Auto merge of #37132 - petrochenkov:intern, r=alexcrichton

Get rid of double indirection in string interner

Author: bors

src/liballoc/rc.rs

@@ -230,13 +230,14 @@ use core::hash::{Hash, Hasher};
 use core::intrinsics::{abort, assume};
 use core::marker;
 use core::marker::Unsize;
-use core::mem::{self, align_of_val, forget, size_of_val, uninitialized};
+use core::mem::{self, align_of_val, forget, size_of, size_of_val, uninitialized};
 use core::ops::Deref;
 use core::ops::CoerceUnsized;
 use core::ptr::{self, Shared};
 use core::convert::From;
 
 use heap::deallocate;
+use raw_vec::RawVec;
 
 struct RcBox<T: ?Sized> {
     strong: Cell<usize>,
@@ -365,6 +366,31 @@ impl<T> Rc<T> {
     }
 }
 
+impl Rc<str> {
+    /// Constructs a new `Rc<str>` from a string slice.
+    #[doc(hidden)]
+    #[unstable(feature = "rustc_private",
+               reason = "for internal use in rustc",
+               issue = "0")]
+    pub fn __from_str(value: &str) -> Rc<str> {
+        unsafe {
+            // Allocate enough space for `RcBox<str>`.
+            let aligned_len = 2 + (value.len() + size_of::<usize>() - 1) / size_of::<usize>();
+            let vec = RawVec::<usize>::with_capacity(aligned_len);
+            let ptr = vec.ptr();
+            forget(vec);
+            // Initialize fields of `RcBox<str>`.
+            *ptr.offset(0) = 1; // strong: Cell::new(1)
+            *ptr.offset(1) = 1; // weak: Cell::new(1)
+            ptr::copy_nonoverlapping(value.as_ptr(), ptr.offset(2) as *mut u8, value.len());
+            // Combine the allocation address and the string length into a fat pointer to `RcBox`.
+            let rcbox_ptr: *mut RcBox<str> = mem::transmute([ptr as usize, value.len()]);
+            assert!(aligned_len * size_of::<usize>() == size_of_val(&*rcbox_ptr));
+            Rc { ptr: Shared::new(rcbox_ptr) }
+        }
+    }
+}
+
 impl<T: ?Sized> Rc<T> {
     /// Creates a new [`Weak`][weak] pointer to this value.
     ///

src/librustc_trans/debuginfo/metadata.rs

@@ -128,7 +128,7 @@ impl<'tcx> TypeMap<'tcx> {
 
     // Get the string representation of a UniqueTypeId. This method will fail if
     // the id is unknown.
-    fn get_unique_type_id_as_string(&self, unique_type_id: UniqueTypeId) -> Rc<String> {
+    fn get_unique_type_id_as_string(&self, unique_type_id: UniqueTypeId) -> Rc<str> {
         let UniqueTypeId(interner_key) = unique_type_id;
         self.unique_id_interner.get(interner_key)
     }
@@ -299,7 +299,7 @@ impl<'tcx> TypeMap<'tcx> {
         // Trim to size before storing permanently
         unique_type_id.shrink_to_fit();
 
-        let key = self.unique_id_interner.intern(unique_type_id);
+        let key = self.unique_id_interner.intern(&unique_type_id);
         self.type_to_unique_id.insert(type_, UniqueTypeId(key));
 
         return UniqueTypeId(key);
@@ -367,7 +367,7 @@ impl<'tcx> TypeMap<'tcx> {
         let enum_variant_type_id = format!("{}::{}",
                                            &self.get_unique_type_id_as_string(enum_type_id),
                                            variant_name);
-        let interner_key = self.unique_id_interner.intern(enum_variant_type_id);
+        let interner_key = self.unique_id_interner.intern(&enum_variant_type_id);
         UniqueTypeId(interner_key)
     }
 }

src/libsyntax/parse/token.rs

@@ -478,27 +478,20 @@ pub fn clear_ident_interner() {
 /// somehow.
 #[derive(Clone, PartialEq, Hash, PartialOrd, Eq, Ord)]
 pub struct InternedString {
-    string: Rc<String>,
+    string: Rc<str>,
 }
 
 impl InternedString {
     #[inline]
     pub fn new(string: &'static str) -> InternedString {
         InternedString {
-            string: Rc::new(string.to_owned()),
-        }
-    }
-
-    #[inline]
-    fn new_from_rc_str(string: Rc<String>) -> InternedString {
-        InternedString {
-            string: string,
+            string: Rc::__from_str(string),
         }
     }
 
