1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs
2 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-enable-irbuilder -verify -fopenmp -fopenmp-version=45 -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s
3 // expected-no-diagnostics
4 
5 #ifndef HEADER
6 #define HEADER
7 
8 // CHECK-LABEL: define {{.*}}@workshareloop_unsigned_dynamic_chunked(
9 // CHECK-NEXT:  [[ENTRY:.*]]:
10 // CHECK-NEXT:    %[[A_ADDR:.+]] = alloca float*, align 8
11 // CHECK-NEXT:    %[[B_ADDR:.+]] = alloca float*, align 8
12 // CHECK-NEXT:    %[[C_ADDR:.+]] = alloca float*, align 8
13 // CHECK-NEXT:    %[[D_ADDR:.+]] = alloca float*, align 8
14 // CHECK-NEXT:    %[[I:.+]] = alloca i32, align 4
15 // CHECK-NEXT:    %[[AGG_CAPTURED:.+]] = alloca %struct.anon, align 8
16 // CHECK-NEXT:    %[[AGG_CAPTURED1:.+]] = alloca %struct.anon.0, align 4
17 // CHECK-NEXT:    %[[DOTCOUNT_ADDR:.+]] = alloca i32, align 4
18 // CHECK-NEXT:    %[[P_LASTITER:.+]] = alloca i32, align 4
19 // CHECK-NEXT:    %[[P_LOWERBOUND:.+]] = alloca i32, align 4
20 // CHECK-NEXT:    %[[P_UPPERBOUND:.+]] = alloca i32, align 4
21 // CHECK-NEXT:    %[[P_STRIDE:.+]] = alloca i32, align 4
22 // CHECK-NEXT:    store float* %[[A:.+]], float** %[[A_ADDR]], align 8
23 // CHECK-NEXT:    store float* %[[B:.+]], float** %[[B_ADDR]], align 8
24 // CHECK-NEXT:    store float* %[[C:.+]], float** %[[C_ADDR]], align 8
25 // CHECK-NEXT:    store float* %[[D:.+]], float** %[[D_ADDR]], align 8
26 // CHECK-NEXT:    store i32 33, i32* %[[I]], align 4
27 // CHECK-NEXT:    %[[TMP0:.+]] = getelementptr inbounds %struct.anon, %struct.anon* %[[AGG_CAPTURED]], i32 0, i32 0
28 // CHECK-NEXT:    store i32* %[[I]], i32** %[[TMP0]], align 8
29 // CHECK-NEXT:    %[[TMP1:.+]] = getelementptr inbounds %struct.anon.0, %struct.anon.0* %[[AGG_CAPTURED1]], i32 0, i32 0
30 // CHECK-NEXT:    %[[TMP2:.+]] = load i32, i32* %[[I]], align 4
31 // CHECK-NEXT:    store i32 %[[TMP2]], i32* %[[TMP1]], align 4
32 // CHECK-NEXT:    call void @__captured_stmt(i32* %[[DOTCOUNT_ADDR]], %struct.anon* %[[AGG_CAPTURED]])
33 // CHECK-NEXT:    %[[DOTCOUNT:.+]] = load i32, i32* %[[DOTCOUNT_ADDR]], align 4
34 // CHECK-NEXT:    br label %[[OMP_LOOP_PREHEADER:.+]]
35 // CHECK-EMPTY:
36 // CHECK-NEXT:  [[OMP_LOOP_PREHEADER]]:
37 // CHECK-NEXT:    store i32 1, i32* %[[P_LOWERBOUND]], align 4
38 // CHECK-NEXT:    store i32 %[[DOTCOUNT]], i32* %[[P_UPPERBOUND]], align 4
39 // CHECK-NEXT:    store i32 1, i32* %[[P_STRIDE]], align 4
40 // CHECK-NEXT:    %[[OMP_GLOBAL_THREAD_NUM:.+]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
41 // CHECK-NEXT:    call void @__kmpc_dispatch_init_4u(%struct.ident_t* @1, i32 %[[OMP_GLOBAL_THREAD_NUM]], i32 1073741859, i32 1, i32 %[[DOTCOUNT]], i32 1, i32 5)
