Merge pull request #19495 from lorentey/simplify-rawHashValue

[stdlib] Finalize one-shot hashing interface

Author: lorentey

stdlib/private/StdlibUnittest/StdlibUnittest.swift

@@ -2378,7 +2378,7 @@ public func checkHashable<Instances: Collection>(
           """,
           stackTrace: stackTrace.pushIf(showFrame, file: file, line: line))
         expectEqual(
-          x._rawHashValue(seed: (0, 0)), y._rawHashValue(seed: (0, 0)),
+          x._rawHashValue(seed: 0), y._rawHashValue(seed: 0),
           """
           _rawHashValue expected to match, found to differ
           lhs (at index \(i)): \(x)
@@ -2399,8 +2399,8 @@ public func checkHashable<Instances: Collection>(
           """,
           stackTrace: stackTrace.pushIf(showFrame, file: file, line: line))
         expectTrue(
-          (0..<10 as Range<UInt64>).contains { i in
-            x._rawHashValue(seed: (0, i)) != y._rawHashValue(seed: (0, i))
+          (0..<10).contains { i in
+            x._rawHashValue(seed: i) != y._rawHashValue(seed: i)
           },
           """
           _rawHashValue(seed:) expected to differ, found to match

stdlib/public/SDK/ObjectiveC/ObjectiveC.swift

@@ -240,7 +240,7 @@ extension NSObject : Equatable, Hashable {
     hasher.combine(hashValue)
   }
 
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     // FIXME: We should use self.hash here, but hashValue is currently
     // overridable.
     return self.hashValue._rawHashValue(seed: seed)

stdlib/public/SwiftShims/GlobalObjects.h

@@ -44,8 +44,7 @@ struct _SwiftDictionaryBodyStorage {
   __swift_intptr_t count;
   __swift_intptr_t capacity;
   __swift_intptr_t scale;
-  __swift_uint64_t seed0;
-  __swift_uint64_t seed1;
+  __swift_intptr_t seed;
   void *rawKeys;
   void *rawValues;
 };
@@ -54,8 +53,7 @@ struct _SwiftSetBodyStorage {
   __swift_intptr_t count;
   __swift_intptr_t capacity;
   __swift_intptr_t scale;
-  __swift_uint64_t seed0;
-  __swift_uint64_t seed1;
+  __swift_intptr_t seed;
   void *rawElements;
 };
 

stdlib/public/core/AnyHashable.swift

@@ -49,7 +49,7 @@ internal protocol _AnyHashableBox {
   func _isEqual(to box: _AnyHashableBox) -> Bool?
   var _hashValue: Int { get }
   func _hash(into hasher: inout Hasher)
-  func _rawHashValue(_seed: (UInt64, UInt64)) -> Int
+  func _rawHashValue(_seed: Int) -> Int
 
   var _base: Any { get }
   func _unbox<T: Hashable>() -> T?
@@ -97,7 +97,7 @@ internal struct _ConcreteHashableBox<Base : Hashable> : _AnyHashableBox {
   }
 
   @inlinable // FIXME(sil-serialize-all)
-  func _rawHashValue(_seed: (UInt64, UInt64)) -> Int {
+  func _rawHashValue(_seed: Int) -> Int {
     return _baseHashable._rawHashValue(seed: _seed)
   }
 
@@ -269,7 +269,7 @@ extension AnyHashable : Hashable {
   }
 
   @inlinable // FIXME(sil-serialize-all)
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     return _box._canonicalBox._rawHashValue(_seed: seed)
   }
 }

stdlib/public/core/Array.swift

@@ -1836,7 +1836,7 @@ internal struct _ArrayAnyHashableBox<Element: Hashable>
     }
   }
 
-  func _rawHashValue(_seed: (UInt64, UInt64)) -> Int {
+  func _rawHashValue(_seed: Int) -> Int {
     var hasher = Hasher(_seed: _seed)
     self._hash(into: &hasher)
     return hasher._finalize()

stdlib/public/core/Dictionary.swift

@@ -1584,7 +1584,7 @@ internal struct _DictionaryAnyHashableBox<Key: Hashable, Value: Hashable>
     _canonical.hash(into: &hasher)
   }
 
