summaryrefslogtreecommitdiffstats
path: root/clang/test/OpenMP/nvptx_teams_codegen.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'clang/test/OpenMP/nvptx_teams_codegen.cpp')
-rw-r--r--clang/test/OpenMP/nvptx_teams_codegen.cpp38
1 files changed, 34 insertions, 4 deletions
diff --git a/clang/test/OpenMP/nvptx_teams_codegen.cpp b/clang/test/OpenMP/nvptx_teams_codegen.cpp
index 4e3f2674fc5..af1ae40cdea 100644
--- a/clang/test/OpenMP/nvptx_teams_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_teams_codegen.cpp
@@ -27,6 +27,13 @@ int main (int argc, char **argv) {
return tmain(argv);
}
+// CK1: [[MEM_TY:%.+]] = type { [{{4|8}} x i8] }
+// CK1-DAG: [[GLOBAL_RD:@.+]] = weak global [{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]] zeroinitializer
+// CK1-DAG: [[GLOBAL_RD_PTR:@.+]] = weak unnamed_addr constant i8* getelementptr inbounds ([{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]], [{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]]* [[GLOBAL_RD]], i{{[0-9]+}} 0, i{{[0-9]+}} 0, i{{[0-9]+}} 0, i{{[0-9]+}} 0, i{{[0-9]+}} 0)
+// CK1-DAG: [[KERNEL_PTR:@.+]] = internal addrspace(3) global i8* null
+// CK1-DAG: [[KERNEL_SIZE1:@.+]] = internal unnamed_addr constant i{{64|32}} 4
+// CK1-DAG: [[KERNEL_SIZE2:@.+]] = internal unnamed_addr constant i{{64|32}} {{8|4}}
+
// only nvptx side: do not outline teams region and do not call fork_teams
// CK1: define {{.*}}void @{{[^,]+}}(i{{[0-9]+}} [[ARGC:%.+]])
// CK1: {{.+}} = alloca i{{[0-9]+}}*,
@@ -36,7 +43,11 @@ int main (int argc, char **argv) {
// CK1: store {{.+}} 0, {{.+}},
// CK1: store i{{[0-9]+}} [[ARGC]], i{{[0-9]+}}* [[ARGCADDR]],
// CK1-64: [[CONV:%.+]] = bitcast i{{[0-9]+}}* [[ARGCADDR]] to i{{[0-9]+}}*
-// CK1: call i8* @__kmpc_data_sharing_push_stack(i{{[0-9]+}} 4, i16 0)
+// CK1: [[GLOBAL_RD:%.+]] = load i8*, i8** [[GLOBAL_RD_PTR]],
+// CK1: [[SIZE:%.+]] = load i{{64|32}}, i{{64|32}}* [[KERNEL_SIZE1]],
+// CK1: call void @__kmpc_get_team_static_memory(i8* [[GLOBAL_RD]], i{{64|32}} [[SIZE]], i16 0, i8** addrspacecast (i8* addrspace(3)* [[KERNEL_PTR]] to i8**))
+// CK1: [[KERNEL_RD:%.+]] = load i8*, i8* addrspace(3)* [[KERNEL_PTR]],
+// CK1: [[GLOBALSTACK:%.+]] = getelementptr inbounds i8, i8* [[KERNEL_RD]], i{{64|32}} 0
// CK1-64: [[ARG:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[CONV]]
// CK1-32: [[ARG:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[ARGCADDR]]
// CK1: [[ARGCADDR:%.+]] = getelementptr inbounds %struct.{{.*}}, %struct.{{.*}}* %{{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
@@ -53,7 +64,11 @@ int main (int argc, char **argv) {
// CK1: [[ARGCADDR_PTR:%.+]] = alloca i{{.+}}***,
// CK1: [[ARGCADDR:%.+]] = alloca i{{.+}}**,
// CK1: store i{{.+}}** [[ARGC]], i{{.+}}*** [[ARGCADDR]]
-// CK1: call i8* @__kmpc_data_sharing_push_stack(i{{[0-9]+}} {{4|8}}, i16 0)
+// CK1: [[GLOBAL_RD:%.+]] = load i8*, i8** [[GLOBAL_RD_PTR]],
+// CK1: [[SIZE:%.+]] = load i{{64|32}}, i{{64|32}}* [[KERNEL_SIZE2]],
+// CK1: call void @__kmpc_get_team_static_memory(i8* [[GLOBAL_RD]], i{{64|32}} [[SIZE]], i16 0, i8** addrspacecast (i8* addrspace(3)* [[KERNEL_PTR]] to i8**))
+// CK1: [[KERNEL_RD:%.+]] = load i8*, i8* addrspace(3)* [[KERNEL_PTR]],
+// CK1: [[GLOBALSTACK:%.+]] = getelementptr inbounds i8, i8* [[KERNEL_RD]], i{{64|32}} 0
// CK1: [[ARG:%.+]] = load i{{[0-9]+}}**, i{{[0-9]+}}*** [[ARGCADDR]]
