From b4c74c6603cd1951bffde8b188e7d785a37ab479 Mon Sep 17 00:00:00 2001
From: Gheorghe-Teodor Bercea <gheorghe-teod.bercea@ibm.com>
Date: Tue, 12 Dec 2017 21:38:43 +0000
Subject: [PATCH] [OpenMP] Add function attribute for triggering data sharing.

Summary:
The backend should only emit data sharing code for the cases where it is needed.
A new function attribute is used by Clang to enable data sharing only for the cases where OpenMP semantics require it and there are variables that need to be shared.

Reviewers: hfinkel, Hahnfeld, ABataev, carlo.bertolli, caomhin

Reviewed By: ABataev

Subscribers: cfe-commits, jholewinski

Differential Revision: https://reviews.llvm.org/D41123

llvm-svn: 320527
---
 clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp |  2 ++
 clang/test/OpenMP/nvptx_data_sharing.cpp   | 11 ++++++++---
 2 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp b/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
index bc769f4f4953c..b5fc8d308067e 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -942,6 +942,8 @@ void CGOpenMPRuntimeNVPTX::emitGenericParallelCall(
     llvm::Value *ID = Bld.CreateBitOrPointerCast(WFn, CGM.Int8PtrTy);
 
     if (!CapturedVars.empty()) {
+      // There's somehting to share, add the attribute
+      CGF.CurFn->addFnAttr("has-nvptx-shared-depot");
       // Prepare for parallel region. Indicate the outlined function.
       Address SharedArgs =
           CGF.CreateDefaultAlignTempAlloca(CGF.VoidPtrPtrTy,
diff --git a/clang/test/OpenMP/nvptx_data_sharing.cpp b/clang/test/OpenMP/nvptx_data_sharing.cpp
index 4b8d2df1dab3e..65215cd2c4f97 100644
--- a/clang/test/OpenMP/nvptx_data_sharing.cpp
+++ b/clang/test/OpenMP/nvptx_data_sharing.cpp
@@ -22,7 +22,7 @@ void test_ds(){
 
 /// ========= In the worker function ========= ///
 
-// CK1: define internal void @__omp_offloading_{{.*}}test_ds{{.*}}worker(){{.*}}{
+// CK1: define internal void @__omp_offloading_{{.*}}test_ds{{.*}}worker() [[ATTR1:#.*]] {
 // CK1: [[SHAREDARGS:%.+]] = alloca i8**
 // CK1: call i1 @__kmpc_kernel_parallel(i8** %work_fn, i8*** [[SHAREDARGS]])
 // CK1: [[SHARGSTMP:%.+]] = load i8**, i8*** [[SHAREDARGS]]
@@ -30,7 +30,7 @@ void test_ds(){
 
 /// ========= In the kernel function ========= ///
 
-// CK1: {{.*}}define void @__omp_offloading{{.*}}test_ds{{.*}}()
+// CK1: {{.*}}define void @__omp_offloading{{.*}}test_ds{{.*}}() [[ATTR2:#.*]] {
 // CK1: [[SHAREDARGS1:%.+]] = alloca i8**
 // CK1: call void @__kmpc_kernel_prepare_parallel({{.*}}, i8*** [[SHAREDARGS1]], i32 1)
 // CK1: [[SHARGSTMP1:%.+]] = load i8**, i8*** [[SHAREDARGS1]]
@@ -40,7 +40,7 @@ void test_ds(){
 
 /// ========= In the data sharing wrapper function ========= ///
 
-// CK1: {{.*}}define internal void @__omp_outlined___wrapper({{.*}}i8**){{.*}}{
+// CK1: {{.*}}define internal void @__omp_outlined___wrapper({{.*}}i8**) [[ATTR1]] {
 // CK1: [[SHAREDARGS2:%.+]] = alloca i8**
 // CK1: store i8** %2, i8*** [[SHAREDARGS2]]
 // CK1: [[SHARGSTMP3:%.+]] = load i8**, i8*** [[SHAREDARGS2]]
@@ -49,4 +49,9 @@ void test_ds(){
 // CK1: [[SHARGSTMP6:%.+]] = load i32*, i32** [[SHARGSTMP5]]
 // CK1: call void @__omp_outlined__({{.*}}, i32* [[SHARGSTMP6]])
 
+/// ========= Attributes ========= ///
+
+// CK1-NOT: attributes [[ATTR1]] = { {{.*}}"has-nvptx-shared-depot"{{.*}} }
+// CK1: attributes [[ATTR2]] = { {{.*}}"has-nvptx-shared-depot"{{.*}} }
+
 #endif