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
|
// RUN: %clang_cc1 %s -emit-llvm -o - -triple=armv7-apple-ios -std=c11 | FileCheck %s
// There isn't really anything special about iOS; it just happens to
// only deploy on processors with native atomics support, so it's a good
// way to test those code-paths.
// This work was done in pursuit of <rdar://13338582>.
// CHECK-LABEL: define{{.*}} void @testFloat(ptr
void testFloat(_Atomic(float) *fp) {
// CHECK: [[FP:%.*]] = alloca ptr
// CHECK-NEXT: [[X:%.*]] = alloca float
// CHECK-NEXT: [[F:%.*]] = alloca float
// CHECK-NEXT: store ptr {{%.*}}, ptr [[FP]]
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: store float 1.000000e+00, ptr [[T0]], align 4
__c11_atomic_init(fp, 1.0f);
// CHECK-NEXT: store float 2.000000e+00, ptr [[X]], align 4
_Atomic(float) x = 2.0f;
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: [[T2:%.*]] = load atomic i32, ptr [[T0]] seq_cst, align 4
// CHECK-NEXT: [[T3:%.*]] = bitcast i32 [[T2]] to float
// CHECK-NEXT: store float [[T3]], ptr [[F]]
float f = *fp;
// CHECK-NEXT: [[T0:%.*]] = load float, ptr [[F]], align 4
// CHECK-NEXT: [[T1:%.*]] = load ptr, ptr [[FP]], align 4
// CHECK-NEXT: [[T2:%.*]] = bitcast float [[T0]] to i32
// CHECK-NEXT: store atomic i32 [[T2]], ptr [[T1]] seq_cst, align 4
*fp = f;
// CHECK-NEXT: ret void
}
// CHECK: define{{.*}} void @testComplexFloat(ptr
void testComplexFloat(_Atomic(_Complex float) *fp) {
// CHECK: [[FP:%.*]] = alloca ptr, align 4
// CHECK-NEXT: [[X:%.*]] = alloca [[CF:{ float, float }]], align 8
// CHECK-NEXT: [[F:%.*]] = alloca [[CF]], align 4
// CHECK-NEXT: [[TMP0:%.*]] = alloca [[CF]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = alloca [[CF]], align 8
// CHECK-NEXT: store ptr
// CHECK-NEXT: [[P:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[P]], i32 0, i32 0
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[CF]], ptr [[P]], i32 0, i32 1
// CHECK-NEXT: store float 1.000000e+00, ptr [[T0]]
// CHECK-NEXT: store float 0.000000e+00, ptr [[T1]]
__c11_atomic_init(fp, 1.0f);
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[X]], i32 0, i32 0
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[CF]], ptr [[X]], i32 0, i32 1
// CHECK-NEXT: store float 2.000000e+00, ptr [[T0]]
// CHECK-NEXT: store float 0.000000e+00, ptr [[T1]]
_Atomic(_Complex float) x = 2.0f;
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: [[T2:%.*]] = load atomic i64, ptr [[T0]] seq_cst, align 8
// CHECK-NEXT: store i64 [[T2]], ptr [[TMP0]], align 8
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[TMP0]], i32 0, i32 0
// CHECK-NEXT: [[R:%.*]] = load float, ptr [[T0]]
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[TMP0]], i32 0, i32 1
// CHECK-NEXT: [[I:%.*]] = load float, ptr [[T0]]
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[F]], i32 0, i32 0
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[CF]], ptr [[F]], i32 0, i32 1
// CHECK-NEXT: store float [[R]], ptr [[T0]]
// CHECK-NEXT: store float [[I]], ptr [[T1]]
_Complex float f = *fp;
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[F]], i32 0, i32 0
// CHECK-NEXT: [[R:%.*]] = load float, ptr [[T0]]
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[F]], i32 0, i32 1
// CHECK-NEXT: [[I:%.*]] = load float, ptr [[T0]]
// CHECK-NEXT: [[DEST:%.*]] = load ptr, ptr [[FP]], align 4
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[CF]], ptr [[TMP1]], i32 0, i32 0
