rework implementation for inherent impls for builtin types

Author: lcnr

compiler/rustc_feature/src/builtin_attrs.rs

@@ -636,6 +636,14 @@ pub const BUILTIN_ATTRIBUTES: &[BuiltinAttribute] = &[
         template!(Word), ErrorFollowing,
         "#[rustc_pass_by_value] is used to mark types that must be passed by value instead of reference."
     ),
+    rustc_attr!(
+        rustc_coherence_is_core, AttributeType::CrateLevel, template!(Word), ErrorFollowing,
+        "#![rustc_coherence_is_core] allows inherent methods on builtin types, only intended to be used in `core`."
+    ),
+    rustc_attr!(
+        rustc_allow_incoherent_impl, AttributeType::Normal, template!(Word), ErrorFollowing,
+        "#[rustc_allow_incoherent_impl] has to be added to all impl items of an incoherent inherent impl."
+    ),
     BuiltinAttribute {
         name: sym::rustc_diagnostic_item,
         type_: Normal,

compiler/rustc_metadata/src/rmeta/decoder.rs

@@ -95,6 +95,11 @@ crate struct CrateMetadata {
     /// FIXME: Used only from queries and can use query cache,
     /// so pre-decoding can probably be avoided.
     trait_impls: FxHashMap<(u32, DefIndex), Lazy<[(DefIndex, Option<SimplifiedType>)]>>,
+    /// Inherent impls which do not follow the normal coherence rules.
+    ///
+    /// These can be introduces using either `#![rustc_coherence_is_core]`
+    /// or `#[rustc_allow_incoherent_impl]`.
+    incoherent_impls: FxHashMap<SimplifiedType, Lazy<[DefIndex]>>,
     /// Proc macro descriptions for this crate, if it's a proc macro crate.
     raw_proc_macros: Option<&'static [ProcMacro]>,
     /// Source maps for code from the crate.
@@ -1327,17 +1332,25 @@ impl<'a, 'tcx> CrateMetadataRef<'a> {
 
     /// Decodes all trait impls in the crate (for rustdoc).
     fn get_trait_impls(self) -> impl Iterator<Item = (DefId, DefId, Option<SimplifiedType>)> + 'a {
-        self.cdata.trait_impls.iter().flat_map(move |((trait_cnum_raw, trait_index), impls)| {
+        self.cdata.trait_impls.iter().flat_map(move |(&(trait_cnum_raw, trait_index), impls)| {
             let trait_def_id = DefId {
-                krate: self.cnum_map[CrateNum::from_u32(*trait_cnum_raw)],
-                index: *trait_index,
+                krate: self.cnum_map[CrateNum::from_u32(trait_cnum_raw)],
+                index: trait_index,
             };
             impls.decode(self).map(move |(impl_index, simplified_self_ty)| {
                 (trait_def_id, self.local_def_id(impl_index), simplified_self_ty)
             })
         })
     }
 
+    fn get_incoherent_impls(self, tcx: TyCtxt<'tcx>, simp: SimplifiedType) -> &'tcx [DefId] {
+        if let Some(impls) = self.cdata.incoherent_impls.get(&simp) {
+            tcx.arena.alloc_from_iter(impls.decode(self).map(|idx| self.local_def_id(idx)))
+        } else {
+            &[]
+        }
+    }
+
     fn get_implementations_of_trait(
         self,
         tcx: TyCtxt<'tcx>,
@@ -1754,6 +1767,11 @@ impl CrateMetadata {
             .decode((&blob, sess))
             .map(|trait_impls| (trait_impls.trait_id, trait_impls.impls))
             .collect();
+        let incoherent_impls = root
+            .incoherent_impls
+            .decode((&blob, sess))
+            .map(|incoherent_impls| (incoherent_impls.self_ty, incoherent_impls.impls))
+            .collect();
         let alloc_decoding_state =
             AllocDecodingState::new(root.interpret_alloc_index.decode(&blob).collect());
         let dependencies = Lock::new(cnum_map.iter().cloned().collect());
@@ -1766,6 +1784,7 @@ impl CrateMetadata {
             blob,
             root,
             trait_impls,
+            incoherent_impls,
             raw_proc_macros,
             source_map_import_info: OnceCell::new(),
             def_path_hash_map,