42 // CHECK-NEXT:    br label %[[OMP_LOOP_PREHEADER_OUTER_COND:.+]]
43 // CHECK-EMPTY:
44 // CHECK-NEXT:  [[OMP_LOOP_HEADER:.*]]:
45 // CHECK-NEXT:    %[[OMP_LOOP_IV:.+]] = phi i32 [ %[[LB:.+]], %[[OMP_LOOP_PREHEADER_OUTER_COND]] ], [ %[[OMP_LOOP_NEXT:.+]], %[[OMP_LOOP_INC:.+]] ]
46 // CHECK-NEXT:    br label %[[OMP_LOOP_COND:.+]]
47 // CHECK-EMPTY:
48 // CHECK-NEXT:  [[OMP_LOOP_COND]]:
49 // CHECK-NEXT:    %[[UB:.+]] = load i32, i32* %[[P_UPPERBOUND]], align 4
50 // CHECK-NEXT:    %[[OMP_LOOP_CMP:.+]] = icmp ult i32 %[[OMP_LOOP_IV]], %[[UB]]
51 // CHECK-NEXT:    br i1 %[[OMP_LOOP_CMP]], label %[[OMP_LOOP_BODY:.+]], label %[[OMP_LOOP_PREHEADER_OUTER_COND]]
52 // CHECK-EMPTY:
53 // CHECK-NEXT:  [[OMP_LOOP_BODY]]:
54 // CHECK-NEXT:    call void @__captured_stmt.1(i32* %[[I]], i32 %[[OMP_LOOP_IV]], %struct.anon.0* %[[AGG_CAPTURED1]])
55 // CHECK-NEXT:    %[[TMP3:.+]] = load float*, float** %[[B_ADDR]], align 8
56 // CHECK-NEXT:    %[[TMP4:.+]] = load i32, i32* %[[I]], align 4
57 // CHECK-NEXT:    %[[IDXPROM:.+]] = zext i32 %[[TMP4]] to i64
58 // CHECK-NEXT:    %[[ARRAYIDX:.+]] = getelementptr inbounds float, float* %[[TMP3]], i64 %[[IDXPROM]]
59 // CHECK-NEXT:    %[[TMP5:.+]] = load float, float* %[[ARRAYIDX]], align 4
60 // CHECK-NEXT:    %[[TMP6:.+]] = load float*, float** %[[C_ADDR]], align 8
61 // CHECK-NEXT:    %[[TMP7:.+]] = load i32, i32* %[[I]], align 4
62 // CHECK-NEXT:    %[[IDXPROM2:.+]] = zext i32 %[[TMP7]] to i64
63 // CHECK-NEXT:    %[[ARRAYIDX3:.+]] = getelementptr inbounds float, float* %[[TMP6]], i64 %[[IDXPROM2]]
64 // CHECK-NEXT:    %[[TMP8:.+]] = load float, float* %[[ARRAYIDX3]], align 4
65 // CHECK-NEXT:    %[[MUL:.+]] = fmul float %[[TMP5]], %[[TMP8]]
66 // CHECK-NEXT:    %[[TMP9:.+]] = load float*, float** %[[D_ADDR]], align 8
67 // CHECK-NEXT:    %[[TMP10:.+]] = load i32, i32* %[[I]], align 4
68 // CHECK-NEXT:    %[[IDXPROM4:.+]] = zext i32 %[[TMP10]] to i64
69 // CHECK-NEXT:    %[[ARRAYIDX5:.+]] = getelementptr inbounds float, float* %[[TMP9]], i64 %[[IDXPROM4]]
70 // CHECK-NEXT:    %[[TMP11:.+]] = load float, float* %[[ARRAYIDX5]], align 4
71 // CHECK-NEXT:    %[[MUL6:.+]] = fmul float %[[MUL]], %[[TMP11]]
72 // CHECK-NEXT:    %[[TMP12:.+]] = load float*, float** %[[A_ADDR]], align 8
73 // CHECK-NEXT:    %[[TMP13:.+]] = load i32, i32* %[[I]], align 4
74 // CHECK-NEXT:    %[[IDXPROM7:.+]] = zext i32 %[[TMP13]] to i64
75 // CHECK-NEXT:    %[[ARRAYIDX8:.+]] = getelementptr inbounds float, float* %[[TMP12]], i64 %[[IDXPROM7]]
76 // CHECK-NEXT:    store float %[[MUL6]], float* %[[ARRAYIDX8]], align 4
77 // CHECK-NEXT:    br label %[[OMP_LOOP_INC]]
78 // CHECK-EMPTY:
79 // CHECK-NEXT:  [[OMP_LOOP_INC]]:
80 // CHECK-NEXT:    %[[OMP_LOOP_NEXT]] = add nuw i32 %[[OMP_LOOP_IV]], 1
