[llvm] r217570 - Build correct vector filled with undef nodes

[David Xu]

Author: david.xu
Date: Thu Sep 11 00:10:28 2014
New Revision: 217570

URL: http://llvm.org/viewvc/llvm-project?rev=217570&view=rev
Log:
Build correct vector filled with undef nodes

Added:
    llvm/trunk/test/CodeGen/AArch64/aarch64_tree_tests.ll
Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=217570&r1=217569&r2=217570&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Thu Sep 11 00:10:28 2014
@@ -2664,9 +2664,17 @@ SDValue DAGCombiner::visitAND(SDNode *N)
 
     // fold (and x, 0) -> 0, vector edition
     if (ISD::isBuildVectorAllZeros(N0.getNode()))
-      return N0;
+      // do not return N0, because undef node may exist in N0
+      return DAG.getConstant(
+          APInt::getNullValue(
+              N0.getValueType().getScalarType().getSizeInBits()),
+          N0.getValueType());
     if (ISD::isBuildVectorAllZeros(N1.getNode()))
-      return N1;
+      // do not return N1, because undef node may exist in N1
+      return DAG.getConstant(
+          APInt::getNullValue(
+              N1.getValueType().getScalarType().getSizeInBits()),
+          N1.getValueType());
 
     // fold (and x, -1) -> x, vector edition
     if (ISD::isBuildVectorAllOnes(N0.getNode()))
@@ -3312,9 +3320,17 @@ SDValue DAGCombiner::visitOR(SDNode *N)
 
     // fold (or x, -1) -> -1, vector edition
     if (ISD::isBuildVectorAllOnes(N0.getNode()))
-      return N0;
+      // do not return N0, because undef node may exist in N0
+      return DAG.getConstant(
+          APInt::getAllOnesValue(
+              N0.getValueType().getScalarType().getSizeInBits()),
+          N0.getValueType());
     if (ISD::isBuildVectorAllOnes(N1.getNode()))
-      return N1;
+      // do not return N1, because undef node may exist in N1
+      return DAG.getConstant(
+          APInt::getAllOnesValue(
+              N1.getValueType().getScalarType().getSizeInBits()),
+          N1.getValueType());
 
     // fold (or (shuf A, V_0, MA), (shuf B, V_0, MB)) -> (shuf A, B, Mask1)
     // fold (or (shuf A, V_0, MA), (shuf B, V_0, MB)) -> (shuf B, A, Mask2)

Added: llvm/trunk/test/CodeGen/AArch64/aarch64_tree_tests.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AArch64/aarch64_tree_tests.ll?rev=217570&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/AArch64/aarch64_tree_tests.ll (added)
+++ llvm/trunk/test/CodeGen/AArch64/aarch64_tree_tests.ll Thu Sep 11 00:10:28 2014
@@ -0,0 +1,42 @@
+; RUN: llc < %s | FileCheck %s 
+
+; ModuleID = 'aarch64_tree_tests.bc'
+target datalayout = "e-m:e-i64:64-i128:128-n32:64-S128"
+target triple = "arm64--linux-gnu"
+
+; CHECK-LABLE: @aarch64_tree_tests_and
+; CHECK: .hword	32768                   
+; CHECK: .hword	32767                   
+; CHECK: .hword	4664                    
+; CHECK: .hword	32767                   
+; CHECK: .hword	32768                   
+; CHECK: .hword	32768                   
+; CHECK: .hword	0                       
+; CHECK: .hword	0                      
+
+; Function Attrs: nounwind readnone
+define <8 x i16> @aarch64_tree_tests_and(<8 x i16> %a) {
+entry:
+  %and = and <8 x i16> <i16 0, i16 undef, i16 undef, i16 0, i16 0, i16 undef, i16 undef, i16 0>, %a
+  %ret = add <8 x i16> %and, <i16 -32768, i16 32767, i16 4664, i16 32767, i16 -32768, i16 -32768, i16 0, i16 0>
+  ret <8 x i16> %ret
+}
+
+; CHECK-LABLE: @aarch64_tree_tests_or
+; CHECK: .hword	32768                 
+; CHECK: .hword	32766
+; CHECK: .hword	4664     
+; CHECK: .hword	32766                
+; CHECK: .hword	32768 
+; CHECK: .hword	32768
+; CHECK: .hword	65535            
+; CHECK: .hword	65535
+
+; Function Attrs: nounwind readnone
+define <8 x i16> @aarch64_tree_tests_or(<8 x i16> %a) {
+entry:
+  %or = or <8 x i16> <i16 -1, i16 undef, i16 undef, i16 -1, i16 -1, i16 undef, i16 undef, i16 -1>, %a
+  %ret = add <8 x i16> %or, <i16 -32767, i16 32767, i16 4665, i16 32767, i16 -32767, i16 -32767, i16 0, i16 0>
+  ret <8 x i16> %ret
+}
+