trait item's own generics, then bail if let ty::Param(param) = t.kind() && let param_def_id = self.generics.type_param(param, self.tcx).def_id && self.tcx.parent(param_def_id) == self.trait_item_def_id { return ControlFlow::Break(()); } t.super_visit_with(self) } fn visit_region(&mut self, r: ty::Region<'tcx>) -> ControlFlow { if let ty::ReEarlyBound(param) = r.kind() && let param_def_id = self.generics.region_param(¶m, self.tcx).def_id && self.tcx.parent(param_def_id) == self.trait_item_def_id { return ControlFlow::Break(()); } ControlFlow::Continue(()) } fn visit_const(&mut self, ct: ty::Const<'tcx>) -> ControlFlow { if let ty::ConstKind::Param(param) = ct.kind() && let param_def_id = self.generics.const_param(¶m, self.tcx).def_id && self.tcx.parent(param_def_id) == self.trait_item_def_id { return ControlFlow::Break(()); } ct.super_visit_with(self) } } let generics = tcx.generics_of(trait_item_def_id); let predicates = tcx.predicates_of(trait_item_def_id); let impl_trait_ref = tcx .impl_trait_ref(impl_def_id) .expect("expected impl to correspond to trait") .instantiate_identity(); let param_env = tcx.param_env(impl_def_id); let mut visitor = ReferencesOnlyParentGenerics { tcx, generics, trait_item_def_id }; let predicates_for_trait = predicates.predicates.iter().filter_map(|(pred, span)| { pred.visit_with(&mut visitor).is_continue().then(|| { Obligation::new( tcx, ObligationCause::dummy_with_span(*span), param_env, ty::EarlyBinder::bind(*pred).instantiate(tcx, impl_trait_ref.args), ) }) }); let infcx = tcx.infer_ctxt().ignoring_regions().build(); for obligation in predicates_for_trait { // Ignore overflow error, to be conservative. if let Ok(result) = infcx.evaluate_obligation(&obligation) && !result.may_apply() { return true; } } false } pub fn provide(providers: &mut Providers) { object_safety::provide(providers); vtable::provide(providers); *providers = Providers { specialization_graph_of: specialize::specialization_graph_provider, specializes: specialize::specializes, subst_and_check_impossible_predicates, check_tys_might_be_eq: misc::check_tys_might_be_eq, is_impossible_associated_item, ..*providers }; }