compiler/rustc_metadata/src/rmeta/decoder/cstore_impl.rs

@@ -81,30 +81,42 @@ macro_rules! provide {
 // small trait to work around different signature queries all being defined via
 // the macro above.
 trait IntoArgs {
-    fn into_args(self) -> (DefId, DefId);
+    type Other;
+    fn into_args(self) -> (DefId, Self::Other);
 }
 
 impl IntoArgs for DefId {
-    fn into_args(self) -> (DefId, DefId) {
-        (self, self)
+    type Other = ();
+    fn into_args(self) -> (DefId, ()) {
+        (self, ())
     }
 }
 
 impl IntoArgs for CrateNum {
-    fn into_args(self) -> (DefId, DefId) {
-        (self.as_def_id(), self.as_def_id())
+    type Other = ();
+    fn into_args(self) -> (DefId, ()) {
+        (self.as_def_id(), ())
     }
 }
 
 impl IntoArgs for (CrateNum, DefId) {
+    type Other = DefId;
     fn into_args(self) -> (DefId, DefId) {
         (self.0.as_def_id(), self.1)
     }
 }
 
 impl<'tcx> IntoArgs for ty::InstanceDef<'tcx> {
-    fn into_args(self) -> (DefId, DefId) {
-        (self.def_id(), self.def_id())
+    type Other = ();
+    fn into_args(self) -> (DefId, ()) {
+        (self.def_id(), ())
+    }
+}
+
+impl IntoArgs for (CrateNum, SimplifiedType) {
+    type Other = SimplifiedType;
+    fn into_args(self) -> (DefId, SimplifiedType) {
+        (self.0.as_def_id(), self.1)
     }
 }
 
@@ -199,6 +211,7 @@ provide! { <'tcx> tcx, def_id, other, cdata,
 
     traits_in_crate => { tcx.arena.alloc_from_iter(cdata.get_traits()) }
     implementations_of_trait => { cdata.get_implementations_of_trait(tcx, other) }
+    crate_incoherent_impls => { cdata.get_incoherent_impls(tcx, other) }
 
     dep_kind => {
         let r = *cdata.dep_kind.lock();
@@ -371,7 +384,6 @@ pub(in crate::rmeta) fn provide(providers: &mut Providers) {
                 .alloc_slice(&CStore::from_tcx(tcx).crate_dependencies_in_postorder(LOCAL_CRATE))
         },
         crates: |tcx, ()| tcx.arena.alloc_from_iter(CStore::from_tcx(tcx).crates_untracked()),
-
         ..*providers
     };
 }

compiler/rustc_metadata/src/rmeta/encoder.rs

@@ -2,8 +2,9 @@ use crate::rmeta::def_path_hash_map::DefPathHashMapRef;
 use crate::rmeta::table::{FixedSizeEncoding, TableBuilder};
 use crate::rmeta::*;
 
+use rustc_data_structures::fingerprint::Fingerprint;
 use rustc_data_structures::fx::{FxHashMap, FxIndexSet};
-use rustc_data_structures::stable_hasher::StableHasher;
+use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
 use rustc_data_structures::sync::{join, par_iter, Lrc, ParallelIterator};
 use rustc_hir as hir;
 use rustc_hir::def::DefKind;
@@ -578,6 +579,7 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
     }
 
     fn encode_crate_root(&mut self) -> Lazy<CrateRoot<'tcx>> {
+        let tcx = self.tcx;
         let mut i = self.position();
 
         // Encode the crate deps
@@ -623,8 +625,9 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
         let impls = self.encode_impls();
         let impls_bytes = self.position() - i;
 
-        let tcx = self.tcx;
-
+        i = self.position();
+        let incoherent_impls = self.encode_incoherent_impls();
+        let incoherent_impls_bytes = self.position() - i;
         // Encode MIR.
         i = self.position();
         self.encode_mir();
@@ -734,6 +737,7 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
             source_map,
             traits,
             impls,
+            incoherent_impls,
             exported_symbols,
             interpret_alloc_index,
             tables,
@@ -762,6 +766,7 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
             eprintln!("      source_map bytes: {}", source_map_bytes);
             eprintln!("          traits bytes: {}", traits_bytes);
             eprintln!("           impls bytes: {}", impls_bytes);
+            eprintln!("incoherent_impls bytes: {}", incoherent_impls_bytes);
             eprintln!("    exp. symbols bytes: {}", exported_symbols_bytes);
             eprintln!("  def-path table bytes: {}", def_path_table_bytes);
             eprintln!(" def-path hashes bytes: {}", def_path_hash_map_bytes);
@@ -1813,6 +1818,33 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
         self.lazy(&all_impls)
     }
 
