SyntaxArena: thread safety

Use mutex to guard concurrent mutation to the arena.

#785
rdar://99812330

Author: rintaro

Sources/SwiftParser/Parser.swift

@@ -42,8 +42,10 @@ extension Parser {
     // Extended lifetime is required because `SyntaxArena` in the parser must
     // be alive until `Syntax(raw:)` retains the arena.
     return withExtendedLifetime(parser) {
-      let rawSourceFile =  parser.parseSourceFile()
-      return Syntax(raw: rawSourceFile.raw).as(SourceFileSyntax.self)!
+      parser.arena.assumingSingleThread {
+        let rawSourceFile =  parser.parseSourceFile()
+        return Syntax(raw: rawSourceFile.raw).as(SourceFileSyntax.self)!
+      }
     }
   }
 }

Sources/SwiftParser/Syntax+StringInterpolation.swift

@@ -73,7 +73,9 @@ extension SyntaxExpressibleByStringInterpolation {
       var parser = Parser(buffer)
       // FIXME: When the parser supports incremental parsing, put the
       // interpolatedSyntaxNodes in so we don't have to parse them again.
-      return Self.parse(from: &parser)
+      return parser.arena.assumingSingleThread {
+        return Self.parse(from: &parser)
+      }
     }
   }
 

Sources/SwiftSyntax/SyntaxArena.swift

@@ -10,11 +10,60 @@
 //
 //===----------------------------------------------------------------------===//
 
+#if canImport(Darwin)
+import Darwin
+
+class ScopeGuard {
+  private var lock: os_unfair_lock
+  init() {
+    self.lock = os_unfair_lock()
+  }
+  func withGuard<T>(body: () throws -> T) rethrows -> T {
+    os_unfair_lock_lock(&lock)
+    defer { os_unfair_lock_unlock(&lock)}
+    return try body()
+  }
+}
+
+#elseif canImport(Glibc)
+import Glibc
+
+class ScopeGuard {
+  var lock: pthread_mutex_t
+  init() {
+    self.lock = pthread_mutex_t()
+    pthread_mutex_init(&self.lock, nil)
+  }
+  deinit {
+    pthread_mutex_destroy(&self.lock)
+  }
+  func withGuard<T>(body: () throws -> T) rethrows -> T {
+    pthread_mutex_lock(&lock)
+    defer { pthread_mutex_unlock(&lock) }
+    return try body()
+  }
+}
+#else
+// FIXME: Support other platforms.
+
+/// Dummy mutex that doesn't actually guard at all.
+class ScopeGuard {
+  init() {}
+  func withGuard<T>(body: () throws -> T) rethrows -> T {
+    return try body()
+  }
+}
+#endif
+
 public class SyntaxArena {
 
   @_spi(RawSyntax)
   public typealias ParseTriviaFunction = (_ source: SyntaxText, _ position: TriviaPosition) -> [RawTriviaPiece]
 
+  /// Thread safe guard.
+  private let lock: ScopeGuard
+  private var singleThreadMode: Bool
+
   /// Bump-pointer allocator for all "intern" methods.
   private let allocator: BumpPtrAllocator
   /// Source file buffer the Syntax tree represents.
@@ -30,6 +79,8 @@ public class SyntaxArena {
 
   @_spi(RawSyntax)
   public init(parseTriviaFunction: @escaping ParseTriviaFunction) {
+    lock = ScopeGuard()
+    self.singleThreadMode = false
     allocator = BumpPtrAllocator()
     children = []
     sourceBuffer = .init(start: nil, count: 0)
@@ -41,15 +92,32 @@ public class SyntaxArena {
     self.init(parseTriviaFunction: _defaultParseTriviaFunction(_:_:))
   }
 
+  private func withGuard<R>(body: () throws -> R) rethrows -> R {
+    if self.singleThreadMode {
+      return try body()
+    } else {
+      return try self.lock.withGuard(body: body)
+    }
+  }
+
+  public func assumingSingleThread<R>(body: () throws -> R) rethrows -> R {
+    let oldValue = self.singleThreadMode
+    defer { self.singleThreadMode = oldValue }
+    self.singleThreadMode = true
+    return try body()
+  }
+
   /// Copies a source buffer in to the memory this arena manages, and returns
   /// the interned buffer.
   ///
   /// The interned buffer is guaranteed to be null-terminated.
   /// `contains(address _:)` is faster if the address is inside the memory
   /// range this function returned.
   public func internSourceBuffer(_ buffer: UnsafeBufferPointer<UInt8>) -> UnsafeBufferPointer<UInt8> {
+    let allocated = lock.withGuard {
+      allocator.allocate(UInt8.self, count: buffer.count + /* for NULL */1)
+    }
     precondition(sourceBuffer.baseAddress == nil, "SourceBuffer should only be set once.")
-    let allocated = allocator.allocate(UInt8.self, count: buffer.count + /* for NULL */1)
     _ = allocated.initialize(from: buffer)
 
