Load all DefPathHash -> DefId mappings before dumping dep graph

Fixes rust-lang#79723

Author: Aaron1011

compiler/rustc_incremental/src/assert_dep_graph.rs

@@ -216,6 +216,7 @@ fn check_paths<'tcx>(tcx: TyCtxt<'tcx>, if_this_changed: &Sources, then_this_wou
 }
 
 fn dump_graph(tcx: TyCtxt<'_>) {
+    tcx.build_debug_def_path_hash_map();
     let path: String = env::var("RUST_DEP_GRAPH").unwrap_or_else(|_| "dep_graph".to_string());
     let query = tcx.dep_graph.query();
 

compiler/rustc_incremental/src/persist/load.rs

@@ -210,6 +210,12 @@ pub fn load_query_result_cache<'a>(
     definitions: &Definitions,
 ) -> Option<OnDiskCache<'a>> {
     if sess.opts.incremental.is_none() {
+        // Create a cache so that we can map any `DefPathHash`
+        // to its `DefId` when printing the dep graph
+        if sess.opts.debugging_opts.dump_dep_graph {
+            return Some(OnDiskCache::new_empty(sess.source_map()));
+        }
+        // We have no use for a cache
         return None;
     }
 

compiler/rustc_metadata/src/rmeta/decoder.rs

@@ -1591,6 +1591,29 @@ impl<'a, 'tcx> CrateMetadataRef<'a> {
         })
     }
 
+    fn num_def_ids(&self) -> usize {
+        self.root.tables.def_keys.size()
+    }
+
+    // Only used when certain `-Z` flags are enabled - should *not* be used
+    // otherwise
+    fn debug_all_def_path_hashes_and_def_ids(&self) -> Vec<(DefPathHash, DefId)> {
+        let mut def_path_hashes = self.def_path_hash_cache.lock();
+        let mut def_index_to_data = |index| {
+            (self.def_path_hash_unlocked(index, &mut def_path_hashes), self.local_def_id(index))
+        };
+        if let Some(data) = &self.root.proc_macro_data {
+            std::iter::once(CRATE_DEF_INDEX)
+                .chain(data.macros.decode(self))
+                .map(def_index_to_data)
+                .collect()
+        } else {
+            (0..self.num_def_ids())
+                .map(|index| def_index_to_data(DefIndex::from_usize(index)))
+                .collect()
+        }
+    }
+
     #[inline]
     fn def_path_hash(&self, index: DefIndex) -> DefPathHash {
         let mut def_path_hashes = self.def_path_hash_cache.lock();
@@ -1926,10 +1949,6 @@ impl CrateMetadata {
         self.root.hash
     }
 
-    fn num_def_ids(&self) -> usize {
-        self.root.tables.def_keys.size()
-    }
-
     fn local_def_id(&self, index: DefIndex) -> DefId {
         DefId { krate: self.cnum, index }
     }

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

@@ -529,4 +529,12 @@ impl CrateStore for CStore {
     fn allocator_kind(&self) -> Option<AllocatorKind> {
         self.allocator_kind()
     }
+
+    fn debug_all_def_path_hashes_and_def_ids(&self, cnum: CrateNum) -> Vec<(DefPathHash, DefId)> {
+        self.get_crate_data(cnum).debug_all_def_path_hashes_and_def_ids()
+    }
+
+    fn num_def_ids(&self, cnum: CrateNum) -> usize {
+        self.get_crate_data(cnum).num_def_ids()
+    }
 }

compiler/rustc_middle/src/middle/cstore.rs

@@ -210,6 +210,8 @@ pub trait CrateStore {
     fn encode_metadata(&self, tcx: TyCtxt<'_>) -> EncodedMetadata;
     fn metadata_encoding_version(&self) -> &[u8];
     fn allocator_kind(&self) -> Option<AllocatorKind>;
+    fn debug_all_def_path_hashes_and_def_ids(&self, cnum: CrateNum) -> Vec<(DefPathHash, DefId)>;
+    fn num_def_ids(&self, cnum: CrateNum) -> usize;
 }
 
 pub type CrateStoreDyn = dyn CrateStore + sync::Sync;

compiler/rustc_middle/src/ty/context.rs

@@ -1343,6 +1343,12 @@ impl<'tcx> TyCtxt<'tcx> {
         self.queries.on_disk_cache.as_ref().map(|c| c.serialize(self, encoder)).unwrap_or(Ok(()))
     }
 
+    pub fn build_debug_def_path_hash_map(self) {
+        if let Some(cache) = &self.queries.on_disk_cache {
+            cache.build_debug_def_path_hash_map(self);
+        }
+    }
+
     /// If `true`, we should use the MIR-based borrowck, but also
     /// fall back on the AST borrowck if the MIR-based one errors.
     pub fn migrate_borrowck(self) -> bool {

