Run DCE after a LoopFlatten test to reduce spurious output [nfc]
[llvm-project.git] / llvm / test / MC / AMDGPU / hsa-v5-uses-dynamic-stack.s
blob6edac771faa07efacb616db1163049577b6ef374
1 // RUN: llvm-mc -triple amdgcn-amd-amdhsa -mcpu=gfx904 --amdhsa-code-object-version=5 -mattr=+xnack < %s | FileCheck --check-prefix=ASM %s
2 // RUN: llvm-mc -triple amdgcn-amd-amdhsa -mcpu=gfx904 --amdhsa-code-object-version=5 -mattr=+xnack -filetype=obj < %s > %t
3 // RUN: llvm-readelf -S -r -s %t | FileCheck --check-prefix=READOBJ %s
4 // RUN: llvm-objdump -s -j .rodata %t | FileCheck --check-prefix=OBJDUMP %s
6 // READOBJ: Section Headers
7 // READOBJ: .text PROGBITS {{[0-9a-f]+}} {{[0-9a-f]+}} {{[0-9a-f]+}} {{[0-9]+}} AX {{[0-9]+}} {{[0-9]+}} 256
8 // READOBJ: .rodata PROGBITS {{[0-9a-f]+}} {{[0-9a-f]+}} 000100 {{[0-9]+}} A {{[0-9]+}} {{[0-9]+}} 64
10 // READOBJ: Relocation section '.rela.rodata' at offset
11 // READOBJ: 0000000000000010 {{[0-9a-f]+}}00000005 R_AMDGPU_REL64 0000000000000000 .text + 10
12 // READOBJ: 0000000000000050 {{[0-9a-f]+}}00000005 R_AMDGPU_REL64 0000000000000000 .text + 110
13 // READOBJ: 0000000000000090 {{[0-9a-f]+}}00000005 R_AMDGPU_REL64 0000000000000000 .text + 210
14 // READOBJ: 00000000000000d0 {{[0-9a-f]+}}00000005 R_AMDGPU_REL64 0000000000000000 .text + 310
16 // READOBJ: Symbol table '.symtab' contains {{[0-9]+}} entries:
17 // READOBJ: 0000000000000000 0 FUNC LOCAL PROTECTED 2 minimal
18 // READOBJ-NEXT: 0000000000000100 0 FUNC LOCAL PROTECTED 2 complete
19 // READOBJ-NEXT: 0000000000000200 0 FUNC LOCAL PROTECTED 2 special_sgpr
20 // READOBJ-NEXT: 0000000000000300 0 FUNC LOCAL PROTECTED 2 disabled_user_sgpr
21 // READOBJ-NEXT: 0000000000000000 64 OBJECT LOCAL DEFAULT 3 minimal.kd
22 // READOBJ-NEXT: 0000000000000040 64 OBJECT LOCAL DEFAULT 3 complete.kd
23 // READOBJ-NEXT: 0000000000000080 64 OBJECT LOCAL DEFAULT 3 special_sgpr.kd
24 // READOBJ-NEXT: 00000000000000c0 64 OBJECT LOCAL DEFAULT 3 disabled_user_sgpr.kd
26 // OBJDUMP: Contents of section .rodata
27 // Note, relocation for KERNEL_CODE_ENTRY_BYTE_OFFSET is not resolved here.