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[CF]], ptr [[TMP1]], i32 0, i32 1
// CHECK-NEXT: store float [[R]], ptr [[T0]]
// CHECK-NEXT: store float [[I]], ptr [[T1]]
// CHECK-NEXT: [[T1:%.*]] = load i64, ptr [[TMP1]], align 8
// CHECK-NEXT: store atomic i64 [[T1]], ptr [[DEST]] seq_cst, align 8
*fp = f;
// CHECK-NEXT: ret void
}
typedef struct { short x, y, z, w; } S;
// CHECK: define{{.*}} void @testStruct(ptr
void testStruct(_Atomic(S) *fp) {
// CHECK: [[FP:%.*]] = alloca ptr, align 4
// CHECK-NEXT: [[X:%.*]] = alloca [[S:.*]], align 8
// CHECK-NEXT: [[F:%.*]] = alloca [[S:%.*]], align 2
// CHECK-NEXT: [[TMP0:%.*]] = alloca [[S]], align 8
// CHECK-NEXT: store ptr
// CHECK-NEXT: [[P:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[P]], i32 0, i32 0
// CHECK-NEXT: store i16 1, ptr [[T0]], align 8
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[P]], i32 0, i32 1
// CHECK-NEXT: store i16 2, ptr [[T0]], align 2
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[P]], i32 0, i32 2
// CHECK-NEXT: store i16 3, ptr [[T0]], align 4
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[P]], i32 0, i32 3
// CHECK-NEXT: store i16 4, ptr [[T0]], align 2
__c11_atomic_init(fp, (S){1,2,3,4});
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[X]], i32 0, i32 0
// CHECK-NEXT: store i16 1, ptr [[T0]], align 8
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[X]], i32 0, i32 1
// CHECK-NEXT: store i16 2, ptr [[T0]], align 2
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[X]], i32 0, i32 2
// CHECK-NEXT: store i16 3, ptr [[T0]], align 4
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[S]], ptr [[X]], i32 0, i32 3
// CHECK-NEXT: store i16 4, ptr [[T0]], align 2
_Atomic(S) x = (S){1,2,3,4};
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: [[T2:%.*]] = load atomic i64, ptr [[T0]] seq_cst, align 8
// CHECK-NEXT: store i64 [[T2]], ptr [[F]], align 2
S f = *fp;
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 8 [[TMP0]], ptr align 2 [[F]], i32 8, i1 false)
// CHECK-NEXT: [[T4:%.*]] = load i64, ptr [[TMP0]], align 8
// CHECK-NEXT: store atomic i64 [[T4]], ptr [[T0]] seq_cst, align 8
*fp = f;
// CHECK-NEXT: ret void
}
typedef struct { short x, y, z; } PS;
// CHECK: define{{.*}} void @testPromotedStruct(ptr
void testPromotedStruct(_Atomic(PS) *fp) {
// CHECK: [[FP:%.*]] = alloca ptr, align 4
// CHECK-NEXT: [[X:%.*]] = alloca [[APS:.*]], align 8
// CHECK-NEXT: [[F:%.*]] = alloca [[PS:%.*]], align 2
// CHECK-NEXT: [[TMP0:%.*]] = alloca [[APS]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = alloca [[APS]], align 8
// CHECK-NEXT: store ptr
// CHECK-NEXT: [[P:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[P]], i8 0, i64 8, i1 false)
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[APS]], ptr [[P]], i32 0, i32 0
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[PS]], ptr [[T0]], i32 0, i32 0
// CHECK-NEXT: store i16 1, ptr [[T1]], align 8
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[PS]], ptr [[T0]], i32 0, i32 1
// CHECK-NEXT: store i16 2, ptr [[T1]], align 2
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[PS]], ptr [[T0]], i32 0, i32 2
// CHECK-NEXT: store i16 3, ptr [[T1]], align 4
__c11_atomic_init(fp, (PS){1,2,3});
// CHECK-NEXT: call void @llvm.memset.p0.i32(ptr align 8 [[X]], i8 0, i32 8, i1 false)
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[APS]], ptr [[X]], i32 0, i32 0
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[PS]], ptr [[T0]], i32 0, i32 0
