[OPENMP] Codegen for untied tasks.

If the untied clause is present on a task construct, any thread in the team can resume the task region after a suspension. Patch adds proper codegen for untied tasks.

llvm-svn: 266722

3 files changed, 53 insertions, 13 deletions

diff --git a/clang/test/OpenMP/task_codegen.cpp b/clang/test/OpenMP/task_codegen.cpp
index b262745704c..3bf8253f2a8 100644
--- a/clang/test/OpenMP/task_codegen.cpp
+++ b/clang/test/OpenMP/task_codegen.cpp
@@ -230,6 +230,19 @@ int main() {
     a = 4;
     c = 5;
   }
+// CHECK: [[ORIG_TASK_PTR:%.+]] = call i8* @__kmpc_omp_task_alloc([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i32 0, i64 32, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_T]]{{.*}}*)* [[TASK_ENTRY6:@.+]] to i32 (i32, i8*)*))
+// CHECK: [[DESTRUCTORS_REF_PTR:%.+]] = getelementptr inbounds [[KMP_TASK_T]]{{.*}}* {{%.+}}, i32 0, i32 3
+// CHECK: store i32 (i32, i8*)* null, i32 (i32, i8*)** [[DESTRUCTORS_REF_PTR]]
+// CHECK: call i32 @__kmpc_omp_task([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]])
+#pragma omp task untied
+  {
+    S s1;
+#pragma omp task
+    a = 4;
+#pragma omp taskyield
+    s1 = S();
+#pragma omp taskwait
+  }
   return a;
 }
 // CHECK: define internal i32 [[TASK_ENTRY1]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
@@ -240,16 +253,42 @@ int main() {
 // CHECK: store i32 10, i32* %{{.+}}
 
 // CHECK: define internal i32 [[TASK_ENTRY2]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
-// CHECK: store i32 1, i32* [[A_PTR:@.+]]
+// CHECK: store i32 1, i32* [[A_PTR]]
 
 // CHECK: define internal i32 [[TASK_ENTRY3]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
-// CHECK: store i32 2, i32* [[A_PTR:@.+]]
+// CHECK: store i32 2, i32* [[A_PTR]]
 
 // CHECK: define internal i32 [[TASK_ENTRY4]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
-// CHECK: store i32 3, i32* [[A_PTR:@.+]]
+// CHECK: store i32 3, i32* [[A_PTR]]
 
 // CHECK: define internal i32 [[TASK_ENTRY5]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
-// CHECK: store i32 4, i32* [[A_PTR:@.+]]
+// CHECK: store i32 4, i32* [[A_PTR]]
 // CHECK: store i32 5, i32* [[C_PTR:%.+]], align 128
+
+// CHECK: define internal i32
+// CHECK: store i32 4, i32* [[A_PTR]]
+
+// CHECK: define internal i32 [[TASK_ENTRY6]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
+// CHECK: switch i32 %{{.+}}, label
+// CHECK: load i32*, i32** %
+// CHECK: store i32 1, i32* %
+// CHECK: call i32 @__kmpc_omp_task(%
+
+// CHECK: call i8* @__kmpc_omp_task_alloc(
+// CHECK: store i32 (i32, i8*)* null, i32 (i32, i8*)** %
+// CHECK: call i32 @__kmpc_omp_task(%
+// CHECK: load i32*, i32** %
+// CHECK: store i32 2, i32* %
+// CHECK: call i32 @__kmpc_omp_task(%
+
+// CHECK: call i32 @__kmpc_omp_taskyield(%
+// CHECK: load i32*, i32** %
+// CHECK: store i32 3, i32* %
+// CHECK: call i32 @__kmpc_omp_task(%
+
+// CHECK: call i32 @__kmpc_omp_taskwait(%
+// CHECK: load i32*, i32** %
+// CHECK: store i32 4, i32* %
+// CHECK: call i32 @__kmpc_omp_task(%
 #endif
 
diff --git a/clang/test/OpenMP/task_firstprivate_codegen.cpp b/clang/test/OpenMP/task_firstprivate_codegen.cpp
index e2244140d1e..8abc603b8d9 100644
--- a/clang/test/OpenMP/task_firstprivate_codegen.cpp
+++ b/clang/test/OpenMP/task_firstprivate_codegen.cpp
@@ -422,11 +422,11 @@ int main() {
 // CHECK: ret void
 
 // CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_TMAIN_TY]]* noalias)
-
-// CHECK: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
-// CHECK: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
-// CHECK: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*,
-// CHECK: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*,
+// CHECK: alloca i32*,
+// CHECK-DAG: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
+// CHECK-DAG: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
+// CHECK-DAG: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*,
+// CHECK-DAG: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*,
 // CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_TMAIN_TY]]*, i32**, [2 x i32]**, [2 x [[S_INT_TY]]]**, [[S_INT_TY]]**)* [[PRIVATES_MAP_FN]] to void (i8*, ...)*), void (i8*, ...)** [[MAP_FN_ADDR:%.+]],
 // CHECK: [[MAP_FN:%.+]] = load void (i8*, ...)*, void (i8*, ...)** [[MAP_FN_ADDR]],
 // CHECK: call void (i8*, ...) [[MAP_FN]](i8* %{{.+}}, i32** [[PRIV_T_VAR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]], [2 x [[S_INT_TY]]]** [[PRIV_S_ARR_ADDR]], [[S_INT_TY]]** [[PRIV_VAR_ADDR]])
diff --git a/clang/test/OpenMP/task_private_codegen.cpp b/clang/test/OpenMP/task_private_codegen.cpp
index 1455fd11a91..a2dc8423264 100644
--- a/clang/test/OpenMP/task_private_codegen.cpp
+++ b/clang/test/OpenMP/task_private_codegen.cpp
@@ -331,10 +331,11 @@ int main() {
 
 // CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_TMAIN_TY]]* noalias)
 
-// CHECK: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
-// CHECK: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
-// CHECK: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*,
-// CHECK: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*,
+// CHECK: alloca i32*,
+// CHECK-DAG: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
+// CHECK-DAG: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
+// CHECK-DAG: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]]*,
+// CHECK-DAG: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_INT_TY]]*,
 // CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_TMAIN_TY]]*, i32**, [2 x i32]**, [2 x [[S_INT_TY]]]**, [[S_INT_TY]]**)* [[PRIVATES_MAP_FN]] to void (i8*, ...)*), void (i8*, ...)** [[MAP_FN_ADDR:%.+]],
 // CHECK: [[MAP_FN:%.+]] = load void (i8*, ...)*, void (i8*, ...)** [[MAP_FN_ADDR]],
 // CHECK: call void (i8*, ...) [[MAP_FN]](i8* %{{.+}}, i32** [[PRIV_T_VAR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]], [2 x [[S_INT_TY]]]** [[PRIV_S_ARR_ADDR]], [[S_INT_TY]]** [[PRIV_VAR_ADDR]])