Encode ExpnId using ExpnHash for incr. comp.

Author: cjgillot

compiler/rustc_metadata/src/rmeta/decoder.rs

@@ -79,6 +79,8 @@ crate struct CrateMetadata {
     /// `DefIndex`. See `raw_def_id_to_def_id` for more details about how
     /// this is used.
     def_path_hash_map: OnceCell<UnhashMap<DefPathHash, DefIndex>>,
+    /// Likewise for ExpnHash.
+    expn_hash_map: OnceCell<UnhashMap<ExpnHash, ExpnIndex>>,
     /// Used for decoding interpret::AllocIds in a cached & thread-safe manner.
     alloc_decoding_state: AllocDecodingState,
     /// Caches decoded `DefKey`s.
@@ -1619,6 +1621,41 @@ impl<'a, 'tcx> CrateMetadataRef<'a> {
         self.def_path_hash_unlocked(index, &mut def_path_hashes)
     }
 
+    fn expn_hash_to_expn_id(&self, index_guess: u32, hash: ExpnHash) -> ExpnId {
+        debug_assert_eq!(ExpnId::from_hash(hash), None);
+        let index_guess = ExpnIndex::from_u32(index_guess);
+        let old_hash = self.root.expn_hashes.get(self, index_guess).map(|lazy| lazy.decode(self));
+
+        let index = if old_hash == Some(hash) {
+            // Fast path: the expn and its index is unchanged from the
+            // previous compilation session. There is no need to decode anything
+            // else.
+            index_guess
+        } else {
+            // Slow path: We need to find out the new `DefIndex` of the provided
+            // `DefPathHash`, if its still exists. This requires decoding every `DefPathHash`
+            // stored in this crate.
+            let map = self.cdata.expn_hash_map.get_or_init(|| {
+                let end_id = self.root.expn_hashes.size() as u32;
+                let mut map =
+                    UnhashMap::with_capacity_and_hasher(end_id as usize, Default::default());
+                for i in 0..end_id {
+                    let i = ExpnIndex::from_u32(i);
+                    if let Some(hash) = self.root.expn_hashes.get(self, i) {
+                        map.insert(hash.decode(self), i);
+                    } else {
+                        panic!("Missing expn_hash entry for {:?}", i);
+                    }
+                }
+                map
+            });
+            map[&hash]
+        };
+
+        let data = self.root.expn_data.get(self, index).unwrap().decode(self);
+        rustc_span::hygiene::register_expn_id(data, hash)
+    }
+
     /// Imports the source_map from an external crate into the source_map of the crate
     /// currently being compiled (the "local crate").
     ///
@@ -1857,6 +1894,7 @@ impl CrateMetadata {
             raw_proc_macros,
             source_map_import_info: OnceCell::new(),
             def_path_hash_map: Default::default(),
+            expn_hash_map: Default::default(),
             alloc_decoding_state,
             cnum,
             cnum_map,

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

@@ -18,7 +18,7 @@ use rustc_middle::ty::query::Providers;
 use rustc_middle::ty::{self, TyCtxt, Visibility};
 use rustc_session::utils::NativeLibKind;
 use rustc_session::{Session, StableCrateId};
-use rustc_span::hygiene::{ExpnData, ExpnHash, ExpnId};
+use rustc_span::hygiene::{ExpnHash, ExpnId};
 use rustc_span::source_map::{Span, Spanned};
 use rustc_span::symbol::Symbol;
 
@@ -494,23 +494,6 @@ impl CrateStore for CStore {
     fn as_any(&self) -> &dyn Any {
         self
     }
-    fn decode_expn_data(&self, sess: &Session, expn_id: ExpnId) -> (ExpnData, ExpnHash) {
-        let crate_data = self.get_crate_data(expn_id.krate);
-        (
-            crate_data
-                .root
-                .expn_data
-                .get(&crate_data, expn_id.local_id)
-                .unwrap()
-                .decode((&crate_data, sess)),
-            crate_data
-                .root
-                .expn_hashes
-                .get(&crate_data, expn_id.local_id)
-                .unwrap()
-                .decode((&crate_data, sess)),
-        )
-    }
 
     fn crate_name(&self, cnum: CrateNum) -> Symbol {
         self.get_crate_data(cnum).root.name
@@ -545,6 +528,10 @@ impl CrateStore for CStore {
         self.get_crate_data(cnum).def_path_hash_to_def_id(cnum, index_guess, hash)
     }
 