-  internal func _rawHashValue(_seed: Hasher._Seed) -> Int {
+  internal func _rawHashValue(_seed: Int) -> Int {
     return _canonical._rawHashValue(seed: _seed)
   }
 

stdlib/public/core/DictionaryStorage.swift

@@ -44,7 +44,7 @@ internal class _RawDictionaryStorage: __SwiftNativeNSDictionary {
   internal final var _scale: Int
 
   @usableFromInline
-  internal final var _seed: Hasher._Seed
+  internal final var _seed: Int
 
   @usableFromInline
   @nonobjc
@@ -287,9 +287,8 @@ final internal class _DictionaryStorage<Key: Hashable, Value>
     // FIXME: Use true per-instance seeding instead. Per-capacity seeding still
     // leaves hash values the same in same-sized tables, which may affect
     // operations on two tables at once. (E.g., union.)
-    storage._seed = (
-      Hasher._seed.0 ^ UInt64(truncatingIfNeeded: scale),
-      Hasher._seed.1)
+    storage._seed = scale
+
     // Initialize hash table metadata.
     storage._hashTable.clear()
     return storage

stdlib/public/core/FloatingPointTypes.swift.gyb

@@ -1519,7 +1519,7 @@ extension ${Self} : Hashable {
   }
 
   @inlinable
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     // To satisfy the axiom that equality implies hash equality, we need to
     // finesse the hash value of -0.0 to match +0.0.
     let v = isZero ? 0 : self
@@ -1839,7 +1839,7 @@ internal struct _${Self}AnyHashableBox: _AnyHashableBox {
   }
 
   internal var _hashValue: Int {
-    return _rawHashValue(_seed: Hasher._seed)
+    return _rawHashValue(_seed: 0)
   }
 
   internal func _hash(into hasher: inout Hasher) {
@@ -1848,7 +1848,7 @@ internal struct _${Self}AnyHashableBox: _AnyHashableBox {
     hasher.combine(_value)
   }
 
-  internal func _rawHashValue(_seed: (UInt64, UInt64)) -> Int {
+  internal func _rawHashValue(_seed: Int) -> Int {
     var hasher = Hasher(_seed: _seed)
     _hash(into: &hasher)
     return hasher.finalize()

stdlib/public/core/Hashable.swift

@@ -126,17 +126,13 @@ public protocol Hashable : Equatable {
   // Raw top-level hashing interface. Some standard library types (mostly
   // primitives) specialize this to eliminate small resiliency overheads. (This
   // only matters for tiny keys.)
-  //
-  // FIXME(hasher): Change to take a Hasher instead. To achieve the same
-  // performance, this requires Set and Dictionary to store their fully
-  // initialized local hashers, not just their seeds.
-  func _rawHashValue(seed: (UInt64, UInt64)) -> Int
+  func _rawHashValue(seed: Int) -> Int
 }
 
 extension Hashable {
   @inlinable
   @inline(__always)
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     var hasher = Hasher(_seed: seed)
     hasher.combine(self)
     return hasher._finalize()
@@ -147,7 +143,7 @@ extension Hashable {
 @inlinable
 @inline(__always)
 public func _hashValue<H: Hashable>(for value: H) -> Int {
-  return value._rawHashValue(seed: Hasher._seed)
+  return value._rawHashValue(seed: 0)
 }
 
 // Called by the SwiftValue implementation.

stdlib/public/core/Hasher.swift

@@ -20,20 +20,27 @@ import SwiftShims
 // rdar://problem/38549901
 @usableFromInline
 internal protocol _HasherCore {
-  init(seed: Hasher._Seed)
+  init(rawSeed: (UInt64, UInt64))
   mutating func compress(_ value: UInt64)
   mutating func finalize(tailAndByteCount: UInt64) -> UInt64
+}
 
