gl_tests: Use zero size gl surface for offscreen rendering

gpu/command_buffer/client/fenced_allocator_test.cc

Index: gpu/command_buffer/client/fenced_allocator_test.cc
diff --git a/gpu/command_buffer/client/fenced_allocator_test.cc b/gpu/command_buffer/client/fenced_allocator_test.cc
index 3af9367a942eb28045e80e08f4700a525a0f10a4..6b9f00f41a3e769ad72c59b3033d5eb188ae0891 100644
--- a/gpu/command_buffer/client/fenced_allocator_test.cc
+++ b/gpu/command_buffer/client/fenced_allocator_test.cc
@@ -7,7 +7,6 @@
 #include "base/bind.h"
 #include "base/bind_helpers.h"
 #include "base/memory/aligned_memory.h"
-#include "base/message_loop/message_loop.h"
 #include "gpu/command_buffer/client/cmd_buffer_helper.h"
 #include "gpu/command_buffer/client/fenced_allocator.h"
 #include "gpu/command_buffer/service/cmd_buffer_engine.h"
@@ -17,10 +16,6 @@
 #include "gpu/command_buffer/service/transfer_buffer_manager.h"
 #include "testing/gtest/include/gtest/gtest.h"
 
-#if defined(OS_MACOSX)
-#include "base/mac/scoped_nsautorelease_pool.h"
-#endif
-
 namespace gpu {
 
 using testing::Return;
@@ -74,10 +69,6 @@ class BaseFencedAllocatorTest : public testing::Test {
     return command_buffer_->GetLastState().token;
   }
 
-#if defined(OS_MACOSX)
-  base::mac::ScopedNSAutoreleasePool autorelease_pool_;
-#endif
-  base::MessageLoop message_loop_;
   scoped_ptr<AsyncAPIMock> api_mock_;
   scoped_ptr<TransferBufferManagerInterface> transfer_buffer_manager_;
   scoped_ptr<CommandBufferService> command_buffer_;