Teach LowerBRCOND to recognize (xor (setcc x), 1). The xor inverts the condition. It's normally transformed by the dag combiner, unless the condition is set by a arithmetic op with overflow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@63505 91177308-0d34-0410-b5e6-96231b3b80d8

2 files changed, 44 insertions, 0 deletions

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 19d5aedf69..7590c2e4da 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -5366,6 +5366,17 @@ static bool isAndOrOfSetCCs(SDValue Op, unsigned &Opc) {
           Op.getOperand(1).hasOneUse());
 }
 
+static bool isXor1OfSetCC(SDValue Op) {
+  if (Op.getOpcode() != ISD::XOR)
+    return false;
+  ConstantSDNode *N1C = dyn_cast<ConstantSDNode>(Op.getOperand(1));
+  if (N1C && N1C->getAPIntValue() == 1) {
+    return Op.getOperand(0).getOpcode() == X86ISD::SETCC &&
+      Op.getOperand(0).hasOneUse();
+  }
+  return false;
+}
+
 SDValue X86TargetLowering::LowerBRCOND(SDValue Op, SelectionDAG &DAG) {
   bool addTest = true;
   SDValue Chain = Op.getOperand(0);
@@ -5460,6 +5471,16 @@ SDValue X86TargetLowering::LowerBRCOND(SDValue Op, SelectionDAG &DAG) {
           }
         }
       }
+    } else if (Cond.hasOneUse() && isXor1OfSetCC(Cond)) {
+      // Recognize for xorb (setcc), 1 patterns. The xor inverts the condition.
+      // It should be transformed during dag combiner except when the condition
+      // is set by a arithmetics with overflow node.
+      X86::CondCode CCode =
+        (X86::CondCode)Cond.getOperand(0).getConstantOperandVal(0);
+      CCode = X86::GetOppositeBranchCondition(CCode);
+      CC = DAG.getConstant(CCode, MVT::i8);
+      Cond = Cond.getOperand(0).getOperand(1);
+      addTest = false;
     }
   }
 
diff --git a/test/CodeGen/X86/smul-with-overflow-3.ll b/test/CodeGen/X86/smul-with-overflow-3.ll
new file mode 100644
index 0000000000..aa5e67a029
--- /dev/null
+++ b/test/CodeGen/X86/smul-with-overflow-3.ll
@@ -0,0 +1,23 @@
+; RUN: llvm-as < %s | llc -march=x86 | grep {jno} | count 1
+
+@ok = internal constant [4 x i8] c"%d\0A\00"
+@no = internal constant [4 x i8] c"no\0A\00"
+
+define i1 @func1(i32 %v1, i32 %v2) nounwind {
+entry:
+  %t = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %v1, i32 %v2)
+  %sum = extractvalue {i32, i1} %t, 0
+  %obit = extractvalue {i32, i1} %t, 1
+  br i1 %obit, label %overflow, label %normal
+
+overflow:
+  %t2 = tail call i32 (i8*, ...)* @printf( i8* getelementptr ([4 x i8]* @no, i32 0, i32 0) ) nounwind
+  ret i1 false
+
+normal:
+  %t1 = tail call i32 (i8*, ...)* @printf( i8* getelementptr ([4 x i8]* @ok, i32 0, i32 0), i32 %sum ) nounwind
+  ret i1 true
+}
+
+declare i32 @printf(i8*, ...) nounwind
+declare {i32, i1} @llvm.smul.with.overflow.i32(i32, i32)