-  // FIXME(hasher): Remove once one-shot hashing has a hasher parameter.
-  /// Generate a seed value from the current state of this hasher.
-  ///
-  /// Note that the returned value is not same as the seed that was used to
-  /// initialize the hasher.
-  ///
-  /// This comes handy when type's _hash(into:) implementation needs to perform
-  /// one-shot hashing for some of its components. (E.g., for commutative
-  /// hashing.)
-  func _generateSeed() -> Hasher._Seed
+extension _HasherCore {
+  @inline(__always)
+  internal init() {
+    self.init(rawSeed: Hasher._executionSeed)
+  }
+
+  @inline(__always)
+  internal init(seed: Int) {
+    let executionSeed = Hasher._executionSeed
+    // Prevent sign-extending the supplied seed; this makes testing slightly
+    // easier.
+    let seed = UInt(bitPattern: seed)
+    self.init(rawSeed: (
+      executionSeed.0 ^ UInt64(truncatingIfNeeded: seed),
+      executionSeed.1))
+  }
 }
 
 @inline(__always)
@@ -155,14 +162,24 @@ internal struct _HasherTailBuffer {
 // FIXME: Remove @usableFromInline and @_fixed_layout once Hasher is resilient.
 // rdar://problem/38549901
 @usableFromInline @_fixed_layout
-internal struct _BufferingHasher<Core: _HasherCore> {
+internal struct _BufferingHasher<RawCore: _HasherCore> {
   private var _buffer: _HasherTailBuffer
-  private var _core: Core
+  private var _core: RawCore
 
   @inline(__always)
-  internal init(seed: Hasher._Seed) {
+  internal init(core: RawCore) {
     self._buffer = _HasherTailBuffer()
-    self._core = Core(seed: seed)
+    self._core = core
+  }
+
+  @inline(__always)
+  internal init() {
+    self.init(core: RawCore())
+  }
+
+  @inline(__always)
+  internal init(seed: Int) {
+    self.init(core: RawCore(seed: seed))
   }
 
   @inline(__always)
@@ -249,13 +266,6 @@ internal struct _BufferingHasher<Core: _HasherCore> {
     }
   }
 
-  // Generate a seed value from the current state of this hasher.
-  // FIXME(hasher): Remove
-  @inline(__always)
-  internal func _generateSeed() -> Hasher._Seed {
-    return _core._generateSeed()
-  }
-
   @inline(__always)
   internal mutating func finalize() -> UInt64 {
     return _core.finalize(tailAndByteCount: _buffer.value)
@@ -296,9 +306,6 @@ public struct Hasher {
   @usableFromInline
   internal typealias Core = _BufferingHasher<RawCore>
 
-  @usableFromInline
-  internal typealias _Seed = (UInt64, UInt64)
-
   internal var _core: Core
 
   /// Creates a new hasher.
@@ -307,14 +314,22 @@ public struct Hasher {
   /// startup, usually from a high-quality random source.
   @_effects(releasenone)
   public init() {
-    self._core = Core(seed: Hasher._seed)
+    self._core = Core()
   }
 
   /// Initialize a new hasher using the specified seed value.
+  /// The provided seed is mixed in with the global execution seed.
   @usableFromInline
   @_effects(releasenone)
-  internal init(_seed seed: _Seed) {
-    self._core = Core(seed: seed)
+  internal init(_seed: Int) {
+    self._core = Core(seed: _seed)
+  }
+
+  /// Initialize a new hasher using the specified seed value.
+  @usableFromInline // @testable
+  @_effects(releasenone)
+  internal init(_rawSeed: (UInt64, UInt64)) {
+    self._core = Core(core: RawCore(rawSeed: _rawSeed))
   }
 
   /// Indicates whether we're running in an environment where hashing needs to
@@ -333,8 +348,8 @@ public struct Hasher {
 