+    fn expn_hash_to_expn_id(&self, cnum: CrateNum, index_guess: u32, hash: ExpnHash) -> ExpnId {
+        self.get_crate_data(cnum).expn_hash_to_expn_id(index_guess, hash)
+    }
+
     fn encode_metadata(&self, tcx: TyCtxt<'_>) -> EncodedMetadata {
         encoder::encode_metadata(tcx)
     }

compiler/rustc_middle/src/middle/cstore.rs

@@ -11,8 +11,7 @@ use rustc_hir::definitions::{DefKey, DefPath, DefPathHash};
 use rustc_macros::HashStable;
 use rustc_session::search_paths::PathKind;
 use rustc_session::utils::NativeLibKind;
-use rustc_session::Session;
-use rustc_span::hygiene::{ExpnData, ExpnHash, ExpnId};
+use rustc_span::hygiene::{ExpnHash, ExpnId};
 use rustc_span::symbol::Symbol;
 use rustc_span::Span;
 use rustc_target::spec::Target;
@@ -188,7 +187,6 @@ pub type MetadataLoaderDyn = dyn MetadataLoader + Sync;
 /// during resolve)
 pub trait CrateStore: std::fmt::Debug {
     fn as_any(&self) -> &dyn Any;
-    fn decode_expn_data(&self, sess: &Session, expn_id: ExpnId) -> (ExpnData, ExpnHash);
 
     // Foreign definitions.
     // This information is safe to access, since it's hashed as part of the DefPathHash, which incr.
@@ -209,6 +207,7 @@ pub trait CrateStore: std::fmt::Debug {
         index_guess: u32,
         hash: DefPathHash,
     ) -> Option<DefId>;
+    fn expn_hash_to_expn_id(&self, cnum: CrateNum, index_guess: u32, hash: ExpnHash) -> ExpnId;
 
     // utility functions
     fn encode_metadata(&self, tcx: TyCtxt<'_>) -> EncodedMetadata;

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

@@ -83,14 +83,16 @@ pub struct OnDiskCache<'sess> {
     // `ExpnData` (e.g `ExpnData.krate` may not be `LOCAL_CRATE`). Alternatively,
     // we could look up the `ExpnData` from the metadata of foreign crates,
     // but it seemed easier to have `OnDiskCache` be independent of the `CStore`.
-    expn_data: FxHashMap<u32, AbsoluteBytePos>,
+    expn_data: UnhashMap<ExpnHash, AbsoluteBytePos>,
     // Additional information used when decoding hygiene data.
     hygiene_context: HygieneDecodeContext,
     // Maps `DefPathHash`es to their `RawDefId`s from the *previous*
     // compilation session. This is used as an initial 'guess' when
     // we try to map a `DefPathHash` to its `DefId` in the current compilation
     // session.
     foreign_def_path_hashes: UnhashMap<DefPathHash, RawDefId>,
+    // Likewise for ExpnId.
+    foreign_expn_data: UnhashMap<ExpnHash, u32>,
 
