Upgrade dashmap and hashbrown

ChayimFriedman2 · ChayimFriedman2 · commit 5dac463bda17 · 2025-04-25T12:47:26.000+03:00
And adapt `intern` to the changes in the API.
diff --git a/src/tools/rust-analyzer/Cargo.lock b/src/tools/rust-analyzer/Cargo.lock
@@ -79,7 +79,7 @@ name = "base-db"
 version = "0.0.0"
 dependencies = [
  "cfg",
- "dashmap 5.5.3",
+ "dashmap",
  "intern",
  "la-arena 0.3.1 (registry+https://github.com/rust-lang/crates.io-index)",
  "query-group-macro",
@@ -323,19 +323,6 @@ dependencies = [
  "windows-sys 0.59.0",
 ]
 
-[[package]]
-name = "dashmap"
-version = "5.5.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "978747c1d849a7d2ee5e8adc0159961c48fb7e5db2f06af6723b80123bb53856"
-dependencies = [
- "cfg-if",
- "hashbrown 0.14.5",
- "lock_api",
- "once_cell",
- "parking_lot_core",
-]
-
 [[package]]
 name = "dashmap"
 version = "6.1.0"
@@ -1051,7 +1038,7 @@ dependencies = [
 name = "intern"
 version = "0.0.0"
 dependencies = [
- "dashmap 5.5.3",
+ "dashmap",
  "hashbrown 0.14.5",
  "rustc-hash 2.1.1",
  "triomphe",
@@ -2045,7 +2032,7 @@ checksum = "1be22155f8d9732518b2db2bf379fe6f0b2375e76b08b7c8fe6c1b887d548c24"
 dependencies = [
  "boxcar",
  "crossbeam-queue",
- "dashmap 6.1.0",
+ "dashmap",
  "hashbrown 0.15.2",
  "hashlink",
  "indexmap",
diff --git a/src/tools/rust-analyzer/Cargo.toml b/src/tools/rust-analyzer/Cargo.toml
@@ -158,11 +158,10 @@ triomphe = { version = "0.1.14", default-features = false, features = ["std"] }
 url = "2.5.4"
 xshell = "0.2.7"
 
-
 # We need to freeze the version of the crate, as the raw-api feature is considered unstable
-dashmap = { version = "=5.5.3", features = ["raw-api"] }
+dashmap = { version = "=6.1.0", features = ["raw-api", "inline"] }
 # We need to freeze the version of the crate, as it needs to match with dashmap
-hashbrown = { version = "=0.14.5", features = [
+hashbrown = { version = "0.14.0", features = [
   "inline-more",
 ], default-features = false }
 
diff --git a/src/tools/rust-analyzer/crates/intern/src/lib.rs b/src/tools/rust-analyzer/crates/intern/src/lib.rs
@@ -3,22 +3,20 @@
 //! Eventually this should probably be replaced with salsa-based interning.
 
 use std::{
+    borrow::Borrow,
     fmt::{self, Debug, Display},
-    hash::{BuildHasherDefault, Hash, Hasher},
+    hash::{BuildHasher, BuildHasherDefault, Hash, Hasher},
     ops::Deref,
     sync::OnceLock,
 };
 
 use dashmap::{DashMap, SharedValue};
-use hashbrown::{HashMap, hash_map::RawEntryMut};
+use hashbrown::raw::RawTable;
 use rustc_hash::FxHasher;
 use triomphe::Arc;
 
 type InternMap<T> = DashMap<Arc<T>, (), BuildHasherDefault<FxHasher>>;
-type Guard<T> = dashmap::RwLockWriteGuard<
-    'static,
-    HashMap<Arc<T>, SharedValue<()>, BuildHasherDefault<FxHasher>>,
->;
+type Guard<T> = dashmap::RwLockWriteGuard<'static, RawTable<(Arc<T>, SharedValue<()>)>>;
 
 mod symbol;
 pub use self::symbol::{Symbol, symbols as sym};
@@ -28,54 +26,61 @@ pub struct Interned<T: Internable + ?Sized> {
 }
 
