Added a new "splitting" spiller.

When a call is placed to spill an interval this spiller will first try to
break the interval up into its component values. Single value intervals and
intervals which have already been split (or are the result of previous splits)
are spilled by the default spiller.

Splitting intervals as described above may improve the performance of generated
code in some circumstances. This work is experimental however, and it still
miscompiles many benchmarks. It's not recommended for general use yet.

llvm-svn: 90951

1 files changed, 3 insertions, 1 deletions

diff --git a/llvm/lib/CodeGen/Spiller.h b/llvm/lib/CodeGen/Spiller.h
index c6bd9857dba..dda52e871fe 100644
--- a/llvm/lib/CodeGen/Spiller.h
+++ b/llvm/lib/CodeGen/Spiller.h
@@ -21,6 +21,7 @@ namespace llvm {
   class MachineFunction;
   class MachineInstr;
   class MachineLoopInfo;
+  class SlotIndex;
   class VirtRegMap;
   class VNInfo;
 
@@ -35,7 +36,8 @@ namespace llvm {
     /// Spill the given live range. The method used will depend on the Spiller
     /// implementation selected.
     virtual std::vector<LiveInterval*> spill(LiveInterval *li,
-                                   SmallVectorImpl<LiveInterval*> &spillIs) = 0;
+					     SmallVectorImpl<LiveInterval*> &spillIs,
+                                             SlotIndex *earliestIndex = 0) = 0;
 
   };