misched: add DAG edges from vreg defs to ExitSU.

These edges are not really necessary, but it is consistent with the
way we currently create physreg edges. Scheduler heuristics that
expect a DAG edge to the block terminator could benefit from this
change. Although in the future I hope we have a better mechanism for
modeling latency across scheduling regions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@152895 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/CodeGen/ScheduleDAGInstrs.cpp b/lib/CodeGen/ScheduleDAGInstrs.cpp
index 54828e2..1c455b9 100644
--- a/lib/CodeGen/ScheduleDAGInstrs.cpp
+++ b/lib/CodeGen/ScheduleDAGInstrs.cpp
@@ -201,8 +201,10 @@
 
       if (TRI->isPhysicalRegister(Reg))
         Uses[Reg].push_back(&ExitSU);
-      else
+      else {
         assert(!IsPostRA && "Virtual register encountered after regalloc.");
+        addVRegUseDeps(&ExitSU, i);
+      }
     }
   } else {
     // For others, e.g. fallthrough, conditional branch, assume the exit