Vendor import of clang trunk r256633:vendor/clang/clang-trunk-r256633

https://llvm.org/svn/llvm-project/cfe/trunk@256633

193 files changed, 15277 insertions, 1621 deletions

diff --git a/test/OpenMP/atomic_capture_codegen.cpp b/test/OpenMP/atomic_capture_codegen.cpp
index 6dd9f7ab2a41..f91da3e80f54 100644
--- a/test/OpenMP/atomic_capture_codegen.cpp
+++ b/test/OpenMP/atomic_capture_codegen.cpp
@@ -72,7 +72,10 @@ struct BitFields4_packed {
 typedef float float2 __attribute__((ext_vector_type(2)));
 float2 float2x;
 
-register int rix __asm__("0");
+// Register "0" is currently an invalid register for global register variables.
+// Use "esp" instead of "0".
+// register int rix __asm__("0");
+register int rix __asm__("esp");
 
 int main() {
 // CHECK: [[PREV:%.+]] = atomicrmw add i8* @{{.+}}, i8 1 monotonic
diff --git a/test/OpenMP/atomic_codegen.cpp b/test/OpenMP/atomic_codegen.cpp
index 2ce1f9450b3f..536f2cdffafb 100644
--- a/test/OpenMP/atomic_codegen.cpp
+++ b/test/OpenMP/atomic_codegen.cpp
@@ -1,11 +1,12 @@
 // RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -x c++ -emit-llvm %s -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // expected-no-diagnostics
 
 int a;
 int b;
 
 struct St {
+  unsigned long field;
   St() {}
   ~St() {}
   int &get() { return a; }
@@ -13,6 +14,7 @@ struct St {
 
 // CHECK-LABEL: parallel_atomic_ewc
 void parallel_atomic_ewc() {
+  St s;
 #pragma omp parallel
   {
       // CHECK: invoke void @_ZN2StC1Ev(%struct.St* [[TEMP_ST_ADDR:%.+]])
@@ -47,6 +49,8 @@ void parallel_atomic_ewc() {
       // CHECK: invoke void @_ZN2StD1Ev(%struct.St* [[TEMP_ST_ADDR]])
 #pragma omp atomic
       St().get() %= b;
+#pragma omp atomic
+      s.field++;
       // CHECK: invoke void @_ZN2StC1Ev(%struct.St* [[TEMP_ST_ADDR:%.+]])
       // CHECK: [[SCALAR_ADDR:%.+]] = invoke dereferenceable(4) i32* @_ZN2St3getEv(%struct.St* [[TEMP_ST_ADDR]])
       // CHECK: [[B_VAL:%.+]] = load i32, i32* @b
@@ -108,7 +112,7 @@ void parallel_atomic() {
   // TERM_DEBUG:     call void @__clang_call_terminate
   // TERM_DEBUG:     unreachable
 }
-// TERM_DEBUG-DAG: [[READ_LOC]] = !DILocation(line: [[@LINE-33]],
-// TERM_DEBUG-DAG: [[WRITE_LOC]] = !DILocation(line: [[@LINE-28]],
-// TERM_DEBUG-DAG: [[UPDATE_LOC]] = !DILocation(line: [[@LINE-22]],
-// TERM_DEBUG-DAG: [[CAPTURE_LOC]] = !DILocation(line: [[@LINE-16]],
+// TERM_DEBUG-DAG: [[READ_LOC]] = !DILocation(line: [[@LINE-28]],
+// TERM_DEBUG-DAG: [[WRITE_LOC]] = !DILocation(line: [[@LINE-22]],
+// TERM_DEBUG-DAG: [[UPDATE_LOC]] = !DILocation(line: [[@LINE-16]],
+// TERM_DEBUG-DAG: [[CAPTURE_LOC]] = !DILocation(line: [[@LINE-9]],
diff --git a/test/OpenMP/atomic_messages.cpp b/test/OpenMP/atomic_messages.cpp
index c3e02bc96e86..7f78ad443da6 100644
--- a/test/OpenMP/atomic_messages.cpp
+++ b/test/OpenMP/atomic_messages.cpp
@@ -48,8 +48,8 @@ T read() {
   // expected-note@+1 {{expected built-in assignment operator}}
   foo();
 #pragma omp atomic read
-  // expected-error@+2 {{the statement for 'atomic read' must be an expression statement of form 'v = x;', where v and x are both lvalue expressions with scalar type}}
-  // expected-note@+1 {{expected built-in assignment operator}}
+  // expected-error@+2 2 {{the statement for 'atomic read' must be an expression statement of form 'v = x;', where v and x are both lvalue expressions with scalar type}}
+  // expected-note@+1 2 {{expected built-in assignment operator}}
   a += b;
 #pragma omp atomic read
   // expected-error@+2 {{the statement for 'atomic read' must be an expression statement of form 'v = x;', where v and x are both lvalue expressions with scalar type}}
@@ -93,7 +93,8 @@ int read() {
 #pragma omp atomic read read
   a = b;
 
-  // expected-note@+1 {{in instantiation of function template specialization 'read<S>' requested here}}
+  // expected-note@+2 {{in instantiation of function template specialization 'read<S>' requested here}}
+  // expected-note@+1 {{in instantiation of function template specialization 'read<int>' requested here}}
   return read<int>() + read<S>().a;
 }
 
@@ -147,6 +148,7 @@ int write() {
 #pragma omp atomic write
   a = foo();
 
+  // expected-note@+1 {{in instantiation of function template specialization 'write<int>' requested here}}
   return write<int>();
 }
 
@@ -672,6 +674,7 @@ int capture() {
 #pragma omp atomic capture capture
   b = a /= b;
 
+  // expected-note@+1 {{in instantiation of function template specialization 'capture<int>' requested here}}
   return capture<int>();
 }
 
diff --git a/test/OpenMP/atomic_read_codegen.c b/test/OpenMP/atomic_read_codegen.c
index fc47c82d8913..0cd46e3821fd 100644
--- a/test/OpenMP/atomic_read_codegen.c
+++ b/test/OpenMP/atomic_read_codegen.c
@@ -72,7 +72,10 @@ struct BitFields4_packed {
 typedef float float2 __attribute__((ext_vector_type(2)));
 float2 float2x;
 
-register int rix __asm__("0");
+// Register "0" is currently an invalid register for global register variables.
+// Use "esp" instead of "0".
+// register int rix __asm__("0");
+register int rix __asm__("esp");
 
 int main() {
 // CHECK: load atomic i8, i8*
diff --git a/test/OpenMP/atomic_update_codegen.cpp b/test/OpenMP/atomic_update_codegen.cpp
index df8b538ee165..063b76d7ff54 100644
--- a/test/OpenMP/atomic_update_codegen.cpp
+++ b/test/OpenMP/atomic_update_codegen.cpp
@@ -72,7 +72,10 @@ struct BitFields4_packed {
 typedef float float2 __attribute__((ext_vector_type(2)));
 float2 float2x;
 
-register int rix __asm__("0");
+// Register "0" is currently an invalid register for global register variables.
+// Use "esp" instead of "0".
+// register int rix __asm__("0");
+register int rix __asm__("esp");
 
 int main() {
 // CHECK-NOT: atomicrmw
diff --git a/test/OpenMP/atomic_write_codegen.c b/test/OpenMP/atomic_write_codegen.c
index 1ee26b078212..66172af07e8f 100644
--- a/test/OpenMP/atomic_write_codegen.c
+++ b/test/OpenMP/atomic_write_codegen.c
@@ -72,7 +72,10 @@ struct BitFields4_packed {
 typedef float float2 __attribute__((ext_vector_type(2)));
 float2 float2x;
 
-register int rix __asm__("0");
+// Register "0" is currently an invalid register for global register variables.
+// Use "esp" instead of "0".
+// register int rix __asm__("0");
+register int rix __asm__("esp");
 
 int main() {
 // CHECK: load i8, i8*
diff --git a/test/OpenMP/cancel_ast_print.cpp b/test/OpenMP/cancel_ast_print.cpp
index f244c784e86e..007237a7fb98 100644
--- a/test/OpenMP/cancel_ast_print.cpp
+++ b/test/OpenMP/cancel_ast_print.cpp
@@ -10,11 +10,11 @@ int main (int argc, char **argv) {
 // CHECK: int main(int argc, char **argv) {
 #pragma omp parallel
 {
-#pragma omp cancel parallel
+#pragma omp cancel parallel if(argc)
 }
 // CHECK: #pragma omp parallel
 // CHECK-NEXT: {
-// CHECK-NEXT: #pragma omp cancel parallel
+// CHECK-NEXT: #pragma omp cancel parallel if(argc)
 // CHECK-NEXT: }
 #pragma omp sections
 {
@@ -26,11 +26,11 @@ int main (int argc, char **argv) {
 // CHECK: }
 #pragma omp for
 for (int i = 0; i < argc; ++i) {
-#pragma omp cancel for
+#pragma omp cancel for if(cancel:argc)
 }
 // CHECK: #pragma omp for
 // CHECK-NEXT: for (int i = 0; i < argc; ++i) {
-// CHECK-NEXT: #pragma omp cancel for
+// CHECK-NEXT: #pragma omp cancel for if(cancel: argc)
 // CHECK-NEXT: }
 #pragma omp task
 {
diff --git a/test/OpenMP/cancel_codegen.cpp b/test/OpenMP/cancel_codegen.cpp
index 4bd85a22b1d9..e2dd3673caff 100644
--- a/test/OpenMP/cancel_codegen.cpp
+++ b/test/OpenMP/cancel_codegen.cpp
@@ -6,11 +6,12 @@
 #ifndef HEADER
 #define HEADER
 
+float flag;
 int main (int argc, char **argv) {
 // CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num(
 #pragma omp parallel
 {
-#pragma omp cancel parallel
+#pragma omp cancel parallel if(flag)
   argv[0][0] = argc;
 }
 // CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
@@ -50,9 +51,13 @@ int main (int argc, char **argv) {
 // CHECK: call void @__kmpc_for_static_fini(
 #pragma omp for
 for (int i = 0; i < argc; ++i) {
-#pragma omp cancel for
+#pragma omp cancel for if(cancel: flag)
 }
 // CHECK: call void @__kmpc_for_static_init_4(
+// CHECK: [[FLAG:%.+]] = load float, float* @{{.+}},
+// CHECK: [[BOOL:%.+]] = fcmp une float [[FLAG]], 0.000000e+00
+// CHECK: br i1 [[BOOL]], label %[[THEN:[^,]+]], label %[[ELSE:[^,]+]]
+// CHECK: [[THEN]]
 // CHECK: [[RES:%.+]] = call i32 @__kmpc_cancel(%ident_t* {{[^,]+}}, i32 [[GTID]], i32 2)
 // CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
 // CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
@@ -61,6 +66,8 @@ for (int i = 0; i < argc; ++i) {
 // CHECK: br label
 // CHECK: [[CONTINUE]]
 // CHECK: br label
+// CHECK: [[ELSE]]
+// CHECK: br label
 // CHECK: call void @__kmpc_for_static_fini(
 // CHECK: call void @__kmpc_barrier(%ident_t*
 #pragma omp task
@@ -69,16 +76,41 @@ for (int i = 0; i < argc; ++i) {
 }
 // CHECK: call i8* @__kmpc_omp_task_alloc(
 // CHECK: call i32 @__kmpc_omp_task(
+#pragma omp parallel sections
+{
+#pragma omp cancel sections
+}
+// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
+#pragma omp parallel sections
+{
+#pragma omp cancel sections
+#pragma omp section
+  {
+#pragma omp cancel sections
+  }
+}
+// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
+#pragma omp parallel for
+for (int i = 0; i < argc; ++i) {
+#pragma omp cancel for
+}
+// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
   return argc;
 }
 
 // CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}},
+// CHECK: [[FLAG:%.+]] = load float, float* @{{.+}},
+// CHECK: [[BOOL:%.+]] = fcmp une float [[FLAG]], 0.000000e+00
+// CHECK: br i1 [[BOOL]], label %[[THEN:[^,]+]], label %[[ELSE:[^,]+]]
+// CHECK: [[THEN]]
 // CHECK: [[RES:%.+]] = call i32 @__kmpc_cancel(%ident_t* {{[^,]+}}, i32 {{[^,]+}}, i32 1)
 // CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
 // CHECK: br i1 [[CMP]], label %[[EXIT:[^,]+]],
 // CHECK: [[EXIT]]
 // CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
 // CHECK: br label %[[RETURN:.+]]
+// CHECK: [[ELSE]]
+// CHECK: br label
 // CHECK: [[RETURN]]
 // CHECK: ret void
 
@@ -92,4 +124,44 @@ for (int i = 0; i < argc; ++i) {
 // CHECK: [[RETURN]]
 // CHECK: ret i32 0
 
+// CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}})
+// CHECK: call i32 @__kmpc_single(
+// CHECK-NOT: @__kmpc_cancel
+// CHECK: call void @__kmpc_end_single(
+// CHECK: ret void
+
+// CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}})
+// CHECK: call void @__kmpc_for_static_init_4(
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_cancel(%ident_t* {{[^,]+}}, i32 [[GTID:%.+]], i32 3)
+// CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
+// CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
+// CHECK: [[EXIT]]
+// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
+// CHECK: br label
+// CHECK: [[CONTINUE]]
+// CHECK: br label
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_cancel(%ident_t* {{[^,]+}}, i32 [[GTID]], i32 3)
+// CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
+// CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
+// CHECK: [[EXIT]]
+// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
+// CHECK: br label
+// CHECK: [[CONTINUE]]
+// CHECK: br label
+// CHECK: call void @__kmpc_for_static_fini(
+// CHECK: ret void
+
+// CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}},
+// CHECK: call void @__kmpc_for_static_init_4(
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_cancel(%ident_t* {{[^,]+}}, i32 [[GTID:%.+]], i32 2)
+// CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
+// CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
+// CHECK: [[EXIT]]
+// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
+// CHECK: br label
+// CHECK: [[CONTINUE]]
+// CHECK: br label
+// CHECK: call void @__kmpc_for_static_fini(
+// CHECK: ret void
+
 #endif
diff --git a/test/OpenMP/cancel_if_messages.cpp b/test/OpenMP/cancel_if_messages.cpp
new file mode 100644
index 000000000000..8d1afc9ecddf
--- /dev/null
+++ b/test/OpenMP/cancel_if_messages.cpp
@@ -0,0 +1,66 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp parallel
+  {
+    #pragma omp cancel parallel if // expected-error {{expected '(' after 'if'}}
+    #pragma omp cancel parallel if ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if () // expected-error {{expected expression}}
+    #pragma omp cancel parallel if (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if (argc)) // expected-warning {{extra tokens at the end of '#pragma omp cancel' are ignored}}
+    #pragma omp cancel parallel if (argc > 0 ? argv[1] : argv[2])
+    #pragma omp cancel parallel if (foobool(argc)), if (true) // expected-error {{directive '#pragma omp cancel' cannot contain more than one 'if' clause}}
+    #pragma omp cancel parallel if (S) // expected-error {{'S' does not refer to a value}}
+    #pragma omp cancel parallel if (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(argc)
+    #pragma omp cancel parallel if(cancel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel : argc)
+    #pragma omp cancel parallel if(cancel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp cancel'}}
+    #pragma omp cancel parallel if(cancel : argc) if (cancel:argc) // expected-error {{directive '#pragma omp cancel' cannot contain more than one 'if' clause with 'cancel' name modifier}}
+    #pragma omp cancel parallel if(cancel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+    foo();
+  }
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp parallel
+  {
+    #pragma omp cancel parallel if // expected-error {{expected '(' after 'if'}}
+    #pragma omp cancel parallel if ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if () // expected-error {{expected expression}}
+    #pragma omp cancel parallel if (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if (argc)) // expected-warning {{extra tokens at the end of '#pragma omp cancel' are ignored}}
+    #pragma omp cancel parallel if (argc > 0 ? argv[1] : argv[2])
+    #pragma omp cancel parallel if (foobool(argc)), if (true) // expected-error {{directive '#pragma omp cancel' cannot contain more than one 'if' clause}}
+    #pragma omp cancel parallel if (S1) // expected-error {{'S1' does not refer to a value}}
+    #pragma omp cancel parallel if (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+    #pragma omp cancel parallel if(cancel : argc)
+    #pragma omp cancel parallel if(cancel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp cancel'}}
+    #pragma omp cancel parallel if(cancel : argc) if (cancel:argc)  // expected-error {{directive '#pragma omp cancel' cannot contain more than one 'if' clause with 'cancel' name modifier}}
+    #pragma omp cancel parallel if(cancel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+    foo();
+  }
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/cancellation_point_codegen.cpp b/test/OpenMP/cancellation_point_codegen.cpp
index 453393082ce8..795f69ed24fa 100644
--- a/test/OpenMP/cancellation_point_codegen.cpp
+++ b/test/OpenMP/cancellation_point_codegen.cpp
@@ -11,12 +11,16 @@ int main (int argc, char **argv) {
 #pragma omp parallel
 {
 #pragma omp cancellation point parallel
+#pragma omp cancel parallel
   argv[0][0] = argc;
 }
 // CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
 #pragma omp sections
 {
+  {
 #pragma omp cancellation point sections
+#pragma omp cancel sections
+  }
 }
 // CHECK: call i32 @__kmpc_single(
 // CHECK-NOT: @__kmpc_cancellationpoint
@@ -28,6 +32,7 @@ int main (int argc, char **argv) {
 #pragma omp section
   {
 #pragma omp cancellation point sections
+#pragma omp cancel sections
   }
 }
 // CHECK: call void @__kmpc_for_static_init_4(
@@ -51,6 +56,7 @@ int main (int argc, char **argv) {
 #pragma omp for
 for (int i = 0; i < argc; ++i) {
 #pragma omp cancellation point for
+#pragma omp cancel for
 }
 // CHECK: call void @__kmpc_for_static_init_4(
 // CHECK: [[RES:%.+]] = call i32 @__kmpc_cancellationpoint(%ident_t* {{[^,]+}}, i32 [[GTID]], i32 2)
@@ -66,9 +72,36 @@ for (int i = 0; i < argc; ++i) {
 #pragma omp task
 {
 #pragma omp cancellation point taskgroup
+#pragma omp cancel taskgroup
 }
 // CHECK: call i8* @__kmpc_omp_task_alloc(
 // CHECK: call i32 @__kmpc_omp_task(
+#pragma omp parallel sections
+{
+  {
+#pragma omp cancellation point sections
+#pragma omp cancel sections
+  }
+}
+// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
+#pragma omp parallel sections
+{
+  {
+#pragma omp cancellation point sections
+#pragma omp cancel sections
+  }
+#pragma omp section
+  {
+#pragma omp cancellation point sections
+  }
+}
+// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
+#pragma omp parallel for
+for (int i = 0; i < argc; ++i) {
+#pragma omp cancellation point for
+#pragma omp cancel for
+}
+// CHECK: call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(
   return argc;
 }
 
@@ -92,4 +125,44 @@ for (int i = 0; i < argc; ++i) {
 // CHECK: [[RETURN]]
 // CHECK: ret i32 0
 
+// CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}})
+// CHECK: call i32 @__kmpc_single(
+// CHECK-NOT: @__kmpc_cancellationpoint
+// CHECK: call void @__kmpc_end_single(
+// CHECK: ret void
+
+// CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}})
+// CHECK: call void @__kmpc_for_static_init_4(
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_cancellationpoint(%ident_t* {{[^,]+}}, i32 [[GTID:%.+]], i32 3)
+// CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
+// CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
+// CHECK: [[EXIT]]
+// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
+// CHECK: br label
+// CHECK: [[CONTINUE]]
+// CHECK: br label
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_cancellationpoint(%ident_t* {{[^,]+}}, i32 [[GTID]], i32 3)
+// CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
+// CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
+// CHECK: [[EXIT]]
+// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
+// CHECK: br label
+// CHECK: [[CONTINUE]]
+// CHECK: br label
+// CHECK: call void @__kmpc_for_static_fini(
+// CHECK: ret void
+
+// CHECK: define internal void @{{[^(]+}}(i32* {{[^,]+}}, i32* {{[^,]+}},
+// CHECK: call void @__kmpc_for_static_init_4(
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_cancellationpoint(%ident_t* {{[^,]+}}, i32 [[GTID:%.+]], i32 2)
+// CHECK: [[CMP:%.+]] = icmp ne i32 [[RES]], 0
+// CHECK: br i1 [[CMP]], label %[[EXIT:[^,].+]], label %[[CONTINUE:.+]]
+// CHECK: [[EXIT]]
+// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t*
+// CHECK: br label
+// CHECK: [[CONTINUE]]
+// CHECK: br label
+// CHECK: call void @__kmpc_for_static_fini(
+// CHECK: ret void
+
 #endif
diff --git a/test/OpenMP/critical_ast_print.cpp b/test/OpenMP/critical_ast_print.cpp
index 87a1fe0bcee7..7189fddd7186 100644
--- a/test/OpenMP/critical_ast_print.cpp
+++ b/test/OpenMP/critical_ast_print.cpp
@@ -8,6 +8,26 @@
 
 void foo() {}
 
+// CHECK: template <typename T, int N> int tmain(T argc, char **argv)
+template <typename T, int N>
+int tmain (T argc, char **argv) {
+  T b = argc, c, d, e, f, g;
+  static int a;
+// CHECK: static int a;
+#pragma omp critical
+  a=2;
+// CHECK-NEXT: #pragma omp critical
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: ++a;
+  ++a;
+#pragma omp critical  (the_name) hint(N)
+  foo();
+// CHECK-NEXT: #pragma omp critical (the_name) hint(N)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: return N;
+  return N;
+}
+
 int main (int argc, char **argv) {
   int b = argc, c, d, e, f, g;
   static int a;
@@ -18,12 +38,12 @@ int main (int argc, char **argv) {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: ++a;
   ++a;
-#pragma omp critical  (the_name)
+#pragma omp critical  (the_name1) hint(23)
   foo();
-// CHECK-NEXT: #pragma omp critical (the_name)
+// CHECK-NEXT: #pragma omp critical (the_name1) hint(23)
 // CHECK-NEXT: foo();
-// CHECK-NEXT: return 0;
-  return 0;
+// CHECK-NEXT: return tmain<int, 4>(a, argv);
+  return tmain<int, 4>(a, argv);
 }
 
 #endif
diff --git a/test/OpenMP/critical_codegen.cpp b/test/OpenMP/critical_codegen.cpp
index 5cf3446d9f5f..e44e2202e9a2 100644
--- a/test/OpenMP/critical_codegen.cpp
+++ b/test/OpenMP/critical_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
 #ifndef HEADER
@@ -10,6 +10,7 @@
 // CHECK:       [[IDENT_T_TY:%.+]] = type { i32, i32, i32, i32, i8* }
 // CHECK:       [[UNNAMED_LOCK:@.+]] = common global [8 x i32] zeroinitializer
 // CHECK:       [[THE_NAME_LOCK:@.+]] = common global [8 x i32] zeroinitializer
+// CHECK:       [[THE_NAME_LOCK1:@.+]] = common global [8 x i32] zeroinitializer
 
 // CHECK:       define {{.*}}void [[FOO:@.+]]()
 
@@ -32,11 +33,41 @@ int main() {
 // CHECK:       call {{.*}}void @__kmpc_end_critical([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]], [8 x i32]* [[THE_NAME_LOCK]])
 #pragma omp critical(the_name)
   foo();
+// CHECK:       call {{.*}}void @__kmpc_critical_with_hint([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]], [8 x i32]* [[THE_NAME_LOCK1]], i{{64|32}} 23)
+// CHECK-NEXT:  invoke {{.*}}void [[FOO]]()
+// CHECK:       call {{.*}}void @__kmpc_end_critical([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]], [8 x i32]* [[THE_NAME_LOCK1]])
+#pragma omp critical(the_name1) hint(23)
+  foo();
+// CHECK:       call {{.*}}void @__kmpc_critical([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]], [8 x i32]* [[THE_NAME_LOCK]])
+// CHECK-NOT:   call {{.*}}void @__kmpc_end_critical(
+  if (a)
+#pragma omp critical(the_name)
+    while (1)
+      ;
+// CHECK:  call {{.*}}void [[FOO]]()
+  foo();
 // CHECK-NOT:   call void @__kmpc_critical
 // CHECK-NOT:   call void @__kmpc_end_critical
   return a;
 }
 
+struct S {
+  int a;
+};
+// CHECK-LABEL: critical_ref
+void critical_ref(S &s) {
+  // CHECK: [[S_ADDR:%.+]] = alloca %struct.S*,
+  // CHECK: [[S_REF:%.+]] = load %struct.S*, %struct.S** [[S_ADDR]],
+  // CHECK: [[S_A_REF:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_REF]], i32 0, i32 0
+  ++s.a;
+  // CHECK: call void @__kmpc_critical(
+#pragma omp critical
+  // CHECK: [[S_REF:%.+]] = load %struct.S*, %struct.S** [[S_ADDR]],
+  // CHECK: [[S_A_REF:%.+]] = getelementptr inbounds %struct.S, %struct.S* [[S_REF]], i32 0, i32 0
+  ++s.a;
+  // CHECK: call void @__kmpc_end_critical(
+}
+
 // CHECK-LABEL:      parallel_critical
 // TERM_DEBUG-LABEL: parallel_critical
 void parallel_critical() {
diff --git a/test/OpenMP/critical_messages.cpp b/test/OpenMP/critical_messages.cpp
index e3f2ad70fe98..bb5ec00fdbf1 100644
--- a/test/OpenMP/critical_messages.cpp
+++ b/test/OpenMP/critical_messages.cpp
@@ -2,20 +2,21 @@
 
 int foo();
 
-int main() {
+template<typename T, int N>
+int tmain(int argc, char **argv) { // expected-note {{declared here}}
   #pragma omp critical
   ;
   #pragma omp critical untied // expected-error {{unexpected OpenMP clause 'untied' in directive '#pragma omp critical'}}
   #pragma omp critical unknown // expected-warning {{extra tokens at the end of '#pragma omp critical' are ignored}}
   #pragma omp critical ( // expected-error {{expected identifier}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp critical ( + // expected-error {{expected identifier}} expected-error {{expected ')'}} expected-note {{to match this '('}}
-  #pragma omp critical (name // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp critical (name2 // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp critical (name1)
   foo();
   {
     #pragma omp critical
   } // expected-error {{expected statement}}
-  #pragma omp critical (name) // expected-note {{previous 'critical' region starts here}}
+  #pragma omp critical (name2) // expected-note {{previous 'critical' region starts here}}
   #pragma omp critical
   for (int i = 0; i < 10; ++i) {
     foo();
@@ -23,11 +24,11 @@ int main() {
     #pragma omp for
     for (int j = 0; j < 10; j++) {
       foo();
-      #pragma omp critical(name) // expected-error {{cannot nest 'critical' regions having the same name 'name'}}
+      #pragma omp critical(name2) // expected-error {{cannot nest 'critical' regions having the same name 'name2'}}
       foo();
     }
   }
-  #pragma omp critical (name)
+  #pragma omp critical (name2)
   #pragma omp critical
   for (int i = 0; i < 10; ++i) {
     foo();
@@ -38,7 +39,7 @@ int main() {
       foo();
     }
   }
-  #pragma omp critical (name)
+  #pragma omp critical (name2)
   #pragma omp critical
   for (int i = 0; i < 10; ++i) {
     foo();
@@ -50,9 +51,86 @@ int main() {
     }
   }
 
+  #pragma omp critical (name2) hint // expected-error {{expected '(' after 'hint'}}
+  foo();
+  #pragma omp critical (name2) hint( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+  #pragma omp critical (name2) hint(+ // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+  #pragma omp critical (name2) hint(argc) // expected-error {{expression is not an integral constant expression}} expected-note {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  foo();
+  #pragma omp critical (name) hint(N) // expected-error {{argument to 'hint' clause must be a strictly positive integer value}} expected-error {{constructs with the same name must have a 'hint' clause with the same value}} expected-note {{'hint' clause with value '4'}}
+  foo();
+  #pragma omp critical hint(N) // expected-error {{the name of the construct must be specified in presence of 'hint' clause}}
+  foo();
   return 0;
 }
 
+int main(int argc, char **argv) { // expected-note {{declared here}}
+  #pragma omp critical
+  ;
+  #pragma omp critical untied // expected-error {{unexpected OpenMP clause 'untied' in directive '#pragma omp critical'}}
+  #pragma omp critical unknown // expected-warning {{extra tokens at the end of '#pragma omp critical' are ignored}}
+  #pragma omp critical ( // expected-error {{expected identifier}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp critical ( + // expected-error {{expected identifier}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp critical (name2 // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp critical (name1)
+  foo();
+  {
+    #pragma omp critical
+  } // expected-error {{expected statement}}
+  #pragma omp critical (name2) // expected-note {{previous 'critical' region starts here}}
+  #pragma omp critical
+  for (int i = 0; i < 10; ++i) {
+    foo();
+    #pragma omp parallel
+    #pragma omp for
+    for (int j = 0; j < 10; j++) {
+      foo();
+      #pragma omp critical(name2) // expected-error {{cannot nest 'critical' regions having the same name 'name2'}}
+      foo();
+    }
+  }
+  #pragma omp critical (name2)
+  #pragma omp critical
+  for (int i = 0; i < 10; ++i) {
+    foo();
+    #pragma omp parallel
+    #pragma omp for
+    for (int j = 0; j < 10; j++) {
+      #pragma omp critical
+      foo();
+    }
+  }
+  #pragma omp critical (name2)
+  #pragma omp critical
+  for (int i = 0; i < 10; ++i) {
+    foo();
+    #pragma omp parallel
+    #pragma omp for
+    for (int j = 0; j < 10; j++) {
+      #pragma omp critical (nam)
+      foo();
+    }
+  }
+
+  #pragma omp critical (name2) hint // expected-error {{expected '(' after 'hint'}}
+  foo();
+  #pragma omp critical (name2) hint( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+  #pragma omp critical (name2) hint(+ // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+  #pragma omp critical (name2) hint(argc) // expected-error {{expression is not an integral constant expression}} expected-note {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  foo();
+  #pragma omp critical (name) hint(23) // expected-note {{previous 'hint' clause with value '23'}}
+  foo();
+  #pragma omp critical hint(-5) // expected-error {{argument to 'hint' clause must be a strictly positive integer value}}
+  foo();
+  #pragma omp critical hint(1) // expected-error {{the name of the construct must be specified in presence of 'hint' clause}}
+  foo();
+  return tmain<int, 4>(argc, argv) + tmain<float, -5>(argc, argv); // expected-note {{in instantiation of function template specialization 'tmain<float, -5>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<int, 4>' requested here}}
+}
+
 int foo() {
   L1:
     foo();
@@ -69,4 +147,4 @@ int foo() {
   }
 
   return 0;
-}
+ }
diff --git a/test/OpenMP/distribute_ast_print.cpp b/test/OpenMP/distribute_ast_print.cpp
new file mode 100644
index 000000000000..c3a175a5e5e8
--- /dev/null
+++ b/test/OpenMP/distribute_ast_print.cpp
@@ -0,0 +1,74 @@
+// RUN: %clang_cc1 -verify -fopenmp -ast-print %s | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -std=c++11 -include-pch %t -fsyntax-only -verify %s -ast-print | FileCheck %s
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void foo() {}
+
+template <class T, int N>
+T tmain(T argc) {
+  T b = argc, c, d, e, f, g;
+  static T a;
+// CHECK: static T a;
+#pragma omp distribute
+// CHECK-NEXT: #pragma omp distribute
+  for (int i=0; i < 2; ++i)a=2;
+// CHECK-NEXT: for (int i = 0; i < 2; ++i)
+// CHECK-NEXT: a = 2;
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute private(argc, b), firstprivate(c, d), collapse(2)
+  for (int i = 0; i < 10; ++i)
+  for (int j = 0; j < 10; ++j)foo();
+// CHECK-NEXT: #pragma omp target
+// CHECK-NEXT: #pragma omp teams
+// CHECK-NEXT: #pragma omp distribute private(argc,b) firstprivate(c,d) collapse(2)
+// CHECK-NEXT: for (int i = 0; i < 10; ++i)
+// CHECK-NEXT: for (int j = 0; j < 10; ++j)
+// CHECK-NEXT: foo();
+  for (int i = 0; i < 10; ++i)foo();
+// CHECK-NEXT: for (int i = 0; i < 10; ++i)
+// CHECK-NEXT: foo();
+#pragma omp distribute
+// CHECK: #pragma omp distribute
+  for (int i = 0; i < 10; ++i)foo();
+// CHECK-NEXT: for (int i = 0; i < 10; ++i)
+// CHECK-NEXT: foo();  
+  return T();
+}
+
+int main (int argc, char **argv) {
+  int b = argc, c, d, e, f, g;
+  static int a;
+// CHECK: static int a;
+#pragma omp distribute
+// CHECK-NEXT: #pragma omp distribute
+  for (int i=0; i < 2; ++i)a=2;
+// CHECK-NEXT: for (int i = 0; i < 2; ++i)
+// CHECK-NEXT: a = 2;
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute private(argc,b),firstprivate(argv, c), collapse(2)
+  for (int i = 0; i < 10; ++i)
+  for (int j = 0; j < 10; ++j)foo();
+// CHECK-NEXT: #pragma omp target
+// CHECK-NEXT: #pragma omp teams
+// CHECK-NEXT: #pragma omp distribute private(argc,b) firstprivate(argv,c) collapse(2)
+// CHECK-NEXT: for (int i = 0; i < 10; ++i)
+// CHECK-NEXT: for (int j = 0; j < 10; ++j)
+// CHECK-NEXT: foo();
+  for (int i = 0; i < 10; ++i)foo();
+// CHECK-NEXT: for (int i = 0; i < 10; ++i)
+// CHECK-NEXT: foo();
+#pragma omp distribute
+// CHECK: #pragma omp distribute
+  for (int i = 0; i < 10; ++i)foo();
+// CHECK-NEXT: for (int i = 0; i < 10; ++i)
+// CHECK-NEXT: foo();
+  return (0);
+}
+
+#endif
diff --git a/test/OpenMP/distribute_collapse_messages.cpp b/test/OpenMP/distribute_collapse_messages.cpp
new file mode 100644
index 000000000000..8818d659d441
--- /dev/null
+++ b/test/OpenMP/distribute_collapse_messages.cpp
@@ -0,0 +1,83 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp distribute collapse // expected-error {{expected '(' after 'collapse'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp distribute collapse (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  #pragma omp distribute collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp distribute' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp distribute', but found only 1}}
+  // expected-error@+3 2 {{directive '#pragma omp distribute' cannot contain more than one 'collapse' clause}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp distribute collapse (foobool(argc)), collapse (true), collapse (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp distribute collapse (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse (1)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp distribute collapse (2) // expected-note {{as specified in 'collapse' clause}}
+  foo(); // expected-error {{expected 2 for loops after '#pragma omp distribute'}}
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp distribute collapse // expected-error {{expected '(' after 'collapse'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp distribute collapse ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp distribute collapse () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp distribute collapse (4 // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-note {{as specified in 'collapse' clause}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4]; // expected-error {{expected 4 for loops after '#pragma omp distribute', but found only 1}}
+  #pragma omp distribute collapse (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp distribute' are ignored}}  expected-note {{as specified in 'collapse' clause}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4]; // expected-error {{expected 4 for loops after '#pragma omp distribute', but found only 1}}
+  #pragma omp distribute collapse (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp distribute' cannot contain more than one 'collapse' clause}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  #pragma omp distribute collapse (foobool(argc)), collapse (true), collapse (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp distribute collapse (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp distribute collapse (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp distribute' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp distribute collapse(collapse(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  #pragma omp distribute collapse (2) // expected-note {{as specified in 'collapse' clause}}
+  foo(); // expected-error {{expected 2 for loops after '#pragma omp distribute'}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
+  return tmain<int, char, 1, 0>(argc, argv);
+}
+
diff --git a/test/OpenMP/distribute_firstprivate_messages.cpp b/test/OpenMP/distribute_firstprivate_messages.cpp
new file mode 100644
index 000000000000..5d371abbfa8e
--- /dev/null
+++ b/test/OpenMP/distribute_firstprivate_messages.cpp
@@ -0,0 +1,157 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}} expected-note{{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+  
+public:
+  S2() : a(0) {}
+  S2(const S2 &s2) : a(s2.a) {}
+  static float S2s;
+  static const float S2sc;
+};
+const float S2::S2sc = 0;
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+  S3 &operator=(const S3 &s3);
+  
+public:
+  S3() : a(0) {} // expected-note {{candidate constructor not viable: requires 0 arguments, but 1 was provided}}
+  S3(S3 &s3) : a(s3.a) {} // expected-note {{candidate constructor not viable: 1st argument ('const S3') would lose const qualifier}}
+};
+const S3 c;
+const S3 ca[5];
+extern const int f;
+class S4 {
+  int a;
+  S4();
+  S4(const S4 &s4);
+public:
+  S4(int v):a(v) { }
+};
+class S5 {
+  int a;
+  S5():a(0) {}
+  S5(const S5 &s5):a(s5.a) { }
+public:
+  S5(int v):a(v) { }
+};
+class S6 {
+  int a;
+public:
+  S6() : a(0) { }
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note {{defined as threadprivate or thread local}}
+
+int main(int argc, char **argv) {
+  const int d = 5;
+  const int da[5] = { 0 };
+  S4 e(4);
+  S5 g(5);
+  S6 p;
+  int i;
+  int &j = i;
+  #pragma omp distribute firstprivate // expected-error {{expected '(' after 'firstprivate'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp distribute firstprivate ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp distribute firstprivate () // expected-error {{expected expression}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (argc)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (S1) // expected-error {{'S1' does not refer to a value}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (a, b, c, d, f) // expected-error {{firstprivate variable with incomplete type 'S1'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate (argv[1]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(ba)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(ca) // expected-error {{no matching constructor for initialization of 'S3'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(da)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(S2::S2s)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(S2::S2sc)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(h) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute private(i), firstprivate(i) // expected-error {{private variable cannot be firstprivate}} expected-note{{defined as private}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams shared(i)
+  #pragma omp distribute firstprivate(i)
+  for (j = 0; j < argc; ++j) foo();
+  #pragma omp target
+  #pragma omp teams shared(i)
+  #pragma omp distribute firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) foo(); // expected-error {{loop iteration variable in the associated loop of 'omp distribute' directive may not be firstprivate, predetermined as private}}
+  #pragma omp target
+  #pragma omp teams private(argc) // expected-note {{defined as private}}
+  #pragma omp distribute firstprivate(argc) // expected-error {{private variable in '#pragma omp teams' cannot be firstprivate in '#pragma omp distribute'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams reduction(+:argc) // expected-note {{defined as reduction}}
+  #pragma omp distribute firstprivate(argc) // expected-error {{reduction variable in '#pragma omp teams' cannot be firstprivate in '#pragma omp distribute'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(j)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute lastprivate(argc), firstprivate(argc) // expected-error {{lastprivate variable cannot be firstprivate in '#pragma omp distribute'}} expected-note{{defined as lastprivate}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp target
+  #pragma omp teams
+#pragma omp distribute firstprivate(argc), lastprivate(argc)  // expected-error {{lastprivate variable cannot be firstprivate in '#pragma omp distribute'}} expected-note{{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) foo();
+  return 0;
+}
diff --git a/test/OpenMP/distribute_private_messages.cpp b/test/OpenMP/distribute_private_messages.cpp
new file mode 100644
index 000000000000..94ba4659c684
--- /dev/null
+++ b/test/OpenMP/distribute_private_messages.cpp
@@ -0,0 +1,132 @@
+// RUN: %clang_cc1 -triple x86_64-apple-macos10.7.0 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}} expected-note {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+public:
+  S2():a(0) { }
+  static float S2s; // expected-note {{predetermined as shared}}
+};
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+public:
+  S3():a(0) { }
+};
+const S3 c; // expected-note {{predetermined as shared}}
+const S3 ca[5]; // expected-note {{predetermined as shared}}
+extern const int f;  // expected-note {{predetermined as shared}}
+class S4 {
+  int a;
+  S4(); // expected-note {{implicitly declared private here}}
+public:
+  S4(int v):a(v) { }
+};
+class S5 { 
+  int a;
+  S5():a(0) {} // expected-note {{implicitly declared private here}}
+public:
+  S5(int v):a(v) { }
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note {{defined as threadprivate or thread local}}
+
+
+int main(int argc, char **argv) {
+  const int d = 5;  // expected-note {{predetermined as shared}}
+  const int da[5] = { 0 }; // expected-note {{predetermined as shared}}
+  S4 e(4);
+  S5 g(5);
+  int i;
+  int &j = i;
+ #pragma omp distribute private // expected-error {{expected '(' after 'private'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private () // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (argc)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (a, b, c, d, f) // expected-error {{private variable with incomplete type 'S1'}} expected-error 3 {{shared variable cannot be private}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private (argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(ba)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(ca) // expected-error {{shared variable cannot be private}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(da) // expected-error {{shared variable cannot be private}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(S2::S2s) // expected-error {{shared variable cannot be private}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(h) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute shared(i) // expected-error {{unexpected OpenMP clause 'shared' in directive '#pragma omp distribute'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp target
+  #pragma omp teams
+  {
+    int i; // expected-note {{predetermined as private}}
+    #pragma omp distribute firstprivate(i), private(i) // expected-error {{private variable in '#pragma omp teams' cannot be firstprivate in '#pragma omp distribute'}}
+    for (int k = 0; k < argc; ++k) ++k;
+  }
+  #pragma omp target
+  #pragma omp teams private(i)
+  #pragma omp distribute private(j)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp parallel private(i)
+  #pragma omp target
+  #pragma omp teams firstprivate(i)
+  #pragma omp parallel private(i)
+  #pragma omp target
+  #pragma omp teams reduction(+:i)
+  #pragma omp distribute private(i)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp distribute private(i)
+  for (int k = 0; k < 10; ++k) {
+    #pragma omp target
+    #pragma omp teams private(i)
+    #pragma omp distribute private(i)
+    for (int x = 0; x < 10; ++x) foo();
+  }
+  #pragma omp target
+  #pragma omp teams
+  #pragma omp distribute firstprivate(i)
+  for (int k = 0; k < 10; ++k) {
+    #pragma omp target
+    #pragma omp teams firstprivate(i)
+    #pragma omp distribute private(i)
+    for (int x = 0; x < 10; ++x) foo();
+  }
+  #pragma omp target
+  #pragma omp teams reduction(+:i)
+  #pragma omp distribute
+  for (int k = 0; k < 10; ++k) {
+    #pragma omp target
+    #pragma omp teams reduction(+:i)
+    #pragma omp distribute private(i)
+    for (int x = 0; x < 10; ++x) foo();
+  }
+
+  return 0;
+}
diff --git a/test/OpenMP/driver.c b/test/OpenMP/driver.c
new file mode 100644
index 000000000000..f84541bef8ba
--- /dev/null
+++ b/test/OpenMP/driver.c
@@ -0,0 +1,10 @@
+// Test that by default -fnoopenmp-use-tls is passed to frontend.
+//
+// RUN: %clang %s -### -o %t.o 2>&1 -fopenmp=libomp | FileCheck --check-prefix=CHECK-DEFAULT %s
+// CHECK-DEFAULT: -cc1
+// CHECK-DEFAULT-NOT: -fnoopenmp-use-tls
+//
+// RUN: %clang %s -### -o %t.o 2>&1 -fopenmp=libomp -fnoopenmp-use-tls | FileCheck --check-prefix=CHECK-NO-TLS %s
+// CHECK-NO-TLS: -cc1
+// CHECK-NO-TLS-SAME: -fnoopenmp-use-tls
+//
diff --git a/test/OpenMP/flush_codegen.cpp b/test/OpenMP/flush_codegen.cpp
index cd411cea4573..d0ab9f1525a8 100644
--- a/test/OpenMP/flush_codegen.cpp
+++ b/test/OpenMP/flush_codegen.cpp
@@ -1,6 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
 #ifndef HEADER
diff --git a/test/OpenMP/for_ast_print.cpp b/test/OpenMP/for_ast_print.cpp
index c482ec56b9ca..8fd82e7f028b 100644
--- a/test/OpenMP/for_ast_print.cpp
+++ b/test/OpenMP/for_ast_print.cpp
@@ -13,27 +13,37 @@ T tmain(T argc) {
   T b = argc, c, d, e, f, g;
   static T a;
 // CHECK: static T a;
-#pragma omp for schedule(dynamic)
-  // CHECK-NEXT: #pragma omp for schedule(dynamic)
+#pragma omp for schedule(dynamic) linear(a)
+  // CHECK-NEXT: #pragma omp for schedule(dynamic) linear(a)
   for (int i = 0; i < 2; ++i)
     a = 2;
 // CHECK-NEXT: for (int i = 0; i < 2; ++i)
 // CHECK-NEXT: a = 2;
 #pragma omp parallel
-#pragma omp for private(argc, b), firstprivate(c, d), lastprivate(d, f) collapse(N) schedule(static, N) ordered nowait
-  for (int i = 0; i < 10; ++i)
-    for (int j = 0; j < 10; ++j)
-      for (int j = 0; j < 10; ++j)
-        for (int j = 0; j < 10; ++j)
-          for (int j = 0; j < 10; ++j)
+#pragma omp for private(argc, b), firstprivate(c, d), lastprivate(d, f) collapse(N) schedule(static, N) ordered(N) nowait
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
             foo();
   // CHECK-NEXT: #pragma omp parallel
-  // CHECK-NEXT: #pragma omp for private(argc,b) firstprivate(c,d) lastprivate(d,f) collapse(N) schedule(static, N) ordered nowait
-  // CHECK-NEXT: for (int i = 0; i < 10; ++i)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
+  // CHECK-NEXT: #pragma omp for private(argc,b) firstprivate(c,d) lastprivate(d,f) collapse(N) schedule(static, N) ordered(N) nowait
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
   // CHECK-NEXT: foo();
   return T();
 }
@@ -49,12 +59,12 @@ int main(int argc, char **argv) {
 // CHECK-NEXT: for (int i = 0; i < 2; ++i)
 // CHECK-NEXT: a = 2;
 #pragma omp parallel
-#pragma omp for private(argc, b), firstprivate(argv, c), lastprivate(d, f) collapse(2) schedule(auto) ordered nowait
+#pragma omp for private(argc, b), firstprivate(argv, c), lastprivate(d, f) collapse(2) schedule(auto) ordered nowait linear(g:-1)
   for (int i = 0; i < 10; ++i)
     for (int j = 0; j < 10; ++j)
       foo();
   // CHECK-NEXT: #pragma omp parallel
-  // CHECK-NEXT: #pragma omp for private(argc,b) firstprivate(argv,c) lastprivate(d,f) collapse(2) schedule(auto) ordered nowait
+  // CHECK-NEXT: #pragma omp for private(argc,b) firstprivate(argv,c) lastprivate(d,f) collapse(2) schedule(auto) ordered nowait linear(g: -1)
   // CHECK-NEXT: for (int i = 0; i < 10; ++i)
   // CHECK-NEXT: for (int j = 0; j < 10; ++j)
   // CHECK-NEXT: foo();
diff --git a/test/OpenMP/for_codegen.cpp b/test/OpenMP/for_codegen.cpp
index 9db157011075..98761f56c762 100644
--- a/test/OpenMP/for_codegen.cpp
+++ b/test/OpenMP/for_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 //
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
@@ -357,7 +357,7 @@ void parallel_for(float *a) {
   // TERM_DEBUG:     unwind label %[[TERM_LPAD:.+]],
   // TERM_DEBUG-NOT: __kmpc_global_thread_num
   // TERM_DEBUG:     call void @__kmpc_for_static_fini({{.+}}), !dbg [[DBG_LOC_END:![0-9]+]]
-  // TERM_DEBUG:     call {{.+}} @__kmpc_cancel_barrier({{.+}}), !dbg [[DBG_LOC_CANCEL:![0-9]+]]
+  // TERM_DEBUG:     call {{.+}} @__kmpc_barrier({{.+}}), !dbg [[DBG_LOC_CANCEL:![0-9]+]]
   // TERM_DEBUG:     [[TERM_LPAD]]
   // TERM_DEBUG:     call void @__clang_call_terminate
   // TERM_DEBUG:     unreachable
@@ -404,6 +404,25 @@ void for_with_global_lcv() {
     k = i;
     k = j;
   }
+  char &cnt = i;
+#pragma omp for
+  for (cnt = 0; cnt < 2; ++cnt)
+    k = cnt;
+}
+
+// CHECK-LABEL: for_with_references
+void for_with_references() {
+// CHECK: [[I:%.+]] = alloca i8,
+// CHECK: [[CNT:%.+]] = alloca i8*,
+// CHECK: [[CNT_PRIV:%.+]] = alloca i8,
+// CHECK: call void @__kmpc_for_static_init_4(
+// CHECK-NOT: load i8, i8* [[CNT]],
+// CHECK: call void @__kmpc_for_static_fini(
+  char i = 0;
+  char &cnt = i;
+#pragma omp for
+  for (cnt = 0; cnt < 2; ++cnt)
+    k = cnt;
 }
 
 struct Bool {
diff --git a/test/OpenMP/for_collapse_messages.cpp b/test/OpenMP/for_collapse_messages.cpp
index 0a7443319e8c..d40c305c6271 100644
--- a/test/OpenMP/for_collapse_messages.cpp
+++ b/test/OpenMP/for_collapse_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp for collapse (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp for collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp for' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp for collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp for', but found only 1}}
   // expected-error@+3 2 {{directive '#pragma omp for' cannot contain more than one 'collapse' clause}}
-  // expected-error@+2 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp for collapse (foobool(argc)), collapse (true), collapse (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for collapse (1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for collapse (N) // expected-error {{argument to 'collapse' clause must be a positive integer value}}
+  #pragma omp for collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for collapse (2) // expected-note {{as specified in 'collapse' clause}}
   foo(); // expected-error {{expected 2 for loops after '#pragma omp for'}}
@@ -63,7 +63,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp for' cannot contain more than one 'collapse' clause}}
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp for collapse (foobool(argc)), collapse (true), collapse (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for collapse (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/for_firstprivate_codegen.cpp b/test/OpenMP/for_firstprivate_codegen.cpp
index 0d5f83ed4ec3..01a93559451a 100644
--- a/test/OpenMP/for_firstprivate_codegen.cpp
+++ b/test/OpenMP/for_firstprivate_codegen.cpp
@@ -16,6 +16,7 @@ struct St {
 };
 
 volatile int g = 1212;
+volatile int &g1 = g;
 
 template <class T>
 struct S {
@@ -30,7 +31,6 @@ struct S {
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
 
 template <typename T>
 T tmain() {
@@ -38,7 +38,7 @@ T tmain() {
   T t_var = T();
   T vec[] = {1, 2};
   S<T> s_arr[] = {1, 2};
-  S<T> var(3);
+  S<T> &var = test;
 #pragma omp parallel
 #pragma omp for firstprivate(t_var, vec, s_arr, var)
   for (int i = 0; i < 2; ++i) {
@@ -58,22 +58,24 @@ int vec[] = {1, 2};
 S<float> s_arr[] = {1, 2};
 // CHECK-DAG: [[VAR:@.+]] = global [[S_FLOAT_TY]] zeroinitializer,
 S<float> var(3);
+// CHECK: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
 // CHECK: ([[S_FLOAT_TY]]*)* [[S_FLOAT_TY_DESTR:@[^ ]+]] {{[^,]+}}, {{.+}}([[S_FLOAT_TY]]* [[TEST]]
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for firstprivate(g)
+#pragma omp for firstprivate(g, g1, sivar)
   for (int i = 0; i < 2; ++i) {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR_REF:%.+]])
     // Skip temp vars for loop
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
@@ -81,25 +83,47 @@ int main() {
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[SIVAR2_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+
+    // LAMBDA:  store i{{[0-9]+}}* [[SIVAR_REF]], i{{[0-9]+}}** %{{.+}},
+    // LAMBDA:  [[SIVAR2_PRIVATE_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
+
+
     // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]]
     // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // LAMBDA: [[SIVAR2_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR_REF]]
+    // LAMBDA: store i{{[0-9]+}} [[SIVAR2_VAL]], i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR]]
+
     // LAMBDA: call void @__kmpc_barrier(
     g = 1;
+    g1 = 1;
+    sivar = 2;
     // LAMBDA: call void @__kmpc_for_static_init_4(
+
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call void @__kmpc_for_static_fini(
-    // LAMBDA: call i32 @__kmpc_cancel_barrier(
+    // LAMBDA: call void @__kmpc_barrier(
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      g1 = 2;
+      sivar = 4;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
+
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -110,11 +134,11 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for firstprivate(g)
+#pragma omp for firstprivate(g, g1, sivar)
   for (int i = 0; i < 2; ++i) {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR_REF:%.+]])
     // Skip temp vars for loop
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
@@ -122,34 +146,56 @@ int main() {
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[SIVAR2_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+
+    // BLOCKS: store i{{[0-9]+}}* [[SIVAR_REF]], i{{[0-9]+}}** %{{.+}},
+    // BLOCKS: [[SIVAR_REF_ADDRR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
+
     // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]]
     // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+
+    // BLOCKS: [[SIVAR2_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF_ADDRR]]
+    // BLOCKS: store i{{[0-9]+}} {{.+}}, i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR]]
+
     // BLOCKS: call void @__kmpc_barrier(
     g = 1;
+    g1 =1;
+    sivar = 2;
     // BLOCKS: call void @__kmpc_for_static_init_4(
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR]],
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR2_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
     // BLOCKS: call void @__kmpc_for_static_fini(
-    // BLOCKS: call i32 @__kmpc_cancel_barrier(
+    // BLOCKS: call void @__kmpc_barrier(
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      g1 = 2;
+      sivar = 4;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
   }();
   return 0;
 #else
-#pragma omp for firstprivate(t_var, vec, s_arr, var)
+#pragma omp for firstprivate(t_var, vec, s_arr, var, sivar)
   for (int i = 0; i < 2; ++i) {
     vec[i] = t_var;
     s_arr[i] = var;
+    sivar += i;
   }
   return tmain<int>();
 #endif
@@ -167,6 +213,7 @@ int main() {
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num(
 
 // firstprivate t_var(t_var)
@@ -194,6 +241,10 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 
+// firstprivate (sivar)
+// CHECK: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR]]
+// CHECK: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIV]]
+
 // Synchronization for initialization.
 // CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 
@@ -212,11 +263,11 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
 // Skip temp vars for loop
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -229,22 +280,20 @@ int main() {
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+
 // firstprivate t_var(t_var)
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]],
 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]],
 
 // firstprivate vec(vec)
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
 
 // firstprivate s_arr(s_arr)
-// CHECK: [[S_ARR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF]],
 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2
 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]]
@@ -256,8 +305,7 @@ int main() {
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
 
 // firstprivate var(var)
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]],
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]])
 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
@@ -275,7 +323,7 @@ int main() {
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]*
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 #endif
 
diff --git a/test/OpenMP/for_firstprivate_messages.cpp b/test/OpenMP/for_firstprivate_messages.cpp
index 2ec81104e771..1933de25850e 100644
--- a/test/OpenMP/for_firstprivate_messages.cpp
+++ b/test/OpenMP/for_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (int k = 0; k < argc; ++k)
@@ -115,14 +115,10 @@ int foomain(int argc, char **argv) {
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel
-#pragma omp for linear(i) // expected-error {{unexpected OpenMP clause 'linear' in directive '#pragma omp for'}}
-  for (int k = 0; k < argc; ++k)
-    ++k;
-#pragma omp parallel
   {
     int v = 0;
     int i;                      // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for' directive into a parallel or another task region?}}
-#pragma omp for firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp for firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     for (int k = 0; k < argc; ++k) {
       i = k;
       v += i;
@@ -130,7 +126,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp for firstprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel
@@ -175,7 +171,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (i = 0; i < argc; ++i)
@@ -274,7 +270,7 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp for firstprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
@@ -289,7 +285,7 @@ int main(int argc, char **argv) {
   {
     int v = 0;
     int i;                      // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for' directive into a parallel or another task region?}}
-#pragma omp for firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp for firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     for (int k = 0; k < argc; ++k) {
       i = k;
       v += i;
@@ -307,6 +303,10 @@ int main(int argc, char **argv) {
 #pragma omp for firstprivate(B::x) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
   for (i = 0; i < argc; ++i)
     foo();
+  static int si;
+#pragma omp for firstprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/for_lastprivate_codegen.cpp b/test/OpenMP/for_lastprivate_codegen.cpp
index 880637ea8b94..36abd8df07cb 100644
--- a/test/OpenMP/for_lastprivate_codegen.cpp
+++ b/test/OpenMP/for_lastprivate_codegen.cpp
@@ -18,14 +18,13 @@ struct S {
   ~S() {}
 };
 
-volatile int g = 1212;
+volatile int g __attribute__((aligned(128)))= 1212;
+volatile int &g1 = g;
 float f;
 char cnt;
 
 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[CAP_MAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]* }
 // CHECK: [[S_INT_TY:%.+]] = type { i32 }
-// CHECK: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-DAG: [[X:@.+]] = global double 0.0
 // CHECK-DAG: [[F:@.+]] = global float 0.0
@@ -33,10 +32,10 @@ char cnt;
 template <typename T>
 T tmain() {
   S<T> test;
-  T t_var = T();
-  T vec[] = {1, 2};
-  S<T> s_arr[] = {1, 2};
-  S<T> var(3);
+  T t_var __attribute__((aligned(128))) = T();
+  T vec[] __attribute__((aligned(128))) = {1, 2};
+  S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
+  S<T> &var __attribute__((aligned(128))) = test;
 #pragma omp parallel
 #pragma omp for lastprivate(t_var, vec, s_arr, var)
   for (int i = 0; i < 2; ++i) {
@@ -54,33 +53,44 @@ using A::x;
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
+  // LAMBDA: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
   // LAMBDA-LABEL: @main
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* %{{.+}})
 #pragma omp parallel
-#pragma omp for lastprivate(g)
+#pragma omp for lastprivate(g, g1, sivar)
   for (int i = 0; i < 2; ++i) {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]])
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
-    // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
+    // LAMBDA: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
+
     // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
     // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+
     // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
     g = 1;
+    g1 = 1;
+    sivar = 2;
     // Check for final copying of private values back to original vars.
     // LAMBDA: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
     // LAMBDA: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
@@ -91,17 +101,26 @@ int main() {
     // original g=private_g;
     // LAMBDA: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
     // LAMBDA: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]],
+
+    // original sivar=private_sivar;
+    // LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* %{{.+}},
     // LAMBDA: br label %[[LAST_DONE]]
     // LAMBDA: [[LAST_DONE]]
-    // LAMBDA: call i32 @__kmpc_cancel_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
+    // LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      g1 = 2;
+      sivar = 4;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -112,18 +131,21 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for lastprivate(g)
+#pragma omp for lastprivate(g, g1, sivar)
   for (int i = 0; i < 2; ++i) {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]])
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
-    // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
+    // BLOCKS: [[G1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 4
+    // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: store i{{[0-9]+}}* [[SIVAR]], i{{[0-9]+}}** [[SIVAR_ADDR:%.+]],
+    // BLOCKS: {{.+}} = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_ADDR]]
     // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
     // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
     // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
@@ -134,6 +156,8 @@ int main() {
     // BLOCKS: call void {{%.+}}(i8
     // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
     g = 1;
+    g1 = 1;
+    sivar = 2;
     // Check for final copying of private values back to original vars.
     // BLOCKS: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
     // BLOCKS: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
@@ -144,16 +168,24 @@ int main() {
     // original g=private_g;
     // BLOCKS: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
     // BLOCKS: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]],
+    // BLOCKS: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* %{{.+}},
     // BLOCKS: br label %[[LAST_DONE]]
     // BLOCKS: [[LAST_DONE]]
-    // BLOCKS: call i32 @__kmpc_cancel_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
+    // BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
     g = 1;
+    g1 = 1;
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      g1 = 1;
+      sivar = 4;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -166,10 +198,11 @@ int main() {
   S<float> s_arr[] = {1, 2};
   S<float> var(3);
 #pragma omp parallel
-#pragma omp for lastprivate(t_var, vec, s_arr, var)
+#pragma omp for lastprivate(t_var, vec, s_arr, var, sivar)
   for (int i = 0; i < 2; ++i) {
     vec[i] = t_var;
     s_arr[i] = var;
+    sivar += i;
   }
 #pragma omp parallel
 #pragma omp for lastprivate(A::x, B::x) firstprivate(f) lastprivate(f)
@@ -193,16 +226,15 @@ int main() {
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, %{{.+}}*)* [[MAIN_MICROTASK1:@.+]] to void
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, %{{.+}}*)* [[MAIN_MICROTASK2:@.+]] to void
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, %{{.+}}*)* [[MAIN_MICROTASK3:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i32*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK1:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK2:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK3:@.+]] to void
 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i32* noalias [[GTID_ADDR:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}})
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -212,21 +244,19 @@ int main() {
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VEC_REF:%.+]] = load [2 x i32]*, [2 x i32]** %
+// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+
 // Check for default initialization.
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
 // CHECK-NOT: [[T_VAR_PRIV]]
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK-NOT: [[VEC_PRIV]]
-// CHECK: [[S_ARR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_REF_PTR]],
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_REF_PTR]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
 // <Skip loop body>
@@ -261,17 +291,18 @@ int main() {
 
 // original var=private_var;
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* [[VAR_REF]], [[S_FLOAT_TY]]* {{.*}} [[VAR_PRIV]])
+// CHECK: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_PRIV]],
 // CHECK: br label %[[LAST_DONE]]
 // CHECK: [[LAST_DONE]]
 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
 //
-// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[F_PRIV:%.+]] = alloca float,
 // CHECK-NOT: alloca float
 // CHECK: [[X_PRIV:%.+]] = alloca double,
@@ -308,11 +339,10 @@ int main() {
 // CHECK-NEXT: br label %[[LAST_DONE]]
 // CHECK: [[LAST_DONE]]
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
-// CHECK: define internal void [[MAIN_MICROTASK2]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK2]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[F_PRIV:%.+]] = alloca float,
 // CHECK-NOT: alloca float
 
@@ -340,11 +370,10 @@ int main() {
 // CHECK-NEXT: br label %[[LAST_DONE]]
 // CHECK: [[LAST_DONE]]
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
-// CHECK: define internal void [[MAIN_MICROTASK3]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK3]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[CNT_PRIV:%.+]] = alloca i8,
 
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
@@ -387,43 +416,41 @@ int main() {
 // CHECK-NEXT: br label %[[LAST_DONE]]
 // CHECK: [[LAST_DONE]]
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
-// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
-// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
-// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
+// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
+// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128
+// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128
+// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
+// CHECK: [[VAR_PRIV_REF:%.+]] = alloca [[S_INT_TY]]*,
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+
 // Check for default initialization.
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
 // CHECK-NOT: [[T_VAR_PRIV]]
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK-NOT: [[VEC_PRIV]]
-// CHECK: [[S_ARR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF_PTR]],
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]],
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
+// CHECK: store [[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]** [[VAR_PRIV_REF]]
 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
 // <Skip loop body>
 // CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
@@ -456,14 +483,15 @@ int main() {
 // CHECK: [[S_ARR_BODY_DONE]]
 
