[x86] Add the beginnings of a proper DAG combine to match ADDSUBPS and

ADDSUBPD nodes out of blends of adds and subs.

This allows us to actually form these instructions with SSE3 rather than
only forming them when we had both SSE3 for the ADDSUB instructions and
SSE4.1 for the blend instructions. ;] Kind-of important.

I've adjusted the CPU requirements on one of the tests to demonstrate
this kicking in nicely for an SSE3 cpu configuration.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@217848 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Target/X86/X86ISelLowering.cpp

@@ -19987,6 +19987,55 @@ static SDValue PerformTargetShuffleCombine(SDValue N, SelectionDAG &DAG,
   return SDValue();
 }
 
+/// \brief Try to combine a shuffle into a target-specific add-sub node.
+///
+/// We combine this directly on the abstract vector shuffle nodes so it is
+/// easier to generically match. We also insert dummy vector shuffle nodes for
+/// the operands which explicitly discard the lanes which are unused by this
+/// operation to try to flow through the rest of the combiner the fact that
+/// they're unused.
+static SDValue combineShuffleToAddSub(SDNode *N, SelectionDAG &DAG) {
+  SDLoc DL(N);
+
+  // We only handle target-independent shuffles.
+  // FIXME: It would be easy and harmless to use the target shuffle mask
+  // extraction tool to support more.
+  if (N->getOpcode() != ISD::VECTOR_SHUFFLE)
+    return SDValue();
+
+  auto *SVN = cast<ShuffleVectorSDNode>(N);
+  ArrayRef<int> Mask = SVN->getMask();
+  SDValue V1 = N->getOperand(0);
+  SDValue V2 = N->getOperand(1);
+
+  // We require the first shuffle operand to be the SUB node, and the second to
+  // be the ADD node.
+  // FIXME: We should support the commuted patterns.
+  if (V1->getOpcode() != ISD::FSUB || V2->getOpcode() != ISD::FADD)
+    return SDValue();
+
+  // If there are other uses of these operations we can't fold them.
+  if (!V1->hasOneUse() || !V2->hasOneUse())
+    return SDValue();
+
+  // Ensure that both operations have the same operands. Note that we can
+  // commute the FADD operands.
+  SDValue LHS = V1->getOperand(0), RHS = V1->getOperand(1);
+  if ((V2->getOperand(0) != LHS || V2->getOperand(1) != RHS) &&
+      (V2->getOperand(0) != RHS || V2->getOperand(1) != LHS))
+    return SDValue();
+
+  // We're looking for blends between FADD and FSUB nodes. We insist on these
+  // nodes being lined up in a specific expected pattern.
+  if (!isShuffleEquivalent(Mask, 0, 5, 2, 7))
+    return SDValue();
+
+  // FIXME: Munge the inputs through no-op shuffles that drop the undef lanes to
+  // allow nuking any instructions that feed only those lanes.
+
+  return DAG.getNode(X86ISD::ADDSUB, DL, N->getValueType(0), LHS, RHS);
+}
+
 /// PerformShuffleCombine - Performs several different shuffle combines.
 static SDValue PerformShuffleCombine(SDNode *N, SelectionDAG &DAG,
                                      TargetLowering::DAGCombinerInfo &DCI,
@@ -20001,6 +20050,12 @@ static SDValue PerformShuffleCombine(SDNode *N, SelectionDAG &DAG,
   if (!DCI.isBeforeLegalize() && !TLI.isTypeLegal(VT.getVectorElementType()))
     return SDValue();
 
+  // If we have legalized the vector types, look for blends of FADD and FSUB
+  // nodes that we can fuse into an ADDSUB node.
+  if (TLI.isTypeLegal(VT) && Subtarget->hasSSE3())
+    if (SDValue AddSub = combineShuffleToAddSub(N, DAG))
+      return AddSub;
+
   // Combine 256-bit vector shuffles. This is only profitable when in AVX mode
   if (Subtarget->hasFp256() && VT.is256BitVector() &&
       N->getOpcode() == ISD::VECTOR_SHUFFLE)

test/CodeGen/X86/sse3-avx-addsub.ll

@@ -1,4 +1,4 @@
-; RUN: llc < %s -march=x86-64 -mcpu=corei7 | FileCheck %s -check-prefix=SSE -check-prefix=CHECK
+; RUN: llc < %s -march=x86-64 -mcpu=core2 | FileCheck %s -check-prefix=SSE -check-prefix=CHECK
 ; RUN: llc < %s -march=x86-64 -mcpu=corei7-avx | FileCheck %s -check-prefix=AVX -check-prefix=CHECK
 
 ; Test ADDSUB ISel patterns.