Fix code generation of materialized conditions.

Move the logic for knowing if a condition needs to be materialized
in an optimization pass (so that the information does not change
as a side effect of another optimization).

Also clean-up arm and x86_64 codegen:
- arm: ldr and str are for power-users when a constant is
  in play. We should use LoadFromOffset and StoreToOffset.
- x86_64: fix misuses of movq instead of movl.

Change-Id: I01a03b91803624be2281a344a13ad5efbf4f3ef3

1 files changed, 3 insertions, 0 deletions

diff --git a/compiler/optimizing/liveness_test.cc b/compiler/optimizing/liveness_test.cc
index 2d861696bb..246e7ef309 100644
--- a/compiler/optimizing/liveness_test.cc
+++ b/compiler/optimizing/liveness_test.cc
@@ -21,6 +21,7 @@
 #include "dex_instruction.h"
 #include "nodes.h"
 #include "optimizing_unit_test.h"
+#include "prepare_for_register_allocation.h"
 #include "ssa_liveness_analysis.h"
 #include "utils/arena_allocator.h"
 
@@ -50,6 +51,8 @@ static void TestCode(const uint16_t* data, const char* expected) {
   graph->BuildDominatorTree();
   graph->TransformToSSA();
   graph->FindNaturalLoops();
+  // `Inline` conditions into ifs.
+  PrepareForRegisterAllocation(graph).Run();
   x86::CodeGeneratorX86 codegen(graph);
   SsaLivenessAnalysis liveness(*graph, &codegen);
   liveness.Analyze();