Fixing remaining VC++ 2015 64-bit build breaks

third_party/WebKit/Source/platform/TraceEvent.h

Index: third_party/WebKit/Source/platform/TraceEvent.h
diff --git a/third_party/WebKit/Source/platform/TraceEvent.h b/third_party/WebKit/Source/platform/TraceEvent.h
index c7eda727580cb5d26954f95ce0d12cb2a9f95b91..db4c313a05237c92e29ae7d61785b9bb319d8132 100644
--- a/third_party/WebKit/Source/platform/TraceEvent.h
+++ b/third_party/WebKit/Source/platform/TraceEvent.h
@@ -269,8 +269,8 @@ public:
     typedef MangleBehavior<false> DontMangle;
     typedef MangleBehavior<true> ForceMangle;
 
-    TraceID(const void* id, unsigned* flags) :
-        m_data(static_cast<unsigned long long>(reinterpret_cast<unsigned long>(id)))
+    TraceID(const void* id, unsigned* flags)
+        : m_data(reinterpret_cast<uintptr_t>(id))

[Will Harris, 2015/11/10 18:05:14]

this looks like a behavior change, before the id was truncated to 32-bit even on
64-bit, now it's not.

[brucedawson, 2015/11/10 18:43:26]

Yep, it's a behavior change because the previous code was buggy. They copied
from base\trace_event.h which now does this correctly.

[Will Harris, 2015/11/10 19:45:46]

trace_event.h does

data_(static_cast<unsigned long long>(reinterpret_cast<uintptr_t>(id)))

can't you just match this by changing unsigned long to uintptr_t? It seems
strange to be inconsistent here.

[brucedawson, 2015/11/10 19:54:33]

I guess, although the whole idea of casting from uintptr_t to ulonglong is very
odd - why are we casting to an equal or larger type?

     {
         *flags |= TRACE_EVENT_FLAG_MANGLE_ID;
     }