1 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-64 2 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 3 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-64 4 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-32 5 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 6 // RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap %s --check-prefix CHECK --check-prefix CHECK-32 7 8 // RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 9 // RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 10 // RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 11 // RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 12 // RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 13 // RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY0 %s 14 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}} 15 16 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap %s --check-prefix LAMBDA --check-prefix LAMBDA-64 17 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 18 // RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap %s --check-prefix LAMBDA --check-prefix LAMBDA-64 19 20 // RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 21 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s -Wno-openmp-mapping 22 // RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - -Wno-openmp-mapping | FileCheck -allow-deprecated-dag-overlap --check-prefix SIMD-ONLY1 %s 23 // SIMD-ONLY1-NOT: {{__kmpc|__tgt}} 24 25 // expected-no-diagnostics 26 #ifndef HEADER 27 #define HEADER 28 29 struct St { 30 int a, b; 31 St() : a(0), b(0) {} 32 St(const St &st) : a(st.a + st.b), b(0) {} 33 ~St() {} 34 }; 35 36 volatile int g = 1212; 37 volatile int &g1 = g; 38 39 template <class T> 40 struct S { 41 T f; 42 S(T a) : f(a + g) {} 43 S() : f(g) {} 44 S(const S &s, St t = St()) : f(s.f + t.a) {} 45 operator T() { return T(); } 46 ~S() {} 47 }; 48 49 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float } 50 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} } 51 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} } 52 53 template <typename T> 54 T tmain() { 55 S<T> test; 56 T t_var = T(); 57 T vec[] = {1, 2}; 58 S<T> s_arr[] = {1, 2}; 59 S<T> &var = test; 60 #pragma omp target 61 #pragma omp teams distribute simd firstprivate(t_var, vec, s_arr, var) 62 for (int i = 0; i < 2; ++i) { 63 vec[i] = t_var; 64 s_arr[i] = var; 65 } 66 return T(); 67 } 68 69 // CHECK-DAG: [[TEST:@.+]] ={{.*}} global [[S_FLOAT_TY]] zeroinitializer, 70 S<float> test; 71 // CHECK-DAG: [[T_VAR:@.+]] ={{.*}} global i{{[0-9]+}} 333, 72 int t_var = 333; 73 // CHECK-DAG: [[VEC:@.+]] ={{.*}} global [2 x i{{[0-9]+}}] [i{{[0-9]+}} 1, i{{[0-9]+}} 2], 74 int vec[] = {1, 2}; 75 // CHECK-DAG: [[S_ARR:@.+]] ={{.*}} global [2 x [[S_FLOAT_TY]]] zeroinitializer, 76 S<float> s_arr[] = {1, 2}; 77 // CHECK-DAG: [[VAR:@.+]] ={{.*}} global [[S_FLOAT_TY]] zeroinitializer, 78 S<float> var(3); 79 // CHECK-DAG: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0, 80 81 int main() { 82 static int sivar; 83 #ifdef LAMBDA 84 // LAMBDA: [[G:@.+]] ={{.*}} global i{{[0-9]+}} 1212, 85 // LAMBDA-LABEL: @main 86 // LAMBDA: call void [[OUTER_LAMBDA:@.+]]( 87 [&]() { 88 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]]( 89 // LAMBDA: call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 3, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 90 // LAMBDA: call void @[[LOFFL1:.+]](i{{64|32}} %{{.+}}) 91 // LAMBDA: ret 92 #pragma omp target 93 #pragma omp teams distribute simd firstprivate(g, g1, sivar) 94 for (int i = 0; i < 2; ++i) { 95 // LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} {{%.