28 // minimal
29 // OBJDUMP-NEXT: 0000 00000000 00000000 00000000 00000000
30 // OBJDUMP-NEXT: 0010 00000000 00000000 00000000 00000000
31 // OBJDUMP-NEXT: 0020 00000000 00000000 00000000 00000000
32 // OBJDUMP-NEXT: 0030 0000ac00 80000000 00000000 00000000
33 // complete
34 // OBJDUMP-NEXT: 0040 01000000 01000000 08000000 00000000
35 // OBJDUMP-NEXT: 0050 00000000 00000000 00000000 00000000
36 // OBJDUMP-NEXT: 0060 00000000 00000000 00000000 00000000
37 // OBJDUMP-NEXT: 0070 c2500104 1f0f007f 7f080000 00000000
38 // special_sgpr
39 // OBJDUMP-NEXT: 0080 00000000 00000000 00000000 00000000
40 // OBJDUMP-NEXT: 0090 00000000 00000000 00000000 00000000
41 // OBJDUMP-NEXT: 00a0 00000000 00000000 00000000 00000000
42 // OBJDUMP-NEXT: 00b0 00010000 80000000 00000000 00000000
43 // disabled_user_sgpr
44 // OBJDUMP-NEXT: 00c0 00000000 00000000 00000000 00000000
45 // OBJDUMP-NEXT: 00d0 00000000 00000000 00000000 00000000
46 // OBJDUMP-NEXT: 00e0 00000000 00000000 00000000 00000000
47 // OBJDUMP-NEXT: 00f0 0000ac00 80000000 00000000 00000000
49 .text
50 // ASM: .text
52 .amdgcn_target "amdgcn-amd-amdhsa--gfx904:xnack+"
53 // ASM: .amdgcn_target "amdgcn-amd-amdhsa--gfx904:xnack+"
55 .p2align 8
56 .type minimal,@function
57 minimal:
58 s_endpgm
60 .p2align 8
61 .type complete,@function
62 complete:
63 s_endpgm
65 .p2align 8
66 .type special_sgpr,@function
67 special_sgpr:
68 s_endpgm
70 .p2align 8
71 .type disabled_user_sgpr,@function
72 disabled_user_sgpr:
73 s_endpgm
75 .rodata
76 // ASM: .rodata
78 // Test that only specifying required directives is allowed, and that defaulted
79 // values are omitted.
80 .p2align 6
81 .amdhsa_kernel minimal
82 .amdhsa_next_free_vgpr 0
83 .amdhsa_next_free_sgpr 0
84 .end_amdhsa_kernel
86 // ASM: .amdhsa_kernel minimal
87 // ASM: .amdhsa_next_free_vgpr 0
88 // ASM-NEXT: .amdhsa_next_free_sgpr 0
89 // ASM: .end_amdhsa_kernel
91 // Test that we can specify all available directives with non-default values.
92 .p2align 6
93 .amdhsa_kernel complete
94 .amdhsa_group_segment_fixed_size 1
95 .amdhsa_private_segment_fixed_size 1
96 .amdhsa_kernarg_size 8
97 .amdhsa_user_sgpr_count 15
98 .amdhsa_user_sgpr_private_segment_buffer 1
99 .amdhsa_user_sgpr_dispatch_ptr 1
100 .amdhsa_user_sgpr_queue_ptr 1
101 .amdhsa_user_sgpr_kernarg_segment_ptr 1
102 .amdhsa_user_sgpr_dispatch_id 1
103 .amdhsa_user_sgpr_flat_scratch_init 1
104 .amdhsa_user_sgpr_private_segment_size 1
105 .amdhsa_uses_dynamic_stack 1
106 .amdhsa_system_sgpr_private_segment_wavefront_offset 1
107 .amdhsa_system_sgpr_workgroup_id_x 0
108 .amdhsa_system_sgpr_workgroup_id_y 1
109 .amdhsa_system_sgpr_workgroup_id_z 1
110 .amdhsa_system_sgpr_workgroup_info 1
111 .amdhsa_system_vgpr_workitem_id 1
112 .amdhsa_next_free_vgpr 9
113 .amdhsa_next_free_sgpr 27
114 .amdhsa_reserve_vcc 0
115 .amdhsa_reserve_flat_scratch 0
116 .amdhsa_reserve_xnack_mask 1
117 .amdhsa_float_round_mode_32 1
118 .amdhsa_float_round_mode_16_64 1
119 .amdhsa_float_denorm_mode_32 1
120 .amdhsa_float_denorm_mode_16_64 0
121 .amdhsa_dx10_clamp 0
122 .amdhsa_ieee_mode 0
123 .amdhsa_fp16_overflow 1
124 .amdhsa_exception_fp_ieee_invalid_op 1
125 .amdhsa_exception_fp_denorm_src 1