 // original var=private_var;
-// CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[VAR_REF]], [[S_INT_TY]]* {{.*}} [[VAR_PRIV]])
+// CHECK: [[VAR_PRIV1:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PRIV_REF]],
+// CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[VAR_REF]], [[S_INT_TY]]* {{.*}} [[VAR_PRIV1]])
 // CHECK: br label %[[LAST_DONE]]
 // CHECK: [[LAST_DONE]]
 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 #endif
 
diff --git a/test/OpenMP/for_lastprivate_messages.cpp b/test/OpenMP/for_lastprivate_messages.cpp
index 2fa14b63bd11..79912b6efbca 100644
--- a/test/OpenMP/for_lastprivate_messages.cpp
+++ b/test/OpenMP/for_lastprivate_messages.cpp
@@ -67,7 +67,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (int k = 0; k < argc; ++k)
@@ -117,10 +117,6 @@ int foomain(int argc, char **argv) {
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel
-#pragma omp for linear(i) // expected-error {{unexpected OpenMP clause 'linear' in directive '#pragma omp for'}}
-  for (int k = 0; k < argc; ++k)
-    ++k;
-#pragma omp parallel
   {
     int v = 0;
     int i;                     // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for' directive into a parallel or another task region?}}
@@ -132,7 +128,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp for lastprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel
@@ -165,7 +161,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (i = 0; i < argc; ++i)
@@ -272,7 +268,7 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp for lastprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
@@ -283,5 +279,9 @@ int main(int argc, char **argv) {
 #pragma omp for lastprivate(n) firstprivate(n) // OK
   for (i = 0; i < argc; ++i)
     foo();
+  static int si;
+#pragma omp for lastprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/for_linear_codegen.cpp b/test/OpenMP/for_linear_codegen.cpp
new file mode 100644
index 000000000000..db9788361dd0
--- /dev/null
+++ b/test/OpenMP/for_linear_codegen.cpp
@@ -0,0 +1,266 @@
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s
+// expected-no-diagnostics
+// REQUIRES: x86-registered-target
+#ifndef HEADER
+#define HEADER
+
+template <class T>
+struct S {
+  T f;
+  S(T a) : f(a) {}
+  S() : f() {}
+  S<T> &operator=(const S<T> &);
+  operator T() { return T(); }
+  ~S() {}
+};
+
+volatile int g = 1212;
+volatile int &g1 = g;
+float f;
+char cnt;
+
+// CHECK: [[S_FLOAT_TY:%.+]] = type { float }
+// CHECK: [[S_INT_TY:%.+]] = type { i32 }
+// CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
+// CHECK-DAG: [[F:@.+]] = global float 0.0
+// CHECK-DAG: [[CNT:@.+]] = global i8 0
+template <typename T>
+T tmain() {
+  S<T> test;
+  T *pvar = &test.f;
+  T &lvar = test.f;
+#pragma omp parallel
+#pragma omp for linear(pvar, lvar)
+  for (int i = 0; i < 2; ++i) {
+    ++pvar, ++lvar;
+  }
+  return T();
+}
+
+int main() {
+#ifdef LAMBDA
+  // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
+  // LAMBDA-LABEL: @main
+  // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
+  [&]() {
+  // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
+#pragma omp parallel
+#pragma omp for linear(g, g1:5)
+  for (int i = 0; i < 2; ++i) {
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
+    // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+    // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+    // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]]
+    // LAMBDA: [[CNT:%.+]] = load i32, i32*
+    // LAMBDA: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5
+    // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]]
+    // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]],
+    // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], 5
+    // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
+    // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
+    // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
+    g += 5;
+    g1 += 5;
+    // LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
+    [&]() {
+      // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
+      // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
+      g = 2;
+      g1 = 2;
+      // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
+      // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
+      // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+    }();
+  }
+  }();
+  return 0;
+#elif defined(BLOCKS)
+  // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212,
+  // BLOCKS-LABEL: @main
+  // BLOCKS: call void {{%.+}}(i8
+  ^{
+  // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
+#pragma omp parallel
+#pragma omp for linear(g, g1:5)
+  for (int i = 0; i < 2; ++i) {
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
+    // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+    // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+    // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]]
+    // BLOCKS: [[CNT:%.+]] = load i32, i32*
+    // BLOCKS: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5
+    // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]]
+    // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]],
+    // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], 5
+    // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: call void {{%.+}}(i8
+    // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
+    g += 5;
+    g1 += 5;
+    // BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
+    g = 1;
+    g1 = 5;
+    ^{
+      // BLOCKS: define {{.+}} void {{@.+}}(i8*
+      g = 2;
+      g1 = 2;
+      // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS: ret
+    }();
+  }
+  }();
+  return 0;
+#else
+  S<float> test;
+  float *pvar = &test.f;
+  long long lvar = 0;
+#pragma omp parallel
+#pragma omp for linear(pvar, lvar : 3)
+  for (int i = 0; i < 2; ++i) {
+    pvar += 3, lvar += 3;
+  }
+  return tmain<int>();
+#endif
+}
+
+// CHECK: define i{{[0-9]+}} @main()
+// CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, float**, i64*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: = call {{.+}} [[TMAIN_INT:@.+]]()
+// CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
+// CHECK: ret
+
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, i64* dereferenceable(8) %{{.+}})
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_START:%.+]] = alloca float*,
+// CHECK: [[LVAR_START:%.+]] = alloca i64,
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_PRIV:%.+]] = alloca float*,
+// CHECK: [[LVAR_PRIV:%.+]] = alloca i64,
+// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
+
+// Check for default initialization.
+// CHECK: [[PVAR_REF:%.+]] = load float**, float*** %
+// CHECK: [[LVAR_REF:%.+]] = load i64*, i64** %
+// CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_REF]],
+// CHECK: store float* [[PVAR_VAL]], float** [[PVAR_START]],
+// CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_REF]],
+// CHECK: store i64 [[LVAR_VAL]], i64* [[LVAR_START]],
+// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+// CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3
+// CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64
+// CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 [[IDX]]
+// CHECK: store float* [[PTR]], float** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3
+// CHECK: [[CONV:%.+]] = sext i32 [[MUL]] to i64
+// CHECK: [[VAL:%.+]] = add nsw i64 [[LVAR_VAL]], [[CONV]]
+// CHECK: store i64 [[VAL]], i64* [[LVAR_PRIV]],
+// CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_PRIV]]
+// CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 3
+// CHECK: store float* [[PTR]], float** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_PRIV]],
+// CHECK: [[ADD:%.+]] = add nsw i64 [[LVAR_VAL]], 3
+// CHECK: store i64 [[ADD]], i64* [[LVAR_PRIV]],
+// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: ret void
+
+// CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
+// CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
+// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32**, i32*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
+// CHECK: ret
+//
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32** dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}})
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_START:%.+]] = alloca i32*,
+// CHECK: [[LVAR_START:%.+]] = alloca i32,
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_PRIV:%.+]] = alloca i32*,
+// CHECK: [[LVAR_PRIV:%.+]] = alloca i32,
+// CHECK: [[LVAR_PRIV_REF:%.+]] = alloca i32*,
+// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
+
+// Check for default initialization.
+// CHECK: [[PVAR_REF:%.+]] = load i32**, i32*** %
+// CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_REF]],
+// CHECK: store i32* [[PVAR_VAL]], i32** [[PVAR_START]],
+// CHECK: [[LVAR_REF:%.+]] = load i32*, i32** %
+// CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_REF]],
+// CHECK: store i32 [[LVAR_VAL]], i32* [[LVAR_START]],
+// CHECK: store i32* [[LVAR_PRIV]], i32** [[LVAR_PRIV_REF]],
+
+// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+// CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1
+// CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64
+// CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i64 [[IDX]]
+// CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1
+// CHECK: [[VAL:%.+]] = add nsw i32 [[LVAR_VAL]], [[MUL]]
+// CHECK: store i32 [[VAL]], i32* [[LVAR_PRIV]],
+// CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_PRIV]]
+// CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i32 1
+// CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]],
+// CHECK: [[LVAR_PRIV:%.+]] = load i32*, i32** [[LVAR_PRIV_REF]],
+// CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_PRIV]],
+// CHECK: [[ADD:%.+]] = add nsw i32 [[LVAR_VAL]], 1
+// CHECK: store i32 [[ADD]], i32* [[LVAR_PRIV]],
+// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: ret void
+#endif
+
diff --git a/test/OpenMP/for_linear_messages.cpp b/test/OpenMP/for_linear_messages.cpp
new file mode 100644
index 000000000000..39fb21ef7be0
--- /dev/null
+++ b/test/OpenMP/for_linear_messages.cpp
@@ -0,0 +1,218 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+namespace X {
+  int x;
+};
+
+struct B {
+  static int ib; // expected-note {{'B::ib' declared here}}
+  static int bfoo() { return 8; }
+};
+
+int bfoo() { return 4; }
+
+int z;
+const int C1 = 1;
+const int C2 = 2;
+void test_linear_colons()
+{
+  int B = 0;
+  #pragma omp for linear(B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'}}
+  #pragma omp for linear(B::ib:B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{use of undeclared identifier 'ib'; did you mean 'B::ib'}}
+  #pragma omp for linear(B:ib)
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'?}}
+  #pragma omp for linear(z:B:ib)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp for linear(B:B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp for linear(X::x : ::z)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp for linear(B,::z, X::x)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp for linear(::z)
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{expected variable name}}
+  #pragma omp for linear(B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp for linear(B::ib,B:C1+C2)
+  for (int i = 0; i < 10; ++i) ;
+}
+
+template<int L, class T, class N> T test_template(T* arr, N num) {
+  N i;
+  T sum = (T)0;
+  T ind2 = - num * L; // expected-note {{'ind2' defined here}}
+  // expected-error@+1 {{argument of a linear clause should be of integral or pointer type}}
+#pragma omp for linear(ind2:L)
+  for (i = 0; i < num; ++i) {
+    T cur = arr[(int)ind2];
+    ind2 += L;
+    sum += cur;
+  }
+  return T();
+}
+
+template<int LEN> int test_warn() {
+  int ind2 = 0;
+  // expected-warning@+1 {{zero linear step (ind2 should probably be const)}}
+  #pragma omp for linear(ind2:LEN)
+  for (int i = 0; i < 100; i++) {
+    ind2 += LEN;
+  }
+  return ind2;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+public:
+  S2():a(0) { }
+};
+const S2 b; // expected-note 2 {{'b' defined here}}
+const S2 ba[5];
+class S3 {
+  int a;
+public:
+  S3():a(0) { }
+};
+const S3 ca[5];
+class S4 {
+  int a;
+  S4();
+public:
+  S4(int v):a(v) { }
+};
+class S5 {
+  int a;
+  S5():a(0) {}
+public:
+  S5(int v):a(v) { }
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template<class I, class C> int foomain(I argc, C **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+  #pragma omp for linear // expected-error {{expected '(' after 'linear'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear () // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc : 5)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{linear variable with incomplete type 'S1'}}
+  // expected-error@+1 {{const-qualified variable cannot be linear}}
+  #pragma omp for linear (a, b:B::ib)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(e, g)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(h) // expected-error {{threadprivate or thread local variable cannot be linear}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(i)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp parallel
+  {
+    int v = 0;
+    int i;
+    #pragma omp for linear(v:i)
+    for (int k = 0; k < argc; ++k) { i = k; v += i; }
+  }
+  #pragma omp for linear(j)
+  for (int k = 0; k < argc; ++k) ++k;
+  int v = 0;
+  #pragma omp for linear(v:j)
+  for (int k = 0; k < argc; ++k) { ++k; v += j; }
+  #pragma omp for linear(i)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(i) ordered(1) // expected-error {{'linear' clause cannot be specified along with 'ordered' clause with a parameter}}
+  for (int k = 0; k < argc; ++k) ++k;
+  return 0;
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace C {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  double darr[100];
+  // expected-note@+1 {{in instantiation of function template specialization 'test_template<-4, double, int>' requested here}}
+  test_template<-4>(darr, 4);
+  // expected-note@+1 {{in instantiation of function template specialization 'test_warn<0>' requested here}}
+  test_warn<0>();
+
+  S4 e(4); // expected-note {{'e' defined here}}
+  S5 g(5); // expected-note {{'g' defined here}}
+  int i;
+  int &j = i;
+  #pragma omp for linear // expected-error {{expected '(' after 'linear'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear () // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argc)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{linear variable with incomplete type 'S1'}}
+  // expected-error@+1 {{const-qualified variable cannot be linear}}
+  #pragma omp for linear(a, b)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear (argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{argument of a linear clause should be of integral or pointer type, not 'S4'}}
+  // expected-error@+1 {{argument of a linear clause should be of integral or pointer type, not 'S5'}}
+  #pragma omp for linear(e, g)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(h, C::x) // expected-error 2 {{threadprivate or thread local variable cannot be linear}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp parallel
+  {
+    int i;
+    #pragma omp for linear(i)
+    for (int k = 0; k < argc; ++k) ++k;
+    #pragma omp for linear(i : 4)
+    for (int k = 0; k < argc; ++k) { ++k; i += 4; }
+  }
+  #pragma omp for linear(j)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(i)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp for linear(i) ordered(1) // expected-error {{'linear' clause cannot be specified along with 'ordered' clause with a parameter}}
+  for (int k = 0; k < argc; ++k) ++k;
+
+  foomain<int,char>(argc,argv);
+  return 0;
+}
+
diff --git a/test/OpenMP/for_loop_messages.cpp b/test/OpenMP/for_loop_messages.cpp
index 1a1315507893..895baf57e9cc 100644
--- a/test/OpenMP/for_loop_messages.cpp
+++ b/test/OpenMP/for_loop_messages.cpp
@@ -10,10 +10,13 @@ public:
 };
 
 static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
 #pragma omp threadprivate(sii)
 static int globalii;
 
-register int reg0 __asm__("0");
+// Currently, we cannot use "0" for global register variables.
+// register int reg0 __asm__("0");
+int reg0;
 
 int test_iteration_spaces() {
   const int N = 100;
@@ -66,37 +69,37 @@ int test_iteration_spaces() {
     c[(int)fi] = a[(int)fi] + b[(int)fi];
   }
 #pragma omp parallel
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (int &ref = ii; ref < 10; ref++) {
   }
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (int i; i < 10; i++)
     c[i] = a[i];
 
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (int i = 0, j = 0; i < 10; ++i)
     c[i] = a[i];
 
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (; ii < 10; ++ii)
     c[ii] = a[ii];
 
 #pragma omp parallel
 // expected-warning@+3 {{expression result unused}}
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (ii + 1; ii < 10; ++ii)
     c[ii] = a[ii];
 
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (c[ii] = 0; ii < 10; ++ii)
     c[ii] = a[ii];
@@ -289,7 +292,6 @@ int test_iteration_spaces() {
     c[ii] = a[ii];
 
 #pragma omp parallel
-// expected-error@+3 {{unexpected OpenMP clause 'linear' in directive '#pragma omp for'}}
 // expected-note@+2  {{defined as linear}}
 // expected-error@+2 {{loop iteration variable in the associated loop of 'omp for' directive may not be linear, predetermined as private}}
 #pragma omp for linear(ii)
@@ -308,6 +310,7 @@ int test_iteration_spaces() {
 
 #pragma omp parallel
   {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp for' directive may not be threadprivate or thread local, predetermined as private}}
 #pragma omp for
     for (sii = 0; sii < 10; sii += 1)
       c[sii] = a[sii];
@@ -447,7 +450,7 @@ int test_with_random_access_iterator() {
   for (GoodIter I = begin; I < end; ++I)
     ++I;
 #pragma omp parallel
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (GoodIter &I = begin; I < end; ++I)
     ++I;
@@ -486,7 +489,7 @@ int test_with_random_access_iterator() {
   for (begin = begin0; begin < end; ++begin)
     ++begin;
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for
   for (++begin; begin < end; ++begin)
     ++begin;
diff --git a/test/OpenMP/for_misc_messages.c b/test/OpenMP/for_misc_messages.c
index f5b87514947c..0a7cdf8fc06b 100644
--- a/test/OpenMP/for_misc_messages.c
+++ b/test/OpenMP/for_misc_messages.c
@@ -62,9 +62,8 @@ void test_non_identifiers() {
 #pragma omp for;
   for (i = 0; i < 16; ++i)
     ;
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp for' are ignored}}
 #pragma omp parallel
-// expected-error@+2 {{unexpected OpenMP clause 'linear' in directive '#pragma omp for'}}
-// expected-warning@+1 {{extra tokens at the end of '#pragma omp for' are ignored}}
 #pragma omp for linear(x);
   for (i = 0; i < 16; ++i)
     ;
@@ -176,17 +175,17 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp for collapse(-5)
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp for collapse(0)
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp for collapse(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
@@ -195,7 +194,7 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
 // expected-note@+1 {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for' directive into a parallel or another task region?}}
     for (int j = 0; j < 16; ++j)
-// expected-error@+2 {{private variable cannot be reduction}}
+// expected-error@+2 {{reduction variable must be shared}}
 // expected-error@+1 {{region cannot be closely nested inside 'for' region; perhaps you forget to enclose 'omp for' directive into a parallel region?}}
 #pragma omp for reduction(+ : i, j)
       for (int k = 0; k < 16; ++k)
diff --git a/test/OpenMP/for_ordered_clause.cpp b/test/OpenMP/for_ordered_clause.cpp
new file mode 100644
index 000000000000..8af509ab9aa8
--- /dev/null
+++ b/test/OpenMP/for_ordered_clause.cpp
@@ -0,0 +1,120 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) {                   //expected-note 2 {{declared here}}
+#pragma omp for ordered
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered() // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+// expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+// expected-error@+2 2 {{expression is not an integral constant expression}}
+// expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+#pragma omp for ordered(argc
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+// expected-error@+1 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+#pragma omp for ordered(ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered(1)) // expected-warning {{extra tokens at the end of '#pragma omp for' are ignored}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'ordered' clause}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp for', but found only 1}}
+// expected-error@+3 2 {{directive '#pragma omp for' cannot contain more than one 'ordered' clause}}
+// expected-error@+2 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+// expected-error@+1 2 {{expression is not an integral constant expression}}
+#pragma omp for ordered(foobool(argc)), ordered(true), ordered(-5)
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered(S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+// expected-error@+1 2 {{expression is not an integral constant expression}}
+#pragma omp for ordered(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered(1)
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered(N-1) // expected-error 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered(N) // expected-error {{argument to 'ordered' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp for ordered(2) // expected-note {{as specified in 'ordered' clause}}
+  foo();                    // expected-error {{expected 2 for loops after '#pragma omp for'}}
+#pragma omp for ordered(N) collapse(N + 2) // expected-error {{the parameter of the 'ordered' clause must be greater than or equal to the parameter of the 'collapse' clause}} expected-note {{parameter of the 'collapse' clause}} expected-error {{argument to 'ordered' clause must be a strictly positive integer value}}
+  for (int i = ST; i < N; i++)
+    for (int j = ST; j < N; j++)
+      for (int k = ST; k < N; k++)
+        foo();
+  return argc;
+}
+
+int main(int argc, char **argv) {
+#pragma omp for ordered
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp for ordered( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp for ordered() // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp for ordered(4 // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-note {{as specified in 'ordered' clause}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4]; // expected-error {{expected 4 for loops after '#pragma omp for', but found only 1}}
+#pragma omp for ordered(2 + 2))              // expected-warning {{extra tokens at the end of '#pragma omp for' are ignored}}  expected-note {{as specified in 'ordered' clause}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];    // expected-error {{expected 4 for loops after '#pragma omp for', but found only 1}}
+#pragma omp for ordered(foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+// expected-error@+3 {{expression is not an integral constant expression}}
+// expected-error@+2 2 {{directive '#pragma omp for' cannot contain more than one 'ordered' clause}}
+// expected-error@+1 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+#pragma omp for ordered(foobool(argc)), ordered(true), ordered(-5)
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp for ordered(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+// expected-error@+1 {{expression is not an integral constant expression}}
+#pragma omp for ordered(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+// expected-error@+3 {{statement after '#pragma omp for' must be a for loop}}
+// expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+#pragma omp for ordered(ordered(tmain < int, char, -1, -2 > (argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+#pragma omp for ordered(2) // expected-note {{as specified in 'ordered' clause}}
+  foo();                    // expected-error {{expected 2 for loops after '#pragma omp for'}}
+#pragma omp for ordered(0)              // expected-error {{argument to 'ordered' clause must be a strictly positive integer value}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp for ordered(2) collapse(3) // expected-error {{the parameter of the 'ordered' clause must be greater than or equal to the parameter of the 'collapse' clause}} expected-note {{parameter of the 'collapse' clause}}
+  for (int i = 0; i < 10; i++)
+    for (int j = 0; j < 11; j++)
+      for (int k = 0; k < 12; k++)
+        foo();
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
+  return tmain<int, char, 1, 0>(argc, argv);
+}
+
diff --git a/test/OpenMP/for_private_codegen.cpp b/test/OpenMP/for_private_codegen.cpp
index ed6d87bcbae0..e8f4d3f4c830 100644
--- a/test/OpenMP/for_private_codegen.cpp
+++ b/test/OpenMP/for_private_codegen.cpp
@@ -18,19 +18,17 @@ struct S {
 };
 
 volatile double g;
+volatile double &g1 = g;
 
 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[CAP_MAIN_TY:%.+]] = type { i8 }
-// CHECK: type { i8 }
 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK: [[CAP_TMAIN_TY:%.+]] = type { i8 }
 template <typename T>
 T tmain() {
   S<T> test;
   T t_var = T();
   T vec[] = {1, 2};
   S<T> s_arr[] = {1, 2};
-  S<T> var(3);
+  S<T> &var = test;
 #pragma omp parallel
 #pragma omp for private(t_var, vec, s_arr, s_arr, var, var)
   for (int i = 0; i < 2; ++i) {
@@ -41,34 +39,55 @@ T tmain() {
 }
 
 int main() {
+  static int svar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global double
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
+  static float sfvar;
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for private(g)
+#pragma omp for private(g, g1, svar, sfvar)
   for (int i = 0; i < 2; ++i) {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double,
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // LAMBDA: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float,
     g = 1;
+    g1 = 1;
+    svar = 3;
+    sfvar = 4.0;
     // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4(
     // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]],
+    // LAMBDA: store float 4.0{{.+}}, float* [[SFVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store double* [[G_PRIVATE_ADDR]], double** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SVAR_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SFVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
+    // LAMBDA: store float* [[SFVAR_PRIVATE_ADDR]], float** [[SFVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call {{.*}}void @__kmpc_for_static_fini(
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      g1 = 2;
+      svar = 4;
+      sfvar = 8.0;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
       // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
+      // LAMBDA: [[SVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SVAR_REF]]
+      // LAMBDA: [[SFVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
+      // LAMBDA: [[SFVAR_REF:%.+]] = load float*, float** [[SFVAR_PTR_REF]]
+      // LAMBDA: store float 8.0{{.+}}, float* [[SFVAR_REF]]
     }();
   }
   }();
@@ -78,28 +97,50 @@ int main() {
   // BLOCKS-LABEL: @main
   // BLOCKS: call {{.*}}void {{%.+}}(i8
   ^{
+  static float sfvar;
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* {{.+}})
+  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for private(g)
+#pragma omp for private(g, g1, svar, sfvar)
   for (int i = 0; i < 2; ++i) {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca double,
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // BLOCKS: [[SVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[SFVAR_PRIVATE_ADDR:%.+]] = alloca float,
     g = 1;
+    g1 = 1;
+    svar = 2;
+    sfvar = 3.0;
     // BLOCKS: call {{.*}}void @__kmpc_for_static_init_4(
     // BLOCKS: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]],
+    // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
+    // BLOCKS: store float 3.0{{.+}}, float* [[SFVAR_PRIVATE_ADDR]],
+    // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
     // BLOCKS: double* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SVAR_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
+    // BLOCKS: float* [[SFVAR_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
     // BLOCKS: call {{.*}}void @__kmpc_for_static_fini(
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      g1 = 2;
+      svar = 4;
+      sfvar = 9.0;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store double 2.0{{.+}}, double*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SVAR]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
+      // BLOCKS: store float 9.0{{.+}}, float*
+      // BLOCKS-NOT: [[SFVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -110,9 +151,9 @@ int main() {
   int t_var = 0;
   int vec[] = {1, 2};
   S<float> s_arr[] = {1, 2};
-  S<float> var(3);
+  S<float> &var = test;
 #pragma omp parallel
-#pragma omp for private(t_var, vec, s_arr, s_arr, var, var)
+#pragma omp for private(t_var, vec, s_arr, s_arr, var, var, svar)
   for (int i = 0; i < 2; ++i) {
     vec[i] = t_var;
     s_arr[i] = var;
@@ -130,19 +171,19 @@ int main() {
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK-NOT: alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK-NOT: alloca [[S_FLOAT_TY]],
+// CHECK: [[S_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
@@ -161,11 +202,11 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
diff --git a/test/OpenMP/for_private_messages.cpp b/test/OpenMP/for_private_messages.cpp
index 225a1beb99ea..3015f819b544 100644
--- a/test/OpenMP/for_private_messages.cpp
+++ b/test/OpenMP/for_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp for private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -99,7 +99,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp for private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp for private(i)
@@ -127,7 +127,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp for private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -179,12 +179,16 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp for private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp for private(i)
   for (int k = 0; k < argc; ++k)
     ++k;
+  static int si;
+#pragma omp for private(si) // OK
+  for(int k = 0; k < argc; ++k)
+    si = k + 1;
 
   return 0;
 }
diff --git a/test/OpenMP/for_reduction_codegen.cpp b/test/OpenMP/for_reduction_codegen.cpp
index 7de8dff31c4c..423ab3ce76d0 100644
--- a/test/OpenMP/for_reduction_codegen.cpp
+++ b/test/OpenMP/for_reduction_codegen.cpp
@@ -8,7 +8,8 @@
 #ifndef HEADER
 #define HEADER
 
-volatile double g;
+volatile double g, g_orig;
+volatile double &g1 = g_orig;
 
 template <class T>
 struct S {
@@ -22,8 +23,6 @@ struct S {
 
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { float*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float*, [2 x i{{[0-9]+}}]*, [2 x [[S_FLOAT_TY]]]* }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [[S_INT_TY]]*, [[S_INT_TY]]*, i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]* }
 // CHECK-DAG: [[ATOMIC_REDUCE_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 18, i32 0, i32 0, i8*
 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-DAG: [[REDUCTION_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 18, i32 0, i32 0, i8*
@@ -36,7 +35,8 @@ T tmain() {
   T t_var = T(), t_var1;
   T vec[] = {1, 2};
   S<T> s_arr[] = {1, 2};
-  S<T> var(3), var1;
+  S<T> &var = test;
+  S<T> var1;
 #pragma omp parallel
 #pragma omp for reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1) nowait
   for (int i = 0; i < 2; ++i) {
@@ -59,26 +59,27 @@ int main() {
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for reduction(+:g)
+#pragma omp for reduction(+:g, g1)
     for (int i = 0; i < 2; ++i) {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* %{{.+}})
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double,
 
     // Reduction list for runtime.
-    // LAMBDA: [[RED_LIST:%.+]] = alloca [1 x i8*],
+    // LAMBDA: [[RED_LIST:%.+]] = alloca [2 x i8*],
 
     // LAMBDA: store double 0.0{{.+}}, double* [[G_PRIVATE_ADDR]]
     // LAMBDA: call void @__kmpc_for_static_init_4(
     g = 1;
+    g1 = 1;
     // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store double* [[G_PRIVATE_ADDR]], double** [[G_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call void @__kmpc_for_static_fini(
 
-    // LAMBDA: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i32 0, i32 0
+    // LAMBDA: [[G_PRIV_REF:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[RED_LIST]], i64 0, i64 0
     // LAMBDA: [[BITCAST:%.+]] = bitcast double* [[G_PRIVATE_ADDR]] to i8*
     // LAMBDA: store i8* [[BITCAST]], i8** [[G_PRIV_REF]],
     // LAMBDA: call i32 @__kmpc_reduce(
@@ -104,6 +105,7 @@ int main() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      g1 = 2;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
@@ -118,18 +120,19 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp for reduction(-:g)
+#pragma omp for reduction(-:g, g1)
     for (int i = 0; i < 2; ++i)  {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* %{{.+}})
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca double,
 
     // Reduction list for runtime.
-    // BLOCKS: [[RED_LIST:%.+]] = alloca [1 x i8*],
+    // BLOCKS: [[RED_LIST:%.+]] = alloca [2 x i8*],
 
     // BLOCKS: store double 0.0{{.+}}, double* [[G_PRIVATE_ADDR]]
     g = 1;
+    g1 = 1;
     // BLOCKS: call void @__kmpc_for_static_init_4(
     // BLOCKS: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
@@ -138,7 +141,7 @@ int main() {
     // BLOCKS: call void {{%.+}}(i8
     // BLOCKS: call void @__kmpc_for_static_fini(
 
-    // BLOCKS: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i32 0, i32 0
+    // BLOCKS: [[G_PRIV_REF:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[RED_LIST]], i64 0, i64 0
     // BLOCKS: [[BITCAST:%.+]] = bitcast double* [[G_PRIVATE_ADDR]] to i8*
     // BLOCKS: store i8* [[BITCAST]], i8** [[G_PRIV_REF]],
     // BLOCKS: call i32 @__kmpc_reduce(
@@ -163,6 +166,7 @@ int main() {
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      g1 = 2;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store double 2.0{{.+}}, double*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
@@ -176,13 +180,18 @@ int main() {
   float t_var = 0, t_var1;
   int vec[] = {1, 2};
   S<float> s_arr[] = {1, 2};
-  S<float> var(3), var1;
+  S<float> &var = test;
+  S<float> var1, arrs[10][4];
 #pragma omp parallel
 #pragma omp for reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1)
   for (int i = 0; i < 2; ++i) {
     vec[i] = t_var;
     s_arr[i] = var;
   }
+  int arr[10][vec[1]];
+#pragma omp parallel for reduction(+:arr[1][:vec[1]]) reduction(&:arrs[1:vec[1]][1:2])
+  for (int i = 0; i < 10; ++i)
+    ++arr[1][i];
   return tmain<int>();
 #endif
 }
@@ -190,13 +199,13 @@ int main() {
 // CHECK: define {{.*}}i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, float*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i64, i64, i32*, [2 x i32]*, [10 x [4 x [[S_FLOAT_TY]]]]*)* [[MAIN_MICROTASK1:@.+]] to void
 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float* dereferenceable(4) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, float* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %vec, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca float,
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: [[VAR1_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
@@ -207,23 +216,20 @@ int main() {
 
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR_REF:%.+]] = load float*, float** [[T_VAR_PTR_REF]],
+// CHECK: [[T_VAR_REF:%.+]] = load float*, float** %
+// CHECK: [[VAR1_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+// CHECK: [[T_VAR1_REF:%.+]] = load float*, float** %
+
 // For + reduction operation initial value of private variable is 0.
 // CHECK: store float 0.0{{.+}}, float* [[T_VAR_PRIV]],
 
-// CHECK: [[VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_PTR_REF:%.+]],
 // For & reduction operation initial value of private variable is ones in all bits.
+// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 
-// CHECK: [[VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR1_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_PTR_REF:%.+]],
 // For && reduction operation initial value of private variable is 1.0.
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[VAR1_PRIV]])
 
-// CHECK: [[T_VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR1_REF:%.+]] = load float*, float** [[T_VAR1_PTR_REF]],
 // For min reduction operation initial value of private variable is largest repesentable value.
 // CHECK: store float 0x47EFFFFFE0000000, float* [[T_VAR1_PRIV]],
 
@@ -236,16 +242,16 @@ int main() {
 
 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
 
-// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 0
+// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 0
 // CHECK: [[BITCAST:%.+]] = bitcast float* [[T_VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR_PRIV_REF]],
-// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 1
+// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 1
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR_PRIV_REF]],
-// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 2
+// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 2
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR1_PRIV_REF]],
-// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 3
+// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 3
 // CHECK: [[BITCAST:%.+]] = bitcast float* [[T_VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR1_PRIV_REF]],
 
@@ -372,8 +378,7 @@ int main() {
 // CHECK: [[RED_DONE]]
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 
 // CHECK: ret void
 
@@ -385,38 +390,38 @@ int main() {
 // }
 // CHECK: define internal void [[REDUCTION_FUNC]](i8*, i8*)
 // t_var_lhs = (float*)lhs[0];
-// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR_RHS_REF]],
 // CHECK: [[T_VAR_RHS:%.+]] = bitcast i8* [[T_VAR_RHS_VOID]] to float*
 // t_var_rhs = (float*)rhs[0];
-// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR_LHS_REF]],
 // CHECK: [[T_VAR_LHS:%.+]] = bitcast i8* [[T_VAR_LHS_VOID]] to float*
 
 // var_lhs = (S<float>*)lhs[1];
-// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 1
+// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 1
 // CHECK: [[VAR_RHS_VOID:%.+]] = load i8*, i8** [[VAR_RHS_REF]],
 // CHECK: [[VAR_RHS:%.+]] = bitcast i8* [[VAR_RHS_VOID]] to [[S_FLOAT_TY]]*
 // var_rhs = (S<float>*)rhs[1];
-// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 1
+// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 1
 // CHECK: [[VAR_LHS_VOID:%.+]] = load i8*, i8** [[VAR_LHS_REF]],
 // CHECK: [[VAR_LHS:%.+]] = bitcast i8* [[VAR_LHS_VOID]] to [[S_FLOAT_TY]]*
 
 // var1_lhs = (S<float>*)lhs[2];
-// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 2
+// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 2
 // CHECK: [[VAR1_RHS_VOID:%.+]] = load i8*, i8** [[VAR1_RHS_REF]],
 // CHECK: [[VAR1_RHS:%.+]] = bitcast i8* [[VAR1_RHS_VOID]] to [[S_FLOAT_TY]]*
 // var1_rhs = (S<float>*)rhs[2];
-// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 2
+// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 2
 // CHECK: [[VAR1_LHS_VOID:%.+]] = load i8*, i8** [[VAR1_LHS_REF]],
 // CHECK: [[VAR1_LHS:%.+]] = bitcast i8* [[VAR1_LHS_VOID]] to [[S_FLOAT_TY]]*
 
 // t_var1_lhs = (float*)lhs[3];
-// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_RHS_REF]],
 // CHECK: [[T_VAR1_RHS:%.+]] = bitcast i8* [[T_VAR1_RHS_VOID]] to float*
 // t_var1_rhs = (float*)rhs[3];
-// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_LHS_REF]],
 // CHECK: [[T_VAR1_LHS:%.+]] = bitcast i8* [[T_VAR1_LHS_VOID]] to float*
 
@@ -457,15 +462,218 @@ int main() {
 // CHECK: store float [[UP]], float* [[T_VAR1_LHS]],
 // CHECK: ret void
 
+// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i64 %{{.+}}, i64 %{{.+}}, i32* nonnull %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [10 x [4 x [[S_FLOAT_TY]]]]* dereferenceable(160) %{{.+}})
+
+// Reduction list for runtime.
+// CHECK: [[RED_LIST:%.+]] = alloca [4 x i8*],
+
+// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
+
+// CHECK: [[IDX1:%.+]] = mul nsw i64 1, %{{.+}}
+// CHECK: [[LB1:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[LB1_0:%.+]] = getelementptr inbounds i32, i32* [[LB1]], i64 0
+// CHECK: [[IDX1:%.+]] = mul nsw i64 1, %{{.+}}
+// CHECK: [[UB1:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[UB1_UP:%.+]] = getelementptr inbounds i32, i32* [[UB1]], i64 %
+// CHECK: [[UB_CAST:%.+]] = ptrtoint i32* [[UB1_UP]] to i64
+// CHECK: [[LB_CAST:%.+]] = ptrtoint i32* [[LB1_0]] to i64
+// CHECK: [[DIFF:%.+]] = sub i64 [[UB_CAST]], [[LB_CAST]]
+// CHECK: [[SIZE_1:%.+]] = sdiv exact i64 [[DIFF]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
+// CHECK: [[ARR_SIZE:%.+]] = add nuw i64 [[SIZE_1]], 1
+// CHECK: call i8* @llvm.stacksave()
+// CHECK: [[ARR_PRIV:%.+]] = alloca i32, i64 [[ARR_SIZE]],
+
+// Check initialization of private copy.
+// CHECK: [[END:%.+]] = getelementptr i32, i32* [[ARR_PRIV]], i64 [[ARR_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq i32* [[ARR_PRIV]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi i32*
+// CHECK: store i32 0, i32* %
+// CHECK: [[DONE:%.+]] = icmp eq i32* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// CHECK: [[ARRS_PRIV:%.+]] = alloca [[S_FLOAT_TY]], i64 [[ARRS_SIZE:%.+]],
+
+// Check initialization of private copy.
+// CHECK: [[END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[ARRS_PRIV]], i64 [[ARRS_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[ARRS_PRIV]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi [[S_FLOAT_TY]]*
+// CHECK: call void @_ZN1SIfEC1Ev([[S_FLOAT_TY]]* %
+// CHECK: [[DONE:%.+]] = icmp eq [[S_FLOAT_TY]]* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
+// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+// CHECK: call void @__kmpc_for_static_init_4(
+// Skip checks for internal operations.
+// CHECK: call void @__kmpc_for_static_fini(
+
+// void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
+
+// CHECK: [[ARR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 0
+// CHECK: [[BITCAST:%.+]] = bitcast i32* [[ARR_PRIV]] to i8*
+// CHECK: store i8* [[BITCAST]], i8** [[ARR_PRIV_REF]],
+// CHECK: [[ARR_SIZE_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 1
+// CHECK: [[BITCAST:%.+]] = inttoptr i64 [[ARR_SIZE]] to i8*
+// CHECK: store i8* [[BITCAST]], i8** [[ARR_SIZE_REF]],
+// CHECK: [[ARRS_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 2
+// CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[ARRS_PRIV]] to i8*
+// CHECK: store i8* [[BITCAST]], i8** [[ARRS_PRIV_REF]],
+// CHECK: [[ARRS_SIZE_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 3
+// CHECK: [[BITCAST:%.+]] = inttoptr i64 [[ARRS_SIZE]] to i8*
+// CHECK: store i8* [[BITCAST]], i8** [[ARRS_SIZE_REF]],
+
+// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);
+
+// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
+// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+// CHECK: [[BITCAST:%.+]] = bitcast [4 x i8*]* [[RED_LIST]] to i8*
+// CHECK: [[RES:%.+]] = call i32 @__kmpc_reduce_nowait(%{{.+}}* [[REDUCTION_LOC]], i32 [[GTID]], i32 2, i64 32, i8* [[BITCAST]], void (i8*, i8*)* [[REDUCTION_FUNC:@.+]], [8 x i32]* [[REDUCTION_LOCK]])
+
+// switch(res)
+// CHECK: switch i32 [[RES]], label %[[RED_DONE:.+]] [
+// CHECK: i32 1, label %[[CASE1:.+]]
+// CHECK: i32 2, label %[[CASE2:.+]]
+// CHECK: ]
+
+// case 1:
+// CHECK: [[CASE1]]
+
+// arr[:] += arr_reduction[:];
+// CHECK: [[END:%.+]] = getelementptr i32, i32* [[LB1_0]], i64 [[ARR_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq i32* [[LB1_0]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi i32*
+// CHECK: [[ADD:%.+]] = add nsw i32 %
+// CHECK: store i32 [[ADD]], i32* %
+// CHECK: [[DONE:%.+]] = icmp eq i32* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// arrs[:] = var.operator &(arrs_reduction[:]);
+// CHECK: [[END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[ARRS_LB:%.+]], i64 [[ARRS_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[ARRS_LB]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi [[S_FLOAT_TY]]*
+// CHECK: [[AND:%.+]] = call dereferenceable(4) [[S_FLOAT_TY]]* @_ZN1SIfEanERKS0_([[S_FLOAT_TY]]* %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}})
+// CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[AND]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* [[BITCAST]], i64 4, i32 4, i1 false)
+// CHECK: [[DONE:%.+]] = icmp eq [[S_FLOAT_TY]]* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);
+// CHECK: call void @__kmpc_end_reduce_nowait(%{{.+}}* [[REDUCTION_LOC]], i32 [[GTID]], [8 x i32]* [[REDUCTION_LOCK]])
+
+// break;
+// CHECK: br label %[[RED_DONE]]
+
+// case 2:
+// CHECK: [[CASE2]]
+
+// arr[:] += arr_reduction[:];
+// CHECK: [[END:%.+]] = getelementptr i32, i32* [[LB1_0]], i64 [[ARR_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq i32* [[LB1_0]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi i32*
+// CHECK: atomicrmw add i32* %{{.+}}, i32 %{{.+}} monotonic
+// CHECK: [[DONE:%.+]] = icmp eq i32* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// arrs[:] = var.operator &(arrs_reduction[:]);
+// CHECK: [[END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[ARRS_LB:%.+]], i64 [[ARRS_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[ARRS_LB]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi [[S_FLOAT_TY]]*
+// CHECK: call void @__kmpc_critical(
+// CHECK: [[AND:%.+]] = call dereferenceable(4) [[S_FLOAT_TY]]* @_ZN1SIfEanERKS0_([[S_FLOAT_TY]]* %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}})
+// CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[AND]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* [[BITCAST]], i64 4, i32 4, i1 false)
+// CHECK: call void @__kmpc_end_critical(
+// CHECK: [[DONE:%.+]] = icmp eq [[S_FLOAT_TY]]* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// break;
+// CHECK: br label %[[RED_DONE]]
+// CHECK: [[RED_DONE]]
+
+// Check destruction of private copy.
+// CHECK: [[END:%.+]] = getelementptr inbounds [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[ARRS_PRIV]], i64 [[ARRS_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[ARRS_PRIV]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi [[S_FLOAT_TY]]*
+// CHECK: call void @_ZN1SIfED1Ev([[S_FLOAT_TY]]* %
+// CHECK: [[DONE:%.+]] = icmp eq [[S_FLOAT_TY]]* %{{.+}}, [[ARRS_PRIV]]
+// CHECK: br i1 [[DONE]],
+// CHECK: call void @llvm.stackrestore(i8*
+
+// CHECK: ret void
+
+// void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
+//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
+//  ...
+//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
+//  *(Type<n>-1*)rhs[<n>-1]);
+// }
+// CHECK: define internal void [[REDUCTION_FUNC]](i8*, i8*)
+// arr_rhs = (int*)rhs[0];
+// CHECK: [[ARR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i64 0, i64 0
+// CHECK: [[ARR_RHS_VOID:%.+]] = load i8*, i8** [[ARR_RHS_REF]],
+// CHECK: [[ARR_RHS:%.+]] = bitcast i8* [[ARR_RHS_VOID]] to i32*
+// arr_lhs = (int*)lhs[0];
+// CHECK: [[ARR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i64 0, i64 0
+// CHECK: [[ARR_LHS_VOID:%.+]] = load i8*, i8** [[ARR_LHS_REF]],
+// CHECK: [[ARR_LHS:%.+]] = bitcast i8* [[ARR_LHS_VOID]] to i32*
+
+// arr_size = (size_t)lhs[1];
+// CHECK: [[ARR_SIZE_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 1
+// CHECK: [[ARR_SIZE_VOID:%.+]] = load i8*, i8** [[ARR_SIZE_REF]],
+// CHECK: [[ARR_SIZE:%.+]] = ptrtoint i8* [[ARR_SIZE_VOID]] to i64
+
+// arrs_rhs = (S<float>*)rhs[2];
+// CHECK: [[ARRS_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 2
+// CHECK: [[ARRS_RHS_VOID:%.+]] = load i8*, i8** [[ARRS_RHS_REF]],
+// CHECK: [[ARRS_RHS:%.+]] = bitcast i8* [[ARRS_RHS_VOID]] to [[S_FLOAT_TY]]*
+// arrs_lhs = (S<float>*)lhs[2];
+// CHECK: [[ARRS_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 2
+// CHECK: [[ARRS_LHS_VOID:%.+]] = load i8*, i8** [[ARRS_LHS_REF]],
+// CHECK: [[ARRS_LHS:%.+]] = bitcast i8* [[ARRS_LHS_VOID]] to [[S_FLOAT_TY]]*
+
+// arrs_size = (size_t)lhs[3];
+// CHECK: [[ARRS_SIZE_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 3
+// CHECK: [[ARRS_SIZE_VOID:%.+]] = load i8*, i8** [[ARRS_SIZE_REF]],
+// CHECK: [[ARRS_SIZE:%.+]] = ptrtoint i8* [[ARRS_SIZE_VOID]] to i64
+
+// arr_lhs[:] += arr_rhs[:];
+// CHECK: [[END:%.+]] = getelementptr i32, i32* [[ARR_LHS]], i64 [[ARR_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq i32* [[ARR_LHS]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi i32*
+// CHECK: [[ADD:%.+]] = add nsw i32 %
+// CHECK: store i32 [[ADD]], i32* %
+// CHECK: [[DONE:%.+]] = icmp eq i32* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// arrs_lhs = arrs_lhs.operator &(arrs_rhs);
+// CHECK: [[END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[ARRS_LB:%.+]], i64 [[ARRS_SIZE]]
+// CHECK: [[ISEMPTY:%.+]] = icmp eq [[S_FLOAT_TY]]* [[ARRS_LB]], [[END]]
+// CHECK: br i1 [[ISEMPTY]],
+// CHECK: phi [[S_FLOAT_TY]]*
+// CHECK: [[AND:%.+]] = call dereferenceable(4) [[S_FLOAT_TY]]* @_ZN1SIfEanERKS0_([[S_FLOAT_TY]]* %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}})
+// CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[AND]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* [[BITCAST]], i64 4, i32 4, i1 false)
+// CHECK: [[DONE:%.+]] = icmp eq [[S_FLOAT_TY]]* %{{.+}}, [[END]]
+// CHECK: br i1 [[DONE]],
+
+// CHECK: ret void
+
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_TMAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [[S_INT_TY]]*, [[S_INT_TY]]*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}})
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -481,23 +689,20 @@ int main() {
 
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VAR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+// CHECK: [[T_VAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+
 // For + reduction operation initial value of private variable is 0.
 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[T_VAR_PRIV]],
 
-// CHECK: [[VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PTR_REF:%.+]],
 // For & reduction operation initial value of private variable is ones in all bits.
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[VAR_PRIV]])
 
-// CHECK: [[VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PTR_REF:%.+]],
 // For && reduction operation initial value of private variable is 1.0.
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[VAR1_PRIV]])
 
-// CHECK: [[T_VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR1_PTR_REF]],
 // For min reduction operation initial value of private variable is largest repesentable value.
 // CHECK: store i{{[0-9]+}} 2147483647, i{{[0-9]+}}* [[T_VAR1_PRIV]],
 
@@ -509,16 +714,16 @@ int main() {
 
 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
 
-// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 0
+// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 0
 // CHECK: [[BITCAST:%.+]] = bitcast i{{[0-9]+}}* [[T_VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR_PRIV_REF]],
-// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 1
+// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 1
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR_PRIV_REF]],
-// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 2
+// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 2
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR1_PRIV_REF]],
-// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 3
+// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 3
 // CHECK: [[BITCAST:%.+]] = bitcast i{{[0-9]+}}* [[T_VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR1_PRIV_REF]],
 
@@ -616,7 +821,6 @@ int main() {
 // CHECK: [[RED_DONE]]
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]*
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
 // void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
@@ -627,38 +831,38 @@ int main() {
 // }
 // CHECK: define internal void [[REDUCTION_FUNC]](i8*, i8*)
 // t_var_lhs = (i{{[0-9]+}}*)lhs[0];
-// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR_RHS_REF]],
 // CHECK: [[T_VAR_RHS:%.+]] = bitcast i8* [[T_VAR_RHS_VOID]] to i{{[0-9]+}}*
 // t_var_rhs = (i{{[0-9]+}}*)rhs[0];
-// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR_LHS_REF]],
 // CHECK: [[T_VAR_LHS:%.+]] = bitcast i8* [[T_VAR_LHS_VOID]] to i{{[0-9]+}}*
 
 // var_lhs = (S<i{{[0-9]+}}>*)lhs[1];
-// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 1
+// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 1
 // CHECK: [[VAR_RHS_VOID:%.+]] = load i8*, i8** [[VAR_RHS_REF]],
 // CHECK: [[VAR_RHS:%.+]] = bitcast i8* [[VAR_RHS_VOID]] to [[S_INT_TY]]*
 // var_rhs = (S<i{{[0-9]+}}>*)rhs[1];
-// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 1
+// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 1
 // CHECK: [[VAR_LHS_VOID:%.+]] = load i8*, i8** [[VAR_LHS_REF]],
 // CHECK: [[VAR_LHS:%.+]] = bitcast i8* [[VAR_LHS_VOID]] to [[S_INT_TY]]*
 
 // var1_lhs = (S<i{{[0-9]+}}>*)lhs[2];
-// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 2
+// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 2
 // CHECK: [[VAR1_RHS_VOID:%.+]] = load i8*, i8** [[VAR1_RHS_REF]],
 // CHECK: [[VAR1_RHS:%.+]] = bitcast i8* [[VAR1_RHS_VOID]] to [[S_INT_TY]]*
 // var1_rhs = (S<i{{[0-9]+}}>*)rhs[2];
-// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 2
+// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 2
 // CHECK: [[VAR1_LHS_VOID:%.+]] = load i8*, i8** [[VAR1_LHS_REF]],
 // CHECK: [[VAR1_LHS:%.+]] = bitcast i8* [[VAR1_LHS_VOID]] to [[S_INT_TY]]*
 
 // t_var1_lhs = (i{{[0-9]+}}*)lhs[3];
-// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_RHS_REF]],
 // CHECK: [[T_VAR1_RHS:%.+]] = bitcast i8* [[T_VAR1_RHS_VOID]] to i{{[0-9]+}}*
 // t_var1_rhs = (i{{[0-9]+}}*)rhs[3];
-// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_LHS_REF]],
 // CHECK: [[T_VAR1_LHS:%.+]] = bitcast i8* [[T_VAR1_LHS_VOID]] to i{{[0-9]+}}*
 
diff --git a/test/OpenMP/for_reduction_messages.cpp b/test/OpenMP/for_reduction_messages.cpp
index 9c15e37989b2..317f88ce11ba 100644
--- a/test/OpenMP/for_reduction_messages.cpp
+++ b/test/OpenMP/for_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -ferror-limit 150 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -ferror-limit 150 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 150 -o - %s
 
 void foo() {
 }
@@ -54,6 +56,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable: no known conversion from 'int' to 'S6' for 1st argument}}
+#endif
   int a;
 
 public:
@@ -64,6 +69,8 @@ public:
 S3 h, k;
 #pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
 
+char *get();
+
 template <class T>       // expected-note {{declared here}}
 T tmain(T argc) {
   const T d = T();       // expected-note 4 {{'d' defined here}}
@@ -128,27 +135,27 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp for reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp for reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp for reduction(max : qa[1])
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp for reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp for reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp for reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -160,7 +167,7 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp for reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -184,7 +191,7 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp for reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -196,6 +203,14 @@ T tmain(T argc) {
 #pragma omp for reduction(+ : fl) // expected-error 2 {{reduction variable must be shared}}
   for (int i = 0; i < 10; ++i)
     foo();
+#pragma omp parallel private(qa)  // expected-note 2 {{defined as private}}
+#pragma omp for reduction(+ : qa[1], get()[0]) // expected-error 2 {{reduction variable must be shared}} expected-error {{expected variable name as a base of the array subscript}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp parallel shared(qa)
+#pragma omp for reduction(+ : qa[1], qa[0]) // expected-error 2 {{variable can appear only once in OpenMP 'reduction' clause}} expected-note 2 {{previously referenced here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
 #pragma omp parallel reduction(* : fl) // expected-note 2 {{defined as reduction}}
 #pragma omp for reduction(+ : fl)      // expected-error 2 {{reduction variable must be shared}}
   for (int i = 0; i < 10; ++i)
@@ -277,27 +292,27 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp for reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp for reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp for reduction(max : argv[1])
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp for reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp for reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp for reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -309,7 +324,7 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp for reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -341,7 +356,7 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp for reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -353,10 +368,22 @@ int main(int argc, char **argv) {
 #pragma omp for reduction(+ : fl) // expected-error {{reduction variable must be shared}}
   for (int i = 0; i < 10; ++i)
     foo();
+#pragma omp parallel private(argv)  // expected-note {{defined as private}}
+#pragma omp for reduction(+ : argv[1], get()[0]) // expected-error {{reduction variable must be shared}} expected-error {{expected variable name as a base of the array subscript}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp parallel shared(qa)
+#pragma omp for reduction(+ : qa[1], qa[0]) // expected-error {{variable can appear only once in OpenMP 'reduction' clause}} expected-note {{previously referenced here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
 #pragma omp parallel reduction(* : fl) // expected-note {{defined as reduction}}
 #pragma omp for reduction(+ : fl)      // expected-error {{reduction variable must be shared}}
   for (int i = 0; i < 10; ++i)
     foo();
+  static int m=0;
+#pragma omp for reduction(+:m)
+  for (int i = 0; i < 10; ++i)
+    m++;
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/for_schedule_messages.cpp b/test/OpenMP/for_schedule_messages.cpp
index c4490e88f02b..8946ce3aedfb 100644
--- a/test/OpenMP/for_schedule_messages.cpp
+++ b/test/OpenMP/for_schedule_messages.cpp
@@ -13,13 +13,23 @@ template <class T, typename S, int N, int ST> // expected-note {{declared here}}
 T tmain(T argc, S **argv) {
   #pragma omp for schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for schedule (monotonic // expected-error {{expected ')'}} expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-warning {{missing ':' after schedule modifier - ignoring}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic, // expected-error {{expected ')'}} expected-error {{expected 'simd' in OpenMP clause 'schedule'}} expected-warning {{missing ':' after schedule modifier - ignoring}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (simd: // expected-error {{expected ')'}} expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (monotonic, auto // expected-error {{expected ')'}} expected-warning {{missing ':' after schedule modifier - ignoring}} expected-error {{expected 'simd' in OpenMP clause 'schedule'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic: auto, // expected-error {{expected ')'}} expected-error {{'nonmonotonic' modifier can only be specified with 'dynamic' or 'guided' schedule kind}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -28,7 +38,7 @@ T tmain(T argc, S **argv) {
   // expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
   #pragma omp for schedule (guided argc
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp for schedule (static, ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for schedule (dynamic, 1)) // expected-warning {{extra tokens at the end of '#pragma omp for' are ignored}}
@@ -36,17 +46,35 @@ T tmain(T argc, S **argv) {
   #pragma omp for schedule (guided, (ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+2 2 {{directive '#pragma omp for' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp for schedule (static, foobool(argc)), schedule (dynamic, true), schedule (guided, -5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for schedule (static, S) // expected-error {{'S' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp for' are ignored}}
+  #pragma omp for schedule (static, S) // expected-error {{'S' does not refer to a value}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+1 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp for schedule (guided, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for schedule (dynamic, 1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for schedule (static, N) // expected-error {{argument to 'schedule' clause must be a positive integer value}}
+  #pragma omp for schedule (static, N) // expected-error {{argument to 'schedule' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic: static) // expected-error {{'nonmonotonic' modifier can only be specified with 'dynamic' or 'guided' schedule kind}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic: auto) // expected-error {{'nonmonotonic' modifier can only be specified with 'dynamic' or 'guided' schedule kind}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic: runtime) // expected-error {{'nonmonotonic' modifier can only be specified with 'dynamic' or 'guided' schedule kind}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (monotonic, nonmonotonic: auto) // expected-error {{modifier 'nonmonotonic' cannot be used along with modifier 'monotonic'}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic, monotonic: auto) // expected-error {{modifier 'monotonic' cannot be used along with modifier 'nonmonotonic'}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic, nonmonotonic: auto) // expected-error {{modifier 'nonmonotonic' cannot be used along with modifier 'nonmonotonic'}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (monotonic, monotonic: auto) // expected-error {{modifier 'monotonic' cannot be used along with modifier 'monotonic'}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for schedule (nonmonotonic: guided) ordered // expected-error {{'schedule' clause with 'nonmonotonic' modifier cannot be specified if an 'ordered' clause is specified}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for ordered(1) schedule(nonmonotonic: dynamic) // expected-error {{'schedule' clause with 'nonmonotonic' modifier cannot be specified if an 'ordered' clause is specified}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -54,13 +82,13 @@ T tmain(T argc, S **argv) {
 int main(int argc, char **argv) {
   #pragma omp for schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
@@ -73,17 +101,17 @@ int main(int argc, char **argv) {
   #pragma omp for schedule (dynamic, foobool(1) > 0 ? 1 : 2)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+2 2 {{directive '#pragma omp for' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp for schedule (guided, foobool(argc)), schedule (static, true), schedule (dynamic, -5)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for schedule (guided, S1) // expected-error {{'S1' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp for' are ignored}}
+  #pragma omp for schedule (guided, S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+1 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp for schedule (static, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{statement after '#pragma omp for' must be a for loop}}
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
-  #pragma omp for schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
   foo();
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
   return tmain<int, char, 1, 0>(argc, argv);
diff --git a/test/OpenMP/for_simd_aligned_messages.cpp b/test/OpenMP/for_simd_aligned_messages.cpp
index d41ecc1f28ee..1007b3c545cc 100644
--- a/test/OpenMP/for_simd_aligned_messages.cpp
+++ b/test/OpenMP/for_simd_aligned_messages.cpp
@@ -50,7 +50,7 @@ template<int L, class T, class N> T test_template(T* arr, N num) {
   T sum = (T)0;
   T ind2 = - num * L;
   // Negative number is passed as L.
-  // expected-error@+1 {{argument to 'aligned' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
   #pragma omp for simd aligned(arr:L)
   for (i = 0; i < num; ++i) {
     T cur = arr[(int)ind2];
@@ -65,7 +65,7 @@ template<int L, class T, class N> T test_template(T* arr, N num) {
 
 template<int LEN> int test_warn() {
   int *ind2 = 0;
-  // expected-error@+1 {{argument to 'aligned' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
   #pragma omp for simd aligned(ind2:LEN)
   for (int i = 0; i < 100; i++) {
     ind2 += LEN;
diff --git a/test/OpenMP/for_simd_ast_print.cpp b/test/OpenMP/for_simd_ast_print.cpp
index 21c15a2fd677..725c727e261e 100644
--- a/test/OpenMP/for_simd_ast_print.cpp
+++ b/test/OpenMP/for_simd_ast_print.cpp
@@ -35,8 +35,8 @@ template<class T> struct S {
 // CHECK: T res;
 // CHECK: T val;
 // CHECK: T lin = 0;
-    #pragma omp for simd private(val)  safelen(7) linear(lin : -5) lastprivate(res)
-// CHECK-NEXT: #pragma omp for simd private(val) safelen(7) linear(lin: -5) lastprivate(res)
+    #pragma omp for simd private(val)  safelen(7) linear(lin : -5) lastprivate(res) simdlen(5)
+// CHECK-NEXT: #pragma omp for simd private(val) safelen(7) linear(lin: -5) lastprivate(res) simdlen(5)
     for (T i = 7; i < m_a; ++i) {
       val = v[i-7] + m_a;
       res = val;
@@ -44,8 +44,8 @@ template<class T> struct S {
     }
     const T clen = 3;
 // CHECK: T clen = 3;
-    #pragma omp for simd safelen(clen-1)
-// CHECK-NEXT: #pragma omp for simd safelen(clen - 1)
+    #pragma omp for simd safelen(clen-1) simdlen(clen-1)
+// CHECK-NEXT: #pragma omp for simd safelen(clen - 1) simdlen(clen - 1)
     for(T i = clen+2; i < 20; ++i) {
 // CHECK-NEXT: for (T i = clen + 2; i < 20; ++i) {
       v[i] = v[v-clen] + 1;
@@ -62,7 +62,7 @@ template<class T> struct S {
 template<int LEN> struct S2 {
   static void func(int n, float *a, float *b, float *c) {
     int k1 = 0, k2 = 0;
-#pragma omp for simd safelen(LEN) linear(k1,k2:LEN) aligned(a:LEN)
+#pragma omp for simd safelen(LEN) linear(k1,k2:LEN) aligned(a:LEN) simdlen(LEN)
     for(int i = 0; i < n; i++) {
       c[i] = a[i] + b[i];
       c[k1] = a[k1] + b[k1];
@@ -77,7 +77,7 @@ template<int LEN> struct S2 {
 // CHECK: template <int LEN = 4> struct S2 {
 // CHECK-NEXT: static void func(int n, float *a, float *b, float *c)     {
 // CHECK-NEXT:   int k1 = 0, k2 = 0;
-// CHECK-NEXT: #pragma omp for simd safelen(4) linear(k1,k2: 4) aligned(a: 4)
+// CHECK-NEXT: #pragma omp for simd safelen(4) linear(k1,k2: 4) aligned(a: 4) simdlen(4)
 // CHECK-NEXT:   for (int i = 0; i < n; i++) {
 // CHECK-NEXT:     c[i] = a[i] + b[i];
 // CHECK-NEXT:     c[k1] = a[k1] + b[k1];
@@ -114,8 +114,8 @@ int main (int argc, char **argv) {
 // CHECK-NEXT: foo();
   const int CLEN = 4;
 // CHECK-NEXT: const int CLEN = 4;
-  #pragma omp for simd aligned(a:CLEN) linear(a:CLEN) safelen(CLEN) collapse( 1 )
-// CHECK-NEXT: #pragma omp for simd aligned(a: CLEN) linear(a: CLEN) safelen(CLEN) collapse(1)
+  #pragma omp for simd aligned(a:CLEN) linear(a:CLEN) safelen(CLEN) collapse( 1 ) simdlen(CLEN)
+// CHECK-NEXT: #pragma omp for simd aligned(a: CLEN) linear(a: CLEN) safelen(CLEN) collapse(1) simdlen(CLEN)
   for (int i = 0; i < 10; ++i)foo();
 // CHECK-NEXT: for (int i = 0; i < 10; ++i)
 // CHECK-NEXT: foo();
diff --git a/test/OpenMP/for_simd_codegen.cpp b/test/OpenMP/for_simd_codegen.cpp
index a5644013868c..cc8193d31803 100644
--- a/test/OpenMP/for_simd_codegen.cpp
+++ b/test/OpenMP/for_simd_codegen.cpp
@@ -1,7 +1,7 @@
-// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
-// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm -fexceptions -fcxx-exceptions -o - < %s | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t < %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify -emit-llvm -o - < %s | FileCheck %s
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm -o - < %s | FileCheck %s --check-prefix=TERM_DEBUG
 // REQUIRES: x86-registered-target
 // expected-no-diagnostics
 #ifndef HEADER
@@ -303,7 +303,7 @@ void simple(float *a, float *b, float *c, float *d) {
 // CHECK: [[A_PRIV_VAL:%.+]] = load i32, i32* [[A_PRIV]],
 // CHECK-NEXT: store i32 [[A_PRIV_VAL]], i32* %{{.+}},
 // CHECK-NEXT: br label
-// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t* {{.+}}, i32 %{{.+}})
+// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
   }
   int R;
   #pragma omp parallel
@@ -354,7 +354,7 @@ void simple(float *a, float *b, float *c, float *d) {
 // CHECK: [[RED:%.+]] = mul nsw i32 %{{.+}}, [[R_PRIV_VAL]]
 // CHECK-NEXT: store i32 [[RED]], i32* %{{.+}},
 // CHECK-NEXT: call void @__kmpc_end_reduce(
-// CHECK: call i32 @__kmpc_cancel_barrier(%ident_t* {{.+}}, i32 %{{.+}})
+// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
   }
 }
 
@@ -683,4 +683,3 @@ void parallel_simd(float *a) {
 }
 // TERM_DEBUG: !{{[0-9]+}} = !DILocation(line: [[@LINE-11]],
 #endif // HEADER
-
diff --git a/test/OpenMP/for_simd_collapse_messages.cpp b/test/OpenMP/for_simd_collapse_messages.cpp
index 7bb9b04cda7a..5c9d058b975c 100644
--- a/test/OpenMP/for_simd_collapse_messages.cpp
+++ b/test/OpenMP/for_simd_collapse_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp for simd collapse (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp for simd collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp for simd collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp for simd', but found only 1}}
   // expected-error@+3 2 {{directive '#pragma omp for simd' cannot contain more than one 'collapse' clause}}
-  // expected-error@+2 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp for simd collapse (foobool(argc)), collapse (true), collapse (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd collapse (1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd collapse (N) // expected-error {{argument to 'collapse' clause must be a positive integer value}}
+  #pragma omp for simd collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd collapse (2) // expected-note {{as specified in 'collapse' clause}}
   foo(); // expected-error {{expected 2 for loops after '#pragma omp for simd'}}
@@ -63,7 +63,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp for simd' cannot contain more than one 'collapse' clause}}
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp for simd collapse (foobool(argc)), collapse (true), collapse (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for simd collapse (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/for_simd_firstprivate_messages.cpp b/test/OpenMP/for_simd_firstprivate_messages.cpp
index 6b49b11eaff6..cb74ee081710 100644
--- a/test/OpenMP/for_simd_firstprivate_messages.cpp
+++ b/test/OpenMP/for_simd_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for simd firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (int k = 0; k < argc; ++k)
@@ -122,7 +122,7 @@ int foomain(int argc, char **argv) {
   {
     int v = 0;
     int i;                      // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for simd' directive into a parallel or another task region?}}
-#pragma omp for simd firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp for simd firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     for (int k = 0; k < argc; ++k) {
       i = k;
       v += i;
@@ -130,7 +130,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for simd firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp for simd firstprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel
@@ -168,7 +168,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for simd firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (i = 0; i < argc; ++i)
@@ -271,7 +271,7 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp for simd firstprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
@@ -286,7 +286,7 @@ int main(int argc, char **argv) {
   {
     int v = 0;
     int i;                      // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for simd' directive into a parallel or another task region?}}
-#pragma omp for simd firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp for simd firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     for (int k = 0; k < argc; ++k) {
       i = k;
       v += i;
@@ -300,6 +300,10 @@ int main(int argc, char **argv) {
 #pragma omp for simd firstprivate(i)       // expected-error {{firstprivate variable must be shared}}
   for (i = 0; i < argc; ++i)
     foo();
+  static int si;
+#pragma omp for simd firstprivate(si)
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/for_simd_lastprivate_messages.cpp b/test/OpenMP/for_simd_lastprivate_messages.cpp
index 2aa4dd12045c..d48c07d92b39 100644
--- a/test/OpenMP/for_simd_lastprivate_messages.cpp
+++ b/test/OpenMP/for_simd_lastprivate_messages.cpp
@@ -67,7 +67,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (int k = 0; k < argc; ++k)
@@ -132,7 +132,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for simd lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp for simd lastprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel
@@ -158,7 +158,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp for simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (i = 0; i < argc; ++i)
@@ -261,7 +261,7 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp for simd lastprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
@@ -272,5 +272,9 @@ int main(int argc, char **argv) {
 #pragma omp for simd lastprivate(n) firstprivate(n) // OK
   for (i = 0; i < argc; ++i)
     foo();
+  static int si;
+#pragma omp for simd lastprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/for_simd_linear_messages.cpp b/test/OpenMP/for_simd_linear_messages.cpp
index 189c8b0be88e..44370a15664f 100644
--- a/test/OpenMP/for_simd_linear_messages.cpp
+++ b/test/OpenMP/for_simd_linear_messages.cpp
@@ -102,7 +102,7 @@ template<class I, class C> int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp for simd linear // expected-error {{expected '(' after 'linear'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp for simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -138,7 +138,7 @@ template<class I, class C> int foomain(I argc, C **argv) {
     #pragma omp for simd linear(v:i)
     for (int k = 0; k < argc; ++k) { i = k; v += i; }
   }
-  #pragma omp for simd linear(j) // expected-error {{arguments of OpenMP clause 'linear' cannot be of reference type}}
+  #pragma omp for simd linear(j)
   for (int k = 0; k < argc; ++k) ++k;
   int v = 0;
   #pragma omp for simd linear(v:j)
@@ -166,7 +166,7 @@ int main(int argc, char **argv) {
   S4 e(4); // expected-note {{'e' defined here}}
   S5 g(5); // expected-note {{'g' defined here}}
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp for simd linear // expected-error {{expected '(' after 'linear'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp for simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -203,7 +203,7 @@ int main(int argc, char **argv) {
     #pragma omp for simd linear(i : 4)
     for (int k = 0; k < argc; ++k) { ++k; i += 4; }
   }
-  #pragma omp for simd linear(j) // expected-error {{arguments of OpenMP clause 'linear' cannot be of reference type 'int &'}}
+  #pragma omp for simd linear(j)
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp for simd linear(i)
   for (int k = 0; k < argc; ++k) ++k;
diff --git a/test/OpenMP/for_simd_loop_messages.cpp b/test/OpenMP/for_simd_loop_messages.cpp
index da7e6843a0c4..958c7f60d0eb 100644
--- a/test/OpenMP/for_simd_loop_messages.cpp
+++ b/test/OpenMP/for_simd_loop_messages.cpp
@@ -10,6 +10,7 @@ public:
 };
 
 static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
 #pragma omp threadprivate(sii)
 static int globalii;
 
@@ -63,37 +64,37 @@ int test_iteration_spaces() {
     c[(int)fi] = a[(int)fi] + b[(int)fi];
   }
 #pragma omp parallel
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (int &ref = ii; ref < 10; ref++) {
   }
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (int i; i < 10; i++)
     c[i] = a[i];
 
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (int i = 0, j = 0; i < 10; ++i)
     c[i] = a[i];
 
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (; ii < 10; ++ii)
     c[ii] = a[ii];
 
 #pragma omp parallel
 // expected-warning@+3 {{expression result unused}}
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (ii + 1; ii < 10; ++ii)
     c[ii] = a[ii];
 
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (c[ii] = 0; ii < 10; ++ii)
     c[ii] = a[ii];
@@ -306,6 +307,7 @@ int test_iteration_spaces() {
 
 #pragma omp parallel
   {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp for simd' directive may not be threadprivate or thread local, predetermined as linear}}
 #pragma omp for simd
     for (sii = 0; sii < 10; sii += 1)
       c[sii] = a[sii];
@@ -430,7 +432,7 @@ int test_with_random_access_iterator() {
   for (GoodIter I = begin; I < end; ++I)
     ++I;
 #pragma omp parallel
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (GoodIter &I = begin; I < end; ++I)
     ++I;
@@ -469,7 +471,7 @@ int test_with_random_access_iterator() {
   for (begin = begin0; begin < end; ++begin)
     ++begin;
 #pragma omp parallel
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp for simd
   for (++begin; begin < end; ++begin)
     ++begin;
diff --git a/test/OpenMP/for_simd_misc_messages.c b/test/OpenMP/for_simd_misc_messages.c
index e87a78976e2c..ca1e366ca092 100644
--- a/test/OpenMP/for_simd_misc_messages.c
+++ b/test/OpenMP/for_simd_misc_messages.c
@@ -153,20 +153,117 @@ void test_safelen() {
 #pragma omp for simd safelen(foo())
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp for simd safelen(-5)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp for simd safelen(0)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp for simd safelen(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
 }
 
