[OPENMP][NVPTX]Improve emission of the globalized variables for

target/teams/distribute regions.

Target/teams/distribute regions exist for all the time the kernel is
executed. Thus, if the variable is declared in their context and then
escape it, we can allocate global memory statically instead of
allocating it dynamically.
Patch captures all the globalized variables in target/teams/distribute
contexts, merges them into the records, one per each target region.
Those records are then joined into the union, one per compilation unit
(to save the global memory). Those units are organized into
2 x dimensional arrays, where the first dimension is
the number of blocks per SM and the second one is the number of SMs.
Runtime functions manage this global memory space between the executing
teams.

llvm-svn: 345978

1 files changed, 12 insertions, 1 deletions

diff --git a/clang/test/OpenMP/nvptx_parallel_codegen.cpp b/clang/test/OpenMP/nvptx_parallel_codegen.cpp
index d1a3104407d..5f0f09af870 100644
--- a/clang/test/OpenMP/nvptx_parallel_codegen.cpp
+++ b/clang/test/OpenMP/nvptx_parallel_codegen.cpp
@@ -71,6 +71,12 @@ int bar(int n){
   return a;
 }
 
+// CHECK: [[MEM_TY:%.+]] = type { [4 x i8] }
+// CHECK-DAG: [[GLOBAL_RD:@.+]] = weak global [{{[0-9]+}} x [{{[0-9]+}} x [[MEM_TY]]]] zeroinitializer
+// CHECK-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)
+// CHECK-DAG: [[KERNEL_PTR:@.+]] = internal addrspace(3) global i8* null
+// CHECK-DAG: [[KERNEL_SIZE:@.+]] = internal unnamed_addr constant i{{64|32}} 4
+
 // CHECK-NOT: define {{.*}}void {{@__omp_offloading_.+template.+l17}}_worker()
 
 // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+template.+l26}}_worker()
@@ -318,12 +324,17 @@ int bar(int n){
 // CHECK-32: [[A_ADDR:%.+]] = alloca i32,
 // CHECK-64: [[A_ADDR:%.+]] = alloca i64,
 // CHECK-64: [[CONV:%.+]] = bitcast i64* [[A_ADDR]] to i32*
-// CHECK: [[STACK:%.+]] = call i8* @__kmpc_data_sharing_push_stack(i{{64|32}} 4, i16 0)
+// CHECK: [[GLOBAL_RD:%.+]] = load i8*, i8** [[GLOBAL_RD_PTR]],
+// CHECK: [[SIZE:%.+]] = load i{{64|32}}, i{{64|32}}* [[KERNEL_SIZE]],
+// CHECK: 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**))
+// CHECK: [[KERNEL_RD:%.+]] = load i8*, i8* addrspace(3)* [[KERNEL_PTR]],
+// CHECK: [[STACK:%.+]] = getelementptr inbounds i8, i8* [[KERNEL_RD]], i{{64|32}} 0
 // CHECK: [[BC:%.+]] = bitcast i8* [[STACK]] to %struct._globalized_locals_ty*
 // CHECK-32: [[A:%.+]] = load i32, i32* [[A_ADDR]],
 // CHECK-64: [[A:%.+]] = load i32, i32* [[CONV]],
 // CHECK: [[GLOBAL_A_ADDR:%.+]] = getelementptr inbounds %struct._globalized_locals_ty, %struct._globalized_locals_ty* [[BC]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: store i32 [[A]], i32* [[GLOBAL_A_ADDR]],
+// CHECK: call void @__kmpc_restore_team_static_memory(i16 0)
 
 // CHECK-LABEL: define internal void @{{.+}}(i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable{{.*}})
 // CHECK:  [[CC:%.+]] = alloca i32,