1 ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
2 ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s
3 ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx810 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s
6 define amdgpu_ps void @struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
7 ; CHECK-LABEL: name: struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
8 ; CHECK: bb.1 (%ir-block.0):
9 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
11 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
12 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
13 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
14 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
15 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
16 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
17 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
18 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
19 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
20 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
21 ; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 4)
22 ; CHECK-NEXT: S_ENDPGM 0
23 call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
27 define amdgpu_ps void @struct_buffer_store_v2f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
28 ; CHECK-LABEL: name: struct_buffer_store_v2f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
29 ; CHECK: bb.1 (%ir-block.0):
30 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
32 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
33 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
34 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
35 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
36 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
37 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
38 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
39 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
40 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
41 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
42 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
43 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
44 ; CHECK-NEXT: BUFFER_STORE_DWORDX2_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 4)
45 ; CHECK-NEXT: S_ENDPGM 0
46 call void @llvm.amdgcn.struct.buffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
50 define amdgpu_ps void @struct_buffer_store_v3f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
51 ; CHECK-LABEL: name: struct_buffer_store_v3f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
52 ; CHECK: bb.1 (%ir-block.0):
53 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
55 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
56 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
57 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
58 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2
59 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
60 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
61 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
62 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5
63 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3
64 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
65 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
66 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6
67 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
68 ; CHECK-NEXT: BUFFER_STORE_DWORDX3_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY9]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 4)
69 ; CHECK-NEXT: S_ENDPGM 0
70 call void @llvm.amdgcn.struct.buffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
74 define amdgpu_ps void @struct_buffer_store_v4f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
75 ; CHECK-LABEL: name: struct_buffer_store_v4f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
76 ; CHECK: bb.1 (%ir-block.0):
77 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
79 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
80 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
81 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
82 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
83 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
84 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
85 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
86 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
87 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5
88 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
89 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
90 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY $vgpr5
91 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:sreg_32 = COPY $sgpr6
92 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1
93 ; CHECK-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 4)
94 ; CHECK-NEXT: S_ENDPGM 0
95 call void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
99 define amdgpu_ps void @struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x float> inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
100 ; CHECK-LABEL: name: struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset
101 ; CHECK: bb.1 (%ir-block.0):
102 ; CHECK-NEXT: successors: %bb.2(0x80000000)
103 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
105 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
106 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
107 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
108 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
109 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
110 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
111 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
112 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
113 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
114 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
115 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
116 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr7
117 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY $vgpr4
118 ; CHECK-NEXT: [[COPY11:%[0-9]+]]:vreg_128 = COPY [[REG_SEQUENCE]]
119 ; CHECK-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[COPY8]]
120 ; CHECK-NEXT: [[COPY13:%[0-9]+]]:vgpr_32 = COPY [[COPY9]]
121 ; CHECK-NEXT: [[COPY14:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE1]].sub0_sub1
122 ; CHECK-NEXT: [[COPY15:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE1]].sub2_sub3
123 ; CHECK-NEXT: [[S_MOV_B64_:%[0-9]+]]:sreg_64_xexec = S_MOV_B64 $exec
126 ; CHECK-NEXT: successors: %bb.3(0x40000000), %bb.2(0x40000000)
128 ; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY14]].sub0, implicit $exec
129 ; CHECK-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY14]].sub1, implicit $exec
130 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
131 ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY14]], implicit $exec
132 ; CHECK-NEXT: [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY15]].sub0, implicit $exec
133 ; CHECK-NEXT: [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY15]].sub1, implicit $exec
134 ; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
135 ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE3]], [[COPY15]], implicit $exec
136 ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
137 ; CHECK-NEXT: [[REG_SEQUENCE4:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
138 ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY10]], implicit $exec
139 ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY10]], implicit $exec
