[OPENMP][NVPTX] Fix emission of __kmpc_global_thread_num() for non-SPMD

mode.

__kmpc_global_thread_num() should be called before initialization of the
runtime.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@343857 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp b/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
index f0f0a735f8a79fc1aa997d835f38050bae7ececd..b75891e4f21e1c477fe36c3927e4639fc7010930 100644 (file)
--- a/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ b/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -1083,12 +1083,15 @@ void CGOpenMPRuntimeNVPTX::emitNonSPMDKernel(const OMPExecutableDirective &D,
                          CGOpenMPRuntimeNVPTX::WorkerFunctionState &WST)
         : EST(EST), WST(WST) {}
     void Enter(CodeGenFunction &CGF) override {
-      static_cast<CGOpenMPRuntimeNVPTX &>(CGF.CGM.getOpenMPRuntime())
-          .emitNonSPMDEntryHeader(CGF, EST, WST);
+      auto &RT = static_cast<CGOpenMPRuntimeNVPTX &>(CGF.CGM.getOpenMPRuntime());
+      RT.emitNonSPMDEntryHeader(CGF, EST, WST);
+      // Skip target region initialization.
+      RT.setLocThreadIdInsertPt(CGF, /*AtCurrentPoint=*/true);
     }
     void Exit(CodeGenFunction &CGF) override {
-      static_cast<CGOpenMPRuntimeNVPTX &>(CGF.CGM.getOpenMPRuntime())
-          .emitNonSPMDEntryFooter(CGF, EST);
+      auto &RT = static_cast<CGOpenMPRuntimeNVPTX &>(CGF.CGM.getOpenMPRuntime());
+      RT.clearLocThreadIdInsertPt(CGF);
+      RT.emitNonSPMDEntryFooter(CGF, EST);
     }
   } Action(EST, WST);
   CodeGen.setAction(Action);

diff --git a/test/OpenMP/nvptx_teams_codegen.cpp b/test/OpenMP/nvptx_teams_codegen.cpp
index 2c50f037b2280b1e30114796ab37beb90c8a95fd..4e3f2674fc5f55169ce052b2673ba97deafab788 100644 (file)
--- a/test/OpenMP/nvptx_teams_codegen.cpp
+++ b/test/OpenMP/nvptx_teams_codegen.cpp
@@ -105,7 +105,6 @@ int main (int argc, char **argv) {
 // CK2: [[AADDR:%.+]] = alloca i{{[0-9]+}},
 // CK2: [[BADDR:%.+]] = alloca i{{[0-9]+}},
 // CK2: [[ARGCADDR:%.+]] = alloca i{{[0-9]+}},
-// CK2:  {{%.+}} = call i32 @__kmpc_global_thread_num(
 // 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_IN]], i{{[0-9]+}}* [[ARGCADDR]],
@@ -117,6 +116,7 @@ int main (int argc, char **argv) {
 // 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
 // CK2:  store i{{[0-9]+}} [[ARG]], i{{[0-9]+}}* [[ARGCADDR]],
+// CK2:  {{%.+}} = call i32 @__kmpc_global_thread_num(
 // CK2:  store i{{[0-9]+}}* [[ARGCADDR]], i{{[0-9]+}}** [[ARGCADDR_PTR]],
 // CK2:  [[ARGCADDR_PTR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[ARGCADDR_PTR]],
 // CK2: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[ARGCADDR_PTR_REF]],
@@ -129,7 +129,6 @@ int main (int argc, char **argv) {
 // CK2: [[AADDR:%.+]] = alloca i{{[0-9]+}},
 // CK2: [[BADDR:%.+]] = alloca i{{[0-9]+}},
 // CK2: [[ARGCADDR:%.+]] = alloca i{{[0-9]+}}**,
-// CK2: {{%.+}} = call i32 @__kmpc_global_thread_num(
 // 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]],
@@ -137,6 +136,7 @@ int main (int argc, char **argv) {
 // 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]],
+// CK2: {{%.+}} = call i32 @__kmpc_global_thread_num(
 // CK2: store i{{[0-9]+}}*** [[ARGCADDR]], i{{[0-9]+}}**** [[ARGCADDR_PTR]],
 // CK2: [[ARGCADDR_PTR_REF:%.+]] = load i{{[0-9]+}}***, i{{[0-9]+}}**** [[ARGCADDR_PTR]],
 // CK2: store i{{[0-9]+}}** null, i{{[0-9]+}}*** [[ARGCADDR_PTR_REF]],