xref: /llvm-project/llvm/test/CodeGen/ARM/vlddup.ll (revision 2d790df1057be4acfe96d60aef5dfef3db9c4d9e)
1; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
2
3define <8 x i8> @vld1dupi8(i8* %A) nounwind {
4;CHECK: vld1dupi8:
5;Check the (default) alignment value.
6;CHECK: vld1.8 {d16[]}, [r0]
7	%tmp1 = load i8* %A, align 8
8	%tmp2 = insertelement <8 x i8> undef, i8 %tmp1, i32 0
9	%tmp3 = shufflevector <8 x i8> %tmp2, <8 x i8> undef, <8 x i32> zeroinitializer
10        ret <8 x i8> %tmp3
11}
12
13define <4 x i16> @vld1dupi16(i16* %A) nounwind {
14;CHECK: vld1dupi16:
15;Check the alignment value.  Max for this instruction is 16 bits:
16;CHECK: vld1.16 {d16[]}, [r0, :16]
17	%tmp1 = load i16* %A, align 8
18	%tmp2 = insertelement <4 x i16> undef, i16 %tmp1, i32 0
19	%tmp3 = shufflevector <4 x i16> %tmp2, <4 x i16> undef, <4 x i32> zeroinitializer
20        ret <4 x i16> %tmp3
21}
22
23define <2 x i32> @vld1dupi32(i32* %A) nounwind {
24;CHECK: vld1dupi32:
25;Check the alignment value.  Max for this instruction is 32 bits:
26;CHECK: vld1.32 {d16[]}, [r0, :32]
27	%tmp1 = load i32* %A, align 8
28	%tmp2 = insertelement <2 x i32> undef, i32 %tmp1, i32 0
29	%tmp3 = shufflevector <2 x i32> %tmp2, <2 x i32> undef, <2 x i32> zeroinitializer
30        ret <2 x i32> %tmp3
31}
32
33define <16 x i8> @vld1dupQi8(i8* %A) nounwind {
34;CHECK: vld1dupQi8:
35;Check the (default) alignment value.
36;CHECK: vld1.8 {d16[], d17[]}, [r0]
37	%tmp1 = load i8* %A, align 8
38	%tmp2 = insertelement <16 x i8> undef, i8 %tmp1, i32 0
39	%tmp3 = shufflevector <16 x i8> %tmp2, <16 x i8> undef, <16 x i32> zeroinitializer
40        ret <16 x i8> %tmp3
41}
42
43%struct.__neon_int8x8x2_t = type { <8 x i8>, <8 x i8> }
44%struct.__neon_int2x32x2_t = type { <2 x i32>, <2 x i32> }
45
46define <8 x i8> @vld2dupi8(i8* %A) nounwind {
47;CHECK: vld2dupi8:
48;Check the (default) alignment value.
49;CHECK: vld2.8 {d16[], d17[]}, [r0]
50	%tmp0 = tail call %struct.__neon_int8x8x2_t @llvm.arm.neon.vld2lane.v8i8(i8* %A, <8 x i8> undef, <8 x i8> undef, i32 0, i32 1)
51	%tmp1 = extractvalue %struct.__neon_int8x8x2_t %tmp0, 0
52	%tmp2 = shufflevector <8 x i8> %tmp1, <8 x i8> undef, <8 x i32> zeroinitializer
53	%tmp3 = extractvalue %struct.__neon_int8x8x2_t %tmp0, 1
54	%tmp4 = shufflevector <8 x i8> %tmp3, <8 x i8> undef, <8 x i32> zeroinitializer
55        %tmp5 = add <8 x i8> %tmp2, %tmp4
56        ret <8 x i8> %tmp5
57}
58
59define <2 x i32> @vld2dupi32(i32* %A) nounwind {
60;CHECK: vld2dupi32:
61;Check the alignment value.  Max for this instruction is 64 bits:
62;CHECK: vld2.32 {d16[], d17[]}, [r0, :64]
63	%tmp0 = tail call %struct.__neon_int2x32x2_t @llvm.arm.neon.vld2lane.v2i32(i32* %A, <2 x i32> undef, <2 x i32> undef, i32 0, i32 16)
64	%tmp1 = extractvalue %struct.__neon_int2x32x2_t %tmp0, 0
65	%tmp2 = shufflevector <2 x i32> %tmp1, <2 x i32> undef, <2 x i32> zeroinitializer
66	%tmp3 = extractvalue %struct.__neon_int2x32x2_t %tmp0, 1
67	%tmp4 = shufflevector <2 x i32> %tmp3, <2 x i32> undef, <2 x i32> zeroinitializer
68        %tmp5 = add <2 x i32> %tmp2, %tmp4
69        ret <2 x i32> %tmp5
70}
71
72declare %struct.__neon_int8x8x2_t @llvm.arm.neon.vld2lane.v8i8(i8*, <8 x i8>, <8 x i8>, i32, i32) nounwind readonly
73declare %struct.__neon_int2x32x2_t @llvm.arm.neon.vld2lane.v2i32(i32*, <2 x i32>, <2 x i32>, i32, i32) nounwind readonly
74