+void test_simdlen() {
+  int i;
+// expected-error@+1 {{expected '('}}
+#pragma omp for simd simdlen
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}}
+#pragma omp for simd simdlen()
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(,
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}}  expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(, )
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp for simd' are ignored}}
+// expected-error@+1 {{expected '('}}
+#pragma omp for simd simdlen 4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(4
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(4,
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(4, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp for simd simdlen(4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(4 4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(4, , 4)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp for simd simdlen(4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp for simd simdlen(4, 8)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp for simd simdlen(2.5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp for simd simdlen(foo())
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp for simd simdlen(-5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp for simd simdlen(0)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp for simd simdlen(5 - 5)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_safelen_simdlen() {
+  int i;
+// expected-error@+1 {{the value of 'simdlen' parameter must be less than or equal to the value of the 'safelen' parameter}}
+#pragma omp for simd simdlen(6) safelen(5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{the value of 'simdlen' parameter must be less than or equal to the value of the 'safelen' parameter}}
+#pragma omp for simd safelen(5) simdlen(6)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
 void test_collapse() {
   int i;
 #pragma omp parallel
@@ -259,17 +356,17 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp for simd collapse(-5)
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp for simd collapse(0)
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp for simd collapse(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
@@ -278,7 +375,7 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
 // expected-note@+1 {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for simd' directive into a parallel or another task region?}}
     for (int j = 0; j < 16; ++j)
-// expected-error@+2 {{private variable cannot be reduction}}
+// expected-error@+2 {{reduction variable must be shared}}
 // expected-error@+1 {{OpenMP constructs may not be nested inside a simd region}}
 #pragma omp for simd reduction(+ : i, j)
       for (int k = 0; k < 16; ++k)
diff --git a/test/OpenMP/for_simd_private_messages.cpp b/test/OpenMP/for_simd_private_messages.cpp
index 96885c859e99..15e235c2f8f4 100644
--- a/test/OpenMP/for_simd_private_messages.cpp
+++ b/test/OpenMP/for_simd_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp for simd private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -99,7 +99,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for simd private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp for simd private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp for simd private(i)
@@ -120,7 +120,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp for simd private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -169,12 +169,16 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp for simd private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp for simd private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp for simd private(i)
   for (int k = 0; k < argc; ++k)
     ++k;
+  static int m;
+#pragma omp for simd private(m)
+  for (int k = 0; k < argc; ++k)
+    m = k + 2;
 
   return 0;
 }
diff --git a/test/OpenMP/for_simd_reduction_messages.cpp b/test/OpenMP/for_simd_reduction_messages.cpp
index 74952776d508..000960fb55b5 100644
--- a/test/OpenMP/for_simd_reduction_messages.cpp
+++ b/test/OpenMP/for_simd_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 %s
 
 void foo() {
 }
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -112,7 +118,7 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp for simd reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -128,27 +134,27 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp for simd reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp for simd reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp for simd reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp for simd reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp for simd reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp for simd reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -160,7 +166,7 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -184,7 +190,7 @@ T tmain(T argc) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp for simd reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -261,7 +267,7 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp for simd reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -277,27 +283,27 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp for simd reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp for simd reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp for simd reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp for simd reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp for simd reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp for simd reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -309,7 +315,7 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
@@ -337,7 +343,7 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel
-#pragma omp for simd reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp for simd reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -353,6 +359,10 @@ int main(int argc, char **argv) {
 #pragma omp for simd reduction(+ : fl)      // expected-error {{reduction variable must be shared}}
   for (int i = 0; i < 10; ++i)
     foo();
+  static int m;
+#pragma omp for simd reduction(+ : m)
+  for (int i = 0; i < 10; ++i)
+    m++;
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/for_simd_safelen_messages.cpp b/test/OpenMP/for_simd_safelen_messages.cpp
index 27a87b559bc9..d70e90198ad6 100644
--- a/test/OpenMP/for_simd_safelen_messages.cpp
+++ b/test/OpenMP/for_simd_safelen_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp for simd safelen (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
   #pragma omp for simd safelen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd safelen (1)) // expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp for simd safelen ((ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+3 2 {{directive '#pragma omp for simd' cannot contain more than one 'safelen' clause}}
-  // expected-error@+2 2 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp for simd safelen (foobool(argc)), safelen (true), safelen (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd safelen (4)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd safelen (N) // expected-error {{argument to 'safelen' clause must be a positive integer value}}
+  #pragma omp for simd safelen (N) // expected-error {{argument to 'safelen' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -61,7 +61,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp for simd' cannot contain more than one 'safelen' clause}}
-  // expected-error@+1 2 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
   #pragma omp for simd safelen (foobool(argc)), safelen (true), safelen (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for simd safelen (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/for_simd_schedule_messages.cpp b/test/OpenMP/for_simd_schedule_messages.cpp
index 1367676ad58c..76ba3a42eda2 100644
--- a/test/OpenMP/for_simd_schedule_messages.cpp
+++ b/test/OpenMP/for_simd_schedule_messages.cpp
@@ -13,13 +13,13 @@ template <class T, typename S, int N, int ST> // expected-note {{declared here}}
 T tmain(T argc, S **argv) {
   #pragma omp for simd schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -28,7 +28,7 @@ T tmain(T argc, S **argv) {
   // expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
   #pragma omp for simd schedule (guided argc
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp for simd schedule (static, ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd schedule (dynamic, 1)) // expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
@@ -36,17 +36,17 @@ T tmain(T argc, S **argv) {
   #pragma omp for simd schedule (guided, (ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+2 2 {{directive '#pragma omp for simd' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp for simd schedule (static, foobool(argc)), schedule (dynamic, true), schedule (guided, -5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd schedule (static, S) // expected-error {{'S' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
+  #pragma omp for simd schedule (static, S) // expected-error {{'S' does not refer to a value}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+1 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp for simd schedule (guided, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp for simd schedule (dynamic, 1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp for simd schedule (static, N) // expected-error {{argument to 'schedule' clause must be a positive integer value}}
+  #pragma omp for simd schedule (static, N) // expected-error {{argument to 'schedule' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -54,13 +54,13 @@ T tmain(T argc, S **argv) {
 int main(int argc, char **argv) {
   #pragma omp for simd schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for simd schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp for simd schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
@@ -73,17 +73,17 @@ int main(int argc, char **argv) {
   #pragma omp for simd schedule (dynamic, foobool(1) > 0 ? 1 : 2)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+2 2 {{directive '#pragma omp for simd' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp for simd schedule (guided, foobool(argc)), schedule (static, true), schedule (dynamic, -5)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp for simd schedule (guided, S1) // expected-error {{'S1' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
+  #pragma omp for simd schedule (guided, S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+1 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp for simd schedule (static, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{statement after '#pragma omp for simd' must be a for loop}}
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
-  #pragma omp for simd schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp for simd schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
   foo();
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
   return tmain<int, char, 1, 0>(argc, argv);
diff --git a/test/OpenMP/for_simd_simdlen_messages.cpp b/test/OpenMP/for_simd_simdlen_messages.cpp
new file mode 100644
index 000000000000..c72e54681192
--- /dev/null
+++ b/test/OpenMP/for_simd_simdlen_messages.cpp
@@ -0,0 +1,79 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp for simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp for simd simdlen (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp for simd simdlen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen (1)) // expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen ((ST > 0) ? 1 + ST : 2)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 2 {{directive '#pragma omp for simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+2 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp for simd simdlen (foobool(argc)), simdlen (true), simdlen (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp for simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen (4)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp for simd simdlen (N) // expected-error {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp for simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp for simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp for simd simdlen () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp for simd simdlen (4 // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp for simd simdlen (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp for simd' are ignored}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp for simd simdlen (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp for simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+1 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp for simd simdlen (foobool(argc)), simdlen (true), simdlen (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp for simd simdlen (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp for simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp for simd' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp for simd simdlen(simdlen(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 12, 4>' requested here}}
+  return tmain<int, char, 12, 4>(argc, argv);
+}
+
diff --git a/test/OpenMP/function-attr.cpp b/test/OpenMP/function-attr.cpp
new file mode 100644
index 000000000000..87aba40f2b7d
--- /dev/null
+++ b/test/OpenMP/function-attr.cpp
@@ -0,0 +1,54 @@
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10  -stack-protector 2 -emit-llvm -o - %s | FileCheck %s
+
+// Check that function attributes are added to the OpenMP runtime functions.
+
+template <class T>
+struct S {
+  T f;
+  S(T a) : f(a) {}
+  S() : f() {}
+  operator T() { return T(); }
+  ~S() {}
+};
+
+// CHECK: define internal void @.omp.copyprivate.copy_func(i8*, i8*) [[ATTR0:#[0-9]+]] {
+
+void foo0();
+
+int foo1() {
+  char a;
+
+#pragma omp parallel
+  a = 2;
+#pragma omp single copyprivate(a)
+  foo0();
+
+  return 0;
+}
+
+// CHECK: define internal void @.omp_task_privates_map.({{.*}}) [[ATTR3:#[0-9]+]] {
+// CHECK: define internal i32 @.omp_task_entry.({{.*}}) [[ATTR0]] {
+// CHECK: define internal i32 @.omp_task_destructor.({{.*}}) [[ATTR0]] {
+
+int foo2() {
+  S<double> s_arr[] = {1, 2};
+  S<double> var(3);
+#pragma omp task private(s_arr, var)
+  s_arr[0] = var;
+  return 0;
+}
+
+// CHECK: define internal void @.omp.reduction.reduction_func(i8*, i8*) [[ATTR0]] {
+
+float foo3(int n, float *a, float *b) {
+  int i;
+  float result;
+
+#pragma omp parallel for private(i) reduction(+:result)
+  for (i=0; i < n; i++)
+    result = result + (a[i] * b[i]);
+  return result;
+}
+
+// CHECK: attributes [[ATTR0]] = {{{.*}} sspstrong {{.*}}}
+// CHECK: attributes [[ATTR3]] = {{{.*}} sspstrong {{.*}}}
diff --git a/test/OpenMP/linking.c b/test/OpenMP/linking.c
index 778216d723b8..81706d4f6201 100644
--- a/test/OpenMP/linking.c
+++ b/test/OpenMP/linking.c
@@ -69,3 +69,21 @@
 // CHECK-LD-OVERRIDE-64: "-lgomp" "-lrt" "-lgcc"
 // CHECK-LD-OVERRIDE-64: "-lpthread" "-lc"
 //
+// RUN: %clang -no-canonical-prefixes %s -### -o %t.o 2>&1 \
+// RUN:     -fopenmp=libomp -target x86_64-msvc-win32 \
+// RUN:   | FileCheck --check-prefix=CHECK-MSVC-LINK-64 %s
+// CHECK-MSVC-LINK-64: link.exe
+// CHECK-MSVC-LINK-64-SAME: -nodefaultlib:vcomp.lib
+// CHECK-MSVC-LINK-64-SAME: -nodefaultlib:vcompd.lib
+// CHECK-MSVC-LINK-64-SAME: -libpath:{{.+}}/../lib
+// CHECK-MSVC-LINK-64-SAME: -defaultlib:libomp.lib
+//
+// RUN: %clang -no-canonical-prefixes %s -### -o %t.o 2>&1 \
+// RUN:     -fopenmp=libiomp5 -target x86_64-msvc-win32 \
+// RUN:   | FileCheck --check-prefix=CHECK-MSVC-ILINK-64 %s
+// CHECK-MSVC-ILINK-64: link.exe
+// CHECK-MSVC-ILINK-64-SAME: -nodefaultlib:vcomp.lib
+// CHECK-MSVC-ILINK-64-SAME: -nodefaultlib:vcompd.lib
+// CHECK-MSVC-ILINK-64-SAME: -libpath:{{.+}}/../lib
+// CHECK-MSVC-ILINK-64-SAME: -defaultlib:libiomp5md.lib
+//
diff --git a/test/OpenMP/master_codegen.cpp b/test/OpenMP/master_codegen.cpp
index c364c5278baf..b491fb921c5c 100644
--- a/test/OpenMP/master_codegen.cpp
+++ b/test/OpenMP/master_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
 #ifndef HEADER
diff --git a/test/OpenMP/nesting_of_regions.cpp b/test/OpenMP/nesting_of_regions.cpp
index 8a3bacf0f115..6445c6b1c6d9 100644
--- a/test/OpenMP/nesting_of_regions.cpp
+++ b/test/OpenMP/nesting_of_regions.cpp
@@ -100,6 +100,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'parallel' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp parallel
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'parallel' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // SIMD DIRECTIVE
 #pragma omp simd
@@ -227,6 +239,18 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside a simd region}}
     ++a;
   }
+#pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // FOR DIRECTIVE
 #pragma omp for
@@ -377,6 +401,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'for' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'for' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // FOR SIMD DIRECTIVE
 #pragma omp for simd
@@ -504,6 +540,18 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside a simd region}}
     ++a;
   }
+#pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // SECTIONS DIRECTIVE
 #pragma omp sections
@@ -661,6 +709,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'sections' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp sections
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp sections
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'sections' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // SECTION DIRECTIVE
 #pragma omp section // expected-error {{orphaned 'omp section' directives are prohibited, it must be closely nested to a sections region}}
@@ -854,6 +914,20 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'section' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp sections
+  {
+#pragma omp section
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp sections
+  {
+#pragma omp section
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'section' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // SINGLE DIRECTIVE
 #pragma omp single
@@ -994,6 +1068,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'single' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp single
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp single
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'single' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // MASTER DIRECTIVE
 #pragma omp master
@@ -1134,6 +1220,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'master' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp master
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp master
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'master' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // CRITICAL DIRECTIVE
 #pragma omp critical
@@ -1288,6 +1386,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'critical' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp critical
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp critical
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'critical' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // PARALLEL FOR DIRECTIVE
 #pragma omp parallel for
@@ -1443,6 +1553,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'parallel for' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp parallel for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'parallel for' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // PARALLEL FOR SIMD DIRECTIVE
 #pragma omp parallel for simd
@@ -1598,6 +1720,18 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside a simd region}}
     ++a;
   }
+#pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // PARALLEL SECTIONS DIRECTIVE
 #pragma omp parallel sections
@@ -1744,6 +1878,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'parallel sections' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp parallel sections
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel sections
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'parallel sections' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // TASK DIRECTIVE
 #pragma omp task
@@ -1836,6 +1982,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'task' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp task
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp task
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'task' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // ORDERED DIRECTIVE
 #pragma omp ordered
@@ -1976,6 +2134,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'ordered' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp ordered
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp ordered
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'ordered' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // ATOMIC DIRECTIVE
 #pragma omp atomic
@@ -2145,6 +2315,22 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside an atomic region}}
     ++a;
   }
+#pragma omp atomic
+  // expected-error@+2 {{the statement for 'atomic' must be an expression statement of form '++x;', '--x;', 'x++;', 'x--;', 'x binop= expr;', 'x = x binop expr' or 'x = expr binop x', where x is an l-value expression with scalar type}}
+  // expected-note@+1 {{expected an expression statement}}
+  {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside an atomic region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp atomic
+  // expected-error@+2 {{the statement for 'atomic' must be an expression statement of form '++x;', '--x;', 'x++;', 'x--;', 'x binop= expr;', 'x = x binop expr' or 'x = expr binop x', where x is an l-value expression with scalar type}}
+  // expected-note@+1 {{expected an expression statement}}
+  {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside an atomic region}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // TARGET DIRECTIVE
 #pragma omp target
@@ -2250,6 +2436,18 @@ void foo() {
 #pragma omp teams  // expected-note {{nested teams construct here}}
     ++a;
   }
+#pragma omp target
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp target
+  { 
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'target' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // TEAMS DIRECTIVE
 #pragma omp target
@@ -2370,6 +2568,365 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'teams' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp target
+#pragma omp teams
+  {
+#pragma omp taskloop // expected-error {{region cannot be closely nested inside 'teams' region; perhaps you forget to enclose 'omp taskloop' directive into a parallel region?}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp distribute
+  for (int j = 0; j < 10; ++j)
+    ;        
+
+// TASKLOOP DIRECTIVE
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp for' directive into a parallel region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for simd // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp for simd' directive into a parallel region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp sections // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp sections' directive into a parallel region?}}
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp section // expected-error {{'omp section' directive must be closely nested to a sections region, not a taskloop region}}
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp single // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp single' directive into a parallel region?}}
+    {
+      bar();
+    }
+  }
+
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp master // expected-error {{region cannot be closely nested inside 'taskloop' region}}
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp critical
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    {
+#pragma omp single // OK
+      {
+        bar();
+      }
+#pragma omp for // OK
+      for (int i = 0; i < 10; ++i)
+        ;
+#pragma omp sections // OK
+      {
+        bar();
+      }
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel sections
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp task
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskyield
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp barrier // expected-error {{region cannot be closely nested inside 'taskloop' region}}
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskwait
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp flush
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp ordered // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp ordered' directive into a for or a parallel for region with 'ordered' clause?}}
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp atomic
+    ++a;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp target
+    ++a;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp teams // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
+    ++a;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+// DISTRIBUTE DIRECTIVE
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'distribute' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams  
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp sections
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp section // expected-error {{'omp section' directive must be closely nested to a sections region, not a distribute region}}
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp single
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp master
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp critical
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    {
+#pragma omp single
+      {
+	bar();
+      }
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel sections
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp task
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskyield
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp barrier
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskwait
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp flush
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp ordered // expected-error {{region cannot be closely nested inside 'distribute' region; perhaps you forget to enclose 'omp ordered' directive into a for or a parallel for region with 'ordered' clause?}}
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp atomic
+    ++a;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp target
+    ++a;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp teams // expected-error {{region cannot be closely nested inside 'distribute' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
+    ++a;
+  }
 }
 
 void foo() {
@@ -2469,6 +3026,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'parallel' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp parallel
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'parallel' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // SIMD DIRECTIVE
 #pragma omp simd
@@ -2589,6 +3158,18 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside a simd region}}
     ++a;
   }
+#pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // FOR DIRECTIVE
 #pragma omp for
@@ -2729,6 +3310,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'for' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'for' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // FOR SIMD DIRECTIVE
 #pragma omp for simd
@@ -2849,6 +3442,18 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside a simd region}}
     ++a;
   }
+#pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // SECTIONS DIRECTIVE
 #pragma omp sections
@@ -2981,6 +3586,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'sections' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp sections
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp sections
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'sections' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // SECTION DIRECTIVE
 #pragma omp section // expected-error {{orphaned 'omp section' directives are prohibited, it must be closely nested to a sections region}}
@@ -3180,6 +3797,22 @@ void foo() {
       ++a;
     }
   }
+#pragma omp sections
+  {
+#pragma omp section
+    {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+      ++a;
+    }
+  }
+#pragma omp sections
+  {
+#pragma omp section
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'section' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // SINGLE DIRECTIVE
 #pragma omp single
@@ -3310,6 +3943,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'single' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp single
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp single
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'single' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // MASTER DIRECTIVE
 #pragma omp master
@@ -3450,6 +4095,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'master' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp master
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp master
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'master' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // CRITICAL DIRECTIVE
 #pragma omp critical
@@ -3609,6 +4266,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'critical' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp critical
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp critical
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'critical' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // PARALLEL FOR DIRECTIVE
 #pragma omp parallel for
@@ -3764,6 +4433,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'parallel for' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp parallel for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel for
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'parallel for' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // PARALLEL FOR SIMD DIRECTIVE
 #pragma omp parallel for simd
@@ -3919,6 +4600,18 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside a simd region}}
     ++a;
   }
+#pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    for (int j = 0; j < 10; ++j)
+      ;
+  }
 
 // PARALLEL SECTIONS DIRECTIVE
 #pragma omp parallel sections
@@ -4061,6 +4754,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'parallel sections' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp parallel sections
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp parallel sections
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'parallel sections' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // TASK DIRECTIVE
 #pragma omp task
@@ -4152,6 +4857,18 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'task' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp task
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp task
+  {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'task' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // ATOMIC DIRECTIVE
 #pragma omp atomic
@@ -4321,6 +5038,22 @@ void foo() {
 #pragma omp teams // expected-error {{OpenMP constructs may not be nested inside an atomic region}}
     ++a;
   }
+#pragma omp atomic
+  // expected-error@+2 {{the statement for 'atomic' must be an expression statement of form '++x;', '--x;', 'x++;', 'x--;', 'x binop= expr;', 'x = x binop expr' or 'x = expr binop x', where x is an l-value expression with scalar type}}
+  // expected-note@+1 {{expected an expression statement}}
+  {
+#pragma omp taskloop // expected-error {{OpenMP constructs may not be nested inside an atomic region}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp atomic
+  // expected-error@+2 {{the statement for 'atomic' must be an expression statement of form '++x;', '--x;', 'x++;', 'x--;', 'x binop= expr;', 'x = x binop expr' or 'x = expr binop x', where x is an l-value expression with scalar type}}
+  // expected-note@+1 {{expected an expression statement}}
+  {
+#pragma omp distribute // expected-error {{OpenMP constructs may not be nested inside an atomic region}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // TARGET DIRECTIVE
 #pragma omp target
@@ -4426,6 +5159,18 @@ void foo() {
 #pragma omp teams // expected-note {{nested teams construct here}}
     ++a;
   }
+#pragma omp target
+  {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp target
+  { 
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'target' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
 
 // TEAMS DIRECTIVE
 #pragma omp target
@@ -4546,6 +5291,366 @@ void foo() {
 #pragma omp teams // expected-error {{region cannot be closely nested inside 'teams' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
     ++a;
   }
+#pragma omp target
+#pragma omp teams
+  {
+#pragma omp taskloop // expected-error {{region cannot be closely nested inside 'teams' region; perhaps you forget to enclose 'omp taskloop' directive into a parallel region?}}
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp distribute
+  for (int j = 0; j < 10; ++j)
+    ;
+
+// TASKLOOP DIRECTIVE
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp for' directive into a parallel region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for simd // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp for simd' directive into a parallel region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp sections // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp sections' directive into a parallel region?}}
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp section // expected-error {{'omp section' directive must be closely nested to a sections region, not a taskloop region}}
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp single // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp single' directive into a parallel region?}}
+    {
+      bar();
+    }
+  }
+
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp master // expected-error {{region cannot be closely nested inside 'taskloop' region}}
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp critical
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    {
+#pragma omp single // OK
+      {
+        bar();
+      }
+#pragma omp for // OK
+      for (int i = 0; i < 10; ++i)
+        ;
+#pragma omp sections // OK
+      {
+        bar();
+      }
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel sections
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp task
+    {
+      bar();
+    }
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskyield
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp barrier // expected-error {{region cannot be closely nested inside 'taskloop' region}}
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskwait
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp flush
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp ordered // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp ordered' directive into a for or a parallel for region with 'ordered' clause?}}
+    bar();
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp atomic
+    ++a;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp target
+    ++a;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp teams // expected-error {{region cannot be closely nested inside 'taskloop' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
+    ++a;
+  }
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    ++a;
+  }
+
+// DISTRIBUTE DIRECTIVE
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp distribute // expected-error {{region cannot be closely nested inside 'distribute' region; perhaps you forget to enclose 'omp distribute' directive into a teams region?}}
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp for simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams  
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp sections
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp section // expected-error {{'omp section' directive must be closely nested to a sections region, not a distribute region}}
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp single
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp master
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp critical
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel
+    {
+#pragma omp single
+      {
+	bar();
+      }
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel for simd
+    for (int i = 0; i < 10; ++i)
+      ;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp parallel sections
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp task
+    {
+      bar();
+    }
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskyield
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp barrier
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp taskwait
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp flush
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp ordered // expected-error {{region cannot be closely nested inside 'distribute' region; perhaps you forget to enclose 'omp ordered' directive into a for or a parallel for region with 'ordered' clause?}}
+    bar();
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp atomic
+    ++a;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp target
+    ++a;
+  }
+#pragma omp target
+#pragma omp teams
+#pragma omp distribute
+  for (int i = 0; i < 10; ++i) {
+#pragma omp teams // expected-error {{region cannot be closely nested inside 'distribute' region; perhaps you forget to enclose 'omp teams' directive into a target region?}}
+    ++a;
+  }
   return foo<int>();
 }
 
diff --git a/test/OpenMP/openmp_check.cpp b/test/OpenMP/openmp_check.cpp
new file mode 100644
index 000000000000..c9b5eb0b9cba
--- /dev/null
+++ b/test/OpenMP/openmp_check.cpp
@@ -0,0 +1,15 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+int nested(int a) {
+#pragma omp parallel
+  ++a;
+
+  auto F = [&]() { // expected-error {{expected expression}} expected-error {{expected ';' at end of declaration}} expected-warning {{'auto' type specifier is a C++11 extension}}
+#pragma omp parallel
+    {
+#pragma omp target
+      ++a;
+    }
+  };
+  F(); // expected-error {{C++ requires a type specifier for all declarations}}
+  return a; // expected-error {{expected unqualified-id}}
+}// expected-error {{extraneous closing brace ('}')}}
diff --git a/test/OpenMP/ordered_ast_print.cpp b/test/OpenMP/ordered_ast_print.cpp
index 0006080b2082..97fe7007e262 100644
--- a/test/OpenMP/ordered_ast_print.cpp
+++ b/test/OpenMP/ordered_ast_print.cpp
@@ -8,7 +8,7 @@
 
 void foo() {}
 
-template <class T>
+template <class T, int N>
 T tmain (T argc) {
   T b = argc, c, d, e, f, g;
   static T a;
@@ -18,6 +18,36 @@ T tmain (T argc) {
   {
     a=2;
   }
+  #pragma omp for ordered
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered threads
+  {
+    a=2;
+  }
+  #pragma omp simd
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered simd
+  {
+    a=2;
+  }
+  #pragma omp for simd
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered simd
+  {
+    a=2;
+  }
+  #pragma omp parallel for simd
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered simd
+  {
+    a=2;
+  }
+  #pragma omp parallel for ordered(1)
+  for (int i =0 ; i < argc; ++i) {
+  #pragma omp ordered depend(source)
+  #pragma omp ordered depend(sink:i+N)
+    a = 2;
+  }
   return (0);
 }
 
@@ -28,7 +58,36 @@ T tmain (T argc) {
 // CHECK-NEXT: {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: }
-
+// CHECK-NEXT: #pragma omp for ordered
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered threads
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp for simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp parallel for simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp parallel for ordered(1)
+// CHECK-NEXT: for (int i = 0; i < argc; ++i) {
+// CHECK-NEXT: #pragma omp ordered depend(source)
+// CHECK-NEXT: #pragma omp ordered depend(sink : i + 3)
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
 // CHECK: static T a;
 // CHECK-NEXT: #pragma omp for ordered
 // CHECK-NEXT: for (int i = 0; i < argc; ++i)
@@ -36,7 +95,38 @@ T tmain (T argc) {
 // CHECK-NEXT: {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp for ordered
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered threads
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp for simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp parallel for simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp parallel for ordered(1)
+// CHECK-NEXT: for (int i = 0; i < argc; ++i) {
+// CHECK-NEXT: #pragma omp ordered depend(source)
+// CHECK-NEXT: #pragma omp ordered depend(sink : i + N)
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
 
+// CHECK-LABEL: int main(
 int main (int argc, char **argv) {
   int b = argc, c, d, e, f, g;
   static int a;
@@ -47,13 +137,73 @@ int main (int argc, char **argv) {
   {
     a=2;
   }
+  #pragma omp for ordered
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered threads
+  {
+    a=2;
+  }
+  #pragma omp simd
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered simd
+  {
+    a=2;
+  }
+  #pragma omp for simd
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered simd
+  {
+    a=2;
+  }
+  #pragma omp parallel for simd
+  for (int i =0 ; i < argc; ++i)
+  #pragma omp ordered simd
+  {
+    a=2;
+  }
+  #pragma omp parallel for ordered(1)
+  for (int i =0 ; i < argc; ++i) {
+  #pragma omp ordered depend(source)
+  #pragma omp ordered depend(sink: i - 5)
+    a = 2;
+  }
 // CHECK-NEXT: #pragma omp for ordered
 // CHECK-NEXT: for (int i = 0; i < argc; ++i)
 // CHECK-NEXT: #pragma omp ordered
 // CHECK-NEXT: {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: }
-  return tmain(argc);
+// CHECK-NEXT: #pragma omp for ordered
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered threads
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp for simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp parallel for simd
+// CHECK-NEXT: for (int i = 0; i < argc; ++i)
+// CHECK-NEXT: #pragma omp ordered simd
+// CHECK-NEXT: {
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+// CHECK-NEXT: #pragma omp parallel for ordered(1)
+// CHECK-NEXT: for (int i = 0; i < argc; ++i) {
+// CHECK-NEXT: #pragma omp ordered depend(source)
+// CHECK-NEXT: #pragma omp ordered depend(sink : i - 5)
+// CHECK-NEXT: a = 2;
+// CHECK-NEXT: }
+  return tmain<int, 3>(argc);
 }
 
 #endif
diff --git a/test/OpenMP/ordered_codegen.cpp b/test/OpenMP/ordered_codegen.cpp
index ff8a8047cae4..e77c1bed97ca 100644
--- a/test/OpenMP/ordered_codegen.cpp
+++ b/test/OpenMP/ordered_codegen.cpp
@@ -92,7 +92,7 @@ void dynamic1(float *a, float *b, float *c, float *d) {
 // CHECK-NOT: !llvm.mem.parallel_loop_access
 // CHECK-NEXT: call void @__kmpc_end_ordered([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]])
 // ... end of ordered region ...
-    #pragma omp ordered
+    #pragma omp ordered threads
     a[i] = b[i] * c[i] * d[i];
 // CHECK: [[IV1_2:%.+]] = load i64, i64* [[OMP_IV]]{{.*}}
 // CHECK-NEXT: [[ADD1_2:%.+]] = add i64 [[IV1_2]], 1
@@ -197,7 +197,7 @@ void runtime(float *a, float *b, float *c, float *d) {
 // CHECK-NOT: !llvm.mem.parallel_loop_access
 // CHECK-NEXT: call void @__kmpc_end_ordered([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]])
 // ... end of ordered region ...
-    #pragma omp ordered
+    #pragma omp ordered threads
     a[i] = b[i] * c[i] * d[i];
 // CHECK: [[IV1_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}
 // CHECK-NEXT: [[ADD1_2:%.+]] = add nsw i32 [[IV1_2]], 1
@@ -213,5 +213,22 @@ void runtime(float *a, float *b, float *c, float *d) {
 // CHECK: ret void
 }
 
+float f[10];
+// CHECK-LABEL: foo_simd
+void foo_simd(int low, int up) {
+  // CHECK: store float 0.000000e+00, float* %{{.+}}, align {{[0-9]+}}, !llvm.mem.parallel_loop_access !
+  // CHECK-NEXT: call void [[CAP_FUNC:@.+]](i32* %{{.+}}) #{{[0-9]+}}, !llvm.mem.parallel_loop_access !
+#pragma omp simd
+  for (int i = low; i < up; ++i) {
+    f[i] = 0.0;
+#pragma omp ordered simd
+    f[i] = 1.0;
+  }
+}
+
+// CHECK: define internal void [[CAP_FUNC]](i32* dereferenceable({{[0-9]+}}) %{{.+}}) #
+// CHECK: store float 1.000000e+00, float* %{{.+}}, align
+// CHECK-NEXT: ret void
+
 #endif // HEADER
 
diff --git a/test/OpenMP/ordered_messages.cpp b/test/OpenMP/ordered_messages.cpp
index 72e59b7c7646..36f9bb2c7777 100644
--- a/test/OpenMP/ordered_messages.cpp
+++ b/test/OpenMP/ordered_messages.cpp
@@ -4,6 +4,7 @@ int foo();
 
 template <class T>
 T foo() {
+ T k;
   #pragma omp for ordered
   for (int i = 0; i < 10; ++i) {
     L1:
@@ -24,11 +25,115 @@ T foo() {
       foo();
     }
   }
-
+  #pragma omp for ordered
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads threads // expected-error {{directive '#pragma omp ordered' cannot contain more than one 'threads' clause}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for ordered(1) // expected-note {{'ordered' clause with specified parameter}}
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for ordered(1) // expected-note {{'ordered' clause with specified parameter}}
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{'ordered' directive with 'threads' clause cannot be closely nested inside ordered region with specified parameter}}
+    {
+      foo();
+    }
+  }
+  #pragma omp ordered simd simd // expected-error {{directive '#pragma omp ordered' cannot contain more than one 'simd' clause}}
+  {
+    foo();
+  }
+  #pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+    #pragma omp ordered depend(source) // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  }
+#pragma omp parallel for ordered
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered depend(source) // expected-error {{'ordered' directive with 'depend' clause cannot be closely nested inside ordered region without specified parameter}}
+    #pragma omp ordered depend(sink : i) // expected-error {{'ordered' directive with 'depend' clause cannot be closely nested inside ordered region without specified parameter}}
+  }
+#pragma omp parallel for ordered(2) // expected-note 5 {{'ordered' clause with specified parameter}}
+  for (int i = 0; i < 10; ++i) {
+    for (int j = 0; j < 10; ++j) {
+#pragma omp ordered depend // expected-error {{expected '(' after 'depend'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+#pragma omp ordered depend( // expected-error {{expected ')'}} expected-error {{expected 'source' or 'sink' in OpenMP clause 'depend'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}} expected-warning {{missing ':' or ')' after dependency type - ignoring}} expected-note {{to match this '('}}
+#pragma omp ordered depend(source // expected-error {{expected ')'}} expected-note {{to match this '('}}
+#pragma omp ordered depend(sink // expected-error {{expected expression}} expected-warning {{missing ':' or ')' after dependency type - ignoring}} expected-error {{expected ')'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}} expected-note {{to match this '('}}
+#pragma omp ordered depend(sink : // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected expression}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+#pragma omp ordered depend(sink : i // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected 'j' loop iteration variable}}
+#pragma omp ordered depend(sink : i) // expected-error {{expected 'j' loop iteration variable}}
+#pragma omp ordered depend(source)
+                           if (i == j)
+#pragma omp ordered depend(source) // expected-error {{'#pragma omp ordered' with 'depend' clause cannot be an immediate substatement}}
+                             ;
+                           if (i == j)
+#pragma omp ordered depend(sink : i, j) // expected-error {{'#pragma omp ordered' with 'depend' clause cannot be an immediate substatement}}
+                             ;
+#pragma omp ordered depend(source) threads // expected-error {{'depend' clauses cannot be mixed with 'threads' clause}}
+#pragma omp ordered simd depend(source) // expected-error {{'depend' clauses cannot be mixed with 'simd' clause}}
+#pragma omp ordered depend(source) depend(source) // expected-error {{directive '#pragma omp ordered' cannot contain more than one 'depend' clause with 'source' dependence}}
+#pragma omp ordered depend(in : i) // expected-error {{expected 'source' or 'sink' in OpenMP clause 'depend'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+#pragma omp ordered depend(sink : i, j)
+#pragma omp ordered depend(sink : j, i) // expected-error {{expected 'i' loop iteration variable}} expected-error {{expected 'j' loop iteration variable}}
+#pragma omp ordered depend(sink : i, j, k) // expected-error {{unexpected expression: number of expressions is larger than the number of associated loops}}
+#pragma omp ordered depend(sink : i+foo(), j/4) // expected-error {{expression is not an integral constant expression}} expected-error {{expected '+' or '-' operation}}
+#pragma omp ordered depend(sink : i*0, j-4)// expected-error {{expected '+' or '-' operation}}
+#pragma omp ordered depend(sink : i-0, j+sizeof(T)) depend(sink : i-0, j+sizeof(T))
+#pragma omp ordered depend(sink : i-0, j+sizeof(T)) depend(source) // expected-error {{'depend(source)' clause cannot be mixed with 'depend(sink:vec)' clauses}}
+#pragma omp ordered depend(source) depend(sink : i-0, j+sizeof(T)) // expected-error {{'depend(sink:vec)' clauses cannot be mixed with 'depend(source)' clause}}
+    }
+  }
   return T();
 }
 
 int foo() {
+int k;
   #pragma omp for ordered
   for (int i = 0; i < 10; ++i) {
     L1:
@@ -49,6 +154,110 @@ int foo() {
       foo();
     }
   }
+  #pragma omp for ordered
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads threads // expected-error {{directive '#pragma omp ordered' cannot contain more than one 'threads' clause}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for ordered(1) // expected-note {{'ordered' clause with specified parameter}}
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for ordered(1) // expected-note {{'ordered' clause with specified parameter}}
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{'ordered' directive with 'threads' clause cannot be closely nested inside ordered region with specified parameter}}
+    {
+      foo();
+    }
+  }
+  #pragma omp ordered simd simd // expected-error {{directive '#pragma omp ordered' cannot contain more than one 'simd' clause}}
+  {
+    foo();
+  }
+  #pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+  }
+  #pragma omp parallel for simd
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered threads // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+    {
+      foo();
+    }
+    #pragma omp ordered depend(source) // expected-error {{OpenMP constructs may not be nested inside a simd region}}
+  }
+#pragma omp parallel for ordered
+  for (int i = 0; i < 10; ++i) {
+    #pragma omp ordered depend(source) // expected-error {{'ordered' directive with 'depend' clause cannot be closely nested inside ordered region without specified parameter}}
+    #pragma omp ordered depend(sink : i) // expected-error {{'ordered' directive with 'depend' clause cannot be closely nested inside ordered region without specified parameter}}
+  }
+#pragma omp parallel for ordered(2) // expected-note 5 {{'ordered' clause with specified parameter}}
+  for (int i = 0; i < 10; ++i) {
+    for (int j = 0; j < 10; ++j) {
+#pragma omp ordered depend // expected-error {{expected '(' after 'depend'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+#pragma omp ordered depend( // expected-error {{expected ')'}} expected-error {{expected 'source' or 'sink' in OpenMP clause 'depend'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}} expected-warning {{missing ':' or ')' after dependency type - ignoring}} expected-note {{to match this '('}}
+#pragma omp ordered depend(source // expected-error {{expected ')'}} expected-note {{to match this '('}}
+#pragma omp ordered depend(sink // expected-error {{expected expression}} expected-warning {{missing ':' or ')' after dependency type - ignoring}} expected-error {{expected ')'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}} expected-note {{to match this '('}}
+#pragma omp ordered depend(sink : // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected expression}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+#pragma omp ordered depend(sink : i // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected 'j' loop iteration variable}}
+#pragma omp ordered depend(sink : i) // expected-error {{expected 'j' loop iteration variable}}
+#pragma omp ordered depend(source)
+                           if (i == j)
+#pragma omp ordered depend(source) // expected-error {{'#pragma omp ordered' with 'depend' clause cannot be an immediate substatement}}
+                             ;
+                           if (i == j)
+#pragma omp ordered depend(sink : i, j) // expected-error {{'#pragma omp ordered' with 'depend' clause cannot be an immediate substatement}}
+                             ;
+#pragma omp ordered depend(source) threads // expected-error {{'depend' clauses cannot be mixed with 'threads' clause}}
+#pragma omp ordered simd depend(source) // expected-error {{'depend' clauses cannot be mixed with 'simd' clause}}
+#pragma omp ordered depend(source) depend(source) // expected-error {{directive '#pragma omp ordered' cannot contain more than one 'depend' clause with 'source' dependence}}
+#pragma omp ordered depend(in : i) // expected-error {{expected 'source' or 'sink' in OpenMP clause 'depend'}} expected-error {{'ordered' directive without any clauses cannot be closely nested inside ordered region with specified parameter}}
+#pragma omp ordered depend(sink : i, j)
+#pragma omp ordered depend(sink : j, i) // expected-error {{expected 'i' loop iteration variable}} expected-error {{expected 'j' loop iteration variable}}
+#pragma omp ordered depend(sink : i, j, k) // expected-error {{unexpected expression: number of expressions is larger than the number of associated loops}}
+#pragma omp ordered depend(sink : i+foo(), j/4) // expected-error {{expression is not an integral constant expression}} expected-error {{expected '+' or '-' operation}}
+#pragma omp ordered depend(sink : i*0, j-4)// expected-error {{expected '+' or '-' operation}}
+#pragma omp ordered depend(sink : i-0, j+sizeof(int)) depend(sink : i-0, j+sizeof(int))
+#pragma omp ordered depend(sink : i-0, j+sizeof(int)) depend(source) // expected-error {{'depend(source)' clause cannot be mixed with 'depend(sink:vec)' clauses}}
+#pragma omp ordered depend(source) depend(sink : i-0, j+sizeof(int)) // expected-error {{'depend(sink:vec)' clauses cannot be mixed with 'depend(source)' clause}}
+    }
+  }
 
-  return foo<int>();
+  return foo<int>(); // expected-note {{in instantiation of function template specialization 'foo<int>' requested here}}
 }
diff --git a/test/OpenMP/parallel_ast_print.cpp b/test/OpenMP/parallel_ast_print.cpp
index c3bd416a6d5f..1e46fba48ca3 100644
--- a/test/OpenMP/parallel_ast_print.cpp
+++ b/test/OpenMP/parallel_ast_print.cpp
@@ -33,11 +33,12 @@ T tmain(T argc, T *argv) {
   T b = argc, c, d, e, f, g;
   static T a;
   S<T> s;
+  T arr[C][10], arr1[C];
 #pragma omp parallel
   a=2;
-#pragma omp parallel default(none), private(argc,b) firstprivate(argv) shared (d) if (argc > 0) num_threads(C) copyin(S<T>::TS) proc_bind(master) reduction(+:c) reduction(max:e)
+#pragma omp parallel default(none), private(argc,b) firstprivate(argv) shared (d) if (parallel:argc > 0) num_threads(C) copyin(S<T>::TS) proc_bind(master) reduction(+:c, arr1[argc]) reduction(max:e, arr[:C][0:10])
   foo();
-#pragma omp parallel if (C) num_threads(s) proc_bind(close) reduction(^:e, f) reduction(&& : g)
+#pragma omp parallel if (C) num_threads(s) proc_bind(close) reduction(^:e, f, arr[0:C][:argc]) reduction(&& : g)
   foo();
   return 0;
 }
@@ -46,31 +47,34 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: int b = argc, c, d, e, f, g;
 // CHECK-NEXT: static int a;
 // CHECK-NEXT: S<int> s;
+// CHECK-NEXT: int arr[5][10], arr1[5];
 // CHECK-NEXT: #pragma omp parallel
 // CHECK-NEXT: a = 2;
-// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) num_threads(5) copyin(S<int>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) shared(d) if(parallel: argc > 0) num_threads(5) copyin(S<int>::TS) proc_bind(master) reduction(+: c,arr1[argc]) reduction(max: e,arr[:5][0:10])
 // CHECK-NEXT: foo()
-// CHECK-NEXT: #pragma omp parallel if(5) num_threads(s) proc_bind(close) reduction(^: e,f) reduction(&&: g)
+// CHECK-NEXT: #pragma omp parallel if(5) num_threads(s) proc_bind(close) reduction(^: e,f,arr[0:5][:argc]) reduction(&&: g)
 // CHECK-NEXT: foo()
 // CHECK: template <typename T = long, int C = 1> long tmain(long argc, long *argv) {
 // CHECK-NEXT: long b = argc, c, d, e, f, g;
 // CHECK-NEXT: static long a;
 // CHECK-NEXT: S<long> s;
+// CHECK-NEXT: long arr[1][10], arr1[1];
 // CHECK-NEXT: #pragma omp parallel
 // CHECK-NEXT: a = 2;
-// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) num_threads(1) copyin(S<long>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) shared(d) if(parallel: argc > 0) num_threads(1) copyin(S<long>::TS) proc_bind(master) reduction(+: c,arr1[argc]) reduction(max: e,arr[:1][0:10])
 // CHECK-NEXT: foo()
-// CHECK-NEXT: #pragma omp parallel if(1) num_threads(s) proc_bind(close) reduction(^: e,f) reduction(&&: g)
+// CHECK-NEXT: #pragma omp parallel if(1) num_threads(s) proc_bind(close) reduction(^: e,f,arr[0:1][:argc]) reduction(&&: g)
 // CHECK-NEXT: foo()
 // CHECK: template <typename T, int C> T tmain(T argc, T *argv) {
 // CHECK-NEXT: T b = argc, c, d, e, f, g;
 // CHECK-NEXT: static T a;
 // CHECK-NEXT: S<T> s;
+// CHECK-NEXT: T arr[C][10], arr1[C];
 // CHECK-NEXT: #pragma omp parallel
 // CHECK-NEXT: a = 2;
-// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) num_threads(C) copyin(S<T>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) shared(d) if(parallel: argc > 0) num_threads(C) copyin(S<T>::TS) proc_bind(master) reduction(+: c,arr1[argc]) reduction(max: e,arr[:C][0:10])
 // CHECK-NEXT: foo()
-// CHECK-NEXT: #pragma omp parallel if(C) num_threads(s) proc_bind(close) reduction(^: e,f) reduction(&&: g)
+// CHECK-NEXT: #pragma omp parallel if(C) num_threads(s) proc_bind(close) reduction(^: e,f,arr[0:C][:argc]) reduction(&&: g)
 // CHECK-NEXT: foo()
 
 enum Enum { };
@@ -80,17 +84,40 @@ int main (int argc, char **argv) {
   int b = argc, c, d, e, f, g;
   static int a;
   #pragma omp threadprivate(a)
+  int arr[10][argc], arr1[2];
   Enum ee;
 // CHECK: Enum ee;
 #pragma omp parallel
 // CHECK-NEXT: #pragma omp parallel
   a=2;
 // CHECK-NEXT: a = 2;
-#pragma omp parallel default(none), private(argc,b) firstprivate(argv) if (argc > 0) num_threads(ee) copyin(a) proc_bind(spread) reduction(| : c, d) reduction(* : e)
-// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) if(argc > 0) num_threads(ee) copyin(a) proc_bind(spread) reduction(|: c,d) reduction(*: e)
+#pragma omp parallel default(none), private(argc,b) firstprivate(argv) if (parallel: argc > 0) num_threads(ee) copyin(a) proc_bind(spread) reduction(| : c, d, arr1[argc]) reduction(* : e, arr[:10][0:argc])
+// CHECK-NEXT: #pragma omp parallel default(none) private(argc,b) firstprivate(argv) if(parallel: argc > 0) num_threads(ee) copyin(a) proc_bind(spread) reduction(|: c,d,arr1[argc]) reduction(*: e,arr[:10][0:argc])
   foo();
 // CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp parallel if(b) num_threads(c) proc_bind(close) reduction(^: e,f) reduction(&&: g,arr[0:argc][:10])
+// CHECK-NEXT: foo()
+#pragma omp parallel if (b) num_threads(c) proc_bind(close) reduction(^:e, f) reduction(&& : g, arr[0:argc][:10])
+  foo();
   return tmain<int, 5>(b, &b) + tmain<long, 1>(x, &x);
 }
 
+template <class T>
+struct Foo {
+  int foo;
+};
+
+void foo(const Foo<int> &arg) {
+// CHECK: #pragma omp parallel
+#pragma omp parallel
+  {
+// CHECK: #pragma omp for schedule(static)
+#pragma omp for schedule(static)
+    for (int idx = 0; idx < 1234; ++idx) {
+      //arg.foo = idx;
+      idx = arg.foo;
+    }
+  }
+}
+
 #endif
diff --git a/test/OpenMP/parallel_codegen.cpp b/test/OpenMP/parallel_codegen.cpp
index 6486e44a9dac..04c5c2c85ae2 100644
--- a/test/OpenMP/parallel_codegen.cpp
+++ b/test/OpenMP/parallel_codegen.cpp
@@ -1,18 +1,14 @@
-// RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -triple %itanium_abi_triple -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-DEBUG %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-DEBUG %s
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
 #ifndef HEADER
 #define HEADER
 // CHECK-DAG: %ident_t = type { i32, i32, i32, i32, i8* }
-// CHECK-DAG: %struct.anon = type { i32* }
-// CHECK-DAG: %struct.anon.0 = type { i8*** }
 // CHECK-DAG: [[STR:@.+]] = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00"
 // CHECK-DAG: [[DEF_LOC_2:@.+]] = private unnamed_addr constant %ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* [[STR]], i32 0, i32 0) }
 // CHECK-DEBUG-DAG: %ident_t = type { i32, i32, i32, i32, i8* }
-// CHECK-DEBUG-DAG: %struct.anon = type { i32* }
-// CHECK-DEBUG-DAG: %struct.anon.0 = type { i8*** }
 // CHECK-DEBUG-DAG: [[STR:@.+]] = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00"
 // CHECK-DEBUG-DAG: [[DEF_LOC_2:@.+]] = private unnamed_addr constant %ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* [[STR]], i32 0, i32 0) }
 // CHECK-DEBUG-DAG: [[LOC1:@.+]] = private unnamed_addr constant [{{.+}} x i8] c";{{.*}}parallel_codegen.cpp;main;[[@LINE+14]];9;;\00"
@@ -35,56 +31,42 @@ int main (int argc, char **argv) {
 }
 
 // CHECK-LABEL: define {{[a-z\_\b]*[ ]?i32}} @main({{i32[ ]?[a-z]*}} %argc, i8** %argv)
-// CHECK:       [[AGG_CAPTURED:%.+]] = alloca %struct.anon
-// CHECK:       [[ARGC_REF:%.+]] = getelementptr inbounds %struct.anon, %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
-// CHECK-NEXT:  store i32* {{%[a-z0-9.]+}}, i32** [[ARGC_REF]]
-// CHECK-NEXT:  [[BITCAST:%.+]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
-// CHECK-NEXT:  call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC_2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8* [[BITCAST]])
+// CHECK: store i32 %argc, i32* [[ARGC_ADDR:%.+]],
+// CHECK:  call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC_2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i32* [[ARGC_ADDR]])
 // CHECK-NEXT:  [[ARGV:%.+]] = load i8**, i8*** {{%[a-z0-9.]+}}
 // CHECK-NEXT:  [[RET:%.+]] = call {{[a-z\_\b]*[ ]?i32}} [[TMAIN:@.+tmain.+]](i8** [[ARGV]])
 // CHECK-NEXT:  ret i32 [[RET]]
 // CHECK-NEXT:  }
 // CHECK-DEBUG-LABEL: define i32 @main(i32 %argc, i8** %argv)
-// CHECK-DEBUG-DAG:   [[AGG_CAPTURED:%.+]] = alloca %struct.anon
-// CHECK-DEBUG-DAG:   [[LOC_2_ADDR:%.+]] = alloca %ident_t
+// CHECK-DEBUG:       [[LOC_2_ADDR:%.+]] = alloca %ident_t
 // CHECK-DEBUG:       [[KMPC_LOC_VOIDPTR:%.+]] = bitcast %ident_t* [[LOC_2_ADDR]] to i8*
 // CHECK-DEBUG-NEXT:  [[KMPC_DEFAULT_LOC_VOIDPTR:%.+]] = bitcast %ident_t* [[DEF_LOC_2]] to i8*
-// CHECK-DEBUG-NEXT:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[KMPC_LOC_VOIDPTR]], i8* [[KMPC_DEFAULT_LOC_VOIDPTR]], i64 ptrtoint (%ident_t* getelementptr (%ident_t, %ident_t* null, i32 1) to i64), i32 8, i1 false)
-// CHECK-DEBUG:       [[ARGC_REF:%.+]] = getelementptr inbounds %struct.anon, %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
-// CHECK-DEBUG-NEXT:  store i32* {{%[a-z0-9.]+}}, i32** [[ARGC_REF]]
-// CHECK-DEBUG-NEXT:  [[KMPC_LOC_PSOURCE_REF:%.+]] = getelementptr inbounds %ident_t, %ident_t* [[LOC_2_ADDR]], i32 0, i32 4
+// CHECK-DEBUG-NEXT:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[KMPC_LOC_VOIDPTR]], i8* [[KMPC_DEFAULT_LOC_VOIDPTR]], i64 24, i32 8, i1 false)
+// CHECK-DEBUG:       store i32 %argc, i32* [[ARGC_ADDR:%.+]],
+// CHECK-DEBUG:       [[KMPC_LOC_PSOURCE_REF:%.+]] = getelementptr inbounds %ident_t, %ident_t* [[LOC_2_ADDR]], i32 0, i32 4
 // CHECK-DEBUG-NEXT:  store i8* getelementptr inbounds ([{{.+}} x i8], [{{.+}} x i8]* [[LOC1]], i32 0, i32 0), i8** [[KMPC_LOC_PSOURCE_REF]]
-// CHECK-DEBUG-NEXT:  [[BITCAST:%.+]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
-// CHECK-DEBUG-NEXT:  call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[LOC_2_ADDR]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8* [[BITCAST]])
+// CHECK-DEBUG:  call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[LOC_2_ADDR]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i32* [[ARGC_ADDR]])
 // CHECK-DEBUG-NEXT:  [[ARGV:%.+]] = load i8**, i8*** {{%[a-z0-9.]+}}
 // CHECK-DEBUG-NEXT:  [[RET:%.+]] = call i32 [[TMAIN:@.+tmain.+]](i8** [[ARGV]])
 // CHECK-DEBUG-NEXT:  ret i32 [[RET]]
 // CHECK-DEBUG-NEXT:  }
 
-// CHECK:       define internal {{.*}}void [[OMP_OUTLINED]](i32* %.global_tid., i32* %.bound_tid., %struct.anon* %__context)
-// CHECK:       #[[FN_ATTRS:[0-9]+]]
-// CHECK:       [[CONTEXT_ADDR:%.+]] = alloca %struct.anon*
-// CHECK:       store %struct.anon* %__context, %struct.anon** [[CONTEXT_ADDR]]
-// CHECK:       [[CONTEXT_PTR:%.+]] = load %struct.anon*, %struct.anon** [[CONTEXT_ADDR]]
-// CHECK-NEXT:  [[ARGC_PTR_REF:%.+]] = getelementptr inbounds %struct.anon, %struct.anon* [[CONTEXT_PTR]], i32 0, i32 0
-// CHECK-NEXT:  [[ARGC_REF:%.+]] = load i32*, i32** [[ARGC_PTR_REF]]
+// CHECK:       define internal {{.*}}void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) [[ARGC_ADDR:%[^)]+]])
+// CHECK-SAME:       #[[FN_ATTRS:[0-9]+]]
+// CHECK:       store i32* [[ARGC_ADDR]], i32** [[ARGC_PTR_ADDR:%.+]],
+// CHECK:       [[ARGC_REF:%.+]] = load i32*, i32** [[ARGC_PTR_ADDR]]
 // CHECK-NEXT:  [[ARGC:%.+]] = load i32, i32* [[ARGC_REF]]
 // CHECK-NEXT:  invoke {{.*}}void [[FOO:@.+foo.+]](i32{{[ ]?[a-z]*}} [[ARGC]])
-// CHECK:       call {{.+}} @__kmpc_cancel_barrier(
 // CHECK:       ret void
 // CHECK:       call {{.*}}void @{{.+terminate.*|abort}}(
 // CHECK-NEXT:  unreachable
 // CHECK-NEXT:  }
-// CHECK-DEBUG:       define internal void [[OMP_OUTLINED]](i32* %.global_tid., i32* %.bound_tid., %struct.anon* %__context)
-// CHECK-DEBUG:       #[[FN_ATTRS:[0-9]+]]
-// CHECK-DEBUG:       [[CONTEXT_ADDR:%.+]] = alloca %struct.anon*
-// CHECK-DEBUG:       store %struct.anon* %__context, %struct.anon** [[CONTEXT_ADDR]]
-// CHECK-DEBUG:       [[CONTEXT_PTR:%.+]] = load %struct.anon*, %struct.anon** [[CONTEXT_ADDR]]
-// CHECK-DEBUG-NEXT:  [[ARGC_PTR_REF:%.+]] = getelementptr inbounds %struct.anon, %struct.anon* [[CONTEXT_PTR]], i32 0, i32 0
-// CHECK-DEBUG-NEXT:  [[ARGC_REF:%.+]] = load i32*, i32** [[ARGC_PTR_REF]]
+// CHECK-DEBUG:       define internal void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) [[ARGC_ADDR:%[^)]+]])
+// CHECK-DEBUG-SAME:  #[[FN_ATTRS:[0-9]+]]
+// CHECK-DEBUG:       store i32* [[ARGC_ADDR]], i32** [[ARGC_PTR_ADDR:%.+]],
+// CHECK-DEBUG:  [[ARGC_REF:%.+]] = load i32*, i32** [[ARGC_PTR_ADDR]]
 // CHECK-DEBUG-NEXT:  [[ARGC:%.+]] = load i32, i32* [[ARGC_REF]]
 // CHECK-DEBUG-NEXT:  invoke void [[FOO:@.+foo.+]](i32 [[ARGC]])
-// CHECK-DEBUG:       call {{.+}} @__kmpc_cancel_barrier(
 // CHECK-DEBUG:       ret void
 // CHECK-DEBUG:       call void @{{.+terminate.*|abort}}(
 // CHECK-DEBUG-NEXT:  unreachable
@@ -96,50 +78,36 @@ int main (int argc, char **argv) {
 // CHECK-DEBUG-DAG: declare void @__kmpc_fork_call(%ident_t*, i32, void (i32*, i32*, ...)*, ...)
 
 // CHECK:       define linkonce_odr {{[a-z\_\b]*[ ]?i32}} [[TMAIN]](i8** %argc)
-// CHECK:       [[AGG_CAPTURED:%.+]] = alloca %struct.anon.0
-// CHECK:       [[ARGC_REF:%.+]] = getelementptr inbounds %struct.anon.0, %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 0
-// CHECK-NEXT:  store i8*** {{%[a-z0-9.]+}}, i8**** [[ARGC_REF]]
-// CHECK-NEXT:  [[BITCAST:%.+]] = bitcast %struct.anon.0* [[AGG_CAPTURED]] to i8*
-// CHECK-NEXT:  call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC_2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon.0*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8* [[BITCAST]])
+// CHECK:       store i8** %argc, i8*** [[ARGC_ADDR:%.+]],
+// CHECK-NEXT:  call {{.*}}void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[DEF_LOC_2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8***)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8*** [[ARGC_ADDR]])
 // CHECK-NEXT:  ret i32 0
 // CHECK-NEXT:  }
 // CHECK-DEBUG:       define linkonce_odr i32 [[TMAIN]](i8** %argc)
-// CHECK-DEBUG-DAG:   [[AGG_CAPTURED:%.+]] = alloca %struct.anon.0
 // CHECK-DEBUG-DAG:   [[LOC_2_ADDR:%.+]] = alloca %ident_t
 // CHECK-DEBUG:       [[KMPC_LOC_VOIDPTR:%.+]] = bitcast %ident_t* [[LOC_2_ADDR]] to i8*
 // CHECK-DEBUG-NEXT:  [[KMPC_DEFAULT_LOC_VOIDPTR:%.+]] = bitcast %ident_t* [[DEF_LOC_2]] to i8*
-// CHECK-DEBUG-NEXT:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[KMPC_LOC_VOIDPTR]], i8* [[KMPC_DEFAULT_LOC_VOIDPTR]], i64 ptrtoint (%ident_t* getelementptr (%ident_t, %ident_t* null, i32 1) to i64), i32 8, i1 false)
-// CHECK-DEBUG:       [[ARGC_REF:%.+]] = getelementptr inbounds %struct.anon.0, %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 0
-// CHECK-DEBUG-NEXT:  store i8*** {{%[a-z0-9.]+}}, i8**** [[ARGC_REF]]
-// CHECK-DEBUG-NEXT:  [[KMPC_LOC_PSOURCE_REF:%.+]] = getelementptr inbounds %ident_t, %ident_t* [[LOC_2_ADDR]], i32 0, i32 4
+// CHECK-DEBUG-NEXT:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[KMPC_LOC_VOIDPTR]], i8* [[KMPC_DEFAULT_LOC_VOIDPTR]], i64 24, i32 8, i1 false)
+// CHECK-DEBUG-NEXT:  store i8** %argc, i8*** [[ARGC_ADDR:%.+]],
+// CHECK-DEBUG:  [[KMPC_LOC_PSOURCE_REF:%.+]] = getelementptr inbounds %ident_t, %ident_t* [[LOC_2_ADDR]], i32 0, i32 4
 // CHECK-DEBUG-NEXT:  store i8* getelementptr inbounds ([{{.+}} x i8], [{{.+}} x i8]* [[LOC2]], i32 0, i32 0), i8** [[KMPC_LOC_PSOURCE_REF]]
-// CHECK-DEBUG-NEXT:  [[BITCAST:%.+]] = bitcast %struct.anon.0* [[AGG_CAPTURED]] to i8*
-// CHECK-DEBUG-NEXT:  call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[LOC_2_ADDR]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon.0*)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8* [[BITCAST]])
+// CHECK-DEBUG-NEXT:  call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%ident_t* [[LOC_2_ADDR]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8***)* [[OMP_OUTLINED:@.+]] to void (i32*, i32*, ...)*), i8*** [[ARGC_ADDR]])
 // CHECK-DEBUG-NEXT:  ret i32 0
 // CHECK-DEBUG-NEXT:  }
 
-// CHECK:       define internal {{.*}}void [[OMP_OUTLINED]](i32* %.global_tid., i32* %.bound_tid., %struct.anon.0* %__context)
-// CHECK:       [[CONTEXT_ADDR:%.+]] = alloca %struct.anon.0*
-// CHECK:       store %struct.anon.0* %__context, %struct.anon.0** [[CONTEXT_ADDR]]
-// CHECK:       [[CONTEXT_PTR:%.+]] = load %struct.anon.0*, %struct.anon.0** [[CONTEXT_ADDR]]
-// CHECK-NEXT:  [[ARGC_PTR_REF:%.+]] = getelementptr inbounds %struct.anon.0, %struct.anon.0* [[CONTEXT_PTR]], i32 0, i32 0
-// CHECK-NEXT:  [[ARGC_REF:%.+]] = load i8***, i8**** [[ARGC_PTR_REF]]
+// CHECK:       define internal {{.*}}void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i8*** dereferenceable({{4|8}}) %argc)
+// CHECK:       store i8*** %argc, i8**** [[ARGC_PTR_ADDR:%.+]],
+// CHECK:       [[ARGC_REF:%.+]] = load i8***, i8**** [[ARGC_PTR_ADDR]]
 // CHECK-NEXT:  [[ARGC:%.+]] = load i8**, i8*** [[ARGC_REF]]
 // CHECK-NEXT:  invoke {{.*}}void [[FOO1:@.+foo.+]](i8** [[ARGC]])
-// CHECK:       call {{.+}} @__kmpc_cancel_barrier(
 // CHECK:       ret void
 // CHECK:       call {{.*}}void @{{.+terminate.*|abort}}(
 // CHECK-NEXT:  unreachable
 // CHECK-NEXT:  }
-// CHECK-DEBUG:       define internal void [[OMP_OUTLINED]](i32* %.global_tid., i32* %.bound_tid., %struct.anon.0* %__context)
-// CHECK-DEBUG:       [[CONTEXT_ADDR:%.+]] = alloca %struct.anon.0*
-// CHECK-DEBUG:       store %struct.anon.0* %__context, %struct.anon.0** [[CONTEXT_ADDR]]
-// CHECK-DEBUG:       [[CONTEXT_PTR:%.+]] = load %struct.anon.0*, %struct.anon.0** [[CONTEXT_ADDR]]
-// CHECK-DEBUG-NEXT:  [[ARGC_PTR_REF:%.+]] = getelementptr inbounds %struct.anon.0, %struct.anon.0* [[CONTEXT_PTR]], i32 0, i32 0
-// CHECK-DEBUG-NEXT:  [[ARGC_REF:%.+]] = load i8***, i8**** [[ARGC_PTR_REF]]
+// CHECK-DEBUG:       define internal void [[OMP_OUTLINED]](i32* noalias %.global_tid., i32* noalias %.bound_tid., i8*** dereferenceable({{4|8}}) %argc)
+// CHECK-DEBUG:       store i8*** %argc, i8**** [[ARGC_PTR_ADDR:%.+]],
+// CHECK-DEBUG:       [[ARGC_REF:%.+]] = load i8***, i8**** [[ARGC_PTR_ADDR]]
 // CHECK-DEBUG-NEXT:  [[ARGC:%.+]] = load i8**, i8*** [[ARGC_REF]]
 // CHECK-DEBUG-NEXT:  invoke void [[FOO1:@.+foo.+]](i8** [[ARGC]])
-// CHECK-DEBUG:       call {{.+}} @__kmpc_cancel_barrier(
 // CHECK-DEBUG:       ret void
 // CHECK-DEBUG:       call void @{{.+terminate.*|abort}}(
 // CHECK-DEBUG-NEXT:  unreachable
diff --git a/test/OpenMP/parallel_copyin_codegen.cpp b/test/OpenMP/parallel_copyin_codegen.cpp
index 4aacb809573f..ff76cfe4dd6c 100644
--- a/test/OpenMP/parallel_copyin_codegen.cpp
+++ b/test/OpenMP/parallel_copyin_codegen.cpp
@@ -11,13 +11,12 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=TLS-LAMBDA %s
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple %itanium_abi_triple -emit-llvm %s -o - | FileCheck -check-prefix=TLS-BLOCKS %s
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DARRAY -triple x86_64-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=TLS-ARRAY %s
-// REQUIRES: tls
 // expected-no-diagnostics
 #ifndef ARRAY
 #ifndef HEADER
 #define HEADER
 
-volatile int g = 1212;
+volatile int g __attribute__((aligned(128))) = 1212;
 #pragma omp threadprivate(g)
 
 template <class T>
@@ -58,10 +57,10 @@ template <typename T>
 T tmain() {
   S<T> test;
   test = S<T>();
-  static T t_var = 333;
-  static T vec[] = {3, 3};
-  static S<T> s_arr[] = {1, 2};
-  static S<T> var(3);
+  static T t_var __attribute__((aligned(128))) = 333;
+  static T vec[] __attribute__((aligned(128))) = {3, 3};
+  static S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
+  static S<T> var __attribute__((aligned(128))) (3);
 #pragma omp threadprivate(t_var, vec, s_arr, var)
 #pragma omp parallel copyin(t_var, vec, s_arr, var)
   {
@@ -78,28 +77,24 @@ int main() {
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
-  // TLS-LAMBDA: [[G_CAPTURE_TY:%.+]] = type { i{{[0-9]+}}* }
   // TLS-LAMBDA: [[G:@.+]] = {{.*}}thread_local {{.*}}global i{{[0-9]+}} 1212,
   // TLS-LAMBDA-LABEL: @main
   // TLS-LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8*
+  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 
-  // TLS-LAMBDA-DAG: [[G_CPY_ADDR:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* [[G_CAPTURE:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-  // TLS-LAMBDA-DAG: [[G_CPY_VAL:%.+]] = call i{{[0-9]+}}* [[G_CTOR:@.+]]()
-  // TLS-LAMBDA:     store i{{[0-9]+}}* [[G_CPY_VAL]], i{{[0-9]+}}** [[G_CPY_ADDR]]
-  // TLS-LAMBDA:     [[G_CAPTURE_CAST:%.+]] = bitcast [[G_CAPTURE_TY]]* [[G_CAPTURE]] to i8*
-  // TLS-LAMBDA:     call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[G_CAPTURE_CAST]]
+  // TLS-LAMBDA:     [[G_CPY_VAL:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR:@.+]]()
+  // TLS-LAMBDA:     call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G_CPY_VAL]])
 
-  // TLS-LAMBDA:     define {{.*}}i{{[0-9]+}}* [[G_CTOR]]() {
+  // TLS-LAMBDA:     define {{.*}}i{{[0-9]+}}* [[G_CTOR]]()
   // TLS-LAMBDA:     ret i{{[0-9]+}}* [[G]]
   // TLS-LAMBDA:     }
 
 #pragma omp parallel copyin(g)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
-    // TLS-LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, [[G_CAPTURE_TY]]* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
+    // TLS-LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
 
     // threadprivate_g = g;
     // LAMBDA: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[G]]
@@ -107,24 +102,23 @@ int main() {
     // LAMBDA: icmp ne i{{[0-9]+}} ptrtoint (i{{[0-9]+}}* [[G]] to i{{[0-9]+}}), %{{.+}}
     // LAMBDA: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
     // LAMBDA: [[NOT_MASTER]]
-    // LAMBDA: load i{{[0-9]+}}, i{{[0-9]+}}* [[G]],
-    // LAMBDA: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}},
+    // LAMBDA: load i{{[0-9]+}}, i{{[0-9]+}}* [[G]], align 128
+    // LAMBDA: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, align 128
     // LAMBDA: [[DONE]]
 
-    // TLS-LAMBDA-DAG: [[G_CAPTURE_FIELD:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-    // TLS-LAMBDA-DAG: [[G_CAPTURE_SRC:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_CAPTURE_FIELD]]
-    // TLS-LAMBDA-DAG: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]()
+    // TLS-LAMBDA-DAG: [[G_CAPTURE_SRC:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+    // TLS-LAMBDA-DAG: [[G_CAPTURE_DST:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR]]()
     // TLS-LAMBDA-DAG: [[G_CAPTURE_SRCC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_SRC]] to i{{[0-9]+}}
     // TLS-LAMBDA-DAG: [[G_CAPTURE_DSTC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_DST]] to i{{[0-9]+}}
     // TLS-LAMBDA: icmp ne i{{[0-9]+}} {{%.+}}, {{%.+}}
     // TLS-LAMBDA: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
     // TLS-LAMBDA: [[NOT_MASTER]]
     // TLS-LAMBDA: load i{{[0-9]+}}, i{{[0-9]+}}* [[G_CAPTURE_SRC]],
-    // TLS-LAMBDA: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* [[G_CAPTURE_DST]],
+    // TLS-LAMBDA: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* [[G_CAPTURE_DST]], align 128
     // TLS-LAMBDA: [[DONE]]
 
-    // LAMBDA: call {{.*}}i32 @__kmpc_cancel_barrier(
-    // TLS-LAMBDA: call {{.*}}i32 @__kmpc_cancel_barrier(
+    // LAMBDA: call {{.*}}void @__kmpc_barrier(
+    // TLS-LAMBDA: call {{.*}}void @__kmpc_barrier(
     g = 1;
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}*
     // TLS-LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}*
@@ -134,8 +128,8 @@ int main() {
       g = 2;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
 
-      // TLS-LAMBDA: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]()
-      // TLS-LAMBDA: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_CAPTURE_DST]]
+      // TLS-LAMBDA: [[G_CAPTURE_DST:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR]]()
+      // TLS-LAMBDA: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_CAPTURE_DST]], align 128
     }();
   }
   }();
@@ -145,27 +139,23 @@ int main() {
   // BLOCKS-LABEL: @main
   // BLOCKS: call {{.*}}void {{%.+}}(i8
 
