[DAGCombiner] Fix PR25763 - vector comparison constant folding + sign-extension

PR25763 demonstrated an issue with D14683 - vector comparison constant folding only works for i1 results, so we need to split off the sign-extension of the result to the required type. Luckily this can be done with the existing type legalization code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@255289 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index 771bb00d86ac0cc50c7ab3d2c837264f01a3fbc5..4596b8eba1a0d946980a73fdc2d838c164562c46 100644 (file)
--- a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -3338,12 +3338,15 @@ SDValue SelectionDAG::FoldConstantVectorArithmetic(unsigned Opcode, SDLoc DL,
       !std::all_of(Ops.begin(), Ops.end(), IsScalarOrSameVectorSize))
     return SDValue();
 
+  // If we are comparing vectors, then the result needs to be a i1 boolean
+  // that is then sign-extended back to the legal result type.
+  EVT SVT = (Opcode == ISD::SETCC ? MVT::i1 : VT.getScalarType());
+
   // Find legal integer scalar type for constant promotion and
   // ensure that its scalar size is at least as large as source.
-  EVT SVT = VT.getScalarType();
-  EVT LegalSVT = SVT;
-  if (SVT.isInteger()) {
-    LegalSVT = TLI->getTypeToTransformTo(*getContext(), SVT);
+  EVT LegalSVT = VT.getScalarType();
+  if (LegalSVT.isInteger()) {
+    LegalSVT = TLI->getTypeToTransformTo(*getContext(), LegalSVT);
     if (LegalSVT.bitsLT(SVT))
       return SDValue();
   }
@@ -3380,7 +3383,7 @@ SDValue SelectionDAG::FoldConstantVectorArithmetic(unsigned Opcode, SDLoc DL,
 
     // Legalize the (integer) scalar constant if necessary.
     if (LegalSVT != SVT)
-      ScalarResult = getNode(ISD::ANY_EXTEND, DL, LegalSVT, ScalarResult);
+      ScalarResult = getNode(ISD::SIGN_EXTEND, DL, LegalSVT, ScalarResult);
 
     // Scalar folding only succeeded if the result is a constant or UNDEF.
     if (ScalarResult.getOpcode() != ISD::UNDEF &&

diff --git a/test/CodeGen/AArch64/fold-constants.ll b/test/CodeGen/AArch64/fold-constants.ll
index 3f70f0a7e9f9e003c4bc6a4112156406f9f15ba0..c0fec4d171cd160c3828120323726006f02e25cc 100644 (file)
--- a/test/CodeGen/AArch64/fold-constants.ll
+++ b/test/CodeGen/AArch64/fold-constants.ll
@@ -16,3 +16,19 @@ entry:
   %vget_lane = extractelement <1 x i64> %4, i32 0
   ret i64 %vget_lane
 }
+
+; PR25763 - folding constant vector comparisons with sign-extended result
+define <8 x i16> @dotests_458() {
+; CHECK-LABEL: dotests_458
+; CHECK:       movi d0, #0x00000000ff0000
+; CHECK-NEXT:  sshll v0.8h, v0.8b, #0
+; CHECK-NEXT:  ret
+entry:
+  %vclz_v.i = call <8 x i8> @llvm.ctlz.v8i8(<8 x i8> <i8 127, i8 38, i8 -1, i8 -128, i8 127, i8 0, i8 0, i8 0>, i1 false) #6
+  %vsra_n = lshr <8 x i8> %vclz_v.i, <i8 5, i8 5, i8 5, i8 5, i8 5, i8 5, i8 5, i8 5>
+  %name_6 = or <8 x i8> %vsra_n, <i8 127, i8 -128, i8 -1, i8 67, i8 84, i8 127, i8 -1, i8 0>
+  %cmp.i603 = icmp slt <8 x i8> %name_6, <i8 -57, i8 -128, i8 127, i8 -128, i8 -1, i8 0, i8 -1, i8 -1>
+  %vmovl.i4.i = sext <8 x i1> %cmp.i603 to <8 x i16>
+  ret <8 x i16> %vmovl.i4.i
+}
+declare <8 x i8> @llvm.ctlz.v8i8(<8 x i8>, i1)