[llvm-commits] [llvm] r128146 - in /llvm/trunk: lib/Transforms/Scalar/ScalarReplAggregates.cpp test/Transforms/ScalarRepl/vector_promote.ll

[Cameron Zwarich]

Author: zwarich
Date: Wed Mar 23 00:25:55 2011
New Revision: 128146

URL: http://llvm.org/viewvc/llvm-project?rev=128146&view=rev
Log:
Fix PR9464 by correcting some math that just happened to be right in most cases
that were hit in practice.

Modified:
    llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp
    llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll

Modified: llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp?rev=128146&r1=128145&r2=128146&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp Wed Mar 23 00:25:55 2011
@@ -654,18 +654,18 @@
 /// getScaledElementType - Gets a scaled element type for a partial vector
 /// access of an alloca. The input type must be an integer or float, and
 /// the resulting type must be an integer, float or double.
-static const Type *getScaledElementType(const Type *OldTy, unsigned Scale) {
+static const Type *getScaledElementType(const Type *OldTy,
+                                        unsigned NewBitWidth) {
   assert((OldTy->isIntegerTy() || OldTy->isFloatTy()) && "Partial vector "
          "accesses must be scaled from integer or float elements.");
 
   LLVMContext &Context = OldTy->getContext();
-  unsigned Size = OldTy->getPrimitiveSizeInBits() * Scale;
 
   if (OldTy->isIntegerTy())
-    return Type::getIntNTy(Context, Size);
-  if (Size == 32)
+    return Type::getIntNTy(Context, NewBitWidth);
+  if (NewBitWidth == 32)
     return Type::getFloatTy(Context);
-  if (Size == 64)
+  if (NewBitWidth == 64)
     return Type::getDoubleTy(Context);
 
   llvm_unreachable("Invalid type for a partial vector access of an alloca!");
@@ -703,9 +703,9 @@
                             "from a nonzero offset.");
 
       const Type *ToElementTy = cast<VectorType>(ToType)->getElementType();
-      unsigned Scale = AllocaSize / ToTypeSize;
-      const Type *CastElementTy = getScaledElementType(ToElementTy, Scale);
-      unsigned NumCastVectorElements = VTy->getNumElements() / Scale;
+      const Type *CastElementTy = getScaledElementType(ToElementTy,
+                                                       ToTypeSize * 8);
+      unsigned NumCastVectorElements = AllocaSize / ToTypeSize;
 
       LLVMContext &Context = FromVal->getContext();
       const Type *CastTy = VectorType::get(CastElementTy,
@@ -841,9 +841,8 @@
 
       const Type *ToElementTy =
         cast<VectorType>(SV->getType())->getElementType();
-      unsigned Scale = VecSize / ValSize;
-      const Type *CastElementTy = getScaledElementType(ToElementTy, Scale);
-      unsigned NumCastVectorElements = VTy->getNumElements() / Scale;
+      const Type *CastElementTy = getScaledElementType(ToElementTy, ValSize);
+      unsigned NumCastVectorElements = VecSize / ValSize;
 
       LLVMContext &Context = SV->getContext();
       const Type *OldCastTy = VectorType::get(CastElementTy,

Modified: llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll?rev=128146&r1=128145&r2=128146&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll (original)
+++ llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll Wed Mar 23 00:25:55 2011
@@ -171,3 +171,19 @@
 ; CHECK: @test11
 ; CHECK-NOT: alloca
 }
+
+define void @test12() {
+entry:
+  %a = alloca <64 x i8>, align 64
+  store <64 x i8> undef, <64 x i8>* %a, align 64
+  %p = bitcast <64 x i8>* %a to <16 x i8>*
+  %0 = load <16 x i8>* %p, align 64
+  store <16 x i8> undef, <16 x i8>* %p, align 64
+  %q = bitcast <16 x i8>* %p to <64 x i8>*
+  %1 = load <64 x i8>* %q, align 64
+  ret void
+; CHECK: @test12
+; CHECK-NOT: alloca
+; CHECK: extractelement <4 x i128>
+; CHECK: insertelement <4 x i128>
+}