Fixing remaining VC++ 2015 64-bit build breaks

components/startup_metric_utils/browser/startup_metric_utils.cc

Index: components/startup_metric_utils/browser/startup_metric_utils.cc
diff --git a/components/startup_metric_utils/browser/startup_metric_utils.cc b/components/startup_metric_utils/browser/startup_metric_utils.cc
index 920471640878449790df9bfa4b1837890440cc09..0d8bc8993ea0ecefa2974c9a21693abfa90dccf3 100644
--- a/components/startup_metric_utils/browser/startup_metric_utils.cc
+++ b/components/startup_metric_utils/browser/startup_metric_utils.cc
@@ -136,7 +136,7 @@ bool GetHardFaultCountForCurrentProcess(uint32_t* hard_fault_count,
     DCHECK_LE(index + sizeof(SYSTEM_PROCESS_INFORMATION_EX), buffer.size());
     SYSTEM_PROCESS_INFORMATION_EX* proc_info =
         reinterpret_cast<SYSTEM_PROCESS_INFORMATION_EX*>(buffer.data() + index);
-    if (reinterpret_cast<DWORD>(proc_info->UniqueProcessId) == proc_id) {
+    if (reinterpret_cast<uintptr_t>(proc_info->UniqueProcessId) == proc_id) {

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

UniqueProcessId is a HANDLE, shouldn't this be a double cast similar to others,
or is there an implicit cast from uintptr_t -> DWORD happening? No warnings for
that?

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

There's no truncation because proc_id is extended to uintptr_t size to do the
comparison. But that means this code is wrong if the sign bit is set, so I
should add another cast. It probably only affects the INVALID_HANDLE value case,
but still.

Yet another reason to have a truncation function or macro.

       *hard_fault_count = proc_info->HardFaultCount;
       return true;
     }