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

@@ -8,6 +8,7 @@ use rustc_data_structures::fingerprint::{Fingerprint, FingerprintDecoder, Finger
 use rustc_data_structures::fx::{FxHashMap, FxHashSet, FxIndexSet};
 use rustc_data_structures::sync::{HashMapExt, Lock, Lrc, OnceCell};
 use rustc_data_structures::thin_vec::ThinVec;
+use rustc_data_structures::unhash::UnhashMap;
 use rustc_errors::Diagnostic;
 use rustc_hir::def_id::{CrateNum, DefId, DefIndex, LocalDefId, LOCAL_CRATE};
 use rustc_hir::definitions::DefPathHash;
@@ -108,6 +109,19 @@ pub struct OnDiskCache<'sess> {
     // may no longer exist in the current compilation session, so
     // we use `Option<DefId>` so that we can cache a lookup failure.
     def_path_hash_to_def_id_cache: Lock<FxHashMap<DefPathHash, Option<DefId>>>,
+
+    // A map from a `DefPathHash` to its corresponding `DefId`, for all
+    // known `DefId`s (both foreign and local). Building this map takes
+    // a significant amount of time, so it is only initialized when
+    // needed by certain `-Z` flags (currently `-Z dump-dep-graph`).
+    // When this `OnceCell` is not initialized, the other fields of
+    // this struct (e.g. `foreign_def_path_hahses`) are used to map
+    // `DefPathHashes` to `DefId`s
+    //
+    // This field should *not* be used directly - the `def_path_hash_to_def_id`
+    // function should be used to map a `DefPathHash` to its corresponding
+    // `DefId`
+    debug_def_path_hash_to_def_id: OnceCell<UnhashMap<DefPathHash, DefId>>,
 }
 
 // This type is used only for serialization and deserialization.
@@ -215,6 +229,7 @@ impl<'sess> OnDiskCache<'sess> {
             latest_foreign_def_path_hashes: Default::default(),
             local_def_path_hash_to_def_id: make_local_def_path_hash_map(definitions),
             def_path_hash_to_def_id_cache: Default::default(),
+            debug_def_path_hash_to_def_id: OnceCell::new(),
         }
     }
 
@@ -237,6 +252,7 @@ impl<'sess> OnDiskCache<'sess> {
             latest_foreign_def_path_hashes: Default::default(),
             local_def_path_hash_to_def_id: Default::default(),
             def_path_hash_to_def_id_cache: Default::default(),
+            debug_def_path_hash_to_def_id: OnceCell::new(),
         }
     }
 
@@ -474,6 +490,20 @@ impl<'sess> OnDiskCache<'sess> {
             .insert(hash, RawDefId { krate: def_id.krate.as_u32(), index: def_id.index.as_u32() });
     }
 
+    pub fn build_debug_def_path_hash_map(&self, tcx: TyCtxt<'tcx>) {
+        let crates = tcx.cstore.crates_untracked();
+
+        let capacity = tcx.definitions.def_path_table().num_def_ids()
+            + crates.iter().map(|cnum| tcx.cstore.num_def_ids(*cnum)).sum::<usize>();
+        let mut map = UnhashMap::with_capacity_and_hasher(capacity, Default::default());
+
+        map.extend(tcx.definitions.def_path_table().all_def_path_hashes_and_def_ids(LOCAL_CRATE));
+        for cnum in &crates {
+            map.extend(tcx.cstore.debug_all_def_path_hashes_and_def_ids(*cnum).into_iter());
+        }
+        self.debug_def_path_hash_to_def_id.set(map).unwrap();
+    }
+
     /// If the given `dep_node`'s hash still exists in the current compilation,
     /// and its current `DefId` is foreign, calls `store_foreign_def_id` with it.
     ///
@@ -625,6 +655,14 @@ impl<'sess> OnDiskCache<'sess> {
             Entry::Occupied(e) => *e.get(),
             Entry::Vacant(e) => {
                 debug!("def_path_hash_to_def_id({:?})", hash);
+
+                if let Some(debug_map) = self.debug_def_path_hash_to_def_id.get() {
+                    if let Some(def_id) = debug_map.get(&hash).copied() {
+                        e.insert(Some(def_id));
+                        return Some(def_id);
+                    }
+                }
+
                 // Check if the `DefPathHash` corresponds to a definition in the current
                 // crate
                 if let Some(def_id) = self.local_def_path_hash_to_def_id.get(&hash).cloned() {

src/test/run-make/dump-dep-graph/Makefile

@@ -0,0 +1,15 @@
+include ../../run-make-fulldeps/tools.mk
+
+# Tests that the dep graph contains printed def paths,
+# instead of just hashes
+
+all: main.rs
+	cp main.rs $(TMPDIR)/
+	cd $(TMPDIR)
+	$(RUSTC) -Z dump-dep-graph $<
+	# Make sure that we can find a local function
+	# (my_function) and something from a foreign crate
+	# (compiler_builtins) in the dep graph
+	grep my_function dep_graph.txt
+	grep compiler_builtins dep_graph.txt
+	rm dep_graph.txt dep_graph.dot

src/test/run-make/dump-dep-graph/main.rs

@@ -0,0 +1,2 @@
+fn my_function() {}
+fn main() {}