+}}, i{{64|32}} {{%.+}}) 96 // LAMBDA: {{%.+}} = alloca i{{[0-9]+}}, 97 // LAMBDA: {{%.+}} = alloca i{{[0-9]+}}, 98 // LAMBDA: {{%.+}} = alloca i{{[0-9]+}}, 99 // LAMBDA: [[G_CAST:%.+]] = alloca i{{[0-9]+}}, 100 // LAMBDA: [[G1_CAST:%.+]] = alloca i{{[0-9]+}}, 101 // LAMBDA: [[SIVAR_CAST:%.+]] = alloca i{{[0-9]+}}, 102 // LAMBDA-DAG: [[G_CAST_VAL:%.+]] = load{{.+}} [[G_CAST]], 103 // LAMBDA-DAG: [[G1_CAST_VAL:%.+]] = load{{.+}} [[G1_CAST]], 104 // LAMBDA-DAG: [[SIVAR_CAST_VAL:%.+]] = load{{.+}} [[SIVAR_CAST]], 105 // LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 3, {{.+}} @[[LOUTL1:.+]] to {{.+}}, {{.+}} [[G_CAST_VAL]], {{.+}} [[G1_CAST_VAL]], {{.+}} [[SIVAR_CAST_VAL]]) 106 // LAMBDA: ret void 107 108 // LAMBDA: define internal void @[[LOUTL1]]({{.+}}) 109 // Skip global and bound tid vars 110 // LAMBDA: {{.+}} = alloca i32*, 111 // LAMBDA: {{.+}} = alloca i32*, 112 // LAMBDA: [[G_ADDR:%.+]] = alloca i{{[0-9]+}}, 113 // LAMBDA: [[G1_ADDR:%.+]] = alloca i{{[0-9]+}}, 114 // LAMBDA: [[SIVAR_ADDR:%.+]] = alloca i{{[0-9]+}}, 115 // LAMBDA: [[G1_TMP:%.+]] = alloca i32*, 116 // skip loop vars 117 // LAMBDA-DAG: store {{.+}}, {{.+}} [[G_ADDR]], 118 // LAMBDA-DAG: store {{.+}}, {{.+}} [[G1_ADDR]], 119 // LAMBDA-DAG: store {{.+}}, {{.+}} [[SIVAR_ADDR]], 120 // LAMBDA-DAG: [[G_CONV:%.+]] = bitcast {{.+}} [[G_ADDR]] to 121 // LAMBDA-DAG: [[G1_CONV:%.+]] = bitcast {{.+}} [[G1_ADDR]] to 122 // LAMBDA-DAG: [[SIVAR_CONV:%.+]] = bitcast {{.+}} [[SIVAR_ADDR]] to 123 // LAMBDA-DAG: store{{.+}} [[G1_CONV]], {{.+}} [[G1_TMP]], 124 g = 1; 125 g1 = 1; 126 sivar = 2; 127 // LAMBDA: call void @__kmpc_for_static_init_4( 128 // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G_CONV]], 129 // LAMBDA-DAG: [[G1:%.+]] = load{{.+}}, {{.+}}* [[G1_TMP]] 130 // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G1]], 131 // LAMBDA-DAG: store{{.+}} 2, {{.+}} [[SIVAR_CONV]], 132 // LAMBDA-DAG: [[G1_REF:%.+]] = load{{.+}}, {{.+}} [[G1_TMP]], 133 // LAMBDA-DAG: store{{.+}} 1, {{.+}} [[G1_REF]], 134 // LAMBDA: call void [[INNER_LAMBDA:@.+]]( 135 // LAMBDA: call void @__kmpc_for_static_fini( 136 [&]() { 137 // LAMBDA: define {{.+}} void [[INNER_LAMBDA]]({{.+}} [[ARG_PTR:%.+]]) 138 // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]], 139 g = 2; 140 g1 = 2; 141 sivar = 4; 142 // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]] 143 144 // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0 145 // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]] 146 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]] 147 // LAMBDA: [[G1_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1 148 // LAMBDA: [[G1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G1_PTR_REF]] 149 // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G1_REF]] 150 // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2 151 // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]] 152 // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]] 153 }(); 154 } 155 }(); 156 return 0; 157 #else 158 #pragma omp target 159 #pragma omp teams distribute simd firstprivate(t_var, vec, s_arr, var, sivar) 160 for (int i = 0; i < 2; ++i) { 161 vec[i] = t_var; 162 s_arr[i] = var; 163 sivar += i; 164 } 165 return tmain<int>(); 166 #endif 167 } 168 169 // CHECK: define {{.