1 ; Test the MSA element insertion intrinsics that are encoded with the ELM
4 ; RUN: llc -march=mips -mattr=+msa,+fp64 < %s | FileCheck %s
6 @llvm_mips_insert_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
7 @llvm_mips_insert_b_ARG3 = global i32 27, align 16
8 @llvm_mips_insert_b_RES = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
10 define void @llvm_mips_insert_b_test() nounwind {
12 %0 = load <16 x i8>* @llvm_mips_insert_b_ARG1
13 %1 = load i32* @llvm_mips_insert_b_ARG3
14 %2 = tail call <16 x i8> @llvm.mips.insert.b(<16 x i8> %0, i32 1, i32 %1)
15 store <16 x i8> %2, <16 x i8>* @llvm_mips_insert_b_RES
19 declare <16 x i8> @llvm.mips.insert.b(<16 x i8>, i32, i32) nounwind
21 ; CHECK: llvm_mips_insert_b_test:
22 ; CHECK-DAG: lw [[R1:\$[0-9]+]], 0(
23 ; CHECK-DAG: ld.b [[R2:\$w[0-9]+]], 0(
24 ; CHECK-DAG: insert.b [[R2]][1], [[R1]]
25 ; CHECK-DAG: st.b [[R2]], 0(
26 ; CHECK: .size llvm_mips_insert_b_test
28 @llvm_mips_insert_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
29 @llvm_mips_insert_h_ARG3 = global i32 27, align 16
30 @llvm_mips_insert_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
32 define void @llvm_mips_insert_h_test() nounwind {
34 %0 = load <8 x i16>* @llvm_mips_insert_h_ARG1
35 %1 = load i32* @llvm_mips_insert_h_ARG3
36 %2 = tail call <8 x i16> @llvm.mips.insert.h(<8 x i16> %0, i32 1, i32 %1)
37 store <8 x i16> %2, <8 x i16>* @llvm_mips_insert_h_RES
41 declare <8 x i16> @llvm.mips.insert.h(<8 x i16>, i32, i32) nounwind
43 ; CHECK: llvm_mips_insert_h_test:
44 ; CHECK-DAG: lw [[R1:\$[0-9]+]], 0(
45 ; CHECK-DAG: ld.h [[R2:\$w[0-9]+]], 0(
46 ; CHECK-DAG: insert.h [[R2]][1], [[R1]]
47 ; CHECK-DAG: st.h [[R2]], 0(
48 ; CHECK: .size llvm_mips_insert_h_test
50 @llvm_mips_insert_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
51 @llvm_mips_insert_w_ARG3 = global i32 27, align 16
52 @llvm_mips_insert_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
54 define void @llvm_mips_insert_w_test() nounwind {
56 %0 = load <4 x i32>* @llvm_mips_insert_w_ARG1
57 %1 = load i32* @llvm_mips_insert_w_ARG3
58 %2 = tail call <4 x i32> @llvm.mips.insert.w(<4 x i32> %0, i32 1, i32 %1)
59 store <4 x i32> %2, <4 x i32>* @llvm_mips_insert_w_RES
63 declare <4 x i32> @llvm.mips.insert.w(<4 x i32>, i32, i32) nounwind
65 ; CHECK: llvm_mips_insert_w_test:
66 ; CHECK-DAG: lw [[R1:\$[0-9]+]], 0(
67 ; CHECK-DAG: ld.w [[R2:\$w[0-9]+]], 0(
68 ; CHECK-DAG: insert.w [[R2]][1], [[R1]]
69 ; CHECK-DAG: st.w [[R2]], 0(
70 ; CHECK: .size llvm_mips_insert_w_test
72 @llvm_mips_insert_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
73 @llvm_mips_insert_d_ARG3 = global i64 27, align 16
74 @llvm_mips_insert_d_RES = global <2 x i64> <i64 0, i64 0>, align 16
76 define void @llvm_mips_insert_d_test() nounwind {
78 %0 = load <2 x i64>* @llvm_mips_insert_d_ARG1
79 %1 = load i64* @llvm_mips_insert_d_ARG3
80 %2 = tail call <2 x i64> @llvm.mips.insert.d(<2 x i64> %0, i32 1, i64 %1)
81 store <2 x i64> %2, <2 x i64>* @llvm_mips_insert_d_RES
85 declare <2 x i64> @llvm.mips.insert.d(<2 x i64>, i32, i64) nounwind
87 ; CHECK: llvm_mips_insert_d_test:
88 ; CHECK-DAG: lw [[R1:\$[0-9]+]], 0(
89 ; CHECK-DAG: lw [[R2:\$[0-9]+]], 4(
90 ; CHECK-DAG: ld.w [[R3:\$w[0-9]+]],
91 ; CHECK-DAG: insert.w [[R3]][2], [[R1]]
92 ; CHECK-DAG: insert.w [[R3]][3], [[R2]]
93 ; CHECK-DAG: st.w [[R3]],
94 ; CHECK: .size llvm_mips_insert_d_test
96 @llvm_mips_insve_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
97 @llvm_mips_insve_b_ARG3 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
98 @llvm_mips_insve_b_RES = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
100 define void @llvm_mips_insve_b_test() nounwind {