   /// The 128-bit hash seed used to initialize the hasher state. Initialized
   /// once during process startup.
-  @inlinable
-  internal static var _seed: _Seed {
+  @inlinable // @testable
+  internal static var _executionSeed: (UInt64, UInt64) {
     @inline(__always)
     get {
       // The seed itself is defined in C++ code so that it is initialized during
@@ -427,17 +442,9 @@ public struct Hasher {
     return Int(truncatingIfNeeded: core.finalize())
   }
 
-  // Generate a seed value from the current state of this hasher.
-  // FIXME(hasher): Remove
-  @_effects(readnone)
-  @usableFromInline
-  internal func _generateSeed() -> Hasher._Seed {
-    return _core._generateSeed()
-  }
-
   @_effects(readnone)
   @usableFromInline
-  internal static func _hash(seed: _Seed, _ value: UInt64) -> Int {
+  internal static func _hash(seed: Int, _ value: UInt64) -> Int {
     var core = RawCore(seed: seed)
     core.compress(value)
     let tbc = _HasherTailBuffer(tail: 0, byteCount: 8)
@@ -446,7 +453,7 @@ public struct Hasher {
 
   @_effects(readnone)
   @usableFromInline
-  internal static func _hash(seed: _Seed, _ value: UInt) -> Int {
+  internal static func _hash(seed: Int, _ value: UInt) -> Int {
     var core = RawCore(seed: seed)
 #if arch(i386) || arch(arm)
     _sanityCheck(UInt.bitWidth < UInt64.bitWidth)
@@ -464,7 +471,7 @@ public struct Hasher {
   @_effects(readnone)
   @usableFromInline
   internal static func _hash(
-    seed: _Seed,
+    seed: Int,
     bytes value: UInt64,
     count: Int) -> Int {
     _sanityCheck(count >= 0 && count < 8)
@@ -476,7 +483,7 @@ public struct Hasher {
   @_effects(readnone)
   @usableFromInline
   internal static func _hash(
-    seed: _Seed,
+    seed: Int,
     bytes: UnsafeRawBufferPointer) -> Int {
     var core = Core(seed: seed)
     core.combine(bytes: bytes)

stdlib/public/core/IntegerTypes.swift.gyb

@@ -1673,7 +1673,7 @@ extension ${Self} : Hashable {
   }
 
   @inlinable
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     % if bits == 64:
     return Hasher._hash(seed: seed, UInt64(_value))
     % elif bits == word_bits:
@@ -1905,7 +1905,7 @@ internal struct _IntegerAnyHashableBox<
     _value.hash(into: &hasher)
   }
 
-  internal func _rawHashValue(_seed: (UInt64, UInt64)) -> Int {
+  internal func _rawHashValue(_seed: Int) -> Int {
     _sanityCheck(Base.self == UInt64.self || Base.self == Int64.self,
       "self isn't canonical")
     return _value._rawHashValue(seed: _seed)

stdlib/public/core/NewtypeWrapper.swift

@@ -34,7 +34,7 @@ extension _SwiftNewtypeWrapper where Self: Hashable, Self.RawValue: Hashable {
   }
 
   @inlinable // FIXME(sil-serialize-all)
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     return rawValue._rawHashValue(seed: seed)
   }
 }
@@ -75,7 +75,7 @@ where Base: _SwiftNewtypeWrapper & Hashable, Base.RawValue: Hashable {
     _preconditionFailure("_hash(into:) called on non-canonical AnyHashable box")
   }
 
-  func _rawHashValue(_seed: (UInt64, UInt64)) -> Int {
+  func _rawHashValue(_seed: Int) -> Int {
     _preconditionFailure("_rawHashValue(_seed:) called on non-canonical AnyHashable box")
   }
 

stdlib/public/core/Pointer.swift

@@ -236,7 +236,7 @@ extension _Pointer /*: Hashable */ {
   }
 
   @inlinable
-  public func _rawHashValue(seed: (UInt64, UInt64)) -> Int {
+  public func _rawHashValue(seed: Int) -> Int {
     return Hasher._hash(seed: seed, UInt(bitPattern: self))
   }
 }