Add -reorder-basic-blocks option and fix nop insertion

src/IceTargetLoweringX86BaseImpl.h

Index: src/IceTargetLoweringX86BaseImpl.h
diff --git a/src/IceTargetLoweringX86BaseImpl.h b/src/IceTargetLoweringX86BaseImpl.h
index 77048b089c1a4c9934568f8dd973429138e36f9c..2bc2386bf82890dbe8d15243b59d677b7eb4714a 100644
--- a/src/IceTargetLoweringX86BaseImpl.h
+++ b/src/IceTargetLoweringX86BaseImpl.h
@@ -406,6 +406,12 @@ template <class Machine> void TargetX86Base<Machine>::translateO2() {
   Func->contractEmptyNodes();
   Func->reorderNodes();
 
+  // Shuffle basic block order if -reorder-basic-blocks is enabled.

[Jim Stichnoth, 2015/07/30 16:30:51]

Call Func->shuffleNodes() unconditionally, and check the flag and do the dump
inside that method.

[qining, 2015/07/30 20:25:55]

Done. Done the same with nop insertion.

+  if (Ctx->getFlags().shouldReorderBasicBlocks()) {
+    Func->shuffleNodes();
+    Func->dump("After basic block shuffling");
+  }
+
   // Branch optimization.  This needs to be done just before code emission.  In
   // particular, no transformations that insert or reorder CfgNodes should be
   // done after branch optimization.  We go ahead and do it before nop insertion
@@ -449,6 +455,12 @@ template <class Machine> void TargetX86Base<Machine>::translateOm1() {
     return;
   Func->dump("After stack frame mapping");
 
+  // Shuffle basic block order if -reorder-basic-blocks is enabled.
+  if (Ctx->getFlags().shouldReorderBasicBlocks()) {
+    Func->shuffleNodes();
+    Func->dump("After basic block shuffling");
+  }
+
   // Nop insertion
   if (Ctx->getFlags().shouldDoNopInsertion()) {
     Func->doNopInsertion();