Added compile-time define to add "initial-exec" attribute.

third_party/tcmalloc/chromium/src/thread_cache.h

 // Copyright (c) 2008, Google Inc.
 // 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
@@ skipping 241 matching lines @@
   // a thread is destroyed.
   // We also give a hint to the compiler to use the "initial exec" TLS
   // model.  This is faster than the default TLS model, at the cost that
   // you cannot dlopen this library.  (To see the difference, look at
   // the CPU use of __tls_get_addr with and without this attribute.)
   // Since we don't really use dlopen in google code -- and using dlopen
   // on a malloc replacement is asking for trouble in any case -- that's
   // a good tradeoff for us.
 #ifdef HAVE_TLS
   static __thread ThreadCache* threadlocal_heap_
-# ifdef HAVE___ATTRIBUTE__
+  // http://code.google.com/p/chromium/issues/detail?id=124489
+  // This code is being linked into pyautolib.so file for Chrome. A bug has
+  // been filed here:
+  // http://code.google.com/p/chromium/issues/detail?id=124489
+  // Ideally pyautolib.so should not link against this code. The current
+  // workaround is to pass in -DPGO_GENERATE.

[gpike, 2012/04/24 18:38:21]

Please improve the comment. Under what circumstances is -DPGO_GENERATE supposed
to be used? Is Chrome still going to get the performance benefit of
initial-exec?

+#if defined(HAVE___ATTRIBUTE__) && ! defined(PGO_GENERATE)

[Alexander Potapenko, 2012/04/24 10:38:15]

I think "!defined" is more common in Chromium.

[asharif1, 2012/04/24 17:50:12]

Done.

    __attribute__ ((tls_model ("initial-exec")))
 # endif
    ;
 #endif
 
   // Thread-specific key.  Initialization here is somewhat tricky
   // because some Linux startup code invokes malloc() before it
   // is in a good enough state to handle pthread_keycreate().
   // Therefore, we use TSD keys only after tsd_inited is set to true.
   // Until then, we use a slow path to get the heap object.
@@ skipping 139 matching lines @@
 // because we may be in the thread destruction code and may have
 // already cleaned up the cache for this thread.
 inline ThreadCache* ThreadCache::GetCacheIfPresent() {
   if (!tsd_inited_) return NULL;
   return GetThreadHeap();
 }
 
 }  // namespace tcmalloc
 
 #endif  // TCMALLOC_THREAD_CACHE_H_