Change stack alignment on linux to 16 bytes to keep gcc 4.4 happy....

src/platform-linux.cc

 // Copyright 2006-2008 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 70 matching lines @@
   srandom(static_cast<unsigned int>(seed));
 }
 
 
 double OS::nan_value() {
   return NAN;
 }
 
 
 int OS::ActivationFrameAlignment() {
-  // Floating point code runs faster if the stack is 8-byte aligned.
+#ifdef V8_TARGET_ARCH_ARM
   // On EABI ARM targets this is required for fp correctness in the
   // runtime system.
   return 8;
+#else
+  // With gcc 4.4 the tree vectorization optimiser can generate code
+  // that requires 16 byte alignment such as movdqa on x86.
+  return 16;
+#endif
 }
 
 
 // We keep the lowest and highest addresses mapped as a quick way of
 // determining that pointers are outside the heap (used mostly in assertions
 // and verification).  The estimate is conservative, ie, not all addresses in
 // 'allocated' space are actually allocated to our heap.  The range is
 // [lowest, highest), inclusive on the low and and exclusive on the high end.
 static void* lowest_ever_allocated = reinterpret_cast<void*>(-1);
 static void* highest_ever_allocated = reinterpret_cast<void*>(0);
@@ skipping 586 matching lines @@
   }
 
   // This sampler is no longer the active sampler.
   active_sampler_ = NULL;
   active_ = false;
 }
 
 #endif  // ENABLE_LOGGING_AND_PROFILING
 
 } }  // namespace v8::internal