Perf tweaks to compiling and oat writing.
Make hot quick compiler routines inlinable.
Remove computation/use of SSA strings.
Use vector insert when writing to the end of a vector in the output stream, to
avoid a memset followed by a memcpy.
Generating boot.oat/art these changes improve performance by around 2.5%.
Change-Id: I3d0bdb01333efe8f0eda4bdf97225e0b307f934d
diff --git a/src/compiler/dex/frontend.cc b/src/compiler/dex/frontend.cc
index ca751ab..e015645 100644
--- a/src/compiler/dex/frontend.cc
+++ b/src/compiler/dex/frontend.cc
@@ -18,7 +18,7 @@
#include "compiler/driver/compiler_driver.h"
#include "compiler_internals.h"
-#include "dataflow_iterator.h"
+#include "dataflow_iterator-inl.h"
#if defined(ART_USE_PORTABLE_COMPILER)
#include "compiler/llvm/llvm_compilation_unit.h"
#include "compiler/dex/portable/mir_to_gbc.h"