1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
|
// RUN: echo -n "GPU binary would be here." > %t
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=8.0 -fcuda-include-gpubinary %t -o - \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s \
// RUN: --check-prefixes=ALL,LNX,NORDC,CUDA,CUDANORDC,CUDA-OLD
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=8.0 -fcuda-include-gpubinary %t \
// RUN: -o - -DNOGLOBALS \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s \
// RUN: -check-prefixes=NOGLOBALS,CUDANOGLOBALS
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=8.0 -fgpu-rdc -fcuda-include-gpubinary %t \
// RUN: -o - \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s \
// RUN: --check-prefixes=ALL,LNX,RDC,CUDA,CUDARDC,CUDA-OLD
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=8.0 -o - \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s -check-prefix=NOGPUBIN
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=9.2 -fcuda-include-gpubinary %t -o - \
// RUN: | FileCheck %s -allow-deprecated-dag-overlap \
// RUN: --check-prefixes=ALL,LNX,NORDC,CUDA,CUDANORDC,CUDA-NEW
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=9.2 -fcuda-include-gpubinary %t -o - -DNOGLOBALS \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s \
// RUN: --check-prefixes=NOGLOBALS,CUDANOGLOBALS
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=9.2 -fgpu-rdc -fcuda-include-gpubinary %t -o - \
// RUN: | FileCheck %s -allow-deprecated-dag-overlap \
// RUN: --check-prefixes=ALL,LNX,RDC,CUDA,CUDARDC,CUDA-NEW
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s -std=c++17 \
// RUN: -target-sdk-version=9.2 -fcuda-include-gpubinary %t -o - \
// RUN: | FileCheck %s -allow-deprecated-dag-overlap \
// RUN: --check-prefixes=ALL,LNX,NORDC,CUDA,CUDANORDC,CUDA-NEW,LNX_17,NORDC17
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s -std=c++17 \
// RUN: -target-sdk-version=9.2 -fgpu-rdc -fcuda-include-gpubinary %t -o - \
// RUN: | FileCheck %s -allow-deprecated-dag-overlap \
// RUN: --check-prefixes=ALL,LNX,RDC,CUDA,CUDARDC,CUDA-NEW,LNX_17,RDC17
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -target-sdk-version=9.2 -o - \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s -check-prefix=NOGPUBIN
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -fcuda-include-gpubinary %t -o - -x hip\
// RUN: | FileCheck -allow-deprecated-dag-overlap %s --check-prefixes=ALL,LNX,NORDC,HIP,HIPEF
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -fcuda-include-gpubinary %t -o - -DNOGLOBALS -x hip \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s -check-prefixes=NOGLOBALS,HIPNOGLOBALS
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s \
// RUN: -fgpu-rdc -fcuda-include-gpubinary %t -o - -x hip \
// RUN: | FileCheck -allow-deprecated-dag-overlap %s --check-prefixes=ALL,LNX,RDC,HIP,HIPEF
// RUN: %clang_cc1 -triple x86_64-linux-gnu -emit-llvm %s -o - -x hip\
// RUN: | FileCheck -allow-deprecated-dag-overlap %s -check-prefixes=ALL,LNX,NORDC,HIP,HIPNEF
// RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -aux-triple amdgcn -emit-llvm %s \
// RUN: -fcuda-include-gpubinary %t -o - -x hip\
// RUN: | FileCheck -allow-deprecated-dag-overlap %s --check-prefixes=ALL,WIN
// RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -aux-triple amdgcn -emit-llvm %s \
// RUN: -o - -x hip\
// RUN: | FileCheck -allow-deprecated-dag-overlap %s --check-prefixes=ALL,WIN,HIP,HIPNEF
#include "Inputs/cuda.h"
// HIPNEF: $__hip_gpubin_handle = comdat any
#ifndef NOGLOBALS
// NORDC-DAG: @device_var = internal global i32
// RDC-DAG: @device_var = global i32
// WIN-DAG: @"?device_var@@3HA" = internal global i32
__device__ int device_var;
// NORDC-DAG: @constant_var = internal global i32
// RDC-DAG: @constant_var = global i32
// WIN-DAG: @"?constant_var@@3HA" = internal global i32
__constant__ int constant_var;
// NORDC-DAG: @shared_var = internal global i32
// RDC-DAG: @shared_var = global i32
// WIN-DAG: @"?shared_var@@3HA" = internal global i32
__shared__ int shared_var;
// Make sure host globals don't get internalized...