*}}i{{[0-9]+}} @main() 170 // CHECK: call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 5, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 171 // CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}}) 172 // CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]() 173 // CHECK: ret 174 175 // CHECK: define{{.*}} void @[[OFFL1]]({{.+}}) 176 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 177 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}]*, 178 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*, 179 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]]*, 180 // CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 181 // CHECK: [[T_VAR_CAST:%.+]] = alloca i{{[0-9]+}}, 182 // CHECK: [[SIVAR_CAST:%.+]] = alloca i{{[0-9]+}}, 183 184 // CHECK-DAG: [[VEC_TE_PAR:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PRIV]], 185 // CHECK-DAG: [[T_VAR_TE_PAR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_CAST]], 186 // CHECK-DAG: [[S_ARR_TE_PAR:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_PRIV]], 187 // CHECK-DAG: [[VAR_TE_PAR:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_PRIV]], 188 // CHECK-DAG: [[SIVAR_TE_PAR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_CAST]], 189 190 // CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 5, {{.+}} @[[OUTL1:.+]] to {{.+}}, [2 x i{{[0-9]+}}]* [[VEC_TE_PAR]], i{{[0-9]+}} [[T_VAR_TE_PAR]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_TE_PAR]], [[S_FLOAT_TY]]* [[VAR_TE_PAR]], i{{[0-9]+}} [[SIVAR_TE_PAR]]) 191 // CHECK: ret void 192 193 // CHECK: define internal void @[[OUTL1]]({{.+}}) 194 // Skip global and bound tid vars 195 // CHECK: {{.+}} = alloca i32*, 196 // CHECK: {{.+}} = alloca i32*, 197 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 198 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}, 199 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_FLOAT_TY]]]*, 200 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_FLOAT_TY]]*, 201 // CHECK: [[SIVAR_ADDR:%.+]] = alloca i{{[0-9]+}}, 202 // Skip temp vars for loop 203 // CHECK: alloca i{{[0-9]+}}, 204 // CHECK: alloca i{{[0-9]+}}, 205 // CHECK: alloca i{{[0-9]+}}, 206 // CHECK: alloca i{{[0-9]+}}, 207 // CHECK: alloca i{{[0-9]+}}, 208 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 209 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]], 210 // CHECK: [[AGG_TMP1:%.+]] = alloca [[ST_TY]], 211 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]], 212 // CHECK: [[AGG_TMP2:%.+]] = alloca [[ST_TY]], 213 214 // param copy 215 // CHECK: store [2 x i{{[0-9]+}}]* {{.+}}, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 216 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[T_VAR_ADDR]], 217 // CHECK: store [2 x [[S_FLOAT_TY]]]* {{.+}}, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 218 // CHECK: store [[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]** [[VAR_ADDR]], 219 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[SIVAR_ADDR]], 220 221 // T_VAR and SIVAR 222 // CHECK-64-DAG: [[CONV_TVAR:%.+]] = bitcast i64* [[T_VAR_ADDR]] to i32* 223 // CHECK-64-DAG: [[CONV_SIVAR:%.+]] = bitcast i64* [[SIVAR_ADDR]] to i32* 224 225 // preparation vars 226 // CHECK-DAG: [[VEC_ADDR_VAL:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 227 // CHECK-DAG: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_ADDR]], 228 // CHECK-DAG: [[VAR_ADDR_REF:%.+]] = load{{.+}} [[VAR_ADDR]], 229 230 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2 231 // CHECK-DAG: [[VEC_DEST_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 232 // CHECK-DAG: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_VAL]] to i8* 233 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VEC_DEST_PRIV]], i8* align {{[0-9]+}} [[VEC_SRC]], {{.