-  // TLS-BLOCKS: [[G_CAPTURE_TY:%.+]] = type { i{{[0-9]+}}* }
   // TLS-BLOCKS: [[G:@.+]] = {{.*}}thread_local {{.*}}global i{{[0-9]+}} 1212,
   // TLS-BLOCKS-LABEL: @main
   // TLS-BLOCKS: call {{.*}}void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8*
+  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 
-  // TLS-BLOCKS-DAG: [[G_CPY_ADDR:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* [[G_CAPTURE:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-  // TLS-BLOCKS-DAG: [[G_CPY_VAL:%.+]] = call i{{[0-9]+}}* [[G_CTOR:@.+]]()
-  // TLS-BLOCKS:     store i{{[0-9]+}}* [[G_CPY_VAL]], i{{[0-9]+}}** [[G_CPY_ADDR]]
-  // TLS-BLOCKS:     [[G_CAPTURE_CAST:%.+]] = bitcast [[G_CAPTURE_TY]]* [[G_CAPTURE]] to i8*
-  // TLS-BLOCKS:     call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[G_CAPTURE_CAST]]
+  // TLS-BLOCKS:     [[G_CPY_VAL:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR:@.+]]()
+  // TLS-BLOCKS:     call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G_CPY_VAL]])
 
-  // TLS-BLOCKS:     define {{.*}}i{{[0-9]+}}* [[G_CTOR]]() {
+  // TLS-BLOCKS:     define {{.*}}i{{[0-9]+}}* [[G_CTOR]]()
   // TLS-BLOCKS:     ret i{{[0-9]+}}* [[G]]
   // TLS-BLOCKS:     }
 #pragma omp parallel copyin(g)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
-    // TLS-BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, [[G_CAPTURE_TY]]* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
+    // TLS-BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
 
     // threadprivate_g = g;
     // BLOCKS: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[G]]
@@ -173,30 +163,29 @@ int main() {
     // BLOCKS: icmp ne i{{[0-9]+}} ptrtoint (i{{[0-9]+}}* [[G]] to i{{[0-9]+}}), %{{.+}}
     // BLOCKS: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
     // BLOCKS: [[NOT_MASTER]]
-    // BLOCKS: load i{{[0-9]+}}, i{{[0-9]+}}* [[G]],
-    // BLOCKS: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}},
+    // BLOCKS: load i{{[0-9]+}}, i{{[0-9]+}}* [[G]], align 128
+    // BLOCKS: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, align 128
     // BLOCKS: [[DONE]]
 
-    // TLS-BLOCKS-DAG: [[G_CAPTURE_FIELD:%.+]] = getelementptr inbounds [[G_CAPTURE_TY]], [[G_CAPTURE_TY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-    // TLS-BLOCKS-DAG: [[G_CAPTURE_SRC:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_CAPTURE_FIELD]]
-    // TLS-BLOCKS-DAG: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]()
+    // TLS-BLOCKS-DAG: [[G_CAPTURE_SRC:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+    // TLS-BLOCKS-DAG: [[G_CAPTURE_DST:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR]]()
     // TLS-BLOCKS-DAG: [[G_CAPTURE_SRCC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_SRC]] to i{{[0-9]+}}
     // TLS-BLOCKS-DAG: [[G_CAPTURE_DSTC:%.+]] = ptrtoint i{{[0-9]+}}* [[G_CAPTURE_DST]] to i{{[0-9]+}}
     // TLS-BLOCKS: icmp ne i{{[0-9]+}} {{%.+}}, {{%.+}}
     // TLS-BLOCKS: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
     // TLS-BLOCKS: [[NOT_MASTER]]
     // TLS-BLOCKS: load i{{[0-9]+}}, i{{[0-9]+}}* [[G_CAPTURE_SRC]],
-    // TLS-BLOCKS: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* [[G_CAPTURE_DST]],
+    // TLS-BLOCKS: store volatile i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* [[G_CAPTURE_DST]], align 128
     // TLS-BLOCKS: [[DONE]]
 
-    // BLOCKS: call {{.*}}i32 @__kmpc_cancel_barrier(
-    // TLS-BLOCKS: call {{.*}}i32 @__kmpc_cancel_barrier(
+    // BLOCKS: call {{.*}}void @__kmpc_barrier(
+    // TLS-BLOCKS: call {{.*}}void @__kmpc_barrier(
     g = 1;
     // BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}*
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
 
-    // TLS-BLOCKS: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]()
+    // TLS-BLOCKS: [[G_CAPTURE_DST:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR]]()
     // TLS-BLOCKS: store volatile i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_CAPTURE_DST]]
     // TLS-BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // TLS-BLOCKS: call {{.*}}void {{%.+}}(i8
@@ -211,7 +200,7 @@ int main() {
       // BLOCKS: ret
 
       // TLS-BLOCKS-NOT: [[G]]{{[[^:word:]]}}
-      // TLS-BLOCKS: [[G_CAPTURE_DST:%.+]] = call i{{[0-9]+}}* [[G_CTOR]]()
+      // TLS-BLOCKS: [[G_CAPTURE_DST:%.+]] = call{{( cxx_fast_tlscc)?}} i{{[0-9]+}}* [[G_CTOR]]()
       // TLS-BLOCKS: store volatile i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_CAPTURE_DST]]
       // TLS-BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // TLS-BLOCKS: ret
@@ -241,8 +230,8 @@ int main() {
 // CHECK-LABEL: @main
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* [[TEST]], [[S_FLOAT_TY]]*
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[MAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[MAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*))
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*))
 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
@@ -250,28 +239,18 @@ int main() {
 // TLS-CHECK-LABEL: @main
 // TLS-CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN:@.+]]([[S_FLOAT_TY]]* [[TEST]], [[S_FLOAT_TY]]*
-// TLS-CHECK-DAG: store i{{[0-9]+}}* [[T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]],
-// TLS-CHECK-DAG: store [2 x i{{[0-9]+}}]* [[VEC]], [2 x i{{[0-9]+}}]** [[VEC_FIELD:%.+]],
-// TLS-CHECK-DAG: store [2 x [[S_FLOAT_TY]]]* [[S_ARR]], [2 x [[S_FLOAT_TY]]]** [[S_ARR_FIELD:%.+]],
-// TLS-CHECK-DAG: store [[S_FLOAT_TY]]* [[VAR]], [[S_FLOAT_TY]]** [[VAR_FIELD:%.+]],
-// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK-DAG: [[VEC_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK-DAG: [[S_ARR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK-DAG: [[VAR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[MAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// TLS-CHECK-DAG: store i{{[0-9]+}}* [[T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]],
-// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[MAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
+// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*)* [[MAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*),
+// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*)* [[MAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*),
 // TLS-CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // TLS-CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // TLS-CHECK: ret
 
-// CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
 
-// TLS-CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// TLS-CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
 // threadprivate_t_var = t_var;
@@ -283,8 +262,11 @@ int main() {
 // CHECK: load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR]],
 // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}},
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]]
+// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** %
+// TLS-CHECK: [[MASTER_REF2:%.+]] = load [2 x i32]*, [2 x i32]** %
+// TLS-CHECK: [[MASTER_REF3:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** %
+// TLS-CHECK: [[MASTER_REF4:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+
 // TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}}
 // TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[T_VAR]] to i{{[0-9]+}})
 // TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
@@ -296,9 +278,7 @@ int main() {
 // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[VEC]]
 // CHECK: call void @llvm.memcpy{{.*}}(i8* %{{.+}}, i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*),
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x i32]*, [2 x i32]** [[MASTER_FIELD]]
-// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x i32]* [[MASTER_REF]] to i8*
+// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x i32]* [[MASTER_REF2]] to i8*
 // TLS-CHECK: call void @llvm.memcpy{{.*}}(i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*), i8* [[MASTER_CAST]]
 
 // threadprivate_s_arr = s_arr;
@@ -311,9 +291,7 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}})
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[MASTER_FIELD]]
-// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[MASTER_REF]] to [[S_FLOAT_TY]]*
+// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[MASTER_REF3]] to [[S_FLOAT_TY]]*
 // TLS-CHECK-DAG: [[S_ARR_SRC_BEGIN:%.+]] = phi [[S_FLOAT_TY]]* {{.*}}[[MASTER_CAST]]
 // TLS-CHECK-DAG: [[S_ARR_DST_BEGIN:%.+]] = phi [[S_FLOAT_TY]]* {{.*}}getelementptr inbounds ([2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0)
 // TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}})
@@ -328,24 +306,22 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{%.+}}, [[S_FLOAT_TY]]* {{.*}}[[VAR]])
 // CHECK: [[DONE]]
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[MASTER_FIELD]]
-// TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.*}}[[VAR]], [[S_FLOAT_TY]]* {{.*}}[[MASTER_REF]])
+// TLS-CHECK: call {{.*}} [[S_FLOAT_TY_COPY_ASSIGN]]([[S_FLOAT_TY]]* {{.*}}[[VAR]], [[S_FLOAT_TY]]* {{.*}}[[MASTER_REF4]])
 
-// CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // CHECK: ret void
 
 // TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
-// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// TLS-CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // TLS-CHECK: ret void
 
-// CHECK: define internal {{.*}}void [[MAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// CHECK: define internal {{.*}}void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
 
-// TLS-CHECK: define internal {{.*}}void [[MAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// TLS-CHECK: define internal {{.*}}void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
 // threadprivate_t_var = t_var;
@@ -358,8 +334,8 @@ int main() {
 // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}},
 // CHECK: [[DONE]]
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]]
+// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** %
+
 // TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}}
 // TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[T_VAR]] to i{{[0-9]+}})
 // TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
@@ -368,44 +344,34 @@ int main() {
 // TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[T_VAR]]
 // TLS-CHECK: [[DONE]]
 
-// CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // CHECK: ret void
 
 // TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
-// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// TLS-CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // TLS-CHECK: ret void
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[TEST]], [[S_INT_TY]]*
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*))
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*))
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 
 // TLS-CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // TLS-CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN:@.+]]([[S_INT_TY]]* [[TEST]], [[S_INT_TY]]*
-// TLS-CHECK-DAG: store i{{[0-9]+}}* [[TMAIN_T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]],
-// TLS-CHECK-DAG: store [2 x i{{[0-9]+}}]* [[TMAIN_VEC]], [2 x i{{[0-9]+}}]** [[VEC_FIELD:%.+]],
-// TLS-CHECK-DAG: store [2 x [[S_INT_TY]]]* [[TMAIN_S_ARR]], [2 x [[S_INT_TY]]]** [[S_ARR_FIELD:%.+]],
-// TLS-CHECK-DAG: store [[S_INT_TY]]* [[TMAIN_VAR]], [[S_INT_TY]]** [[VAR_FIELD:%.+]],
-// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK-DAG: [[VEC_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK-DAG: [[S_ARR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK-DAG: [[VAR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// TLS-CHECK-DAG: store i{{[0-9]+}}* [[TMAIN_T_VAR]], i{{[0-9]+}}** [[T_VAR_FIELD:%.+]],
-// TLS-CHECK-DAG: [[T_VAR_FIELD]] = getelementptr inbounds {{.*}}
-// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, {{%.+}}*)* [[TMAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
+// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void (i32*, i32*, ...)*),
+// TLS-CHECK:     call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*)* [[TMAIN_MICROTASK1:@.+]] to void (i32*, i32*, ...)*),
 //
-// CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
 //
-// TLS-CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// TLS-CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
 // threadprivate_t_var = t_var;
@@ -414,25 +380,26 @@ int main() {
 // CHECK: icmp ne i{{[0-9]+}} ptrtoint (i{{[0-9]+}}* [[TMAIN_T_VAR]] to i{{[0-9]+}}), %{{.+}}
 // CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
 // CHECK: [[NOT_MASTER]]
-// CHECK: load i{{[0-9]+}}, i{{[0-9]+}}* [[TMAIN_T_VAR]],
-// CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}},
+// CHECK: load i{{[0-9]+}}, i{{[0-9]+}}* [[TMAIN_T_VAR]], align 128
+// CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}}, align 128
+
+// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** %
+// TLS-CHECK: [[MASTER_REF1:%.+]] = load [2 x i32]*, [2 x i32]** %
+// TLS-CHECK: [[MASTER_REF2:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+// TLS-CHECK: [[MASTER_REF3:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]]
 // TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}}
 // TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[TMAIN_T_VAR]] to i{{[0-9]+}})
 // TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
 // TLS-CHECK: [[NOT_MASTER]]
-// TLS-CHECK: [[MASTER_VAL:%.+]] = load i32, i32* [[MASTER_REF]]
-// TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[TMAIN_T_VAR]]
+// TLS-CHECK: [[MASTER_VAL:%.+]] = load i32, i32* [[MASTER_REF]],
+// TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[TMAIN_T_VAR]], align 128
 
 // threadprivate_vec = vec;
 // CHECK: call {{.*}}i8* @__kmpc_threadprivate_cached({{.+}} [[TMAIN_VEC]]
 // CHECK: call {{.*}}void @llvm.memcpy{{.*}}(i8* %{{.+}}, i8* bitcast ([2 x i{{[0-9]+}}]* [[TMAIN_VEC]] to i8*),
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x i32]*, [2 x i32]** [[MASTER_FIELD]]
-// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x i32]* [[MASTER_REF]] to i8*
+// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x i32]* [[MASTER_REF1]] to i8*
 // TLS-CHECK: call void @llvm.memcpy{{.*}}(i8* bitcast ([2 x i{{[0-9]+}}]* [[TMAIN_VEC]] to i8*), i8* [[MASTER_CAST]]
 
 // threadprivate_s_arr = s_arr;
@@ -445,9 +412,7 @@ int main() {
 // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}})
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[MASTER_FIELD]]
-// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[MASTER_REF]] to [[S_INT_TY]]*
+// TLS-CHECK: [[MASTER_CAST:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[MASTER_REF2]] to [[S_INT_TY]]*
 // TLS-CHECK-DAG: [[S_ARR_SRC_BEGIN:%.+]] = phi [[S_INT_TY]]* {{.*}}[[MASTER_CAST]]
 // TLS-CHECK-DAG: [[S_ARR_DST_BEGIN:%.+]] = phi [[S_INT_TY]]* {{.*}}getelementptr inbounds ([2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[TMAIN_S_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0)
 // TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}})
@@ -462,24 +427,22 @@ int main() {
 // CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{%.+}}, [[S_INT_TY]]* {{.*}}[[TMAIN_VAR]])
 // CHECK: [[DONE]]
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[MASTER_FIELD]]
-// TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.*}}[[TMAIN_VAR]], [[S_INT_TY]]* {{.*}}[[MASTER_REF]])
+// TLS-CHECK: call {{.*}} [[S_INT_TY_COPY_ASSIGN]]([[S_INT_TY]]* {{.*}}[[TMAIN_VAR]], [[S_INT_TY]]* {{.*}}[[MASTER_REF3]])
 
-// CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // CHECK: ret void
 
 // TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
-// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// TLS-CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // TLS-CHECK: ret void
 
-// CHECK: define internal {{.*}}void [[TMAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// CHECK: define internal {{.*}}void [[TMAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
 
-// TLS-CHECK: define internal {{.*}}void [[TMAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, {{%.+}}* %{{.+}})
+// TLS-CHECK: define internal {{.*}}void [[TMAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // TLS-CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
 // threadprivate_t_var = t_var;
@@ -492,8 +455,8 @@ int main() {
 // CHECK: store i{{[0-9]+}} %{{.+}}, i{{[0-9]+}}* %{{.+}},
 // CHECK: [[DONE]]
 
-// TLS-CHECK: [[MASTER_FIELD:%.+]] = getelementptr inbounds {{.+}}
-// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** [[MASTER_FIELD]]
+// TLS-CHECK: [[MASTER_REF:%.+]] = load i32*, i32** %
+
 // TLS-CHECK: [[MASTER_LONG:%.+]] = ptrtoint i32* [[MASTER_REF]] to i{{[0-9]+}}
 // TLS-CHECK: icmp ne i{{[0-9]+}} [[MASTER_LONG]], ptrtoint (i{{[0-9]+}}* [[TMAIN_T_VAR]] to i{{[0-9]+}})
 // TLS-CHECK: br i1 %{{.+}}, label %[[NOT_MASTER:.+]], label %[[DONE:.+]]
@@ -502,18 +465,17 @@ int main() {
 // TLS-CHECK: store i32 [[MASTER_VAL]], i32* [[TMAIN_T_VAR]]
 // TLS-CHECK: [[DONE]]
 
-// CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // CHECK: ret void
 
 // TLS-CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_ADDR]],
 // TLS-CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]],
-// TLS-CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
+// TLS-CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i32 [[GTID]])
 // TLS-CHECK: ret void
 
 #endif
 #else
 // ARRAY-LABEL: array_func
-// TLS-ARRAY: [[ARR_CAPTURETY:%.+]] = type { [2 x {{.*}}]*, [2 x {{.*}}]* }
 // TLS-ARRAY-LABEL: array_func
 
 struct St {
@@ -531,15 +493,10 @@ void array_func() {
 // ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* bitcast ([2 x i32]* @{{.+}} to i8*), i64 8, i32 4, i1 false)
 // ARRAY: call dereferenceable(8) %struct.St* @{{.+}}(%struct.St* %{{.+}}, %struct.St* dereferenceable(8) %{{.+}})
 
-// TLS-ARRAY-DAG: [[ARR_CAP1:%.+]] = getelementptr inbounds [[ARR_CAPTURETY]], [[ARR_CAPTURETY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// TLS-ARRAY-DAG: [[ARR_CAP2:%.+]] = getelementptr inbounds [[ARR_CAPTURETY]], [[ARR_CAPTURETY]]* {{.*}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// TLS-ARRAY-DAG: store [2 x {{.*}}]* @{{.*}}, [2 x {{.*}}]** [[ARR_CAP1]]
-// TLS-ARRAY-DAG: store [2 x {{.*}}]* @{{.*}}, [2 x {{.*}}]** [[ARR_CAP2]]
 // TLS-ARRAY: @__kmpc_fork_call(
-// TLS-ARRAY-DAG: call void @llvm.memcpy.p0i8.p0i8.i64(i8* bitcast ([2 x i32]* @{{.+}} to i8*), i8* [[REF:%.+]], i64 8, i32 4, i1 false)
-// TLS-ARRAY-DAG: [[REF]] = bitcast [2 x i32]* [[REFT:%.+]] to i8*
-// TLS-ARRAY-DAG: [[REFT]] = load [2 x i32]*, [2 x i32]** [[FIELDADDR:%.+]],
-// TLS-ARRAY-DAG: [[FIELDADDR]] = getelementptr inbounds [[ARR_CAPTURETY]], [[ARR_CAPTURETY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}}
+// TLS-ARRAY: [[REFT:%.+]] = load [2 x i32]*, [2 x i32]** [[ADDR:%.+]],
+// TLS-ARRAY: [[REF:%.+]] = bitcast [2 x i32]* [[REFT]] to i8*
+// TLS-ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* bitcast ([2 x i32]* @{{.+}} to i8*), i8* [[REF]], i64 8, i32 4, i1 false)
 // TLS-ARRAY: call dereferenceable(8) %struct.St* @{{.+}}(%struct.St* %{{.+}}, %struct.St* dereferenceable(8) %{{.+}})
 
 #pragma omp threadprivate(a, s)
diff --git a/test/OpenMP/parallel_firstprivate_codegen.cpp b/test/OpenMP/parallel_firstprivate_codegen.cpp
index 6e0f014f4d9e..d0da8cea87ac 100644
--- a/test/OpenMP/parallel_firstprivate_codegen.cpp
+++ b/test/OpenMP/parallel_firstprivate_codegen.cpp
@@ -16,7 +16,7 @@ struct St {
   ~St() {}
 };
 
-volatile int g = 1212;
+volatile int g __attribute__((aligned(128))) = 1212;
 
 template <class T>
 struct S {
@@ -31,17 +31,15 @@ struct S {
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { [2 x i{{[0-9]+}}]*, i{{[0-9]+}}*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]* }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { [2 x i{{[0-9]+}}]*, i{{[0-9]+}}*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 
 template <typename T>
 T tmain() {
   S<T> test;
-  T t_var = T();
-  T vec[] = {1, 2};
-  S<T> s_arr[] = {1, 2};
-  S<T> var(3);
+  T t_var __attribute__((aligned(128))) = T();
+  T vec[] __attribute__((aligned(128))) = {1, 2};
+  S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
+  S<T> var __attribute__((aligned(128))) (3);
 #pragma omp parallel firstprivate(t_var, vec, s_arr, var)
   {
     vec[0] = t_var;
@@ -53,40 +51,46 @@ T tmain() {
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: [[G_LOCAL_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[AGG_CAPTURED:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-  // LAMBDA: store i{{[0-9]+}}* [[G]], i{{[0-9]+}}** [[G_LOCAL_REF]]
-  // LAMBDA: [[ARG:%.+]] = bitcast %{{.+}}* [[AGG_CAPTURED]] to i8*
-  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
-#pragma omp parallel firstprivate(g)
+  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 2, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]], {{.+}})
+#pragma omp parallel firstprivate(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
-    // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    // LAMBDA: [[ARG:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_REF]]
-    // LAMBDA: [[G_REF_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-    // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR]]
-    // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]]
-    // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
-    // LAMBDA: call {{.*}}i32 @__kmpc_cancel_barrier(
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
+    // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]]
+    // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_REF_ADDR:%.+]]
+    // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]], align 128
+    // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
+    // LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF]]
+    // LAMBDA: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
+    // LAMBDA: call {{.*}}void @__kmpc_barrier(
     g = 1;
+    sivar = 2;
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 4;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
-      // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -97,33 +101,42 @@ int main() {
   // BLOCKS: call {{.*}}void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: [[G_LOCAL_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[AGG_CAPTURED:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-  // BLOCKS: store i{{[0-9]+}}* [[G]], i{{[0-9]+}}** [[G_LOCAL_REF]]
-  // BLOCKS: [[ARG:%.+]] = bitcast %{{.+}}* [[AGG_CAPTURED]] to i8*
-  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
-#pragma omp parallel firstprivate(g)
+  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 2, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]], {{.+}})
+#pragma omp parallel firstprivate(g, sivar)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
-    // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    // BLOCKS: [[ARG:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_REF]]
-    // BLOCKS: [[G_REF_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-    // BLOCKS: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR]]
-    // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]]
-    // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
-    // BLOCKS: call {{.*}}i32 @__kmpc_cancel_barrier(
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
+    // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}}, align 128
+    // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]]
+    // BLOCKS: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_REF_ADDR:%.+]]
+    // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G_REF]], align 128
+    // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
+    // BLOCK: [[SIVAR_REF_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // BLOCK: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_REF_ADDR]]
+    // BLOCKS: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF]],
+    // BLOCKS: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
+    // BLOCKS: call {{.*}}void @__kmpc_barrier(
     g = 1;
+    sivar = 2;
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 4;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -135,10 +148,11 @@ int main() {
   int vec[] = {1, 2};
   S<float> s_arr[] = {1, 2};
   S<float> var(3);
-#pragma omp parallel firstprivate(t_var, vec, s_arr, var)
+#pragma omp parallel firstprivate(t_var, vec, s_arr, var, sivar)
   {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 2;
   }
 #pragma omp parallel firstprivate(t_var)
   {}
@@ -149,29 +163,29 @@ int main() {
 // CHECK: define {{.*}}i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i32*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal {{.*}}void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x [[S_FLOAT_TY]]]* dereferenceable(8) %{{.+}}, [[S_FLOAT_TY]]* dereferenceable(4) %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]])
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR7_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
+
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+// CHECK: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]],
 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]],
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
-// CHECK: [[S_ARR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_FLOAT_TY]]]*, [2 x [[S_FLOAT_TY]]]** [[S_ARR_REF_PTR]],
 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_BEGIN:%.+]] = bitcast [2 x [[S_FLOAT_TY]]]* [[S_ARR_REF]] to [[S_FLOAT_TY]]*
 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_FLOAT_TY]], [[S_FLOAT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2
@@ -182,42 +196,41 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR:@.+]]([[S_FLOAT_TY]]* {{.+}}, [[S_FLOAT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR:@.+]]([[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_REF_PTR]],
 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]])
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
-// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
-// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+
+// CHECK: [[SIVAR_REF_ADDR:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF]],
+// CHECK: store i{{[0-9]+}} [[SIVAR_REF_ADDR]], i{{[0-9]+}}* [[SIVAR7_PRIV]],
+// CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR7_PRIV]],
+
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
 // CHECK: ret void
-
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call {{.*}}void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i32*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
-// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
-// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
-// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
+// CHECK: define internal {{.*}}void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
+// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
+// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128
+// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128
+// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
-// CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]],
-// CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]],
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
+
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+
+// CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]], align 128
+// CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]], align 128
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
-// CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
-// CHECK: [[S_ARR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF_PTR]],
+// CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]], i{{[0-9]+}} {{[0-9]+}}, i{{[0-9]+}} 128,
 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_BEGIN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_REF]] to [[S_INT_TY]]*
 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2
@@ -228,44 +241,62 @@ int main() {
 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR:@.+]]([[S_INT_TY]]* {{.+}}, [[S_INT_TY]]* {{.+}}, [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]],
 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]])
 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call {{.*}}i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call {{.*}}void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]*
 // CHECK: ret void
 
 #endif
 #else
-// ARRAY-LABEL: array_func
 struct St {
   int a, b;
   St() : a(0), b(0) {}
   St(const St &) { }
   ~St() {}
+  void St_func(St s[2], int n, long double vla1[n]) {
+    double vla2[n][n] __attribute__((aligned(128)));
+    a = b;
+#pragma omp parallel firstprivate(s, vla1, vla2)
+    vla1[b] = vla2[1][n - 1] = a = b;
+  }
 };
 
+// ARRAY-LABEL: array_func
 void array_func(float a[3], St s[2], int n, long double vla1[n]) {
-  double vla2[n];
+  double vla2[n][n] __attribute__((aligned(128)));
 // ARRAY: @__kmpc_fork_call(
-// ARRAY: [[PRIV_A:%.+]] = alloca float*
-// ARRAY: [[PRIV_S:%.+]] = alloca %struct.St*
-// ARRAY: [[PRIV_VLA1:%.+]] = alloca x86_fp80*
-// ARRAY: store float* %{{.+}}, float** [[PRIV_A]],
-// ARRAY: store %struct.St* %{{.+}}, %struct.St** [[PRIV_S]],
-// ARRAY: store x86_fp80* %{{.+}}, x86_fp80** [[PRIV_VLA1]],
+// ARRAY-DAG: [[PRIV_A:%.+]] = alloca float**,
+// ARRAY-DAG: [[PRIV_S:%.+]] = alloca %struct.St**,
+// ARRAY-DAG: [[PRIV_VLA1:%.+]] = alloca x86_fp80**,
+// ARRAY-DAG: [[PRIV_VLA2:%.+]] = alloca double*,
+// ARRAY-DAG: store float** %{{.+}}, float*** [[PRIV_A]],
+// ARRAY-DAG: store %struct.St** %{{.+}}, %struct.St*** [[PRIV_S]],
+// ARRAY-DAG: store x86_fp80** %{{.+}}, x86_fp80*** [[PRIV_VLA1]],
+// ARRAY-DAG: store double* %{{.+}}, double** [[PRIV_VLA2]],
 // ARRAY: call i8* @llvm.stacksave()
 // ARRAY: [[SIZE:%.+]] = mul nuw i64 %{{.+}}, 8
-// ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* %{{.+}}, i64 [[SIZE]], i32 8, i1 false)
-// ARRAY: call void @llvm.stackrestore(i8*
+// ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* %{{.+}}, i64 [[SIZE]], i32 128, i1 false)
 #pragma omp parallel firstprivate(a, s, vla1, vla2)
+  s[0].St_func(s, n, vla1);
   ;
 }
+
+// ARRAY-LABEL: St_func
+// ARRAY: @__kmpc_fork_call(
+// ARRAY-DAG: [[PRIV_S:%.+]] = alloca %struct.St**,
+// ARRAY-DAG: [[PRIV_VLA1:%.+]] = alloca x86_fp80**,
+// ARRAY-DAG: [[PRIV_VLA2:%.+]] = alloca double*,
+// ARRAY-DAG: store %struct.St** %{{.+}}, %struct.St*** [[PRIV_S]],
+// ARRAY-DAG: store x86_fp80** %{{.+}}, x86_fp80*** [[PRIV_VLA1]],
+// ARRAY-DAG: store double* %{{.+}}, double** [[PRIV_VLA2]],
+// ARRAY: call i8* @llvm.stacksave()
+// ARRAY: [[SIZE:%.+]] = mul nuw i64 %{{.+}}, 8
+// ARRAY: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %{{.+}}, i8* %{{.+}}, i64 [[SIZE]], i32 128, i1 false)
 #endif
 
 
diff --git a/test/OpenMP/parallel_firstprivate_messages.cpp b/test/OpenMP/parallel_firstprivate_messages.cpp
index c6f8dbe138d5..fc0eb4c7d358 100644
--- a/test/OpenMP/parallel_firstprivate_messages.cpp
+++ b/test/OpenMP/parallel_firstprivate_messages.cpp
@@ -61,7 +61,8 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
+  static int m;
   #pragma omp parallel firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   #pragma omp parallel firstprivate ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel firstprivate () // expected-error {{expected expression}}
@@ -83,7 +84,8 @@ int main(int argc, char **argv) {
   foo();
   #pragma omp parallel shared(i)
   #pragma omp parallel firstprivate(i)
-  #pragma omp parallel firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+  #pragma omp parallel firstprivate(j)
+  #pragma omp parallel firstprivate(m)
   foo();
 
   return 0;
diff --git a/test/OpenMP/parallel_for_ast_print.cpp b/test/OpenMP/parallel_for_ast_print.cpp
index f2899ee16ef5..c4be521455df 100644
--- a/test/OpenMP/parallel_for_ast_print.cpp
+++ b/test/OpenMP/parallel_for_ast_print.cpp
@@ -15,25 +15,35 @@ T tmain(T argc) {
 // CHECK: static T a;
   static T g;
 #pragma omp threadprivate(g)
-#pragma omp parallel for schedule(dynamic) default(none) copyin(g)
-  // CHECK: #pragma omp parallel for schedule(dynamic) default(none) copyin(g)
+#pragma omp parallel for schedule(dynamic) default(none) copyin(g) linear(a)
+  // CHECK: #pragma omp parallel for schedule(dynamic) default(none) copyin(g) linear(a)
   for (int i = 0; i < 2; ++i)
     a = 2;
 // CHECK-NEXT: for (int i = 0; i < 2; ++i)
 // CHECK-NEXT: a = 2;
-#pragma omp parallel for private(argc, b), firstprivate(c, d), lastprivate(d, f) collapse(N) schedule(static, N) ordered if (argc) num_threads(N) default(shared) shared(e) reduction(+ : h)
-  for (int i = 0; i < 10; ++i)
-    for (int j = 0; j < 10; ++j)
-      for (int j = 0; j < 10; ++j)
-        for (int j = 0; j < 10; ++j)
-          for (int j = 0; j < 10; ++j)
+#pragma omp parallel for private(argc, b), firstprivate(c, d), lastprivate(d, f) collapse(N) schedule(static, N) ordered(N) if (parallel :argc) num_threads(N) default(shared) shared(e) reduction(+ : h)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
             foo();
-  // CHECK-NEXT: #pragma omp parallel for private(argc,b) firstprivate(c,d) lastprivate(d,f) collapse(N) schedule(static, N) ordered if(argc) num_threads(N) default(shared) shared(e) reduction(+: h)
-  // CHECK-NEXT: for (int i = 0; i < 10; ++i)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
-  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
+  // CHECK-NEXT: #pragma omp parallel for private(argc,b) firstprivate(c,d) lastprivate(d,f) collapse(N) schedule(static, N) ordered(N) if(parallel: argc) num_threads(N) default(shared) shared(e) reduction(+: h)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
   // CHECK-NEXT: foo();
   return T();
 }
@@ -44,18 +54,18 @@ int main(int argc, char **argv) {
 // CHECK: static int a;
   static float g;
 #pragma omp threadprivate(g)
-#pragma omp parallel for schedule(guided, argc) default(none) copyin(g)
-  // CHECK: #pragma omp parallel for schedule(guided, argc) default(none) copyin(g)
+#pragma omp parallel for schedule(guided, argc) default(none) copyin(g) linear(a)
+  // CHECK: #pragma omp parallel for schedule(guided, argc) default(none) copyin(g) linear(a)
   for (int i = 0; i < 2; ++i)
     a = 2;
 // CHECK-NEXT: for (int i = 0; i < 2; ++i)
 // CHECK-NEXT: a = 2;
-#pragma omp parallel for private(argc, b), firstprivate(argv, c), lastprivate(d, f) collapse(2) schedule(auto) ordered if (argc) num_threads(a) default(shared) shared(e) reduction(+ : h)
+#pragma omp parallel for private(argc, b), firstprivate(argv, c), lastprivate(d, f) collapse(2) schedule(auto) ordered if (argc) num_threads(a) default(shared) shared(e) reduction(+ : h) linear(a:-5)
   for (int i = 0; i < 10; ++i)
     for (int j = 0; j < 10; ++j)
       foo();
-  // CHECK-NEXT: #pragma omp parallel for private(argc,b) firstprivate(argv,c) lastprivate(d,f) collapse(2) schedule(auto) ordered if(argc) num_threads(a) default(shared) shared(e) reduction(+: h)
-  // CHECK-NEXT: for (int i = 0; i < 10; ++i)
+  // CHECK-NEXT: #pragma omp parallel for private(argc,b) firstprivate(argv,c) lastprivate(d,f) collapse(2) schedule(auto) ordered if(argc) num_threads(a) default(shared) shared(e) reduction(+: h) linear(a: -5)
+ // CHECK-NEXT: for (int i = 0; i < 10; ++i)
   // CHECK-NEXT: for (int j = 0; j < 10; ++j)
   // CHECK-NEXT: foo();
   return (tmain<int, 5>(argc) + tmain<char, 1>(argv[0][0]));
diff --git a/test/OpenMP/parallel_for_codegen.cpp b/test/OpenMP/parallel_for_codegen.cpp
index 9fda8404c6e7..036fa37cf91b 100644
--- a/test/OpenMP/parallel_for_codegen.cpp
+++ b/test/OpenMP/parallel_for_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -O1 -fopenmp -emit-llvm %s -o - | FileCheck %s --check-prefix=CLEANUP
 // REQUIRES: x86-registered-target
 // expected-no-diagnostics
@@ -9,25 +9,19 @@
 #define HEADER
 
 // CHECK-DAG: [[IDENT_T_TY:%.+]] = type { i32, i32, i32, i32, i8* }
-// CHECK-DAG: [[CAP_TY:%.+]] = type { i8* }
 
 // CHECK-LABEL: with_var_schedule
 void with_var_schedule() {
   double a = 5;
 // CHECK: [[CHUNK_SIZE:%.+]] = fptosi double %{{.+}}to i8
 // CHECK: store i8 %{{.+}}, i8* [[CHUNK:%.+]],
-// CHECK: [[CHUNK_REF:%.+]] = getelementptr inbounds [[CAP_TY]], [[CAP_TY]]* [[CAP_ARG:%.+]], i{{.+}} 0, i{{.+}} 0
-// CHECK: store i8* [[CHUNK]], i8** [[CHUNK_REF]],
-// CHECK: [[BITCAST:%.+]] = bitcast [[CAP_TY]]* [[CAP_ARG]] to i8*
-// CHECK: call void {{.+}} @__kmpc_fork_call({{.+}}, i8* [[BITCAST]])
+// CHECK: call void {{.+}} @__kmpc_fork_call({{.+}}, i8* [[CHUNK]])
 
-// CHECK: [[CHUNK_REF:%.+]] = getelementptr inbounds [[CAP_TY]], [[CAP_TY]]* %{{.+}}, i{{.+}} 0, i{{.+}} 0
-// CHECK: [[CHUNK:%.+]] = load i8*, i8** [[CHUNK_REF]],
+// CHECK: [[CHUNK:%.+]] = load i8*, i8** %
 // CHECK: [[CHUNK_VAL:%.+]] = load i8, i8* [[CHUNK]],
 // CHECK: [[CHUNK_SIZE:%.+]] = sext i8 [[CHUNK_VAL]] to i64
 // CHECK: call void @__kmpc_for_static_init_8u([[IDENT_T_TY]]* [[DEFAULT_LOC:@[^,]+]], i32 [[GTID:%[^,]+]], i32 33, i32* [[IS_LAST:%[^,]+]], i64* [[OMP_LB:%[^,]+]], i64* [[OMP_UB:%[^,]+]], i64* [[OMP_ST:%[^,]+]], i64 1, i64 [[CHUNK_SIZE]])
 // CHECK: call void @__kmpc_for_static_fini([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]])
-// CHECK: __kmpc_barrier
 #pragma omp parallel for schedule(static, char(a))
   for (unsigned long long i = 1; i < 2; ++i) {
   }
@@ -36,8 +30,8 @@ void with_var_schedule() {
 // CHECK-LABEL: define {{.*void}} @{{.*}}without_schedule_clause{{.*}}(float* {{.+}}, float* {{.+}}, float* {{.+}}, float* {{.+}})
 void without_schedule_clause(float *a, float *b, float *c, float *d) {
   #pragma omp parallel for
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -73,15 +67,14 @@ void without_schedule_clause(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[LOOP1_END]]
 // CHECK: call void @__kmpc_for_static_fini([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]])
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
 // CHECK-LABEL: define {{.*void}} @{{.*}}static_not_chunked{{.*}}(float* {{.+}}, float* {{.+}}, float* {{.+}}, float* {{.+}})
 void static_not_chunked(float *a, float *b, float *c, float *d) {
   #pragma omp parallel for schedule(static)
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -117,15 +110,14 @@ void static_not_chunked(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[LOOP1_END]]
 // CHECK: call void @__kmpc_for_static_fini([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]])
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
 // CHECK-LABEL: define {{.*void}} @{{.*}}static_chunked{{.*}}(float* {{.+}}, float* {{.+}}, float* {{.+}}, float* {{.+}})
 void static_chunked(float *a, float *b, float *c, float *d) {
   #pragma omp parallel for schedule(static, 5)
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -180,15 +172,14 @@ void static_chunked(float *a, float *b, float *c, float *d) {
 
 // CHECK: [[O_LOOP1_END]]
 // CHECK: call void @__kmpc_for_static_fini([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 [[GTID]])
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
 // CHECK-LABEL: define {{.*void}} @{{.*}}dynamic1{{.*}}(float* {{.+}}, float* {{.+}}, float* {{.+}}, float* {{.+}})
 void dynamic1(float *a, float *b, float *c, float *d) {
   #pragma omp parallel for schedule(dynamic)
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -225,15 +216,14 @@ void dynamic1(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[LOOP1_END]]
 // CHECK: [[O_LOOP1_END]]
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
 // CHECK-LABEL: define {{.*void}} @{{.*}}guided7{{.*}}(float* {{.+}}, float* {{.+}}, float* {{.+}}, float* {{.+}})
 void guided7(float *a, float *b, float *c, float *d) {
   #pragma omp parallel for schedule(guided, 7)
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -270,7 +260,6 @@ void guided7(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[LOOP1_END]]
 // CHECK: [[O_LOOP1_END]]
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
@@ -279,8 +268,8 @@ void test_auto(float *a, float *b, float *c, float *d) {
   unsigned int x = 0;
   unsigned int y = 0;
   #pragma omp parallel for schedule(auto) collapse(2)
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, i32* dereferenceable(4) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -320,9 +309,6 @@ void test_auto(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[LOOP1_END]]
 // CHECK: [[O_LOOP1_END]]
-// CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
-// CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
@@ -330,8 +316,8 @@ void test_auto(float *a, float *b, float *c, float *d) {
 void runtime(float *a, float *b, float *c, float *d) {
   int x = 0;
   #pragma omp parallel for collapse(2) schedule(runtime)
-// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: call void ([[IDENT_T_TY]]*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call([[IDENT_T_TY]]* [[DEFAULT_LOC:[@%].+]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, float**, float**, float**, float**)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*),
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}}, float** dereferenceable(8) %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 // CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
@@ -367,9 +353,6 @@ void runtime(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[LOOP1_END]]
 // CHECK: [[O_LOOP1_END]]
-// CHECK: [[GTID_REF:%.+]] = load i32*, i32** [[GTID_REF_ADDR]],
-// CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_REF]],
-// CHECK: call {{.+}} @__kmpc_barrier([[IDENT_T_TY]]* [[DEFAULT_LOC_BARRIER:[@%].+]], i32 [[GTID]])
 // CHECK: ret void
 }
 
@@ -378,29 +361,27 @@ int foo() {return 0;};
 
 // TERM_DEBUG-LABEL: parallel_for
 // CLEANUP: parallel_for
-void parallel_for(float *a) {
-#pragma omp parallel for schedule(static, 5)
+void parallel_for(float *a, int n) {
+  float arr[n];
+#pragma omp parallel for schedule(static, 5) private(arr)
   // TERM_DEBUG-NOT: __kmpc_global_thread_num
   // TERM_DEBUG:     call void @__kmpc_for_static_init_4u({{.+}}), !dbg [[DBG_LOC_START:![0-9]+]]
   // TERM_DEBUG:     invoke i32 {{.*}}foo{{.*}}()
   // TERM_DEBUG:     unwind label %[[TERM_LPAD:.+]],
   // TERM_DEBUG-NOT: __kmpc_global_thread_num
   // TERM_DEBUG:     call void @__kmpc_for_static_fini({{.+}}), !dbg [[DBG_LOC_END:![0-9]+]]
-  // TERM_DEBUG:     call {{.+}} @__kmpc_barrier({{.+}}), !dbg [[DBG_LOC_CANCEL:![0-9]+]]
   // TERM_DEBUG:     [[TERM_LPAD]]
   // TERM_DEBUG:     call void @__clang_call_terminate
   // TERM_DEBUG:     unreachable
   // CLEANUP-NOT: __kmpc_global_thread_num
   // CLEANUP:     call void @__kmpc_for_static_init_4u({{.+}})
   // CLEANUP:     call void @__kmpc_for_static_fini({{.+}})
-  // CLEANUP:     call {{.+}} @__kmpc_barrier({{.+}})
   for (unsigned i = 131071; i <= 2147483647; i += 127)
-    a[i] += foo();
+    a[i] += foo() + arr[i];
 }
 // Check source line corresponds to "#pragma omp parallel for schedule(static, 5)" above:
 // TERM_DEBUG-DAG: [[DBG_LOC_START]] = !DILocation(line: [[@LINE-4]],
-// TERM_DEBUG-DAG: [[DBG_LOC_END]] = !DILocation(line: [[@LINE-20]],
-// TERM_DEBUG-DAG: [[DBG_LOC_CANCEL]] = !DILocation(line: [[@LINE-21]],
+// TERM_DEBUG-DAG: [[DBG_LOC_END]] = !DILocation(line: [[@LINE-18]],
 
 #endif // HEADER
 
diff --git a/test/OpenMP/parallel_for_collapse_messages.cpp b/test/OpenMP/parallel_for_collapse_messages.cpp
index 042b819fe58e..6e5f71ff1672 100644
--- a/test/OpenMP/parallel_for_collapse_messages.cpp
+++ b/test/OpenMP/parallel_for_collapse_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp parallel for collapse (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp parallel for collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp parallel for collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp parallel for', but found only 1}}
   // expected-error@+3 2 {{directive '#pragma omp parallel for' cannot contain more than one 'collapse' clause}}
-  // expected-error@+2 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp parallel for collapse (foobool(argc)), collapse (true), collapse (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for collapse (1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for collapse (N) // expected-error {{argument to 'collapse' clause must be a positive integer value}}
+  #pragma omp parallel for collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for collapse (2) // expected-note {{as specified in 'collapse' clause}}
   foo(); // expected-error {{expected 2 for loops after '#pragma omp parallel for'}}
@@ -63,7 +63,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp parallel for' cannot contain more than one 'collapse' clause}}
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp parallel for collapse (foobool(argc)), collapse (true), collapse (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for collapse (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/parallel_for_firstprivate_messages.cpp b/test/OpenMP/parallel_for_firstprivate_messages.cpp
index 2c762b49b39e..c9a69dab2cb2 100644
--- a/test/OpenMP/parallel_for_firstprivate_messages.cpp
+++ b/test/OpenMP/parallel_for_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -102,9 +102,6 @@ int foomain(int argc, char **argv) {
 #pragma omp parallel for firstprivate(h) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
   for (int k = 0; k < argc; ++k)
     ++k;
-#pragma omp parallel for linear(i) // expected-error {{unexpected OpenMP clause 'linear' in directive '#pragma omp parallel for'}}
-  for (int k = 0; k < argc; ++k)
-    ++k;
 #pragma omp parallel
   {
     int v = 0;
@@ -117,7 +114,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp parallel for firstprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for firstprivate(i)
@@ -153,7 +150,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (i = 0; i < argc; ++i)
     foo();
@@ -228,7 +225,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for firstprivate(xa) // OK: may be firstprivate
   for (i = 0; i < argc; ++i)
     foo();
-#pragma omp parallel for firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp parallel for firstprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel for lastprivate(g) firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
@@ -255,6 +252,10 @@ int main(int argc, char **argv) {
 #pragma omp parallel for firstprivate(i) // expected-note {{defined as firstprivate}}
   for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp parallel for' directive may not be firstprivate, predetermined as private}}
     foo();
+  static int si;
+#pragma omp parallel for firstprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/parallel_for_if_messages.cpp b/test/OpenMP/parallel_for_if_messages.cpp
index cf2a3b431dc7..2b7a5f7e2b3e 100644
--- a/test/OpenMP/parallel_for_if_messages.cpp
+++ b/test/OpenMP/parallel_for_if_messages.cpp
@@ -34,6 +34,20 @@ int tmain(T argc, S **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for if(argc)
   for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel for'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel for' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  for (i = 0; i < argc; ++i) foo();
 
   return 0;
 }
@@ -64,6 +78,20 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for if(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel for'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel for' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  for (i = 0; i < argc; ++i) foo();
 
   return tmain(argc, argv);
 }
diff --git a/test/OpenMP/parallel_for_lastprivate_messages.cpp b/test/OpenMP/parallel_for_lastprivate_messages.cpp
index 09bed02a05f9..ccfe2ea407b1 100644
--- a/test/OpenMP/parallel_for_lastprivate_messages.cpp
+++ b/test/OpenMP/parallel_for_lastprivate_messages.cpp
@@ -17,7 +17,7 @@ public:
   S2(S2 &s2) : a(s2.a) {}
   S2 &operator=(const S2 &);
   const S2 &operator=(const S2 &) const;
-  static float S2s;
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note {{static data member is predetermined as shared}}
@@ -67,7 +67,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                        // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -104,9 +104,6 @@ int foomain(int argc, char **argv) {
 #pragma omp parallel for lastprivate(h) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
   for (int k = 0; k < argc; ++k)
     ++k;
-#pragma omp parallel for linear(i) // expected-error {{unexpected OpenMP clause 'linear' in directive '#pragma omp parallel for'}}
-  for (int k = 0; k < argc; ++k)
-    ++k;
 #pragma omp parallel
   {
     int v = 0;
@@ -119,7 +116,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp parallel for lastprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for lastprivate(i)
@@ -144,7 +141,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i;                        // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (i = 0; i < argc; ++i)
     foo();
@@ -191,7 +188,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for lastprivate(xa) // OK
   for (i = 0; i < argc; ++i)
     foo();
-#pragma omp parallel for lastprivate(S2::S2s)
+#pragma omp parallel for lastprivate(S2::S2s) // expected-error {{shared variable cannot be lastprivate}}
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel for lastprivate(S2::S2sc) // expected-error {{shared variable cannot be lastprivate}}
@@ -223,7 +220,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for lastprivate(xa)
   for (i = 0; i < argc; ++i)
     foo();
-#pragma omp parallel for lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp parallel for lastprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel for firstprivate(m) lastprivate(m) // expected-error {{'operator=' is a private member of 'S3'}}
@@ -232,5 +229,10 @@ int main(int argc, char **argv) {
 #pragma omp parallel for lastprivate(n) firstprivate(n) // OK
   for (i = 0; i < argc; ++i)
     foo();
+  static int si;
+#pragma omp parallel for lastprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 2;
+
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/parallel_for_linear_codegen.cpp b/test/OpenMP/parallel_for_linear_codegen.cpp
new file mode 100644
index 000000000000..940d6032a14a
--- /dev/null
+++ b/test/OpenMP/parallel_for_linear_codegen.cpp
@@ -0,0 +1,249 @@
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=LAMBDA %s
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=BLOCKS %s
+// expected-no-diagnostics
+// REQUIRES: x86-registered-target
+#ifndef HEADER
+#define HEADER
+
+template <class T>
+struct S {
+  T f;
+  S(T a) : f(a) {}
+  S() : f() {}
+  S<T> &operator=(const S<T> &);
+  operator T() { return T(); }
+  ~S() {}
+};
+
+volatile int g = 1212;
+float f;
+char cnt;
+
+// CHECK: [[S_FLOAT_TY:%.+]] = type { float }
+// CHECK: [[S_INT_TY:%.+]] = type { i32 }
+// CHECK-DAG: [[F:@.+]] = global float 0.0
+// CHECK-DAG: [[CNT:@.+]] = global i8 0
+template <typename T>
+T tmain() {
+  S<T> test;
+  T *pvar = &test.f;
+  T lvar = T();
+#pragma omp parallel for linear(pvar, lvar)
+  for (int i = 0; i < 2; ++i) {
+    ++pvar, ++lvar;
+  }
+  return T();
+}
+
+int main() {
+#ifdef LAMBDA
+  // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
+  // LAMBDA-LABEL: @main
+  // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
+  [&]() {
+  // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]])
+#pragma omp parallel for linear(g:5)
+  for (int i = 0; i < 2; ++i) {
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: alloca i{{[0-9]+}},
+    // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: store i32 0,
+    // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
+    // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+    // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+    // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]]
+    // LAMBDA: [[CNT:%.+]] = load i32, i32*
+    // LAMBDA: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5
+    // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]]
+    // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // LAMBDA: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]],
+    // LAMBDA: [[ADD:%.+]] = add nsw i32 [[VAL]], 5
+    // LAMBDA: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
+    // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
+    // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
+    g += 5;
+    [&]() {
+      // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
+      // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
+      g = 2;
+      // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
+      // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
+      // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+    }();
+  }
+  }();
+  return 0;
+#elif defined(BLOCKS)
+  // BLOCKS: [[G:@.+]] = global i{{[0-9]+}} 1212,
+  // BLOCKS-LABEL: @main
+  // BLOCKS: call void {{%.+}}(i8
+  ^{
+  // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]])
+#pragma omp parallel for linear(g:5)
+  for (int i = 0; i < 2; ++i) {
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: [[G_START_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: alloca i{{[0-9]+}},
+    // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: store i32 0,
+    // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
+    // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
+    // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+    // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_START_ADDR]]
+    // BLOCKS: [[CNT:%.+]] = load i32, i32*
+    // BLOCKS: [[MUL:%.+]] = mul nsw i32 [[CNT]], 5
+    // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], [[MUL]]
+    // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // BLOCKS: [[VAL:%.+]] = load i32, i32* [[G_PRIVATE_ADDR]],
+    // BLOCKS: [[ADD:%.+]] = add nsw i32 [[VAL]], 5
+    // BLOCKS: store i32 [[ADD]], i32* [[G_PRIVATE_ADDR]],
+    // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: call void {{%.+}}(i8
+    // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
+    g += 5;
+    g = 1;
+    ^{
+      // BLOCKS: define {{.+}} void {{@.+}}(i8*
+      g = 2;
+      // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS: ret
+    }();
+  }
+  }();
+  return 0;
+#else
+  S<float> test;
+  float *pvar = &test.f;
+  long long lvar = 0;
+#pragma omp parallel for linear(pvar, lvar : 3)
+  for (int i = 0; i < 2; ++i) {
+    pvar += 3, lvar += 3;
+  }
+  return tmain<int>();
+#endif
+}
+
+// CHECK: define i{{[0-9]+}} @main()
+// CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, float**, i64*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: = call {{.+}} [[TMAIN_INT:@.+]]()
+// CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
+// CHECK: ret
+
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, float** dereferenceable(8) %{{.+}}, i64* dereferenceable(8) %{{.+}})
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_START:%.+]] = alloca float*,
+// CHECK: [[LVAR_START:%.+]] = alloca i64,
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_PRIV:%.+]] = alloca float*,
+// CHECK: [[LVAR_PRIV:%.+]] = alloca i64,
+// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
+
+// CHECK: [[PVAR_REF:%.+]] = load float**, float*** %
+// CHECK: [[LVAR_REF:%.+]] = load i64*, i64** %
+
+// Check for default initialization.
+// CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_REF]],
+// CHECK: store float* [[PVAR_VAL]], float** [[PVAR_START]],
+// CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_REF]],
+// CHECK: store i64 [[LVAR_VAL]], i64* [[LVAR_START]],
+// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+// CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3
+// CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64
+// CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 [[IDX]]
+// CHECK: store float* [[PTR]], float** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 3
+// CHECK: [[CONV:%.+]] = sext i32 [[MUL]] to i64
+// CHECK: [[VAL:%.+]] = add nsw i64 [[LVAR_VAL]], [[CONV]]
+// CHECK: store i64 [[VAL]], i64* [[LVAR_PRIV]],
+// CHECK: [[PVAR_VAL:%.+]] = load float*, float** [[PVAR_PRIV]]
+// CHECK: [[PTR:%.+]] = getelementptr inbounds float, float* [[PVAR_VAL]], i64 3
+// CHECK: store float* [[PTR]], float** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i64, i64* [[LVAR_PRIV]],
+// CHECK: [[ADD:%.+]] = add nsw i64 [[LVAR_VAL]], 3
+// CHECK: store i64 [[ADD]], i64* [[LVAR_PRIV]],
+// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
+// CHECK: ret void
+
+// CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
+// CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
+// CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 2, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32**, i32*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
+// CHECK: ret
+//
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32** dereferenceable(8) %{{.+}}, i32* dereferenceable(4) %{{.+}})
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_START:%.+]] = alloca i32*,
+// CHECK: [[LVAR_START:%.+]] = alloca i32,
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: alloca i{{[0-9]+}},
+// CHECK: [[PVAR_PRIV:%.+]] = alloca i32*,
+// CHECK: [[LVAR_PRIV:%.+]] = alloca i32,
+// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
+
+// CHECK: [[PVAR_REF:%.+]] = load i32**, i32*** %
+// CHECK: [[LVAR_REF:%.+]] = load i32*, i32** %
+
+// Check for default initialization.
+// CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_REF]],
+// CHECK: store i32* [[PVAR_VAL]], i32** [[PVAR_START]],
+// CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_REF]],
+// CHECK: store i32 [[LVAR_VAL]], i32* [[LVAR_START]],
+// CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID:%.+]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+// CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1
+// CHECK: [[IDX:%.+]] = sext i32 [[MUL]] to i64
+// CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i64 [[IDX]]
+// CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_START]],
+// CHECK: [[CNT:%.+]] = load i32, i32*
+// CHECK: [[MUL:%.+]] = mul nsw i32 [[CNT]], 1
+// CHECK: [[VAL:%.+]] = add nsw i32 [[LVAR_VAL]], [[MUL]]
+// CHECK: store i32 [[VAL]], i32* [[LVAR_PRIV]],
+// CHECK: [[PVAR_VAL:%.+]] = load i32*, i32** [[PVAR_PRIV]]
+// CHECK: [[PTR:%.+]] = getelementptr inbounds i32, i32* [[PVAR_VAL]], i32 1
+// CHECK: store i32* [[PTR]], i32** [[PVAR_PRIV]],
+// CHECK: [[LVAR_VAL:%.+]] = load i32, i32* [[LVAR_PRIV]],
+// CHECK: [[ADD:%.+]] = add nsw i32 [[LVAR_VAL]], 1
+// CHECK: store i32 [[ADD]], i32* [[LVAR_PRIV]],
+// CHECK: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 %{{.+}})
+// CHECK: ret void
+#endif
+
diff --git a/test/OpenMP/parallel_for_linear_messages.cpp b/test/OpenMP/parallel_for_linear_messages.cpp
new file mode 100644
index 000000000000..7272aada92a0
--- /dev/null
+++ b/test/OpenMP/parallel_for_linear_messages.cpp
@@ -0,0 +1,269 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+namespace X {
+int x;
+};
+
+struct B {
+  static int ib; // expected-note {{'B::ib' declared here}}
+  static int bfoo() { return 8; }
+};
+
+int bfoo() { return 4; }
+
+int z;
+const int C1 = 1;
+const int C2 = 2;
+void test_linear_colons() {
+  int B = 0;
+#pragma omp parallel for linear(B : bfoo())
+  for (int i = 0; i < 10; ++i)
+    ;
+// expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'}}
+#pragma omp parallel for linear(B::ib : B : bfoo())
+  for (int i = 0; i < 10; ++i)
+    ;
+// expected-error@+1 {{use of undeclared identifier 'ib'; did you mean 'B::ib'}}
+#pragma omp parallel for linear(B : ib)
+  for (int i = 0; i < 10; ++i)
+    ;
+// expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'?}}
+#pragma omp parallel for linear(z : B : ib)
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp parallel for linear(B : B::bfoo())
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp parallel for linear(X::x : ::z)
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp parallel for linear(B, ::z, X::x)
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp parallel for linear(::z)
+  for (int i = 0; i < 10; ++i)
+    ;
+// expected-error@+1 {{expected variable name}}
+#pragma omp parallel for linear(B::bfoo())
+  for (int i = 0; i < 10; ++i)
+    ;
+#pragma omp parallel for linear(B::ib, B : C1 + C2)
+  for (int i = 0; i < 10; ++i)
+    ;
+}
+
+template <int L, class T, class N>
+T test_template(T *arr, N num) {
+  N i;
+  T sum = (T)0;
+  T ind2 = -num * L; // expected-note {{'ind2' defined here}}
+// expected-error@+1 {{argument of a linear clause should be of integral or pointer type}}
+#pragma omp parallel for linear(ind2 : L)
+  for (i = 0; i < num; ++i) {
+    T cur = arr[(int)ind2];
+    ind2 += L;
+    sum += cur;
+  }
+  return T();
+}
+
+template <int LEN>
+int test_warn() {
+  int ind2 = 0;
+// expected-warning@+1 {{zero linear step (ind2 should probably be const)}}
+#pragma omp parallel for linear(ind2 : LEN)
+  for (int i = 0; i < 100; i++) {
+    ind2 += LEN;
+  }
+  return ind2;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+};
+const S2 b; // expected-note 2 {{'b' defined here}}
+const S2 ba[5];
+class S3 {
+  int a;
+
+public:
+  S3() : a(0) {}
+};
+const S3 ca[5];
+class S4 {
+  int a;
+  S4();
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5() : a(0) {}
+
+public:
+  S5(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(I argc, C **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+#pragma omp parallel for linear // expected-error {{expected '(' after 'linear'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc : 5)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+// expected-error@+2 {{linear variable with incomplete type 'S1'}}
+// expected-error@+1 {{const-qualified variable cannot be linear}}
+#pragma omp parallel for linear(a, b : B::ib)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(e, g)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(h) // expected-error {{threadprivate or thread local variable cannot be linear}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp parallel for linear(v : i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel for linear(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  int v = 0;
+#pragma omp parallel for linear(v : j)
+  for (int k = 0; k < argc; ++k) {
+    ++k;
+    v += j;
+  }
+#pragma omp parallel for linear(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  return 0;
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace C {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  double darr[100];
+  // expected-note@+1 {{in instantiation of function template specialization 'test_template<-4, double, int>' requested here}}
+  test_template<-4>(darr, 4);
+  // expected-note@+1 {{in instantiation of function template specialization 'test_warn<0>' requested here}}
+  test_warn<0>();
+
+  S4 e(4); // expected-note {{'e' defined here}}
+  S5 g(5); // expected-note {{'g' defined here}}
+  int i;
+  int &j = i;
+#pragma omp parallel for linear // expected-error {{expected '(' after 'linear'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+// expected-error@+2 {{linear variable with incomplete type 'S1'}}
+// expected-error@+1 {{const-qualified variable cannot be linear}}
+#pragma omp parallel for linear(a, b)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+// expected-error@+2 {{argument of a linear clause should be of integral or pointer type, not 'S4'}}
+// expected-error@+1 {{argument of a linear clause should be of integral or pointer type, not 'S5'}}
+#pragma omp parallel for linear(e, g)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(h, C::x) // expected-error 2 {{threadprivate or thread local variable cannot be linear}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int i;
+#pragma omp parallel for linear(i)
+    for (int k = 0; k < argc; ++k)
+      ++k;
+#pragma omp parallel for linear(i : 4)
+    for (int k = 0; k < argc; ++k) {
+      ++k;
+      i += 4;
+    }
+  }
+#pragma omp parallel for linear(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel for linear(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+
+  foomain<int, char>(argc, argv);
+  return 0;
+}
+
diff --git a/test/OpenMP/parallel_for_loop_messages.cpp b/test/OpenMP/parallel_for_loop_messages.cpp
index 7d14a3c9b74c..2bb32bdfc176 100644
--- a/test/OpenMP/parallel_for_loop_messages.cpp
+++ b/test/OpenMP/parallel_for_loop_messages.cpp
@@ -10,6 +10,7 @@ public:
 };
 
 static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
 #pragma omp threadprivate(sii)
 static int globalii;
 
@@ -54,32 +55,32 @@ int test_iteration_spaces() {
   for (double fi = 0; fi < 10.0; fi++) {
     c[(int)fi] = a[(int)fi] + b[(int)fi];
   }
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (int &ref = ii; ref < 10; ref++) {
   }
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (int i; i < 10; i++)
     c[i] = a[i];
 
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (int i = 0, j = 0; i < 10; ++i)
     c[i] = a[i];
 
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (; ii < 10; ++ii)
     c[ii] = a[ii];
 
 // expected-warning@+3 {{expression result unused}}
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (ii + 1; ii < 10; ++ii)
     c[ii] = a[ii];
 
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (c[ii] = 0; ii < 10; ++ii)
     c[ii] = a[ii];
@@ -242,7 +243,6 @@ int test_iteration_spaces() {
   for (ii = 0; ii < 10; ii++)
     c[ii] = a[ii];
 
-// expected-error@+3 {{unexpected OpenMP clause 'linear' in directive '#pragma omp parallel for'}}
 // expected-note@+2  {{defined as linear}}
 // expected-error@+2 {{loop iteration variable in the associated loop of 'omp parallel for' directive may not be linear, predetermined as private}}
 #pragma omp parallel for linear(ii)
@@ -258,6 +258,7 @@ int test_iteration_spaces() {
     c[ii] = a[ii];
 
   {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp parallel for' directive may not be threadprivate or thread local, predetermined as private}}
 #pragma omp parallel for
     for (sii = 0; sii < 10; sii += 1)
       c[sii] = a[sii];
@@ -375,7 +376,7 @@ int test_with_random_access_iterator() {
 #pragma omp parallel for
   for (GoodIter I = begin; I < end; ++I)
     ++I;
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (GoodIter &I = begin; I < end; ++I)
     ++I;
@@ -406,7 +407,7 @@ int test_with_random_access_iterator() {
 #pragma omp parallel for
   for (begin = begin0; begin < end; ++begin)
     ++begin;
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for
   for (++begin; begin < end; ++begin)
     ++begin;
diff --git a/test/OpenMP/parallel_for_misc_messages.c b/test/OpenMP/parallel_for_misc_messages.c
index ee6f2e81bb8d..1a773be617c7 100644
--- a/test/OpenMP/parallel_for_misc_messages.c
+++ b/test/OpenMP/parallel_for_misc_messages.c
@@ -59,11 +59,6 @@ void test_non_identifiers() {
 #pragma omp parallel for;
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+2 {{unexpected OpenMP clause 'linear' in directive '#pragma omp parallel for'}}
-// expected-warning@+1 {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
-#pragma omp parallel for linear(x);
-  for (i = 0; i < 16; ++i)
-    ;
 
