Split crate_hash from index_hir.

Author: cjgillot

compiler/rustc_middle/src/arena.rs

@@ -91,7 +91,7 @@ macro_rules! arena_types {
             [] predicates: rustc_middle::ty::PredicateInner<$tcx>,
 
             // HIR query types
-            [few] indexed_hir: rustc_middle::hir::map::IndexedHir<$tcx>,
+            [few] indexed_hir: rustc_middle::hir::IndexedHir<$tcx>,
             [few] hir_definitions: rustc_hir::definitions::Definitions,
             [] hir_owner: rustc_middle::hir::Owner<$tcx>,
             [] hir_owner_nodes: rustc_middle::hir::OwnerNodes<$tcx>,

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

@@ -2,21 +2,19 @@ use crate::arena::Arena;
 use crate::hir::map::{Entry, HirOwnerData, Map};
 use crate::hir::{Owner, OwnerNodes, ParentedNode};
 use crate::ich::StableHashingContext;
-use crate::middle::cstore::CrateStore;
 use rustc_data_structures::fingerprint::Fingerprint;
 use rustc_data_structures::fx::FxHashMap;
 use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
-use rustc_data_structures::svh::Svh;
 use rustc_hir as hir;
+use rustc_hir::def_id::LocalDefId;
 use rustc_hir::def_id::CRATE_DEF_INDEX;
-use rustc_hir::def_id::{LocalDefId, LOCAL_CRATE};
-use rustc_hir::definitions::{self, DefPathHash};
+use rustc_hir::definitions;
 use rustc_hir::intravisit::{self, NestedVisitorMap, Visitor};
 use rustc_hir::*;
 use rustc_index::vec::{Idx, IndexVec};
-use rustc_session::{CrateDisambiguator, Session};
+use rustc_session::Session;
 use rustc_span::source_map::SourceMap;
-use rustc_span::{Span, Symbol, DUMMY_SP};
+use rustc_span::{Span, DUMMY_SP};
 
 use std::iter::repeat;
 
@@ -40,10 +38,6 @@ pub(super) struct NodeCollector<'a, 'hir> {
     definitions: &'a definitions::Definitions,
 
     hcx: StableHashingContext<'a>,
-
-    // We are collecting HIR hashes here so we can compute the
-    // crate hash from them later on.
-    hir_body_nodes: Vec<(DefPathHash, Fingerprint)>,
 }
 
 fn insert_vec_map<K: Idx, V: Clone>(map: &mut IndexVec<K, Option<V>>, k: K, v: V) {
@@ -58,34 +52,13 @@ fn insert_vec_map<K: Idx, V: Clone>(map: &mut IndexVec<K, Option<V>>, k: K, v: V
 
 fn hash_body(
     hcx: &mut StableHashingContext<'_>,
-    def_path_hash: DefPathHash,
     item_like: impl for<'a> HashStable<StableHashingContext<'a>>,
-    hir_body_nodes: &mut Vec<(DefPathHash, Fingerprint)>,
 ) -> Fingerprint {
-    let hash = {
-        let mut stable_hasher = StableHasher::new();
-        hcx.while_hashing_hir_bodies(true, |hcx| {
-            item_like.hash_stable(hcx, &mut stable_hasher);
-        });
-        stable_hasher.finish()
-    };
-    hir_body_nodes.push((def_path_hash, hash));
-    hash
-}
-
-fn upstream_crates(cstore: &dyn CrateStore) -> Vec<(Symbol, Fingerprint, Svh)> {
-    let mut upstream_crates: Vec<_> = cstore
-        .crates_untracked()
-        .iter()
-        .map(|&cnum| {
-            let name = cstore.crate_name_untracked(cnum);
-            let disambiguator = cstore.crate_disambiguator_untracked(cnum).to_fingerprint();
-            let hash = cstore.crate_hash_untracked(cnum);
-            (name, disambiguator, hash)
-        })
-        .collect();
-    upstream_crates.sort_unstable_by_key(|&(name, dis, _)| (name.as_str(), dis));
-    upstream_crates
+    let mut stable_hasher = StableHasher::new();
+    hcx.while_hashing_hir_bodies(true, |hcx| {
+        item_like.hash_stable(hcx, &mut stable_hasher);
+    });
+    stable_hasher.finish()
 }
 
 impl<'a, 'hir> NodeCollector<'a, 'hir> {
@@ -96,11 +69,6 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
         definitions: &'a definitions::Definitions,
         mut hcx: StableHashingContext<'a>,
     ) -> NodeCollector<'a, 'hir> {
-        let root_mod_def_path_hash =
-            definitions.def_path_hash(LocalDefId { local_def_index: CRATE_DEF_INDEX });
-
-        let mut hir_body_nodes = Vec::new();
-
         let hash = {
             let Crate {
                 ref item,
@@ -120,7 +88,7 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
                 attrs: _,
             } = *krate;
 