81 // CHECK-NEXT:    br label %[[OMP_LOOP_HEADER]]
82 // CHECK-EMPTY:
83 // CHECK-NEXT:  [[OMP_LOOP_EXIT:.*]]:
84 // CHECK-NEXT:    %[[OMP_GLOBAL_THREAD_NUM9:.+]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
85 // CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @2, i32 %[[OMP_GLOBAL_THREAD_NUM9]])
86 // CHECK-NEXT:    br label %[[OMP_LOOP_AFTER:.+]]
87 // CHECK-EMPTY:
88 // CHECK-NEXT:  [[OMP_LOOP_AFTER]]:
89 // CHECK-NEXT:    ret void
90 // CHECK-EMPTY:
91 // CHECK-NEXT:  [[OMP_LOOP_PREHEADER_OUTER_COND]]:
92 // CHECK-NEXT:    %[[TMP14:.+]] = call i32 @__kmpc_dispatch_next_4u(%struct.ident_t* @1, i32 %[[OMP_GLOBAL_THREAD_NUM]], i32* %[[P_LASTITER]], i32* %[[P_LOWERBOUND]], i32* %[[P_UPPERBOUND]], i32* %[[P_STRIDE]])
93 // CHECK-NEXT:    %[[TMP15:.+]] = icmp ne i32 %[[TMP14]], 0
94 // CHECK-NEXT:    %[[TMP16:.+]] = load i32, i32* %[[P_LOWERBOUND]], align 4
95 // CHECK-NEXT:    %[[LB]] = sub i32 %[[TMP16]], 1
96 // CHECK-NEXT:    br i1 %[[TMP15]], label %[[OMP_LOOP_HEADER]], label %[[OMP_LOOP_EXIT]]
97 // CHECK-NEXT:  }
98 
workshareloop_unsigned_dynamic_chunked(float * a,float * b,float * c,float * d)99 extern "C" void workshareloop_unsigned_dynamic_chunked(float *a, float *b, float *c, float *d) {
100 #pragma omp for schedule(dynamic, 5)
101   for (unsigned i = 33; i < 32000000; i += 7) {
102     a[i] = b[i] * c[i] * d[i];
103   }
104 }
105 
106 #endif // HEADER
107 
108 // CHECK-LABEL: define {{.*}}@__captured_stmt(
109 // CHECK-NEXT:  [[ENTRY:.*]]:
110 // CHECK-NEXT:    %[[DISTANCE_ADDR:.+]] = alloca i32*, align 8
111 // CHECK-NEXT:    %[[__CONTEXT_ADDR:.+]] = alloca %struct.anon*, align 8
112 // CHECK-NEXT:    %[[DOTSTART:.+]] = alloca i32, align 4
113 // CHECK-NEXT:    %[[DOTSTOP:.+]] = alloca i32, align 4
114 // CHECK-NEXT:    %[[DOTSTEP:.+]] = alloca i32, align 4
115 // CHECK-NEXT:    store i32* %[[DISTANCE:.+]], i32** %[[DISTANCE_ADDR]], align 8
116 // CHECK-NEXT:    store %struct.anon* %[[__CONTEXT:.+]], %struct.anon** %[[__CONTEXT_ADDR]], align 8
117 // CHECK-NEXT:    %[[TMP0:.+]] = load %struct.anon*, %struct.anon** %[[__CONTEXT_ADDR]], align 8
118 // CHECK-NEXT:    %[[TMP1:.+]] = getelementptr inbounds %struct.anon, %struct.anon* %[[TMP0]], i32 0, i32 0
119 // CHECK-NEXT:    %[[TMP2:.+]] = load i32*, i32** %[[TMP1]], align 8
120 // CHECK-NEXT:    %[[TMP3:.+]] = load i32, i32* %[[TMP2]], align 4
121 // CHECK-NEXT:    store i32 %[[TMP3]], i32* %[[DOTSTART]], align 4
122 // CHECK-NEXT:    store i32 32000000, i32* %[[DOTSTOP]], align 4
123 // CHECK-NEXT:    store i32 7, i32* %[[DOTSTEP]], align 4
124 // CHECK-NEXT:    %[[TMP4:.+]] = load i32, i32* %[[DOTSTART]], align 4
125 // CHECK-NEXT:    %[[TMP5:.+]] = load i32, i32* %[[DOTSTOP]], align 4