     // The *next* compilation sessison's `foreign_def_path_hashes` - at
     // the end of our current compilation session, this will get written
@@ -118,8 +120,9 @@ struct Footer {
     // See `OnDiskCache.syntax_contexts`
     syntax_contexts: FxHashMap<u32, AbsoluteBytePos>,
     // See `OnDiskCache.expn_data`
-    expn_data: FxHashMap<u32, AbsoluteBytePos>,
+    expn_data: UnhashMap<ExpnHash, AbsoluteBytePos>,
     foreign_def_path_hashes: UnhashMap<DefPathHash, RawDefId>,
+    foreign_expn_data: UnhashMap<ExpnHash, u32>,
 }
 
 pub type EncodedQueryResultIndex = Vec<(SerializedDepNodeIndex, AbsoluteBytePos)>;
@@ -217,6 +220,7 @@ impl<'sess> OnDiskCache<'sess> {
             alloc_decoding_state: AllocDecodingState::new(footer.interpret_alloc_index),
             syntax_contexts: footer.syntax_contexts,
             expn_data: footer.expn_data,
+            foreign_expn_data: footer.foreign_expn_data,
             hygiene_context: Default::default(),
             foreign_def_path_hashes: footer.foreign_def_path_hashes,
             latest_foreign_def_path_hashes: Default::default(),
@@ -236,7 +240,8 @@ impl<'sess> OnDiskCache<'sess> {
             prev_diagnostics_index: Default::default(),
             alloc_decoding_state: AllocDecodingState::new(Vec::new()),
             syntax_contexts: FxHashMap::default(),
-            expn_data: FxHashMap::default(),
+            expn_data: UnhashMap::default(),
+            foreign_expn_data: UnhashMap::default(),
             hygiene_context: Default::default(),
             foreign_def_path_hashes: Default::default(),
             latest_foreign_def_path_hashes: Default::default(),
@@ -350,7 +355,8 @@ impl<'sess> OnDiskCache<'sess> {
             };
 
             let mut syntax_contexts = FxHashMap::default();
-            let mut expn_ids = FxHashMap::default();
+            let mut expn_data = UnhashMap::default();
+            let mut foreign_expn_data = UnhashMap::default();
 
             // Encode all hygiene data (`SyntaxContextData` and `ExpnData`) from the current
             // session.
@@ -363,13 +369,14 @@ impl<'sess> OnDiskCache<'sess> {
                     syntax_contexts.insert(index, pos);
                     Ok(())
                 },
-                |encoder, index, expn_data, hash| -> FileEncodeResult {
-                    if index.krate == LOCAL_CRATE {
+                |encoder, expn_id, data, hash| -> FileEncodeResult {
+                    if expn_id.krate == LOCAL_CRATE {
                         let pos = AbsoluteBytePos::new(encoder.position());
-                        encoder.encode_tagged(TAG_EXPN_DATA, &(expn_data, hash))?;
-                        expn_ids.insert(index.local_id.as_u32(), pos);
+                        encoder.encode_tagged(TAG_EXPN_DATA, &data)?;
+                        expn_data.insert(hash, pos);
+                    } else {
+                        foreign_expn_data.insert(hash, expn_id.local_id.as_u32());
                     }
-                    // TODO Handle foreign expansions.
                     Ok(())
                 },
             )?;
@@ -387,7 +394,8 @@ impl<'sess> OnDiskCache<'sess> {
                     diagnostics_index,
                     interpret_alloc_index,
                     syntax_contexts,
-                    expn_data: expn_ids,
+                    expn_data,
+                    foreign_expn_data,
                     foreign_def_path_hashes,
                 },
             )?;
@@ -549,6 +557,7 @@ impl<'sess> OnDiskCache<'sess> {
             alloc_decoding_session: self.alloc_decoding_state.new_decoding_session(),
             syntax_contexts: &self.syntax_contexts,
             expn_data: &self.expn_data,
+            foreign_expn_data: &self.foreign_expn_data,
             hygiene_context: &self.hygiene_context,
         };
         f(&mut decoder)
@@ -643,7 +652,8 @@ pub struct CacheDecoder<'a, 'tcx> {
     file_index_to_stable_id: &'a FxHashMap<SourceFileIndex, EncodedSourceFileId>,
     alloc_decoding_session: AllocDecodingSession<'a>,
     syntax_contexts: &'a FxHashMap<u32, AbsoluteBytePos>,
-    expn_data: &'a FxHashMap<u32, AbsoluteBytePos>,
+    expn_data: &'a UnhashMap<ExpnHash, AbsoluteBytePos>,
+    foreign_expn_data: &'a UnhashMap<ExpnHash, u32>,
     hygiene_context: &'a HygieneDecodeContext,
 }
 