 // expected-warning@+1 {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
 #pragma omp parallel for private(x);
@@ -153,15 +148,15 @@ void test_collapse() {
 #pragma omp parallel for collapse(foo())
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp parallel for collapse(-5)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp parallel for collapse(0)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp parallel for collapse(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
@@ -170,8 +165,7 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
 // expected-note@+1 {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for' directive into a parallel or another task region?}}
     for (int j = 0; j < 16; ++j)
-// expected-error@+3 {{reduction variable must be shared}}
-// expected-error@+2 {{private variable cannot be reduction}}
+// expected-error@+2 2 {{reduction variable must be shared}}
 // expected-error@+1 {{region cannot be closely nested inside 'parallel for' region; perhaps you forget to enclose 'omp for' directive into a parallel region?}}
 #pragma omp for reduction(+ : i, j)
       for (int k = 0; k < 16; ++k)
diff --git a/test/OpenMP/parallel_for_num_threads_messages.cpp b/test/OpenMP/parallel_for_num_threads_messages.cpp
index 60c7dfb6e3cd..10a4e1bfb1aa 100644
--- a/test/OpenMP/parallel_for_num_threads_messages.cpp
+++ b/test/OpenMP/parallel_for_num_threads_messages.cpp
@@ -24,7 +24,7 @@ T tmain(T argc, S **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for num_threads ((argc > 0) ? argv[1] : argv[2]) // expected-error 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   for (i = 0; i < argc; ++i) foo();
-  #pragma omp parallel for num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel for num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for num_threads (S) // expected-error {{'S' does not refer to a value}}
   for (i = 0; i < argc; ++i) foo();
@@ -32,7 +32,7 @@ T tmain(T argc, S **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for num_threads (argc)
   for (i = 0; i < argc; ++i) foo();
-  #pragma omp parallel for num_threads (N) // expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel for num_threads (N) // expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   for (i = 0; i < argc; ++i) foo();
 
   return argc;
@@ -52,7 +52,7 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for num_threads (argc > 0 ? argv[1] : argv[2]) // expected-error {{integral }}
   for (i = 0; i < argc; ++i) foo();
-  #pragma omp parallel for num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel for num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for num_threads (S1) // expected-error {{'S1' does not refer to a value}}
   for (i = 0; i < argc; ++i) foo();
diff --git a/test/OpenMP/parallel_for_ordered_messages.cpp b/test/OpenMP/parallel_for_ordered_messages.cpp
new file mode 100644
index 000000000000..3729eb9a68a3
--- /dev/null
+++ b/test/OpenMP/parallel_for_ordered_messages.cpp
@@ -0,0 +1,104 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) {                   //expected-note 2 {{declared here}}
+#pragma omp parallel for ordered
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered() // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+// expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+// expected-error@+2 2 {{expression is not an integral constant expression}}
+// expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+#pragma omp parallel for ordered(argc
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+// expected-error@+1 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+#pragma omp parallel for ordered(ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered(1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'ordered' clause}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp parallel for', but found only 1}}
+// expected-error@+3 2 {{directive '#pragma omp parallel for' cannot contain more than one 'ordered' clause}}
+// expected-error@+2 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+// expected-error@+1 2 {{expression is not an integral constant expression}}
+#pragma omp parallel for ordered(foobool(argc)), ordered(true), ordered(-5)
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered(S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+// expected-error@+1 2 {{expression is not an integral constant expression}}
+#pragma omp parallel for ordered(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered(1)
+  for (int i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered(N) // expected-error {{argument to 'ordered' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - ST];
+#pragma omp parallel for ordered(2) // expected-note {{as specified in 'ordered' clause}}
+  foo();                            // expected-error {{expected 2 for loops after '#pragma omp parallel for'}}
+  return argc;
+}
+
+int main(int argc, char **argv) {
+#pragma omp parallel for ordered
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp parallel for ordered( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp parallel for ordered() // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp parallel for ordered(4 // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-note {{as specified in 'ordered' clause}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4]; // expected-error {{expected 4 for loops after '#pragma omp parallel for', but found only 1}}
+#pragma omp parallel for ordered(2 + 2))      // expected-warning {{extra tokens at the end of '#pragma omp parallel for' are ignored}}  expected-note {{as specified in 'ordered' clause}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];            // expected-error {{expected 4 for loops after '#pragma omp parallel for', but found only 1}}
+#pragma omp parallel for ordered(foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+// expected-error@+3 {{expression is not an integral constant expression}}
+// expected-error@+2 2 {{directive '#pragma omp parallel for' cannot contain more than one 'ordered' clause}}
+// expected-error@+1 2 {{argument to 'ordered' clause must be a strictly positive integer value}}
+#pragma omp parallel for ordered(foobool(argc)), ordered(true), ordered(-5)
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+#pragma omp parallel for ordered(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+// expected-error@+1 {{expression is not an integral constant expression}}
+#pragma omp parallel for ordered(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++)
+    argv[0][i] = argv[0][i] - argv[0][i - 4];
+// expected-error@+3 {{statement after '#pragma omp parallel for' must be a for loop}}
+// expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+#pragma omp parallel for ordered(ordered(tmain < int, char, -1, -2 > (argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+#pragma omp parallel for ordered(2) // expected-note {{as specified in 'ordered' clause}}
+  foo();                            // expected-error {{expected 2 for loops after '#pragma omp parallel for'}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
+  return tmain<int, char, 1, 0>(argc, argv);
+}
+
diff --git a/test/OpenMP/parallel_for_private_messages.cpp b/test/OpenMP/parallel_for_private_messages.cpp
index 17344f62d7f1..efc827b0d5e5 100644
--- a/test/OpenMP/parallel_for_private_messages.cpp
+++ b/test/OpenMP/parallel_for_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -99,7 +99,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp parallel for private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for private(i)
@@ -120,7 +120,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -169,12 +169,16 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp parallel for private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for private(i)
   for (int k = 0; k < argc; ++k)
     ++k;
+  static int m;
+#pragma omp parallel for private(m)
+  for (int k = 0; k < argc; ++k)
+    m = k + 2;
 
   return 0;
 }
diff --git a/test/OpenMP/parallel_for_reduction_messages.cpp b/test/OpenMP/parallel_for_reduction_messages.cpp
index 26fa48cf5697..22251b446300 100644
--- a/test/OpenMP/parallel_for_reduction_messages.cpp
+++ b/test/OpenMP/parallel_for_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 100 -o - %s
 
 void foo() {
 }
@@ -16,7 +18,7 @@ class S2 {
 public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
-  static float S2s;
+  static float S2s; // expected-note 2 {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note 2 {{'S2sc' defined here}}
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -102,7 +108,7 @@ T tmain(T argc) {
 #pragma omp parallel for reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp parallel for reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for reduction(foo : argc) //expected-error {{incorrect reduction identifier, expected one of '+', '-', '*', '&', '|', '^', '&&', '||', 'min' or 'max'}}
@@ -114,31 +120,31 @@ T tmain(T argc) {
 #pragma omp parallel for reduction(^ : T) // expected-error {{'T' does not refer to a value}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel for reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp parallel for reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel for reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel for reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp parallel for reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(&& : S2::S2s)
+#pragma omp parallel for reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for reduction(+ : h, k) // expected-error {{threadprivate or thread local variable cannot be reduction}}
@@ -157,7 +163,7 @@ T tmain(T argc) {
 #pragma omp parallel for reduction(+ : p), reduction(+ : p) // expected-error 3 {{variable can appear only once in OpenMP 'reduction' clause}} expected-note 3 {{previously referenced here}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -224,7 +230,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp parallel for reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for reduction(~ : argc) // expected-error {{expected unqualified-id}}
@@ -236,31 +242,31 @@ int main(int argc, char **argv) {
 #pragma omp parallel for reduction(^ : S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel for reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp parallel for reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel for reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel for reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp parallel for reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(&& : S2::S2s)
+#pragma omp parallel for reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for reduction(& : e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{invalid operands to binary expression ('S4' and 'S4')}} expected-error {{calling a private constructor of class 'S5'}} expected-error {{invalid operands to binary expression ('S5' and 'S5')}}
@@ -282,7 +288,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for reduction(+ : p), reduction(+ : p) // expected-error {{variable can appear only once in OpenMP 'reduction' clause}} expected-note {{previously referenced here}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -298,6 +304,10 @@ int main(int argc, char **argv) {
 #pragma omp parallel for reduction(+ : fl)
   for (int i = 0; i < 10; ++i)
     foo();
+  static int m;
+#pragma omp parallel for reduction(+ : m) // OK
+  for (int i = 0; i < 10; ++i)
+    m++;
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/parallel_for_schedule_messages.cpp b/test/OpenMP/parallel_for_schedule_messages.cpp
index c12c4282686b..f446609615da 100644
--- a/test/OpenMP/parallel_for_schedule_messages.cpp
+++ b/test/OpenMP/parallel_for_schedule_messages.cpp
@@ -13,13 +13,13 @@ template <class T, typename S, int N, int ST> // expected-note {{declared here}}
 T tmain(T argc, S **argv) {
   #pragma omp parallel for schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp parallel for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -28,7 +28,7 @@ T tmain(T argc, S **argv) {
   // expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
   #pragma omp parallel for schedule (guided argc
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp parallel for schedule (static, ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for schedule (dynamic, 1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
@@ -36,17 +36,17 @@ T tmain(T argc, S **argv) {
   #pragma omp parallel for schedule (guided, (ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+2 2 {{directive '#pragma omp parallel for' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp parallel for schedule (static, foobool(argc)), schedule (dynamic, true), schedule (guided, -5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for schedule (static, S) // expected-error {{'S' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
+  #pragma omp parallel for schedule (static, S) // expected-error {{'S' does not refer to a value}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+1 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp parallel for schedule (guided, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for schedule (dynamic, 1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for schedule (static, N) // expected-error {{argument to 'schedule' clause must be a positive integer value}}
+  #pragma omp parallel for schedule (static, N) // expected-error {{argument to 'schedule' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -54,13 +54,13 @@ T tmain(T argc, S **argv) {
 int main(int argc, char **argv) {
   #pragma omp parallel for schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp parallel for schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
@@ -73,17 +73,17 @@ int main(int argc, char **argv) {
   #pragma omp parallel for schedule (dynamic, foobool(1) > 0 ? 1 : 2)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+2 2 {{directive '#pragma omp parallel for' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp parallel for schedule (guided, foobool(argc)), schedule (static, true), schedule (dynamic, -5)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for schedule (guided, S1) // expected-error {{'S1' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp parallel for' are ignored}}
+  #pragma omp parallel for schedule (guided, S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+1 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp parallel for schedule (static, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{statement after '#pragma omp parallel for' must be a for loop}}
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
-  #pragma omp parallel for schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
   foo();
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
   return tmain<int, char, 1, 0>(argc, argv);
diff --git a/test/OpenMP/parallel_for_simd_aligned_messages.cpp b/test/OpenMP/parallel_for_simd_aligned_messages.cpp
index e1b9602f0ca7..8bffd21f4655 100644
--- a/test/OpenMP/parallel_for_simd_aligned_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_aligned_messages.cpp
@@ -50,7 +50,7 @@ template<int L, class T, class N> T test_template(T* arr, N num) {
   T sum = (T)0;
   T ind2 = - num * L;
   // Negative number is passed as L.
-  // expected-error@+1 {{argument to 'aligned' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd aligned(arr:L)
   for (i = 0; i < num; ++i) {
     T cur = arr[(int)ind2];
@@ -65,7 +65,7 @@ template<int L, class T, class N> T test_template(T* arr, N num) {
 
 template<int LEN> int test_warn() {
   int *ind2 = 0;
-  // expected-error@+1 {{argument to 'aligned' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd aligned(ind2:LEN)
   for (int i = 0; i < 100; i++) {
     ind2 += LEN;
diff --git a/test/OpenMP/parallel_for_simd_ast_print.cpp b/test/OpenMP/parallel_for_simd_ast_print.cpp
index cd62fc51af7f..e23c2cbd2428 100644
--- a/test/OpenMP/parallel_for_simd_ast_print.cpp
+++ b/test/OpenMP/parallel_for_simd_ast_print.cpp
@@ -14,8 +14,8 @@ template<class T, class N> T reduct(T* arr, N num) {
   N myind;
   T sum = (T)0;
 // CHECK: T sum = (T)0;
-#pragma omp parallel for simd private(myind, g_ind), linear(ind), aligned(arr)
-// CHECK-NEXT: #pragma omp parallel for simd private(myind,g_ind) linear(ind) aligned(arr)
+#pragma omp parallel for simd private(myind, g_ind), linear(ind), aligned(arr) if (parallel :num)
+// CHECK-NEXT: #pragma omp parallel for simd private(myind,g_ind) linear(ind) aligned(arr) if(parallel: num)
   for (i = 0; i < num; ++i) {
     myind = ind;
     T cur = arr[myind];
@@ -35,8 +35,8 @@ template<class T> struct S {
 // CHECK: T res;
 // CHECK: T val;
 // CHECK: T lin = 0;
-    #pragma omp parallel for simd private(val)  safelen(7) linear(lin : -5) lastprivate(res)
-// CHECK-NEXT: #pragma omp parallel for simd private(val) safelen(7) linear(lin: -5) lastprivate(res)
+    #pragma omp parallel for simd private(val)  safelen(7) linear(lin : -5) lastprivate(res) simdlen(5) if(7)
+// CHECK-NEXT: #pragma omp parallel for simd private(val) safelen(7) linear(lin: -5) lastprivate(res) simdlen(5) if(7)
     for (T i = 7; i < m_a; ++i) {
       val = v[i-7] + m_a;
       res = val;
@@ -44,8 +44,8 @@ template<class T> struct S {
     }
     const T clen = 3;
 // CHECK: T clen = 3;
-    #pragma omp parallel for simd safelen(clen-1)
-// CHECK-NEXT: #pragma omp parallel for simd safelen(clen - 1)
+    #pragma omp parallel for simd safelen(clen-1) simdlen(clen-1)
+// CHECK-NEXT: #pragma omp parallel for simd safelen(clen - 1) simdlen(clen - 1)
     for(T i = clen+2; i < 20; ++i) {
 // CHECK-NEXT: for (T i = clen + 2; i < 20; ++i) {
       v[i] = v[v-clen] + 1;
@@ -62,7 +62,7 @@ template<class T> struct S {
 template<int LEN> struct S2 {
   static void func(int n, float *a, float *b, float *c) {
     int k1 = 0, k2 = 0;
-#pragma omp parallel for simd safelen(LEN) linear(k1,k2:LEN) aligned(a:LEN)
+#pragma omp parallel for simd safelen(LEN) linear(k1,k2:LEN) aligned(a:LEN) simdlen(LEN)
     for(int i = 0; i < n; i++) {
       c[i] = a[i] + b[i];
       c[k1] = a[k1] + b[k1];
@@ -77,7 +77,7 @@ template<int LEN> struct S2 {
 // CHECK: template <int LEN = 4> struct S2 {
 // CHECK-NEXT: static void func(int n, float *a, float *b, float *c)     {
 // CHECK-NEXT:   int k1 = 0, k2 = 0;
-// CHECK-NEXT: #pragma omp parallel for simd safelen(4) linear(k1,k2: 4) aligned(a: 4)
+// CHECK-NEXT: #pragma omp parallel for simd safelen(4) linear(k1,k2: 4) aligned(a: 4) simdlen(4)
 // CHECK-NEXT:   for (int i = 0; i < n; i++) {
 // CHECK-NEXT:     c[i] = a[i] + b[i];
 // CHECK-NEXT:     c[k1] = a[k1] + b[k1];
@@ -92,17 +92,17 @@ int main (int argc, char **argv) {
   int k1=0,k2=0;
   static int *a;
 // CHECK: static int *a;
-#pragma omp parallel for simd
-// CHECK-NEXT: #pragma omp parallel for simd
+#pragma omp parallel for simd if(parallel :b)
+// CHECK-NEXT: #pragma omp parallel for simd if(parallel: b)
   for (int i=0; i < 2; ++i)*a=2;
 // CHECK-NEXT: for (int i = 0; i < 2; ++i)
 // CHECK-NEXT: *a = 2;
 #pragma omp  parallel
-#pragma omp parallel for simd private(argc, b),lastprivate(d,f) collapse(2) aligned(a : 4) ,firstprivate( g )
+#pragma omp parallel for simd private(argc, b),lastprivate(d,f) collapse(2) aligned(a : 4) ,firstprivate( g ) if(g)
   for (int i = 0; i < 10; ++i)
   for (int j = 0; j < 10; ++j) {foo(); k1 += 8; k2 += 8;}
 // CHECK-NEXT: #pragma omp parallel
-// CHECK-NEXT: #pragma omp parallel for simd private(argc,b) lastprivate(d,f) collapse(2) aligned(a: 4) firstprivate(g)
+// CHECK-NEXT: #pragma omp parallel for simd private(argc,b) lastprivate(d,f) collapse(2) aligned(a: 4) firstprivate(g) if(g)
 // CHECK-NEXT: for (int i = 0; i < 10; ++i)
 // CHECK-NEXT: for (int j = 0; j < 10; ++j) {
 // CHECK-NEXT: foo();
@@ -114,8 +114,8 @@ int main (int argc, char **argv) {
 // CHECK-NEXT: foo();
   const int CLEN = 4;
 // CHECK-NEXT: const int CLEN = 4;
-  #pragma omp parallel for simd aligned(a:CLEN) linear(a:CLEN) safelen(CLEN) collapse( 1 )
-// CHECK-NEXT: #pragma omp parallel for simd aligned(a: CLEN) linear(a: CLEN) safelen(CLEN) collapse(1)
+  #pragma omp parallel for simd aligned(a:CLEN) linear(a:CLEN) safelen(CLEN) collapse( 1 ) simdlen(CLEN)
+// CHECK-NEXT: #pragma omp parallel for simd aligned(a: CLEN) linear(a: CLEN) safelen(CLEN) collapse(1) simdlen(CLEN)
   for (int i = 0; i < 10; ++i)foo();
 // CHECK-NEXT: for (int i = 0; i < 10; ++i)
 // CHECK-NEXT: foo();
diff --git a/test/OpenMP/parallel_for_simd_codegen.cpp b/test/OpenMP/parallel_for_simd_codegen.cpp
index eb1e493230e3..53f18d09b751 100644
--- a/test/OpenMP/parallel_for_simd_codegen.cpp
+++ b/test/OpenMP/parallel_for_simd_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // REQUIRES: x86-registered-target
 // expected-no-diagnostics
 #ifndef HEADER
@@ -63,7 +63,6 @@ void simple(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[SIMPLE_LOOP1_END]]
 // CHECK: call void @__kmpc_for_static_fini(%ident_t* {{.+}}, i32 %{{.+}})
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 
   long long k = get_val();
 
@@ -112,21 +111,19 @@ void simple(float *a, float *b, float *c, float *d) {
 // CHECK: [[LIN0_2:%.+]] = load i64, i64* [[LIN0]]
 // CHECK-NEXT: [[LIN_ADD2:%.+]] = add nsw i64 [[LIN0_2]], 27
 // CHECK-NEXT: store i64 [[LIN_ADD2]], i64* %{{.+}}
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 
   int lin = 12;
   #pragma omp parallel for simd linear(lin : get_val()), linear(g_ptr)
 
 // Init linear private var.
 // CHECK: [[LIN_VAR:%.+]] = load i32*, i32** %
-// CHECK-NEXT: [[LIN_LOAD:%.+]] = load i32, i32* [[LIN_VAR]]
+// CHECK: [[LIN_LOAD:%.+]] = load i32, i32* [[LIN_VAR]]
 // CHECK-NEXT: store i32 [[LIN_LOAD]], i32* [[LIN_START:%[^,]+]]
 // Remember linear step.
 // CHECK: [[CALL_VAL:%.+]] = invoke
 // CHECK: store i64 [[CALL_VAL]], i64* [[LIN_STEP:%[^,]+]]
 
-// CHECK: [[GLIN_VAR:%.+]] = load double**, double*** %
-// CHECK-NEXT: [[GLIN_LOAD:%.+]] = load double*, double** [[GLIN_VAR]]
+// CHECK: [[GLIN_LOAD:%.+]] = load double*, double** [[GLIN_VAR:%.+]],
 // CHECK-NEXT: store double* [[GLIN_LOAD]], double** [[GLIN_START:%[^,]+]]
 
 // CHECK: call void @__kmpc_for_static_init_8u(%ident_t* {{[^,]+}}, i32 %{{[^,]+}}, i32 34, i32* %{{[^,]+}}, i64* [[LB:%[^,]+]], i64* [[UB:%[^,]+]], i64* [[STRIDE:%[^,]+]], i64 1, i64 1)
@@ -179,14 +176,11 @@ void simple(float *a, float *b, float *c, float *d) {
 // CHECK: call void @__kmpc_for_static_fini(%ident_t* {{.+}}, i32 %{{.+}})
 //
 // Linear start and step are used to calculate final value of the linear variables.
-// CHECK: [[LIN_VAR:%.+]] = load i32*, i32** %
 // CHECK: [[LINSTART:.+]] = load i32, i32* [[LIN_START]]
 // CHECK: [[LINSTEP:.+]] = load i64, i64* [[LIN_STEP]]
 // CHECK: store i32 {{.+}}, i32* [[LIN_VAR]],
-// CHECK: [[GLIN_VAR:%.+]] = load double**, double*** %
 // CHECK: [[GLINSTART:.+]] = load double*, double** [[GLIN_START]]
 // CHECK: store double* {{.*}}[[GLIN_VAR]]
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 
   #pragma omp parallel for simd
 // CHECK: call void @__kmpc_for_static_init_4(%ident_t* {{[^,]+}}, i32 %{{[^,]+}}, i32 34, i32* %{{[^,]+}}, i32* [[LB:%[^,]+]], i32* [[UB:%[^,]+]], i32* [[STRIDE:%[^,]+]], i32 1, i32 1)
@@ -223,7 +217,6 @@ void simple(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[SIMPLE_LOOP4_END]]
 // CHECK: call void @__kmpc_for_static_fini(%ident_t* {{.+}}, i32 %{{.+}})
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 
   #pragma omp parallel for simd
 // CHECK: call void @__kmpc_for_static_init_4(%ident_t* {{[^,]+}}, i32 %{{[^,]+}}, i32 34, i32* %{{[^,]+}}, i32* [[LB:%[^,]+]], i32* [[UB:%[^,]+]], i32* [[STRIDE:%[^,]+]], i32 1, i32 1)
@@ -260,7 +253,6 @@ void simple(float *a, float *b, float *c, float *d) {
   }
 // CHECK: [[SIMPLE_LOOP5_END]]
 // CHECK: call void @__kmpc_for_static_fini(%ident_t* {{.+}}, i32 %{{.+}})
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 
 // CHECK-NOT: mul i32 %{{.+}}, 10
   #pragma omp parallel for simd
@@ -315,7 +307,6 @@ void simple(float *a, float *b, float *c, float *d) {
 // CHECK: [[A_PRIV_VAL:%.+]] = load i32, i32* [[A_PRIV]],
 // CHECK-NEXT: store i32 [[A_PRIV_VAL]], i32* %{{.+}},
 // CHECK-NEXT: br label
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
   }
   int R;
   {
@@ -364,7 +355,6 @@ void simple(float *a, float *b, float *c, float *d) {
 // CHECK: [[RED:%.+]] = mul nsw i32 %{{.+}}, [[R_PRIV_VAL]]
 // CHECK-NEXT: store i32 [[RED]], i32* %{{.+}},
 // CHECK-NEXT: call void @__kmpc_end_reduce_nowait(
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
   }
 }
 
@@ -473,7 +463,6 @@ void iter_simple(IterDouble ia, IterDouble ib, IterDouble ic) {
   }
 // CHECK: [[IT_END]]
 // CHECK: call void @__kmpc_for_static_fini(%ident_t* {{.+}}, i32 %{{.+}})
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 // CHECK: ret void
 }
 
@@ -552,7 +541,6 @@ void collapsed(float *a, float *b, float *c, float *d) {
 // CHECK: store i32 3, i32* [[I:%[^,]+]]
 // CHECK: store i32 5, i32* [[I:%[^,]+]]
 // CHECK: store i16 9, i16* [[I:%[^,]+]]
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 // CHECK: ret void
 }
 
@@ -672,7 +660,6 @@ void widened(float *a, float *b, float *c, float *d) {
 // CHECK-NEXT: br label {{%.+}}
 // CHECK: [[T1_END]]
 // CHECK: call void @__kmpc_for_static_fini(%ident_t* {{.+}}, i32 %{{.+}})
-// CHECK: call void @__kmpc_barrier(%ident_t* {{.+}}, i32 %{{.+}})
 // CHECK: ret void
 //
 // TERM_DEBUG-LABEL: bar
diff --git a/test/OpenMP/parallel_for_simd_collapse_messages.cpp b/test/OpenMP/parallel_for_simd_collapse_messages.cpp
index 22090e6b23ef..4f04cca5635e 100644
--- a/test/OpenMP/parallel_for_simd_collapse_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_collapse_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp parallel for simd collapse (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp parallel for simd collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp parallel for simd', but found only 1}}
   // expected-error@+3 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'collapse' clause}}
-  // expected-error@+2 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp parallel for simd collapse (foobool(argc)), collapse (true), collapse (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd collapse (1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd collapse (N) // expected-error {{argument to 'collapse' clause must be a positive integer value}}
+  #pragma omp parallel for simd collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd collapse (2) // expected-note {{as specified in 'collapse' clause}}
   foo(); // expected-error {{expected 2 for loops after '#pragma omp parallel for simd'}}
@@ -63,7 +63,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'collapse' clause}}
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd collapse (foobool(argc)), collapse (true), collapse (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for simd collapse (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/parallel_for_simd_firstprivate_messages.cpp b/test/OpenMP/parallel_for_simd_firstprivate_messages.cpp
index 2ff32246dc17..7a08572f7d1a 100644
--- a/test/OpenMP/parallel_for_simd_firstprivate_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for simd firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -116,7 +116,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for simd firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp parallel for simd firstprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for simd firstprivate(i)
@@ -152,7 +152,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for simd firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   for (i = 0; i < argc; ++i)
     foo();
@@ -226,7 +226,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd firstprivate(xa) // OK: may be firstprivate
   for (i = 0; i < argc; ++i)
     foo();
-#pragma omp parallel for simd firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp parallel for simd firstprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel for simd lastprivate(g) firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
@@ -253,6 +253,10 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd firstprivate(i)  // expected-note {{defined as firstprivate}}
   for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp parallel for simd' directive may not be firstprivate, predetermined as linear}}
     foo();
+  static int si;
+#pragma omp parallel for simd firstprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 2;
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/parallel_for_simd_if_messages.cpp b/test/OpenMP/parallel_for_simd_if_messages.cpp
index ca327cff7206..829b825aaf50 100644
--- a/test/OpenMP/parallel_for_simd_if_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_if_messages.cpp
@@ -34,6 +34,20 @@ int tmain(T argc, S **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd if(argc)
   for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel for simd'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel for simd' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  for (i = 0; i < argc; ++i) foo();
 
   return 0;
 }
@@ -64,6 +78,20 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd if(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc)
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel for simd'}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel for simd' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  for (i = 0; i < argc; ++i) foo();
+  #pragma omp parallel for simd if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  for (i = 0; i < argc; ++i) foo();
 
   return tmain(argc, argv);
 }
diff --git a/test/OpenMP/parallel_for_simd_lastprivate_messages.cpp b/test/OpenMP/parallel_for_simd_lastprivate_messages.cpp
index e85e28b3f18a..bd1a6d505572 100644
--- a/test/OpenMP/parallel_for_simd_lastprivate_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_lastprivate_messages.cpp
@@ -16,7 +16,7 @@ public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
   const S2 &operator=(const S2 &) const;
-  static float S2s;
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note {{static data member is predetermined as shared}}
@@ -66,7 +66,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                        // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -118,7 +118,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for simd lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp parallel for simd lastprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for simd lastprivate(i)
@@ -143,7 +143,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i;                        // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (i = 0; i < argc; ++i)
     foo();
@@ -190,7 +190,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd lastprivate(xa) // OK
   for (i = 0; i < argc; ++i)
     foo();
-#pragma omp parallel for simd lastprivate(S2::S2s)
+#pragma omp parallel for simd lastprivate(S2::S2s) // expected-error {{shared variable cannot be lastprivate}}
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel for simd lastprivate(S2::S2sc) // expected-error {{shared variable cannot be lastprivate}}
@@ -222,7 +222,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd lastprivate(xa)
   for (i = 0; i < argc; ++i)
     foo();
-#pragma omp parallel for simd lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp parallel for simd lastprivate(j)
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel for simd firstprivate(m) lastprivate(m) // expected-error {{'operator=' is a private member of 'S3'}}
@@ -231,5 +231,9 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd lastprivate(n) firstprivate(n) // OK
   for (i = 0; i < argc; ++i)
     foo();
+  static int si;
+#pragma omp parallel for simd lastprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 3;
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/parallel_for_simd_linear_messages.cpp b/test/OpenMP/parallel_for_simd_linear_messages.cpp
index 55b0c3d7f9c8..858f53f4674d 100644
--- a/test/OpenMP/parallel_for_simd_linear_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_linear_messages.cpp
@@ -102,7 +102,7 @@ template<class I, class C> int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp parallel for simd linear // expected-error {{expected '(' after 'linear'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp parallel for simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -138,7 +138,7 @@ template<class I, class C> int foomain(I argc, C **argv) {
     #pragma omp parallel for simd linear(v:i)
     for (int k = 0; k < argc; ++k) { i = k; v += i; }
   }
-  #pragma omp parallel for simd linear(j) // expected-error {{arguments of OpenMP clause 'linear' cannot be of reference type}}
+  #pragma omp parallel for simd linear(j)
   for (int k = 0; k < argc; ++k) ++k;
   int v = 0;
   #pragma omp parallel for simd linear(v:j)
@@ -166,7 +166,7 @@ int main(int argc, char **argv) {
   S4 e(4); // expected-note {{'e' defined here}}
   S5 g(5); // expected-note {{'g' defined here}}
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp parallel for simd linear // expected-error {{expected '(' after 'linear'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp parallel for simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -203,7 +203,7 @@ int main(int argc, char **argv) {
     #pragma omp parallel for simd linear(i : 4)
     for (int k = 0; k < argc; ++k) { ++k; i += 4; }
   }
-  #pragma omp parallel for simd linear(j) // expected-error {{arguments of OpenMP clause 'linear' cannot be of reference type 'int &'}}
+  #pragma omp parallel for simd linear(j)
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp parallel for simd linear(i)
   for (int k = 0; k < argc; ++k) ++k;
diff --git a/test/OpenMP/parallel_for_simd_loop_messages.cpp b/test/OpenMP/parallel_for_simd_loop_messages.cpp
index 85d8574ba9f4..f4929d97d484 100644
--- a/test/OpenMP/parallel_for_simd_loop_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_loop_messages.cpp
@@ -10,6 +10,7 @@ public:
 };
 
 static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
 #pragma omp threadprivate(sii)
 static int globalii;
 
@@ -54,32 +55,32 @@ int test_iteration_spaces() {
   for (double fi = 0; fi < 10.0; fi++) {
     c[(int)fi] = a[(int)fi] + b[(int)fi];
   }
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (int &ref = ii; ref < 10; ref++) {
   }
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (int i; i < 10; i++)
     c[i] = a[i];
 
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (int i = 0, j = 0; i < 10; ++i)
     c[i] = a[i];
 
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (; ii < 10; ++ii)
     c[ii] = a[ii];
 
 // expected-warning@+3 {{expression result unused}}
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (ii + 1; ii < 10; ++ii)
     c[ii] = a[ii];
 
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (c[ii] = 0; ii < 10; ++ii)
     c[ii] = a[ii];
@@ -259,6 +260,7 @@ int test_iteration_spaces() {
     c[ii] = a[ii];
 
   {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp parallel for simd' directive may not be threadprivate or thread local, predetermined as linear}}
 #pragma omp parallel for simd
     for (sii = 0; sii < 10; sii += 1)
       c[sii] = a[sii];
@@ -375,7 +377,7 @@ int test_with_random_access_iterator() {
 #pragma omp parallel for simd
   for (GoodIter I = begin; I < end; ++I)
     ++I;
-// expected-error@+2 {{variable must be of integer or random access iterator type}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (GoodIter &I = begin; I < end; ++I)
     ++I;
@@ -406,7 +408,7 @@ int test_with_random_access_iterator() {
 #pragma omp parallel for simd
   for (begin = begin0; begin < end; ++begin)
     ++begin;
-// expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
 #pragma omp parallel for simd
   for (++begin; begin < end; ++begin)
     ++begin;
diff --git a/test/OpenMP/parallel_for_simd_misc_messages.c b/test/OpenMP/parallel_for_simd_misc_messages.c
index ed9ac4b5242a..378c48ff8a4c 100644
--- a/test/OpenMP/parallel_for_simd_misc_messages.c
+++ b/test/OpenMP/parallel_for_simd_misc_messages.c
@@ -153,20 +153,117 @@ void test_safelen() {
 #pragma omp parallel for simd safelen(foo())
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp parallel for simd safelen(-5)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp parallel for simd safelen(0)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp parallel for simd safelen(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
 }
 
+void test_simdlen() {
+  int i;
+// expected-error@+1 {{expected '('}}
+#pragma omp parallel for simd simdlen
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}}
+#pragma omp parallel for simd simdlen()
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(,
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}}  expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(, )
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
+// expected-error@+1 {{expected '('}}
+#pragma omp parallel for simd simdlen 4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(4
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(4,
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(4, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel for simd simdlen(4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(4 4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(4, , 4)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel for simd simdlen(4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp parallel for simd simdlen(4, 8)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp parallel for simd simdlen(2.5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp parallel for simd simdlen(foo())
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp parallel for simd simdlen(-5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp parallel for simd simdlen(0)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp parallel for simd simdlen(5 - 5)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_safelen_simdlen() {
+  int i;
+// expected-error@+1 {{the value of 'simdlen' parameter must be less than or equal to the value of the 'safelen' parameter}}
+#pragma omp parallel for simd simdlen(6) safelen(5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{the value of 'simdlen' parameter must be less than or equal to the value of the 'safelen' parameter}}
+#pragma omp parallel for simd safelen(5) simdlen(6)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
 void test_collapse() {
   int i;
 #pragma omp parallel
@@ -259,17 +356,17 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp parallel for simd collapse(-5)
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp parallel for simd collapse(0)
   for (i = 0; i < 16; ++i)
     ;
 #pragma omp parallel
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp parallel for simd collapse(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
diff --git a/test/OpenMP/parallel_for_simd_num_threads_messages.cpp b/test/OpenMP/parallel_for_simd_num_threads_messages.cpp
index 5b5d33464ef8..940565cc10c3 100644
--- a/test/OpenMP/parallel_for_simd_num_threads_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_num_threads_messages.cpp
@@ -24,7 +24,7 @@ T tmain(T argc, S **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd num_threads ((argc > 0) ? argv[1] : argv[2]) // expected-error 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   for (i = 0; i < argc; ++i) foo();
-  #pragma omp parallel for simd num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel for simd num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd num_threads (S) // expected-error {{'S' does not refer to a value}}
   for (i = 0; i < argc; ++i) foo();
@@ -32,7 +32,7 @@ T tmain(T argc, S **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd num_threads (argc)
   for (i = 0; i < argc; ++i) foo();
-  #pragma omp parallel for simd num_threads (N) // expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel for simd num_threads (N) // expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   for (i = 0; i < argc; ++i) foo();
 
   return argc;
@@ -52,7 +52,7 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd num_threads (argc > 0 ? argv[1] : argv[2]) // expected-error {{integral }}
   for (i = 0; i < argc; ++i) foo();
-  #pragma omp parallel for simd num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel for simd num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   for (i = 0; i < argc; ++i) foo();
   #pragma omp parallel for simd num_threads (S1) // expected-error {{'S1' does not refer to a value}}
   for (i = 0; i < argc; ++i) foo();
diff --git a/test/OpenMP/parallel_for_simd_private_messages.cpp b/test/OpenMP/parallel_for_simd_private_messages.cpp
index 130736a33a07..a031d407ec1b 100644
--- a/test/OpenMP/parallel_for_simd_private_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for simd private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -99,7 +99,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for simd private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp parallel for simd private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for simd private(i)
@@ -120,7 +120,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;           // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel for simd private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -169,12 +169,16 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel for simd private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp parallel for simd private(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp parallel for simd private(i)
   for (int k = 0; k < argc; ++k)
     ++k;
+  static int m;
+#pragma omp parallel for simd private(m) // OK
+  for (int k = 0; k < argc; ++k)
+    m = k + 3;
 
   return 0;
 }
diff --git a/test/OpenMP/parallel_for_simd_reduction_messages.cpp b/test/OpenMP/parallel_for_simd_reduction_messages.cpp
index 75733a3647ac..e2e9e1bca38b 100644
--- a/test/OpenMP/parallel_for_simd_reduction_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -o - %s
 
 void foo() {
 }
@@ -16,7 +18,7 @@ class S2 {
 public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
-  static float S2s;
+  static float S2s; // expected-note 2 {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note 2 {{'S2sc' defined here}}
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -102,7 +108,7 @@ T tmain(T argc) {
 #pragma omp parallel for simd reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp parallel for simd reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for simd reduction(foo : argc) //expected-error {{incorrect reduction identifier, expected one of '+', '-', '*', '&', '|', '^', '&&', '||', 'min' or 'max'}}
@@ -114,31 +120,31 @@ T tmain(T argc) {
 #pragma omp parallel for simd reduction(^ : T) // expected-error {{'T' does not refer to a value}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel for simd reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for simd reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp parallel for simd reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel for simd reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel for simd reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp parallel for simd reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for simd reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(&& : S2::S2s)
+#pragma omp parallel for simd reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for simd reduction(+ : h, k) // expected-error {{threadprivate or thread local variable cannot be reduction}}
@@ -157,7 +163,7 @@ T tmain(T argc) {
 #pragma omp parallel for simd reduction(+ : p), reduction(+ : p) // expected-error 3 {{variable can appear only once in OpenMP 'reduction' clause}} expected-note 3 {{previously referenced here}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for simd reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -224,7 +230,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp parallel for simd reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for simd reduction(~ : argc) // expected-error {{expected unqualified-id}}
@@ -236,31 +242,31 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd reduction(^ : S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel for simd reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for simd reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp parallel for simd reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel for simd reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel for simd reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp parallel for simd reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for simd reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(&& : S2::S2s)
+#pragma omp parallel for simd reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for simd reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel for simd reduction(& : e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{invalid operands to binary expression ('S4' and 'S4')}} expected-error {{calling a private constructor of class 'S5'}} expected-error {{invalid operands to binary expression ('S5' and 'S5')}}
@@ -282,7 +288,7 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd reduction(+ : p), reduction(+ : p) // expected-error {{variable can appear only once in OpenMP 'reduction' clause}} expected-note {{previously referenced here}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp parallel for simd reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel for simd reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -298,6 +304,10 @@ int main(int argc, char **argv) {
 #pragma omp parallel for simd reduction(+ : fl)
   for (int i = 0; i < 10; ++i)
     foo();
+  static int m;
+#pragma omp parallel for simd reduction(+ : m) // OK
+  for (int i = 0; i < 10; ++i)
+    m++;
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/parallel_for_simd_safelen_messages.cpp b/test/OpenMP/parallel_for_simd_safelen_messages.cpp
index eb0aa5ae8b5e..45f2fa2b6277 100644
--- a/test/OpenMP/parallel_for_simd_safelen_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_safelen_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp parallel for simd safelen (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd safelen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd safelen (1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp parallel for simd safelen ((ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+3 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'safelen' clause}}
-  // expected-error@+2 2 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp parallel for simd safelen (foobool(argc)), safelen (true), safelen (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd safelen (4)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd safelen (N) // expected-error {{argument to 'safelen' clause must be a positive integer value}}
+  #pragma omp parallel for simd safelen (N) // expected-error {{argument to 'safelen' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -61,7 +61,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'safelen' clause}}
-  // expected-error@+1 2 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd safelen (foobool(argc)), safelen (true), safelen (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for simd safelen (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/parallel_for_simd_schedule_messages.cpp b/test/OpenMP/parallel_for_simd_schedule_messages.cpp
index 36befb213559..4135427ce10b 100644
--- a/test/OpenMP/parallel_for_simd_schedule_messages.cpp
+++ b/test/OpenMP/parallel_for_simd_schedule_messages.cpp
@@ -13,13 +13,13 @@ template <class T, typename S, int N, int ST> // expected-note {{declared here}}
 T tmain(T argc, S **argv) {
   #pragma omp parallel for simd schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp parallel for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -28,7 +28,7 @@ T tmain(T argc, S **argv) {
   // expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
   #pragma omp parallel for simd schedule (guided argc
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd schedule (static, ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd schedule (dynamic, 1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
@@ -36,17 +36,17 @@ T tmain(T argc, S **argv) {
   #pragma omp parallel for simd schedule (guided, (ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+2 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd schedule (static, foobool(argc)), schedule (dynamic, true), schedule (guided, -5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd schedule (static, S) // expected-error {{'S' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
+  #pragma omp parallel for simd schedule (static, S) // expected-error {{'S' does not refer to a value}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+1 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp parallel for simd schedule (guided, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp parallel for simd schedule (dynamic, 1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp parallel for simd schedule (static, N) // expected-error {{argument to 'schedule' clause must be a positive integer value}}
+  #pragma omp parallel for simd schedule (static, N) // expected-error {{argument to 'schedule' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -54,13 +54,13 @@ T tmain(T argc, S **argv) {
 int main(int argc, char **argv) {
   #pragma omp parallel for simd schedule // expected-error {{expected '(' after 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for simd schedule ( // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}}
+  #pragma omp parallel for simd schedule () // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for simd schedule (auto // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto' or 'runtime' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for simd schedule (auto_dynamic // expected-error {{expected 'static', 'dynamic', 'guided', 'auto', 'runtime', 'monotonic', 'nonmonotonic' or 'simd' in OpenMP clause 'schedule'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp parallel for simd schedule (auto,  // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
@@ -73,17 +73,17 @@ int main(int argc, char **argv) {
   #pragma omp parallel for simd schedule (dynamic, foobool(1) > 0 ? 1 : 2)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+2 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'schedule' clause}}
-  // expected-error@+1 {{argument to 'schedule' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'schedule' clause must be a strictly positive integer value}}
   #pragma omp parallel for simd schedule (guided, foobool(argc)), schedule (static, true), schedule (dynamic, -5)
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
-  #pragma omp parallel for simd schedule (guided, S1) // expected-error {{'S1' does not refer to a value}} expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
+  #pragma omp parallel for simd schedule (guided, S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+1 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp parallel for simd schedule (static, argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{statement after '#pragma omp parallel for simd' must be a for loop}}
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
-  #pragma omp parallel for simd schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel for simd schedule(dynamic, schedule(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
   foo();
   // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
   return tmain<int, char, 1, 0>(argc, argv);
diff --git a/test/OpenMP/parallel_for_simd_simdlen_messages.cpp b/test/OpenMP/parallel_for_simd_simdlen_messages.cpp
new file mode 100644
index 000000000000..dd1cf0feaa49
--- /dev/null
+++ b/test/OpenMP/parallel_for_simd_simdlen_messages.cpp
@@ -0,0 +1,79 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp parallel for simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp parallel for simd simdlen (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp parallel for simd simdlen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen (1)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen ((ST > 0) ? 1 + ST : 2)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+2 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp parallel for simd simdlen (foobool(argc)), simdlen (true), simdlen (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp parallel for simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen (4)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp parallel for simd simdlen (N) // expected-error {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp parallel for simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp parallel for simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp parallel for simd simdlen () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp parallel for simd simdlen (4 // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp parallel for simd simdlen (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp parallel for simd' are ignored}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp parallel for simd simdlen (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp parallel for simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+1 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp parallel for simd simdlen (foobool(argc)), simdlen (true), simdlen (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp parallel for simd simdlen (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp parallel for simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp parallel for simd' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp parallel for simd simdlen(simdlen(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 12, 4>' requested here}}
+  return tmain<int, char, 12, 4>(argc, argv);
+}
+
diff --git a/test/OpenMP/parallel_if_codegen.cpp b/test/OpenMP/parallel_if_codegen.cpp
index ace20ecd8317..9ecf57211f85 100644
--- a/test/OpenMP/parallel_if_codegen.cpp
+++ b/test/OpenMP/parallel_if_codegen.cpp
@@ -16,14 +16,14 @@ int Arg;
 
 // CHECK-LABEL: define {{.*}}void @{{.+}}gtid_test
 void gtid_test() {
-// CHECK:  call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 1, {{.+}}* [[GTID_TEST_REGION1:@.+]] to void
+// CHECK:  call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 0, {{.+}}* [[GTID_TEST_REGION1:@.+]] to void
 #pragma omp parallel
-#pragma omp parallel if (false)
+#pragma omp parallel if (parallel: false)
   gtid_test();
 // CHECK: ret void
 }
 
-// CHECK: define internal {{.*}}void [[GTID_TEST_REGION1]](i{{.+}}* [[GTID_PARAM:%.+]], i
+// CHECK: define internal {{.*}}void [[GTID_TEST_REGION1]](i{{.+}}* noalias [[GTID_PARAM:%.+]], i32* noalias
 // CHECK: store i{{[0-9]+}}* [[GTID_PARAM]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]],
 // CHECK: [[GTID_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_ADDR]]
@@ -43,7 +43,7 @@ int tmain(T Arg) {
   fn1();
 #pragma omp parallel if (false)
   fn2();
-#pragma omp parallel if (Arg)
+#pragma omp parallel if (parallel: Arg)
   fn3();
   return 0;
 }
@@ -51,7 +51,7 @@ int tmain(T Arg) {
 // CHECK-LABEL: define {{.*}}i{{[0-9]+}} @main()
 int main() {
 // CHECK: [[GTID:%.+]] = call {{.*}}i32 @__kmpc_global_thread_num(
-// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 1, void {{.+}}* [[CAP_FN4:@.+]] to void
+// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 0, void {{.+}}* [[CAP_FN4:@.+]] to void
 #pragma omp parallel if (true)
   fn4();
 // CHECK: call {{.*}}void @__kmpc_serialized_parallel(%{{.+}}* @{{.+}}, i32 [[GTID]])
@@ -63,7 +63,7 @@ int main() {
 
 // CHECK: br i1 %{{.+}}, label %[[OMP_THEN:.+]], label %[[OMP_ELSE:.+]]
 // CHECK: [[OMP_THEN]]
-// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 1, void {{.+}}* [[CAP_FN6:@.+]] to void
+// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 0, void {{.+}}* [[CAP_FN6:@.+]] to void
 // CHECK: br label %[[OMP_END:.+]]
 // CHECK: [[OMP_ELSE]]
 // CHECK: call {{.*}}void @__kmpc_serialized_parallel(%{{.+}}* @{{.+}}, i32 [[GTID]])
@@ -92,14 +92,14 @@ int main() {
 
 // CHECK-LABEL: define {{.+}} @{{.+}}tmain
 // CHECK: [[GTID:%.+]] = call {{.*}}i32 @__kmpc_global_thread_num(
-// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 1, void {{.+}}* [[CAP_FN1:@.+]] to void
+// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 0, void {{.+}}* [[CAP_FN1:@.+]] to void
 // CHECK: call {{.*}}void @__kmpc_serialized_parallel(%{{.+}}* @{{.+}}, i32 [[GTID]])
 // CHECK: store i32 [[GTID]], i32* [[GTID_ADDR:%.+]],
 // CHECK: call void [[CAP_FN2:@.+]](i32* [[GTID_ADDR]],
 // CHECK: call {{.*}}void @__kmpc_end_serialized_parallel(%{{.+}}* @{{.+}}, i32 [[GTID]])
 // CHECK: br i1 %{{.+}}, label %[[OMP_THEN:.+]], label %[[OMP_ELSE:.+]]
 // CHECK: [[OMP_THEN]]
-// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 1, void {{.+}}* [[CAP_FN3:@.+]] to void
+// CHECK: call {{.*}}void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 0, void {{.+}}* [[CAP_FN3:@.+]] to void
 // CHECK: br label %[[OMP_END:.+]]
 // CHECK: [[OMP_ELSE]]
 // CHECK: call {{.*}}void @__kmpc_serialized_parallel(%{{.+}}* @{{.+}}, i32 [[GTID]])
diff --git a/test/OpenMP/parallel_if_messages.cpp b/test/OpenMP/parallel_if_messages.cpp
index 97096dfae3e8..1ffc567b5916 100644
--- a/test/OpenMP/parallel_if_messages.cpp
+++ b/test/OpenMP/parallel_if_messages.cpp
@@ -22,6 +22,13 @@ int tmain(T argc, S **argv) {
   #pragma omp parallel if (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel if(argc)
+  #pragma omp parallel if(parallel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel : argc)
+  #pragma omp parallel if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel'}}
+  #pragma omp parallel if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  #pragma omp parallel if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
   foo();
 
   return 0;
@@ -40,6 +47,13 @@ int main(int argc, char **argv) {
   #pragma omp parallel if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel if (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel if(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel // expected-warning {{missing ':' after directive name modifier - ignoring}} expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp parallel if(parallel : argc)
+  #pragma omp parallel if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel'}}
+  #pragma omp parallel if(parallel : argc) if (parallel:argc)  // expected-error {{directive '#pragma omp parallel' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  #pragma omp parallel if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
   foo();
 
   return tmain(argc, argv);
diff --git a/test/OpenMP/parallel_num_threads_messages.cpp b/test/OpenMP/parallel_num_threads_messages.cpp
index 180d9cd8035a..adcf6ca4253f 100644
--- a/test/OpenMP/parallel_num_threads_messages.cpp
+++ b/test/OpenMP/parallel_num_threads_messages.cpp
@@ -19,11 +19,11 @@ T tmain(T argc, S **argv) {
   #pragma omp parallel num_threads (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel num_threads (argc)) // expected-warning {{extra tokens at the end of '#pragma omp parallel' are ignored}}
   #pragma omp parallel num_threads ((argc > 0) ? argv[1] : argv[2]) // expected-error 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
-  #pragma omp parallel num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   #pragma omp parallel num_threads (S) // expected-error {{'S' does not refer to a value}}
   #pragma omp parallel num_threads (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp parallel num_threads (argc)
-  #pragma omp parallel num_threads (N) // expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel num_threads (N) // expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   #pragma omp parallel redef_num_threads (argc, argc)
   foo();
 
@@ -37,7 +37,7 @@ int main(int argc, char **argv) {
   #pragma omp parallel num_threads (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel num_threads (argc)) // expected-warning {{extra tokens at the end of '#pragma omp parallel' are ignored}}
   #pragma omp parallel num_threads (argc > 0 ? argv[1] : argv[2]) // expected-error {{integral }}
-  #pragma omp parallel num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   #pragma omp parallel num_threads (S1) // expected-error {{'S1' does not refer to a value}}
   #pragma omp parallel num_threads (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
   #pragma omp parallel num_threads (num_threads(tmain<int, char, -1>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}} expected-note {{in instantiation of function template specialization 'tmain<int, char, -1>' requested here}}
diff --git a/test/OpenMP/parallel_private_codegen.cpp b/test/OpenMP/parallel_private_codegen.cpp
index d2d02a7c3d37..1d195be97d6e 100644
--- a/test/OpenMP/parallel_private_codegen.cpp
+++ b/test/OpenMP/parallel_private_codegen.cpp
@@ -16,19 +16,17 @@ struct S {
   ~S() {}
 };
 
-volatile int g = 1212;
+volatile int g __attribute__((aligned(128))) = 1212;
 
 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[CAP_MAIN_TY:%.+]] = type { i8 }
 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK: [[CAP_TMAIN_TY:%.+]] = type { i8 }
 template <typename T>
 T tmain() {
   S<T> test;
-  T t_var = T();
-  T vec[] = {1, 2};
-  S<T> s_arr[] = {1, 2};
-  S<T> var(3);
+  T t_var __attribute__((aligned(128))) = T();
+  T vec[] __attribute__((aligned(128))) = {1, 2};
+  S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
+  S<T> var __attribute__((aligned(128))) (3);
 #pragma omp parallel private(t_var, vec, s_arr, var)
   {
     vec[0] = t_var;
@@ -38,6 +36,7 @@ T tmain() {
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
@@ -45,26 +44,35 @@ int main() {
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
   // LAMBDA-NOT: = getelementptr inbounds %{{.+}},
-  // LAMBDA: [[ARG:%.+]] = bitcast %{{.+}}* %{{.+}} to i8*
-  // LAMBDA: call{{.*}} void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
-#pragma omp parallel private(g)
+  // LAMBDA: call{{.*}} void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
+#pragma omp parallel private(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     g = 1;
+    sivar = 2;
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
+
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 4;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -76,25 +84,33 @@ int main() {
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
   // BLOCKS-NOT: = getelementptr inbounds %{{.+}},
-  // BLOCKS: [[ARG:%.+]] = bitcast %{{.+}}* %{{.+}} to i8*
-  // BLOCKS: call{{.*}} void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
-#pragma omp parallel private(g)
+  // BLOCKS: call{{.*}} void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
+#pragma omp parallel private(g, sivar)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     g = 1;
+    sivar = 20;
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 20, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call{{.*}} void {{%.+}}(i8
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 40;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 40, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -106,10 +122,11 @@ int main() {
   int vec[] = {1, 2};
   S<float> s_arr[] = {1, 2};
   S<float> var(3);
-#pragma omp parallel private(t_var, vec, s_arr, var)
+#pragma omp parallel private(t_var, vec, s_arr, var, sivar)
   {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 3;
   }
   return tmain<int>();
 #endif
@@ -118,17 +135,17 @@ int main() {
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
@@ -145,18 +162,19 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
-// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
-// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
-// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
+// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
+// CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}], align 128
+// CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]], align 128
+// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
+// CHECK-NOT: [[SIVAR_PRIV]]
 // CHECK: {{.+}}:
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
diff --git a/test/OpenMP/parallel_private_messages.cpp b/test/OpenMP/parallel_private_messages.cpp
index a637b5018f53..ab535b475265 100644
--- a/test/OpenMP/parallel_private_messages.cpp
+++ b/test/OpenMP/parallel_private_messages.cpp
@@ -13,7 +13,7 @@ class S2 {
   mutable int a;
 public:
   S2():a(0) { }
-  static float S2s;
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
 };
 const S2 b;
 const S2 ba[5];
@@ -55,7 +55,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g[] = {5, 6};
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp parallel private // expected-error {{expected '(' after 'private'}}
   #pragma omp parallel private ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp parallel private () // expected-error {{expected expression}}
@@ -69,7 +69,7 @@ int main(int argc, char **argv) {
   #pragma omp parallel private(ba)
   #pragma omp parallel private(ca) // expected-error {{shared variable cannot be private}}
   #pragma omp parallel private(da) // expected-error {{shared variable cannot be private}}
-  #pragma omp parallel private(S2::S2s)
+  #pragma omp parallel private(S2::S2s) // expected-error {{shared variable cannot be private}}
   #pragma omp parallel private(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
   #pragma omp parallel private(threadvar, B::x) // expected-error 2 {{threadprivate or thread local variable cannot be private}}
   #pragma omp parallel shared(i), private(i) // expected-error {{shared variable cannot be private}} expected-note {{defined as shared}}
@@ -77,13 +77,16 @@ int main(int argc, char **argv) {
   #pragma omp parallel firstprivate(i) private(i) // expected-error {{firstprivate variable cannot be private}} expected-note {{defined as firstprivate}}
   foo();
   #pragma omp parallel private(i)
-  #pragma omp parallel private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type 'int &'}}
+  #pragma omp parallel private(j)
   foo();
   #pragma omp parallel firstprivate(i)
   for (int k = 0; k < 10; ++k) {
     #pragma omp parallel private(i)
     foo();
   }
+  static int m;
+  #pragma omp parallel private(m) // OK
+  foo();
 
   return 0;
 }
diff --git a/test/OpenMP/parallel_reduction_codegen.cpp b/test/OpenMP/parallel_reduction_codegen.cpp
index 04d19ebea8c5..b9744b634118 100644
--- a/test/OpenMP/parallel_reduction_codegen.cpp
+++ b/test/OpenMP/parallel_reduction_codegen.cpp
@@ -8,7 +8,7 @@
 #ifndef HEADER
 #define HEADER
 
-volatile int g = 1212;
+volatile int g __attribute__((aligned(128))) = 1212;
 
 template <class T>
 struct S {
@@ -22,9 +22,6 @@ struct S {
 
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { [2 x i{{[0-9]+}}]*, float*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float* }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { [2 x i{{[0-9]+}}]*, i{{[0-9]+}}*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*, [[S_INT_TY]]*, i{{[0-9]+}}* }
-// CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-DAG: [[REDUCTION_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 18, i32 0, i32 0, i8*
 // CHECK-DAG: [[REDUCTION_LOCK:@.+]] = common global [8 x i32] zeroinitializer
 
@@ -32,10 +29,10 @@ template <typename T>
 T tmain() {
   T t;
   S<T> test;
-  T t_var = T(), t_var1;
+  T t_var __attribute__((aligned(128))) = T(), t_var1 __attribute__((aligned(128)));
   T vec[] = {1, 2};
-  S<T> s_arr[] = {1, 2};
-  S<T> var(3), var1;
+  S<T> s_arr[]  = {1, 2};
+  S<T> var __attribute__((aligned(128))) (3), var1 __attribute__((aligned(128)));
 #pragma omp parallel reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1)
   {
     vec[0] = t_var;
@@ -51,30 +48,24 @@ int main() {
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: [[G_LOCAL_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[AGG_CAPTURED:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-  // LAMBDA: store i{{[0-9]+}}* [[G]], i{{[0-9]+}}** [[G_LOCAL_REF]]
-  // LAMBDA: [[ARG:%.+]] = bitcast %{{.+}}* [[AGG_CAPTURED]] to i8*
-  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]])
 #pragma omp parallel reduction(+:g)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
 
     // Reduction list for runtime.
     // LAMBDA: [[RED_LIST:%.+]] = alloca [1 x i8*],
 
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    // LAMBDA: [[ARG:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_REF]]
-    // LAMBDA: [[G_REF_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-    // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR]]
-    // LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]]
+    // LAMBDA: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
     g = 1;
-    // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
 
-    // LAMBDA: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i32 0, i32 0
+    // LAMBDA: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i64 0, i64 0
     // LAMBDA: [[BITCAST:%.+]] = bitcast i32* [[G_PRIVATE_ADDR]] to i8*
     // LAMBDA: store i8* [[BITCAST]], i8** [[G_PRIV_REF]],
     // LAMBDA: call i32 @__kmpc_reduce_nowait(
@@ -112,31 +103,25 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: [[G_LOCAL_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[AGG_CAPTURED:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-  // BLOCKS: store i{{[0-9]+}}* [[G]], i{{[0-9]+}}** [[G_LOCAL_REF]]
-  // BLOCKS: [[ARG:%.+]] = bitcast %{{.+}}* [[AGG_CAPTURED]] to i8*
-  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* [[ARG]])
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i32* [[G]])
 #pragma omp parallel reduction(-:g)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
 
     // Reduction list for runtime.
     // BLOCKS: [[RED_LIST:%.+]] = alloca [1 x i8*],
 
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    // BLOCKS: [[ARG:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_REF]]
-    // BLOCKS: [[G_REF_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
-    // BLOCKS: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR]]
-    // BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // BLOCKS: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_REF_ADDR:%.+]]
+    // BLOCKS: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
     g = 1;
-    // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]], align 128
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
 
-    // BLOCKS: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i32 0, i32 0
+    // BLOCKS: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i64 0, i64 0
     // BLOCKS: [[BITCAST:%.+]] = bitcast i32* [[G_PRIVATE_ADDR]] to i8*
     // BLOCKS: store i8* [[BITCAST]], i8** [[G_PRIV_REF]],
     // BLOCKS: call i32 @__kmpc_reduce_nowait(
@@ -178,6 +163,12 @@ int main() {
     vec[0] = t_var;
     s_arr[0] = var;
   }
+  if (var1)
+#pragma omp parallel reduction(+ : t_var) reduction(& : var) reduction(&& : var1) reduction(min : t_var1)
+    while (1) {
+      vec[0] = t_var;
+      s_arr[0] = var;
+    }
   return tmain<int>();
 #endif
 }
@@ -185,13 +176,13 @@ int main() {
 // CHECK: define {{.*}}i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, float*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, float*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float*)* [[MAIN_MICROTASK1:@.+]] to void
 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca float,
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: [[VAR1_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
@@ -202,23 +193,20 @@ int main() {
 
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR_REF:%.+]] = load float*, float** [[T_VAR_PTR_REF]],
+// CHECK: [[T_VAR_REF:%.+]] = load float*, float** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+// CHECK: [[VAR1_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+// CHECK: [[T_VAR1_REF:%.+]] = load float*, float** %
+
 // For + reduction operation initial value of private variable is 0.
 // CHECK: store float 0.0{{.+}}, float* [[T_VAR_PRIV]],
 
-// CHECK: [[VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_PTR_REF:%.+]],
 // For & reduction operation initial value of private variable is ones in all bits.
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 
-// CHECK: [[VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR1_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** [[VAR_PTR_REF:%.+]],
 // For && reduction operation initial value of private variable is 1.0.
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[VAR1_PRIV]])
 
-// CHECK: [[T_VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR1_REF:%.+]] = load float*, float** [[T_VAR1_PTR_REF]],
 // For min reduction operation initial value of private variable is largest repesentable value.
 // CHECK: store float 0x47EFFFFFE0000000, float* [[T_VAR1_PRIV]],
 
@@ -226,16 +214,16 @@ int main() {
 
 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
 
-// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 0
+// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 0
 // CHECK: [[BITCAST:%.+]] = bitcast float* [[T_VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR_PRIV_REF]],
-// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 1
+// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 1
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR_PRIV_REF]],
-// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 2
+// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 2
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_FLOAT_TY]]* [[VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR1_PRIV_REF]],
-// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 3
+// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 3
 // CHECK: [[BITCAST:%.+]] = bitcast float* [[T_VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR1_PRIV_REF]],
 
@@ -359,7 +347,6 @@ int main() {
 // break;
 // CHECK: br label %[[RED_DONE]]
 // CHECK: [[RED_DONE]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -373,38 +360,38 @@ int main() {
 // }
 // CHECK: define internal void [[REDUCTION_FUNC]](i8*, i8*)
 // t_var_lhs = (float*)lhs[0];
-// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR_RHS_REF]],
 // CHECK: [[T_VAR_RHS:%.+]] = bitcast i8* [[T_VAR_RHS_VOID]] to float*
 // t_var_rhs = (float*)rhs[0];
-// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR_LHS_REF]],
 // CHECK: [[T_VAR_LHS:%.+]] = bitcast i8* [[T_VAR_LHS_VOID]] to float*
 
 // var_lhs = (S<float>*)lhs[1];
-// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 1
+// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 1
 // CHECK: [[VAR_RHS_VOID:%.+]] = load i8*, i8** [[VAR_RHS_REF]],
 // CHECK: [[VAR_RHS:%.+]] = bitcast i8* [[VAR_RHS_VOID]] to [[S_FLOAT_TY]]*
 // var_rhs = (S<float>*)rhs[1];
-// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 1
+// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 1
 // CHECK: [[VAR_LHS_VOID:%.+]] = load i8*, i8** [[VAR_LHS_REF]],
 // CHECK: [[VAR_LHS:%.+]] = bitcast i8* [[VAR_LHS_VOID]] to [[S_FLOAT_TY]]*
 
 // var1_lhs = (S<float>*)lhs[2];
-// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 2
+// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 2
 // CHECK: [[VAR1_RHS_VOID:%.+]] = load i8*, i8** [[VAR1_RHS_REF]],
 // CHECK: [[VAR1_RHS:%.+]] = bitcast i8* [[VAR1_RHS_VOID]] to [[S_FLOAT_TY]]*
 // var1_rhs = (S<float>*)rhs[2];
-// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 2
+// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 2
 // CHECK: [[VAR1_LHS_VOID:%.+]] = load i8*, i8** [[VAR1_LHS_REF]],
 // CHECK: [[VAR1_LHS:%.+]] = bitcast i8* [[VAR1_LHS_VOID]] to [[S_FLOAT_TY]]*
 
 // t_var1_lhs = (float*)lhs[3];
-// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_RHS_REF]],
 // CHECK: [[T_VAR1_RHS:%.+]] = bitcast i8* [[T_VAR1_RHS_VOID]] to float*
 // t_var1_rhs = (float*)rhs[3];
-// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_LHS_REF]],
 // CHECK: [[T_VAR1_LHS:%.+]] = bitcast i8* [[T_VAR1_LHS_VOID]] to float*
 
@@ -445,42 +432,67 @@ int main() {
 // CHECK: store float [[UP]], float* [[T_VAR1_LHS]],
 // CHECK: ret void
 
+// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
+// CHECK: [[T_VAR_PRIV:%.+]] = alloca float,
+// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[VAR1_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[T_VAR1_PRIV:%.+]] = alloca float,
+
+// CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
+
+// CHECK: [[T_VAR_REF:%.+]] = load float*, float** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+// CHECK: [[VAR1_REF:%.+]] = load [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]** %
+// CHECK: [[T_VAR1_REF:%.+]] = load float*, float** %
+
+// For + reduction operation initial value of private variable is 0.
+// CHECK: store float 0.0{{.+}}, float* [[T_VAR_PRIV]],
+
+// For & reduction operation initial value of private variable is ones in all bits.
+// CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
+
+// For && reduction operation initial value of private variable is 1.0.
+// CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[VAR1_PRIV]])
+
+// For min reduction operation initial value of private variable is largest repesentable value.
+// CHECK: store float 0x47EFFFFFE0000000, float* [[T_VAR1_PRIV]],
+
+// CHECK-NOT: call i32 @__kmpc_reduce
+
+// CHECK: ret void
+
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_TMAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [2 x i32]*, i32*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*, [[S_INT_TY]]*, i32*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
-// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
-// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
-// CHECK: [[VAR1_PRIV:%.+]] = alloca [[S_INT_TY]],
-// CHECK: [[T_VAR1_PRIV:%.+]] = alloca i{{[0-9]+}},
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
+// CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
+// CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
+// CHECK: [[VAR1_PRIV:%.+]] = alloca [[S_INT_TY]], align 128
+// CHECK: [[T_VAR1_PRIV:%.+]] = alloca i{{[0-9]+}}, align 128
 
 // Reduction list for runtime.
 // CHECK: [[RED_LIST:%.+]] = alloca [4 x i8*],
 
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+// CHECK: [[VAR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+// CHECK: [[T_VAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+
 // For + reduction operation initial value of private variable is 0.
 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[T_VAR_PRIV]],
 
-// CHECK: [[VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PTR_REF:%.+]],
 // For & reduction operation initial value of private variable is ones in all bits.
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[VAR_PRIV]])
 
-// CHECK: [[VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PTR_REF:%.+]],
 // For && reduction operation initial value of private variable is 1.0.
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[VAR1_PRIV]])
 
-// CHECK: [[T_VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR1_PTR_REF]],
 // For min reduction operation initial value of private variable is largest repesentable value.
 // CHECK: store i{{[0-9]+}} 2147483647, i{{[0-9]+}}* [[T_VAR1_PRIV]],
 
@@ -488,16 +500,16 @@ int main() {
 
 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
 
-// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 0
+// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 0
 // CHECK: [[BITCAST:%.+]] = bitcast i{{[0-9]+}}* [[T_VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR_PRIV_REF]],
-// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 1
+// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 1
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR_PRIV_REF]],
-// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 2
+// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 2
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR1_PRIV_REF]],
-// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 3
+// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 3
 // CHECK: [[BITCAST:%.+]] = bitcast i{{[0-9]+}}* [[T_VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR1_PRIV_REF]],
 
@@ -594,7 +606,6 @@ int main() {
 // break;
 // CHECK: br label %[[RED_DONE]]
 // CHECK: [[RED_DONE]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]*
@@ -608,38 +619,38 @@ int main() {
 // }
 // CHECK: define internal void [[REDUCTION_FUNC]](i8*, i8*)
 // t_var_lhs = (i{{[0-9]+}}*)lhs[0];
-// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR_RHS_REF]],
 // CHECK: [[T_VAR_RHS:%.+]] = bitcast i8* [[T_VAR_RHS_VOID]] to i{{[0-9]+}}*
 // t_var_rhs = (i{{[0-9]+}}*)rhs[0];
-// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR_LHS_REF]],
 // CHECK: [[T_VAR_LHS:%.+]] = bitcast i8* [[T_VAR_LHS_VOID]] to i{{[0-9]+}}*
 
 // var_lhs = (S<i{{[0-9]+}}>*)lhs[1];
-// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 1
+// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 1
 // CHECK: [[VAR_RHS_VOID:%.+]] = load i8*, i8** [[VAR_RHS_REF]],
 // CHECK: [[VAR_RHS:%.+]] = bitcast i8* [[VAR_RHS_VOID]] to [[S_INT_TY]]*
 // var_rhs = (S<i{{[0-9]+}}>*)rhs[1];
-// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 1
+// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 1
 // CHECK: [[VAR_LHS_VOID:%.+]] = load i8*, i8** [[VAR_LHS_REF]],
 // CHECK: [[VAR_LHS:%.+]] = bitcast i8* [[VAR_LHS_VOID]] to [[S_INT_TY]]*
 
 // var1_lhs = (S<i{{[0-9]+}}>*)lhs[2];
-// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 2
+// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 2
 // CHECK: [[VAR1_RHS_VOID:%.+]] = load i8*, i8** [[VAR1_RHS_REF]],
 // CHECK: [[VAR1_RHS:%.+]] = bitcast i8* [[VAR1_RHS_VOID]] to [[S_INT_TY]]*
 // var1_rhs = (S<i{{[0-9]+}}>*)rhs[2];
-// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 2
+// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 2
 // CHECK: [[VAR1_LHS_VOID:%.+]] = load i8*, i8** [[VAR1_LHS_REF]],
 // CHECK: [[VAR1_LHS:%.+]] = bitcast i8* [[VAR1_LHS_VOID]] to [[S_INT_TY]]*
 
