perf(hstr): Use thin arc for hash and length (#10033)

**Description:**

This would improve performance and reduce allocation.

**Related issue:**

 - Closes #10030

Author: kdy1

.changeset/eight-steaks-yawn.md

@@ -0,0 +1,6 @@
+---
+swc_core: minor
+hstr: minor
+---
+
+perf(hstr): Use thin arc for hash and length

crates/hstr/src/dynamic.rs

@@ -1,64 +1,62 @@
 use std::{
     borrow::Cow,
-    fmt::Debug,
+    ffi::c_void,
     hash::{BuildHasherDefault, Hash, Hasher},
+    mem::ManuallyDrop,
+    ops::Deref,
     ptr::NonNull,
 };
 
 use rustc_hash::FxHasher;
-use triomphe::Arc;
+use triomphe::{HeaderWithLength, ThinArc};
 
 use crate::{
     tagged_value::{TaggedValue, MAX_INLINE_LEN},
     Atom, INLINE_TAG_INIT, LEN_OFFSET, TAG_MASK,
 };
 
-#[derive(Debug)]
-pub(crate) struct Entry {
-    pub string: Box<str>,
+pub(crate) struct Metadata {
     pub hash: u64,
 }
 
-impl Entry {
-    pub unsafe fn cast(ptr: TaggedValue) -> *const Entry {
-        ptr.get_ptr().cast()
-    }
+#[derive(Clone)]
+pub(crate) struct Item(ThinArc<HeaderWithLength<Metadata>, u8>);
 
-    pub unsafe fn deref_from<'i>(ptr: TaggedValue) -> &'i Entry {
-        &*Self::cast(ptr)
-    }
+impl Deref for Item {
+    type Target = <ThinArc<HeaderWithLength<Metadata>, u8> as Deref>::Target;
 
-    pub unsafe fn restore_arc(v: TaggedValue) -> Arc<Entry> {
-        let ptr = v.get_ptr() as *const Entry;
-        Arc::from_raw(ptr)
+    fn deref(&self) -> &Self::Target {
+        &self.0
     }
 }
 
-impl PartialEq for Entry {
-    fn eq(&self, other: &Self) -> bool {
-        // Assumption: `store_id` and `alias` don't matter for equality within a single
-        // store (what we care about here). Compare hash first because that's cheaper.
-        self.hash == other.hash && self.string == other.string
+/// TODO: Use real weak pointer
+type WeakItem = Item;
+
+impl Hash for Item {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        state.write_u64(self.0.header.header.header.hash);
     }
 }
 
-impl Eq for Entry {}
+pub(crate) unsafe fn deref_from(ptr: TaggedValue) -> ManuallyDrop<Item> {
+    let item = restore_arc(ptr);
 
-impl Hash for Entry {
-    fn hash<H: Hasher>(&self, state: &mut H) {
-        // Assumption: type H is an EntryHasher
-        state.write_u64(self.hash);
-    }
+    ManuallyDrop::new(item)
+}
+
+pub(crate) unsafe fn restore_arc(v: TaggedValue) -> Item {
+    let ptr = v.get_ptr();
+    Item(ThinArc::from_raw(ptr))
 }
 
 /// A store that stores [Atom]s. Can be merged with other [AtomStore]s for
 /// better performance.
 ///
 ///
 /// # Merging [AtomStore]
-#[derive(Debug)]
 pub struct AtomStore {
-    pub(crate) data: hashbrown::HashMap<Arc<Entry>, (), BuildEntryHasher>,
+    pub(crate) data: hashbrown::HashMap<WeakItem, (), BuildEntryHasher>,
 }
 
 impl Default for AtomStore {
@@ -96,11 +94,11 @@ where
 
     let hash = calc_hash(&text);
     let entry = storage.insert_entry(text, hash);
-    let entry = Arc::into_raw(entry);
+    let entry = ThinArc::into_raw(entry.0) as *mut c_void;
 