+    fn encode_incoherent_impls(&mut self) -> Lazy<[IncoherentImpls]> {
+        debug!("EncodeContext::encode_traits_and_impls()");
+        empty_proc_macro!(self);
+        let tcx = self.tcx;
+        let mut ctx = tcx.create_stable_hashing_context();
+        let mut all_impls: Vec<_> = tcx.crate_inherent_impls(()).incoherent_impls.iter().collect();
+        all_impls.sort_by_cached_key(|&(&simp, _)| {
+            let mut hasher = StableHasher::new();
+            simp.hash_stable(&mut ctx, &mut hasher);
+            hasher.finish::<Fingerprint>();
+        });
+        let all_impls: Vec<_> = all_impls
+            .into_iter()
+            .map(|(&simp, impls)| {
+                let mut impls: Vec<_> =
+                    impls.into_iter().map(|def_id| def_id.local_def_index).collect();
+                impls.sort_by_cached_key(|&local_def_index| {
+                    tcx.hir().def_path_hash(LocalDefId { local_def_index })
+                });
+
+                IncoherentImpls { self_ty: simp, impls: self.lazy(impls) }
+            })
+            .collect();
+
+        self.lazy(&all_impls)
+    }
+
     // Encodes all symbols exported from this crate into the metadata.
     //
     // This pass is seeded off the reachability list calculated in the

compiler/rustc_metadata/src/rmeta/mod.rs

@@ -212,6 +212,7 @@ crate struct CrateRoot<'tcx> {
     foreign_modules: Lazy<[ForeignModule]>,
     traits: Lazy<[DefIndex]>,
     impls: Lazy<[TraitImpls]>,
+    incoherent_impls: Lazy<[IncoherentImpls]>,
     interpret_alloc_index: Lazy<[u32]>,
     proc_macro_data: Option<ProcMacroData>,
 
@@ -251,6 +252,12 @@ crate struct TraitImpls {
     impls: Lazy<[(DefIndex, Option<SimplifiedType>)]>,
 }
 
+#[derive(MetadataEncodable, MetadataDecodable)]
+crate struct IncoherentImpls {
+    self_ty: SimplifiedType,
+    impls: Lazy<[DefIndex]>,
+}
+
 /// Define `LazyTables` and `TableBuilders` at the same time.
 macro_rules! define_tables {
     ($($name:ident: Table<$IDX:ty, $T:ty>),+ $(,)?) => {

compiler/rustc_middle/src/hir/map/mod.rs

@@ -579,6 +579,10 @@ impl<'hir> Map<'hir> {
         self.attrs(CRATE_HIR_ID)
     }
 
+    pub fn rustc_coherence_is_core(self) -> bool {
+        self.krate_attrs().iter().any(|attr| attr.has_name(sym::rustc_coherence_is_core))
+    }
+
     pub fn get_module(self, module: LocalDefId) -> (&'hir Mod<'hir>, Span, HirId) {
         let hir_id = HirId::make_owner(module);
         match self.tcx.hir_owner(module).map(|o| o.node) {

compiler/rustc_middle/src/query/mod.rs

@@ -684,6 +684,10 @@ rustc_queries! {
         separate_provide_extern
     }
 
+    query incoherent_impls(key: SimplifiedType) -> &'tcx [DefId] {
+        desc { |tcx| "collecting all inherent impls for `{:?}`", key }
+    }
+
     /// The result of unsafety-checking this `LocalDefId`.
     query unsafety_check_result(key: LocalDefId) -> &'tcx mir::UnsafetyCheckResult {
         desc { |tcx| "unsafety-checking `{}`", tcx.def_path_str(key.to_def_id()) }
@@ -1469,6 +1473,15 @@ rustc_queries! {
         separate_provide_extern
     }
 