// CHECK-NEXT: store i16 1, ptr [[T1]], align 8
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[PS]], ptr [[T0]], i32 0, i32 1
// CHECK-NEXT: store i16 2, ptr [[T1]], align 2
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[PS]], ptr [[T0]], i32 0, i32 2
// CHECK-NEXT: store i16 3, ptr [[T1]], align 4
_Atomic(PS) x = (PS){1,2,3};
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: [[T2:%.*]] = load atomic i64, ptr [[T0]] seq_cst, align 8
// CHECK-NEXT: store i64 [[T2]], ptr [[TMP0]], align 8
// CHECK-NEXT: [[T0:%.*]] = getelementptr inbounds [[APS]], ptr [[TMP0]], i32 0, i32 0
// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 2 [[F]], ptr align 8 [[T0]], i32 6, i1 false)
PS f = *fp;
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[FP]]
// CHECK-NEXT: call void @llvm.memset.p0.i32(ptr align 8 [[TMP1]], i8 0, i32 8, i1 false)
// CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds [[APS]], ptr [[TMP1]], i32 0, i32 0
// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 8 [[T1]], ptr align 2 [[F]], i32 6, i1 false)
// CHECK-NEXT: [[T5:%.*]] = load i64, ptr [[TMP1]], align 8
// CHECK-NEXT: store atomic i64 [[T5]], ptr [[T0]] seq_cst, align 8
*fp = f;
// CHECK-NEXT: ret void
}
PS test_promoted_load(_Atomic(PS) *addr) {
// CHECK-LABEL: @test_promoted_load(ptr noalias sret(%struct.PS) align 2 %agg.result, ptr noundef %addr)
// CHECK: [[ADDR_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[ATOMIC_RES:%.*]] = alloca { %struct.PS, [2 x i8] }, align 8
// CHECK: store ptr %addr, ptr [[ADDR_ARG]], align 4
// CHECK: [[ADDR:%.*]] = load ptr, ptr [[ADDR_ARG]], align 4
// CHECK: [[VAL:%.*]] = load atomic i64, ptr [[ADDR]] seq_cst, align 8
// CHECK: store i64 [[VAL]], ptr [[ATOMIC_RES]], align 8
// CHECK: call void @llvm.memcpy.p0.p0.i32(ptr align 2 %agg.result, ptr align 8 [[ATOMIC_RES]], i32 6, i1 false)
return __c11_atomic_load(addr, 5);
}
void test_promoted_store(_Atomic(PS) *addr, PS *val) {
// CHECK-LABEL: @test_promoted_store(ptr noundef %addr, ptr noundef %val)
// CHECK: [[ADDR_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[VAL_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[NONATOMIC_TMP:%.*]] = alloca %struct.PS, align 2
// CHECK: [[ATOMIC_VAL:%.*]] = alloca { %struct.PS, [2 x i8] }, align 8
// CHECK: store ptr %addr, ptr [[ADDR_ARG]], align 4
// CHECK: store ptr %val, ptr [[VAL_ARG]], align 4
// CHECK: [[ADDR:%.*]] = load ptr, ptr [[ADDR_ARG]], align 4
// CHECK: [[VAL:%.*]] = load ptr, ptr [[VAL_ARG]], align 4
// CHECK: call void @llvm.memcpy.p0.p0.i32(ptr align 2 [[NONATOMIC_TMP]], ptr align 2 [[VAL]], i32 6, i1 false)
// CHECK: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[ATOMIC_VAL]], ptr align 2 [[NONATOMIC_TMP]], i64 6, i1 false)
// CHECK: [[VAL64:%.*]] = load i64, ptr [[ATOMIC_VAL]], align 8
// CHECK: store atomic i64 [[VAL64]], ptr [[ADDR]] seq_cst, align 8
__c11_atomic_store(addr, *val, 5);
}
PS test_promoted_exchange(_Atomic(PS) *addr, PS *val) {
// CHECK-LABEL: @test_promoted_exchange(ptr noalias sret(%struct.PS) align 2 %agg.result, ptr noundef %addr, ptr noundef %val)
// CHECK: [[ADDR_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[VAL_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[NONATOMIC_TMP:%.*]] = alloca %struct.PS, align 2
// CHECK: [[ATOMIC_VAL:%.*]] = alloca { %struct.PS, [2 x i8] }, align 8
// CHECK: [[ATOMIC_RES:%.*]] = alloca { %struct.PS, [2 x i8] }, align 8
// CHECK: store ptr %addr, ptr [[ADDR_ARG]], align 4