 // t_var1_lhs = (i{{[0-9]+}}*)lhs[3];
-// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_RHS_REF]],
 // CHECK: [[T_VAR1_RHS:%.+]] = bitcast i8* [[T_VAR1_RHS_VOID]] to i{{[0-9]+}}*
 // t_var1_rhs = (i{{[0-9]+}}*)rhs[3];
-// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_LHS_REF]],
 // CHECK: [[T_VAR1_LHS:%.+]] = bitcast i8* [[T_VAR1_LHS_VOID]] to i{{[0-9]+}}*
 
diff --git a/test/OpenMP/parallel_reduction_messages.cpp b/test/OpenMP/parallel_reduction_messages.cpp
index c93cfbd9efaf..b29f7c98a574 100644
--- a/test/OpenMP/parallel_reduction_messages.cpp
+++ b/test/OpenMP/parallel_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 100 -o - %s
 
 void foo() {
 }
@@ -16,7 +18,7 @@ class S2 {
 public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
-  static float S2s;
+  static float S2s; // expected-note 2 {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note 2 {{'S2sc' defined here}}
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -93,7 +99,7 @@ T tmain(T argc) {
   foo();
 #pragma omp parallel reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{invalid operands to binary expression ('float' and 'float')}}
   foo();
-#pragma omp parallel reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp parallel reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   foo();
 #pragma omp parallel reduction(foo : argc) //expected-error {{incorrect reduction identifier, expected one of '+', '-', '*', '&', '|', '^', '&&', '||', 'min' or 'max'}}
   foo();
@@ -101,23 +107,23 @@ T tmain(T argc) {
   foo();
 #pragma omp parallel reduction(^ : T) // expected-error {{'T' does not refer to a value}}
   foo();
-#pragma omp parallel reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   foo();
-#pragma omp parallel reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   foo();
-#pragma omp parallel reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp parallel reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   foo();
-#pragma omp parallel reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   foo();
-#pragma omp parallel reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   foo();
-#pragma omp parallel reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp parallel reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   foo();
 #pragma omp parallel reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   foo();
-#pragma omp parallel reduction(&& : S2::S2s)
+#pragma omp parallel reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   foo();
-#pragma omp parallel reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp parallel reduction(+ : h, k) // expected-error {{threadprivate or thread local variable cannot be reduction}}
   foo();
@@ -130,7 +136,7 @@ T tmain(T argc) {
   foo();
 #pragma omp parallel reduction(+ : p), reduction(+ : p) // expected-error 3 {{variable can appear only once in OpenMP 'reduction' clause}} expected-note 3 {{previously referenced here}}
   foo();
-#pragma omp parallel reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp parallel shared(i)
 #pragma omp parallel reduction(min : i)
@@ -188,7 +194,7 @@ int main(int argc, char **argv) {
   foo();
 #pragma omp parallel reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   foo();
-#pragma omp parallel reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp parallel reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   foo();
 #pragma omp parallel reduction(~ : argc) // expected-error {{expected unqualified-id}}
   foo();
@@ -196,23 +202,23 @@ int main(int argc, char **argv) {
   foo();
 #pragma omp parallel reduction(^ : S1) // expected-error {{'S1' does not refer to a value}}
   foo();
-#pragma omp parallel reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   foo();
-#pragma omp parallel reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   foo();
-#pragma omp parallel reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp parallel reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   foo();
-#pragma omp parallel reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   foo();
-#pragma omp parallel reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   foo();
-#pragma omp parallel reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp parallel reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   foo();
 #pragma omp parallel reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   foo();
-#pragma omp parallel reduction(&& : S2::S2s)
+#pragma omp parallel reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   foo();
-#pragma omp parallel reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp parallel reduction(& : e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{nvalid operands to binary expression ('S4' and 'S4')}} expected-error {{calling a private constructor of class 'S5'}} expected-error {{invalid operands to binary expression ('S5' and 'S5')}}
   foo();
@@ -227,7 +233,7 @@ int main(int argc, char **argv) {
   foo();
 #pragma omp parallel reduction(+ : p), reduction(+ : p) // expected-error {{variable can appear only once in OpenMP 'reduction' clause}} expected-note {{previously referenced here}}
   foo();
-#pragma omp parallel reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp parallel shared(i)
 #pragma omp parallel reduction(min : i)
@@ -243,6 +249,9 @@ int main(int argc, char **argv) {
   for (int i = 0; i < 10; ++i)
 #pragma omp parallel reduction(+ : fl)
     foo();
+  static int m;
+#pragma omp parallel reduction(+ : m) // OK
+  m++;
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/parallel_sections_ast_print.cpp b/test/OpenMP/parallel_sections_ast_print.cpp
index 7667e45bfdbc..9f5c1fadbeb8 100644
--- a/test/OpenMP/parallel_sections_ast_print.cpp
+++ b/test/OpenMP/parallel_sections_ast_print.cpp
@@ -37,7 +37,7 @@ T tmain(T argc, T *argv) {
   {
     a = 2;
   }
-#pragma omp parallel sections default(none), private(argc, b) firstprivate(argv) shared(d) if (argc > 0) num_threads(C) copyin(S < T > ::TS) proc_bind(master) reduction(+ : c) reduction(max : e)
+#pragma omp parallel sections default(none), private(argc, b) firstprivate(argv) shared(d) if (parallel: argc > 0) num_threads(C) copyin(S < T > ::TS) proc_bind(master) reduction(+ : c) reduction(max : e)
   {
     foo();
   }
@@ -58,7 +58,7 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: }
-// CHECK-NEXT: #pragma omp parallel sections default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) num_threads(5) copyin(S<int>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp parallel sections default(none) private(argc,b) firstprivate(argv) shared(d) if(parallel: argc > 0) num_threads(5) copyin(S<int>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
 // CHECK-NEXT: {
 // CHECK-NEXT: foo();
 // CHECK-NEXT: }
@@ -76,7 +76,7 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: }
-// CHECK-NEXT: #pragma omp parallel sections default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) num_threads(1) copyin(S<long>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp parallel sections default(none) private(argc,b) firstprivate(argv) shared(d) if(parallel: argc > 0) num_threads(1) copyin(S<long>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
 // CHECK-NEXT: {
 // CHECK-NEXT: foo();
 // CHECK-NEXT: }
@@ -94,7 +94,7 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: {
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: }
-// CHECK-NEXT: #pragma omp parallel sections default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) num_threads(C) copyin(S<T>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp parallel sections default(none) private(argc,b) firstprivate(argv) shared(d) if(parallel: argc > 0) num_threads(C) copyin(S<T>::TS) proc_bind(master) reduction(+: c) reduction(max: e)
 // CHECK-NEXT: {
 // CHECK-NEXT: foo();
 // CHECK-NEXT: }
diff --git a/test/OpenMP/parallel_sections_codegen.cpp b/test/OpenMP/parallel_sections_codegen.cpp
index 2e9e1f95aec5..b8c1e39d8f9e 100644
--- a/test/OpenMP/parallel_sections_codegen.cpp
+++ b/test/OpenMP/parallel_sections_codegen.cpp
@@ -5,7 +5,6 @@
 // REQUIRES: x86-registered-target
 #ifndef HEADER
 #define HEADER
-// CHECK: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-LABEL: foo
 void foo() {};
 // CHECK-LABEL: bar
@@ -22,9 +21,9 @@ T tmain() {
 
 // CHECK-LABEL: @main
 int main() {
-// CHECK: call void (%{{.+}}*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %{{.+}}*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*), i8* %{{.+}})
+// CHECK: call void (%{{.+}}*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* [[OMP_PARALLEL_FUNC:@.+]] to void (i32*, i32*, ...)*))
 // CHECK-LABEL: }
-// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* [[GTID_PARAM_ADDR:%.+]], i32* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: define internal void [[OMP_PARALLEL_FUNC]](i32* noalias [[GTID_PARAM_ADDR:%.+]], i32* noalias %{{.+}})
 // CHECK: store i32* [[GTID_PARAM_ADDR]], i32** [[GTID_REF_ADDR:%.+]],
 #pragma omp parallel sections
   {
@@ -73,7 +72,6 @@ int main() {
 // CHECK:      [[INNER_LOOP_END]]
   }
 // CHECK:      call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
-// CHECK:      call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]],
   return tmain<int>();
 }
 
@@ -89,7 +87,6 @@ int main() {
 // CHECK:       call void @__kmpc_end_single(
 // CHECK-NEXT:  br label %[[END]]
 // CHECK:       [[END]]
-// CHECK-NEXT:  call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]],
 // CHECK-NEXT:  ret
 // CHECK:       [[TERM_LPAD]]
 // CHECK:       call void @__clang_call_terminate(i8*
diff --git a/test/OpenMP/parallel_sections_firstprivate_messages.cpp b/test/OpenMP/parallel_sections_firstprivate_messages.cpp
index 84d18596c99c..b733aabb7f25 100644
--- a/test/OpenMP/parallel_sections_firstprivate_messages.cpp
+++ b/test/OpenMP/parallel_sections_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i;                              // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel sections firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   {
     foo();
@@ -130,7 +130,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel sections firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp parallel sections firstprivate(j)
   {
     foo();
   }
@@ -171,7 +171,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i;                              // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel sections firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   {
     foo();
@@ -266,7 +266,7 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp parallel sections firstprivate(j)
   {
     foo();
   }
@@ -298,6 +298,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int r;
+#pragma omp parallel sections firstprivate(r) // OK
+  {
+    foo();
+  }
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/parallel_sections_if_messages.cpp b/test/OpenMP/parallel_sections_if_messages.cpp
index 03182b4c8fe1..c2e3f5ec99bd 100644
--- a/test/OpenMP/parallel_sections_if_messages.cpp
+++ b/test/OpenMP/parallel_sections_if_messages.cpp
@@ -55,6 +55,30 @@ int tmain(T argc, S **argv) {
   {
     foo();
   }
+  #pragma omp parallel sections if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc)
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel sections'}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel sections' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  {
+    foo();
+  }
 
   return 0;
 }
@@ -108,6 +132,30 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  #pragma omp parallel sections if(parallel : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc)
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp parallel sections'}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc) if (parallel:argc) // expected-error {{directive '#pragma omp parallel sections' cannot contain more than one 'if' clause with 'parallel' name modifier}}
+  {
+    foo();
+  }
+  #pragma omp parallel sections if(parallel : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  {
+    foo();
+  }
 
   return tmain(argc, argv);
 }
diff --git a/test/OpenMP/parallel_sections_lastprivate_messages.cpp b/test/OpenMP/parallel_sections_lastprivate_messages.cpp
index 10cc36e5ea7f..af3c5e2a575b 100644
--- a/test/OpenMP/parallel_sections_lastprivate_messages.cpp
+++ b/test/OpenMP/parallel_sections_lastprivate_messages.cpp
@@ -16,7 +16,7 @@ public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
   const S2 &operator=(const S2 &) const;
-  static float S2s;
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note {{static data member is predetermined as shared}}
@@ -66,7 +66,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                             // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel sections lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   {
     foo();
@@ -131,7 +131,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel sections lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp parallel sections lastprivate(j)
   {
     foo();
   }
@@ -158,7 +158,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i;                             // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel sections lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   {
     foo();
@@ -220,7 +220,7 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections lastprivate(S2::S2s)
+#pragma omp parallel sections lastprivate(S2::S2s) // expected-error {{shared variable cannot be lastprivate}}
   {
     foo();
   }
@@ -262,7 +262,7 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp parallel sections lastprivate(j)
   {
     foo();
   }
@@ -274,5 +274,10 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int r;
+#pragma omp parallel sections lastprivate(r) // OK
+  {
+    foo();
+  }
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/parallel_sections_num_threads_messages.cpp b/test/OpenMP/parallel_sections_num_threads_messages.cpp
index a50025660da9..cda3841ace97 100644
--- a/test/OpenMP/parallel_sections_num_threads_messages.cpp
+++ b/test/OpenMP/parallel_sections_num_threads_messages.cpp
@@ -23,7 +23,7 @@ T tmain(T argc, S **argv) {
   {foo();}
   #pragma omp parallel sections num_threads ((argc > 0) ? argv[1] : argv[2]) // expected-error 2 {{expression must have integral or unscoped enumeration type, not 'char *'}}
   {foo();}
-  #pragma omp parallel sections num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel sections' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel sections num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel sections' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   {foo();}
   #pragma omp parallel sections num_threads (S) // expected-error {{'S' does not refer to a value}}
   {foo();}
@@ -31,7 +31,7 @@ T tmain(T argc, S **argv) {
   {foo();}
   #pragma omp parallel sections num_threads (argc)
   {foo();}
-  #pragma omp parallel sections num_threads (N) // expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel sections num_threads (N) // expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   {foo();}
 
   return argc;
@@ -50,7 +50,7 @@ int main(int argc, char **argv) {
   {foo();}
   #pragma omp parallel sections num_threads (argc > 0 ? argv[1] : argv[2]) // expected-error {{integral }}
   {foo();}
-  #pragma omp parallel sections num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel sections' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a positive integer value}}
+  #pragma omp parallel sections num_threads (foobool(argc)), num_threads (true), num_threads (-5) // expected-error 2 {{directive '#pragma omp parallel sections' cannot contain more than one 'num_threads' clause}} expected-error {{argument to 'num_threads' clause must be a strictly positive integer value}}
   {foo();}
   #pragma omp parallel sections num_threads (S1) // expected-error {{'S1' does not refer to a value}}
   {foo();}
diff --git a/test/OpenMP/parallel_sections_private_messages.cpp b/test/OpenMP/parallel_sections_private_messages.cpp
index d9c4404bd984..ac9280e74ce1 100644
--- a/test/OpenMP/parallel_sections_private_messages.cpp
+++ b/test/OpenMP/parallel_sections_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                         // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel sections private // expected-error {{expected '(' after 'private'}}
   {
     foo();
@@ -112,7 +112,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel sections private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp parallel sections private(j)
   {
     foo();
   }
@@ -135,7 +135,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;                         // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel sections private // expected-error {{expected '(' after 'private'}}
   {
     foo();
@@ -198,7 +198,7 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp parallel sections private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp parallel sections private(j)
   {
     foo();
   }
@@ -206,6 +206,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int m;
+#pragma omp parallel sections private(m)
+  {
+    foo();
+  }
 
   return 0;
 }
diff --git a/test/OpenMP/parallel_sections_reduction_messages.cpp b/test/OpenMP/parallel_sections_reduction_messages.cpp
index 2ed828dabfa7..eff1849d7132 100644
--- a/test/OpenMP/parallel_sections_reduction_messages.cpp
+++ b/test/OpenMP/parallel_sections_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 100 -o - %s
 
 void foo() {
 }
@@ -16,7 +18,7 @@ class S2 {
 public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
-  static float S2s;
+  static float S2s; // expected-note 2 {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note 2 {{'S2sc' defined here}}
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -111,7 +117,7 @@ T tmain(T argc) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp parallel sections reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   {
     foo();
   }
@@ -127,27 +133,27 @@ T tmain(T argc) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel sections reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel sections reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp parallel sections reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel sections reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel sections reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp parallel sections reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   {
     foo();
   }
@@ -155,11 +161,11 @@ T tmain(T argc) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(&& : S2::S2s)
+#pragma omp parallel sections reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel sections reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -184,7 +190,7 @@ T tmain(T argc) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel sections reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -264,7 +270,7 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp parallel sections reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   {
     foo();
   }
@@ -280,27 +286,27 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp parallel sections reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp parallel sections reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp parallel sections reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp parallel sections reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp parallel sections reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp parallel sections reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   {
     foo();
   }
@@ -308,11 +314,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(&& : S2::S2s)
+#pragma omp parallel sections reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   {
     foo();
   }
-#pragma omp parallel sections reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel sections reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -341,7 +347,7 @@ int main(int argc, char **argv) {
   {
     foo();
   }
-#pragma omp parallel sections reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp parallel sections reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -361,6 +367,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int m;
+#pragma omp parallel sections reduction(+ : m) // OK
+  {
+    foo();
+  }
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/sections_codegen.cpp b/test/OpenMP/sections_codegen.cpp
index 2c257e83ae30..44fdefeee733 100644
--- a/test/OpenMP/sections_codegen.cpp
+++ b/test/OpenMP/sections_codegen.cpp
@@ -95,8 +95,7 @@ int main() {
 // CHECK:       call void @__kmpc_end_single(
 // CHECK-NEXT:  br label %[[END]]
 // CHECK:       [[END]]
-// CHECK-NEXT:  call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_SINGLE_LOC]],
-// CHECK:  call i32 @__kmpc_cancel_barrier(
+// CHECK-NEXT:  call void @__kmpc_barrier(%{{.+}}* [[IMPLICIT_BARRIER_SINGLE_LOC]],
 // CHECK:  ret
 // CHECK:       [[TERM_LPAD]]
 // CHECK:       call void @__clang_call_terminate(i8*
diff --git a/test/OpenMP/sections_firstprivate_codegen.cpp b/test/OpenMP/sections_firstprivate_codegen.cpp
index af9c1fe5fbda..f673597f660b 100644
--- a/test/OpenMP/sections_firstprivate_codegen.cpp
+++ b/test/OpenMP/sections_firstprivate_codegen.cpp
@@ -30,7 +30,6 @@ struct S {
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
 
 template <typename T>
 T tmain() {
@@ -59,50 +58,70 @@ int vec[] = {1, 2};
 S<float> s_arr[] = {1, 2};
 // CHECK-DAG: [[VAR:@.+]] = global [[S_FLOAT_TY]] zeroinitializer,
 S<float> var(3);
+// CHECK-DAG: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
 // CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-DAG: [[SECTIONS_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 194, i32 0, i32 0, i8*
 
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
 // CHECK: ([[S_FLOAT_TY]]*)* [[S_FLOAT_TY_DESTR:@[^ ]+]] {{[^,]+}}, {{.+}}([[S_FLOAT_TY]]* [[TEST]]
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp sections firstprivate(g)
+#pragma omp sections firstprivate(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR_REF:%.+]])
     // Skip temp vars for loop
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
+
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[SIVAR1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_REF]], i{{[0-9]+}}** [[SIVAR_REF_ADDR:%.+]],
+    // LAMBDA: [[SIVAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_REF_ADDR]],
     // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]]
     // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+
+    // LAMBDA: [[SIVAR1_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR1_REF]]
+    // LAMBDA: store i{{[0-9]+}} [[SIVAR1_VAL]], i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]]
+
     // LAMBDA: call void @__kmpc_barrier(
-    g = 1;
+    {
+      g = 1;
+      sivar = 10;
+    }
     // LAMBDA: call void @__kmpc_for_static_init_4(
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 10, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call void @__kmpc_for_static_fini(
-    // LAMBDA: call i32 @__kmpc_cancel_barrier(
+    // LAMBDA: call void @__kmpc_barrier(
 #pragma omp section
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 20;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 20, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -113,11 +132,11 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp sections firstprivate(g)
+#pragma omp sections firstprivate(g, sivar)
    {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR_REF:%.+]])
     // Skip temp vars for loop
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
@@ -125,36 +144,56 @@ int main() {
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[SIVAR1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+
+    // BLOCKS: store i{{[0-9]+}}* [[SIVAR_REF]], i{{[0-9]+}}** {{.+}},
+    // BLOCKS: [[SIVAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** {{.+}},
+
     // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]]
     // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+
+    // BLOCKS: [[SIVAR1_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR1_REF]],
+    // BLOCKS: store i{{[0-9]+}} [[SIVAR1_VAL]], i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     // BLOCKS: call void @__kmpc_barrier(
-    g = 1;
+    {
+      g = 1;
+      sivar = 10;
+    }
     // BLOCKS: call void @__kmpc_for_static_init_4(
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 10, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
     // BLOCKS: call void @__kmpc_for_static_fini(
-    // BLOCKS: call i32 @__kmpc_cancel_barrier(
+    // BLOCKS: call void @__kmpc_barrier(
 #pragma omp section
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 20;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 20, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
   }();
   return 0;
 #else
-#pragma omp sections firstprivate(t_var, vec, s_arr, var) nowait
+#pragma omp sections firstprivate(t_var, vec, s_arr, var, sivar) nowait
   {
     {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 31;
     }
   }
   return tmain<int>();
@@ -168,12 +207,12 @@ int main() {
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 
 // CHECK: call i32 @__kmpc_single(
 // firstprivate t_var(t_var)
 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR]],
 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]],
-
 // firstprivate vec(vec)
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* bitcast ([2 x i{{[0-9]+}}]* [[VEC]] to i8*),
@@ -195,6 +234,10 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 
+// firstprivate isvar
+// CHEC: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR]],
+// CHEC: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIV]],
+
 // ~(firstprivate var), ~(firstprivate s_arr)
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -209,11 +252,11 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // Skip temp vars for loop
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -226,22 +269,21 @@ int main() {
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+
 // firstprivate t_var(t_var)
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]],
 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]],
 
 // firstprivate vec(vec)
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
 
 // firstprivate s_arr(s_arr)
-// CHECK: [[S_ARR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF]],
 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2
 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]]
@@ -253,8 +295,6 @@ int main() {
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
 
 // firstprivate var(var)
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]],
 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]])
 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
@@ -272,7 +312,7 @@ int main() {
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]*
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[SECTIONS_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[SECTIONS_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 #endif
 
diff --git a/test/OpenMP/sections_firstprivate_messages.cpp b/test/OpenMP/sections_firstprivate_messages.cpp
index ff7614967ee8..cd2b4b883873 100644
--- a/test/OpenMP/sections_firstprivate_messages.cpp
+++ b/test/OpenMP/sections_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp sections firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   {
@@ -135,7 +135,7 @@ int foomain(int argc, char **argv) {
   {
     int v = 0;
     int i;                           // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp sections' directive into a parallel or another task region?}}
-#pragma omp sections firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp sections firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     {
       foo();
     }
@@ -143,7 +143,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp sections firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp sections firstprivate(j)
   {
     foo();
   }
@@ -186,7 +186,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp sections firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   {
@@ -304,7 +304,7 @@ int main(int argc, char **argv) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp sections firstprivate(j)
   {
     foo();
   }
@@ -322,7 +322,7 @@ int main(int argc, char **argv) {
   {
     int v = 0;
     int i;                           // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp sections' directive into a parallel or another task region?}}
-#pragma omp sections firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp sections firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     {
       foo();
     }
@@ -338,6 +338,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int r;
+#pragma omp sections firstprivate(r) // OK
+  {
+    foo();
+  }
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/sections_lastprivate_codegen.cpp b/test/OpenMP/sections_lastprivate_codegen.cpp
index c2f68aa0fe7d..a1ff007fd61d 100644
--- a/test/OpenMP/sections_lastprivate_codegen.cpp
+++ b/test/OpenMP/sections_lastprivate_codegen.cpp
@@ -21,10 +21,8 @@ struct S {
 volatile int g = 1212;
 
 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[CAP_MAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]* }
+// CHECK [[CAP_MAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}}* }
 // CHECK: [[S_INT_TY:%.+]] = type { i32 }
-// CHECK: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
-// CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-DAG: [[SINGLE_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 322, i32 0, i32 0, i8*
 // CHECK-DAG: [[SECTIONS_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 194, i32 0, i32 0, i8*
 // CHECK-DAG: [[X:@.+]] = global double 0.0
@@ -53,33 +51,45 @@ using A::x;
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp sections lastprivate(g)
+#pragma omp sections lastprivate(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias [[GTID:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR_REF:%.+]])
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: alloca i{{[0-9]+}},
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    // LAMBDA: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
-    // LAMBDA: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-    // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+    // LAMBDA: [[SIVAR1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_REF]], i{{[0-9]+}}** %{{.+}},
+    // LAMBDA: [[SIVAR_REF_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
+
+    // LAMBDA: [[GTID_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}, align 8
+    // LAMBDA: [[GTID_ADDR_REF:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_ADDR]], align 4
+
+    // LAMBDA: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID_ADDR_REF]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 13, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
-    // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
-    g = 1;
+    // LAMBDA: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID_ADDR_REF]])
+    {
+      g = 1;
+      sivar = 13;
+    }
     // Check for final copying of private values back to original vars.
     // LAMBDA: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
     // LAMBDA: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
@@ -90,18 +100,26 @@ int main() {
     // original g=private_g;
     // LAMBDA: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
     // LAMBDA: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]],
+
+    // original sivar = private sivar;
+    // LAMBDA: [[SIVAR1_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} [[SIVAR1_VAL]], i{{[0-9]+}}* [[SIVAR_REF_ADDR]],
     // LAMBDA: br label %[[LAST_DONE]]
     // LAMBDA: [[LAST_DONE]]
-    // LAMBDA: call i32 @__kmpc_cancel_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
+    // LAMBDA: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID_ADDR_REF]])
 #pragma omp section
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 23;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 23, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -112,28 +130,39 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp sections lastprivate(g)
+#pragma omp sections lastprivate(g, sivar)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias [[GTID:%.+]], i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR:%.+]])
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: alloca i{{[0-9]+}},
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    // BLOCKS: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}}
-    // BLOCKS: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-    // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
+    // BLOCKS: [[SIVAR1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+
+    // BLOCKS: store i{{[0-9]+}}* [[SIVAR]], i{{[0-9]+}}** [[SIVAR_ADDR:%.+]],
+    // BLOCKS: [[SIVAR_REF_ADDR:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_ADDR]],
+
+    // BLOCKS: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID:%.+]], align 8
+    // BLOCKS: [[GTID_ADDR_REF:%.+]] = load i32, i32* [[GTID_ADDR]], align 4
+    // BLOCKS: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 [[GTID_ADDR_REF]], i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 17, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
-    // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID]])
-    g = 1;
+    // BLOCKS: call void @__kmpc_for_static_fini(%{{.+}}* @{{.+}}, i32 [[GTID_ADDR_REF]])
+    {
+      g = 1;
+      sivar = 17;
+    }
     // Check for final copying of private values back to original vars.
     // BLOCKS: [[IS_LAST_VAL:%.+]] = load i32, i32* [[IS_LAST_ADDR]],
     // BLOCKS: [[IS_LAST_ITER:%.+]] = icmp ne i32 [[IS_LAST_VAL]], 0
@@ -144,16 +173,24 @@ int main() {
     // original g=private_g;
     // BLOCKS: [[G_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
     // BLOCKS: store volatile i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G]],
+
+    // original sivar = private sivar;
+    // BLOCKS: [[SIVAR1_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} [[SIVAR1_VAL]], i{{[0-9]+}}* [[SIVAR_REF_ADDR]],
     // BLOCKS: br label %[[LAST_DONE]]
     // BLOCKS: [[LAST_DONE]]
-    // BLOCKS: call i32 @__kmpc_cancel_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID]])
+    // BLOCKS: call void @__kmpc_barrier(%{{.+}}* @{{.+}}, i{{[0-9]+}} [[GTID_ADDR_REF]])
 #pragma omp section
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 29;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 29, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -166,11 +203,12 @@ int main() {
   S<float> s_arr[] = {1, 2};
   S<float> var(3);
 #pragma omp parallel
-#pragma omp sections lastprivate(t_var, vec, s_arr, var)
+#pragma omp sections lastprivate(t_var, vec, s_arr, var, sivar)
   {
     {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 31;
     }
   }
 #pragma omp parallel
@@ -187,14 +225,15 @@ int main() {
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, %{{.+}}*)* [[MAIN_MICROTASK1:@.+]] to void
+
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 5, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*, [[S_FLOAT_TY]]*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
+
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK1:@.+]] to void
 // CHECK: = call {{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // CHECK-NOT: alloca i{{[0-9]+}},
 // CHECK-NOT: alloca [2 x i{{[0-9]+}}],
 // CHECK-NOT: alloca [2 x [[S_FLOAT_TY]]],
@@ -205,10 +244,8 @@ int main() {
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
 // CHECK: call i32 @__kmpc_single(
 
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
+// CHECK-DAG: getelementptr inbounds [2 x i32], [2 x i32]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
+// CHECK-DAG: getelementptr inbounds [2 x [[S_FLOAT_TY]]], [2 x [[S_FLOAT_TY]]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
 
 // <Skip loop body>
 
@@ -217,12 +254,11 @@ int main() {
 
 // CHECK: call void @__kmpc_end_single(
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[SINGLE_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[SINGLE_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
 //
-// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, %{{.+}}* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK1]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[X_PRIV:%.+]] = alloca double,
 // CHECK-NOT: alloca double
 
@@ -248,18 +284,17 @@ int main() {
 // CHECK-NEXT: br label %[[LAST_DONE]]
 // CHECK: [[LAST_DONE]]
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[SECTIONS_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[SECTIONS_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -271,19 +306,16 @@ int main() {
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+
 // Check for default initialization.
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
 // CHECK-NOT: [[T_VAR_PRIV]]
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK-NOT: [[VEC_PRIV]]
-// CHECK: [[S_ARR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR_REF:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF_PTR]],
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_INT_TY]]*
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[S_ARR_PRIV_ITEM]])
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK: call {{.+}} @__kmpc_for_static_init_4(%{{.+}}* @{{.+}}, i32 %{{.+}}, i32 34, i32* [[IS_LAST_ADDR:%.+]], i32* %{{.+}}, i32* %{{.+}}, i32* %{{.+}}, i32 1, i32 1)
 // <Skip loop body>
@@ -309,6 +341,10 @@ int main() {
 // CHECK: [[S_ARR_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_REF]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = bitcast [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]] to [[S_INT_TY]]*
 // CHECK: [[S_ARR_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 2
+
+// CHK: [[SIVAR_REF:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_BEGIN]], i{{[0-9]+}} 4
+// CHK: store i{{[0-9]+}}* [[SIVAR]], i{{[0-9]+}} [[SIVAR_REF]]
+
 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_BEGIN]], [[S_ARR_END]]
 // CHECK: br i1 [[IS_EMPTY]], label %[[S_ARR_BODY_DONE:.+]], label %[[S_ARR_BODY:.+]]
 // CHECK: [[S_ARR_BODY]]
@@ -324,10 +360,7 @@ int main() {
 // CHECK-DAG: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]*
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[SECTIONS_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_REF]]
-// CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[SECTIONS_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 #endif
 
diff --git a/test/OpenMP/sections_lastprivate_messages.cpp b/test/OpenMP/sections_lastprivate_messages.cpp
index 870dd158aeeb..5f6b420cb8ea 100644
--- a/test/OpenMP/sections_lastprivate_messages.cpp
+++ b/test/OpenMP/sections_lastprivate_messages.cpp
@@ -66,7 +66,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp sections lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   {
@@ -144,7 +144,7 @@ int foomain(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp sections lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp sections lastprivate(j)
   {
     foo();
   }
@@ -172,7 +172,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp sections lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   {
@@ -300,7 +300,7 @@ int main(int argc, char **argv) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp sections lastprivate(j)
   {
     foo();
   }
@@ -314,5 +314,10 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int r;
+#pragma omp sections lastprivate(r) // OK
+  {
+    foo();
+  }
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/sections_private_codegen.cpp b/test/OpenMP/sections_private_codegen.cpp
index 9b579a2aa661..cd2218832bcf 100644
--- a/test/OpenMP/sections_private_codegen.cpp
+++ b/test/OpenMP/sections_private_codegen.cpp
@@ -20,9 +20,7 @@ struct S {
 volatile double g;
 
 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[CAP_MAIN_TY:%.+]] = type { i8 }
 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK: [[CAP_TMAIN_TY:%.+]] = type { i8 }
 template <typename T>
 T tmain() {
   S<T> test;
@@ -41,24 +39,31 @@ T tmain() {
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global double
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp sections private(g)
+#pragma omp sections private(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double,
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    g = 1;
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    {
+      g = 1;
+      sivar = 11;
+    }
     // LAMBDA: call {{.*}}void @__kmpc_for_static_init_4(
     // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 11, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store double* [[G_PRIVATE_ADDR]], double** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call {{.*}}void @__kmpc_for_static_fini(
 #pragma omp section
@@ -66,10 +71,16 @@ int main() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 22;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
+
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
       // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
+
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 22, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -80,28 +91,39 @@ int main() {
   // BLOCKS: call {{.*}}void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* {{.+}})
+  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp sections private(g)
+#pragma omp sections private(g, sivar)
     {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca double,
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
-    g = 1;
+    // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    {
+      g = 1;
+      sivar = 111;
+    }
     // BLOCKS: call {{.*}}void @__kmpc_for_static_init_4(
     // BLOCKS: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 111, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: double* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
     // BLOCKS: call {{.*}}void @__kmpc_for_static_fini(
 #pragma omp section
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 222;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store double 2.0{{.+}}, double*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 222, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -114,11 +136,12 @@ int main() {
   S<float> s_arr[] = {1, 2};
   S<float> var(3);
 #pragma omp parallel
-#pragma omp sections private(t_var, vec, s_arr, s_arr, var, var)
+#pragma omp sections private(t_var, vec, s_arr, s_arr, var, var, sivar)
   {
     {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 2;
     }
   }
   return tmain<int>();
@@ -128,23 +151,24 @@ int main() {
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK-NOT: alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK-NOT: alloca [[S_FLOAT_TY]],
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 // CHECK: call i32 @__kmpc_single(
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
+// CHECK-NOT: [[SIVAR_PRIV]]
 // CHECK: {{.+}}:
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
@@ -159,11 +183,11 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: alloca i32,
 // CHECK: alloca i32,
 // CHECK: alloca i32,
diff --git a/test/OpenMP/sections_private_messages.cpp b/test/OpenMP/sections_private_messages.cpp
index 7854b3d8bdd3..f13bbdb012e2 100644
--- a/test/OpenMP/sections_private_messages.cpp
+++ b/test/OpenMP/sections_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp sections private // expected-error {{expected '(' after 'private'}}
   {
     foo();
@@ -112,7 +112,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp sections private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp sections private(j)
   {
     foo();
   }
@@ -135,7 +135,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;                // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp sections private // expected-error {{expected '(' after 'private'}}
   {
     foo();
@@ -198,7 +198,7 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp sections private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp sections private(j)
   {
     foo();
   }
@@ -206,6 +206,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int m;
+#pragma omp sections private(m)
+  {
+    foo();
+  }
 
   return 0;
 }
diff --git a/test/OpenMP/sections_reduction_codegen.cpp b/test/OpenMP/sections_reduction_codegen.cpp
index 4d404dbcc6d1..f67977c39580 100644
--- a/test/OpenMP/sections_reduction_codegen.cpp
+++ b/test/OpenMP/sections_reduction_codegen.cpp
@@ -22,10 +22,7 @@ struct S {
 
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { float*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float*, [2 x i{{[0-9]+}}]*, [2 x [[S_FLOAT_TY]]]* }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [[S_INT_TY]]*, [[S_INT_TY]]*, i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]* }
 // CHECK-DAG: [[ATOMIC_REDUCE_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 18, i32 0, i32 0, i8*
-// CHECK-DAG: [[IMPLICIT_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 66, i32 0, i32 0, i8*
 // CHECK-DAG: [[SINGLE_BARRIER_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 322, i32 0, i32 0, i8*
 // CHECK-DAG: [[REDUCTION_LOC:@.+]] = private unnamed_addr constant %{{.+}} { i32 0, i32 18, i32 0, i32 0, i8*
 // CHECK-DAG: [[REDUCTION_LOCK:@.+]] = common global [8 x i32] zeroinitializer
@@ -55,11 +52,11 @@ int main() {
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
 #pragma omp sections reduction(+:g)
     {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* %{{.+}})
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double,
 
     // Reduction list for runtime.
@@ -74,7 +71,7 @@ int main() {
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call void @__kmpc_for_static_fini(
 
-    // LAMBDA: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i32 0, i32 0
+    // LAMBDA: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i64 0, i64 0
     // LAMBDA: [[BITCAST:%.+]] = bitcast double* [[G_PRIVATE_ADDR]] to i8*
     // LAMBDA: store i8* [[BITCAST]], i8** [[G_PRIV_REF]],
     // LAMBDA: call i32 @__kmpc_reduce(
@@ -115,11 +112,11 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
 #pragma omp sections reduction(-:g)
     {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* %{{.+}})
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca double,
 
     // Reduction list for runtime.
@@ -135,7 +132,7 @@ int main() {
     // BLOCKS: call void {{%.+}}(i8
     // BLOCKS: call void @__kmpc_for_static_fini(
 
-    // BLOCKS: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i32 0, i32 0
+    // BLOCKS: [[G_PRIV_REF:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[RED_LIST]], i64 0, i64 0
     // BLOCKS: [[BITCAST:%.+]] = bitcast double* [[G_PRIVATE_ADDR]] to i8*
     // BLOCKS: store i8* [[BITCAST]], i8** [[G_PRIV_REF]],
     // BLOCKS: call i32 @__kmpc_reduce(
@@ -192,13 +189,12 @@ int main() {
 // CHECK: define {{.*}}i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, float*, [[S_FLOAT_TY]]*, [[S_FLOAT_TY]]*, float*, [2 x i32]*, [2 x [[S_FLOAT_TY]]]*)* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call {{.*}}i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call {{.*}} [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // CHECK-NOT: alloca float,
 // CHECK-NOT: alloca [[S_FLOAT_TY]],
 // CHECK-NOT: alloca [[S_FLOAT_TY]],
@@ -210,30 +206,23 @@ int main() {
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
 // CHECK: call i32 @__kmpc_single(
 
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK-DAG: getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-
 // CHECK-NOT: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-NOT: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
 
 // CHECK: call void @__kmpc_end_single(
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[SINGLE_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[SINGLE_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 
 // CHECK: ret void
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_TMAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 6, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [[S_INT_TY]]*, [[S_INT_TY]]*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
 // CHECK: alloca i{{[0-9]+}},
@@ -249,23 +238,20 @@ int main() {
 
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+// CHECK: [[VAR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+// CHECK: [[T_VAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+
 // For + reduction operation initial value of private variable is 0.
 // CHECK: store i{{[0-9]+}} 0, i{{[0-9]+}}* [[T_VAR_PRIV]],
 
-// CHECK: [[VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PTR_REF:%.+]],
 // For & reduction operation initial value of private variable is ones in all bits.
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[VAR_PRIV]])
 
-// CHECK: [[VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[VAR1_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_PTR_REF:%.+]],
 // For && reduction operation initial value of private variable is 1.0.
 // CHECK: call {{.*}} [[S_INT_TY_CONSTR:@.+]]([[S_INT_TY]]* [[VAR1_PRIV]])
 
-// CHECK: [[T_VAR1_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} {{[0-9]+}}
-// CHECK: [[T_VAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR1_PTR_REF]],
 // For min reduction operation initial value of private variable is largest repesentable value.
 // CHECK: store i{{[0-9]+}} 2147483647, i{{[0-9]+}}* [[T_VAR1_PRIV]],
 
@@ -277,16 +263,16 @@ int main() {
 
 // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
 
-// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 0
+// CHECK: [[T_VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 0
 // CHECK: [[BITCAST:%.+]] = bitcast i{{[0-9]+}}* [[T_VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR_PRIV_REF]],
-// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 1
+// CHECK: [[VAR_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 1
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR_PRIV_REF]],
-// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 2
+// CHECK: [[VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 2
 // CHECK: [[BITCAST:%.+]] = bitcast [[S_INT_TY]]* [[VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[VAR1_PRIV_REF]],
-// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i32 0, i32 3
+// CHECK: [[T_VAR1_PRIV_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST]], i64 0, i64 3
 // CHECK: [[BITCAST:%.+]] = bitcast i{{[0-9]+}}* [[T_VAR1_PRIV]] to i8*
 // CHECK: store i8* [[BITCAST]], i8** [[T_VAR1_PRIV_REF]],
 
@@ -384,7 +370,6 @@ int main() {
 // CHECK: [[RED_DONE]]
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_INT_TY_DESTR]]([[S_INT_TY]]*
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 
 // void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
@@ -395,38 +380,38 @@ int main() {
 // }
 // CHECK: define internal void [[REDUCTION_FUNC]](i8*, i8*)
 // t_var_lhs = (i{{[0-9]+}}*)lhs[0];
-// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR_RHS_REF]],
 // CHECK: [[T_VAR_RHS:%.+]] = bitcast i8* [[T_VAR_RHS_VOID]] to i{{[0-9]+}}*
 // t_var_rhs = (i{{[0-9]+}}*)rhs[0];
-// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i32 0, i32 0
+// CHECK: [[T_VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS:%.+]], i64 0, i64 0
 // CHECK: [[T_VAR_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR_LHS_REF]],
 // CHECK: [[T_VAR_LHS:%.+]] = bitcast i8* [[T_VAR_LHS_VOID]] to i{{[0-9]+}}*
 
 // var_lhs = (S<i{{[0-9]+}}>*)lhs[1];
-// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 1
+// CHECK: [[VAR_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 1
 // CHECK: [[VAR_RHS_VOID:%.+]] = load i8*, i8** [[VAR_RHS_REF]],
 // CHECK: [[VAR_RHS:%.+]] = bitcast i8* [[VAR_RHS_VOID]] to [[S_INT_TY]]*
 // var_rhs = (S<i{{[0-9]+}}>*)rhs[1];
-// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 1
+// CHECK: [[VAR_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 1
 // CHECK: [[VAR_LHS_VOID:%.+]] = load i8*, i8** [[VAR_LHS_REF]],
 // CHECK: [[VAR_LHS:%.+]] = bitcast i8* [[VAR_LHS_VOID]] to [[S_INT_TY]]*
 
 // var1_lhs = (S<i{{[0-9]+}}>*)lhs[2];
-// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 2
+// CHECK: [[VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 2
 // CHECK: [[VAR1_RHS_VOID:%.+]] = load i8*, i8** [[VAR1_RHS_REF]],
 // CHECK: [[VAR1_RHS:%.+]] = bitcast i8* [[VAR1_RHS_VOID]] to [[S_INT_TY]]*
 // var1_rhs = (S<i{{[0-9]+}}>*)rhs[2];
-// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 2
+// CHECK: [[VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 2
 // CHECK: [[VAR1_LHS_VOID:%.+]] = load i8*, i8** [[VAR1_LHS_REF]],
 // CHECK: [[VAR1_LHS:%.+]] = bitcast i8* [[VAR1_LHS_VOID]] to [[S_INT_TY]]*
 
 // t_var1_lhs = (i{{[0-9]+}}*)lhs[3];
-// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_RHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_RHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_RHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_RHS_REF]],
 // CHECK: [[T_VAR1_RHS:%.+]] = bitcast i8* [[T_VAR1_RHS_VOID]] to i{{[0-9]+}}*
 // t_var1_rhs = (i{{[0-9]+}}*)rhs[3];
-// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i32 0, i32 3
+// CHECK: [[T_VAR1_LHS_REF:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[RED_LIST_LHS]], i64 0, i64 3
 // CHECK: [[T_VAR1_LHS_VOID:%.+]] = load i8*, i8** [[T_VAR1_LHS_REF]],
 // CHECK: [[T_VAR1_LHS:%.+]] = bitcast i8* [[T_VAR1_LHS_VOID]] to i{{[0-9]+}}*
 
diff --git a/test/OpenMP/sections_reduction_messages.cpp b/test/OpenMP/sections_reduction_messages.cpp
index a0f56d5a1e2a..79473d4e5d28 100644
--- a/test/OpenMP/sections_reduction_messages.cpp
+++ b/test/OpenMP/sections_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -ferror-limit 150 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -ferror-limit 150 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 150 -o - %s
 
 void foo() {
 }
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -121,7 +127,7 @@ T tmain(T argc) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp sections reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   {
     foo();
   }
@@ -141,32 +147,32 @@ T tmain(T argc) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp sections reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp sections reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp sections reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp sections reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp sections reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp sections reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   {
     foo();
   }
@@ -181,7 +187,7 @@ T tmain(T argc) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp sections reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -211,7 +217,7 @@ T tmain(T argc) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp sections reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -301,7 +307,7 @@ int main(int argc, char **argv) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp sections reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   {
     foo();
   }
@@ -321,32 +327,32 @@ int main(int argc, char **argv) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp sections reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp sections reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp sections reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp sections reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp sections reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp sections reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   {
     foo();
   }
@@ -361,7 +367,7 @@ int main(int argc, char **argv) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp sections reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -396,7 +402,7 @@ int main(int argc, char **argv) {
     foo();
   }
 #pragma omp parallel
-#pragma omp sections reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp sections reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   {
     foo();
   }
@@ -416,6 +422,11 @@ int main(int argc, char **argv) {
   {
     foo();
   }
+  static int m;
+#pragma omp sections reduction(+ : m) // OK
+  {
+    foo();
+  }
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/simd_aligned_messages.cpp b/test/OpenMP/simd_aligned_messages.cpp
index 408cc2e360b5..6be7529ad511 100644
--- a/test/OpenMP/simd_aligned_messages.cpp
+++ b/test/OpenMP/simd_aligned_messages.cpp
@@ -50,7 +50,7 @@ template<int L, class T, class N> T test_template(T* arr, N num) {
   T sum = (T)0;
   T ind2 = - num * L;
   // Negative number is passed as L.
-  // expected-error@+1 {{argument to 'aligned' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
   #pragma omp simd aligned(arr:L)
   for (i = 0; i < num; ++i) {
     T cur = arr[(int)ind2];
@@ -65,7 +65,7 @@ template<int L, class T, class N> T test_template(T* arr, N num) {
 
 template<int LEN> int test_warn() {
   int *ind2 = 0;
-  // expected-error@+1 {{argument to 'aligned' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
   #pragma omp simd aligned(ind2:LEN)
   for (int i = 0; i < 100; i++) {
     ind2 += LEN;
diff --git a/test/OpenMP/simd_ast_print.cpp b/test/OpenMP/simd_ast_print.cpp
index 069862b6dfbc..cabbe338db2a 100644
--- a/test/OpenMP/simd_ast_print.cpp
+++ b/test/OpenMP/simd_ast_print.cpp
@@ -12,10 +12,11 @@ template<class T, class N> T reduct(T* arr, N num) {
   N i;
   N ind;
   N myind;
+  N &ref = i;
   T sum = (T)0;
 // CHECK: T sum = (T)0;
-#pragma omp simd private(myind, g_ind), linear(ind), aligned(arr)
-// CHECK-NEXT: #pragma omp simd private(myind,g_ind) linear(ind) aligned(arr)
+#pragma omp simd private(myind, g_ind), linear(ind), aligned(arr), linear(uval(ref))
+// CHECK-NEXT: #pragma omp simd private(myind,g_ind) linear(ind) aligned(arr) linear(uval(ref))
   for (i = 0; i < num; ++i) {
     myind = ind;
     T cur = arr[myind];
@@ -32,11 +33,13 @@ template<class T> struct S {
     T res;
     T val;
     T lin = 0;
+    T &ref = res;
 // CHECK: T res;
 // CHECK: T val;
 // CHECK: T lin = 0;
-    #pragma omp simd private(val)  safelen(7) linear(lin : -5) lastprivate(res)
-// CHECK-NEXT: #pragma omp simd private(val) safelen(7) linear(lin: -5) lastprivate(res)
+// CHECK: T &ref = res;
+    #pragma omp simd private(val)  safelen(7) linear(lin : -5) lastprivate(res) simdlen(5) linear(ref(ref))
+// CHECK-NEXT: #pragma omp simd private(val) safelen(7) linear(lin: -5) lastprivate(res) simdlen(5) linear(ref(ref))
     for (T i = 7; i < m_a; ++i) {
       val = v[i-7] + m_a;
       res = val;
@@ -44,8 +47,8 @@ template<class T> struct S {
     }
     const T clen = 3;
 // CHECK: T clen = 3;
-    #pragma omp simd safelen(clen-1)
-// CHECK-NEXT: #pragma omp simd safelen(clen - 1)
+    #pragma omp simd safelen(clen-1) simdlen(clen-1)
+// CHECK-NEXT: #pragma omp simd safelen(clen - 1) simdlen(clen - 1)
     for(T i = clen+2; i < 20; ++i) {
 // CHECK-NEXT: for (T i = clen + 2; i < 20; ++i) {
       v[i] = v[v-clen] + 1;
@@ -62,7 +65,7 @@ template<class T> struct S {
 template<int LEN> struct S2 {
   static void func(int n, float *a, float *b, float *c) {
     int k1 = 0, k2 = 0;
-#pragma omp simd safelen(LEN) linear(k1,k2:LEN) aligned(a:LEN)
+#pragma omp simd safelen(LEN) linear(k1,k2:LEN) aligned(a:LEN) simdlen(LEN)
     for(int i = 0; i < n; i++) {
       c[i] = a[i] + b[i];
       c[k1] = a[k1] + b[k1];
@@ -77,7 +80,7 @@ template<int LEN> struct S2 {
 // CHECK: template <int LEN = 4> struct S2 {
 // CHECK-NEXT: static void func(int n, float *a, float *b, float *c)     {
 // CHECK-NEXT:   int k1 = 0, k2 = 0;
-// CHECK-NEXT: #pragma omp simd safelen(4) linear(k1,k2: 4) aligned(a: 4)
+// CHECK-NEXT: #pragma omp simd safelen(4) linear(k1,k2: 4) aligned(a: 4) simdlen(4)
 // CHECK-NEXT:   for (int i = 0; i < n; i++) {
 // CHECK-NEXT:     c[i] = a[i] + b[i];
 // CHECK-NEXT:     c[k1] = a[k1] + b[k1];
@@ -90,6 +93,7 @@ template<int LEN> struct S2 {
 int main (int argc, char **argv) {
   int b = argc, c, d, e, f, g;
   int k1=0,k2=0;
+  int &ref = b;
   static int *a;
 // CHECK: static int *a;
 #pragma omp simd
@@ -112,8 +116,8 @@ int main (int argc, char **argv) {
 // CHECK-NEXT: foo();
   const int CLEN = 4;
 // CHECK-NEXT: const int CLEN = 4;
-  #pragma omp simd aligned(a:CLEN) linear(a:CLEN) safelen(CLEN) collapse( 1 )
-// CHECK-NEXT: #pragma omp simd aligned(a: CLEN) linear(a: CLEN) safelen(CLEN) collapse(1)
+  #pragma omp simd aligned(a:CLEN) linear(a:CLEN) safelen(CLEN) collapse( 1 ) simdlen(CLEN) linear(val(ref): CLEN)
+// CHECK-NEXT: #pragma omp simd aligned(a: CLEN) linear(a: CLEN) safelen(CLEN) collapse(1) simdlen(CLEN) linear(val(ref): CLEN)
   for (int i = 0; i < 10; ++i)foo();
 // CHECK-NEXT: for (int i = 0; i < 10; ++i)
 // CHECK-NEXT: foo();
diff --git a/test/OpenMP/simd_codegen.cpp b/test/OpenMP/simd_codegen.cpp
index e67ad5d4c7e0..62028339f514 100644
--- a/test/OpenMP/simd_codegen.cpp
+++ b/test/OpenMP/simd_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // REQUIRES: x86-registered-target
 // expected-no-diagnostics
 #ifndef HEADER
@@ -481,6 +481,139 @@ void widened(float *a, float *b, float *c, float *d) {
 // CHECK: ret void
 }
 
+// CHECK-LABEL: define {{.*void}} @{{.*}}linear{{.*}}(float* {{.+}})
+void linear(float *a) {
+  // CHECK: [[VAL_ADDR:%.+]] = alloca i64,
+  // CHECK: [[K_ADDR:%.+]] = alloca i64*,
+  long long val = 0;
+  long long &k = val;
+
+  #pragma omp simd linear(k : 3)
+// CHECK: store i64* [[VAL_ADDR]], i64** [[K_ADDR]],
+// CHECK: store i32 0, i32* [[OMP_IV:%[^,]+]]
+// CHECK: [[K_REF:%.+]] = load i64*, i64** [[K_ADDR]],
+// CHECK: [[K0LOAD:%.+]] = load i64, i64* [[K_REF]]
+// CHECK-NEXT: store i64 [[K0LOAD]], i64* [[LIN0:%[^,]+]]
+
+// CHECK: [[IV:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID:[0-9]+]]
+// CHECK-NEXT: [[CMP2:%.+]] = icmp slt i32 [[IV]], 9
+// CHECK-NEXT: br i1 [[CMP2]], label %[[SIMPLE_LOOP_BODY:.+]], label %[[SIMPLE_LOOP_END:[^,]+]]
+  for (int i = 10; i > 1; i--) {
+// CHECK: [[SIMPLE_LOOP_BODY]]
+// Start of body: calculate i from IV:
+// CHECK: [[IV_0:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// FIXME: It is interesting, why the following "mul 1" was not constant folded?
+// CHECK-NEXT: [[IV_1:%.+]] = mul nsw i32 [[IV_0]], 1
+// CHECK-NEXT: [[LC_I_1:%.+]] = sub nsw i32 10, [[IV_1]]
+// CHECK-NEXT: store i32 [[LC_I_1]], i32* {{.+}}, !llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+//
+// CHECK-NEXT: [[LIN0_1:%.+]] = load i64, i64* [[LIN0]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[IV_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[LIN_MUL1:%.+]] = mul nsw i32 [[IV_2]], 3
+// CHECK-NEXT: [[LIN_EXT1:%.+]] = sext i32 [[LIN_MUL1]] to i64
+// CHECK-NEXT: [[LIN_ADD1:%.+]] = add nsw i64 [[LIN0_1]], [[LIN_EXT1]]
+// Update of the privatized version of linear variable!
+// CHECK-NEXT: store i64 [[LIN_ADD1]], i64* [[K_PRIVATIZED:%[^,]+]]
+    a[k]++;
+    k = k + 3;
+// CHECK: [[IV_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[ADD2_2:%.+]] = add nsw i32 [[IV_2]], 1
+// CHECK-NEXT: store i32 [[ADD2_2]], i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// br label {{.+}}, !llvm.loop ![[SIMPLE_LOOP_ID]]
+  }
+// CHECK: [[SIMPLE_LOOP_END]]
+//
+// Update linear vars after loop, as the loop was operating on a private version.
+// CHECK: [[K_REF:%.+]] = load i64*, i64** [[K_ADDR]],
+// CHECK: store i64* [[K_REF]], i64** [[K_PRIV_REF:%.+]],
+// CHECK: [[LIN0_2:%.+]] = load i64, i64* [[LIN0]]
+// CHECK-NEXT: [[LIN_ADD2:%.+]] = add nsw i64 [[LIN0_2]], 27
+// CHECK-NEXT: [[K_REF:%.+]] = load i64*, i64** [[K_PRIV_REF]],
+// CHECK-NEXT: store i64 [[LIN_ADD2]], i64* [[K_REF]]
+//
+
+  #pragma omp simd linear(val(k) : 3)
+// CHECK: store i32 0, i32* [[OMP_IV:%[^,]+]]
+// CHECK: [[K_REF:%.+]] = load i64*, i64** [[K_ADDR]],
+// CHECK: [[K0LOAD:%.+]] = load i64, i64* [[K_REF]]
+// CHECK-NEXT: store i64 [[K0LOAD]], i64* [[LIN0:%[^,]+]]
+
+// CHECK: [[IV:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID:[0-9]+]]
+// CHECK-NEXT: [[CMP2:%.+]] = icmp slt i32 [[IV]], 9
+// CHECK-NEXT: br i1 [[CMP2]], label %[[SIMPLE_LOOP_BODY:.+]], label %[[SIMPLE_LOOP_END:[^,]+]]
+  for (int i = 10; i > 1; i--) {
+// CHECK: [[SIMPLE_LOOP_BODY]]
+// Start of body: calculate i from IV:
+// CHECK: [[IV_0:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// FIXME: It is interesting, why the following "mul 1" was not constant folded?
+// CHECK-NEXT: [[IV_1:%.+]] = mul nsw i32 [[IV_0]], 1
+// CHECK-NEXT: [[LC_I_1:%.+]] = sub nsw i32 10, [[IV_1]]
+// CHECK-NEXT: store i32 [[LC_I_1]], i32* {{.+}}, !llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+//
+// CHECK-NEXT: [[LIN0_1:%.+]] = load i64, i64* [[LIN0]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[IV_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[LIN_MUL1:%.+]] = mul nsw i32 [[IV_2]], 3
+// CHECK-NEXT: [[LIN_EXT1:%.+]] = sext i32 [[LIN_MUL1]] to i64
+// CHECK-NEXT: [[LIN_ADD1:%.+]] = add nsw i64 [[LIN0_1]], [[LIN_EXT1]]
+// Update of the privatized version of linear variable!
+// CHECK-NEXT: store i64 [[LIN_ADD1]], i64* [[K_PRIVATIZED:%[^,]+]]
+    a[k]++;
+    k = k + 3;
+// CHECK: [[IV_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[ADD2_2:%.+]] = add nsw i32 [[IV_2]], 1
+// CHECK-NEXT: store i32 [[ADD2_2]], i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// br label {{.+}}, !llvm.loop ![[SIMPLE_LOOP_ID]]
+  }
+// CHECK: [[SIMPLE_LOOP_END]]
+//
+// Update linear vars after loop, as the loop was operating on a private version.
+// CHECK: [[K_REF:%.+]] = load i64*, i64** [[K_ADDR]],
+// CHECK: store i64* [[K_REF]], i64** [[K_PRIV_REF:%.+]],
+// CHECK: [[LIN0_2:%.+]] = load i64, i64* [[LIN0]]
+// CHECK-NEXT: [[LIN_ADD2:%.+]] = add nsw i64 [[LIN0_2]], 27
+// CHECK-NEXT: [[K_REF:%.+]] = load i64*, i64** [[K_PRIV_REF]],
+// CHECK-NEXT: store i64 [[LIN_ADD2]], i64* [[K_REF]]
+//
+  #pragma omp simd linear(uval(k) : 3)
+// CHECK: store i32 0, i32* [[OMP_IV:%[^,]+]]
+// CHECK: [[K0LOAD:%.+]] = load i64, i64* [[VAL_ADDR]]
+// CHECK-NEXT: store i64 [[K0LOAD]], i64* [[LIN0:%[^,]+]]
+
+// CHECK: [[IV:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID:[0-9]+]]
+// CHECK-NEXT: [[CMP2:%.+]] = icmp slt i32 [[IV]], 9
+// CHECK-NEXT: br i1 [[CMP2]], label %[[SIMPLE_LOOP_BODY:.+]], label %[[SIMPLE_LOOP_END:[^,]+]]
+  for (int i = 10; i > 1; i--) {
+// CHECK: [[SIMPLE_LOOP_BODY]]
+// Start of body: calculate i from IV:
+// CHECK: [[IV_0:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// FIXME: It is interesting, why the following "mul 1" was not constant folded?
+// CHECK-NEXT: [[IV_1:%.+]] = mul nsw i32 [[IV_0]], 1
+// CHECK-NEXT: [[LC_I_1:%.+]] = sub nsw i32 10, [[IV_1]]
+// CHECK-NEXT: store i32 [[LC_I_1]], i32* {{.+}}, !llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+//
+// CHECK-NEXT: [[LIN0_1:%.+]] = load i64, i64* [[LIN0]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[IV_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[LIN_MUL1:%.+]] = mul nsw i32 [[IV_2]], 3
+// CHECK-NEXT: [[LIN_EXT1:%.+]] = sext i32 [[LIN_MUL1]] to i64
+// CHECK-NEXT: [[LIN_ADD1:%.+]] = add nsw i64 [[LIN0_1]], [[LIN_EXT1]]
+// Update of the privatized version of linear variable!
+// CHECK-NEXT: store i64 [[LIN_ADD1]], i64* [[K_PRIVATIZED:%[^,]+]]
+    a[k]++;
+    k = k + 3;
+// CHECK: [[IV_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// CHECK-NEXT: [[ADD2_2:%.+]] = add nsw i32 [[IV_2]], 1
+// CHECK-NEXT: store i32 [[ADD2_2]], i32* [[OMP_IV]]{{.*}}!llvm.mem.parallel_loop_access ![[SIMPLE_LOOP_ID]]
+// br label {{.+}}, !llvm.loop ![[SIMPLE_LOOP_ID]]
+  }
+// CHECK: [[SIMPLE_LOOP_END]]
+//
+// Update linear vars after loop, as the loop was operating on a private version.
+// CHECK: [[LIN0_2:%.+]] = load i64, i64* [[LIN0]]
+// CHECK-NEXT: [[LIN_ADD2:%.+]] = add nsw i64 [[LIN0_2]], 27
+// CHECK-NEXT: store i64 [[LIN_ADD2]], i64* [[VAL_ADDR]]
+//
+}
+
 // TERM_DEBUG-LABEL: bar
 int bar() {return 0;};
 
diff --git a/test/OpenMP/simd_collapse_messages.cpp b/test/OpenMP/simd_collapse_messages.cpp
index 56ff201d112d..e34f0a15b20c 100644
--- a/test/OpenMP/simd_collapse_messages.cpp
+++ b/test/OpenMP/simd_collapse_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp simd collapse (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp simd collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp simd collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp simd' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp simd collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp simd', but found only 1}}
   // expected-error@+3 2 {{directive '#pragma omp simd' cannot contain more than one 'collapse' clause}}
-  // expected-error@+2 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp simd collapse (foobool(argc)), collapse (true), collapse (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp simd collapse (1)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp simd collapse (N) // expected-error {{argument to 'collapse' clause must be a positive integer value}}
+  #pragma omp simd collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp simd collapse (2) // expected-note {{as specified in 'collapse' clause}}
   foo(); // expected-error {{expected 2 for loops after '#pragma omp simd'}}
@@ -63,7 +63,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp simd' cannot contain more than one 'collapse' clause}}
-  // expected-error@+1 2 {{argument to 'collapse' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
   #pragma omp simd collapse (foobool(argc)), collapse (true), collapse (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp simd collapse (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/simd_lastprivate_messages.cpp b/test/OpenMP/simd_lastprivate_messages.cpp
index ca26db0d1447..7cc5ba81f78b 100644
--- a/test/OpenMP/simd_lastprivate_messages.cpp
+++ b/test/OpenMP/simd_lastprivate_messages.cpp
@@ -66,7 +66,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (int k = 0; k < argc; ++k)
     ++k;
@@ -118,7 +118,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp simd lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp simd lastprivate(j)
   for (int k = 0; k < argc; ++k)
     ++k;
 #pragma omp simd lastprivate(i)
@@ -134,7 +134,7 @@ int main(int argc, char **argv) {
   S5 g(5);
   S3 m;
   int i;
-  int &j = i;                // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
   for (i = 0; i < argc; ++i)
     foo();
@@ -210,7 +210,11 @@ int main(int argc, char **argv) {
   for (i = 0; i < argc; ++i)
     foo();
 #pragma omp parallel
-#pragma omp simd lastprivate(j) // expected-error {{arguments of OpenMP clause 'lastprivate' cannot be of reference type}}
+#pragma omp simd lastprivate(j)
+  for (i = 0; i < argc; ++i)
+    foo();
+  static int t;
+#pragma omp simd lastprivate(t) // OK
   for (i = 0; i < argc; ++i)
     foo();
   return 0;
diff --git a/test/OpenMP/simd_linear_messages.cpp b/test/OpenMP/simd_linear_messages.cpp
index 6fac26290b4e..792f78aa3db6 100644
--- a/test/OpenMP/simd_linear_messages.cpp
+++ b/test/OpenMP/simd_linear_messages.cpp
@@ -102,16 +102,26 @@ template<class I, class C> int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp simd linear // expected-error {{expected '(' after 'linear'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (val // expected-error {{use of undeclared identifier 'val'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (uval( // expected-error {{expected expression}} expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (ref() // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (foo() // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear () // expected-error {{expected expression}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int k = 0; k < argc; ++k) ++k;
-  #pragma omp simd linear (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp simd linear (val argc // expected-error {{use of undeclared identifier 'val'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (val(argc, // expected-error {{expected expression}} expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
   for (int k = 0; k < argc; ++k) ++k;
@@ -121,15 +131,15 @@ template<class I, class C> int foomain(I argc, C **argv) {
   for (int k = 0; k < argc; ++k) ++k;
   // expected-error@+2 {{linear variable with incomplete type 'S1'}}
   // expected-error@+1 {{const-qualified variable cannot be linear}}
-  #pragma omp simd linear (a, b:B::ib)
+  #pragma omp simd linear (val(a, b):B::ib)
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear (argv[1]) // expected-error {{expected variable name}}
   for (int k = 0; k < argc; ++k) ++k;
-  #pragma omp simd linear(e, g)
+  #pragma omp simd linear(ref(e, g)) // expected-error 2 {{variable of non-reference type 'int' can be used only with 'val' modifier, but used with 'ref'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear(h) // expected-error {{threadprivate or thread local variable cannot be linear}}
   for (int k = 0; k < argc; ++k) ++k;
-  #pragma omp simd linear(i)
+  #pragma omp simd linear(uval(i)) // expected-error {{variable of non-reference type 'int' can be used only with 'val' modifier, but used with 'uval'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp parallel
   {
@@ -138,7 +148,9 @@ template<class I, class C> int foomain(I argc, C **argv) {
     #pragma omp simd linear(v:i)
     for (int k = 0; k < argc; ++k) { i = k; v += i; }
   }
-  #pragma omp simd linear(j) // expected-error {{arguments of OpenMP clause 'linear' cannot be of reference type}}
+  #pragma omp simd linear(ref(j))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(uval(j))
   for (int k = 0; k < argc; ++k) ++k;
   int v = 0;
   #pragma omp simd linear(v:j)
@@ -156,6 +168,20 @@ namespace C {
 using A::x;
 }
 
+void linear_modifiers(int argc) {
+  int &f = argc;
+  #pragma omp simd linear(f)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(val(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(uval(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(ref(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(foo(f)) // expected-error {{expected one of 'ref', val' or 'uval' modifiers}}
+  for (int k = 0; k < argc; ++k) ++k;
+}
+
 int f;
 int main(int argc, char **argv) {
   double darr[100];
@@ -167,7 +193,7 @@ int main(int argc, char **argv) {
   S4 e(4); // expected-note {{'e' defined here}}
   S5 g(5); // expected-note {{'g' defined here}}
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp simd linear(f) linear(f) // expected-error {{linear variable cannot be linear}} expected-note {{defined as linear}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear // expected-error {{expected '(' after 'linear'}}
@@ -176,6 +202,12 @@ int main(int argc, char **argv) {
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear () // expected-error {{expected expression}}
   for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (val // expected-error {{use of undeclared identifier 'val'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (ref()) // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear (foo()) // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -194,24 +226,24 @@ int main(int argc, char **argv) {
   for (int k = 0; k < argc; ++k) ++k;
   // expected-error@+2 {{argument of a linear clause should be of integral or pointer type, not 'S4'}}
   // expected-error@+1 {{argument of a linear clause should be of integral or pointer type, not 'S5'}}
-  #pragma omp simd linear(e, g)
+  #pragma omp simd linear(val(e, g))
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear(h, C::x) // expected-error 2 {{threadprivate or thread local variable cannot be linear}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp parallel
   {
     int i;
-    #pragma omp simd linear(i)
+    #pragma omp simd linear(val(i))
     for (int k = 0; k < argc; ++k) ++k;
-    #pragma omp simd linear(i : 4)
+    #pragma omp simd linear(uval(i) : 4) // expected-error {{variable of non-reference type 'int' can be used only with 'val' modifier, but used with 'uval'}}
     for (int k = 0; k < argc; ++k) { ++k; i += 4; }
   }
-  #pragma omp simd linear(j) // expected-error {{arguments of OpenMP clause 'linear' cannot be of reference type 'int &'}}
+  #pragma omp simd linear(ref(j))
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd linear(i)
   for (int k = 0; k < argc; ++k) ++k;
 
-  foomain<int,char>(argc,argv);
+  foomain<int,char>(argc,argv); // expected-note {{in instantiation of function template specialization 'foomain<int, char>' requested here}}
   return 0;
 }
 
diff --git a/test/OpenMP/simd_loop_messages.cpp b/test/OpenMP/simd_loop_messages.cpp
index bd7dd5bf52d1..96b4cafc6bd8 100644
--- a/test/OpenMP/simd_loop_messages.cpp
+++ b/test/OpenMP/simd_loop_messages.cpp
@@ -1,6 +1,7 @@
 // RUN: %clang_cc1 -fsyntax-only -fopenmp -x c++ -std=c++11 -fexceptions -fcxx-exceptions -verify %s
 
 static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
 #pragma omp threadprivate(sii)
 static int globalii;
 
@@ -45,32 +46,32 @@ int test_iteration_spaces() {
   for (double fi = 0; fi < 10.0; fi++) {
     c[(int)fi] = a[(int)fi] + b[(int)fi];
   }
-  // expected-error@+2 {{variable must be of integer or random access iterator type}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (int &ref = ii; ref < 10; ref++) {
   }
-  // expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (int i; i < 10; i++)
     c[i] = a[i];
 
-  // expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (int i = 0, j = 0; i < 10; ++i)
     c[i] = a[i];
 
-  // expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (;ii < 10; ++ii)
     c[ii] = a[ii];
 
   // expected-warning@+3 {{expression result unused}}
-  // expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (ii + 1;ii < 10; ++ii)
     c[ii] = a[ii];
 
-  // expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (c[ii] = 0;ii < 10; ++ii)
     c[ii] = a[ii];
@@ -252,6 +253,7 @@ int test_iteration_spaces() {
 
   #pragma omp parallel
   {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp simd' directive may not be threadprivate or thread local, predetermined as linear}}
     #pragma omp simd
     for (sii = 0; sii < 10; sii+=1)
       c[sii] = a[sii];
@@ -364,7 +366,7 @@ int test_with_random_access_iterator() {
   #pragma omp simd
   for (GoodIter I = begin; I < end; ++I)
     ++I;
-  // expected-error@+2 {{variable must be of integer or random access iterator type}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (GoodIter &I = begin; I < end; ++I)
     ++I;
@@ -393,7 +395,7 @@ int test_with_random_access_iterator() {
   #pragma omp simd
   for (begin = GoodIter(1,2); begin < end; ++begin)
     ++begin;
-  // expected-error@+2 {{initialization clause of OpenMP for loop must be of the form 'var = init' or 'T var = init'}}
+  // expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
   #pragma omp simd
   for (++begin; begin < end; ++begin)
     ++begin;
diff --git a/test/OpenMP/simd_metadata.c b/test/OpenMP/simd_metadata.c
index e7e35dd54d74..6ed660747dcb 100644
--- a/test/OpenMP/simd_metadata.c
+++ b/test/OpenMP/simd_metadata.c
@@ -1,6 +1,9 @@
 // RUN: %clang_cc1 -fopenmp -triple x86_64-unknown-unknown -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=X86
 // RUN: %clang_cc1 -fopenmp -triple x86_64-unknown-unknown -target-feature +avx -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=X86-AVX
 // RUN: %clang_cc1 -fopenmp -triple x86_64-unknown-unknown -target-feature +avx512f -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=X86-AVX512
+// RUN: %clang_cc1 -fopenmp -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=X86
+// RUN: %clang_cc1 -fopenmp -triple i386-unknown-unknown -target-feature +avx -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=X86-AVX
+// RUN: %clang_cc1 -fopenmp -triple i386-unknown-unknown -target-feature +avx512f -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=X86-AVX512
 // RUN: %clang_cc1 -fopenmp -triple powerpc64-unknown-unknown -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=PPC
 // RUN: %clang_cc1 -fopenmp -triple powerpc64-unknown-unknown -target-abi elfv1-qpx -emit-llvm %s -o - | FileCheck %s -check-prefix=CHECK -check-prefix=PPC-QPX
 