+    /// Collects all incoherent impls for the given crate and type.
+    ///
+    /// Do not call this directly, but instead use the `incoherent_impls` query.
+    /// This query is only used to get the data necessary for that query.
+    query crate_incoherent_impls(key: (CrateNum, SimplifiedType)) -> &'tcx [DefId] {
+        desc { |tcx| "collecting all impls for a type in a crate" }
+        separate_provide_extern
+    }
+
     query is_dllimport_foreign_item(def_id: DefId) -> bool {
         desc { |tcx| "is_dllimport_foreign_item({})", tcx.def_path_str(def_id) }
     }

compiler/rustc_middle/src/ty/impls_ty.rs

@@ -4,6 +4,7 @@
 use crate::middle::region;
 use crate::mir;
 use crate::ty;
+use crate::ty::fast_reject::SimplifiedType;
 use rustc_data_structures::fingerprint::Fingerprint;
 use rustc_data_structures::fx::FxHashMap;
 use rustc_data_structures::stable_hasher::HashingControls;
@@ -55,6 +56,18 @@ where
     }
 }
 
+impl<'a> ToStableHashKey<StableHashingContext<'a>> for SimplifiedType {
+    type KeyType = Fingerprint;
+
+    #[inline]
+    fn to_stable_hash_key(&self, hcx: &StableHashingContext<'a>) -> Fingerprint {
+        let mut hasher = StableHasher::new();
+        let mut hcx: StableHashingContext<'a> = hcx.clone();
+        self.hash_stable(&mut hcx, &mut hasher);
+        hasher.finish()
+    }
+}
+
 impl<'a, 'tcx> HashStable<StableHashingContext<'a>> for ty::subst::GenericArg<'tcx> {
     fn hash_stable(&self, hcx: &mut StableHashingContext<'a>, hasher: &mut StableHasher) {
         self.unpack().hash_stable(hcx, hasher);

compiler/rustc_middle/src/ty/mod.rs

@@ -25,6 +25,7 @@ use crate::middle::privacy::AccessLevels;
 use crate::mir::{Body, GeneratorLayout};
 use crate::traits::{self, Reveal};
 use crate::ty;
+use crate::ty::fast_reject::SimplifiedType;
 use crate::ty::subst::{GenericArg, InternalSubsts, Subst, SubstsRef};
 use crate::ty::util::Discr;
 use rustc_ast as ast;
@@ -2335,6 +2336,7 @@ pub fn provide(providers: &mut ty::query::Providers) {
     super::middle::provide(providers);
     *providers = ty::query::Providers {
         trait_impls_of: trait_def::trait_impls_of_provider,
+        incoherent_impls: trait_def::incoherent_impls_provider,
         type_uninhabited_from: inhabitedness::type_uninhabited_from,
         const_param_default: consts::const_param_default,
         vtable_allocation: vtable::vtable_allocation_provider,
@@ -2350,6 +2352,7 @@ pub fn provide(providers: &mut ty::query::Providers) {
 #[derive(Clone, Debug, Default, HashStable)]
 pub struct CrateInherentImpls {
     pub inherent_impls: LocalDefIdMap<Vec<DefId>>,
+    pub incoherent_impls: FxHashMap<SimplifiedType, Vec<LocalDefId>>,
 }
 
 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, TyEncodable, HashStable)]

compiler/rustc_middle/src/ty/trait_def.rs

@@ -2,9 +2,11 @@ use crate::traits::specialization_graph;
 use crate::ty::fast_reject::{self, SimplifiedType, TreatParams};
 use crate::ty::fold::TypeFoldable;
 use crate::ty::{Ident, Ty, TyCtxt};
+use hir::def_id::LOCAL_CRATE;
 use rustc_hir as hir;
 use rustc_hir::def_id::DefId;
 use rustc_hir::definitions::DefPathHash;
+use std::iter;
 
 use rustc_data_structures::fx::FxIndexMap;
 use rustc_errors::ErrorGuaranteed;
@@ -257,3 +259,19 @@ pub(super) fn trait_impls_of_provider(tcx: TyCtxt<'_>, trait_id: DefId) -> Trait
 
     impls
 }
+
+// Query provider for `incoherent_impls`.
+#[instrument(level = "debug", skip(tcx))]
+pub(super) fn incoherent_impls_provider(tcx: TyCtxt<'_>, simp: SimplifiedType) -> &[DefId] {
+    let mut impls = Vec::new();
+
+    for cnum in iter::once(LOCAL_CRATE).chain(tcx.crates(()).iter().copied()) {
+        for &impl_def_id in tcx.crate_incoherent_impls((cnum, simp)) {
+            impls.push(impl_def_id)
+        }
+    }
+
+    debug!(?impls);
+
+    tcx.arena.alloc_slice(&impls)
+}