 impl<T: Internable> Interned<T> {
+    #[inline]
     pub fn new(obj: T) -> Self {
-        let (mut shard, hash) = Self::select(&obj);
-        // Atomically,
-        // - check if `obj` is already in the map
-        //   - if so, clone its `Arc` and return it
-        //   - if not, box it up, insert it, and return a clone
-        // This needs to be atomic (locking the shard) to avoid races with other thread, which could
-        // insert the same object between us looking it up and inserting it.
-        match shard.raw_entry_mut().from_key_hashed_nocheck(hash, &obj) {
-            RawEntryMut::Occupied(occ) => Self { arc: occ.key().clone() },
-            RawEntryMut::Vacant(vac) => Self {
-                arc: vac.insert_hashed_nocheck(hash, Arc::new(obj), SharedValue::new(())).0.clone(),
-            },
-        }
+        Self::new_generic(obj)
     }
 }
 
 impl Interned<str> {
+    #[inline]
     pub fn new_str(s: &str) -> Self {
-        let (mut shard, hash) = Self::select(s);
+        Self::new_generic(s)
+    }
+}
+
+impl<T: Internable + ?Sized> Interned<T> {
+    #[inline]
+    pub fn new_generic<U>(obj: U) -> Self
+    where
+        U: Borrow<T>,
+        Arc<T>: From<U>,
+    {
+        let storage = T::storage().get();
+        let (mut shard, hash) = Self::select(storage, obj.borrow());
         // Atomically,
         // - check if `obj` is already in the map
         //   - if so, clone its `Arc` and return it
         //   - if not, box it up, insert it, and return a clone
         // This needs to be atomic (locking the shard) to avoid races with other thread, which could
         // insert the same object between us looking it up and inserting it.
-        match shard.raw_entry_mut().from_key_hashed_nocheck(hash, s) {
-            RawEntryMut::Occupied(occ) => Self { arc: occ.key().clone() },
-            RawEntryMut::Vacant(vac) => Self {
-                arc: vac.insert_hashed_nocheck(hash, Arc::from(s), SharedValue::new(())).0.clone(),
+        let bucket = match shard.find_or_find_insert_slot(
+            hash,
+            |(other, _)| **other == *obj.borrow(),
+            |(x, _)| Self::hash(storage, x),
+        ) {
+            Ok(bucket) => bucket,
+            // SAFETY: The slot came from `find_or_find_insert_slot()`, and the table wasn't modified since then.
+            Err(insert_slot) => unsafe {
+                shard.insert_in_slot(hash, insert_slot, (Arc::from(obj), SharedValue::new(())))
             },
-        }
+        };
+        // SAFETY: We just retrieved/inserted this bucket.
+        unsafe { Self { arc: bucket.as_ref().0.clone() } }
     }
-}
 
-impl<T: Internable + ?Sized> Interned<T> {
     #[inline]
-    fn select(obj: &T) -> (Guard<T>, u64) {
-        let storage = T::storage().get();
-        let hash = {
-            let mut hasher = std::hash::BuildHasher::build_hasher(storage.hasher());
-            obj.hash(&mut hasher);
-            hasher.finish()
-        };
+    fn select(storage: &'static InternMap<T>, obj: &T) -> (Guard<T>, u64) {
+        let hash = Self::hash(storage, obj);
         let shard_idx = storage.determine_shard(hash as usize);
         let shard = &storage.shards()[shard_idx];
         (shard.write(), hash)
     }
+
+    #[inline]
+    fn hash(storage: &'static InternMap<T>, obj: &T) -> u64 {
+        storage.hasher().hash_one(obj)
+    }
 }
 
 impl<T: Internable + ?Sized> Drop for Interned<T> {
@@ -93,21 +98,20 @@ impl<T: Internable + ?Sized> Drop for Interned<T> {
 impl<T: Internable + ?Sized> Interned<T> {
     #[cold]
     fn drop_slow(&mut self) {
-        let (mut shard, hash) = Self::select(&self.arc);
+        let storage = T::storage().get();
+        let (mut shard, hash) = Self::select(storage, &self.arc);
 
         if Arc::count(&self.arc) != 2 {
             // Another thread has interned another copy
             return;
         }
 