@@ -794,27 +804,43 @@ impl<'a, 'tcx> Decodable<CacheDecoder<'a, 'tcx>> for SyntaxContext {
 
 impl<'a, 'tcx> Decodable<CacheDecoder<'a, 'tcx>> for ExpnId {
     fn decode(decoder: &mut CacheDecoder<'a, 'tcx>) -> Result<Self, String> {
-        let krate = CrateNum::decode(decoder)?;
-        let index = u32::decode(decoder)?;
-
-        let expn_data = decoder.expn_data;
-        let tcx = decoder.tcx;
-        rustc_span::hygiene::decode_expn_id_incrcomp(
-            krate,
-            index,
-            decoder.hygiene_context,
-            |index| -> Result<(ExpnData, ExpnHash), _> {
-                // This closure is invoked if we haven't already decoded the data for the `ExpnId` we are deserializing.
-                // We look up the position of the associated `ExpnData` and decode it.
-                let pos = expn_data
-                    .get(&index)
-                    .unwrap_or_else(|| panic!("Bad index {:?} (map {:?})", index, expn_data));
-
-                decoder
-                    .with_position(pos.to_usize(), |decoder| decode_tagged(decoder, TAG_EXPN_DATA))
-            },
-            |expn_id| tcx.untracked_resolutions.cstore.decode_expn_data(tcx.sess, expn_id),
-        )
+        let hash = ExpnHash::decode(decoder)?;
+        if hash.is_root() {
+            return Ok(ExpnId::root());
+        }
+
+        if let Some(expn_id) = ExpnId::from_hash(hash) {
+            return Ok(expn_id);
+        }
+
+        let krate = decoder.cnum_map[&hash.stable_crate_id()];
+
+        let expn_id = if krate == LOCAL_CRATE {
+            // We look up the position of the associated `ExpnData` and decode it.
+            let pos = decoder
+                .expn_data
+                .get(&hash)
+                .unwrap_or_else(|| panic!("Bad hash {:?} (map {:?})", hash, decoder.expn_data));
+
+            let data: ExpnData = decoder
+                .with_position(pos.to_usize(), |decoder| decode_tagged(decoder, TAG_EXPN_DATA))?;
+            rustc_span::hygiene::register_local_expn_id(data, hash)
+        } else {
+            let index_guess = decoder.foreign_expn_data[&hash];
+            decoder.tcx.untracked_resolutions.cstore.expn_hash_to_expn_id(krate, index_guess, hash)
+        };
+
+        #[cfg(debug_assertions)]
+        {
+            use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
+            let mut hcx = decoder.tcx.create_stable_hashing_context();
+            let mut hasher = StableHasher::new();
+            expn_id.expn_data().hash_stable(&mut hcx, &mut hasher);
+            let local_hash: u64 = hasher.finish();
+            debug_assert_eq!(hash.local_hash(), local_hash);
+        }
+
+        Ok(expn_id)
     }
 }
 
@@ -990,8 +1016,7 @@ where
 {
     fn encode(&self, s: &mut CacheEncoder<'a, 'tcx, E>) -> Result<(), E::Error> {
         s.hygiene_context.schedule_expn_data_for_encoding(*self);
-        self.krate.encode(s)?;
-        self.local_id.as_u32().encode(s)
+        self.expn_hash().encode(s)
     }
 }
 

compiler/rustc_span/src/def_id.rs

@@ -136,7 +136,7 @@ impl Borrow<Fingerprint> for DefPathHash {
 /// further trouble.
 #[derive(Copy, Clone, Hash, PartialEq, Eq, PartialOrd, Ord, Debug)]
 #[derive(HashStable_Generic, Encodable, Decodable)]
-pub struct StableCrateId(u64);
+pub struct StableCrateId(pub(crate) u64);
 
 impl StableCrateId {
     pub fn to_u64(self) -> u64 {