+}}) 234 235 // firstprivate(s_arr) 236 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], 237 // CHECK-DAG: [[S_ARR_ADDR_BGN:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_ADDR_REF]] to 238 // CHECK-DAG: [[S_ARR_FIN:%.+]] = icmp{{.+}} [[S_ARR_PRIV_BGN]], 239 // CHECK-DAG: [[S_ARR_SRC_COPY:%.+]] = phi{{.+}} [ [[S_ARR_ADDR_BGN]], {{.+}} ], [ [[S_ARR_SRC:%.+]], {{.+}} ] 240 // CHECK-DAG: [[S_ARR_DST_COPY:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}}], [ [[S_ARR_DST:%.+]], {{.+}} ] 241 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]]) 242 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_DST_COPY]], {{.+}} [[S_ARR_SRC_COPY]], {{.+}} [[AGG_TMP1]]) 243 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]]) 244 // CHECK-DAG: [[S_ARR_DST]] = getelementptr {{.+}} [[S_ARR_DST_COPY]], 245 // CHECK-DAG: [[S_ARR_SRC]] = getelementptr {{.+}} [[S_ARR_SRC_COPY]], 246 247 // firstprivate(var) 248 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]]) 249 // CHECK-DAG: call void @{{.+}}({{.+}} [[VAR_PRIV]], {{.+}} [[VAR_ADDR_REF]], {{.+}} [[AGG_TMP2]]) 250 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]]) 251 252 // CHECK: call void @__kmpc_for_static_init_4( 253 // CHECK-32-DAG: {{.+}} = {{.+}} [[T_VAR_ADDR]] 254 // CHECK-64-DAG: {{.+}} = {{.+}} [[CONV_TVAR]] 255 // CHECK-DAG: {{.+}} = {{.+}} [[VEC_PRIV]] 256 // CHECK-DAG: {{.+}} = {{.+}} [[S_ARR_PRIV]] 257 // CHECK-DAG: {{.+}} = {{.+}} [[VAR_PRIV]] 258 // CHECK-32-DAG: {{.+}} = {{.+}} [[SIVAR_ADDR]] 259 // CHECK-64-DAG: {{.+}} = {{.+}} [[CONV_SIVAR]] 260 // CHECK: call void @__kmpc_for_static_fini( 261 // CHECK: ret void 262 263 // CHECK: define{{.*}} i{{[0-9]+}} @[[TMAIN_INT]]() 264 // CHECK: call i32 @__tgt_target_teams_mapper(%struct.ident_t* @{{.+}}, i64 -1, i8* @{{[^,]+}}, i32 4, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i8** null, i8** null, i32 0, i32 1) 265 // CHECK: call void @[[TOFFL1:.+]](i{{64|32}} %{{.+}}) 266 // CHECK: ret 267 268 // CHECK: define {{.*}}void @[[TOFFL1]]({{.+}}) 269 // CHECK: [[TT_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, 270 // CHECK: [[TVEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}]*, 271 // CHECK: [[TS_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]]*, 272 // CHECK: [[TVAR_PRIV:%.+]] = alloca [[S_INT_TY]]*, 273 // CHECK: [[TT_VAR_CAST:%.+]] = alloca i{{[0-9]+}}, 274 275 // CHECK-DAG: [[TVEC_TE_PAR:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[TVEC_PRIV]], 276 // CHECK-DAG: [[TT_VAR_TE_PAR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[TT_VAR_CAST]], 277 // CHECK-DAG: [[TS_ARR_TE_PAR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[TS_ARR_PRIV]], 278 // CHECK-DAG: [[TVAR_TE_PAR:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[TVAR_PRIV]], 279 280 // CHECK: [[TVAR_TE_PAR:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** % 281 // CHECK: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 4, {{.+}} @[[TOUTL1:.+]] to {{.+}}, [2 x i{{[0-9]+}}]* [[TVEC_TE_PAR]], i{{[0-9]+}} [[TT_VAR_TE_PAR]], [2 x [[S_INT_TY]]]* [[TS_ARR_TE_PAR]], [[S_INT_TY]]* [[TVAR_TE_PAR]]) 282 // CHECK: ret void 283 284 // CHECK: define internal void @[[TOUTL1]]({{.+}}) 285 // Skip global and bound tid vars 286 // CHECK: {{.+}} = alloca i32*, 287 // CHECK: {{.+}} = alloca i32*, 288 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i{{[0-9]+}}]*, 289 // CHECK: [[T_VAR_ADDR:%.