126 .amdhsa_exception_fp_ieee_div_zero 1
127 .amdhsa_exception_fp_ieee_overflow 1
128 .amdhsa_exception_fp_ieee_underflow 1
129 .amdhsa_exception_fp_ieee_inexact 1
130 .amdhsa_exception_int_div_zero 1
131 .end_amdhsa_kernel
133 // ASM: .amdhsa_kernel complete
134 // ASM-NEXT: .amdhsa_group_segment_fixed_size 1
135 // ASM-NEXT: .amdhsa_private_segment_fixed_size 1
136 // ASM-NEXT: .amdhsa_kernarg_size 8
137 // ASM-NEXT: .amdhsa_user_sgpr_count 15
138 // ASM-NEXT: .amdhsa_user_sgpr_private_segment_buffer 1
139 // ASM-NEXT: .amdhsa_user_sgpr_dispatch_ptr 1
140 // ASM-NEXT: .amdhsa_user_sgpr_queue_ptr 1
141 // ASM-NEXT: .amdhsa_user_sgpr_kernarg_segment_ptr 1
142 // ASM-NEXT: .amdhsa_user_sgpr_dispatch_id 1
143 // ASM-NEXT: .amdhsa_user_sgpr_flat_scratch_init 1
144 // ASM-NEXT: .amdhsa_user_sgpr_private_segment_size 1
145 // ASM-NEXT: .amdhsa_uses_dynamic_stack 1
146 // ASM-NEXT: .amdhsa_system_sgpr_private_segment_wavefront_offset 1
147 // ASM-NEXT: .amdhsa_system_sgpr_workgroup_id_x 0
148 // ASM-NEXT: .amdhsa_system_sgpr_workgroup_id_y 1
149 // ASM-NEXT: .amdhsa_system_sgpr_workgroup_id_z 1
150 // ASM-NEXT: .amdhsa_system_sgpr_workgroup_info 1
151 // ASM-NEXT: .amdhsa_system_vgpr_workitem_id 1
152 // ASM-NEXT: .amdhsa_next_free_vgpr 9
153 // ASM-NEXT: .amdhsa_next_free_sgpr 27
154 // ASM-NEXT: .amdhsa_reserve_vcc 0
155 // ASM-NEXT: .amdhsa_reserve_flat_scratch 0
156 // ASM-NEXT: .amdhsa_reserve_xnack_mask 1
157 // ASM-NEXT: .amdhsa_float_round_mode_32 1
158 // ASM-NEXT: .amdhsa_float_round_mode_16_64 1
159 // ASM-NEXT: .amdhsa_float_denorm_mode_32 1
160 // ASM-NEXT: .amdhsa_float_denorm_mode_16_64 0
161 // ASM-NEXT: .amdhsa_dx10_clamp 0
162 // ASM-NEXT: .amdhsa_ieee_mode 0
163 // ASM-NEXT: .amdhsa_fp16_overflow 1
164 // ASM-NEXT: .amdhsa_exception_fp_ieee_invalid_op 1
165 // ASM-NEXT: .amdhsa_exception_fp_denorm_src 1
166 // ASM-NEXT: .amdhsa_exception_fp_ieee_div_zero 1
167 // ASM-NEXT: .amdhsa_exception_fp_ieee_overflow 1
168 // ASM-NEXT: .amdhsa_exception_fp_ieee_underflow 1
169 // ASM-NEXT: .amdhsa_exception_fp_ieee_inexact 1
170 // ASM-NEXT: .amdhsa_exception_int_div_zero 1
171 // ASM-NEXT: .end_amdhsa_kernel
173 // Test that we are including special SGPR usage in the granulated count.
174 .p2align 6
175 .amdhsa_kernel special_sgpr
176 // Same next_free_sgpr as "complete", but...
177 .amdhsa_next_free_sgpr 27
178 // ...on GFX9 this should require an additional 6 SGPRs, pushing us from
179 // 3 granules to 4
180 .amdhsa_reserve_flat_scratch 1
182 .amdhsa_reserve_vcc 0
183 .amdhsa_reserve_xnack_mask 1
185 .amdhsa_float_denorm_mode_16_64 0
186 .amdhsa_dx10_clamp 0
187 .amdhsa_ieee_mode 0
188 .amdhsa_next_free_vgpr 0
189 .end_amdhsa_kernel
191 // ASM: .amdhsa_kernel special_sgpr
192 // ASM: .amdhsa_next_free_vgpr 0
193 // ASM-NEXT: .amdhsa_next_free_sgpr 27
194 // ASM-NEXT: .amdhsa_reserve_vcc 0
195 // ASM-NEXT: .amdhsa_reserve_xnack_mask 1
196 // ASM: .amdhsa_float_denorm_mode_16_64 0
197 // ASM-NEXT: .amdhsa_dx10_clamp 0
198 // ASM-NEXT: .amdhsa_ieee_mode 0
199 // ASM: .end_amdhsa_kernel
201 // Test that explicitly disabling user_sgpr's does not affect the user_sgpr
202 // count, i.e. this should produce the same descriptor as minimal.