// LNX-DAG: @host_var ={{.*}} global i32
// WIN-DAG: @"?host_var@@3HA" = dso_local global i32
int host_var;
// ... and that extern vars remain external.
// LNX-DAG: @ext_host_var = external global i32
// WIN-DAG: @"?ext_host_var@@3HA" = external dso_local global i32
extern int ext_host_var;
// external device-side variables -> extern references to their shadows.
// LNX-DAG: @ext_device_var = external global i32
// WIN-DAG: @"?ext_device_var@@3HA" = external dso_local global i32
extern __device__ int ext_device_var;
// LNX-DAG: @ext_device_var = external global i32
// WIN-DAG: @"?ext_constant_var@@3HA" = external dso_local global i32
extern __constant__ int ext_constant_var;
// external device-side variables with definitions should generate
// definitions for the shadows.
// NORDC-DAG: @ext_device_var_def = internal global i32 undef,
// RDC-DAG: @ext_device_var_def = global i32 undef,
// WIN-DAG: @"?ext_device_var_def@@3HA" = internal global i32 undef
extern __device__ int ext_device_var_def;
__device__ int ext_device_var_def = 1;
// NORDC-DAG: @ext_device_var_def = internal global i32 undef,
// RDC-DAG: @ext_device_var_def = global i32 undef,
// WIN-DAG: @"?ext_constant_var_def@@3HA" = internal global i32 undef
__constant__ int ext_constant_var_def = 2;
#if __cplusplus > 201402L
// NORDC17: @inline_var = internal global i32 undef, comdat, align 4{{$}}
// RDC17: @inline_var = linkonce_odr global i32 undef, comdat, align 4{{$}}
// NORDC17-NOT: @inline_var2 =
// RDC17-NOT: @inline_var2 =
// NORDC17: @_ZN1C17member_inline_varE = internal constant i32 undef, comdat, align 4{{$}}
// RDC17: @_ZN1C17member_inline_varE = linkonce_odr constant i32 undef, comdat, align 4{{$}}
// Check inline variable ODR-used by host is emitted on host and registered.
__device__ inline int inline_var = 3;
// Check inline variable not ODR-used by host is not emitted on host or registered.
__device__ inline int inline_var2 = 5;
struct C {
__device__ static constexpr int member_inline_var = 4;
};
#endif
void use_pointers() {
const int *p;
p = &device_var;
p = &constant_var;
p = &shared_var;
p = &host_var;
p = &ext_device_var;
p = &ext_constant_var;
p = &ext_host_var;
#if __cplusplus > 201402L
p = &inline_var;
decltype(inline_var2) tmp;
p = &C::member_inline_var;
#endif
}
__device__ void device_use() {
#if __cplusplus > 201402L
const int *p = &inline_var2;
#endif
}
// Make sure that all parts of GPU code init/cleanup are there:
// * constant unnamed string with the device-side kernel name to be passed to
// __hipRegisterFunction/__cudaRegisterFunction.
// ALL: @0 = private unnamed_addr constant [18 x i8] c"_Z10kernelfunciii\00"
// * constant unnamed string with the device-side kernel name to be passed to
// __hipRegisterVar/__cudaRegisterVar.
// ALL: @1 = private unnamed_addr constant [11 x i8] c"device_var\00"
// ALL: @2 = private unnamed_addr constant [13 x i8] c"constant_var\00"
// ALL: @3 = private unnamed_addr constant [19 x i8] c"ext_device_var_def\00"
// ALL: @4 = private unnamed_addr constant [21 x i8] c"ext_constant_var_def\00"
// * constant unnamed string with GPU binary
// CUDA: @[[FATBIN:.*]] = private constant{{.*}} c"GPU binary would be here.",
// HIPEF: @[[FATBIN:.*]] = private constant{{.*}} c"GPU binary would be here.",{{.*}}align 4096
// HIPNEF: @[[FATBIN:__hip_fatbin]] = external constant i8, section ".hip_fatbin"
// CUDANORDC-SAME: section ".nv_fatbin", align 8
// CUDARDC-SAME: section "__nv_relfatbin", align 8
// * constant struct that wraps GPU binary
// ALL: @__[[PREFIX:cuda|hip]]_fatbin_wrapper = internal constant
// LNX-SAME: { i32, i32, ptr, ptr }
// CUDA-SAME: { i32 1180844977, i32 1,
// HIP-SAME: { i32 1212764230, i32 1,
// CUDA-SAME: ptr @[[FATBIN]],
// HIPEF-SAME: ptr @[[FATBIN]],
// HIPNEF-SAME: ptr @[[FATBIN]],
// LNX-SAME: ptr null }
// CUDA-SAME: section ".nvFatBinSegment"
// HIP-SAME: section ".hipFatBinSegment"
// * variable to save GPU binary handle after initialization
// CUDANORDC: @__[[PREFIX]]_gpubin_handle = internal global ptr null
// HIPNEF: @__[[PREFIX]]_gpubin_handle = linkonce hidden global ptr null
// * constant unnamed string with NVModuleID
// CUDARDC: [[MODULE_ID_GLOBAL:@.*]] = private constant
// CUDARDC-SAME: c"[[MODULE_ID:.+]]\00", section "__nv_module_id", align 32
// * Make sure our constructor was added to global ctor list.
