Mac: Require child AllocateGpuMemoryBuffer to not fail

content/common/gpu/gpu_memory_buffer_factory_io_surface.cc

Index: content/common/gpu/gpu_memory_buffer_factory_io_surface.cc
diff --git a/content/common/gpu/gpu_memory_buffer_factory_io_surface.cc b/content/common/gpu/gpu_memory_buffer_factory_io_surface.cc
index f8a74d626e63c047b934b34e51e9d22825b0da37..5db66aa2cac1d6f8d17fad6b41056198d47bf18d 100644
--- a/content/common/gpu/gpu_memory_buffer_factory_io_surface.cc
+++ b/content/common/gpu/gpu_memory_buffer_factory_io_surface.cc
@@ -50,13 +50,12 @@ GpuMemoryBufferFactoryIOSurface::CreateGpuMemoryBuffer(
     gfx::PluginWindowHandle surface_handle) {
   base::ScopedCFTypeRef<IOSurfaceRef> io_surface(
       gfx::IOSurfaceManager::CreateIOSurface(size, format));
-  if (!io_surface)
-    return gfx::GpuMemoryBufferHandle();
+  CHECK(io_surface);

[reveman, 2015/12/01 20:49:04]

This makes it possible for a malicious renderer to crash the GPU process by
requesting an IOSurface to be allocated that is too large.

 
-  if (!gfx::IOSurfaceManager::GetInstance()->RegisterIOSurface(id, client_id,
-                                                               io_surface)) {
-    return gfx::GpuMemoryBufferHandle();
-  }
+  bool register_result =
+      gfx::IOSurfaceManager::GetInstance()->RegisterIOSurface(id, client_id,
+                                                              io_surface);
+  CHECK(register_result);

[reveman, 2015/12/01 20:49:04]

Is it possible that the browser decides to start a new gpu process before this
process has been shutdown? This can fail in that case without being a real
problem.

 
   {
     base::AutoLock lock(io_surfaces_lock_);