     // NULL terminate.
@@ -69,20 +137,27 @@ public class SyntaxArena {
   /// Allocates a buffer of `RawSyntax?` with the given count, then returns the
   /// uninitlialized memory range as a `UnsafeMutableBufferPointer<RawSyntax?>`.
   func allocateRawSyntaxBuffer(count: Int) -> UnsafeMutableBufferPointer<RawSyntax?> {
-    return allocator.allocate(RawSyntax?.self, count: count)
+    return self.withGuard {
+      allocator.allocate(RawSyntax?.self, count: count)
+    }
   }
 
   /// Allcates a buffer of `RawTriviaPiece` with the given count, then returns
   /// the uninitialized memory range as a `UnsafeMutableBufferPointer<RawTriviaPiece>`.
   func allocateRawTriviaPieceBuffer(
-    count: Int) -> UnsafeMutableBufferPointer<RawTriviaPiece> {
-      return allocator.allocate(RawTriviaPiece.self, count: count)
+    count: Int
+  ) -> UnsafeMutableBufferPointer<RawTriviaPiece> {
+    return self.withGuard {
+      allocator.allocate(RawTriviaPiece.self, count: count)
     }
+  }
 
   /// Allcates a buffer of `UInt8` with the given count, then returns the
   /// uninitialized memory range as a `UnsafeMutableBufferPointer<UInt8>`.
   func allocateTextBuffer(count: Int) -> UnsafeMutableBufferPointer<UInt8> {
-    return allocator.allocate(UInt8.self, count: count)
+    return self.withGuard {
+      allocator.allocate(UInt8.self, count: count)
+    }
   }
 
   /// Copies the contents of a `SyntaxText` to the memory this arena manages,
@@ -114,7 +189,9 @@ public class SyntaxArena {
   /// Copies a `RawSyntaxData` to the memory this arena manages, and retuns the
   /// pointer to the destination.
   func intern(_ value: RawSyntaxData) -> UnsafePointer<RawSyntaxData> {
-    let allocated = allocator.allocate(RawSyntaxData.self, count: 1).baseAddress!
+    let allocated = lock.withGuard {
+      allocator.allocate(RawSyntaxData.self, count: 1).baseAddress!
+    }
     allocated.initialize(to: value)
     return UnsafePointer(allocated)
   }
@@ -128,21 +205,26 @@ public class SyntaxArena {
   /// See also `RawSyntax.layout()`.
   func addChild(_ arenaRef: SyntaxArenaRef) {
     if SyntaxArenaRef(self) == arenaRef { return }
-
     let other = arenaRef.value
 
-    precondition(
-      !self.hasParent,
-      "an arena can't have a new child once it's owned by other arenas")
-    
-    other.hasParent = true
-    children.insert(other)
+    other.withGuard {
+      self.withGuard {
+        precondition(
+          !self.hasParent,
+          "an arena can't have a new child once it's owned by other arenas")
+
+        other.hasParent = true
+        children.insert(other)
+      }
+    }
   }
 
   /// Recursively checks if this arena contains given `arena` as a descendant.
   func contains(arena: SyntaxArena) -> Bool {
-    return children.contains { child in
-      child === arena  || child.contains(arena: arena)
+    self.withGuard {
+      children.contains { child in
+        child === arena  || child.contains(arena: arena)
+      }
     }
   }
 

Tests/SwiftSyntaxTest/MultithreadingTests.swift

@@ -1,5 +1,6 @@
 import XCTest
-import SwiftSyntax
+@_spi(RawSyntax) import SwiftSyntax
+
 
 public class MultithreadingTests: XCTestCase {
 
@@ -15,6 +16,24 @@ public class MultithreadingTests: XCTestCase {
     }
   }
 
+  public func testConcurrentArena() {
+    let arena = SyntaxArena()
+
+    DispatchQueue.concurrentPerform(iterations: 100) { i in
+      var identStr = " ident\(i) "
+      let tokenRaw = identStr.withSyntaxText { text in
+        RawTokenSyntax(
+          kind: .identifier,
+          wholeText: arena.intern(text),
+          textRange: 1..<(text.count-1),
+          presence: .present,
+          arena: arena)
+      }
+      let token = Syntax(raw: RawSyntax(tokenRaw)).as(TokenSyntax.self)!
+      XCTAssertEqual(token.text, "ident\(i)")
+    }
+  }
+
   public func testTwoAccesses() {
     let tuple = TupleTypeSyntax(
       leftParen: .leftParenToken(),