1 ; Test the MSA intrinsics that are encoded with the 2R instruction format.
3 ; RUN: llc -march=mips -mattr=+msa,+fp64,+mips32r2 -relocation-model=pic < %s | FileCheck %s
4 ; RUN: llc -march=mipsel -mattr=+msa,+fp64,+mips32r2 -relocation-model=pic < %s | FileCheck %s
6 @llvm_mips_nloc_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_nloc_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
9 define void @llvm_mips_nloc_b_test() nounwind {
11 %0 = load <16 x i8>, ptr @llvm_mips_nloc_b_ARG1
12 %1 = tail call <16 x i8> @llvm.mips.nloc.b(<16 x i8> %0)
13 store <16 x i8> %1, ptr @llvm_mips_nloc_b_RES
17 declare <16 x i8> @llvm.mips.nloc.b(<16 x i8>) nounwind
19 ; CHECK: llvm_mips_nloc_b_test:
20 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nloc_b_ARG1)
21 ; CHECK-DAG: ld.b [[WS:\$w[0-9]+]], 0([[R1]])
22 ; CHECK-DAG: nloc.b [[WD:\$w[0-9]+]], [[WS]]
23 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nloc_b_RES)
24 ; CHECK-DAG: st.b [[WD]], 0([[R2]])
25 ; CHECK: .size llvm_mips_nloc_b_test
27 @llvm_mips_nloc_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
28 @llvm_mips_nloc_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
30 define void @llvm_mips_nloc_h_test() nounwind {
32 %0 = load <8 x i16>, ptr @llvm_mips_nloc_h_ARG1
33 %1 = tail call <8 x i16> @llvm.mips.nloc.h(<8 x i16> %0)
34 store <8 x i16> %1, ptr @llvm_mips_nloc_h_RES
38 declare <8 x i16> @llvm.mips.nloc.h(<8 x i16>) nounwind
40 ; CHECK: llvm_mips_nloc_h_test:
41 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nloc_h_ARG1)
42 ; CHECK-DAG: ld.h [[WS:\$w[0-9]+]], 0([[R1]])
43 ; CHECK-DAG: nloc.h [[WD:\$w[0-9]+]], [[WS]]
44 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nloc_h_RES)
45 ; CHECK-DAG: st.h [[WD]], 0([[R2]])
46 ; CHECK: .size llvm_mips_nloc_h_test
48 @llvm_mips_nloc_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
49 @llvm_mips_nloc_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
51 define void @llvm_mips_nloc_w_test() nounwind {
53 %0 = load <4 x i32>, ptr @llvm_mips_nloc_w_ARG1
54 %1 = tail call <4 x i32> @llvm.mips.nloc.w(<4 x i32> %0)
55 store <4 x i32> %1, ptr @llvm_mips_nloc_w_RES
59 declare <4 x i32> @llvm.mips.nloc.w(<4 x i32>) nounwind
61 ; CHECK: llvm_mips_nloc_w_test:
62 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nloc_w_ARG1)
63 ; CHECK-DAG: ld.w [[WS:\$w[0-9]+]], 0([[R1]])
64 ; CHECK-DAG: nloc.w [[WD:\$w[0-9]+]], [[WS]]
65 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nloc_w_RES)
66 ; CHECK-DAG: st.w [[WD]], 0([[R2]])
67 ; CHECK: .size llvm_mips_nloc_w_test
69 @llvm_mips_nloc_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
70 @llvm_mips_nloc_d_RES = global <2 x i64> <i64 0, i64 0>, align 16
72 define void @llvm_mips_nloc_d_test() nounwind {
74 %0 = load <2 x i64>, ptr @llvm_mips_nloc_d_ARG1
75 %1 = tail call <2 x i64> @llvm.mips.nloc.d(<2 x i64> %0)
76 store <2 x i64> %1, ptr @llvm_mips_nloc_d_RES
80 declare <2 x i64> @llvm.mips.nloc.d(<2 x i64>) nounwind
82 ; CHECK: llvm_mips_nloc_d_test:
83 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nloc_d_ARG1)
84 ; CHECK-DAG: ld.d [[WS:\$w[0-9]+]], 0([[R1]])
85 ; CHECK-DAG: nloc.d [[WD:\$w[0-9]+]], [[WS]]