-        match shard.raw_entry_mut().from_key_hashed_nocheck(hash, &self.arc) {
-            RawEntryMut::Occupied(occ) => occ.remove(),
-            RawEntryMut::Vacant(_) => unreachable!(),
-        };
+        shard.remove_entry(hash, |(other, _)| **other == *self.arc);
 
         // Shrink the backing storage if the shard is less than 50% occupied.
         if shard.len() * 2 < shard.capacity() {
-            shard.shrink_to_fit();
+            let len = shard.len();
+            shard.shrink_to(len, |(x, _)| Self::hash(storage, x));
         }
     }
 }
diff --git a/src/tools/rust-analyzer/crates/intern/src/symbol.rs b/src/tools/rust-analyzer/crates/intern/src/symbol.rs
@@ -2,16 +2,15 @@
 //! supporting compile time declaration of symbols that will never be freed.
 
 use std::{
-    borrow::Borrow,
     fmt,
-    hash::{BuildHasherDefault, Hash, Hasher},
+    hash::{BuildHasher, BuildHasherDefault, Hash},
     mem::{self, ManuallyDrop},
     ptr::NonNull,
     sync::OnceLock,
 };
 
 use dashmap::{DashMap, SharedValue};
-use hashbrown::{HashMap, hash_map::RawEntryMut};
+use hashbrown::raw::RawTable;
 use rustc_hash::FxHasher;
 use triomphe::Arc;
 
@@ -127,31 +126,39 @@ impl fmt::Debug for Symbol {
 const _: () = assert!(size_of::<Symbol>() == size_of::<NonNull<()>>());
 const _: () = assert!(align_of::<Symbol>() == align_of::<NonNull<()>>());
 
-static MAP: OnceLock<DashMap<SymbolProxy, (), BuildHasherDefault<FxHasher>>> = OnceLock::new();
+type Map = DashMap<Symbol, (), BuildHasherDefault<FxHasher>>;
+static MAP: OnceLock<Map> = OnceLock::new();
 
 impl Symbol {
     pub fn intern(s: &str) -> Self {
-        let (mut shard, hash) = Self::select_shard(s);
+        let storage = MAP.get_or_init(symbols::prefill);
+        let (mut shard, hash) = Self::select_shard(storage, s);
         // Atomically,
         // - check if `obj` is already in the map
         //   - if so, copy out its entry, conditionally bumping the backing Arc and return it
         //   - if not, put it into a box and then into an Arc, insert it, bump the ref-count and return the copy
         // This needs to be atomic (locking the shard) to avoid races with other thread, which could
         // insert the same object between us looking it up and inserting it.
-        match shard.raw_entry_mut().from_key_hashed_nocheck(hash, s) {
-            RawEntryMut::Occupied(occ) => Self { repr: increase_arc_refcount(occ.key().0) },
-            RawEntryMut::Vacant(vac) => Self {
-                repr: increase_arc_refcount(
-                    vac.insert_hashed_nocheck(
-                        hash,
-                        SymbolProxy(TaggedArcPtr::arc(Arc::new(Box::<str>::from(s)))),
+        let bucket = match shard.find_or_find_insert_slot(
+            hash,
+            |(other, _)| other.as_str() == s,
+            |(x, _)| Self::hash(storage, x.as_str()),
+        ) {
+            Ok(bucket) => bucket,
+            // SAFETY: The slot came from `find_or_find_insert_slot()`, and the table wasn't modified since then.
+            Err(insert_slot) => unsafe {
+                shard.insert_in_slot(
+                    hash,
+                    insert_slot,
+                    (
+                        Symbol { repr: TaggedArcPtr::arc(Arc::new(Box::<str>::from(s))) },
                         SharedValue::new(()),
-                    )
-                    .0
-                    .0,
-                ),
+                    ),
+                )
             },
-        }
+        };
+        // SAFETY: We just retrieved/inserted this bucket.
+        unsafe { bucket.as_ref().0.clone() }
     }
 
     pub fn integer(i: usize) -> Self {
@@ -180,58 +187,52 @@ impl Symbol {
         symbols::__empty.clone()
     }
 