// LNX: @llvm.global_ctors = appending global {{.*}}@__[[PREFIX]]_module_ctor
// * Alias to global symbol containing the NVModuleID.
// CUDARDC: @__fatbinwrap[[MODULE_ID]] ={{.*}} alias { i32, i32, ptr, ptr }
// CUDARDC-SAME: ptr @__[[PREFIX]]_fatbin_wrapper
// Test that we build the correct number of calls to cudaSetupArgument followed
// by a call to cudaLaunch.
// LNX: define{{.*}}kernelfunc
// New launch sequence stores arguments into local buffer and passes array of
// pointers to them directly to cudaLaunchKernel
// CUDA-NEW: alloca
// CUDA-NEW: store
// CUDA-NEW: store
// CUDA-NEW: store
// CUDA-NEW: call{{.*}}__cudaPopCallConfiguration
// CUDA-NEW: call{{.*}}cudaLaunchKernel
// Legacy style launch sequence sets up arguments by passing them to
// [cuda|hip]SetupArgument.
// CUDA-OLD: call{{.*}}[[PREFIX]]SetupArgument
// CUDA-OLD: call{{.*}}[[PREFIX]]SetupArgument
// CUDA-OLD: call{{.*}}[[PREFIX]]SetupArgument
// CUDA-OLD: call{{.*}}[[PREFIX]]Launch
// HIP: call{{.*}}[[PREFIX]]SetupArgument
// HIP: call{{.*}}[[PREFIX]]SetupArgument
// HIP: call{{.*}}[[PREFIX]]SetupArgument
// HIP: call{{.*}}[[PREFIX]]Launch
__global__ void kernelfunc(int i, int j, int k) {}
// Test that we've built correct kernel launch sequence.
// LNX: define{{.*}}hostfunc
// CUDA-OLD: call{{.*}}[[PREFIX]]ConfigureCall
// CUDA-NEW: call{{.*}}__cudaPushCallConfiguration
// HIP: call{{.*}}[[PREFIX]]ConfigureCall
// LNX: call{{.*}}kernelfunc
void hostfunc(void) { kernelfunc<<<1, 1>>>(1, 1, 1); }
#endif
// Test that we've built a function to register kernels and global vars.
// ALL: define internal void @__[[PREFIX]]_register_globals
// ALL: call{{.*}}[[PREFIX]]RegisterFunction(ptr %0, {{.*}}kernelfunc{{[^,]*}}, {{[^@]*}}@0
// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(ptr %0, {{.*}}device_var{{[^,]*}}, {{[^@]*}}@1, {{.*}}i32 0, {{i32|i64}} 4, i32 0, i32 0
// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(ptr %0, {{.*}}constant_var{{[^,]*}}, {{[^@]*}}@2, {{.*}}i32 0, {{i32|i64}} 4, i32 1, i32 0
// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(ptr %0, {{.*}}ext_device_var_def{{[^,]*}}, {{[^@]*}}@3, {{.*}}i32 0, {{i32|i64}} 4, i32 0, i32 0
// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(ptr %0, {{.*}}ext_constant_var_def{{[^,]*}}, {{[^@]*}}@4, {{.*}}i32 0, {{i32|i64}} 4, i32 1, i32 0
// LNX_17-DAG: [[PREFIX]]RegisterVar(ptr %0, {{.*}}inline_var
// LNX_17-NOT: [[PREFIX]]RegisterVar(ptr %0, {{.*}}inline_var2
// ALL: ret void
// Test that we've built a constructor.