-            hash_body(&mut hcx, root_mod_def_path_hash, item, &mut hir_body_nodes)
+            hash_body(&mut hcx, item)
         };
 
         let mut collector = NodeCollector {
@@ -131,7 +99,6 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
             current_dep_node_owner: LocalDefId { local_def_index: CRATE_DEF_INDEX },
             definitions,
             hcx,
-            hir_body_nodes,
             map: (0..definitions.def_index_count())
                 .map(|_| HirOwnerData { signature: None, with_bodies: None })
                 .collect(),
@@ -147,53 +114,13 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
 
     pub(super) fn finalize_and_compute_crate_hash(
         mut self,
-        crate_disambiguator: CrateDisambiguator,
-        cstore: &dyn CrateStore,
-        commandline_args_hash: u64,
-    ) -> (IndexVec<LocalDefId, HirOwnerData<'hir>>, Svh) {
+    ) -> IndexVec<LocalDefId, HirOwnerData<'hir>> {
         // Insert bodies into the map
         for (id, body) in self.krate.bodies.iter() {
             let bodies = &mut self.map[id.hir_id.owner].with_bodies.as_mut().unwrap().bodies;
             assert!(bodies.insert(id.hir_id.local_id, body).is_none());
         }
-
-        self.hir_body_nodes.sort_unstable_by_key(|bn| bn.0);
-
-        let node_hashes = self.hir_body_nodes.iter().fold(
-            Fingerprint::ZERO,
-            |combined_fingerprint, &(def_path_hash, fingerprint)| {
-                combined_fingerprint.combine(def_path_hash.0.combine(fingerprint))
-            },
-        );
-
-        let upstream_crates = upstream_crates(cstore);
-
-        // We hash the final, remapped names of all local source files so we
-        // don't have to include the path prefix remapping commandline args.
-        // If we included the full mapping in the SVH, we could only have
-        // reproducible builds by compiling from the same directory. So we just
-        // hash the result of the mapping instead of the mapping itself.
-        let mut source_file_names: Vec<_> = self
-            .source_map
-            .files()
-            .iter()
-            .filter(|source_file| source_file.cnum == LOCAL_CRATE)
-            .map(|source_file| source_file.name_hash)
-            .collect();
-
-        source_file_names.sort_unstable();
-
-        let crate_hash_input = (
-            ((node_hashes, upstream_crates), source_file_names),
-            (commandline_args_hash, crate_disambiguator.to_fingerprint()),
-        );
-
-        let mut stable_hasher = StableHasher::new();
-        crate_hash_input.hash_stable(&mut self.hcx, &mut stable_hasher);
-        let crate_hash: Fingerprint = stable_hasher.finish();
-
-        let svh = Svh::new(crate_hash.to_smaller_hash());
-        (self.map, svh)
+        self.map
     }
 
     fn insert_entry(&mut self, id: HirId, entry: Entry<'hir>, hash: Fingerprint) {
@@ -294,10 +221,7 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
         f: F,
     ) {
         let prev_owner = self.current_dep_node_owner;
-
-        let def_path_hash = self.definitions.def_path_hash(dep_node_owner);
-
-        let hash = hash_body(&mut self.hcx, def_path_hash, item_like, &mut self.hir_body_nodes);
+        let hash = hash_body(&mut self.hcx, item_like);
 
         self.current_dep_node_owner = dep_node_owner;
         f(self, hash);

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

@@ -1,8 +1,11 @@
 use self::collector::NodeCollector;
 