203 .p2align 6
204 .amdhsa_kernel disabled_user_sgpr
205 .amdhsa_user_sgpr_private_segment_buffer 0
206 .amdhsa_next_free_vgpr 0
207 .amdhsa_next_free_sgpr 0
208 .end_amdhsa_kernel
210 // ASM: .amdhsa_kernel disabled_user_sgpr
211 // ASM: .amdhsa_next_free_vgpr 0
212 // ASM-NEXT: .amdhsa_next_free_sgpr 0
213 // ASM: .end_amdhsa_kernel
215 .section .foo
217 .byte .amdgcn.gfx_generation_number
218 // ASM: .byte 9
220 .byte .amdgcn.gfx_generation_minor
221 // ASM: .byte 0
223 .byte .amdgcn.gfx_generation_stepping
224 // ASM: .byte 4
226 .byte .amdgcn.next_free_vgpr
227 // ASM: .byte 0
228 .byte .amdgcn.next_free_sgpr
229 // ASM: .byte 0
231 v_mov_b32_e32 v7, s10
233 .byte .amdgcn.next_free_vgpr
234 // ASM: .byte 8
235 .byte .amdgcn.next_free_sgpr
236 // ASM: .byte 11
238 .set .amdgcn.next_free_vgpr, 0
239 .set .amdgcn.next_free_sgpr, 0
241 .byte .amdgcn.next_free_vgpr
242 // ASM: .byte 0
243 .byte .amdgcn.next_free_sgpr
244 // ASM: .byte 0
246 v_mov_b32_e32 v16, s3
248 .byte .amdgcn.next_free_vgpr
249 // ASM: .byte 17
250 .byte .amdgcn.next_free_sgpr
251 // ASM: .byte 4
253 // Metadata
255 .amdgpu_metadata
256 amdhsa.version:
259 amdhsa.kernels:
260 - .name: amd_kernel_code_t_test_all
261 .symbol: amd_kernel_code_t_test_all@kd
262 .kernarg_segment_size: 8
263 .group_segment_fixed_size: 16
264 .private_segment_fixed_size: 32
265 .uses_dynamic_stack: true
266 .kernarg_segment_align: 64
267 .wavefront_size: 128
268 .sgpr_count: 14
269 .vgpr_count: 40
270 .max_flat_workgroup_size: 256
271 - .name: amd_kernel_code_t_minimal
272 .symbol: amd_kernel_code_t_minimal@kd
273 .kernarg_segment_size: 8
274 .group_segment_fixed_size: 16
275 .private_segment_fixed_size: 32
276 .uses_dynamic_stack: true
277 .kernarg_segment_align: 64
278 .wavefront_size: 128
279 .sgpr_count: 14
280 .vgpr_count: 40
281 .max_flat_workgroup_size: 256
282 .end_amdgpu_metadata
284 // ASM: .amdgpu_metadata
285 // ASM: amdhsa.kernels:
286 // ASM: - .group_segment_fixed_size: 16
287 // ASM: .kernarg_segment_align: 64
288 // ASM: .kernarg_segment_size: 8
289 // ASM: .max_flat_workgroup_size: 256
290 // ASM: .name: amd_kernel_code_t_test_all
291 // ASM: .private_segment_fixed_size: 32
292 // ASM: .sgpr_count: 14
293 // ASM: .symbol: 'amd_kernel_code_t_test_all@kd'
294 // ASM: .uses_dynamic_stack: true
295 // ASM: .vgpr_count: 40
296 // ASM: .wavefront_size: 128
297 // ASM: - .group_segment_fixed_size: 16
298 // ASM: .kernarg_segment_align: 64
299 // ASM: .kernarg_segment_size: 8
300 // ASM: .max_flat_workgroup_size: 256
301 // ASM: .name: amd_kernel_code_t_minimal
302 // ASM: .private_segment_fixed_size: 32
303 // ASM: .sgpr_count: 14
304 // ASM: .symbol: 'amd_kernel_code_t_minimal@kd'
305 // ASM: .uses_dynamic_stack: true
306 // ASM: .vgpr_count: 40
307 // ASM: .wavefront_size: 128
308 // ASM: amdhsa.version:
309 // ASM-NEXT: - 3
310 // ASM-NEXT: - 0
311 // ASM: .end_amdgpu_metadata