implement lvalue to rvalue conversion for ocuvector components. We can now compile stuff

like this:

typedef __attribute__(( ocu_vector_type(4) )) float float4;
float4 test1(float4 V) {
  return V.wzyx+V;
}

to:
_test1:
        pshufd  $27, %xmm0, %xmm1
        addps   %xmm0, %xmm1
        movaps  %xmm1, %xmm0
        ret

and:

_test1:
        mfspr r2, 256
        oris r3, r2, 4096
        mtspr 256, r3
        li r3, lo16(LCPI1_0)
        lis r4, ha16(LCPI1_0)
        lvx v3, r4, r3
        vperm v3, v2, v2, v3
        vaddfp v2, v3, v2
        mtspr 256, r2
        blr 

llvm-svn: 40771

2 files changed, 28 insertions, 1 deletions

diff --git a/clang/CodeGen/CGExpr.cpp b/clang/CodeGen/CGExpr.cpp
index 1a841f9d81d..7d8dc8c33b9 100644
--- a/clang/CodeGen/CGExpr.cpp
+++ b/clang/CodeGen/CGExpr.cpp
@@ -282,6 +282,30 @@ RValue CodeGenFunction::EmitLoadOfLValue(LValue LV, QualType ExprType) {
     return RValue::get(Builder.CreateExtractElement(Vec, LV.getVectorIdx(),
                                                     "vecext"));
   }
+
+  // If this is a reference to a subset of the elements of a vector, either
+  // shuffle the input or extract/insert them as appropriate.
+  if (LV.isOCUVectorComp()) {
+    llvm::Value *Vec = Builder.CreateLoad(LV.getOCUVectorAddr(), "tmp");
+    unsigned NumElts = cast<VectorType>(ExprType)->getNumElements();
+    
+    // Start out with an undef of the result type.
+    llvm::Value *Result = llvm::UndefValue::get(ConvertType(ExprType));
+    
+    unsigned EncFields = LV.getOCUVectorComp();
+    
+    // Extract/Insert each element of the result.
+    for (unsigned i = 0; i != NumElts; ++i) {
+      unsigned InIdx = OCUVectorComponent::getAccessedFieldNo(i, EncFields);
+      llvm::Value *Elt = llvm::ConstantInt::get(llvm::Type::Int32Ty, InIdx);
+      Elt = Builder.CreateExtractElement(Vec, Elt, "tmp");
+
+      llvm::Value *OutIdx = llvm::ConstantInt::get(llvm::Type::Int32Ty, i);
+      Result = Builder.CreateInsertElement(Result, Elt, OutIdx, "tmp");
+    }
+    
+    return RValue::get(Result);
+  }
   
   assert(0 && "Bitfield ref not impl!");
 }
@@ -505,6 +529,8 @@ RValue CodeGenFunction::EmitExpr(const Expr *E) {
     return EmitLoadOfLValue(E);
   case Expr::ArraySubscriptExprClass:
     return EmitArraySubscriptExprRV(cast<ArraySubscriptExpr>(E));
+  case Expr::OCUVectorComponentClass:
+    return EmitLoadOfLValue(E);
   case Expr::PreDefinedExprClass:
   case Expr::StringLiteralClass:
     return RValue::get(EmitLValue(E).getAddress());
diff --git a/clang/CodeGen/CodeGenFunction.h b/clang/CodeGen/CodeGenFunction.h
index 37614e95f06..919e68c91d0 100644
--- a/clang/CodeGen/CodeGenFunction.h
+++ b/clang/CodeGen/CodeGenFunction.h
@@ -141,6 +141,7 @@ public:
   llvm::Value *getVectorAddr() const { assert(isVectorElt()); return V; }
   llvm::Value *getVectorIdx() const { assert(isVectorElt()); return VectorIdx; }
   // ocu vector components.
+  llvm::Value *getOCUVectorAddr() const { assert(isOCUVectorComp()); return V; }
   unsigned getOCUVectorComp() const {
     assert(isOCUVectorComp());
     return VectorComp;
@@ -164,7 +165,7 @@ public:
   
   static LValue MakeOCUVectorComp(llvm::Value *Vec, unsigned Components) {
     LValue R;
-    R.LVType = VectorElt;
+    R.LVType = OCUVectorComp;
     R.V = Vec;
     R.VectorComp = Components;
     return R;