140 ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U32_e64_]], [[S_AND_B64_]], implicit-def $scc
141 ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
142 ; CHECK-NEXT: [[REG_SEQUENCE5:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY12]], %subreg.sub0, [[COPY13]], %subreg.sub1
143 ; CHECK-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact [[COPY11]], [[REG_SEQUENCE5]], [[REG_SEQUENCE4]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 4)
144 ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
145 ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
148 ; CHECK-NEXT: successors: %bb.4(0x80000000)
150 ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
153 ; CHECK-NEXT: S_ENDPGM 0
154 call void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
158 define amdgpu_ps void @struct_buffer_store_i8_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
159 ; CHECK-LABEL: name: struct_buffer_store_i8_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
160 ; CHECK: bb.1 (%ir-block.0):
161 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
163 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
164 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
165 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
166 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
167 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
168 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
169 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
170 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
171 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
172 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
173 ; CHECK-NEXT: BUFFER_STORE_BYTE_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 4)
174 ; CHECK-NEXT: S_ENDPGM 0
175 %val.trunc = trunc i32 %val to i8
176 call void @llvm.amdgcn.struct.buffer.store.i8(i8 %val.trunc, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
180 define amdgpu_ps void @struct_buffer_store_i16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
181 ; CHECK-LABEL: name: struct_buffer_store_i16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
182 ; CHECK: bb.1 (%ir-block.0):
183 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
185 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
186 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
187 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
188 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
189 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
190 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
191 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
192 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
193 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
194 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
195 ; CHECK-NEXT: BUFFER_STORE_SHORT_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 4)
196 ; CHECK-NEXT: S_ENDPGM 0
197 %val.trunc = trunc i32 %val to i16
198 call void @llvm.amdgcn.struct.buffer.store.i16(i16 %val.trunc, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
202 define amdgpu_ps void @struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
203 ; CHECK-LABEL: name: struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc
204 ; CHECK: bb.1 (%ir-block.0):
205 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
207 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
208 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
209 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
210 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
211 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
212 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
213 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
214 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
215 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
216 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
217 ; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 4)
218 ; CHECK-NEXT: S_ENDPGM 0
219 call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 1)
223 define amdgpu_ps void @struct_buffer_store_v2f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
224 ; CHECK-LABEL: name: struct_buffer_store_v2f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
225 ; CHECK: bb.1 (%ir-block.0):
226 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
228 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
229 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
230 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
231 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
232 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
233 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
234 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
235 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
236 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
237 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
238 ; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 4)
239 ; CHECK-NEXT: S_ENDPGM 0
240 call void @llvm.amdgcn.struct.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
245 ; define amdgpu_ps void @struct_buffer_store_v3f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
246 ; call void @llvm.amdgcn.struct.buffer.store.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
250 define amdgpu_ps void @struct_buffer_store_v4f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
251 ; CHECK-LABEL: name: struct_buffer_store_v4f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
252 ; CHECK: bb.1 (%ir-block.0):
253 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
255 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
256 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
257 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
258 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
259 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
260 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
261 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
262 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
263 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
264 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
265 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
266 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
267 ; CHECK-NEXT: BUFFER_STORE_DWORDX2_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 4)
268 ; CHECK-NEXT: S_ENDPGM 0
269 call void @llvm.amdgcn.struct.buffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
273 declare void @llvm.amdgcn.struct.buffer.store.i8(i8, <4 x i32>, i32, i32, i32, i32 immarg)
274 declare void @llvm.amdgcn.struct.buffer.store.i16(i16, <4 x i32>, i32, i32, i32, i32 immarg)
276 declare void @llvm.amdgcn.struct.buffer.store.f16(half, <4 x i32>, i32, i32, i32, i32 immarg)
277 declare void @llvm.amdgcn.struct.buffer.store.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
278 declare void @llvm.amdgcn.struct.buffer.store.v3f16(<3 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
279 declare void @llvm.amdgcn.struct.buffer.store.v4f16(<4 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
281 declare void @llvm.amdgcn.struct.buffer.store.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)
282 declare void @llvm.amdgcn.struct.buffer.store.v2f32(<2 x float>, <4 x i32>, i32, i32, i32, i32 immarg)
283 declare void @llvm.amdgcn.struct.buffer.store.v3f32(<3 x float>, <4 x i32>, i32, i32, i32, i32 immarg)
284 declare void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float>, <4 x i32>, i32, i32, i32, i32 immarg)