     #[inline]
     pub fn new_from_name(name: ast::Name) -> InternedString {
-        with_ident_interner(|interner| InternedString::new_from_rc_str(interner.get(name)))
+        with_ident_interner(|interner| InternedString { string: interner.get(name) })
     }
 }
 

src/libsyntax/util/interner.rs

@@ -14,23 +14,13 @@
 
 use ast::Name;
 
-use std::borrow::Borrow;
 use std::collections::HashMap;
 use std::rc::Rc;
 
-#[derive(PartialEq, Eq, Hash)]
-struct RcStr(Rc<String>);
-
-impl Borrow<str> for RcStr {
-    fn borrow(&self) -> &str {
-        &self.0
-    }
-}
-
 #[derive(Default)]
 pub struct Interner {
-    names: HashMap<RcStr, Name>,
-    strings: Vec<Rc<String>>,
+    names: HashMap<Rc<str>, Name>,
+    strings: Vec<Rc<str>>,
 }
 
 /// When traits can extend traits, we should extend index<Name,T> to get []
@@ -47,22 +37,22 @@ impl Interner {
         this
     }
 
-    pub fn intern<T: Borrow<str> + Into<String>>(&mut self, string: T) -> Name {
-        if let Some(&name) = self.names.get(string.borrow()) {
+    pub fn intern(&mut self, string: &str) -> Name {
+        if let Some(&name) = self.names.get(string) {
             return name;
         }
 
         let name = Name(self.strings.len() as u32);
-        let string = Rc::new(string.into());
+        let string = Rc::__from_str(string);
         self.strings.push(string.clone());
-        self.names.insert(RcStr(string), name);
+        self.names.insert(string, name);
         name
     }
 
     pub fn gensym(&mut self, string: &str) -> Name {
         let gensym = Name(self.strings.len() as u32);
         // leave out of `names` to avoid colliding
-        self.strings.push(Rc::new(string.to_owned()));
+        self.strings.push(Rc::__from_str(string));
         gensym
     }
 
@@ -75,7 +65,7 @@ impl Interner {
         gensym
     }
 
-    pub fn get(&self, name: Name) -> Rc<String> {
+    pub fn get(&self, name: Name) -> Rc<str> {
         self.strings[name.0 as usize].clone()
     }
 
@@ -109,13 +99,13 @@ mod tests {
         assert_eq!(i.gensym("dog"), Name(4));
         // gensym tests again with gensym_copy:
         assert_eq!(i.gensym_copy(Name(2)), Name(5));
-        assert_eq!(*i.get(Name(5)), "zebra");
+        assert_eq!(&*i.get(Name(5)), "zebra");
         assert_eq!(i.gensym_copy(Name(2)), Name(6));
-        assert_eq!(*i.get(Name(6)), "zebra");
-        assert_eq!(*i.get(Name(0)), "dog");
-        assert_eq!(*i.get(Name(1)), "cat");
-        assert_eq!(*i.get(Name(2)), "zebra");
-        assert_eq!(*i.get(Name(3)), "zebra");
-        assert_eq!(*i.get(Name(4)), "dog");
+        assert_eq!(&*i.get(Name(6)), "zebra");
+        assert_eq!(&*i.get(Name(0)), "dog");
+        assert_eq!(&*i.get(Name(1)), "cat");
+        assert_eq!(&*i.get(Name(2)), "zebra");
+        assert_eq!(&*i.get(Name(3)), "zebra");
+        assert_eq!(&*i.get(Name(4)), "dog");
     }
 }