Changeset 26760 in webkit

Timestamp:

Oct 18, 2007, 7:14:09 PM (17 years ago)

Author:

mjs

Message:

Reviewed by Adam.

• use declspec(thread) for fast thread-local storage on Windows

• 2.2% speedup on sunspider (on Windows)
• 7% speedup on the string section
• 6% speedup on JS iBench

• fixed <rdar://problem/5473084> PLT on Windows got 2.5% slower between r25406 and r25422
• fixed at least some of <rdar://5527965? i-Bench JS was 14% slower in 310A11 than 310A10

• wtf/FastMalloc.cpp: (WTF::getThreadHeap): (WTF::setThreadHeap): (WTF::TCMalloc_ThreadCache::GetCache): (WTF::TCMalloc_ThreadCache::GetCacheIfPresent): (WTF::TCMalloc_ThreadCache::CreateCacheIfNecessary):

Location:

trunk/JavaScriptCore

Files:

• ChangeLog (modified) (1 diff)
• wtf/FastMalloc.cpp (modified) (6 diffs)

trunk/JavaScriptCore/ChangeLog

@@ +1 @@
+2007-10-18  Maciej Stachowiak  <mjs@apple.com>
+
+        Reviewed by Adam.
+        
+        - use __declspec(thread) for fast thread-local storage on Windows
+        
+        - 2.2% speedup on sunspider (on Windows)
+        - 7% speedup on the string section
+        - 6% speedup on JS iBench
+        
+        - fixed <rdar://problem/5473084> PLT on Windows got 2.5% slower between r25406 and r25422
+        - fixed at least some of <rdar://5527965? i-Bench JS was 14% slower in 310A11 than 310A10
+        
+        
+        * wtf/FastMalloc.cpp:
+        (WTF::getThreadHeap):
+        (WTF::setThreadHeap):
+        (WTF::TCMalloc_ThreadCache::GetCache):
+        (WTF::TCMalloc_ThreadCache::GetCacheIfPresent):
+        (WTF::TCMalloc_ThreadCache::CreateCacheIfNecessary):
+
 2007-10-17  Darin Adler  <darin@apple.com>
 

trunk/JavaScriptCore/wtf/FastMalloc.cpp

@@ -1185 +1185 @@
   static TCMalloc_ThreadCache* GetCache();
   static TCMalloc_ThreadCache* GetCacheIfPresent();
-  static void*                 CreateCacheIfNecessary();
+  static TCMalloc_ThreadCache* CreateCacheIfNecessary();
   static void                  DeleteCache(void* ptr);
   static void                  RecomputeThreadCacheSize();
@@ -1293 +1293 @@
 static bool tsd_inited = false;
 static pthread_key_t heap_key;
+#if COMPILER(MSVC)
+__declspec(thread) TCMalloc_ThreadCache* threadHeap;
+#endif
+
+static ALWAYS_INLINE TCMalloc_ThreadCache* getThreadHeap()
+{
+#if COMPILER(MSVC)
+    return threadHeap;
+#else
+    return static_cast<TCMalloc_ThreadCache*>(pthread_getspecific(heap_key));
+#endif
+}
+
+static ALWAYS_INLINE void setThreadHeap(TCMalloc_ThreadCache* heap)
+{
+    // still do pthread_setspecific when using MSVC fast TLS to
+    // benefit from the delete callback.
+    pthread_setspecific(heap_key, heap);
+#if COMPILER(MSVC)
+    threadHeap = heap;
+#endif
+}
 
 // Allocator for thread heaps
@@ -1558 +1580 @@
 
 ALWAYS_INLINE TCMalloc_ThreadCache* TCMalloc_ThreadCache::GetCache() {
-  void* ptr = NULL;
+  TCMalloc_ThreadCache* ptr = NULL;
   if (!tsd_inited) {
     InitModule();
   } else {
-    ptr = pthread_getspecific(heap_key);
+      ptr = getThreadHeap();
   }
   if (ptr == NULL) ptr = CreateCacheIfNecessary();
-  return static_cast<TCMalloc_ThreadCache*>(ptr);
+  return ptr;
 }
 
@@ -1573 +1595 @@
 inline TCMalloc_ThreadCache* TCMalloc_ThreadCache::GetCacheIfPresent() {
   if (!tsd_inited) return NULL;
-  return static_cast<TCMalloc_ThreadCache*>(pthread_getspecific(heap_key));
+  return getThreadHeap();
 }
 
@@ -1638 +1660 @@
 }
 
-void* TCMalloc_ThreadCache::CreateCacheIfNecessary() {
+TCMalloc_ThreadCache* TCMalloc_ThreadCache::CreateCacheIfNecessary() {
   // Initialize per-thread data if necessary
   TCMalloc_ThreadCache* heap = NULL;
@@ -1681 +1703 @@
   if (!heap->setspecific_ && tsd_inited) {
     heap->setspecific_ = true;
-    pthread_setspecific(heap_key, heap);
+    setThreadHeap(heap);
   }
   return heap;