86 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nloc_d_RES)
87 ; CHECK-DAG: st.d [[WD]], 0([[R2]])
88 ; CHECK: .size llvm_mips_nloc_d_test
90 @llvm_mips_nlzc_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
91 @llvm_mips_nlzc_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
93 define void @llvm_mips_nlzc_b_test() nounwind {
95 %0 = load <16 x i8>, ptr @llvm_mips_nlzc_b_ARG1
96 %1 = tail call <16 x i8> @llvm.mips.nlzc.b(<16 x i8> %0)
97 store <16 x i8> %1, ptr @llvm_mips_nlzc_b_RES
101 declare <16 x i8> @llvm.mips.nlzc.b(<16 x i8>) nounwind
103 ; CHECK: llvm_mips_nlzc_b_test:
104 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nlzc_b_ARG1)
105 ; CHECK-DAG: ld.b [[WS:\$w[0-9]+]], 0([[R1]])
106 ; CHECK-DAG: nlzc.b [[WD:\$w[0-9]+]], [[WS]]
107 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nlzc_b_RES)
108 ; CHECK-DAG: st.b [[WD]], 0([[R2]])
109 ; CHECK: .size llvm_mips_nlzc_b_test
111 @llvm_mips_nlzc_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
112 @llvm_mips_nlzc_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
114 define void @llvm_mips_nlzc_h_test() nounwind {
116 %0 = load <8 x i16>, ptr @llvm_mips_nlzc_h_ARG1
117 %1 = tail call <8 x i16> @llvm.mips.nlzc.h(<8 x i16> %0)
118 store <8 x i16> %1, ptr @llvm_mips_nlzc_h_RES
122 declare <8 x i16> @llvm.mips.nlzc.h(<8 x i16>) nounwind
124 ; CHECK: llvm_mips_nlzc_h_test:
125 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nlzc_h_ARG1)
126 ; CHECK-DAG: ld.h [[WS:\$w[0-9]+]], 0([[R1]])
127 ; CHECK-DAG: nlzc.h [[WD:\$w[0-9]+]], [[WS]]
128 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nlzc_h_RES)
129 ; CHECK-DAG: st.h [[WD]], 0([[R2]])
130 ; CHECK: .size llvm_mips_nlzc_h_test
132 @llvm_mips_nlzc_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
133 @llvm_mips_nlzc_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
135 define void @llvm_mips_nlzc_w_test() nounwind {
137 %0 = load <4 x i32>, ptr @llvm_mips_nlzc_w_ARG1
138 %1 = tail call <4 x i32> @llvm.mips.nlzc.w(<4 x i32> %0)
139 store <4 x i32> %1, ptr @llvm_mips_nlzc_w_RES
143 declare <4 x i32> @llvm.mips.nlzc.w(<4 x i32>) nounwind
145 ; CHECK: llvm_mips_nlzc_w_test:
146 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nlzc_w_ARG1)
147 ; CHECK-DAG: ld.w [[WS:\$w[0-9]+]], 0([[R1]])
148 ; CHECK-DAG: nlzc.w [[WD:\$w[0-9]+]], [[WS]]
149 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nlzc_w_RES)
150 ; CHECK-DAG: st.w [[WD]], 0([[R2]])
151 ; CHECK: .size llvm_mips_nlzc_w_test
153 @llvm_mips_nlzc_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
154 @llvm_mips_nlzc_d_RES = global <2 x i64> <i64 0, i64 0>, align 16
156 define void @llvm_mips_nlzc_d_test() nounwind {
158 %0 = load <2 x i64>, ptr @llvm_mips_nlzc_d_ARG1
159 %1 = tail call <2 x i64> @llvm.mips.nlzc.d(<2 x i64> %0)
160 store <2 x i64> %1, ptr @llvm_mips_nlzc_d_RES
164 declare <2 x i64> @llvm.mips.nlzc.d(<2 x i64>) nounwind
166 ; CHECK: llvm_mips_nlzc_d_test:
167 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_nlzc_d_ARG1)
168 ; CHECK-DAG: ld.d [[WS:\$w[0-9]+]], 0([[R1]])
169 ; CHECK-DAG: nlzc.d [[WD:\$w[0-9]+]], [[WS]]
170 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_nlzc_d_RES)