126 // CHECK-NEXT:    %[[CMP:.+]] = icmp ult i32 %[[TMP4]], %[[TMP5]]
127 // CHECK-NEXT:    br i1 %[[CMP]], label %[[COND_TRUE:.+]], label %[[COND_FALSE:.+]]
128 // CHECK-EMPTY:
129 // CHECK-NEXT:  [[COND_TRUE]]:
130 // CHECK-NEXT:    %[[TMP6:.+]] = load i32, i32* %[[DOTSTOP]], align 4
131 // CHECK-NEXT:    %[[TMP7:.+]] = load i32, i32* %[[DOTSTART]], align 4
132 // CHECK-NEXT:    %[[SUB:.+]] = sub i32 %[[TMP6]], %[[TMP7]]
133 // CHECK-NEXT:    %[[TMP8:.+]] = load i32, i32* %[[DOTSTEP]], align 4
134 // CHECK-NEXT:    %[[SUB1:.+]] = sub i32 %[[TMP8]], 1
135 // CHECK-NEXT:    %[[ADD:.+]] = add i32 %[[SUB]], %[[SUB1]]
136 // CHECK-NEXT:    %[[TMP9:.+]] = load i32, i32* %[[DOTSTEP]], align 4
137 // CHECK-NEXT:    %[[DIV:.+]] = udiv i32 %[[ADD]], %[[TMP9]]
138 // CHECK-NEXT:    br label %[[COND_END:.+]]
139 // CHECK-EMPTY:
140 // CHECK-NEXT:  [[COND_FALSE]]:
141 // CHECK-NEXT:    br label %[[COND_END]]
142 // CHECK-EMPTY:
143 // CHECK-NEXT:  [[COND_END]]:
144 // CHECK-NEXT:    %[[COND:.+]] = phi i32 [ %[[DIV]], %[[COND_TRUE]] ], [ 0, %[[COND_FALSE]] ]
145 // CHECK-NEXT:    %[[TMP10:.+]] = load i32*, i32** %[[DISTANCE_ADDR]], align 8
146 // CHECK-NEXT:    store i32 %[[COND]], i32* %[[TMP10]], align 4
147 // CHECK-NEXT:    ret void
148 // CHECK-NEXT:  }
149 
150 
151 // CHECK-LABEL: define {{.*}}@__captured_stmt.1(
152 // CHECK-NEXT:  [[ENTRY:.*]]:
153 // CHECK-NEXT:    %[[LOOPVAR_ADDR:.+]] = alloca i32*, align 8
154 // CHECK-NEXT:    %[[LOGICAL_ADDR:.+]] = alloca i32, align 4
155 // CHECK-NEXT:    %[[__CONTEXT_ADDR:.+]] = alloca %struct.anon.0*, align 8
156 // CHECK-NEXT:    store i32* %[[LOOPVAR:.+]], i32** %[[LOOPVAR_ADDR]], align 8
157 // CHECK-NEXT:    store i32 %[[LOGICAL:.+]], i32* %[[LOGICAL_ADDR]], align 4
158 // CHECK-NEXT:    store %struct.anon.0* %[[__CONTEXT:.+]], %struct.anon.0** %[[__CONTEXT_ADDR]], align 8
159 // CHECK-NEXT:    %[[TMP0:.+]] = load %struct.anon.0*, %struct.anon.0** %[[__CONTEXT_ADDR]], align 8
160 // CHECK-NEXT:    %[[TMP1:.+]] = getelementptr inbounds %struct.anon.0, %struct.anon.0* %[[TMP0]], i32 0, i32 0
161 // CHECK-NEXT:    %[[TMP2:.+]] = load i32, i32* %[[TMP1]], align 4
162 // CHECK-NEXT:    %[[TMP3:.+]] = load i32, i32* %[[LOGICAL_ADDR]], align 4
163 // CHECK-NEXT:    %[[MUL:.+]] = mul i32 7, %[[TMP3]]
164 // CHECK-NEXT:    %[[ADD:.+]] = add i32 %[[TMP2]], %[[MUL]]
165 // CHECK-NEXT:    %[[TMP4:.+]] = load i32*, i32** %[[LOOPVAR_ADDR]], align 8
166 // CHECK-NEXT:    store i32 %[[ADD]], i32* %[[TMP4]], align 4
167 // CHECK-NEXT:    ret void
168 // CHECK-NEXT:  }
169 
170 
171 // CHECK: ![[META0:[0-9]+]] = !{i32 1, !"wchar_size", i32 4}
172 // CHECK: ![[META1:[0-9]+]] = !{i32 7, !"openmp", i32 45}
173 // CHECK: ![[META2:[0-9]+]] =
174