[DAGCombiner] Fix ReplaceExtractVectorEltOfLoadWithNarrowedLoad for BE

For little-endian, the function would convert (extract_vector_elt (load X), Y)
to X + Y*sizeof(elt).  For big-endian it would instead use
X + sizeof(vec) - Y*sizeof(elt).  The big-endian case wasn't right since
vector index order always follows memory/array order, even for big-endian.
(Note that the current handling has to be wrong for Y==0 since it would
access beyond the end of the vector.)

Original patch by Richard Sandiford.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@236529 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 9319b81d00ec7655566b9355bcd302a0a217a1d9..4669d6c2f18dd5e77d93ff04a805508d963795d3 100644 (file)
--- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -11413,19 +11413,12 @@ SDValue DAGCombiner::ReplaceExtractVectorEltOfLoadWithNarrowedLoad(
   if (auto *ConstEltNo = dyn_cast<ConstantSDNode>(EltNo)) {
     int Elt = ConstEltNo->getZExtValue();
     unsigned PtrOff = VecEltVT.getSizeInBits() * Elt / 8;
-    if (TLI.isBigEndian())
-      PtrOff = InVecVT.getSizeInBits() / 8 - PtrOff;
     Offset = DAG.getConstant(PtrOff, DL, PtrType);
     MPI = OriginalLoad->getPointerInfo().getWithOffset(PtrOff);
   } else {
     Offset = DAG.getNode(
         ISD::MUL, DL, EltNo.getValueType(), EltNo,
         DAG.getConstant(VecEltVT.getStoreSize(), DL, EltNo.getValueType()));
-    if (TLI.isBigEndian())
-      Offset = DAG.getNode(
-          ISD::SUB, DL, EltNo.getValueType(),
-          DAG.getConstant(InVecVT.getStoreSize(), DL, EltNo.getValueType()),
-          Offset);
     MPI = OriginalLoad->getPointerInfo();
   }
   NewPtr = DAG.getNode(ISD::ADD, DL, PtrType, NewPtr, Offset);

diff --git a/test/CodeGen/SystemZ/vec-extract-01.ll b/test/CodeGen/SystemZ/vec-extract-01.ll
new file mode 100644 (file)
index 0000000..549392f
--- /dev/null
+++ b/test/CodeGen/SystemZ/vec-extract-01.ll
@@ -0,0 +1,13 @@
+; Verify ReplaceExtractVectorEltOfLoadWithNarrowedLoad fixes
+;
+; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
+
+; Test a memory copy of a v2i32 (via the constant pool).
+define void @f1(<2 x i32> *%dest) {
+; CHECK-LABEL: f1:
+; CHECK: lgrl [[REG:%r[0-5]]], {{[._A-Za-z0-9]}}
+; CHECK: stg [[REG]], 0(%r2)
+; CHECK: br %r14
+  store <2 x i32> <i32 1000000, i32 99999>, <2 x i32> *%dest
+  ret void
+}