// LNX: define internal void @__[[PREFIX]]_module_ctor
// In separate mode it calls __[[PREFIX]]RegisterFatBinary(&__[[PREFIX]]_fatbin_wrapper)
// HIP only register fat binary once.
// HIP: load ptr, ptr @__hip_gpubin_handle
// HIP-NEXT: icmp eq ptr {{.*}}, null
// HIP-NEXT: br i1 {{.*}}, label %if, label %exit
// HIP: if:
// CUDANORDC: call{{.*}}[[PREFIX]]RegisterFatBinary{{.*}}__[[PREFIX]]_fatbin_wrapper
// .. stores return value in __[[PREFIX]]_gpubin_handle
// CUDANORDC-NEXT: store{{.*}}__[[PREFIX]]_gpubin_handle
// .. and then calls __[[PREFIX]]_register_globals
// HIP: call{{.*}}[[PREFIX]]RegisterFatBinary{{.*}}__[[PREFIX]]_fatbin_wrapper
// .. stores return value in __[[PREFIX]]_gpubin_handle
// HIP-NEXT: store{{.*}}__[[PREFIX]]_gpubin_handle
// .. and then calls __[[PREFIX]]_register_globals
// HIP-NEXT: br label %exit
// HIP: exit:
// HIP-NEXT: load ptr, ptr @__hip_gpubin_handle
// CUDANORDC-NEXT: call void @__[[PREFIX]]_register_globals
// HIP-NEXT: call void @__[[PREFIX]]_register_globals
// * In separate mode we also register a destructor.
// CUDANORDC-NEXT: call i32 @atexit(ptr @__[[PREFIX]]_module_dtor)
// HIP-NEXT: call i32 @atexit(ptr @__[[PREFIX]]_module_dtor)
// With relocatable device code we call __[[PREFIX]]RegisterLinkedBinary%NVModuleID%
// CUDARDC: call{{.*}}__[[PREFIX]]RegisterLinkedBinary[[MODULE_ID]](
// CUDARDC-SAME: __[[PREFIX]]_register_globals, {{.*}}__[[PREFIX]]_fatbin_wrapper
// CUDARDC-SAME: [[MODULE_ID_GLOBAL]]
// Test that we've created destructor.
// CUDANORDC: define internal void @__[[PREFIX]]_module_dtor
// HIP: define internal void @__[[PREFIX]]_module_dtor
// CUDANORDC: load{{.*}}__[[PREFIX]]_gpubin_handle
// HIP: load{{.*}}__[[PREFIX]]_gpubin_handle
// CUDANORDC-NEXT: call void @__[[PREFIX]]UnregisterFatBinary
// HIP-NEXT: icmp ne ptr {{.*}}, null
// HIP-NEXT: br i1 {{.*}}, label %if, label %exit
// HIP: if:
// HIP-NEXT: call void @__[[PREFIX]]UnregisterFatBinary
// HIP-NEXT: store ptr null, ptr @__hip_gpubin_handle
// HIP-NEXT: br label %exit
// HIP: exit:
// There should be no __[[PREFIX]]_register_globals if we have no
// device-side globals, but we still need to register GPU binary.
// Skip GPU binary string first.
// CUDANOGLOBALS-NOT: @{{.*}} = private constant{{.*}}
// HIPNOGLOBALS-NOT: @{{.*}} = internal constant{{.*}}
// NOGLOBALS-NOT: define internal void @__{{.*}}_register_globals
// NOGLOBALS-NOT: define internal void @__{{cuda|hip}}_module_ctor
// NOGLOBALS-NOT: call{{.*}}{{cuda|hip}}RegisterFatBinary{{.*}}__{{cuda|hip}}_fatbin_wrapper
// NOGLOBALS-NOT: call void @__{{cuda|hip}}_register_globals
// NOGLOBALS-NOT: define internal void @__{{cuda|hip}}_module_dtor
// NOGLOBALS-NOT: call void @__{{cuda|hip}}UnregisterFatBinary
// There should be no constructors/destructors if we have no GPU binary.
// NOGPUBIN-NOT: define internal void @__{{cuda|hip}}_register_globals
// NOGPUBIN-NOT: define internal void @__{{cuda|hip}}_module_ctor
// NOGPUBIN-NOT: define internal void @__{{cuda|hip}}_module_dtor
|