171 ; CHECK-DAG: st.d [[WD]], 0([[R2]])
172 ; CHECK: .size llvm_mips_nlzc_d_test
174 @llvm_mips_pcnt_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
175 @llvm_mips_pcnt_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
177 define void @llvm_mips_pcnt_b_test() nounwind {
179 %0 = load <16 x i8>, ptr @llvm_mips_pcnt_b_ARG1
180 %1 = tail call <16 x i8> @llvm.mips.pcnt.b(<16 x i8> %0)
181 store <16 x i8> %1, ptr @llvm_mips_pcnt_b_RES
185 declare <16 x i8> @llvm.mips.pcnt.b(<16 x i8>) nounwind
187 ; CHECK: llvm_mips_pcnt_b_test:
188 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_pcnt_b_ARG1)
189 ; CHECK-DAG: ld.b [[WS:\$w[0-9]+]], 0([[R1]])
190 ; CHECK-DAG: pcnt.b [[WD:\$w[0-9]+]], [[WS]]
191 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_pcnt_b_RES)
192 ; CHECK-DAG: st.b [[WD]], 0([[R2]])
193 ; CHECK: .size llvm_mips_pcnt_b_test
195 @llvm_mips_pcnt_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
196 @llvm_mips_pcnt_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
198 define void @llvm_mips_pcnt_h_test() nounwind {
200 %0 = load <8 x i16>, ptr @llvm_mips_pcnt_h_ARG1
201 %1 = tail call <8 x i16> @llvm.mips.pcnt.h(<8 x i16> %0)
202 store <8 x i16> %1, ptr @llvm_mips_pcnt_h_RES
206 declare <8 x i16> @llvm.mips.pcnt.h(<8 x i16>) nounwind
208 ; CHECK: llvm_mips_pcnt_h_test:
209 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_pcnt_h_ARG1)
210 ; CHECK-DAG: ld.h [[WS:\$w[0-9]+]], 0([[R1]])
211 ; CHECK-DAG: pcnt.h [[WD:\$w[0-9]+]], [[WS]]
212 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_pcnt_h_RES)
213 ; CHECK-DAG: st.h [[WD]], 0([[R2]])
214 ; CHECK: .size llvm_mips_pcnt_h_test
216 @llvm_mips_pcnt_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
217 @llvm_mips_pcnt_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
219 define void @llvm_mips_pcnt_w_test() nounwind {
221 %0 = load <4 x i32>, ptr @llvm_mips_pcnt_w_ARG1
222 %1 = tail call <4 x i32> @llvm.mips.pcnt.w(<4 x i32> %0)
223 store <4 x i32> %1, ptr @llvm_mips_pcnt_w_RES
227 declare <4 x i32> @llvm.mips.pcnt.w(<4 x i32>) nounwind
229 ; CHECK: llvm_mips_pcnt_w_test:
230 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_pcnt_w_ARG1)
231 ; CHECK-DAG: ld.w [[WS:\$w[0-9]+]], 0([[R1]])
232 ; CHECK-DAG: pcnt.w [[WD:\$w[0-9]+]], [[WS]]
233 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_pcnt_w_RES)
234 ; CHECK-DAG: st.w [[WD]], 0([[R2]])
235 ; CHECK: .size llvm_mips_pcnt_w_test
237 @llvm_mips_pcnt_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
238 @llvm_mips_pcnt_d_RES = global <2 x i64> <i64 0, i64 0>, align 16
240 define void @llvm_mips_pcnt_d_test() nounwind {
242 %0 = load <2 x i64>, ptr @llvm_mips_pcnt_d_ARG1
243 %1 = tail call <2 x i64> @llvm.mips.pcnt.d(<2 x i64> %0)
244 store <2 x i64> %1, ptr @llvm_mips_pcnt_d_RES
248 declare <2 x i64> @llvm.mips.pcnt.d(<2 x i64>) nounwind
250 ; CHECK: llvm_mips_pcnt_d_test:
251 ; CHECK-DAG: lw [[R1:\$[0-9]+]], %got(llvm_mips_pcnt_d_ARG1)
252 ; CHECK-DAG: ld.d [[WS:\$w[0-9]+]], 0([[R1]])
253 ; CHECK-DAG: pcnt.d [[WD:\$w[0-9]+]], [[WS]]
254 ; CHECK-DAG: lw [[R2:\$[0-9]+]], %got(llvm_mips_pcnt_d_RES)
255 ; CHECK-DAG: st.d [[WD]], 0([[R2]])
256 ; CHECK: .size llvm_mips_pcnt_d_test