-    let ptr: NonNull<Entry> = unsafe {
+    let ptr: NonNull<c_void> = unsafe {
         // Safety: Arc::into_raw returns a non-null pointer
-        NonNull::new_unchecked(entry as *mut Entry)
+        NonNull::new_unchecked(entry)
     };
     debug_assert!(0 == ptr.as_ptr() as u8 & TAG_MASK);
     Atom {
@@ -109,22 +107,24 @@ where
 }
 
 pub(crate) trait Storage {
-    fn insert_entry(self, text: Cow<str>, hash: u64) -> Arc<Entry>;
+    fn insert_entry(self, text: Cow<str>, hash: u64) -> Item;
 }
 
 impl Storage for &'_ mut AtomStore {
     #[inline(never)]
-    fn insert_entry(self, text: Cow<str>, hash: u64) -> Arc<Entry> {
+    fn insert_entry(self, text: Cow<str>, hash: u64) -> Item {
         let (entry, _) = self
             .data
             .raw_entry_mut()
-            .from_hash(hash, |key| key.hash == hash && *key.string == *text)
+            .from_hash(hash, |key| {
+                key.header.header.header.hash == hash && key.slice == *text.as_bytes()
+            })
             .or_insert_with(move || {
                 (
-                    Arc::new(Entry {
-                        string: text.into_owned().into_boxed_str(),
-                        hash,
-                    }),
+                    Item(ThinArc::from_header_and_slice(
+                        HeaderWithLength::new(Metadata { hash }, text.len()),
+                        text.as_bytes(),
+                    )),
                     (),
                 )
             });

crates/hstr/src/lib.rs

@@ -1,20 +1,21 @@
 #![cfg_attr(feature = "atom_size_128", feature(integer_atomics))]
 //! See [Atom] for more information.
 
+use core::str;
 use std::{
     fmt::{Debug, Display},
     hash::Hash,
-    mem::{self, forget},
+    mem::{self, forget, transmute},
     num::NonZeroU8,
     ops::Deref,
+    str::from_utf8_unchecked,
 };
 
 use debug_unreachable::debug_unreachable;
 use once_cell::sync::Lazy;
-use tagged_value::TaggedValue;
 
 pub use crate::dynamic::AtomStore;
-use crate::dynamic::Entry;
+use crate::tagged_value::TaggedValue;
 
 mod dynamic;
 mod global_store;
@@ -233,7 +234,13 @@ impl Atom {
     #[inline(never)]
     fn get_hash(&self) -> u64 {
         match self.tag() {
-            DYNAMIC_TAG => unsafe { Entry::deref_from(self.unsafe_data) }.hash,
+            DYNAMIC_TAG => {
+                unsafe { crate::dynamic::deref_from(self.unsafe_data) }
+                    .header
+                    .header
+                    .header
+                    .hash
+            }
             STATIC_TAG => {
                 todo!("static hash")
             }
@@ -249,7 +256,10 @@ impl Atom {
     #[inline(never)]
     fn as_str(&self) -> &str {
         match self.tag() {
-            DYNAMIC_TAG => &unsafe { Entry::deref_from(self.unsafe_data) }.string,
+            DYNAMIC_TAG => unsafe {
+                let item = crate::dynamic::deref_from(self.unsafe_data);
+                from_utf8_unchecked(transmute::<&[u8], &'static [u8]>(&item.slice))
+            },
             STATIC_TAG => {
                 todo!("static as_str")
             }
@@ -277,14 +287,14 @@ impl PartialEq for Atom {
         }
 
         if self.is_dynamic() && other.is_dynamic() {
-            let te = unsafe { Entry::deref_from(self.unsafe_data) };
-            let oe = unsafe { Entry::deref_from(other.unsafe_data) };
+            let te = unsafe { crate::dynamic::deref_from(self.unsafe_data) };
+            let oe = unsafe { crate::dynamic::deref_from(other.unsafe_data) };
 
-            if te.hash != oe.hash {
+            if te.header.header.header.hash != oe.header.header.header.hash {
                 return false;
             }
 
-            return te.string == oe.string;
+            return te.slice == oe.slice;
         }
 
         if self.get_hash() != other.get_hash() {
@@ -310,7 +320,7 @@ impl Drop for Atom {
     #[inline(always)]
     fn drop(&mut self) {
         if self.is_dynamic() {
-            unsafe { drop(Entry::restore_arc(self.unsafe_data)) }
+            unsafe { drop(crate::dynamic::restore_arc(self.unsafe_data)) }
         }
     }
 }
@@ -327,7 +337,7 @@ impl Atom {
     pub(crate) fn from_alias(alias: TaggedValue) -> Self {
         if alias.tag() & TAG_MASK == DYNAMIC_TAG {
             unsafe {
-                let arc = Entry::restore_arc(alias);
+                let arc = crate::dynamic::restore_arc(alias);
                 forget(arc.clone());
                 forget(arc);
             }