@@ -36,8 +39,69 @@ void h1(float *c, float *a, double b[], int size)
   for (int i = 0; i < size; ++i) {
     c[i] = a[i] * a[i] + b[i] * b[t];
     ++t;
+  }
+// do not emit parallel_loop_access metadata due to usage of safelen clause.
+// CHECK-NOT: store float {{.+}}, float* {{.+}}, align {{.+}}, !llvm.mem.parallel_loop_access {{![0-9]+}}
+#pragma omp simd safelen(16) linear(t) aligned(c:32) aligned(a,b) simdlen(8)
+// CHECK:         [[C_PTRINT:%.+]] = ptrtoint
+// CHECK-NEXT:    [[C_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[C_PTRINT]], 31
+// CHECK-NEXT:    [[C_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[C_MASKEDPTR]], 0
+// CHECK-NEXT:    call void @llvm.assume(i1 [[C_MASKCOND]])
+// CHECK:         [[A_PTRINT:%.+]] = ptrtoint
+
+// X86-NEXT:     [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15
+// X86-AVX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 31
+// X86-AVX512-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 63
+// PPC-NEXT:     [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15
+// PPC-QPX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15
+
+// CHECK-NEXT:    [[A_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[A_MASKEDPTR]], 0
+// CHECK-NEXT:    call void @llvm.assume(i1 [[A_MASKCOND]])
+// CHECK:         [[B_PTRINT:%.+]] = ptrtoint
+
+// X86-NEXT:      [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15
+// X86-AVX-NEXT:  [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31
+// X86-AVX512-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 63
+// PPC-NEXT:      [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15
+// PPC-QPX-NEXT:  [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31
+
+// CHECK-NEXT:    [[B_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[B_MASKEDPTR]], 0
+// CHECK-NEXT:    call void @llvm.assume(i1 [[B_MASKCOND]])
+  for (int i = 0; i < size; ++i) {
+    c[i] = a[i] * a[i] + b[i] * b[t];
+    ++t;
+  }
 // do not emit parallel_loop_access metadata due to usage of safelen clause.
 // CHECK-NOT: store float {{.+}}, float* {{.+}}, align {{.+}}, !llvm.mem.parallel_loop_access {{![0-9]+}}
+#pragma omp simd linear(t) aligned(c:32) aligned(a,b) simdlen(8)
+// CHECK:         [[C_PTRINT:%.+]] = ptrtoint
+// CHECK-NEXT:    [[C_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[C_PTRINT]], 31
+// CHECK-NEXT:    [[C_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[C_MASKEDPTR]], 0
+// CHECK-NEXT:    call void @llvm.assume(i1 [[C_MASKCOND]])
+// CHECK:         [[A_PTRINT:%.+]] = ptrtoint
+
+// X86-NEXT:     [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15
+// X86-AVX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 31
+// X86-AVX512-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 63
+// PPC-NEXT:     [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15
+// PPC-QPX-NEXT: [[A_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[A_PTRINT]], 15
+
+// CHECK-NEXT:    [[A_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[A_MASKEDPTR]], 0
+// CHECK-NEXT:    call void @llvm.assume(i1 [[A_MASKCOND]])
+// CHECK:         [[B_PTRINT:%.+]] = ptrtoint
+
+// X86-NEXT:      [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15
+// X86-AVX-NEXT:  [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31
+// X86-AVX512-NEXT: [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 63
+// PPC-NEXT:      [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 15
+// PPC-QPX-NEXT:  [[B_MASKEDPTR:%.+]] = and i{{[0-9]+}} [[B_PTRINT]], 31
+
+// CHECK-NEXT:    [[B_MASKCOND:%.+]] = icmp eq i{{[0-9]+}} [[B_MASKEDPTR]], 0
+// CHECK-NEXT:    call void @llvm.assume(i1 [[B_MASKCOND]])
+  for (int i = 0; i < size; ++i) {
+    c[i] = a[i] * a[i] + b[i] * b[t];
+    ++t;
+// CHECK: store float {{.+}}, float* {{.+}}, align {{.+}}, !llvm.mem.parallel_loop_access {{![0-9]+}}
   }
 }
 
@@ -70,6 +134,9 @@ void h3(float *c, float *a, float *b, int size)
 // CHECK: [[LOOP_H1_HEADER:![0-9]+]] = distinct !{[[LOOP_H1_HEADER]], [[LOOP_WIDTH_16:![0-9]+]], [[LOOP_VEC_ENABLE:![0-9]+]]}
 // CHECK: [[LOOP_WIDTH_16]] = !{!"llvm.loop.vectorize.width", i32 16}
 // CHECK: [[LOOP_VEC_ENABLE]] = !{!"llvm.loop.vectorize.enable", i1 true}
+// CHECK: [[LOOP_H1_HEADER:![0-9]+]] = distinct !{[[LOOP_H1_HEADER]], [[LOOP_WIDTH_8:![0-9]+]], [[LOOP_VEC_ENABLE]]}
+// CHECK: [[LOOP_WIDTH_8]] = !{!"llvm.loop.vectorize.width", i32 8}
+// CHECK: [[LOOP_H1_HEADER:![0-9]+]] = distinct !{[[LOOP_H1_HEADER]], [[LOOP_WIDTH_8]], [[LOOP_VEC_ENABLE]]}
 //
 // Metadata for h2:
 // CHECK: [[LOOP_H2_HEADER]] = distinct !{[[LOOP_H2_HEADER]], [[LOOP_VEC_ENABLE]]}
diff --git a/test/OpenMP/simd_misc_messages.c b/test/OpenMP/simd_misc_messages.c
index 0b4dd7f3b533..1e1548291947 100644
--- a/test/OpenMP/simd_misc_messages.c
+++ b/test/OpenMP/simd_misc_messages.c
@@ -152,20 +152,117 @@ void test_safelen() {
 #pragma omp simd safelen(foo())
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp simd safelen(-5)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp simd safelen(0)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
 #pragma omp simd safelen(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
 }
 
+void test_simdlen() {
+  int i;
+// expected-error@+1 {{expected '('}}
+#pragma omp simd simdlen
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}}
+#pragma omp simd simdlen()
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(,
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expected expression}}  expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(, )
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp simd' are ignored}}
+// expected-error@+1 {{expected '('}}
+#pragma omp simd simdlen 4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(4
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(4,
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(4, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp simd simdlen(4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(4 4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(4, , 4)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp simd simdlen(4)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}}
+#pragma omp simd simdlen(4, 8)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp simd simdlen(2.5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp simd simdlen(foo())
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp simd simdlen(-5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp simd simdlen(0)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+#pragma omp simd simdlen(5 - 5)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_safelen_simdlen() {
+  int i;
+// expected-error@+1 {{the value of 'simdlen' parameter must be less than or equal to the value of the 'safelen' parameter}}
+#pragma omp simd simdlen(6) safelen(5)
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{the value of 'simdlen' parameter must be less than or equal to the value of the 'safelen' parameter}}
+#pragma omp simd safelen(5) simdlen(6)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
 void test_collapse() {
   int i;
 // expected-error@+1 {{expected '('}}
@@ -241,15 +338,15 @@ void test_collapse() {
 #pragma omp simd collapse(foo())
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp simd collapse(-5)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp simd collapse(0)
   for (i = 0; i < 16; ++i)
     ;
-// expected-error@+1 {{argument to 'collapse' clause must be a positive integer value}}
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
 #pragma omp simd collapse(5 - 5)
   for (i = 0; i < 16; ++i)
     ;
@@ -259,12 +356,18 @@ void test_collapse() {
   for (i = 0; i < 16; ++i)
     // expected-note@+1 {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp for' directive into a parallel or another task region?}}
     for (int j = 0; j < 16; ++j)
-// expected-error@+3 {{reduction variable must be shared}}
-// expected-error@+2 {{private variable cannot be reduction}}
+// expected-error@+2 2 {{reduction variable must be shared}}
 // expected-error@+1 {{OpenMP constructs may not be nested inside a simd region}}
 #pragma omp for reduction(+ : i, j)
       for (int k = 0; k < 16; ++k)
         i += j;
+#pragma omp parallel
+#pragma omp for
+  for (i = 0; i < 16; ++i)
+    for (int j = 0; j < 16; ++j)
+#pragma omp simd reduction(+ : i, j)
+      for (int k = 0; k < 16; ++k)
+        i += j;
 }
 
 void test_linear() {
@@ -678,3 +781,17 @@ void test_loop_messages() {
   }
 }
 
+void linear_modifiers(int argc) {
+  int f;
+  #pragma omp simd linear(f)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(val(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(uval(f)) // expected-error {{expected 'val' modifier}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(ref(f)) // expected-error {{expected 'val' modifier}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp simd linear(foo(f)) // expected-error {{expected 'val' modifier}}
+  for (int k = 0; k < argc; ++k) ++k;
+}
+
diff --git a/test/OpenMP/simd_private_messages.cpp b/test/OpenMP/simd_private_messages.cpp
index adef373b6a4e..3442d182ed02 100644
--- a/test/OpenMP/simd_private_messages.cpp
+++ b/test/OpenMP/simd_private_messages.cpp
@@ -42,7 +42,7 @@ template<class I, class C> int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp simd private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd private ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -78,7 +78,7 @@ template<class I, class C> int foomain(I argc, C **argv) {
   }
   #pragma omp parallel shared(i)
   #pragma omp parallel private(i)
-  #pragma omp simd private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+  #pragma omp simd private(j)
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd private(i)
   for (int k = 0; k < argc; ++k) ++k;
@@ -97,7 +97,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp simd private // expected-error {{expected '(' after 'private'}}
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd private ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -132,7 +132,7 @@ int main(int argc, char **argv) {
   }
   #pragma omp parallel shared(i)
   #pragma omp parallel private(i)
-  #pragma omp simd private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+  #pragma omp simd private(j)
   for (int k = 0; k < argc; ++k) ++k;
   #pragma omp simd private(i)
   for (int k = 0; k < argc; ++k) ++k;
diff --git a/test/OpenMP/simd_reduction_messages.cpp b/test/OpenMP/simd_reduction_messages.cpp
index aeb2b23bb0ca..e082921cf383 100644
--- a/test/OpenMP/simd_reduction_messages.cpp
+++ b/test/OpenMP/simd_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 %s
 
 void foo() {
 }
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -102,7 +108,7 @@ T tmain(T argc) {
 #pragma omp simd reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{invalid operands to binary expression ('float' and 'float')}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp simd reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp simd reduction(foo : argc) //expected-error {{incorrect reduction identifier, expected one of '+', '-', '*', '&', '|', '^', '&&', '||', 'min' or 'max'}}
@@ -114,22 +120,22 @@ T tmain(T argc) {
 #pragma omp simd reduction(^ : T) // expected-error {{'T' does not refer to a value}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp simd reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp simd reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp simd reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp simd reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp simd reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp simd reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp simd reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
@@ -138,7 +144,7 @@ T tmain(T argc) {
 #pragma omp simd reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp simd reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp simd reduction(+ : h, k) // expected-error {{threadprivate or thread local variable cannot be reduction}}
@@ -160,7 +166,7 @@ T tmain(T argc) {
 #pragma omp simd reduction(+ : p), reduction(+ : p) // expected-error 3 {{variable can appear only once in OpenMP 'reduction' clause}} expected-note 3 {{previously referenced here}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp simd reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
@@ -227,7 +233,7 @@ int main(int argc, char **argv) {
 #pragma omp simd reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp simd reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp simd reduction(~ : argc) // expected-error {{expected unqualified-id}}
@@ -239,22 +245,22 @@ int main(int argc, char **argv) {
 #pragma omp simd reduction(^ : S1) // expected-error {{'S1' does not refer to a value}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp simd reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp simd reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp simd reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp simd reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp simd reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp simd reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp simd reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
@@ -263,7 +269,7 @@ int main(int argc, char **argv) {
 #pragma omp simd reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp simd reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp simd reduction(& : e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{invalid operands to binary expression ('S4' and 'S4')}} expected-error {{calling a private constructor of class 'S5'}} expected-error {{invalid operands to binary expression ('S5' and 'S5')}}
@@ -285,7 +291,7 @@ int main(int argc, char **argv) {
 #pragma omp simd reduction(+ : p), reduction(+ : p) // expected-error {{variable can appear only once in OpenMP 'reduction' clause}} expected-note {{previously referenced here}}
   for (int i = 0; i < 10; ++i)
     foo();
-#pragma omp simd reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp simd reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   for (int i = 0; i < 10; ++i)
     foo();
 #pragma omp parallel shared(i)
diff --git a/test/OpenMP/simd_safelen_messages.cpp b/test/OpenMP/simd_safelen_messages.cpp
index b7300c337532..aa31b7da9b7f 100644
--- a/test/OpenMP/simd_safelen_messages.cpp
+++ b/test/OpenMP/simd_safelen_messages.cpp
@@ -22,7 +22,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
   #pragma omp simd safelen (argc 
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  // expected-error@+1 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
   #pragma omp simd safelen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp simd safelen (1)) // expected-warning {{extra tokens at the end of '#pragma omp simd' are ignored}}
@@ -30,7 +30,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   #pragma omp simd safelen ((ST > 0) ? 1 + ST : 2)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   // expected-error@+3 2 {{directive '#pragma omp simd' cannot contain more than one 'safelen' clause}}
-  // expected-error@+2 2 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+2 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
   // expected-error@+1 2 {{expression is not an integral constant expression}}
   #pragma omp simd safelen (foobool(argc)), safelen (true), safelen (-5)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
@@ -41,7 +41,7 @@ T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   #pragma omp simd safelen (4)
   for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
-  #pragma omp simd safelen (N) // expected-error {{argument to 'safelen' clause must be a positive integer value}}
+  #pragma omp simd safelen (N) // expected-error {{argument to 'safelen' clause must be a strictly positive integer value}}
   for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
   return argc;
 }
@@ -61,7 +61,7 @@ int main(int argc, char **argv) {
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   // expected-error@+3 {{expression is not an integral constant expression}}
   // expected-error@+2 2 {{directive '#pragma omp simd' cannot contain more than one 'safelen' clause}}
-  // expected-error@+1 2 {{argument to 'safelen' clause must be a positive integer value}}
+  // expected-error@+1 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
   #pragma omp simd safelen (foobool(argc)), safelen (true), safelen (-5) 
   for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
   #pragma omp simd safelen (S1) // expected-error {{'S1' does not refer to a value}}
diff --git a/test/OpenMP/simd_simdlen_messages.cpp b/test/OpenMP/simd_simdlen_messages.cpp
new file mode 100644
index 000000000000..91656f87b583
--- /dev/null
+++ b/test/OpenMP/simd_simdlen_messages.cpp
@@ -0,0 +1,79 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp simd simdlen (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp simd simdlen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen (1)) // expected-warning {{extra tokens at the end of '#pragma omp simd' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen ((ST > 0) ? 1 + ST : 2)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 2 {{directive '#pragma omp simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+2 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp simd simdlen (foobool(argc)), simdlen (true), simdlen (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen (4)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp simd simdlen (N) // expected-error {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp simd simdlen () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp simd simdlen (4 // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp simd simdlen (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp simd' are ignored}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp simd simdlen (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+1 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp simd simdlen (foobool(argc)), simdlen (true), simdlen (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp simd simdlen (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp simd' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp simd simdlen(simdlen(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 12, 4>' requested here}}
+  return tmain<int, char, 12, 4>(argc, argv);
+}
+
diff --git a/test/OpenMP/single_codegen.cpp b/test/OpenMP/single_codegen.cpp
index 327f6fafa834..61a93a5a1b73 100644
--- a/test/OpenMP/single_codegen.cpp
+++ b/test/OpenMP/single_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -fnoopenmp-use-tls -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -fnoopenmp-use-tls -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp -fnoopenmp-use-tls -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fnoopenmp-use-tls -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fnoopenmp-use-tls -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // RUN: %clang_cc1 -verify -fopenmp -fnoopenmp-use-tls -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -check-prefix=ARRAY %s
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
@@ -38,7 +38,7 @@ int main() {
   // CHECK-DAG: [[C_ADDR:%.+]] = alloca [[TEST_CLASS_TY]]
   char a;
   char a2[2];
-  TestClass c;
+  TestClass &c = tc;
 
 // CHECK:       [[GTID:%.+]] = call i32 @__kmpc_global_thread_num([[IDENT_T_TY]]* [[DEFAULT_LOC:@.+]])
 // CHECK-DAG:   [[DID_IT:%.+]] = alloca i32,
@@ -81,8 +81,7 @@ int main() {
 // CHECK:       [[A_PTR_REF:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[COPY_LIST]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK:       store i8* [[A_ADDR]], i8** [[A_PTR_REF]],
 // CHECK:       [[C_PTR_REF:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[COPY_LIST]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK:       [[C_PTR_REF_VOID_PTR:%.+]] = bitcast [[TEST_CLASS_TY]]* [[C_ADDR]] to i8*
-// CHECK:       store i8* [[C_PTR_REF_VOID_PTR]], i8** [[C_PTR_REF]],
+// CHECK:       store i8* {{.+}}, i8** [[C_PTR_REF]],
 // CHECK:       [[TC_PTR_REF:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[COPY_LIST]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
 // CHECK:       [[TC_THREADPRIVATE_ADDR_VOID_PTR:%.+]] = call{{.*}} i8* @__kmpc_threadprivate_cached
 // CHECK:       [[TC_THREADPRIVATE_ADDR:%.+]] = bitcast i8* [[TC_THREADPRIVATE_ADDR_VOID_PTR]] to [[TEST_CLASS_TY]]*
diff --git a/test/OpenMP/single_firstprivate_codegen.cpp b/test/OpenMP/single_firstprivate_codegen.cpp
index 01eaea318119..cc72addb5f45 100644
--- a/test/OpenMP/single_firstprivate_codegen.cpp
+++ b/test/OpenMP/single_firstprivate_codegen.cpp
@@ -30,7 +30,6 @@ struct S {
 // CHECK-DAG: [[S_FLOAT_TY:%.+]] = type { float }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
 // CHECK-DAG: [[ST_TY:%.+]] = type { i{{[0-9]+}}, i{{[0-9]+}} }
-// CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i{{[0-9]+}}*, [2 x i{{[0-9]+}}]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
 
 template <typename T>
 T tmain() {
@@ -64,36 +63,49 @@ S<float> var(3);
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
 // CHECK: ([[S_FLOAT_TY]]*)* [[S_FLOAT_TY_DESTR:@[^ ]+]] {{[^,]+}}, {{.+}}([[S_FLOAT_TY]]* [[TEST]]
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global i{{[0-9]+}} 1212,
   // LAMBDA-LABEL: @main
   // LAMBDA: call void [[OUTER_LAMBDA:@.+]](
   [&]() {
 // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+// LAMBDA: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp single firstprivate(g)
+#pragma omp single firstprivate(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[ARG:%.+]])
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // LAMBDA: %{{.+}} = alloca [[CAP_MAIN_TY:%.+]], 
     // LAMBDA: call i32 @__kmpc_single(
     // LAMBDA: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]]
     // LAMBDA: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* {{.*}}
+    // LAMBDA: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]   
     g = 1;
+    sivar = 17;
     // LAMBDA: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 17, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store i{{[0-9]+}}* [[G_PRIVATE_ADDR]], i{{[0-9]+}}** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call void @__kmpc_end_single(
-    // LAMBDA: call i32 @__kmpc_cancel_barrier(
+    // LAMBDA: call void @__kmpc_barrier(
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 31;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[G_PTR_REF]]
       // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 31, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -104,40 +116,55 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
 // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+// BLOCKS: call void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp single firstprivate(g)
+#pragma omp single firstprivate(g, sivar)
    {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}}, i32* dereferenceable(4) [[SIVAR_REF:%.+]])
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: [[SIVAR1_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
+    // BLOCKS: store i{{[0-9]+}}* [[SIVAR_REF]], i{{[0-9]+}}** %{{.+}},
+    // BLOCKS: [[SIVAR1_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %{{.+}},
     // BLOCKS: call i32 @__kmpc_single(
     // BLOCKS: [[G_VAL:%.+]] = load volatile i{{[0-9]+}}, i{{[0-9]+}}* [[G]]
     // BLOCKS: store i{{[0-9]+}} [[G_VAL]], i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
+    // BLOCKS: [[SIVAR1_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR1_REF]],
+    // BLOCKS: store i{{[0-9]+}} [[SIVAR1_VAL]], i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     g = 1;
+    sivar = 37;
     // BLOCKS: store i{{[0-9]+}} 1, i{{[0-9]+}}* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 37, i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: i{{[0-9]+}}* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR1_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
     // BLOCKS: call void @__kmpc_end_single(
-    // BLOCKS: call i32 @__kmpc_cancel_barrier(
+    // BLOCKS: call void @__kmpc_barrier(
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 31;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store i{{[0-9]+}} 2, i{{[0-9]+}}*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 31, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
   }();
   return 0;
 #else
-#pragma omp single firstprivate(t_var, vec, s_arr, var) nowait
+#pragma omp single firstprivate(t_var, vec, s_arr, var, sivar) nowait
   {
     {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 41;
     }
   }
   return tmain<int>();
@@ -151,6 +178,7 @@ int main() {
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 
 // CHECK: call i32 @__kmpc_single(
 // firstprivate t_var(t_var)
@@ -178,6 +206,10 @@ int main() {
 // CHECK: call {{.*}} [[S_FLOAT_TY_COPY_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]], [[S_FLOAT_TY]]* {{.*}} [[VAR]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
 
+// firstprivate isvar
+// CHEC: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR]],
+// CHEC: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIV]],
+
 // ~(firstprivate var), ~(firstprivate s_arr)
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call {{.*}} [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -192,36 +224,36 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 4, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, i32*, [2 x i32]*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call {{.*}} [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}}, i32* dereferenceable(4) %{{.+}}, [2 x i32]* dereferenceable(8) %{{.+}}, [2 x [[S_INT_TY]]]* dereferenceable(8) %{{.+}}, [[S_INT_TY]]* dereferenceable(4) %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_INT_TY]],
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_ADDR:%.+]],
 
+// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** %
+// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** %
+// CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** %
+// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** %
+
 // CHECK: [[GTID_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[GTID_ADDR_ADDR]]
 // CHECK: [[GTID:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[GTID_REF]]
 // CHECK: call i32 @__kmpc_single(
+
 // firstprivate t_var(t_var)
-// CHECK: [[T_VAR_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 0
-// CHECK: [[T_VAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[T_VAR_PTR_REF]],
 // CHECK: [[T_VAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[T_VAR_REF]],
 // CHECK: store i{{[0-9]+}} [[T_VAR_VAL]], i{{[0-9]+}}* [[T_VAR_PRIV]],
 
 // firstprivate vec(vec)
-// CHECK: [[VEC_PTR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[VEC_REF:%.+]] = load [2 x i{{[0-9]+}}]*, [2 x i{{[0-9]+}}]** [[VEC_PTR_REF:%.+]],
 // CHECK: [[VEC_DEST:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_PRIV]] to i8*
 // CHECK: [[VEC_SRC:%.+]] = bitcast [2 x i{{[0-9]+}}]* [[VEC_REF]] to i8*
 // CHECK: call void @llvm.memcpy.{{.+}}(i8* [[VEC_DEST]], i8* [[VEC_SRC]],
 
 // firstprivate s_arr(s_arr)
-// CHECK: [[S_ARR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 2
-// CHECK: [[S_ARR:%.+]] = load [2 x [[S_INT_TY]]]*, [2 x [[S_INT_TY]]]** [[S_ARR_REF]],
 // CHECK: [[S_ARR_PRIV_BEGIN:%.+]] = getelementptr inbounds [2 x [[S_INT_TY]]], [2 x [[S_INT_TY]]]* [[S_ARR_PRIV]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_PRIV_END:%.+]] = getelementptr [[S_INT_TY]], [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], i{{[0-9]+}} 2
 // CHECK: [[IS_EMPTY:%.+]] = icmp eq [[S_INT_TY]]* [[S_ARR_PRIV_BEGIN]], [[S_ARR_PRIV_END]]
@@ -233,8 +265,6 @@ int main() {
 // CHECK: br i1 {{.+}}, label %{{.+}}, label %[[S_ARR_BODY]]
 
 // firstprivate var(var)
-// CHECK: [[VAR_REF_PTR:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_INT_TY]]*, [[S_INT_TY]]** [[VAR_REF_PTR]],
 // CHECK: call {{.*}} [[ST_TY_DEFAULT_CONSTR]]([[ST_TY]]* [[ST_TY_TEMP:%.+]])
 // CHECK: call {{.*}} [[S_INT_TY_COPY_CONSTR]]([[S_INT_TY]]* [[VAR_PRIV]], [[S_INT_TY]]* {{.*}} [[VAR_REF]], [[ST_TY]]* [[ST_TY_TEMP]])
 // CHECK: call {{.*}} [[ST_TY_DESTR]]([[ST_TY]]* [[ST_TY_TEMP]])
@@ -245,8 +275,7 @@ int main() {
 
 // CHECK: call void @__kmpc_end_single(
 
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[SINGLE_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
-// CHECK: call i32 @__kmpc_cancel_barrier(%{{.+}}* [[IMPLICIT_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
+// CHECK: call void @__kmpc_barrier(%{{.+}}* [[SINGLE_BARRIER_LOC]], i{{[0-9]+}} [[GTID]])
 // CHECK: ret void
 #endif
 
diff --git a/test/OpenMP/single_firstprivate_messages.cpp b/test/OpenMP/single_firstprivate_messages.cpp
index 6f21a42208f4..32de9fdc8806 100644
--- a/test/OpenMP/single_firstprivate_messages.cpp
+++ b/test/OpenMP/single_firstprivate_messages.cpp
@@ -65,7 +65,7 @@ int foomain(int argc, char **argv) {
   I e(4);
   C g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp single firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   foo();
@@ -109,13 +109,13 @@ int foomain(int argc, char **argv) {
   {
     int v = 0;
     int i;                         // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp single' directive into a parallel or another task region?}}
-#pragma omp single firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp single firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     foo();
     v += i;
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp single firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp single firstprivate(j)
   foo();
 #pragma omp parallel
 #pragma omp single firstprivate(i)
@@ -148,7 +148,7 @@ int main(int argc, char **argv) {
   S3 m;
   S6 n(2);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp parallel
 #pragma omp single firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   foo();
@@ -220,7 +220,7 @@ int main(int argc, char **argv) {
 #pragma omp single firstprivate(xa) // OK: may be firstprivate
   foo();
 #pragma omp parallel
-#pragma omp single firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp single firstprivate(j)
   foo();
 #pragma omp parallel
 #pragma omp single firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
@@ -232,7 +232,7 @@ int main(int argc, char **argv) {
   {
     int v = 0;
     int i;                         // expected-note {{variable with automatic storage duration is predetermined as private; perhaps you forget to enclose 'omp single' directive into a parallel or another task region?}}
-#pragma omp single firstprivate(i) // expected-error {{private variable cannot be firstprivate}}
+#pragma omp single firstprivate(i) // expected-error {{firstprivate variable must be shared}}
     foo();
     v += i;
   }
@@ -242,6 +242,9 @@ int main(int argc, char **argv) {
 #pragma omp parallel reduction(+ : i) // expected-note {{defined as reduction}}
 #pragma omp single firstprivate(i)    // expected-error {{firstprivate variable must be shared}}
   foo();
+  static int t;
+#pragma omp single firstprivate(t)    // OK
+  foo();
 
   return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
 }
diff --git a/test/OpenMP/single_private_codegen.cpp b/test/OpenMP/single_private_codegen.cpp
index e228b1b14887..c98c5ea1dd67 100644
--- a/test/OpenMP/single_private_codegen.cpp
+++ b/test/OpenMP/single_private_codegen.cpp
@@ -20,9 +20,7 @@ struct S {
 volatile double g;
 
 // CHECK: [[S_FLOAT_TY:%.+]] = type { float }
-// CHECK: [[CAP_MAIN_TY:%.+]] = type { i8 }
 // CHECK: [[S_INT_TY:%.+]] = type { i{{[0-9]+}} }
-// CHECK: [[CAP_TMAIN_TY:%.+]] = type { i8 }
 template <typename T>
 T tmain() {
   S<T> test;
@@ -40,34 +38,43 @@ T tmain() {
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global double
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{.*}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* %{{.+}})
+  // LAMBDA: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp single private(g)
+#pragma omp single private(g, sivar)
   {
-    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // LAMBDA: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = alloca double,
-    // LAMBDA: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     g = 1;
+    sivar = 101;
     // LAMBDA: call {{.*}}i32 @__kmpc_single(
     // LAMBDA: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
+    // LAMBDA: store i{{[0-9]+}} 101, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // LAMBDA: [[G_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: store double* [[G_PRIVATE_ADDR]], double** [[G_PRIVATE_ADDR_REF]]
+    // LAMBDA: [[SIVAR_PRIVATE_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: store i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]], i{{[0-9]+}}** [[SIVAR_PRIVATE_ADDR_REF]]
     // LAMBDA: call{{.*}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG]])
     // LAMBDA: call {{.*}}void @__kmpc_end_single(
     [&]() {
       // LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
       // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
       g = 2;
+      sivar = 211;
       // LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
       // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
       // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
       // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
+      // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+      // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+      // LAMBDA: store i{{[0-9]+}} 211, i{{[0-9]+}}* [[SIVAR_REF]]
     }();
   }
   }();
@@ -78,27 +85,36 @@ int main() {
   // BLOCKS: call {{.*}}void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 1, {{.+}}* [[OMP_REGION:@.+]] to {{.+}}, i8* {{.+}})
+  // BLOCKS: call {{.*}}void {{.+}} @__kmpc_fork_call({{.+}}, i32 0, {{.+}}* [[OMP_REGION:@.+]] to {{.+}})
 #pragma omp parallel
-#pragma omp single private(g)
+#pragma omp single private(g, sivar)
   {
-    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* %{{.+}}, i32* %{{.+}}, %{{.+}}* [[ARG:%.+]])
+    // BLOCKS: define{{.*}} internal{{.*}} void [[OMP_REGION]](i32* noalias %{{.+}}, i32* noalias %{{.+}})
     // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = alloca double,
-    // BLOCKS: store %{{.+}}* [[ARG]], %{{.+}}** [[ARG_REF:%.+]],
+    // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = alloca i{{[0-9]+}},
     g = 1;
+    sivar = 101;
     // BLOCKS: call {{.*}}i32 @__kmpc_single(
     // BLOCKS: store double 1.0{{.+}}, double* [[G_PRIVATE_ADDR]],
+    // BLOCKS: store i{{[0-9]+}} 101, i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]],
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: double* [[G_PRIVATE_ADDR]]
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call {{.*}}void {{%.+}}(i8
     // BLOCKS: call {{.*}}void @__kmpc_end_single(
     ^{
       // BLOCKS: define {{.+}} void {{@.+}}(i8*
       g = 2;
+      sivar = 203;
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
       // BLOCKS: store double 2.0{{.+}}, double*
       // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+      // BLOCKS: store i{{[0-9]+}} 203, i{{[0-9]+}}*
+      // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
       // BLOCKS: ret
     }();
   }
@@ -111,10 +127,11 @@ int main() {
   S<float> s_arr[] = {1, 2};
   S<float> var(3);
 #pragma omp parallel
-#pragma omp single private(t_var, vec, s_arr, s_arr, var, var)
+#pragma omp single private(t_var, vec, s_arr, s_arr, var, var, sivar)
   {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 303;
   }
   return tmain<int>();
 #endif
@@ -123,28 +140,30 @@ int main() {
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: [[TEST:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR:@.+]]([[S_FLOAT_TY]]* [[TEST]])
-// CHECK: %{{.+}} = bitcast [[CAP_MAIN_TY]]*
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_MAIN_TY]]*)* [[MAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[MAIN_MICROTASK:@.+]] to void
 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
 // CHECK: call void [[S_FLOAT_TY_DESTR:@.+]]([[S_FLOAT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_MAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[MAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_FLOAT_TY]]],
 // CHECK-NOT: alloca [2 x [[S_FLOAT_TY]]],
 // CHECK: [[VAR_PRIV:%.+]] = alloca [[S_FLOAT_TY]],
 // CHECK-NOT: alloca [[S_FLOAT_TY]],
+// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: store i{{[0-9]+}}* [[GTID_ADDR]], i{{[0-9]+}}** [[GTID_ADDR_REF:%.+]]
 // CHECK: call i32 @__kmpc_single(
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
+// CHECK-NOT: [[SIVAR_PRIV]]
 // CHECK: {{.+}}:
 // CHECK: [[S_ARR_PRIV_ITEM:%.+]] = phi [[S_FLOAT_TY]]*
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[S_ARR_PRIV_ITEM]])
 // CHECK-NOT: [[T_VAR_PRIV]]
 // CHECK-NOT: [[VEC_PRIV]]
+// CHECK-NOT: [[SIVAR_PRIV]]
 // CHECK: call {{.*}} [[S_FLOAT_TY_DEF_CONSTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]* [[VAR_PRIV]])
 // CHECK-DAG: call void [[S_FLOAT_TY_DESTR]]([[S_FLOAT_TY]]*
@@ -154,11 +173,11 @@ int main() {
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
 // CHECK: call {{.*}} [[S_INT_TY_DEF_CONSTR:@.+]]([[S_INT_TY]]* [[TEST]])
-// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 1, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*, [[CAP_TMAIN_TY]]*)* [[TMAIN_MICROTASK:@.+]] to void
+// CHECK: call void (%{{.+}}*, i{{[0-9]+}}, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)*, ...) @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{[0-9]+}} 0, void (i{{[0-9]+}}*, i{{[0-9]+}}*, ...)* bitcast (void (i{{[0-9]+}}*, i{{[0-9]+}}*)* [[TMAIN_MICROTASK:@.+]] to void
 // CHECK: call void [[S_INT_TY_DESTR:@.+]]([[S_INT_TY]]*
 // CHECK: ret
 //
-// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* [[GTID_ADDR:%.+]], i{{[0-9]+}}* %{{.+}}, [[CAP_TMAIN_TY]]* %{{.+}})
+// CHECK: define internal void [[TMAIN_MICROTASK]](i{{[0-9]+}}* noalias [[GTID_ADDR:%.+]], i{{[0-9]+}}* noalias %{{.+}})
 // CHECK: [[T_VAR_PRIV:%.+]] = alloca i{{[0-9]+}},
 // CHECK: [[VEC_PRIV:%.+]] = alloca [2 x i{{[0-9]+}}],
 // CHECK: [[S_ARR_PRIV:%.+]] = alloca [2 x [[S_INT_TY]]],
diff --git a/test/OpenMP/single_private_messages.cpp b/test/OpenMP/single_private_messages.cpp
index 0e04e0f5aff1..a24cf47cd2a2 100644
--- a/test/OpenMP/single_private_messages.cpp
+++ b/test/OpenMP/single_private_messages.cpp
@@ -47,7 +47,7 @@ int foomain(I argc, C **argv) {
   I e(4);
   I g(5);
   int i;
-  int &j = i;                // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp single private // expected-error {{expected '(' after 'private'}}
   foo();
 #pragma omp single private( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -84,7 +84,7 @@ int foomain(I argc, C **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp single private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp single private(j)
   foo();
 #pragma omp single private(i)
   foo();
@@ -103,7 +103,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;                // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp single private // expected-error {{expected '(' after 'private'}}
   foo();
 #pragma omp single private( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
@@ -138,10 +138,13 @@ int main(int argc, char **argv) {
   }
 #pragma omp parallel shared(i)
 #pragma omp parallel private(i)
-#pragma omp single private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type}}
+#pragma omp single private(j)
   foo();
 #pragma omp single private(i)
   foo();
+  static int m;
+#pragma omp single private(m) // OK
+  foo();
 
   return 0;
 }
diff --git a/test/OpenMP/target_ast_print.cpp b/test/OpenMP/target_ast_print.cpp
index 6955cf31112e..acf032a22120 100644
--- a/test/OpenMP/target_ast_print.cpp
+++ b/test/OpenMP/target_ast_print.cpp
@@ -10,39 +10,77 @@ void foo() {}
 
 template <typename T, int C>
 T tmain(T argc, T *argv) {
+  T i, j, a[20];
 #pragma omp target
   foo();
-#pragma omp target if (argc > 0)
+#pragma omp target if (target:argc > 0)
   foo();
 #pragma omp target if (C)
   foo();
+#pragma omp target map(i)
+  foo();
+#pragma omp target map(a[0:10], i)
+  foo();
+#pragma omp target map(to: i) map(from: j)
+  foo();
+#pragma omp target map(always,alloc: i)
+  foo();
   return 0;
 }
 
 // CHECK: template <typename T = int, int C = 5> int tmain(int argc, int *argv) {
+// CHECK-NEXT: int i, j, a[20]
 // CHECK-NEXT: #pragma omp target
 // CHECK-NEXT: foo();
-// CHECK-NEXT: #pragma omp target if(argc > 0)
+// CHECK-NEXT: #pragma omp target if(target: argc > 0)
 // CHECK-NEXT: foo()
 // CHECK-NEXT: #pragma omp target if(5)
 // CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(tofrom: i)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(tofrom: a[0:10],i)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(to: i) map(from: j)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(always,alloc: i)
+// CHECK-NEXT: foo()
 // CHECK: template <typename T = char, int C = 1> char tmain(char argc, char *argv) {
+// CHECK-NEXT: char i, j, a[20]
 // CHECK-NEXT: #pragma omp target
 // CHECK-NEXT: foo();
-// CHECK-NEXT: #pragma omp target if(argc > 0)
+// CHECK-NEXT: #pragma omp target if(target: argc > 0)
 // CHECK-NEXT: foo()
 // CHECK-NEXT: #pragma omp target if(1)
 // CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(tofrom: i)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(tofrom: a[0:10],i)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(to: i) map(from: j)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(always,alloc: i)
+// CHECK-NEXT: foo()
 // CHECK: template <typename T, int C> T tmain(T argc, T *argv) {
+// CHECK-NEXT: T i, j, a[20]
 // CHECK-NEXT: #pragma omp target
 // CHECK-NEXT: foo();
-// CHECK-NEXT: #pragma omp target if(argc > 0)
+// CHECK-NEXT: #pragma omp target if(target: argc > 0)
 // CHECK-NEXT: foo()
 // CHECK-NEXT: #pragma omp target if(C)
 // CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(tofrom: i)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(tofrom: a[0:10],i)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(to: i) map(from: j)
+// CHECK-NEXT: foo()
+// CHECK-NEXT: #pragma omp target map(always,alloc: i)
+// CHECK-NEXT: foo()
 
 // CHECK-LABEL: int main(int argc, char **argv) {
 int main (int argc, char **argv) {
+  int i, j, a[20];
+// CHECK-NEXT: int i, j, a[20]
 #pragma omp target
 // CHECK-NEXT: #pragma omp target
   foo();
@@ -51,6 +89,32 @@ int main (int argc, char **argv) {
 // CHECK-NEXT: #pragma omp target if(argc > 0)
   foo();
 // CHECK-NEXT: foo();
+
+#pragma omp target map(i) if(argc>0)
+// CHECK-NEXT: #pragma omp target map(tofrom: i) if(argc > 0)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target map(i)
+// CHECK-NEXT: #pragma omp target map(tofrom: i)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target map(a[0:10], i)
+// CHECK-NEXT: #pragma omp target map(tofrom: a[0:10],i)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target map(to: i) map(from: j)
+// CHECK-NEXT: #pragma omp target map(to: i) map(from: j)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target map(always,alloc: i)
+// CHECK-NEXT: #pragma omp target map(always,alloc: i)
+  foo();
+// CHECK-NEXT: foo();
+
   return tmain<int, 5>(argc, &argc) + tmain<char, 1>(argv[0][0], argv[0]);
 }
 