// CK1: [[ARGCADDR:%.+]] = getelementptr inbounds %struct.{{.*}}, %struct.{{.*}}* %{{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CK1: store i{{[0-9]+}}** [[ARG]], i{{[0-9]+}}*** [[ARGCADDR]],
@@ -98,6 +113,13 @@ int main (int argc, char **argv) {
return tmain(argv);
}
+// CK2: [[MEM_TY:%.+]] = type { [{{4|8}} x i8] }
+// CK2-DAG: [[GLOBAL_RD:@.+]] = weak global [{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]] zeroinitializer
+// CK2-DAG: [[GLOBAL_RD_PTR:@.+]] = weak unnamed_addr constant i8* getelementptr inbounds ([{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]], [{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]]* [[GLOBAL_RD]], i{{[0-9]+}} 0, i{{[0-9]+}} 0, i{{[0-9]+}} 0, i{{[0-9]+}} 0, i{{[0-9]+}} 0)
+// CK2-DAG: [[KERNEL_PTR:@.+]] = internal addrspace(3) global i8* null
+// CK2-DAG: [[KERNEL_SIZE1:@.+]] = internal unnamed_addr constant i{{64|32}} 4
+// CK2-DAG: [[KERNEL_SIZE2:@.+]] = internal unnamed_addr constant i{{64|32}} {{8|4}}
+
// CK2: define {{.*}}void @{{[^,]+}}(i{{[0-9]+}} [[A_IN:%.+]], i{{[0-9]+}} [[B_IN:%.+]], i{{[0-9]+}} [[ARGC_IN:.+]])
// CK2: {{.}} = alloca i{{[0-9]+}}*,
// CK2: {{.}} = alloca i{{[0-9]+}}*,
@@ -111,7 +133,11 @@ int main (int argc, char **argv) {
// CK2-64: [[ACONV:%.+]] = bitcast i64* [[AADDR]] to i32*
// CK2-64: [[BCONV:%.+]] = bitcast i64* [[BADDR]] to i32*
// CK2-64: [[CONV:%.+]] = bitcast i64* [[ARGCADDR]] to i32*
-// CK2: call i8* @__kmpc_data_sharing_push_stack(i{{[0-9]+}} 4, i16 0)
+// CK2: [[GLOBAL_RD:%.+]] = load i8*, i8** [[GLOBAL_RD_PTR]],
+// CK2: [[SIZE:%.+]] = load i{{64|32}}, i{{64|32}}* [[KERNEL_SIZE1]],
+// CK2: call void @__kmpc_get_team_static_memory(i8* [[GLOBAL_RD]], i{{64|32}} [[SIZE]], i16 0, i8** addrspacecast (i8* addrspace(3)* [[KERNEL_PTR]] to i8**))
+// CK2: [[KERNEL_RD:%.+]] = load i8*, i8* addrspace(3)* [[KERNEL_PTR]],
+// CK2: [[GLOBALSTACK:%.+]] = getelementptr inbounds i8, i8* [[KERNEL_RD]], i{{64|32}} 0
// CK2-64: [[ARG:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[CONV]]
// CK2-32: [[ARG:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[ARGCADDR]]
// CK2: [[ARGCADDR:%.+]] = getelementptr inbounds %struct.{{.*}}, %struct.{{.*}}* %{{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
@@ -132,7 +158,11 @@ int main (int argc, char **argv) {
// CK2: store i{{[0-9]+}} [[A_IN]], i{{[0-9]+}}* [[AADDR]],
// CK2: store i{{[0-9]+}} [[B_IN]], i{{[0-9]+}}* [[BADDR]],
// CK2: store i{{[0-9]+}}** [[ARGC]], i{{[0-9]+}}*** [[ARGCADDR]],
-// CK2: call i8* @__kmpc_data_sharing_push_stack(i{{[0-9]+}} {{4|8}}, i16 0)
+// CK2: [[GLOBAL_RD:%.+]] = load i8*, i8** [[GLOBAL_RD_PTR]],
+// CK2: [[SIZE:%.+]] = load i{{64|32}}, i{{64|32}}* [[KERNEL_SIZE2]],
+// CK2: call void @__kmpc_get_team_static_memory(i8* [[GLOBAL_RD]], i{{64|32}} [[SIZE]], i16 0, i8** addrspacecast (i8* addrspace(3)* [[KERNEL_PTR]] to i8**))
+// CK2: [[KERNEL_RD:%.+]] = load i8*, i8* addrspace(3)* [[KERNEL_PTR]],
+// CK2: [[GLOBALSTACK:%.+]] = getelementptr inbounds i8, i8* [[KERNEL_RD]], i{{64|32}} 0
// CK2: [[ARG:%.+]] = load i{{[0-9]+}}**, i{{[0-9]+}}*** [[ARGCADDR]]
// CK2: [[ARGCADDR:%.+]] = getelementptr inbounds %struct.{{.*}}, %struct.{{.*}}* %{{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
// CK2: store i{{[0-9]+}}** [[ARG]], i{{[0-9]+}}*** [[ARGCADDR]],
OpenPOWER on IntegriCloud