+]] = alloca i{{[0-9]+}}, 290 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*, 291 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*, 292 // CHECK: [[TMP_VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*, 293 // Skip temp vars for loop 294 // CHECK: alloca i{{[0-9]+}}, 295 // CHECK: alloca i{{[0-9]+}}, 296 // CHECK: alloca i{{[0-9]+}}, 297 // CHECK: alloca i{{[0-9]+}}, 298 // CHECK: alloca i{{[0-9]+}}, 299 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], 300 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], 301 // CHECK: [[AGG_TMP1:%.+]] = alloca [[ST_TY]], 302 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], 303 // CHECK: [[AGG_TMP2:%.+]] = alloca [[ST_TY]], 304 // CHECK: [[TMP:%.+]] = alloca [[S_INT_TY]]*, 305 306 // param copy 307 // CHECK: store [2 x i{{[0-9]+}}]* {{.+}}, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 308 // CHECK: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[T_VAR_ADDR]], 309 // CHECK: store [2 x [[S_INT_TY]]]* {{.+}}, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR]], 310 // CHECK: store [[S_INT_TY]]* {{.+}}, [[S_INT_TY]]** [[VAR_ADDR]], 311 312 313 // T_VAR and preparation variables 314 // CHECK: [[VEC_ADDR_VAL:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_ADDR]], 315 // CHECK-64: [[CONV_TVAR:%.+]] = bitcast i64* [[T_VAR_ADDR]] to i32* 316 // CHECK: [[S_ARR_ADDR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_ADDR]], 317 318 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2 319 // CHECK-DAG: [[VEC_DEST_PRIV:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8* 320 // CHECK-DAG: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_ADDR_VAL]] to i8* 321 // CHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[VEC_DEST_PRIV]], i8* align {{[0-9]+}} [[VEC_SRC]], {{.+}}) 322 323 // firstprivate(s_arr) 324 // CHECK-DAG: [[S_ARR_PRIV_BGN:%.+]] = getelementptr{{.*}} [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], 325 // CHECK-DAG: [[S_ARR_ADDR_BGN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_ADDR_REF]] to 326 // CHECK-DAG: [[S_ARR_FIN:%.+]] = icmp{{.+}} [[S_ARR_PRIV_BGN]], 327 // CHECK-DAG: [[S_ARR_SRC_COPY:%.+]] = phi{{.+}} [ [[S_ARR_ADDR_BGN]], {{.+}} ], [ [[S_ARR_SRC:%.+]], {{.+}} ] 328 // CHECK-DAG: [[S_ARR_DST_COPY:%.+]] = phi{{.+}} [ [[S_ARR_PRIV_BGN]], {{.+}} ], [ [[S_ARR_DST:%.+]], {{.+}} ] 329 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]]) 330 // CHECK-DAG: call void @{{.+}}({{.+}} [[S_ARR_DST_COPY]], {{.+}} [[S_ARR_SRC_COPY]], {{.+}} [[AGG_TMP1]]) 331 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP1]]) 332 // CHECK-DAG: [[S_ARR_DST]] = getelementptr {{.+}} [[S_ARR_DST_COPY]], 333 // CHECK-DAG: [[S_ARR_SRC]] = getelementptr {{.+}} [[S_ARR_SRC_COPY]], 334 335 // firstprivate(var) 336 // CHECK-DAG: [[VAR_ADDR_REF:%.+]] = load{{.+}} [[TMP_VAR_ADDR]], 337 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]]) 338 // CHECK-DAG: call void @{{.+}}({{.+}} [[VAR_PRIV]], {{.+}} [[VAR_ADDR_REF]], {{.+}} [[AGG_TMP2]]) 339 // CHECK-DAG: call void @{{.+}}({{.+}} [[AGG_TMP2]]) 340 // CHECK-DAG: store [[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]** [[TMP]], 341 342 // CHECK: call void @__kmpc_for_static_init_4( 343 // CHECK-32-DAG: {{.+}} = {{.+}} [[T_VAR_ADDR]] 344 // CHECK-64-DAG: {{.+}} = {{.+}} [[CONV_TVAR]] 345 // CHECK-DAG: {{.+}} = {{.+}} [[VEC_PRIV]] 346 // CHECK-DAG: {{.+}} = {{.+}} [[TMP]] 347 // CHECK-DAG: {{.+}} = {{.+}} [[S_ARR_PRIV]] 348 // CHECK: call void @__kmpc_for_static_fini( 349 // CHECK: ret void 350 351 // CHECK: !{!"llvm.loop.vectorize.enable", i1 true} 352 #endif 353