Workaround a gcc 4.4 bug....

src/heap.cc

 // Copyright 2009 the V8 project authors. All rights reserved.
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions are
 // met:
 //
 //     * Redistributions of source code must retain the above copyright
 //       notice, this list of conditions and the following disclaimer.
 //     * Redistributions in binary form must reproduce the above
 //       copyright notice, this list of conditions and the following
 //       disclaimer in the documentation and/or other materials provided
@@ skipping 1239 matching lines @@
   if (obj->IsFailure()) return false;
   Object* elements = AllocateFixedArray(2);
   if (elements->IsFailure()) return false;
   FixedArray::cast(elements)->set(0, Smi::FromInt(0));
   JSObject::cast(obj)->set_elements(FixedArray::cast(elements));
   message_listeners_ = JSObject::cast(obj);
 
   return true;
 }
 
+
+void Heap::CreateCEntryStub() {
+  CEntryStub stub;
+  c_entry_code_ = *stub.GetCode();
+}
+
+
+void Heap::CreateCEntryDebugBreakStub() {
+  CEntryDebugBreakStub stub;
+  c_entry_debug_break_code_ = *stub.GetCode();
+}
+
+
+void Heap::CreateJSEntryStub() {
+  JSEntryStub stub;
+  js_entry_code_ = *stub.GetCode();
+}
+
+
+void Heap::CreateJSConstructEntryStub() {
+  JSConstructEntryStub stub;
+  js_construct_entry_code_ = *stub.GetCode();
+}
+
+
 void Heap::CreateFixedStubs() {
   // Here we create roots for fixed stubs. They are needed at GC
   // for cooking and uncooking (check out frames.cc).
   // The eliminates the need for doing dictionary lookup in the
   // stub cache for these stubs.
   HandleScope scope;
-  {
-    CEntryStub stub;
-    c_entry_code_ = *stub.GetCode();
-  }
-  {
-    CEntryDebugBreakStub stub;
-    c_entry_debug_break_code_ = *stub.GetCode();
-  }
-  {
-    JSEntryStub stub;
-    js_entry_code_ = *stub.GetCode();
-  }
-  {
-    JSConstructEntryStub stub;
-    js_construct_entry_code_ = *stub.GetCode();
-  }
+  // gcc-4.4 has problem to generate the correct vtables if the following

[Erik Corry, 2009/06/24 07:38:31]

Should be "gcc-4.4 has problems generating..."

I don't understand the comment.  It doesn't seem like vtables are being
generated here.

+  // functions are inlined. e.g.,
+  // {  CEntryStub stub;
+  //    c_entry_code_ = *stub.GetCode();
+  // }
+  // {  CEntryDebugBreakStub stub;
+  //    c_entry_debug_break_code_ = *stub.GetCode();
+  // }
+  //
+  // Possible gcc bug: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32327

[Erik Corry, 2009/06/24 07:38:31]

Is this relevant?  According to the bug this was fixed in gcc-4.3.

+  Heap::CreateCEntryStub();
+  Heap::CreateCEntryDebugBreakStub();
+  Heap::CreateJSEntryStub();
+  Heap::CreateJSConstructEntryStub();
 }
 
 
 bool Heap::CreateInitialObjects() {
   Object* obj;
 
   // The -0 value must be set before NumberFromDouble works.
   obj = AllocateHeapNumber(-0.0, TENURED);
   if (obj->IsFailure()) return false;
   minus_zero_value_ = obj;
@@ skipping 2288 matching lines @@
 #ifdef DEBUG
 bool Heap::GarbageCollectionGreedyCheck() {
   ASSERT(FLAG_gc_greedy);
   if (Bootstrapper::IsActive()) return true;
   if (disallow_allocation_failure()) return true;
   return CollectGarbage(0, NEW_SPACE);
 }
 #endif
 
 } }  // namespace v8::internal