+    #[inline]
     pub fn as_str(&self) -> &str {
         self.repr.as_str()
     }
 
     #[inline]
     fn select_shard(
+        storage: &'static Map,
         s: &str,
-    ) -> (
-        dashmap::RwLockWriteGuard<
-            'static,
-            HashMap<SymbolProxy, SharedValue<()>, BuildHasherDefault<FxHasher>>,
-        >,
-        u64,
-    ) {
-        let storage = MAP.get_or_init(symbols::prefill);
-        let hash = {
-            let mut hasher = std::hash::BuildHasher::build_hasher(storage.hasher());
-            s.hash(&mut hasher);
-            hasher.finish()
-        };
+    ) -> (dashmap::RwLockWriteGuard<'static, RawTable<(Symbol, SharedValue<()>)>>, u64) {
+        let hash = Self::hash(storage, s);
         let shard_idx = storage.determine_shard(hash as usize);
         let shard = &storage.shards()[shard_idx];
         (shard.write(), hash)
     }
 
+    #[inline]
+    fn hash(storage: &'static Map, s: &str) -> u64 {
+        storage.hasher().hash_one(s)
+    }
+
     #[cold]
     fn drop_slow(arc: &Arc<Box<str>>) {
-        let (mut shard, hash) = Self::select_shard(arc);
+        let storage = MAP.get_or_init(symbols::prefill);
+        let (mut shard, hash) = Self::select_shard(storage, arc);
 
         match Arc::count(arc) {
-            0 => unreachable!(),
-            1 => unreachable!(),
+            0 | 1 => unreachable!(),
             2 => (),
             _ => {
                 // Another thread has interned another copy
                 return;
             }
         }
 
-        let ptr = match shard.raw_entry_mut().from_key_hashed_nocheck::<str>(hash, arc.as_ref()) {
-            RawEntryMut::Occupied(occ) => occ.remove_entry(),
-            RawEntryMut::Vacant(_) => unreachable!(),
-        }
-        .0
-        .0;
+        let s = &***arc;
+        let (ptr, _) = shard.remove_entry(hash, |(x, _)| x.as_str() == s).unwrap();
+        let ptr = ManuallyDrop::new(ptr);
         // SAFETY: We're dropping, we have ownership.
-        ManuallyDrop::into_inner(unsafe { ptr.try_as_arc_owned().unwrap() });
+        ManuallyDrop::into_inner(unsafe { ptr.repr.try_as_arc_owned().unwrap() });
         debug_assert_eq!(Arc::count(arc), 1);
 
         // Shrink the backing storage if the shard is less than 50% occupied.
         if shard.len() * 2 < shard.capacity() {
-            shard.shrink_to_fit();
+            let len = shard.len();
+            shard.shrink_to(len, |(x, _)| Self::hash(storage, x.as_str()));
         }
     }
 }
@@ -276,22 +277,6 @@ impl fmt::Display for Symbol {
     }
 }
 
-// only exists so we can use `from_key_hashed_nocheck` with a &str
-#[derive(Debug, PartialEq, Eq)]
-struct SymbolProxy(TaggedArcPtr);
-
-impl Hash for SymbolProxy {
-    fn hash<H: Hasher>(&self, state: &mut H) {
-        self.0.as_str().hash(state);
-    }
-}
-
-impl Borrow<str> for SymbolProxy {
-    fn borrow(&self) -> &str {
-        self.0.as_str()
-    }
-}
-
 #[cfg(test)]
 mod tests {
     use super::*;
diff --git a/src/tools/rust-analyzer/crates/intern/src/symbol/symbols.rs b/src/tools/rust-analyzer/crates/intern/src/symbol/symbols.rs