102 %0 = load <16 x i8>* @llvm_mips_insve_b_ARG1
103 %1 = load <16 x i8>* @llvm_mips_insve_b_ARG3
104 %2 = tail call <16 x i8> @llvm.mips.insve.b(<16 x i8> %0, i32 1, <16 x i8> %1)
105 store <16 x i8> %2, <16 x i8>* @llvm_mips_insve_b_RES
109 declare <16 x i8> @llvm.mips.insve.b(<16 x i8>, i32, <16 x i8>) nounwind
111 ; CHECK: llvm_mips_insve_b_test:
112 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_insve_b_ARG1)(
113 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_insve_b_ARG3)(
114 ; CHECK-DAG: ld.b [[R3:\$w[0-9]+]], 0([[R1]])
115 ; CHECK-DAG: ld.b [[R4:\$w[0-9]+]], 0([[R2]])
116 ; CHECK-DAG: insve.b [[R3]][1], [[R4]][0]
117 ; CHECK-DAG: st.b [[R3]],
118 ; CHECK: .size llvm_mips_insve_b_test
120 @llvm_mips_insve_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
121 @llvm_mips_insve_h_ARG3 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
122 @llvm_mips_insve_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
124 define void @llvm_mips_insve_h_test() nounwind {
126 %0 = load <8 x i16>* @llvm_mips_insve_h_ARG1
127 %1 = load <8 x i16>* @llvm_mips_insve_h_ARG3
128 %2 = tail call <8 x i16> @llvm.mips.insve.h(<8 x i16> %0, i32 1, <8 x i16> %1)
129 store <8 x i16> %2, <8 x i16>* @llvm_mips_insve_h_RES
133 declare <8 x i16> @llvm.mips.insve.h(<8 x i16>, i32, <8 x i16>) nounwind
135 ; CHECK: llvm_mips_insve_h_test:
136 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_insve_h_ARG1)(
137 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_insve_h_ARG3)(
138 ; CHECK-DAG: ld.h [[R3:\$w[0-9]+]], 0([[R1]])
139 ; CHECK-DAG: ld.h [[R4:\$w[0-9]+]], 0([[R2]])
140 ; CHECK-DAG: insve.h [[R3]][1], [[R4]][0]
141 ; CHECK-DAG: st.h [[R3]],
142 ; CHECK: .size llvm_mips_insve_h_test
144 @llvm_mips_insve_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
145 @llvm_mips_insve_w_ARG3 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
146 @llvm_mips_insve_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
148 define void @llvm_mips_insve_w_test() nounwind {
150 %0 = load <4 x i32>* @llvm_mips_insve_w_ARG1
151 %1 = load <4 x i32>* @llvm_mips_insve_w_ARG3
152 %2 = tail call <4 x i32> @llvm.mips.insve.w(<4 x i32> %0, i32 1, <4 x i32> %1)
153 store <4 x i32> %2, <4 x i32>* @llvm_mips_insve_w_RES
157 declare <4 x i32> @llvm.mips.insve.w(<4 x i32>, i32, <4 x i32>) nounwind
159 ; CHECK: llvm_mips_insve_w_test:
160 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_insve_w_ARG1)(
161 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_insve_w_ARG3)(
162 ; CHECK-DAG: ld.w [[R3:\$w[0-9]+]], 0([[R1]])
163 ; CHECK-DAG: ld.w [[R4:\$w[0-9]+]], 0([[R2]])
164 ; CHECK-DAG: insve.w [[R3]][1], [[R4]][0]
165 ; CHECK-DAG: st.w [[R3]],
166 ; CHECK: .size llvm_mips_insve_w_test
168 @llvm_mips_insve_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
169 @llvm_mips_insve_d_ARG3 = global <2 x i64> <i64 2, i64 3>, align 16
170 @llvm_mips_insve_d_RES = global <2 x i64> <i64 0, i64 0>, align 16
172 define void @llvm_mips_insve_d_test() nounwind {
174 %0 = load <2 x i64>* @llvm_mips_insve_d_ARG1
175 %1 = load <2 x i64>* @llvm_mips_insve_d_ARG3
176 %2 = tail call <2 x i64> @llvm.mips.insve.d(<2 x i64> %0, i32 1, <2 x i64> %1)
177 store <2 x i64> %2, <2 x i64>* @llvm_mips_insve_d_RES
181 declare <2 x i64> @llvm.mips.insve.d(<2 x i64>, i32, <2 x i64>) nounwind
183 ; CHECK: llvm_mips_insve_d_test:
184 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_insve_d_ARG1)(
185 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_insve_d_ARG3)(
186 ; CHECK-DAG: ld.d [[R3:\$w[0-9]+]], 0([[R1]])
187 ; CHECK-DAG: ld.d [[R4:\$w[0-9]+]], 0([[R2]])
188 ; CHECK-DAG: insve.d [[R3]][1], [[R4]][0]
189 ; CHECK-DAG: st.d [[R3]],
190 ; CHECK: .size llvm_mips_insve_d_test