Use PRIx64 instead of PRIxPTR for uint64 in deep-heap-profile.

third_party/tcmalloc/chromium/src/deep-heap-profile.cc

Index: third_party/tcmalloc/chromium/src/deep-heap-profile.cc
diff --git a/third_party/tcmalloc/chromium/src/deep-heap-profile.cc b/third_party/tcmalloc/chromium/src/deep-heap-profile.cc
index 7201c8f2245d54b50efec136f36815f533140035..a2b1e35d13dd8ac81e7d0c747d700dd4d4aa4943 100644
--- a/third_party/tcmalloc/chromium/src/deep-heap-profile.cc
+++ b/third_party/tcmalloc/chromium/src/deep-heap-profile.cc
@@ -363,9 +363,9 @@ bool DeepHeapProfile::TextBuffer::AppendInt64(int64 v, int d) {
 bool DeepHeapProfile::TextBuffer::AppendPtr(uint64 v, int d) {

[Jeffrey Yasskin, 2013/05/10 00:52:08]

Another possible fix would be to change this to uintptr_t, but your change is
fine.

[Dai Mikurube (NOT FULLTIME), 2013/05/10 09:45:53]

Yeah, it might be better, but we decided to use uint64 for pointers since some
existing code in third_party tcmalloc use uint64 for pointers (e.g.
base/sysinfo.cc:ProcMaps-).

   int appended;
   if (d == 0)
-    appended = snprintf(buffer_ + cursor_, size_ - cursor_, "%"PRIxPTR, v);
+    appended = snprintf(buffer_ + cursor_, size_ - cursor_, "%"PRIx64, v);

[Jeffrey Yasskin, 2013/05/10 00:52:08]

We'll eventually, probably, need to add a space between the "%" and the macro to
avoid a collision with user-defined literals. No need to do this now, and it's
possible the language will fix the problem in general.

[Dai Mikurube (NOT FULLTIME), 2013/05/10 09:45:53]

Ok.

   else
-    appended = snprintf(buffer_ + cursor_, size_ - cursor_, "%0*"PRIxPTR, d, v);
+    appended = snprintf(buffer_ + cursor_, size_ - cursor_, "%0*"PRIx64, d, v);
   return ForwardCursor(appended);
 }