// CHECK: store ptr %val, ptr [[VAL_ARG]], align 4
// CHECK: [[ADDR:%.*]] = load ptr, ptr [[ADDR_ARG]], align 4
// CHECK: [[VAL:%.*]] = load ptr, ptr [[VAL_ARG]], align 4
// CHECK: call void @llvm.memcpy.p0.p0.i32(ptr align 2 [[NONATOMIC_TMP]], ptr align 2 [[VAL]], i32 6, i1 false)
// CHECK: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[ATOMIC_VAL]], ptr align 2 [[NONATOMIC_TMP]], i64 6, i1 false)
// CHECK: [[VAL64:%.*]] = load i64, ptr [[ATOMIC_VAL]], align 8
// CHECK: [[RES:%.*]] = atomicrmw xchg ptr [[ADDR]], i64 [[VAL64]] seq_cst, align 8
// CHECK: store i64 [[RES]], ptr [[ATOMIC_RES]], align 8
// CHECK: call void @llvm.memcpy.p0.p0.i32(ptr align 2 %agg.result, ptr align 8 [[ATOMIC_RES]], i32 6, i1 false)
return __c11_atomic_exchange(addr, *val, 5);
}
_Bool test_promoted_cmpxchg(_Atomic(PS) *addr, PS *desired, PS *new) {
// CHECK: define{{.*}} zeroext i1 @test_promoted_cmpxchg(ptr noundef %addr, ptr noundef %desired, ptr noundef %new) #0 {
// CHECK: [[ADDR_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[DESIRED_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[NEW_ARG:%.*]] = alloca ptr, align 4
// CHECK: [[NONATOMIC_TMP:%.*]] = alloca %struct.PS, align 2
// CHECK: [[ATOMIC_DESIRED:%.*]] = alloca { %struct.PS, [2 x i8] }, align 8
// CHECK: [[ATOMIC_NEW:%.*]] = alloca { %struct.PS, [2 x i8] }, align 8
// CHECK: [[RES_ADDR:%.*]] = alloca i8, align 1
// CHECK: store ptr %addr, ptr [[ADDR_ARG]], align 4
// CHECK: store ptr %desired, ptr [[DESIRED_ARG]], align 4
// CHECK: store ptr %new, ptr [[NEW_ARG]], align 4
// CHECK: [[ADDR:%.*]] = load ptr, ptr [[ADDR_ARG]], align 4
// CHECK: [[DESIRED:%.*]] = load ptr, ptr [[DESIRED_ARG]], align 4
// CHECK: [[NEW:%.*]] = load ptr, ptr [[NEW_ARG]], align 4
// CHECK: call void @llvm.memcpy.p0.p0.i32(ptr align 2 [[NONATOMIC_TMP]], ptr align 2 [[NEW]], i32 6, i1 false)
// CHECK: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[ATOMIC_DESIRED:%.*]], ptr align 2 [[DESIRED]], i64 6, i1 false)
// CHECK: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[ATOMIC_NEW]], ptr align 2 [[NONATOMIC_TMP]], i64 6, i1 false)
// CHECK: [[ATOMIC_DESIRED_VAL64:%.*]] = load i64, ptr [[ATOMIC_DESIRED:%.*]], align 8
// CHECK: [[ATOMIC_NEW_VAL64:%.*]] = load i64, ptr [[ATOMIC_NEW]], align 8
// CHECK: [[RES:%.*]] = cmpxchg ptr [[ADDR]], i64 [[ATOMIC_DESIRED_VAL64]], i64 [[ATOMIC_NEW_VAL64]] seq_cst seq_cst, align 8
// CHECK: [[RES_VAL64:%.*]] = extractvalue { i64, i1 } [[RES]], 0
// CHECK: [[RES_BOOL:%.*]] = extractvalue { i64, i1 } [[RES]], 1
// CHECK: br i1 [[RES_BOOL]], label {{%.*}}, label {{%.*}}
// CHECK: store i64 [[RES_VAL64]], ptr [[ATOMIC_DESIRED]], align 8
// CHECK: br label {{%.*}}
// CHECK: [[RES_BOOL8:%.*]] = zext i1 [[RES_BOOL]] to i8
// CHECK: store i8 [[RES_BOOL8]], ptr [[RES_ADDR]], align 1
// CHECK: [[RES_BOOL8:%.*]] = load i8, ptr [[RES_ADDR]], align 1
// CHECK: [[RETVAL:%.*]] = trunc i8 [[RES_BOOL8]] to i1
// CHECK: ret i1 [[RETVAL]]
return __c11_atomic_compare_exchange_strong(addr, desired, *new, 5, 5);
}
struct Empty {};
struct Empty testEmptyStructLoad(_Atomic(struct Empty)* empty) {
// CHECK-LABEL: @testEmptyStructLoad(
// CHECK-NOT: @__atomic_load
// CHECK: load atomic i8, ptr %{{.*}} seq_cst, align 1
return *empty;
}
void testEmptyStructStore(_Atomic(struct Empty)* empty, struct Empty value) {
// CHECK-LABEL: @testEmptyStructStore(
// CHECK-NOT: @__atomic_store
// CHECK: store atomic i8 %{{.*}}, ptr %{{.*}} seq_cst, align 1
*empty = value;
}
|
