Implement irreducible loop support in optimizing.

So we don't fallback to the interpreter in the presence of
irreducible loops.

Implications:
- A loop pre-header does not necessarily dominate a loop header.
- Non-constant redundant phis will be kept in loop headers, to
  satisfy our linear scan register allocation algorithm.
- while-graph optimizations, such as gvn, licm, lse, and dce
  need to know when they are dealing with irreducible loops.

Change-Id: I2cea8934ce0b40162d215353497c7f77d6c9137e

1 files changed, 2 insertions, 2 deletions

diff --git a/compiler/optimizing/ssa_builder.h b/compiler/optimizing/ssa_builder.h
index 0fcc3a1306..743dabde0f 100644
--- a/compiler/optimizing/ssa_builder.h
+++ b/compiler/optimizing/ssa_builder.h
@@ -49,7 +49,7 @@ static constexpr int kDefaultNumberOfLoops = 2;
  */
 class SsaBuilder : public HGraphVisitor {
  public:
-  explicit SsaBuilder(HGraph* graph, StackHandleScopeCollection* handles)
+  SsaBuilder(HGraph* graph, StackHandleScopeCollection* handles)
       : HGraphVisitor(graph),
         handles_(handles),
         agets_fixed_(false),
@@ -63,7 +63,7 @@ class SsaBuilder : public HGraphVisitor {
     loop_headers_.reserve(kDefaultNumberOfLoops);
   }
 
-  BuildSsaResult BuildSsa();
+  GraphAnalysisResult BuildSsa();
 
   // Returns locals vector for `block`. If it is a catch block, the vector will be
   // prepopulated with catch phis for vregs which are defined in `current_locals_`.