-use crate::hir::{Owner, OwnerNodes};
+use crate::hir::{HirOwnerData, IndexedHir};
+use crate::middle::cstore::CrateStore;
 use crate::ty::TyCtxt;
 use rustc_ast as ast;
+use rustc_data_structures::fingerprint::Fingerprint;
+use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
 use rustc_data_structures::svh::Svh;
 use rustc_hir::def::{DefKind, Res};
 use rustc_hir::def_id::{CrateNum, DefId, LocalDefId, CRATE_DEF_INDEX, LOCAL_CRATE};
@@ -11,7 +14,6 @@ use rustc_hir::intravisit;
 use rustc_hir::intravisit::Visitor;
 use rustc_hir::itemlikevisit::ItemLikeVisitor;
 use rustc_hir::*;
-use rustc_index::vec::IndexVec;
 use rustc_span::hygiene::MacroKind;
 use rustc_span::source_map::Spanned;
 use rustc_span::symbol::{kw, Ident, Symbol};
@@ -86,20 +88,6 @@ fn is_body_owner<'hir>(node: Node<'hir>, hir_id: HirId) -> bool {
     }
 }
 
-#[derive(Debug)]
-pub(super) struct HirOwnerData<'hir> {
-    pub(super) signature: Option<&'hir Owner<'hir>>,
-    pub(super) with_bodies: Option<&'hir mut OwnerNodes<'hir>>,
-}
-
-#[derive(Debug)]
-pub struct IndexedHir<'hir> {
-    /// The SVH of the local crate.
-    pub crate_hash: Svh,
-
-    pub(super) map: IndexVec<LocalDefId, HirOwnerData<'hir>>,
-}
-
 #[derive(Copy, Clone)]
 pub struct Map<'hir> {
     pub(super) tcx: TyCtxt<'hir>,
@@ -935,19 +923,82 @@ pub(super) fn index_hir<'tcx>(tcx: TyCtxt<'tcx>, cnum: CrateNum) -> &'tcx Indexe
 
     let _prof_timer = tcx.sess.prof.generic_activity("build_hir_map");
 
-    let (map, crate_hash) = {
+    let map = {
         let hcx = tcx.create_stable_hashing_context();
 
         let mut collector =
             NodeCollector::root(tcx.sess, &**tcx.arena, tcx.untracked_crate, &tcx.definitions, hcx);
         intravisit::walk_crate(&mut collector, tcx.untracked_crate);
 
-        let crate_disambiguator = tcx.sess.local_crate_disambiguator();
-        let cmdline_args = tcx.sess.opts.dep_tracking_hash(true);
-        collector.finalize_and_compute_crate_hash(crate_disambiguator, &*tcx.cstore, cmdline_args)
+        collector.finalize_and_compute_crate_hash()
     };
 
-    tcx.arena.alloc(IndexedHir { crate_hash, map })
+    tcx.arena.alloc(IndexedHir { map })
+}
+
+pub(super) fn crate_hash(tcx: TyCtxt<'_>, crate_num: CrateNum) -> Svh {
+    let mut hir_body_nodes: Vec<_> = tcx
+        .index_hir(crate_num)
+        .map
+        .iter_enumerated()
+        .filter_map(|(def_id, hod)| {
+            let def_path_hash = tcx.definitions.def_path_hash(def_id);
+            let hash = hod.with_bodies.as_ref()?.hash;
+            Some((def_path_hash, hash))
+        })
+        .collect();
+    hir_body_nodes.sort_unstable_by_key(|bn| bn.0);
+
+    let node_hashes = hir_body_nodes.iter().fold(
+        Fingerprint::ZERO,
+        |combined_fingerprint, &(def_path_hash, fingerprint)| {
+            combined_fingerprint.combine(def_path_hash.0.combine(fingerprint))
+        },
+    );
+
+    let upstream_crates = upstream_crates(&*tcx.cstore);
+
+    // We hash the final, remapped names of all local source files so we
+    // don't have to include the path prefix remapping commandline args.
+    // If we included the full mapping in the SVH, we could only have
+    // reproducible builds by compiling from the same directory. So we just
+    // hash the result of the mapping instead of the mapping itself.
+    let mut source_file_names: Vec<_> = tcx
+        .sess
+        .source_map()
+        .files()
+        .iter()
+        .filter(|source_file| source_file.cnum == LOCAL_CRATE)
+        .map(|source_file| source_file.name_hash)
+        .collect();
+
+    source_file_names.sort_unstable();
+
+    let mut hcx = tcx.create_stable_hashing_context();
+    let mut stable_hasher = StableHasher::new();
+    node_hashes.hash_stable(&mut hcx, &mut stable_hasher);
+    upstream_crates.hash_stable(&mut hcx, &mut stable_hasher);
+    source_file_names.hash_stable(&mut hcx, &mut stable_hasher);
+    tcx.sess.opts.dep_tracking_hash(true).hash_stable(&mut hcx, &mut stable_hasher);
+    tcx.sess.local_crate_disambiguator().to_fingerprint().hash_stable(&mut hcx, &mut stable_hasher);
+
+    let crate_hash: Fingerprint = stable_hasher.finish();
+    Svh::new(crate_hash.to_smaller_hash())
+}
+
+fn upstream_crates(cstore: &dyn CrateStore) -> Vec<(Symbol, Fingerprint, Svh)> {
+    let mut upstream_crates: Vec<_> = cstore
+        .crates_untracked()
+        .iter()
+        .map(|&cnum| {
+            let name = cstore.crate_name_untracked(cnum);
+            let disambiguator = cstore.crate_disambiguator_untracked(cnum).to_fingerprint();
+            let hash = cstore.crate_hash_untracked(cnum);
+            (name, disambiguator, hash)
+        })
+        .collect();
+    upstream_crates.sort_unstable_by_key(|&(name, dis, _)| (name.as_str(), dis));
+    upstream_crates
 }
 
