From e6ccd7b66a55363bbc57a7efcff1c514081a6269 Mon Sep 17 00:00:00 2001
From: Pirama Arumuga Nainar <pirama@google.com>
Date: Tue, 8 Dec 2015 23:07:06 +0000
Subject: [PATCH] Define selection for v4f16, v8f16 scalar_to_vector

Summary:
This fixes failure when trying to select
    insertelement <4 x half> undef, half %a, i64 0
which gets transformed to a scalar_to_vector node.

The accompanying v4 and v8 tests fail instruction selection without this
patch.

Reviewers: ab, jmolloy

Subscribers: srhines, llvm-commits

Differential Revision: http://reviews.llvm.org/D15322

llvm-svn: 255072
---
 llvm/lib/Target/AArch64/AArch64InstrInfo.td       | 5 +++++
 llvm/test/CodeGen/AArch64/fp16-v4-instructions.ll | 9 +++++++++
 llvm/test/CodeGen/AArch64/fp16-v8-instructions.ll | 9 +++++++++
 3 files changed, 23 insertions(+)

diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index cfb0c1b..ed6e171 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -3843,6 +3843,11 @@ def : Pat<(v2i64 (scalar_to_vector (i64 FPR64:$Rn))),
             (v2i64 (INSERT_SUBREG (v2i64 (IMPLICIT_DEF)),
                                   (i64 FPR64:$Rn), dsub))>;
 
+def : Pat<(v4f16 (scalar_to_vector (f16 FPR16:$Rn))),
+          (INSERT_SUBREG (v4f16 (IMPLICIT_DEF)), FPR16:$Rn, hsub)>;
+def : Pat<(v8f16 (scalar_to_vector (f16 FPR16:$Rn))),
+          (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), FPR16:$Rn, hsub)>;
+
 def : Pat<(v4f32 (scalar_to_vector (f32 FPR32:$Rn))),
           (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FPR32:$Rn, ssub)>;
 def : Pat<(v2f32 (scalar_to_vector (f32 FPR32:$Rn))),
diff --git a/llvm/test/CodeGen/AArch64/fp16-v4-instructions.ll b/llvm/test/CodeGen/AArch64/fp16-v4-instructions.ll
index 0dbda15..1249c14 100644
--- a/llvm/test/CodeGen/AArch64/fp16-v4-instructions.ll
+++ b/llvm/test/CodeGen/AArch64/fp16-v4-instructions.ll
@@ -218,4 +218,13 @@ define <4 x half> @uitofp_i64(<4 x i64> %a) #0 {
   ret <4 x half> %1
 }
 
+define void @test_insert_at_zero(half %a, <4 x half>* %b) #0 {
+; CHECK-LABEL: test_insert_at_zero:
+; CHECK-NEXT: str d0, [x0]
+; CHECK-NEXT: ret
+  %1 = insertelement <4 x half> undef, half %a, i64 0
+  store <4 x half> %1, <4 x half>* %b, align 4
+  ret void
+}
+
 attributes #0 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/fp16-v8-instructions.ll b/llvm/test/CodeGen/AArch64/fp16-v8-instructions.ll
index 10a8c22..dfad6bc 100644
--- a/llvm/test/CodeGen/AArch64/fp16-v8-instructions.ll
+++ b/llvm/test/CodeGen/AArch64/fp16-v8-instructions.ll
@@ -358,4 +358,13 @@ define <8 x half> @uitofp_i64(<8 x i64> %a) #0 {
   ret <8 x half> %1
 }
 
+define void @test_insert_at_zero(half %a, <8 x half>* %b) #0 {
+; CHECK-LABEL: test_insert_at_zero:
+; CHECK-NEXT: str q0, [x0]
+; CHECK-NEXT: ret
+  %1 = insertelement <8 x half> undef, half %a, i64 0
+  store <8 x half> %1, <8 x half>* %b, align 4
+  ret void
+}
+
 attributes #0 = { nounwind }
-- 
2.7.4