Rename ValueRequiresCast to ShouldOptimizeCast, to better reflect

what it does.  Enhance it to return false to optimizing vector
sign extensions from vector comparisions, which is the idiom used
to get a splatted vector for a vector comparison.

Doing this breaks vector-casts.ll, add some compensating 
transformations to handle the important case they cover without
depending on this canonicalization.

This fixes rdar://7434900 a serious pessimization of vector compares.

llvm-svn: 95855

1 files changed, 6 insertions, 5 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstCombine.h b/llvm/lib/Transforms/InstCombine/InstCombine.h
index 536790004e8..09accb6b857 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombine.h
+++ b/llvm/lib/Transforms/InstCombine/InstCombine.h
@@ -199,11 +199,12 @@ private:
                                   SmallVectorImpl<Value*> &NewIndices);
   Instruction *FoldOpIntoSelect(Instruction &Op, SelectInst *SI);
                                  
-  /// ValueRequiresCast - Return true if the cast from "V to Ty" actually
-  /// results in any code being generated.  It does not require codegen if V is
-  /// simple enough or if the cast can be folded into other casts.
-  bool ValueRequiresCast(Instruction::CastOps opcode,const Value *V,
-                         const Type *Ty);
+  /// ShouldOptimizeCast - Return true if the cast from "V to Ty" actually
+  /// results in any code being generated and is interesting to optimize out. If
+  /// the cast can be eliminated by some other simple transformation, we prefer
+  /// to do the simplification first.
+  bool ShouldOptimizeCast(Instruction::CastOps opcode,const Value *V,
+                          const Type *Ty);
 
   Instruction *visitCallSite(CallSite CS);
   bool transformConstExprCastCall(CallSite CS);