[AArch64][SVE] Implement all-inactive predicate with PFALSE.

[sdesmalen]

Summary

Instead of using a WHILE XZR, XZR instruction, just emit a PFALSE.

[georges]

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

9649–9650

can remove this todo now?

[sdesmalen]

Removed TODO.

[sdesmalen]

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

9649–9650

good spot, thanks.

[david-arm]

LGTM! Truly a work of art. :)

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

SDValue AArch64TargetLowering::LowerSPLAT_VECTOR(SDValue Op,
 
   // Extend input splat value where needed to fit into a GPR (32b or 64b only)
   // FPRs don't have this restriction.
   switch (ElemVT.getSimpleVT().SimpleTy) {
   case MVT::i1: {
     // The only legal i1 vectors are SVE vectors, so we can use SVE-specific
     // lowering code.
     if (auto *ConstVal = dyn_cast<ConstantSDNode>(SplatVal)) {
+      if (ConstVal->isNullValue())
+        return SDValue(DAG.getMachineNode(AArch64::PFALSE, dl, VT), 0);
       if (ConstVal->isOne())
         return getPTrue(DAG, dl, VT, AArch64SVEPredPattern::all);
-      // TODO: Add special case for constant false
     }
     // The general case of i1.  There isn't any natural way to do this,
     // so we use some trickery with whilelo.
     SplatVal = DAG.getAnyExtOrTrunc(SplatVal, dl, MVT::i64);
     SplatVal = DAG.getNode(ISD::SIGN_EXTEND_INREG, dl, MVT::i64, SplatVal,
                            DAG.getValueType(MVT::i1));
     SDValue ID = DAG.getTargetConstant(Intrinsic::aarch64_sve_whilelo, dl,
                                        MVT::i64);
     return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, dl, VT, ID,

llvm/test/CodeGen/AArch64/sve-intrinsics-int-compares-with-imm.ll

; CHECK-NEXT: ret
   %out = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> %pg,
                                                                       <vscale x 16 x i8> %a,
                                                                       <vscale x 2 x i64> %splat)
   ret <vscale x 16 x i1> %out
 }
 
 define <vscale x 8 x i1> @ir_cmplo_h(<vscale x 8 x i16> %a) {
 ; CHECK-LABEL: ir_cmplo_h
-; CHECK: whilelo p0.h, xzr, xzr
+; CHECK: pfalse p0.b
 ; CHECK-NEXT: ret
   %elt   = insertelement <vscale x 8 x i16> undef, i16 0, i32 0
   %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
   %out = icmp ult <vscale x 8 x i16> %a, %splat
   ret <vscale x 8 x i1> %out
 }
 
 define <vscale x 8 x i1> @int_cmplo_h(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {

llvm/test/CodeGen/AArch64/sve-zeroinit.ll

 ; CHECK-LABEL: test_zeroinit_8xf16
 ; CHECK:       mov z0.h, #0
 ; CHECK-NEXT:  ret
   ret <vscale x 8 x half> zeroinitializer
 }
 
 define <vscale x 2 x i1> @test_zeroinit_2xi1() {
 ; CHECK-LABEL: test_zeroinit_2xi1
-; CHECK:       whilelo p0.d, xzr, xzr
+; CHECK:       pfalse p0.b
 ; CHECK-NEXT:  ret
   ret <vscale x 2 x i1> zeroinitializer
 }
 
 define <vscale x 4 x i1> @test_zeroinit_4xi1() {
 ; CHECK-LABEL: test_zeroinit_4xi1
-; CHECK:       whilelo p0.s, xzr, xzr
+; CHECK:       pfalse p0.b
 ; CHECK-NEXT:  ret
   ret <vscale x 4 x i1> zeroinitializer
 }
 
 define <vscale x 8 x i1> @test_zeroinit_8xi1() {
 ; CHECK-LABEL: test_zeroinit_8xi1
-; CHECK:       whilelo p0.h, xzr, xzr
+; CHECK:       pfalse p0.b
 ; CHECK-NEXT:  ret
   ret <vscale x 8 x i1> zeroinitializer
 }
 
 define <vscale x 16 x i1> @test_zeroinit_16xi1() {
 ; CHECK-LABEL: test_zeroinit_16xi1
-; CHECK:       whilelo p0.b, xzr, xzr
+; CHECK:       pfalse p0.b
 ; CHECK-NEXT:  ret
   ret <vscale x 16 x i1> zeroinitializer
 }