 fn hir_id_to_string(map: &Map<'_>, id: HirId) -> String {

compiler/rustc_middle/src/hir/mod.rs

@@ -19,6 +19,17 @@ use rustc_index::vec::IndexVec;
 use rustc_span::DUMMY_SP;
 use std::collections::BTreeMap;
 
+#[derive(Debug)]
+struct HirOwnerData<'hir> {
+    signature: Option<&'hir Owner<'hir>>,
+    with_bodies: Option<&'hir mut OwnerNodes<'hir>>,
+}
+
+#[derive(Debug)]
+pub struct IndexedHir<'hir> {
+    map: IndexVec<LocalDefId, HirOwnerData<'hir>>,
+}
+
 #[derive(Debug)]
 pub struct Owner<'tcx> {
     parent: HirId,
@@ -117,6 +128,7 @@ pub fn provide(providers: &mut Providers) {
     };
     providers.hir_crate = |tcx, _| tcx.untracked_crate;
     providers.index_hir = map::index_hir;
+    providers.crate_hash = map::crate_hash;
     providers.hir_module_items = |tcx, id| &tcx.untracked_crate.modules[&id];
     providers.hir_owner = |tcx, id| tcx.index_hir(LOCAL_CRATE).map[id].signature;
     providers.hir_owner_nodes = |tcx, id| tcx.index_hir(LOCAL_CRATE).map[id].with_bodies.as_deref();

compiler/rustc_middle/src/query/mod.rs

@@ -28,7 +28,7 @@ rustc_queries! {
 
     /// The indexed HIR. This can be conveniently accessed by `tcx.hir()`.
     /// Avoid calling this query directly.
-    query index_hir(_: CrateNum) -> &'tcx map::IndexedHir<'tcx> {
+    query index_hir(_: CrateNum) -> &'tcx crate::hir::IndexedHir<'tcx> {
         eval_always
         no_hash
         desc { "index HIR" }

compiler/rustc_middle/src/ty/query/mod.rs

@@ -1,6 +1,5 @@
 use crate::dep_graph;
 use crate::hir::exports::Export;
-use crate::hir::map;
 use crate::infer::canonical::{self, Canonical};
 use crate::lint::LintLevelMap;
 use crate::middle::codegen_fn_attrs::CodegenFnAttrs;

compiler/rustc_ty_utils/src/ty.rs

@@ -1,5 +1,4 @@
 use rustc_data_structures::fx::FxIndexSet;
-use rustc_data_structures::svh::Svh;
 use rustc_hir as hir;
 use rustc_hir::def_id::{CrateNum, DefId, LocalDefId, LOCAL_CRATE};
 use rustc_middle::hir::map as hir_map;
@@ -400,10 +399,6 @@ fn original_crate_name(tcx: TyCtxt<'_>, crate_num: CrateNum) -> Symbol {
     tcx.crate_name
 }
 
-fn crate_hash(tcx: TyCtxt<'_>, crate_num: CrateNum) -> Svh {
-    tcx.index_hir(crate_num).crate_hash
-}
-
 fn instance_def_size_estimate<'tcx>(
     tcx: TyCtxt<'tcx>,
     instance_def: ty::InstanceDef<'tcx>,
@@ -551,7 +546,6 @@ pub fn provide(providers: &mut ty::query::Providers) {
         trait_of_item,
         crate_disambiguator,
         original_crate_name,
-        crate_hash,
         instance_def_size_estimate,
         issue33140_self_ty,
         impl_defaultness,