[ARM] [FIX] Add missing f16 vector operations lowering

[dnsampaio]

Summary

Add missing <8xhalf> shufflevectors pattern, when using concat_vector dag node.
As well, allows <8xhalf> and <4xhalf> vldup1 operations.

These instructions are required for v8.2a fp16 lowering of vmul_n_f16, vmulq_n_f16 and vmulq_lane_f16 intrinsics.

[efriedma]

Do we not have some existing file we've been using for fp16 tests?

Code changes look fine.

[dnsampaio]

Moved tests to existing vector test file.

[efriedma]

LGTM

lib/Target/ARM/ARMISelDAGToDAG.cpp

void ARMDAGToDAGISel::SelectVLDDup(SDNode *N, bool IsIntrinsic,
   Align = CurDAG->getTargetConstant(Alignment, dl, MVT::i32);
 
   unsigned OpcodeIndex;
   switch (VT.getSimpleVT().SimpleTy) {
   default: llvm_unreachable("unhandled vld-dup type");
   case MVT::v8i8:
   case MVT::v16i8: OpcodeIndex = 0; break;
   case MVT::v4i16:
-  case MVT::v8i16: OpcodeIndex = 1; break;
+  case MVT::v8i16:
+  case MVT::v4f16:
+  case MVT::v8f16:
+                  OpcodeIndex = 1; break;
   case MVT::v2f32:
   case MVT::v2i32:
   case MVT::v4f32:
   case MVT::v4i32: OpcodeIndex = 2; break;
   case MVT::v1f64:
   case MVT::v1i64: OpcodeIndex = 3; break;
   }
 

lib/Target/ARM/ARMInstrNEON.td

 def : Pat<(v4i32 (concat_vectors DPR:$Dn, DPR:$Dm)),
           (REG_SEQUENCE QPR, DPR:$Dn, dsub_0, DPR:$Dm, dsub_1)>;
 def : Pat<(v8i16 (concat_vectors DPR:$Dn, DPR:$Dm)),
           (REG_SEQUENCE QPR, DPR:$Dn, dsub_0, DPR:$Dm, dsub_1)>;
 def : Pat<(v16i8 (concat_vectors DPR:$Dn, DPR:$Dm)),
           (REG_SEQUENCE QPR, DPR:$Dn, dsub_0, DPR:$Dm, dsub_1)>;
 def : Pat<(v4f32 (concat_vectors DPR:$Dn, DPR:$Dm)),
           (REG_SEQUENCE QPR, DPR:$Dn, dsub_0, DPR:$Dm, dsub_1)>;
+def : Pat<(v8f16 (concat_vectors DPR:$Dn, DPR:$Dm)),
+          (REG_SEQUENCE QPR, DPR:$Dn, dsub_0, DPR:$Dm, dsub_1)>;
 
 //===----------------------------------------------------------------------===//
 // Assembler aliases
 //
 
 def : VFP2InstAlias<"fmdhr${p} $Dd, $Rn",
                     (VSETLNi32 DPR:$Dd, GPR:$Rn, 1, pred:$p)>;
 def : VFP2InstAlias<"fmdlr${p} $Dd, $Rn",

test/CodeGen/ARM/armv8.2a-fp16-vector-intrinsics.ll

 }
 
 define dso_local <8 x half> @test_vrev64q_f16(<8 x half> %a) {
 entry:
   %shuffle.i = shufflevector <8 x half> %a, <8 x half> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
   ret <8 x half> %shuffle.i
 }
 
+define <4 x half> @test_vld_dup1_4xhalf(half* %b) #0 {
+; CHECK-LABEL: test_vld_dup1_4xhalf:
+; CHECK:       vld1.16 {d0[]}, [r0:16]
+; CHECK-NEXT:  bx      lr
+
+entry:
+  %b1 = load half, half* %b, align 2
+  %vecinit = insertelement <4 x half> undef, half %b1, i32 0
+  %vecinit2 = insertelement <4 x half> %vecinit, half %b1, i32 1
+  %vecinit3 = insertelement <4 x half> %vecinit2, half %b1, i32 2
+  %vecinit4 = insertelement <4 x half> %vecinit3, half %b1, i32 3
+  ret <4 x half> %vecinit4
+}
+
+define <8 x half> @test_vld_dup1_8xhalf(half* %b) local_unnamed_addr #0 {
+; CHECK-LABEL: test_vld_dup1_8xhalf:
+; CHECK:       vld1.16 {d0[], d1[]}, [r0:16]
+; CHECK-NEXT:  bx      lr
+
+entry:
+  %b1 = load half, half* %b, align 2
+  %vecinit = insertelement <8 x half> undef, half %b1, i32 0
+  %vecinit8 = shufflevector <8 x half> %vecinit, <8 x half> undef, <8 x i32> zeroinitializer
+  ret <8 x half> %vecinit8
+}
+
+define <8 x half> @test_shufflevector8xhalf(<4 x half> %a) #0 {
+; CHECK-LABEL: test_shufflevector8xhalf:
+; CHECK:       vmov.f64        d1, d0
+; CHECK-NEXT:  bx      lr
+
+entry:
+  %r = shufflevector <4 x half> %a, <4 x half> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  ret <8 x half> %r
+}
+
 declare <4 x half> @llvm.fabs.v4f16(<4 x half>)
 declare <8 x half> @llvm.fabs.v8f16(<8 x half>)
 declare <4 x i16> @llvm.arm.neon.vcvtas.v4i16.v4f16(<4 x half>)
 declare <4 x i16> @llvm.arm.neon.vcvtau.v4i16.v4f16(<4 x half>)
 declare <8 x i16> @llvm.arm.neon.vcvtas.v8i16.v8f16(<8 x half>)
 declare <4 x i16> @llvm.arm.neon.vcvtms.v4i16.v4f16(<4 x half>)
 declare <8 x i16> @llvm.arm.neon.vcvtms.v8i16.v8f16(<8 x half>)
 declare <4 x i16> @llvm.arm.neon.vcvtmu.v4i16.v4f16(<4 x half>)