diff --git a/test/OpenMP/target_codegen.cpp b/test/OpenMP/target_codegen.cpp
new file mode 100644
index 000000000000..bcefa2419dd2
--- /dev/null
+++ b/test/OpenMP/target_codegen.cpp
@@ -0,0 +1,652 @@
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
+// expected-no-diagnostics
+#ifndef HEADER
+#define HEADER
+
+// CHECK-DAG: [[TT:%.+]] = type { i64, i8 }
+// CHECK-DAG: [[S1:%.+]] = type { double }
+
+// We have 8 target regions, but only 7 that actually will generate offloading
+// code, only 6 will have mapped arguments, and only 4 have all-constant map
+// sizes.
+
+// CHECK-DAG: [[SIZET2:@.+]] = private unnamed_addr constant [1 x i{{32|64}}] [i[[SZ:32|64]] 2]
+// CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [1 x i32] [i32 128]
+// CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i[[SZ]]] [i[[SZ]] 4, i[[SZ]] 2]
+// CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i32] [i32 128, i32 128]
+// CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [9 x i32] [i32 128, i32 3, i32 128, i32 3, i32 3, i32 128, i32 128, i32 3, i32 3]
+// CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i[[SZ]]] [i[[SZ]] 4, i[[SZ]] 2, i[[SZ]] 40]
+// CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i32] [i32 128, i32 128, i32 3]
+// CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [4 x i[[SZ]]] [i[[SZ]] 4, i[[SZ]] 2, i[[SZ]] 1, i[[SZ]] 40]
+// CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [4 x i32] [i32 128, i32 128, i32 128, i32 3]
+// CHECK-DAG: [[MAPT7:@.+]] = private unnamed_addr constant [5 x i32] [i32 3, i32 128, i32 128, i32 128, i32 3]
+// CHECK-DAG: @{{.*}} = private constant i8 0
+// CHECK-DAG: @{{.*}} = private constant i8 0
+// CHECK-DAG: @{{.*}} = private constant i8 0
+// CHECK-DAG: @{{.*}} = private constant i8 0
+// CHECK-DAG: @{{.*}} = private constant i8 0
+// CHECK-DAG: @{{.*}} = private constant i8 0
+// CHECK-DAG: @{{.*}} = private constant i8 0
+
+template<typename tx, typename ty>
+struct TT{
+  tx X;
+  ty Y;
+};
+
+// CHECK: define {{.*}}[[FOO:@.+]](
+int foo(int n) {
+  int a = 0;
+  short aa = 0;
+  float b[10];
+  float bn[n];
+  double c[5][10];
+  double cn[5][n];
+  TT<long long, char> d;
+
+  // CHECK:       [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 0, i8** null, i8** null, i[[SZ]]* null, i32* null)
+  // CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+  // CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+  // CHECK-NEXT:  [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+  // CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]]
+  // CHECK:       [[FAIL]]
+  // CHECK:       call void [[HVT0:@.+]]()
+  // CHECK-NEXT:  br label %[[END]]
+  // CHECK:       [[END]]
+  #pragma omp target
+  {
+  }
+
+  // CHECK:       store i32 0, i32* [[RHV:%.+]], align 4
+  // CHECK:       store i32 -1, i32* [[RHV]], align 4
+  // CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+  // CHECK-NEXT:  [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+  // CHECK:       call void [[HVT1:@.+]](i[[SZ]] {{[^,]+}})
+  #pragma omp target if(0)
+  {
+    a += 1;
+  }
+
+  // CHECK-DAG:   [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 1, i8** [[BP:%[^,]+]], i8** [[P:%[^,]+]], i[[SZ]]* getelementptr inbounds ([1 x i[[SZ]]], [1 x i[[SZ]]]* [[SIZET2]], i32 0, i32 0), i32* getelementptr inbounds ([1 x i32], [1 x i32]* [[MAPT2]], i32 0, i32 0))
+  // CHECK-DAG:   [[BP]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[BPR:%[^,]+]], i32 0, i32 0
+  // CHECK-DAG:   [[P]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[PR:%[^,]+]], i32 0, i32 0
+  // CHECK-DAG:   [[BPADDR0:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[BPR]], i32 0, i32 [[IDX0:[0-9]+]]
+  // CHECK-DAG:   [[PADDR0:%.+]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[PR]], i32 0, i32 [[IDX0]]
+  // CHECK-DAG:   store i8* [[BP0:%[^,]+]], i8** [[BPADDR0]]
+  // CHECK-DAG:   store i8* [[P0:%[^,]+]], i8** [[PADDR0]]
+  // CHECK-DAG:   [[BP0]] = inttoptr i[[SZ]] %{{.+}} to i8*
+  // CHECK-DAG:   [[P0]] = inttoptr i[[SZ]] %{{.+}} to i8*
+
+  // CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+  // CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+  // CHECK-NEXT:  [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+  // CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]]
+  // CHECK:       [[FAIL]]
+  // CHECK:       call void [[HVT2:@.+]](i[[SZ]] {{[^,]+}})
+  // CHECK-NEXT:  br label %[[END]]
+  // CHECK:       [[END]]
+  #pragma omp target if(1)
+  {
+    aa += 1;
+  }
+
+  // CHECK:       [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 10
+  // CHECK:       br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
+  // CHECK:       [[IFTHEN]]
+  // CHECK-DAG:   [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 2, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* getelementptr inbounds ([2 x i[[SZ]]], [2 x i[[SZ]]]* [[SIZET3]], i32 0, i32 0), i32* getelementptr inbounds ([2 x i32], [2 x i32]* [[MAPT3]], i32 0, i32 0))
+  // CHECK-DAG:   [[BPR]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP:%[^,]+]], i32 0, i32 0
+  // CHECK-DAG:   [[PR]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P:%[^,]+]], i32 0, i32 0
+
+  // CHECK-DAG:   [[BPADDR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 0
+  // CHECK-DAG:   [[PADDR0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 0
+  // CHECK-DAG:   store i8* [[BP0:%[^,]+]], i8** [[BPADDR0]]
+  // CHECK-DAG:   store i8* [[P0:%[^,]+]], i8** [[PADDR0]]
+  // CHECK-DAG:   [[BP0]] = inttoptr i[[SZ]] %{{.+}} to i8*
+  // CHECK-DAG:   [[P0]] = inttoptr i[[SZ]] %{{.+}} to i8*
+
+  // CHECK-DAG:   [[BPADDR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BP]], i32 0, i32 1
+  // CHECK-DAG:   [[PADDR1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[P]], i32 0, i32 1
+  // CHECK-DAG:   store i8* [[BP1:%[^,]+]], i8** [[BPADDR1]]
+  // CHECK-DAG:   store i8* [[P1:%[^,]+]], i8** [[PADDR1]]
+  // CHECK-DAG:   [[BP1]] = inttoptr i[[SZ]] %{{.+}} to i8*
+  // CHECK-DAG:   [[P1]] = inttoptr i[[SZ]] %{{.+}} to i8*
+  // CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+  // CHECK-NEXT:  br label %[[IFEND:.+]]
+
+  // CHECK:       [[IFELSE]]
+  // CHECK:       store i32 -1, i32* [[RHV]], align 4
+  // CHECK-NEXT:  br label %[[IFEND:.+]]
+
+  // CHECK:       [[IFEND]]
+  // CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+  // CHECK:       [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+  // CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]]
+  // CHECK:       [[FAIL]]
+  // CHECK:       call void [[HVT3:@.+]]({{[^,]+}}, {{[^,]+}})
+  // CHECK-NEXT:  br label %[[END]]
+  // CHECK:       [[END]]
+  #pragma omp target if(n>10)
+  {
+    a += 1;
+    aa += 1;
+  }
+
+  // We capture 3 VLA sizes in this target region
+  // CHECK-64:       [[A_VAL:%.+]] = load i32, i32* %{{.+}},
+  // CHECK-64:       [[A_ADDR:%.+]] = bitcast i[[SZ]]* [[A_CADDR:%.+]] to i32*
+  // CHECK-64:       store i32 [[A_VAL]], i32* [[A_ADDR]],
+  // CHECK-64:       [[A_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CADDR]],
+
+  // CHECK-32:       [[A_VAL:%.+]] = load i32, i32* %{{.+}},
+  // CHECK-32:       store i32 [[A_VAL]], i32* [[A_CADDR:%.+]],
+  // CHECK-32:       [[A_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[A_CADDR]],
+
+  // CHECK:       [[BNSIZE:%.+]] = mul nuw i[[SZ]] [[VLA0:%.+]], 4
+  // CHECK:       [[CNELEMSIZE2:%.+]] = mul nuw i[[SZ]] 5, [[VLA1:%.+]]
+  // CHECK:       [[CNSIZE:%.+]] = mul nuw i[[SZ]] [[CNELEMSIZE2]], 8
+
+  // CHECK:       [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 20
+  // CHECK:       br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]]
+  // CHECK:       [[TRY]]
+  // CHECK-DAG:   [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 9, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* [[SR:%[^,]+]], i32* getelementptr inbounds ([9 x i32], [9 x i32]* [[MAPT4]], i32 0, i32 0))
+  // CHECK-DAG:   [[BPR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP:%[^,]+]], i32 0, i32 0
+  // CHECK-DAG:   [[PR]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P:%[^,]+]], i32 0, i32 0
+  // CHECK-DAG:   [[SR]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S:%[^,]+]], i32 0, i32 0
+
+  // CHECK-DAG:   [[SADDR0:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX0:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX0]]
+  // CHECK-DAG:   [[PADDR0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX0]]
+  // CHECK-DAG:   [[SADDR1:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX1:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX1]]
+  // CHECK-DAG:   [[PADDR1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX1]]
+  // CHECK-DAG:   [[SADDR2:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX2:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX2]]
+  // CHECK-DAG:   [[PADDR2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX2]]
+  // CHECK-DAG:   [[SADDR3:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX3:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX3]]
+  // CHECK-DAG:   [[PADDR3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX3]]
+  // CHECK-DAG:   [[SADDR4:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX4:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX4]]
+  // CHECK-DAG:   [[PADDR4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX4]]
+  // CHECK-DAG:   [[SADDR5:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX5:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX5]]
+  // CHECK-DAG:   [[PADDR5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX5]]
+  // CHECK-DAG:   [[SADDR6:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX6:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX6]]
+  // CHECK-DAG:   [[PADDR6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX6]]
+  // CHECK-DAG:   [[SADDR7:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX7:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX7]]
+  // CHECK-DAG:   [[PADDR7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX7]]
+  // CHECK-DAG:   [[SADDR8:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[S]], i32 0, i32 [[IDX8:[0-9]+]]
+  // CHECK-DAG:   [[BPADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BP]], i32 0, i32 [[IDX8]]
+  // CHECK-DAG:   [[PADDR8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[P]], i32 0, i32 [[IDX8]]
+
+  // The names below are not necessarily consistent with the names used for the
+  // addresses above as some are repeated.
+  // CHECK-DAG:   [[BP0:%[^,]+]] = inttoptr i[[SZ]] [[VLA0]] to i8*
+  // CHECK-DAG:   [[P0:%[^,]+]] = inttoptr i[[SZ]] [[VLA0]] to i8*
+  // CHECK-DAG:   store i8* [[BP0]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P0]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP1:%[^,]+]] = inttoptr i[[SZ]] [[VLA1]] to i8*
+  // CHECK-DAG:   [[P1:%[^,]+]] = inttoptr i[[SZ]] [[VLA1]] to i8*
+  // CHECK-DAG:   store i8* [[BP1]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P1]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   store i8* inttoptr (i[[SZ]] 5 to i8*), i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* inttoptr (i[[SZ]] 5 to i8*), i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP3:%[^,]+]] = inttoptr i[[SZ]] [[A_CVAL]] to i8*
+  // CHECK-DAG:   [[P3:%[^,]+]] = inttoptr i[[SZ]] [[A_CVAL]] to i8*
+  // CHECK-DAG:   store i8* [[BP3]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P3]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] 4, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP4:%[^,]+]] = bitcast [10 x float]* %{{.+}} to i8*
+  // CHECK-DAG:   [[P4:%[^,]+]] = bitcast [10 x float]* %{{.+}} to i8*
+  // CHECK-DAG:   store i8* [[BP4]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P4]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] 40, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP5:%[^,]+]] = bitcast float* %{{.+}} to i8*
+  // CHECK-DAG:   [[P5:%[^,]+]] = bitcast float* %{{.+}} to i8*
+  // CHECK-DAG:   store i8* [[BP5]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P5]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] [[BNSIZE]], i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP6:%[^,]+]] = bitcast [5 x [10 x double]]* %{{.+}} to i8*
+  // CHECK-DAG:   [[P6:%[^,]+]] = bitcast [5 x [10 x double]]* %{{.+}} to i8*
+  // CHECK-DAG:   store i8* [[BP6]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P6]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] 400, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP7:%[^,]+]] = bitcast double* %{{.+}} to i8*
+  // CHECK-DAG:   [[P7:%[^,]+]] = bitcast double* %{{.+}} to i8*
+  // CHECK-DAG:   store i8* [[BP7]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P7]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] [[CNSIZE]], i[[SZ]]* {{%[^,]+}}
+
+  // CHECK-DAG:   [[BP8:%[^,]+]] = bitcast [[TT]]* %{{.+}} to i8*
+  // CHECK-DAG:   [[P8:%[^,]+]] = bitcast [[TT]]* %{{.+}} to i8*
+  // CHECK-DAG:   store i8* [[BP8]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i8* [[P8]], i8** {{%[^,]+}}
+  // CHECK-DAG:   store i[[SZ]] {{12|16}}, i[[SZ]]* {{%[^,]+}}
+
+  // CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+  // CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+  // CHECK-NEXT:  [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+  // CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]]
+
+  // CHECK:       [[FAIL]]
+  // CHECK:       call void [[HVT4:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
+  // CHECK-NEXT:  br label %[[END]]
+  // CHECK:       [[END]]
+  #pragma omp target if(n>20)
+  {
+    a += 1;
+    b[2] += 1.0;
+    bn[3] += 1.0;
+    c[1][2] += 1.0;
+    cn[1][3] += 1.0;
+    d.X += 1;
+    d.Y += 1;
+  }
+
+  return a;
+}
+
+// Check that the offloading functions are emitted and that the arguments are
+// correct and loaded correctly for the target regions in foo().
+
+// CHECK:       define internal void [[HVT0]]()
+
+// CHECK:       define internal void [[HVT1]](i[[SZ]] %{{.+}})
+// Create stack storage and store argument in there.
+// CHECK:       [[AA_ADDR:%.+]] = alloca i[[SZ]], align
+// CHECK:       store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align
+// CHECK-64:    [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i32*
+// CHECK-64:    load i32, i32* [[AA_CADDR]], align
+// CHECK-32:    load i32, i32* [[AA_ADDR]], align
+
+// CHECK:       define internal void [[HVT2]](i[[SZ]] %{{.+}})
+// Create stack storage and store argument in there.
+// CHECK:       [[AA_ADDR:%.+]] = alloca i[[SZ]], align
+// CHECK:       store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align
+// CHECK:       [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16*
+// CHECK:       load i16, i16* [[AA_CADDR]], align
+
+// CHECK:       define internal void [[HVT3]]
+// Create stack storage and store argument in there.
+// CHECK:       [[A_ADDR:%.+]] = alloca i[[SZ]], align
+// CHECK:       [[AA_ADDR:%.+]] = alloca i[[SZ]], align
+// CHECK-DAG:   store i[[SZ]] %{{.+}}, i[[SZ]]* [[A_ADDR]], align
+// CHECK-DAG:   store i[[SZ]] %{{.+}}, i[[SZ]]* [[AA_ADDR]], align
+// CHECK-64-DAG:[[A_CADDR:%.+]] = bitcast i[[SZ]]* [[A_ADDR]] to i32*
+// CHECK-DAG:   [[AA_CADDR:%.+]] = bitcast i[[SZ]]* [[AA_ADDR]] to i16*
+// CHECK-64-DAG:load i32, i32* [[A_CADDR]], align
+// CHECK-32-DAG:load i32, i32* [[A_ADDR]], align
+// CHECK-DAG:   load i16, i16* [[AA_CADDR]], align
+
+// CHECK:       define internal void [[HVT4]]
+// Create local storage for each capture.
+// CHECK:       [[LOCAL_A:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_B:%.+]] = alloca [10 x float]*
+// CHECK:       [[LOCAL_VLA1:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_BN:%.+]] = alloca float*
+// CHECK:       [[LOCAL_C:%.+]] = alloca [5 x [10 x double]]*
+// CHECK:       [[LOCAL_VLA2:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_VLA3:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_CN:%.+]] = alloca double*
+// CHECK:       [[LOCAL_D:%.+]] = alloca [[TT]]*
+// CHECK-DAG:   store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]]
+// CHECK-DAG:   store [10 x float]* [[ARG_B:%.+]], [10 x float]** [[LOCAL_B]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_VLA1:%.+]], i[[SZ]]* [[LOCAL_VLA1]]
+// CHECK-DAG:   store float* [[ARG_BN:%.+]], float** [[LOCAL_BN]]
+// CHECK-DAG:   store [5 x [10 x double]]* [[ARG_C:%.+]], [5 x [10 x double]]** [[LOCAL_C]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_VLA2:%.+]], i[[SZ]]* [[LOCAL_VLA2]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_VLA3:%.+]], i[[SZ]]* [[LOCAL_VLA3]]
+// CHECK-DAG:   store double* [[ARG_CN:%.+]], double** [[LOCAL_CN]]
+// CHECK-DAG:   store [[TT]]* [[ARG_D:%.+]], [[TT]]** [[LOCAL_D]]
+
+// CHECK-64-DAG:[[REF_A:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32*
+// CHECK-DAG:   [[REF_B:%.+]] = load [10 x float]*, [10 x float]** [[LOCAL_B]],
+// CHECK-DAG:   [[VAL_VLA1:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA1]],
+// CHECK-DAG:   [[REF_BN:%.+]] = load float*, float** [[LOCAL_BN]],
+// CHECK-DAG:   [[REF_C:%.+]] = load [5 x [10 x double]]*, [5 x [10 x double]]** [[LOCAL_C]],
+// CHECK-DAG:   [[VAL_VLA2:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA2]],
+// CHECK-DAG:   [[VAL_VLA3:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA3]],
+// CHECK-DAG:   [[REF_CN:%.+]] = load double*, double** [[LOCAL_CN]],
+// CHECK-DAG:   [[REF_D:%.+]] = load [[TT]]*, [[TT]]** [[LOCAL_D]],
+
+// Use captures.
+// CHECK-64-DAG:   load i32, i32* [[REF_A]]
+// CHECK-32-DAG:   load i32, i32* [[LOCAL_A]]
+// CHECK-DAG:   getelementptr inbounds [10 x float], [10 x float]* [[REF_B]], i[[SZ]] 0, i[[SZ]] 2
+// CHECK-DAG:   getelementptr inbounds float, float* [[REF_BN]], i[[SZ]] 3
+// CHECK-DAG:   getelementptr inbounds [5 x [10 x double]], [5 x [10 x double]]* [[REF_C]], i[[SZ]] 0, i[[SZ]] 1
+// CHECK-DAG:   getelementptr inbounds double, double* [[REF_CN]], i[[SZ]] %{{.+}}
+// CHECK-DAG:   getelementptr inbounds [[TT]], [[TT]]* [[REF_D]], i32 0, i32 0
+
+template<typename tx>
+tx ftemplate(int n) {
+  tx a = 0;
+  short aa = 0;
+  tx b[10];
+
+  #pragma omp target if(n>40)
+  {
+    a += 1;
+    aa += 1;
+    b[2] += 1;
+  }
+
+  return a;
+}
+
+static
+int fstatic(int n) {
+  int a = 0;
+  short aa = 0;
+  char aaa = 0;
+  int b[10];
+
+  #pragma omp target if(n>50)
+  {
+    a += 1;
+    aa += 1;
+    aaa += 1;
+    b[2] += 1;
+  }
+
+  return a;
+}
+
+struct S1 {
+  double a;
+
+  int r1(int n){
+    int b = n+1;
+    short int c[2][n];
+
+    #pragma omp target if(n>60)
+    {
+      this->a = (double)b + 1.5;
+      c[1][1] = ++a;
+    }
+
+    return c[1][1] + (int)b;
+  }
+};
+
+// CHECK: define {{.*}}@{{.*}}bar{{.*}}
+int bar(int n){
+  int a = 0;
+
+  // CHECK: call {{.*}}i32 [[FOO]](i32 {{.*}})
+  a += foo(n);
+
+  S1 S;
+  // CHECK: call {{.*}}i32 [[FS1:@.+]]([[S1]]* {{.*}}, i32 {{.*}})
+  a += S.r1(n);
+
+  // CHECK: call {{.*}}i32 [[FSTATIC:@.+]](i32 {{.*}})
+  a += fstatic(n);
+
+  // CHECK: call {{.*}}i32 [[FTEMPLATE:@.+]](i32 {{.*}})
+  a += ftemplate<int>(n);
+
+  return a;
+}
+
+//
+// CHECK: define {{.*}}[[FS1]]
+//
+// CHECK:          i8* @llvm.stacksave()
+// CHECK-64:       [[B_ADDR:%.+]] = bitcast i[[SZ]]* [[B_CADDR:%.+]] to i32*
+// CHECK-64:       store i32 %{{.+}}, i32* [[B_ADDR]],
+// CHECK-64:       [[B_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[B_CADDR]],
+
+// CHECK-32:       store i32 %{{.+}}, i32* [[B_ADDR:%.+]],
+// CHECK-32:       [[B_CVAL:%.+]] = load i[[SZ]], i[[SZ]]* [[B_ADDR]],
+
+// We capture 2 VLA sizes in this target region
+// CHECK:       [[CELEMSIZE2:%.+]] = mul nuw i[[SZ]] 2, [[VLA0:%.+]]
+// CHECK:       [[CSIZE:%.+]] = mul nuw i[[SZ]] [[CELEMSIZE2]], 2
+
+// CHECK:       [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 60
+// CHECK:       br i1 [[IF]], label %[[TRY:[^,]+]], label %[[FAIL:[^,]+]]
+// CHECK:       [[TRY]]
+// CHECK-DAG:   [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 5, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* [[SR:%[^,]+]], i32* getelementptr inbounds ([5 x i32], [5 x i32]* [[MAPT7]], i32 0, i32 0))
+// CHECK-DAG:   [[BPR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP:%.+]], i32 0, i32 0
+// CHECK-DAG:   [[PR]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P:%.+]], i32 0, i32 0
+// CHECK-DAG:   [[SR]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S:%.+]], i32 0, i32 0
+// CHECK-DAG:   [[SADDR0:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX0:[0-9]+]]
+// CHECK-DAG:   [[BPADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX0]]
+// CHECK-DAG:   [[PADDR0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX0]]
+// CHECK-DAG:   [[SADDR1:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX1:[0-9]+]]
+// CHECK-DAG:   [[BPADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX1]]
+// CHECK-DAG:   [[PADDR1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX1]]
+// CHECK-DAG:   [[SADDR2:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX2:[0-9]+]]
+// CHECK-DAG:   [[BPADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX2]]
+// CHECK-DAG:   [[PADDR2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX2]]
+// CHECK-DAG:   [[SADDR3:%.+]] = getelementptr inbounds [5 x i[[SZ]]], [5 x i[[SZ]]]* [[S]], i32 [[IDX3:[0-9]+]]
+// CHECK-DAG:   [[BPADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BP]], i32 [[IDX3]]
+// CHECK-DAG:   [[PADDR3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[P]], i32 [[IDX3]]
+
+// The names below are not necessarily consistent with the names used for the
+// addresses above as some are repeated.
+// CHECK-DAG:   [[BP0:%[^,]+]] = inttoptr i[[SZ]] [[VLA0]] to i8*
+// CHECK-DAG:   [[P0:%[^,]+]] = inttoptr i[[SZ]] [[VLA0]] to i8*
+// CHECK-DAG:   store i8* [[BP0]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i8* [[P0]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}}
+
+// CHECK-DAG:   store i8* inttoptr (i[[SZ]] 2 to i8*), i8** {{%[^,]+}}
+// CHECK-DAG:   store i8* inttoptr (i[[SZ]] 2 to i8*), i8** {{%[^,]+}}
+// CHECK-DAG:   store i[[SZ]] {{4|8}}, i[[SZ]]* {{%[^,]+}}
+
+// CHECK-DAG:   [[BP2:%[^,]+]] = inttoptr i[[SZ]] [[B_CVAL]] to i8*
+// CHECK-DAG:   [[P2:%[^,]+]] = inttoptr i[[SZ]] [[B_CVAL]] to i8*
+// CHECK-DAG:   store i8* [[BP2]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i8* [[P2]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i[[SZ]] 4, i[[SZ]]* {{%[^,]+}}
+
+// CHECK-DAG:   [[BP3:%[^,]+]] = bitcast [[S1]]* %{{.+}} to i8*
+// CHECK-DAG:   [[P3:%[^,]+]] = bitcast [[S1]]* %{{.+}} to i8*
+// CHECK-DAG:   store i8* [[BP3]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i8* [[P3]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i[[SZ]] 8, i[[SZ]]* {{%[^,]+}}
+
+// CHECK-DAG:   [[BP4:%[^,]+]] = bitcast i16* %{{.+}} to i8*
+// CHECK-DAG:   [[P4:%[^,]+]] = bitcast i16* %{{.+}} to i8*
+// CHECK-DAG:   store i8* [[BP4]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i8* [[P4]], i8** {{%[^,]+}}
+// CHECK-DAG:   store i[[SZ]] [[CSIZE]], i[[SZ]]* {{%[^,]+}}
+
+// CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+// CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+// CHECK-NEXT:  [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+// CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:[^,]+]], label %[[END:[^,]+]]
+
+// CHECK:       [[FAIL]]
+// CHECK:       call void [[HVT7:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
+// CHECK-NEXT:  br label %[[END]]
+// CHECK:       [[END]]
+
+//
+// CHECK: define {{.*}}[[FSTATIC]]
+//
+// CHECK:       [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 50
+// CHECK:       br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
+// CHECK:       [[IFTHEN]]
+// CHECK-DAG:   [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 4, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* getelementptr inbounds ([4 x i[[SZ]]], [4 x i[[SZ]]]* [[SIZET6]], i32 0, i32 0), i32* getelementptr inbounds ([4 x i32], [4 x i32]* [[MAPT6]], i32 0, i32 0))
+// CHECK-DAG:   [[BPR]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP:%.+]], i32 0, i32 0
+// CHECK-DAG:   [[PR]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P:%.+]], i32 0, i32 0
+
+// CHECK-DAG:   [[BPADDR0:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 0
+// CHECK-DAG:   [[PADDR0:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 0
+// CHECK-DAG:   store i8* [[BP0:%[^,]+]], i8** [[BPADDR0]]
+// CHECK-DAG:   store i8* [[P0:%[^,]+]], i8** [[PADDR0]]
+// CHECK-DAG:   [[BP0]] = inttoptr i[[SZ]] [[VAL0:%.+]] to i8*
+// CHECK-DAG:   [[P0]] = inttoptr i[[SZ]] [[VAL0]] to i8*
+
+// CHECK-DAG:   [[BPADDR1:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 1
+// CHECK-DAG:   [[PADDR1:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 1
+// CHECK-DAG:   store i8* [[BP1:%[^,]+]], i8** [[BPADDR1]]
+// CHECK-DAG:   store i8* [[P1:%[^,]+]], i8** [[PADDR1]]
+// CHECK-DAG:   [[BP1]] = inttoptr i[[SZ]] [[VAL1:%.+]] to i8*
+// CHECK-DAG:   [[P1]] = inttoptr i[[SZ]] [[VAL1]] to i8*
+
+// CHECK-DAG:   [[BPADDR2:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 2
+// CHECK-DAG:   [[PADDR2:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 2
+// CHECK-DAG:   store i8* [[BP2:%[^,]+]], i8** [[BPADDR2]]
+// CHECK-DAG:   store i8* [[P2:%[^,]+]], i8** [[PADDR2]]
+
+// CHECK-DAG:   [[BPADDR3:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[BP]], i32 0, i32 3
+// CHECK-DAG:   [[PADDR3:%.+]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[P]], i32 0, i32 3
+// CHECK-DAG:   store i8* [[BP3:%[^,]+]], i8** [[BPADDR3]]
+// CHECK-DAG:   store i8* [[P3:%[^,]+]], i8** [[PADDR3]]
+// CHECK-DAG:   [[BP3]] = bitcast [10 x i32]* %{{.+}} to i8*
+// CHECK-DAG:   [[P3]] = bitcast [10 x i32]* %{{.+}} to i8*
+
+// CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+// CHECK-NEXT:  br label %[[IFEND:.+]]
+
+// CHECK:       [[IFELSE]]
+// CHECK:       store i32 -1, i32* [[RHV]], align 4
+// CHECK-NEXT:  br label %[[IFEND:.+]]
+
+// CHECK:       [[IFEND]]
+// CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+// CHECK:       [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+// CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]]
+// CHECK:       [[FAIL]]
+// CHECK:       call void [[HVT6:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}}, {{[^,]+}})
+// CHECK-NEXT:  br label %[[END]]
+// CHECK:       [[END]]
+
+//
+// CHECK: define {{.*}}[[FTEMPLATE]]
+//
+// CHECK:       [[IF:%.+]] = icmp sgt i32 {{[^,]+}}, 40
+// CHECK:       br i1 [[IF]], label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
+// CHECK:       [[IFTHEN]]
+// CHECK-DAG:   [[RET:%.+]] = call i32 @__tgt_target(i32 -1, i8* @{{[^,]+}}, i32 3, i8** [[BPR:%[^,]+]], i8** [[PR:%[^,]+]], i[[SZ]]* getelementptr inbounds ([3 x i[[SZ]]], [3 x i[[SZ]]]* [[SIZET5]], i32 0, i32 0), i32* getelementptr inbounds ([3 x i32], [3 x i32]* [[MAPT5]], i32 0, i32 0))
+// CHECK-DAG:   [[BPR]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP:%.+]], i32 0, i32 0
+// CHECK-DAG:   [[PR]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P:%.+]], i32 0, i32 0
+
+// CHECK-DAG:   [[BPADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 0
+// CHECK-DAG:   [[PADDR0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 0
+// CHECK-DAG:   store i8* [[BP0:%[^,]+]], i8** [[BPADDR0]]
+// CHECK-DAG:   store i8* [[P0:%[^,]+]], i8** [[PADDR0]]
+// CHECK-DAG:   [[BP0]] = inttoptr i[[SZ]] [[VAL0:%.+]] to i8*
+// CHECK-DAG:   [[P0]] = inttoptr i[[SZ]] [[VAL0]] to i8*
+
+// CHECK-DAG:   [[BPADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 1
+// CHECK-DAG:   [[PADDR1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 1
+// CHECK-DAG:   store i8* [[BP1:%[^,]+]], i8** [[BPADDR1]]
+// CHECK-DAG:   store i8* [[P1:%[^,]+]], i8** [[PADDR1]]
+// CHECK-DAG:   [[BP1]] = inttoptr i[[SZ]] [[VAL1:%.+]] to i8*
+// CHECK-DAG:   [[P1]] = inttoptr i[[SZ]] [[VAL1]] to i8*
+
+// CHECK-DAG:   [[BPADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BP]], i32 0, i32 2
+// CHECK-DAG:   [[PADDR2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[P]], i32 0, i32 2
+// CHECK-DAG:   store i8* [[BP2:%[^,]+]], i8** [[BPADDR2]]
+// CHECK-DAG:   store i8* [[P2:%[^,]+]], i8** [[PADDR2]]
+// CHECK-DAG:   [[BP2]] = bitcast [10 x i32]* %{{.+}} to i8*
+// CHECK-DAG:   [[P2]] = bitcast [10 x i32]* %{{.+}} to i8*
+
+// CHECK:       store i32 [[RET]], i32* [[RHV:%.+]], align 4
+// CHECK-NEXT:  br label %[[IFEND:.+]]
+
+// CHECK:       [[IFELSE]]
+// CHECK:       store i32 -1, i32* [[RHV]], align 4
+// CHECK-NEXT:  br label %[[IFEND:.+]]
+
+// CHECK:       [[IFEND]]
+// CHECK:       [[RET2:%.+]] = load i32, i32* [[RHV]], align 4
+// CHECK:       [[ERROR:%.+]] = icmp ne i32 [[RET2]], 0
+// CHECK-NEXT:  br i1 [[ERROR]], label %[[FAIL:.+]], label %[[END:[^,]+]]
+// CHECK:       [[FAIL]]
+// CHECK:       call void [[HVT5:@.+]]({{[^,]+}}, {{[^,]+}}, {{[^,]+}})
+// CHECK-NEXT:  br label %[[END]]
+// CHECK:       [[END]]
+
+
+
+// Check that the offloading functions are emitted and that the arguments are
+// correct and loaded correctly for the target regions of the callees of bar().
+
+// CHECK:       define internal void [[HVT7]]
+// Create local storage for each capture.
+// CHECK:       [[LOCAL_THIS:%.+]] = alloca [[S1]]*
+// CHECK:       [[LOCAL_B:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_VLA1:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_VLA2:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_C:%.+]] = alloca i16*
+// CHECK-DAG:   store [[S1]]* [[ARG_THIS:%.+]], [[S1]]** [[LOCAL_THIS]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_B:%.+]], i[[SZ]]* [[LOCAL_B]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_VLA1:%.+]], i[[SZ]]* [[LOCAL_VLA1]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_VLA2:%.+]], i[[SZ]]* [[LOCAL_VLA2]]
+// CHECK-DAG:   store i16* [[ARG_C:%.+]], i16** [[LOCAL_C]]
+// Store captures in the context.
+// CHECK-DAG:   [[REF_THIS:%.+]] = load [[S1]]*, [[S1]]** [[LOCAL_THIS]],
+// CHECK-64-DAG:[[REF_B:%.+]] = bitcast i[[SZ]]* [[LOCAL_B]] to i32*
+// CHECK-DAG:   [[VAL_VLA1:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA1]],
+// CHECK-DAG:   [[VAL_VLA2:%.+]] = load i[[SZ]], i[[SZ]]* [[LOCAL_VLA2]],
+// CHECK-DAG:   [[REF_C:%.+]] = load i16*, i16** [[LOCAL_C]],
+// Use captures.
+// CHECK-DAG:   getelementptr inbounds [[S1]], [[S1]]* [[REF_THIS]], i32 0, i32 0
+// CHECK-64-DAG:load i32, i32* [[REF_B]]
+// CHECK-32-DAG:load i32, i32* [[LOCAL_B]]
+// CHECK-DAG:   getelementptr inbounds i16, i16* [[REF_C]], i[[SZ]] %{{.+}}
+
+
+// CHECK:       define internal void [[HVT6]]
+// Create local storage for each capture.
+// CHECK:       [[LOCAL_A:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_AA:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_AAA:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_B:%.+]] = alloca [10 x i32]*
+// CHECK-DAG:   store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_AA:%.+]], i[[SZ]]* [[LOCAL_AA]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_AAA:%.+]], i[[SZ]]* [[LOCAL_AAA]]
+// CHECK-DAG:   store [10 x i32]* [[ARG_B:%.+]], [10 x i32]** [[LOCAL_B]]
+// Store captures in the context.
+// CHECK-64-DAG:   [[REF_A:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32*
+// CHECK-DAG:      [[REF_AA:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA]] to i16*
+// CHECK-DAG:      [[REF_AAA:%.+]] = bitcast i[[SZ]]* [[LOCAL_AAA]] to i8*
+// CHECK-DAG:      [[REF_B:%.+]] = load [10 x i32]*, [10 x i32]** [[LOCAL_B]],
+// Use captures.
+// CHECK-64-DAG:   load i32, i32* [[REF_A]]
+// CHECK-DAG:      load i16, i16* [[REF_AA]]
+// CHECK-DAG:      load i8, i8* [[REF_AAA]]
+// CHECK-32-DAG:   load i32, i32* [[LOCAL_A]]
+// CHECK-DAG:      getelementptr inbounds [10 x i32], [10 x i32]* [[REF_B]], i[[SZ]] 0, i[[SZ]] 2
+
+// CHECK:       define internal void [[HVT5]]
+// Create local storage for each capture.
+// CHECK:       [[LOCAL_A:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_AA:%.+]] = alloca i[[SZ]]
+// CHECK:       [[LOCAL_B:%.+]] = alloca [10 x i32]*
+// CHECK-DAG:   store i[[SZ]] [[ARG_A:%.+]], i[[SZ]]* [[LOCAL_A]]
+// CHECK-DAG:   store i[[SZ]] [[ARG_AA:%.+]], i[[SZ]]* [[LOCAL_AA]]
+// CHECK-DAG:   store [10 x i32]* [[ARG_B:%.+]], [10 x i32]** [[LOCAL_B]]
+// Store captures in the context.
+// CHECK-64-DAG:[[REF_A:%.+]] = bitcast i[[SZ]]* [[LOCAL_A]] to i32*
+// CHECK-DAG:   [[REF_AA:%.+]] = bitcast i[[SZ]]* [[LOCAL_AA]] to i16*
+// CHECK-DAG:   [[REF_B:%.+]] = load [10 x i32]*, [10 x i32]** [[LOCAL_B]],
+// Use captures.
+// CHECK-64-DAG:   load i32, i32* [[REF_A]]
+// CHECK-32-DAG:   load i32, i32* [[LOCAL_A]]
+// CHECK-DAG:   load i16, i16* [[REF_AA]]
+// CHECK-DAG:   getelementptr inbounds [10 x i32], [10 x i32]* [[REF_B]], i[[SZ]] 0, i[[SZ]] 2
+#endif
diff --git a/test/OpenMP/target_codegen_global_capture.cpp b/test/OpenMP/target_codegen_global_capture.cpp
new file mode 100644
index 000000000000..29469cdf71ea
--- /dev/null
+++ b/test/OpenMP/target_codegen_global_capture.cpp
@@ -0,0 +1,287 @@
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
+// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
+// expected-no-diagnostics
+#ifndef HEADER
+#define HEADER
+
+
+// CHECK-DAG: [[GA:@.+]] = global double 1.000000e+00
+// CHECK-DAG: [[GB:@.+]] = global double 2.000000e+00
+// CHECK-DAG: [[GC:@.+]] = global double 3.000000e+00
+// CHECK-DAG: [[GD:@.+]] = global double 4.000000e+00
+// CHECK-DAG: [[FA:@.+]] = internal global float 5.000000e+00
+// CHECK-DAG: [[FB:@.+]] = internal global float 6.000000e+00
+// CHECK-DAG: [[FC:@.+]] = internal global float 7.000000e+00
+// CHECK-DAG: [[FD:@.+]] = internal global float 8.000000e+00
+// CHECK-DAG: [[BA:@.+]] = internal global float 9.000000e+00
+// CHECK-DAG: [[BB:@.+]] = internal global float 1.000000e+01
+// CHECK-DAG: [[BC:@.+]] = internal global float 1.100000e+01
+// CHECK-DAG: [[BD:@.+]] = internal global float 1.200000e+01
+double Ga = 1.0;
+double Gb = 2.0;
+double Gc = 3.0;
+double Gd = 4.0;
+
+// CHECK: define {{.*}} @{{.*}}foo{{.*}}(
+// CHECK-SAME: i16 {{[^,]*}}[[A:%[^,]+]],
+// CHECK-SAME: i16 {{[^,]*}}[[B:%[^,]+]],
+// CHECK-SAME: i16 {{[^,]*}}[[C:%[^,]+]],
+// CHECK-SAME: i16 {{[^,]*}}[[D:%[^,]+]])
+// CHECK: [[LA:%.+]] = alloca i16
+// CHECK: [[LB:%.+]] = alloca i16
+// CHECK: [[LC:%.+]] = alloca i16
+// CHECK: [[LD:%.+]] = alloca i16
+int foo(short a, short b, short c, short d){
+  static float Sa = 5.0;
+  static float Sb = 6.0;
+  static float Sc = 7.0;
+  static float Sd = 8.0;
+
+  // CHECK-DAG:    [[VALLB:%.+]] = load i16, i16* [[LB]],
+  // CHECK-64-DAG: [[VALGB:%.+]] = load double, double* @Gb,
+  // CHECK-DAG:    [[VALFB:%.+]] = load float, float* @_ZZ3foossssE2Sb,
+  // CHECK-64-DAG: [[VALGC:%.+]] = load double, double* @Gc,
+  // CHECK-DAG:    [[VALLC:%.+]] = load i16, i16* [[LC]],
+  // CHECK-DAG:    [[VALFC:%.+]] = load float, float* @_ZZ3foossssE2Sc,
+  // CHECK-DAG:    [[VALLD:%.+]] = load i16, i16* [[LD]],
+  // CHECK-64-DAG: [[VALGD:%.+]] = load double, double* @Gd,
+  // CHECK-DAG:    [[VALFD:%.+]] = load float, float* @_ZZ3foossssE2Sd,
+
+  // 3 local vars being captured.
+
+  // CHECK-DAG: store i16 [[VALLB]], i16* [[CONVLB:%.+]],
+  // CHECK-DAG: [[CONVLB]] = bitcast i[[sz:64|32]]* [[CADDRLB:%.+]] to i16*
+  // CHECK-DAG: [[CVALLB:%.+]] = load i[[sz]], i[[sz]]* [[CADDRLB]],
+  // CHECK-DAG: [[CPTRLB:%.+]] = inttoptr i[[sz]] [[CVALLB]] to i8*
+  // CHECK-DAG: store i8* [[CPTRLB]], i8** [[GEPLB:%.+]],
+  // CHECK-DAG: [[GEPLB]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK-DAG: store i16 [[VALLC]], i16* [[CONVLC:%.+]],
+  // CHECK-DAG: [[CONVLC]] = bitcast i[[sz]]* [[CADDRLC:%.+]] to i16*
+  // CHECK-DAG: [[CVALLC:%.+]] = load i[[sz]], i[[sz]]* [[CADDRLC]],
+  // CHECK-DAG: [[CPTRLC:%.+]] = inttoptr i[[sz]] [[CVALLC]] to i8*
+  // CHECK-DAG: store i8* [[CPTRLC]], i8** [[GEPLC:%.+]],
+  // CHECK-DAG: [[GEPLC]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK-DAG: store i16 [[VALLD]], i16* [[CONVLD:%.+]],
+  // CHECK-DAG: [[CONVLD]] = bitcast i[[sz]]* [[CADDRLD:%.+]] to i16*
+  // CHECK-DAG: [[CVALLD:%.+]] = load i[[sz]], i[[sz]]* [[CADDRLD]],
+  // CHECK-DAG: [[CPTRLD:%.+]] = inttoptr i[[sz]] [[CVALLD]] to i8*
+  // CHECK-DAG: store i8* [[CPTRLD]], i8** [[GEPLD:%.+]],
+  // CHECK-DAG: [[GEPLD]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // 3 static vars being captured.
+
+  // CHECK-DAG: store float [[VALFB]], float* [[CONVFB:%.+]],
+  // CHECK-DAG: [[CONVFB]] = bitcast i[[sz]]* [[CADDRFB:%.+]] to float*
+  // CHECK-DAG: [[CVALFB:%.+]] = load i[[sz]], i[[sz]]* [[CADDRFB]],
+  // CHECK-DAG: [[CPTRFB:%.+]] = inttoptr i[[sz]] [[CVALFB]] to i8*
+  // CHECK-DAG: store i8* [[CPTRFB]], i8** [[GEPFB:%.+]],
+  // CHECK-DAG: [[GEPFB]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK-DAG: store float [[VALFC]], float* [[CONVFC:%.+]],
+  // CHECK-DAG: [[CONVFC]] = bitcast i[[sz]]* [[CADDRFC:%.+]] to float*
+  // CHECK-DAG: [[CVALFC:%.+]] = load i[[sz]], i[[sz]]* [[CADDRFC]],
+  // CHECK-DAG: [[CPTRFC:%.+]] = inttoptr i[[sz]] [[CVALFC]] to i8*
+  // CHECK-DAG: store i8* [[CPTRFC]], i8** [[GEPFC:%.+]],
+  // CHECK-DAG: [[GEPFC]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK-DAG: store float [[VALFD]], float* [[CONVFD:%.+]],
+  // CHECK-DAG: [[CONVFD]] = bitcast i[[sz]]* [[CADDRFD:%.+]] to float*
+  // CHECK-DAG: [[CVALFD:%.+]] = load i[[sz]], i[[sz]]* [[CADDRFD]],
+  // CHECK-DAG: [[CPTRFD:%.+]] = inttoptr i[[sz]] [[CVALFD]] to i8*
+  // CHECK-DAG: store i8* [[CPTRFD]], i8** [[GEPFD:%.+]],
+  // CHECK-DAG: [[GEPFD]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // 3 static global vars being captured.
+
+  // CHECK-64-DAG: store double [[VALGB]], double* [[CONVGB:%.+]],
+  // CHECK-64-DAG: [[CONVGB]] = bitcast i[[sz]]* [[CADDRGB:%.+]] to double*
+  // CHECK-64-DAG: [[CVALGB:%.+]] = load i[[sz]], i[[sz]]* [[CADDRGB]],
+  // CHECK-64-DAG: [[CPTRGB:%.+]] = inttoptr i[[sz]] [[CVALGB]] to i8*
+  // CHECK-64-DAG: store i8* [[CPTRGB]], i8** [[GEPGB:%.+]],
+  // CHECK-32-DAG: store i8* bitcast (double* @Gb to i8*), i8** [[GEPGB:%.+]],
+  // CHECK-DAG: [[GEPGB]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK-64-DAG: store double [[VALGC]], double* [[CONVGC:%.+]],
+  // CHECK-64-DAG: [[CONVGC]] = bitcast i[[sz]]* [[CADDRGC:%.+]] to double*
+  // CHECK-64-DAG: [[CVALGC:%.+]] = load i[[sz]], i[[sz]]* [[CADDRGC]],
+  // CHECK-64-DAG: [[CPTRGC:%.+]] = inttoptr i[[sz]] [[CVALGC]] to i8*
+  // CHECK-64-DAG: store i8* [[CPTRGC]], i8** [[GEPGC:%.+]],
+  // CHECK-32-DAG: store i8* bitcast (double* @Gc to i8*), i8** [[GEPGC:%.+]],
+  // CHECK-DAG: [[GEPGC]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK-64-DAG: store double [[VALGD]], double* [[CONVGD:%.+]],
+  // CHECK-64-DAG: [[CONVGD]] = bitcast i[[sz]]* [[CADDRGD:%.+]] to double*
+  // CHECK-64-DAG: [[CVALGD:%.+]] = load i[[sz]], i[[sz]]* [[CADDRGD]],
+  // CHECK-64-DAG: [[CPTRGD:%.+]] = inttoptr i[[sz]] [[CVALGD]] to i8*
+  // CHECK-64-DAG: store i8* [[CPTRGD]], i8** [[GEPGD:%.+]],
+  // CHECK-32-DAG: store i8* bitcast (double* @Gd to i8*), i8** [[GEPGD:%.+]],
+  // CHECK-DAG: [[GEPGD]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+  // CHECK: call i32 @__tgt_target
+  // CHECK: call void [[OFFLOADF:@.+]](
+  // Capture b, Gb, Sb, Gc, c, Sc, d, Gd, Sd
+  #pragma omp target if(Ga>0.0 && a>0 && Sa>0.0)
+  {
+    b += 1;
+    Gb += 1.0;
+    Sb += 1.0;
+
+    // CHECK: define internal void [[OFFLOADF]]({{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}})
+    // The parallel region only uses 3 captures.
+    // CHECK:     call {{.*}}@__kmpc_fork_call(%ident_t* {{.+}}, i32 {{.+}}, void (i32*, i32*, ...)* bitcast ({{.*}}[[PARF:@.+]] to {{.*}}), {{.+}}* %{{.+}}, {{.+}}* %{{.+}}, {{.+}}* %{{.+}})
+    // CHECK:     call void @.omp_outlined.(i32* %{{.+}}, i32* %{{.+}}, {{.+}}* %{{.+}}, {{.+}}* %{{.+}}, {{.+}}* %{{.+}})
+    // Capture d, Gd, Sd,
+
+    // CHECK: define internal void [[PARF]](i32* noalias %{{.*}}, i32* noalias %{{.*}},
+    #pragma omp parallel if(Gc>0.0 && c>0 && Sc>0.0)
+    {
+      d += 1;
+      Gd += 1.0;
+      Sd += 1.0;
+    }
+  }
+  return a + b + c + d + (int)Sa + (int)Sb + (int)Sc + (int)Sd;
+}
+
+// CHECK: define {{.*}} @{{.*}}bar{{.*}}(
+// CHECK-SAME: i16 {{[^,]*}}[[A:%[^,]+]],
+// CHECK-SAME: i16 {{[^,]*}}[[B:%[^,]+]],
+// CHECK-SAME: i16 {{[^,]*}}[[C:%[^,]+]],
+// CHECK-SAME: i16 {{[^,]*}}[[D:%[^,]+]])
+// CHECK: [[LA:%.+]] = alloca i16
+// CHECK: [[LB:%.+]] = alloca i16
+// CHECK: [[LC:%.+]] = alloca i16
+// CHECK: [[LD:%.+]] = alloca i16
+int bar(short a, short b, short c, short d){
+  static float Sa = 9.0;
+  static float Sb = 10.0;
+  static float Sc = 11.0;
+  static float Sd = 12.0;
+
+  // CHECK: call void {{.*}}@__kmpc_fork_call(%ident_t* {{.+}}, i32 {{.+}}, void (i32*, i32*, ...)* bitcast ({{.*}}[[PARF:@.+]] to {{.*}}), i16* %{{.+}}, i16* %{{.+}}, i16* %{{.+}}, i16* %{{.+}})
+  // CHECK: define internal void [[PARF]](i32* noalias %{{.*}}, i32* noalias %{{.*}}, i16* dereferenceable(2) [[A:%.+]], i16* dereferenceable(2) [[B:%.+]], i16* dereferenceable(2) [[C:%.+]], i16* dereferenceable(2) [[D:%.+]])
+  // Capture a, b, c, d
+  // CHECK: [[ALLOCLA:%.+]] = alloca i16
+  // CHECK: [[ALLOCLB:%.+]] = alloca i16
+  // CHECK: [[ALLOCLC:%.+]] = alloca i16
+  // CHECK: [[ALLOCLD:%.+]] = alloca i16
+  // CHECK: [[LLA:%.+]] = load i16*, i16** [[ALLOCLA]],
+  // CHECK: [[LLB:%.+]] = load i16*, i16** [[ALLOCLB]],
+  // CHECK: [[LLC:%.+]] = load i16*, i16** [[ALLOCLC]],
+  // CHECK: [[LLD:%.+]] = load i16*, i16** [[ALLOCLD]],
+  #pragma omp parallel
+  {
+    // CHECK-DAG:    [[VALLB:%.+]] = load i16, i16* [[LLB]],
+    // CHECK-64-DAG: [[VALGB:%.+]] = load double, double* @Gb,
+    // CHECK-DAG:    [[VALFB:%.+]] = load float, float* @_ZZ3barssssE2Sb,
+    // CHECK-64-DAG: [[VALGC:%.+]] = load double, double* @Gc,
+    // CHECK-DAG:    [[VALLC:%.+]] = load i16, i16* [[LLC]],
+    // CHECK-DAG:    [[VALFC:%.+]] = load float, float* @_ZZ3barssssE2Sc,
+    // CHECK-DAG:    [[VALLD:%.+]] = load i16, i16* [[LLD]],
+    // CHECK-64-DAG: [[VALGD:%.+]] = load double, double* @Gd,
+    // CHECK-DAG:    [[VALFD:%.+]] = load float, float* @_ZZ3barssssE2Sd,
+
+    // 3 local vars being captured.
+
+    // CHECK-DAG: store i16 [[VALLB]], i16* [[CONVLB:%.+]],
+    // CHECK-DAG: [[CONVLB]] = bitcast i[[sz:64|32]]* [[CADDRLB:%.+]] to i16*
+    // CHECK-DAG: [[CVALLB:%.+]] = load i[[sz]], i[[sz]]* [[CADDRLB]],
+    // CHECK-DAG: [[CPTRLB:%.+]] = inttoptr i[[sz]] [[CVALLB]] to i8*
+    // CHECK-DAG: store i8* [[CPTRLB]], i8** [[GEPLB:%.+]],
+    // CHECK-DAG: [[GEPLB]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK-DAG: store i16 [[VALLC]], i16* [[CONVLC:%.+]],
+    // CHECK-DAG: [[CONVLC]] = bitcast i[[sz]]* [[CADDRLC:%.+]] to i16*
+    // CHECK-DAG: [[CVALLC:%.+]] = load i[[sz]], i[[sz]]* [[CADDRLC]],
+    // CHECK-DAG: [[CPTRLC:%.+]] = inttoptr i[[sz]] [[CVALLC]] to i8*
+    // CHECK-DAG: store i8* [[CPTRLC]], i8** [[GEPLC:%.+]],
+    // CHECK-DAG: [[GEPLC]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK-DAG: store i16 [[VALLD]], i16* [[CONVLD:%.+]],
+    // CHECK-DAG: [[CONVLD]] = bitcast i[[sz]]* [[CADDRLD:%.+]] to i16*
+    // CHECK-DAG: [[CVALLD:%.+]] = load i[[sz]], i[[sz]]* [[CADDRLD]],
+    // CHECK-DAG: [[CPTRLD:%.+]] = inttoptr i[[sz]] [[CVALLD]] to i8*
+    // CHECK-DAG: store i8* [[CPTRLD]], i8** [[GEPLD:%.+]],
+    // CHECK-DAG: [[GEPLD]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // 3 static vars being captured.
+
+    // CHECK-DAG: store float [[VALFB]], float* [[CONVFB:%.+]],
+    // CHECK-DAG: [[CONVFB]] = bitcast i[[sz]]* [[CADDRFB:%.+]] to float*
+    // CHECK-DAG: [[CVALFB:%.+]] = load i[[sz]], i[[sz]]* [[CADDRFB]],
+    // CHECK-DAG: [[CPTRFB:%.+]] = inttoptr i[[sz]] [[CVALFB]] to i8*
+    // CHECK-DAG: store i8* [[CPTRFB]], i8** [[GEPFB:%.+]],
+    // CHECK-DAG: [[GEPFB]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK-DAG: store float [[VALFC]], float* [[CONVFC:%.+]],
+    // CHECK-DAG: [[CONVFC]] = bitcast i[[sz]]* [[CADDRFC:%.+]] to float*
+    // CHECK-DAG: [[CVALFC:%.+]] = load i[[sz]], i[[sz]]* [[CADDRFC]],
+    // CHECK-DAG: [[CPTRFC:%.+]] = inttoptr i[[sz]] [[CVALFC]] to i8*
+    // CHECK-DAG: store i8* [[CPTRFC]], i8** [[GEPFC:%.+]],
+    // CHECK-DAG: [[GEPFC]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK-DAG: store float [[VALFD]], float* [[CONVFD:%.+]],
+    // CHECK-DAG: [[CONVFD]] = bitcast i[[sz]]* [[CADDRFD:%.+]] to float*
+    // CHECK-DAG: [[CVALFD:%.+]] = load i[[sz]], i[[sz]]* [[CADDRFD]],
+    // CHECK-DAG: [[CPTRFD:%.+]] = inttoptr i[[sz]] [[CVALFD]] to i8*
+    // CHECK-DAG: store i8* [[CPTRFD]], i8** [[GEPFD:%.+]],
+    // CHECK-DAG: [[GEPFD]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // 3 static global vars being captured.
+
+    // CHECK-64-DAG: store double [[VALGB]], double* [[CONVGB:%.+]],
+    // CHECK-64-DAG: [[CONVGB]] = bitcast i[[sz]]* [[CADDRGB:%.+]] to double*
+    // CHECK-64-DAG: [[CVALGB:%.+]] = load i[[sz]], i[[sz]]* [[CADDRGB]],
+    // CHECK-64-DAG: [[CPTRGB:%.+]] = inttoptr i[[sz]] [[CVALGB]] to i8*
+    // CHECK-64-DAG: store i8* [[CPTRGB]], i8** [[GEPGB:%.+]],
+    // CHECK-32-DAG: store i8* bitcast (double* @Gb to i8*), i8** [[GEPGB:%.+]],
+    // CHECK-DAG: [[GEPGB]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK-64-DAG: store double [[VALGC]], double* [[CONVGC:%.+]],
+    // CHECK-64-DAG: [[CONVGC]] = bitcast i[[sz]]* [[CADDRGC:%.+]] to double*
+    // CHECK-64-DAG: [[CVALGC:%.+]] = load i[[sz]], i[[sz]]* [[CADDRGC]],
+    // CHECK-64-DAG: [[CPTRGC:%.+]] = inttoptr i[[sz]] [[CVALGC]] to i8*
+    // CHECK-64-DAG: store i8* [[CPTRGC]], i8** [[GEPGC:%.+]],
+    // CHECK-32-DAG: store i8* bitcast (double* @Gc to i8*), i8** [[GEPGC:%.+]],
+    // CHECK-DAG: [[GEPGC]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK-64-DAG: store double [[VALGD]], double* [[CONVGD:%.+]],
+    // CHECK-64-DAG: [[CONVGD]] = bitcast i[[sz]]* [[CADDRGD:%.+]] to double*
+    // CHECK-64-DAG: [[CVALGD:%.+]] = load i[[sz]], i[[sz]]* [[CADDRGD]],
+    // CHECK-64-DAG: [[CPTRGD:%.+]] = inttoptr i[[sz]] [[CVALGD]] to i8*
+    // CHECK-64-DAG: store i8* [[CPTRGD]], i8** [[GEPGD:%.+]],
+    // CHECK-32-DAG: store i8* bitcast (double* @Gd to i8*), i8** [[GEPGD:%.+]],
+    // CHECK-DAG: [[GEPGD]] = getelementptr inbounds [9 x i8*], [9 x i8*]* %{{.+}}, i32 0, i32 {{[0-8]}}
+
+    // CHECK: call i32 @__tgt_target
+    // CHECK: call void [[OFFLOADF:@.+]](
+    // Capture b, Gb, Sb, Gc, c, Sc, d, Gd, Sd
+    #pragma omp target if(Ga>0.0 && a>0 && Sa>0.0)
+    {
+      b += 1;
+      Gb += 1.0;
+      Sb += 1.0;
+
+      // CHECK: define internal void [[OFFLOADF]]({{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}}, {{.+}} {{.*}}%{{.+}})
+      // CHECK: call void {{.*}}@__kmpc_fork_call(%ident_t* {{.+}}, i32 {{.+}}, void (i32*, i32*, ...)* bitcast ({{.*}}[[PARF:@.+]] to {{.*}})
+
+      // CHECK: define internal void [[PARF]](i32* noalias %{{.*}}, i32* noalias %{{.*}}, {{.+}}* dereferenceable({{.+}}) %{{.+}}, {{.+}}* dereferenceable({{.+}}) %{{.+}}, {{.+}}* dereferenceable({{.+}}) %{{.+}})
+      // Capture d, Gd, Sd
+      #pragma omp parallel if(Gc>0.0 && c>0 && Sc>0.0)
+      {
+        d += 1;
+        Gd += 1.0;
+        Sd += 1.0;
+      }
+    }
+  }
+  return a + b + c + d + (int)Sa + (int)Sb + (int)Sc + (int)Sd;
+}
+
+#endif
diff --git a/test/OpenMP/target_data_ast_print.cpp b/test/OpenMP/target_data_ast_print.cpp
new file mode 100644
index 000000000000..cdff857e569a
--- /dev/null
+++ b/test/OpenMP/target_data_ast_print.cpp
@@ -0,0 +1,173 @@
+// RUN: %clang_cc1 -verify -fopenmp -ast-print %s | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -std=c++11 -include-pch %t -fsyntax-only -verify %s -ast-print | FileCheck %s
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void foo() {}
+
+template <typename T, int C>
+T tmain(T argc, T *argv) {
+  T i, j, b, c, d, e, x[20];
+
+#pragma omp target data
+  i = argc;
+
+#pragma omp target data if (target data: j > 0)
+  foo();
+
+#pragma omp target data if (b)
+  foo();
+
+#pragma omp target data map(c)
+  foo();
+
+#pragma omp target data map(c) if(b>e)
+  foo();
+
+#pragma omp target data map(x[0:10], c)
+  foo();
+
+#pragma omp target data map(to: c) map(from: d)
+  foo();
+
+#pragma omp target data map(always,alloc: e)
+  foo();
+
+// nesting a target region
+#pragma omp target data map(e)
+{
+  #pragma omp target map(always, alloc: e)
+    foo();
+}
+
+  return 0;
+}
+
+// CHECK: template <typename T = int, int C = 5> int tmain(int argc, int *argv) {
+// CHECK-NEXT: int i, j, b, c, d, e, x[20];
+// CHECK-NEXT: #pragma omp target data
+// CHECK-NEXT: i = argc;
+// CHECK-NEXT: #pragma omp target data if(target data: j > 0)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data if(b)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: c)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: c) if(b > e)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: x[0:10],c)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(to: c) map(from: d)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(always,alloc: e)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: e)
+// CHECK-NEXT: {
+// CHECK-NEXT: #pragma omp target map(always,alloc: e)
+// CHECK-NEXT: foo();
+// CHECK: template <typename T = char, int C = 1> char tmain(char argc, char *argv) {
+// CHECK-NEXT: char i, j, b, c, d, e, x[20];
+// CHECK-NEXT: #pragma omp target data
+// CHECK-NEXT: i = argc;
+// CHECK-NEXT: #pragma omp target data if(target data: j > 0)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data if(b)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: c)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: c) if(b > e)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: x[0:10],c)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(to: c) map(from: d)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(always,alloc: e)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: e)
+// CHECK-NEXT: {
+// CHECK-NEXT: #pragma omp target map(always,alloc: e)
+// CHECK-NEXT: foo();
+// CHECK: template <typename T, int C> T tmain(T argc, T *argv) {
+// CHECK-NEXT: T i, j, b, c, d, e, x[20];
+// CHECK-NEXT: #pragma omp target data
+// CHECK-NEXT: i = argc;
+// CHECK-NEXT: #pragma omp target data if(target data: j > 0)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data if(b)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: c)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: c) if(b > e)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: x[0:10],c)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(to: c) map(from: d)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(always,alloc: e)
+// CHECK-NEXT: foo();
+// CHECK-NEXT: #pragma omp target data map(tofrom: e)
+// CHECK-NEXT: {
+// CHECK-NEXT: #pragma omp target map(always,alloc: e)
+// CHECK-NEXT: foo();
+
+int main (int argc, char **argv) {
+  int b = argc, c, d, e, f, g, x[20];
+  static int a;
+// CHECK: static int a;
+
+#pragma omp target data
+// CHECK:      #pragma omp target data
+  a=2;
+// CHECK-NEXT: a = 2;
+#pragma omp target data if (target data: b)
+// CHECK: #pragma omp target data if(target data: b)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target data if (b > g)
+// CHECK: #pragma omp target data if(b > g)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target data map(c)
+// CHECK-NEXT: #pragma omp target data map(tofrom: c)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target data map(c) if(b>g)
+// CHECK-NEXT: #pragma omp target data map(tofrom: c) if(b > g)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target data map(x[0:10], c)
+// CHECK-NEXT: #pragma omp target data map(tofrom: x[0:10],c)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target data map(to: c) map(from: d)
+// CHECK-NEXT: #pragma omp target data map(to: c) map(from: d)
+  foo();
+// CHECK-NEXT: foo();
+
+#pragma omp target data map(always,alloc: e)
+// CHECK-NEXT: #pragma omp target data map(always,alloc: e)
+  foo();
+// CHECK-NEXT: foo();
+
+// nesting a target region
+#pragma omp target data map(e)
+// CHECK-NEXT: #pragma omp target data map(tofrom: e)
+{
+// CHECK-NEXT: {
+  #pragma omp target map(always, alloc: e)
+// CHECK-NEXT: #pragma omp target map(always,alloc: e)
+    foo();
+// CHECK-NEXT: foo();
+}
+  return tmain<int, 5>(argc, &argc) + tmain<char, 1>(argv[0][0], argv[0]);
+}
+
+#endif
diff --git a/test/OpenMP/target_data_device_messages.cpp b/test/OpenMP/target_data_device_messages.cpp
new file mode 100644
index 000000000000..9e8e31a28f5b
--- /dev/null
+++ b/test/OpenMP/target_data_device_messages.cpp
@@ -0,0 +1,28 @@
+// RUN: %clang_cc1 -triple x86_64-apple-macos10.7.0 -verify -fopenmp -ferror-limit 100 -o - %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+int main(int argc, char **argv) {
+  #pragma omp target data device // expected-error {{expected '(' after 'device'}}
+  #pragma omp target data device ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data device () // expected-error {{expected expression}}
+  #pragma omp target data device (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data device (argc)) // expected-warning {{extra tokens at the end of '#pragma omp target data' are ignored}}
+#pragma omp target data device (argc > 0 ? argv[1] : argv[2]) // expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
+  #pragma omp target data device (argc + argc)
+  #pragma omp target data device (argc), device (argc+1) // expected-error {{directive '#pragma omp target data' cannot contain more than one 'device' clause}}
+  #pragma omp target data device (S1) // expected-error {{'S1' does not refer to a value}}
+  #pragma omp target data device (-2) // expected-error {{argument to 'device' clause must be a non-negative integer value}}
+  #pragma omp target device (-10u)
+  #pragma omp target device (3.14) // expected-error {{expression must have integral or unscoped enumeration type, not 'double'}}
+  foo();
+
+  return 0;
+}
diff --git a/test/OpenMP/target_data_if_messages.cpp b/test/OpenMP/target_data_if_messages.cpp
new file mode 100644
index 000000000000..77edefa48b8a
--- /dev/null
+++ b/test/OpenMP/target_data_if_messages.cpp
@@ -0,0 +1,32 @@
+// RUN: %clang_cc1 -triple x86_64-apple-macos10.7.0 -verify -fopenmp -ferror-limit 100 -o - %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+int main(int argc, char **argv) {
+  #pragma omp target data if // expected-error {{expected '(' after 'if'}}
+  #pragma omp target data if ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data if () // expected-error {{expected expression}}
+  #pragma omp target data if (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data if (argc)) // expected-warning {{extra tokens at the end of '#pragma omp target data' are ignored}}
+  #pragma omp target data if (argc > 0 ? argv[1] : argv[2])
+  #pragma omp target data if (argc + argc)
+  #pragma omp target data if (foobool(argc)), if (true) // expected-error {{directive '#pragma omp target data' cannot contain more than one 'if' clause}}
+  #pragma omp target data if (S1) // expected-error {{'S1' does not refer to a value}}
+  #pragma omp target data if (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data if(target data : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data if(target data : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target data if(target data : argc)
+  #pragma omp target data if(target data : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp target data'}}
+  #pragma omp target data if(target data : argc) if (target data:argc) // expected-error {{directive '#pragma omp target data' cannot contain more than one 'if' clause with 'target data' name modifier}}
+  #pragma omp target data if(target data : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
+  foo();
+
+  return 0;
+}
diff --git a/test/OpenMP/target_data_messages.c b/test/OpenMP/target_data_messages.c
new file mode 100644
index 000000000000..cd60d85a9030
--- /dev/null
+++ b/test/OpenMP/target_data_messages.c
@@ -0,0 +1,27 @@
+// RUN: %clang_cc1 -triple x86_64-apple-macos10.7.0 -verify -fopenmp -ferror-limit 100 -o - %s
+
+void foo() { }
+
+int main(int argc, char **argv) {
+  L1:
+    foo();
+  #pragma omp target data
+  {
+    foo();
+    goto L1; // expected-error {{use of undeclared label 'L1'}}
+  }
+  goto L2; // expected-error {{use of undeclared label 'L2'}}
+  #pragma omp target data
+  L2:
+  foo();
+
+  #pragma omp target data(i) // expected-warning {{extra tokens at the end of '#pragma omp target data' are ignored}}
+  {
+    foo();
+  }
+  #pragma omp target unknown // expected-warning {{extra tokens at the end of '#pragma omp target' are ignored}}
+  {
+    foo();
+  }
+  return 0;
+}
diff --git a/test/OpenMP/target_device_messages.cpp b/test/OpenMP/target_device_messages.cpp
new file mode 100644
index 000000000000..fb0f2defa42b
--- /dev/null
+++ b/test/OpenMP/target_device_messages.cpp
@@ -0,0 +1,28 @@
+// RUN: %clang_cc1 -triple x86_64-apple-macos10.7.0 -verify -fopenmp -ferror-limit 100 -o - %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+int main(int argc, char **argv) {
+  #pragma omp target device // expected-error {{expected '(' after 'device'}}
+  #pragma omp target device ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target device () // expected-error {{expected expression}}
+  #pragma omp target device (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target device (argc)) // expected-warning {{extra tokens at the end of '#pragma omp target' are ignored}}
+#pragma omp target device (argc > 0 ? argv[1] : argv[2]) // expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
+  #pragma omp target device (argc + argc)
+  #pragma omp target device (argc), device (argc+1) // expected-error {{directive '#pragma omp target' cannot contain more than one 'device' clause}}
+  #pragma omp target device (S1) // expected-error {{'S1' does not refer to a value}}
+  #pragma omp target device (-2) // expected-error {{argument to 'device' clause must be a non-negative integer value}}
+  #pragma omp target device (-10u)
+  #pragma omp target device (3.14) // expected-error {{expression must have integral or unscoped enumeration type, not 'double'}}
+  foo();
+
+  return 0;
+}
diff --git a/test/OpenMP/target_if_messages.cpp b/test/OpenMP/target_if_messages.cpp
index 5193b64ff720..4ee7302282f6 100644
--- a/test/OpenMP/target_if_messages.cpp
+++ b/test/OpenMP/target_if_messages.cpp
@@ -22,6 +22,12 @@ int tmain(T argc, S **argv) {
   #pragma omp target if (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp target if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp target if(argc)
+  #pragma omp target if(target : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target if(target : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target if(target : argc)
+  #pragma omp target if(target : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp target'}}
+  #pragma omp target if(target : argc) if (target:argc) // expected-error {{directive '#pragma omp target' cannot contain more than one 'if' clause with 'target' name modifier}}
+  #pragma omp target if(target : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
   foo();
 
   return 0;
@@ -40,6 +46,12 @@ int main(int argc, char **argv) {
   #pragma omp target if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp target if (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp target if(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target if(target : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target if(target : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp target if(target : argc)
+  #pragma omp target if(target : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp target'}}
+  #pragma omp target if(target : argc) if (target:argc) // expected-error {{directive '#pragma omp target' cannot contain more than one 'if' clause with 'target' name modifier}}
+  #pragma omp target if(target : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
   foo();
 
   return tmain(argc, argv);
diff --git a/test/OpenMP/target_map_codegen.cpp b/test/OpenMP/target_map_codegen.cpp
new file mode 100644
index 000000000000..2b24c8296406
--- /dev/null
+++ b/test/OpenMP/target_map_codegen.cpp
@@ -0,0 +1,1015 @@
+// expected-no-diagnostics
+#ifndef HEADER
+#define HEADER
+
+///
+/// Implicit maps.
+///
+
+///==========================================================================///
+// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK1 --check-prefix CK1-64
+// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK1 --check-prefix CK1-64
+// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK1 --check-prefix CK1-32
+// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK1 --check-prefix CK1-32
+#ifdef CK1
+
+// CK1-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK1-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK1-LABEL: implicit_maps_integer
+void implicit_maps_integer (int a){
+  int i = a;
+
+  // CK1-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK1-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK1-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK1-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK1-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK1-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK1-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK1-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK1-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK1-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK1: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  #pragma omp target
+  {
+   ++i;
+  }
+}
+
+// CK1: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK1: [[ADDR:%.+]] = alloca i[[sz]],
+// CK1: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK1-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK1-64: {{.+}} = load i32, i32* [[CADDR]],
+// CK1-32: {{.+}} = load i32, i32* [[ADDR]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK2 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK2 --check-prefix CK2-64
+// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK2 --check-prefix CK2-64
+// RUN: %clang_cc1 -DCK2 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK2 --check-prefix CK2-32
+// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK2 --check-prefix CK2-32
+#ifdef CK2
+
+// CK2-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK2-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK2-LABEL: implicit_maps_integer_reference
+void implicit_maps_integer_reference (int a){
+  int &i = a;
+  // CK2-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK2-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK2-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK2-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK2-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK2-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK2-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK2-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK2-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK2-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK2: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  #pragma omp target
+  {
+   ++i;
+  }
+}
+
+// CK2: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK2: [[ADDR:%.+]] = alloca i[[sz]],
+// CK2: [[REF:%.+]] = alloca i32*,
+// CK2: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK2-64: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+// CK2-64: store i32* [[CADDR]], i32** [[REF]],
+// CK2-64: [[RVAL:%.+]] = load i32*, i32** [[REF]],
+// CK2-64: {{.+}} = load i32, i32* [[RVAL]],
+// CK2-32: store i32* [[ADDR]], i32** [[REF]],
+// CK2-32: [[RVAL:%.+]] = load i32*, i32** [[REF]],
+// CK2-32: {{.+}} = load i32, i32* [[RVAL]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK3 --check-prefix CK3-64
+// RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK3 --check-prefix CK3-64
+// RUN: %clang_cc1 -DCK3 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK3 --check-prefix CK3-32
+// RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK3 --check-prefix CK3-32
+#ifdef CK3
+
+// CK3-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK3-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK3-LABEL: implicit_maps_parameter
+void implicit_maps_parameter (int a){
+
+  // CK3-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK3-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK3-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK3-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK3-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK3-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK3-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK3-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK3-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK3-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK3-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK3-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK3: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  #pragma omp target
+  {
+   ++a;
+  }
+}
+
+// CK3: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK3: [[ADDR:%.+]] = alloca i[[sz]],
+// CK3: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK3-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK3-64: {{.+}} = load i32, i32* [[CADDR]],
+// CK3-32: {{.+}} = load i32, i32* [[ADDR]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK4 --check-prefix CK4-64
+// RUN: %clang_cc1 -DCK4 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK4 --check-prefix CK4-64
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK4 --check-prefix CK4-32
+// RUN: %clang_cc1 -DCK4 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK4 --check-prefix CK4-32
+#ifdef CK4
+
+// CK4-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK4-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK4-LABEL: implicit_maps_nested_integer
+void implicit_maps_nested_integer (int a){
+  int i = a;
+
+  // The captures in parallel are by reference. Only the capture in target is by
+  // copy.
+
+  // CK4: call void {{.+}}@__kmpc_fork_call({{.+}} [[KERNELP1:@.+]] to void (i32*, i32*, ...)*), i32* {{.+}})
+  // CK4: define internal void [[KERNELP1]](i32* {{[^,]+}}, i32* {{[^,]+}}, i32* {{[^,]+}})
+  #pragma omp parallel
+  {
+    // CK4-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+    // CK4-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+    // CK4-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+    // CK4-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+    // CK4-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+    // CK4-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+    // CK4-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+    // CK4-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+    // CK4-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+    // CK4-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+    // CK4-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+    // CK4-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+    // CK4: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+    #pragma omp target
+    {
+      #pragma omp parallel
+      {
+        ++i;
+      }
+    }
+  }
+}
+
+// CK4: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK4: [[ADDR:%.+]] = alloca i[[sz]],
+// CK4: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK4-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK4-64: call void {{.+}}@__kmpc_fork_call({{.+}} [[KERNELP2:@.+]] to void (i32*, i32*, ...)*), i32* [[CADDR]])
+// CK4-32: call void {{.+}}@__kmpc_fork_call({{.+}} [[KERNELP2:@.+]] to void (i32*, i32*, ...)*), i32* [[ADDR]])
+// CK4: define internal void [[KERNELP2]](i32* {{[^,]+}}, i32* {{[^,]+}}, i32* {{[^,]+}})
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK5 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK5 --check-prefix CK5-64
+// RUN: %clang_cc1 -DCK5 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK5 --check-prefix CK5-64
+// RUN: %clang_cc1 -DCK5 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK5 --check-prefix CK5-32
+// RUN: %clang_cc1 -DCK5 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK5 --check-prefix CK5-32
+#ifdef CK5
+
+// CK5-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK5-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK5-LABEL: implicit_maps_nested_integer_and_enum
+void implicit_maps_nested_integer_and_enum (int a){
+  enum Bla {
+    SomeEnum = 0x09
+  };
+
+  // Using an enum should not change the mapping information.
+  int  i = a;
+
+  // CK5-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK5-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK5-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK5-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK5-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK5-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK5-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK5-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK5-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK5-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK5-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK5-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK5: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  #pragma omp target
+  {
+    ++i;
+    i += SomeEnum;
+  }
+}
+
+// CK5: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK5: [[ADDR:%.+]] = alloca i[[sz]],
+// CK5: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK5-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK5-64: {{.+}} = load i32, i32* [[CADDR]],
+// CK5-32: {{.+}} = load i32, i32* [[ADDR]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK6 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK6 --check-prefix CK6-64
+// RUN: %clang_cc1 -DCK6 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK6 --check-prefix CK6-64
+// RUN: %clang_cc1 -DCK6 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK6 --check-prefix CK6-32
+// RUN: %clang_cc1 -DCK6 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK6 --check-prefix CK6-32
+#ifdef CK6
+// CK6-DAG: [[GBL:@Gi]] = global i32 0
+// CK6-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK6-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK6-LABEL: implicit_maps_host_global
+int Gi;
+void implicit_maps_host_global (int a){
+  // CK6-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK6-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK6-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK6-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK6-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK6-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK6-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK6-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK6-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK6-64-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK6-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK6-64-DAG: store i32 [[GBLVAL:%.+]], i32* [[CADDR]],
+  // CK6-64-DAG: [[GBLVAL]] = load i32, i32* [[GBL]],
+  // CK6-32-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[GBLVAL:%.+]],
+
+  // CK6: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  #pragma omp target
+  {
+    ++Gi;
+  }
+}
+
+// CK6: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK6: [[ADDR:%.+]] = alloca i[[sz]],
+// CK6: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK6-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK6-64: {{.+}} = load i32, i32* [[CADDR]],
+// CK6-32: {{.+}} = load i32, i32* [[ADDR]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK7 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK7 --check-prefix CK7-64
+// RUN: %clang_cc1 -DCK7 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK7  --check-prefix CK7-64
+// RUN: %clang_cc1 -DCK7 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK7  --check-prefix CK7-32
+// RUN: %clang_cc1 -DCK7 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK7  --check-prefix CK7-32
+#ifdef CK7
+
+// For a 32-bit targets, the value doesn't fit the size of the pointer,
+// therefore it is passed by reference with a map 'to' specification.
+
+// CK7-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 8]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK7-64-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+// Map types: OMP_MAP_TO = 1
+// CK7-32-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 1]
+
+// CK7-LABEL: implicit_maps_double
+void implicit_maps_double (int a){
+  double d = (double)a;
+
+  // CK7-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK7-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK7-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK7-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK7-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+
+  // CK7-64-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK7-64-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK7-64-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK7-64-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK7-64-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK7-64-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to double*
+  // CK7-64-64-DAG: store double {{.+}}, double* [[CADDR]],
+
+  // CK7-32-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK7-32-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK7-32-DAG: [[VALBP]] = bitcast double* [[DECL:%.+]] to i8*
+  // CK7-32-DAG: [[VALP]] = bitcast double* [[DECL]] to i8*
+
+  // CK7-64: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  // CK7-32: call void [[KERNEL:@.+]](double* [[DECL]])
+  #pragma omp target
+  {
+    d += 1.0;
+  }
+}
+
+// CK7-64: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK7-64: [[ADDR:%.+]] = alloca i[[sz]],
+// CK7-64: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK7-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to double*
+// CK7-64: {{.+}} = load double, double* [[CADDR]],
+
+// CK7-32: define internal void [[KERNEL]](double* {{.+}}[[ARG:%.+]])
+// CK7-32: [[ADDR:%.+]] = alloca double*,
+// CK7-32: store double* [[ARG]], double** [[ADDR]],
+// CK7-32: [[REF:%.+]] = load double*, double** [[ADDR]],
+// CK7-32: {{.+}} = load double, double* [[REF]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK8 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK8
+// RUN: %clang_cc1 -DCK8 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK8
+// RUN: %clang_cc1 -DCK8 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK8
+// RUN: %clang_cc1 -DCK8 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK8
+#ifdef CK8
+
+// CK8-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK8-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+// CK8-LABEL: implicit_maps_float
+void implicit_maps_float (int a){
+  float f = (float)a;
+
+  // CK8-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK8-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK8-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK8-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK8-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK8-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK8-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK8-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK8-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK8-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK8-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to float*
+  // CK8-DAG: store float {{.+}}, float* [[CADDR]],
+
+  // CK8: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  #pragma omp target
+  {
+    f += 1.0;
+  }
+}
+
+// CK8: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK8: [[ADDR:%.+]] = alloca i[[sz]],
+// CK8: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK8: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to float*
+// CK8: {{.+}} = load float, float* [[CADDR]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK9 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK9
+// RUN: %clang_cc1 -DCK9 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK9
+// RUN: %clang_cc1 -DCK9 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK9
+// RUN: %clang_cc1 -DCK9 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK9
+#ifdef CK9
+
+// CK9-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 16]
+// Map types: OMP_MAP_TO + OMP_MAP_FROM = 2 + 1
+// CK9-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 3]
+
+// CK9-LABEL: implicit_maps_array
+void implicit_maps_array (int a){
+  double darr[2] = {(double)a, (double)a};
+
+  // CK9-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK9-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK9-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK9-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK9-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK9-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK9-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK9-DAG: [[VALBP]] = bitcast [2 x double]* [[DECL:%.+]] to i8*
+  // CK9-DAG: [[VALP]] = bitcast [2 x double]* [[DECL]] to i8*
+
+  // CK9: call void [[KERNEL:@.+]]([2 x double]* [[DECL]])
+  #pragma omp target
+  {
+    darr[0] += 1.0;
+    darr[1] += 1.0;
+  }
+}
+
+// CK9: define internal void [[KERNEL]]([2 x double]* {{.+}}[[ARG:%.+]])
+// CK9: [[ADDR:%.+]] = alloca [2 x double]*,
+// CK9: store [2 x double]* [[ARG]], [2 x double]** [[ADDR]],
+// CK9: [[REF:%.+]] = load [2 x double]*, [2 x double]** [[ADDR]],
+// CK9: {{.+}} = getelementptr inbounds [2 x double], [2 x double]* [[REF]], i[[sz]] 0, i[[sz]] 0
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK10 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK10
+// RUN: %clang_cc1 -DCK10 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK10
+// RUN: %clang_cc1 -DCK10 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK10
+// RUN: %clang_cc1 -DCK10 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK10
+#ifdef CK10
+
+// CK10-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} {{8|4}}]
+// Map types: OMP_MAP_BYCOPY | OMP_MAP_PTR = 128 + 32
+// CK10-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 160]
+
+// CK10-LABEL: implicit_maps_pointer
+void implicit_maps_pointer (){
+  double *ddyn;
+
+  // CK10-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK10-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK10-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK10-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK10-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK10-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK10-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK10-DAG: [[VALBP]] = bitcast double* [[PTR:%.+]] to i8*
+  // CK10-DAG: [[VALP]] = bitcast double* [[PTR]] to i8*
+
+  // CK10: call void [[KERNEL:@.+]](double* [[PTR]])
+  #pragma omp target
+  {
+    ddyn[0] += 1.0;
+    ddyn[1] += 1.0;
+  }
+}
+
+// CK10: define internal void [[KERNEL]](double* {{.*}}[[ARG:%.+]])
+// CK10: [[ADDR:%.+]] = alloca double*,
+// CK10: store double* [[ARG]], double** [[ADDR]],
+// CK10: [[REF:%.+]] = load double*, double** [[ADDR]],
+// CK10: {{.+}} = getelementptr inbounds double, double* [[REF]], i[[sz]] 0
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK11 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK11
+// RUN: %clang_cc1 -DCK11 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK11
+// RUN: %clang_cc1 -DCK11 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK11
+// RUN: %clang_cc1 -DCK11 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK11
+#ifdef CK11
+
+// CK11-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 16]
+// Map types: OMP_MAP_TO = 1
+// CK11-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 1]
+
+// CK11-LABEL: implicit_maps_double_complex
+void implicit_maps_double_complex (int a){
+  double _Complex dc = (double)a;
+
+  // CK11-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK11-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK11-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK11-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK11-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK11-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK11-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK11-DAG: [[VALBP]] = bitcast { double, double }* [[PTR:%.+]] to i8*
+  // CK11-DAG: [[VALP]] = bitcast { double, double }* [[PTR]] to i8*
+
+  // CK11: call void [[KERNEL:@.+]]({ double, double }* [[PTR]])
+  #pragma omp target
+  {
+   dc *= dc;
+  }
+}
+
+// CK11: define internal void [[KERNEL]]({ double, double }* {{.*}}[[ARG:%.+]])
+// CK11: [[ADDR:%.+]] = alloca { double, double }*,
+// CK11: store { double, double }* [[ARG]], { double, double }** [[ADDR]],
+// CK11: [[REF:%.+]] = load { double, double }*, { double, double }** [[ADDR]],
+// CK11: {{.+}} = getelementptr inbounds { double, double }, { double, double }* [[REF]], i32 0, i32 0
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK12 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK12 --check-prefix CK12-64
+// RUN: %clang_cc1 -DCK12 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK12 --check-prefix CK12-64
+// RUN: %clang_cc1 -DCK12 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK12 --check-prefix CK12-32
+// RUN: %clang_cc1 -DCK12 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK12 --check-prefix CK12-32
+#ifdef CK12
+
+// For a 32-bit targets, the value doesn't fit the size of the pointer,
+// therefore it is passed by reference with a map 'to' specification.
+
+// CK12-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 8]
+// Map types: OMP_MAP_BYCOPY = 128
+// CK12-64-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+// Map types: OMP_MAP_TO = 1
+// CK12-32-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 1]
+
+// CK12-LABEL: implicit_maps_float_complex
+void implicit_maps_float_complex (int a){
+  float _Complex fc = (float)a;
+
+  // CK12-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK12-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK12-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK12-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK12-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+
+  // CK12-64-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK12-64-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK12-64-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK12-64-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK12-64-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK12-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to { float, float }*
+  // CK12-64-DAG: store { float, float } {{.+}}, { float, float }* [[CADDR]],
+
+  // CK12-32-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK12-32-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK12-32-DAG: [[VALBP]] = bitcast { float, float }* [[DECL:%.+]] to i8*
+  // CK12-32-DAG: [[VALP]] = bitcast { float, float }* [[DECL]] to i8*
+
+  // CK12-64: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  // CK12-32: call void [[KERNEL:@.+]]({ float, float }* [[DECL]])
+  #pragma omp target
+  {
+    fc *= fc;
+  }
+}
+
+// CK12-64: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK12-64: [[ADDR:%.+]] = alloca i[[sz]],
+// CK12-64: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK12-64: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to { float, float }*
+// CK12-64: {{.+}} = getelementptr inbounds { float, float }, { float, float }* [[CADDR]], i32 0, i32 0
+
+// CK12-32: define internal void [[KERNEL]]({ float, float }* {{.+}}[[ARG:%.+]])
+// CK12-32: [[ADDR:%.+]] = alloca { float, float }*,
+// CK12-32: store { float, float }* [[ARG]], { float, float }** [[ADDR]],
+// CK12-32: [[REF:%.+]] = load { float, float }*, { float, float }** [[ADDR]],
+// CK12-32: {{.+}} = getelementptr inbounds { float, float }, { float, float }* [[REF]], i32 0, i32 0
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK13 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK13
+// RUN: %clang_cc1 -DCK13 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK13
+// RUN: %clang_cc1 -DCK13 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK13
+// RUN: %clang_cc1 -DCK13 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK13
+#ifdef CK13
+
+// We don't have a constant map size for VLAs.
+// Map types:
+//  - OMP_MAP_BYCOPY = 128 (vla size)
+//  - OMP_MAP_BYCOPY = 128 (vla size)
+//  - OMP_MAP_TO + OMP_MAP_FROM = 2 + 1
+// CK13-DAG: [[TYPES:@.+]] = {{.+}}constant [3 x i32] [i32 128, i32 128, i32 3]
+
+// CK13-LABEL: implicit_maps_variable_length_array
+void implicit_maps_variable_length_array (int a){
+  double vla[2][a];
+
+  // CK13-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 3, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], i[[sz:64|32]]* [[SGEP:%[^,]+]], {{.+}}[[TYPES]]{{.+}})
+  // CK13-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK13-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK13-DAG: [[SGEP]] = getelementptr inbounds {{.+}}[[SS:%[^,]+]], i32 0, i32 0
+
+  // CK13-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK13-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK13-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 0
+  // CK13-DAG: store i8* inttoptr (i[[sz]] 2 to i8*), i8** [[BP0]],
+  // CK13-DAG: store i8* inttoptr (i[[sz]] 2 to i8*), i8** [[P0]],
+  // CK13-DAG: store i[[sz]] {{8|4}}, i[[sz]]* [[S0]],
+
+  // CK13-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
+  // CK13-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
+  // CK13-DAG: [[S1:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 1
+  // CK13-DAG: store i8* [[VALBP1:%.+]], i8** [[BP1]],
+  // CK13-DAG: store i8* [[VALP1:%.+]], i8** [[P1]],
+  // CK13-DAG: [[VALBP1]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK13-DAG: [[VALP1]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK13-DAG: store i[[sz]] {{8|4}}, i[[sz]]* [[S1]],
+
+  // CK13-DAG: [[BP2:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 2
+  // CK13-DAG: [[P2:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 2
+  // CK13-DAG: [[S2:%.+]] = getelementptr inbounds {{.+}}[[SS]], i32 0, i32 2
+  // CK13-DAG: store i8* [[VALBP2:%.+]], i8** [[BP2]],
+  // CK13-DAG: store i8* [[VALP2:%.+]], i8** [[P2]],
+  // CK13-DAG: store i[[sz]] [[VALS2:%.+]], i[[sz]]* [[S2]],
+  // CK13-DAG: [[VALBP2]] = bitcast double* [[DECL:%.+]] to i8*
+  // CK13-DAG: [[VALP2]] = bitcast double* [[DECL]] to i8*
+  // CK13-DAG: [[VALS2]] = mul nuw i[[sz]] %{{.+}}, 8
+
+  // CK13: call void [[KERNEL:@.+]](i[[sz]] {{.+}}, i[[sz]] {{.+}}, double* [[DECL]])
+  #pragma omp target
+  {
+    vla[1][3] += 1.0;
+  }
+}
+
+// CK13: define internal void [[KERNEL]](i[[sz]] [[VLA0:%.+]], i[[sz]] [[VLA1:%.+]], double* {{.+}}[[ARG:%.+]])
+// CK13: [[ADDR0:%.+]] = alloca i[[sz]],
+// CK13: [[ADDR1:%.+]] = alloca i[[sz]],
+// CK13: [[ADDR2:%.+]] = alloca double*,
+// CK13: store i[[sz]] [[VLA0]], i[[sz]]* [[ADDR0]],
+// CK13: store i[[sz]] [[VLA1]], i[[sz]]* [[ADDR1]],
+// CK13: store double* [[ARG]], double** [[ADDR2]],
+// CK13: {{.+}} = load i[[sz]],  i[[sz]]* [[ADDR0]],
+// CK13: {{.+}} = load i[[sz]],  i[[sz]]* [[ADDR1]],
+// CK13: [[REF:%.+]] = load double*, double** [[ADDR2]],
+// CK13: {{.+}} = getelementptr inbounds double, double* [[REF]], i[[sz]] %{{.+}}
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK14 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK14 --check-prefix CK14-64
+// RUN: %clang_cc1 -DCK14 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK14 --check-prefix CK14-64
+// RUN: %clang_cc1 -DCK14 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK14 --check-prefix CK14-32
+// RUN: %clang_cc1 -DCK14 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK14 --check-prefix CK14-32
+#ifdef CK14
+
+// CK14-DAG: [[ST:%.+]] = type { i32, double }
+// CK14-DAG: [[SIZES:@.+]] = {{.+}}constant [2 x i[[sz:64|32]]] [i{{64|32}} {{16|12}}, i{{64|32}} 4]
+// Map types:
+// - OMP_MAP_TO | OMP_MAP_FROM = 1 + 2
+// - OMP_MAP_BYCOPY = 128
+// CK14-DAG: [[TYPES:@.+]] = {{.+}}constant [2 x i32] [i32 3, i32 128]
+
+class SSS {
+public:
+  int a;
+  double b;
+
+  void foo(int c) {
+    #pragma omp target
+    {
+      a += c;
+      b += (double)c;
+    }
+  }
+
+  SSS(int a, double b) : a(a), b(b) {}
+};
+
+// CK14-LABEL: implicit_maps_class
+void implicit_maps_class (int a){
+  SSS sss(a, (double)a);
+
+  // CK14: define {{.*}}void @{{.+}}foo{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}})
+  // CK14-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 2, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK14-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK14-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+
+  // CK14-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK14-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK14-DAG: store i8* [[VALBP0:%.+]], i8** [[BP0]],
+  // CK14-DAG: store i8* [[VALP0:%.+]], i8** [[P0]],
+  // CK14-DAG: [[VALBP0]] = bitcast [[ST]]* [[DECL:%.+]] to i8*
+  // CK14-DAG: [[VALP0]] = bitcast [[ST]]* [[DECL]] to i8*
+
+  // CK14-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
+  // CK14-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
+  // CK14-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK14-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK14-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK14-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK14-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK14-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK14-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK14: call void [[KERNEL:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}})
+  sss.foo(123);
+}
+
+// CK14: define internal void [[KERNEL]]([[ST]]* [[THIS:%.+]], i[[sz]] [[ARG:%.+]])
+// CK14: [[ADDR0:%.+]] = alloca [[ST]]*,
+// CK14: [[ADDR1:%.+]] = alloca i[[sz]],
+// CK14: store [[ST]]* [[THIS]], [[ST]]** [[ADDR0]],
+// CK14: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR1]],
+// CK14: [[REF0:%.+]] = load [[ST]]*, [[ST]]** [[ADDR0]],
+// CK14-64: [[CADDR1:%.+]] = bitcast i[[sz]]* [[ADDR1]] to i32*
+// CK14-64: {{.+}} = load i32,  i32* [[CADDR1]],
+// CK14-32: {{.+}} = load i32, i32* [[ADDR1]],
+// CK14: {{.+}} = getelementptr inbounds [[ST]], [[ST]]* [[REF0]], i32 0, i32 0
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK15 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK15 --check-prefix CK15-64
+// RUN: %clang_cc1 -DCK15 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK15 --check-prefix CK15-64
+// RUN: %clang_cc1 -DCK15 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK15 --check-prefix CK15-32
+// RUN: %clang_cc1 -DCK15 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK15 --check-prefix CK15-32
+#ifdef CK15
+
+// CK15: [[ST:%.+]] = type { i32, double, i32* }
+// CK15: [[SIZES:@.+]] = {{.+}}constant [2 x i[[sz:64|32]]] [i{{64|32}} {{24|16}}, i{{64|32}} 4]
+// Map types:
+// - OMP_MAP_TO | OMP_MAP_FROM = 1 + 2
+// - OMP_MAP_BYCOPY = 128
+// CK15: [[TYPES:@.+]] = {{.+}}constant [2 x i32] [i32 3, i32 128]
+
+// CK15: [[SIZES2:@.+]] = {{.+}}constant [2 x i[[sz]]] [i{{64|32}} {{24|16}}, i{{64|32}} 4]
+// Map types:
+// - OMP_MAP_TO | OMP_MAP_FROM = 1 + 2
+// - OMP_MAP_BYCOPY = 128
+// CK15: [[TYPES2:@.+]] = {{.+}}constant [2 x i32] [i32 3, i32 128]
+
+template<int x>
+class SSST {
+public:
+  int a;
+  double b;
+  int &r;
+
+  void foo(int c) {
+    #pragma omp target
+    {
+      a += c + x;
+      b += (double)(c + x);
+      r += x;
+    }
+  }
+  template<int y>
+  void bar(int c) {
+    #pragma omp target
+    {
+      a += c + x + y;
+      b += (double)(c + x + y);
+      r += x + y;
+    }
+  }
+
+  SSST(int a, double b, int &r) : a(a), b(b), r(r) {}
+};
+
+// CK15-LABEL: implicit_maps_templated_class
+void implicit_maps_templated_class (int a){
+  SSST<123> ssst(a, (double)a, a);
+
+  // CK15: define {{.*}}void @{{.+}}foo{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}})
+  // CK15-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 2, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+
+  // CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK15-DAG: store i8* [[VALBP0:%.+]], i8** [[BP0]],
+  // CK15-DAG: store i8* [[VALP0:%.+]], i8** [[P0]],
+  // CK15-DAG: [[VALBP0]] = bitcast [[ST]]* [[DECL:%.+]] to i8*
+  // CK15-DAG: [[VALP0]] = bitcast [[ST]]* [[DECL]] to i8*
+
+  // CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
+  // CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
+  // CK15-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK15-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK15-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK15-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK15-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK15-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK15-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK15: call void [[KERNEL:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}})
+  ssst.foo(456);
+
+  // CK15: define {{.*}}void @{{.+}}bar{{.+}}([[ST]]* {{[^,]+}}, i32 {{[^,]+}})
+  // CK15-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 2, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES2]]{{.+}}, {{.+}}[[TYPES2]]{{.+}})
+  // CK15-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK15-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+
+  // CK15-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK15-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK15-DAG: store i8* [[VALBP0:%.+]], i8** [[BP0]],
+  // CK15-DAG: store i8* [[VALP0:%.+]], i8** [[P0]],
+  // CK15-DAG: [[VALBP0]] = bitcast [[ST]]* [[DECL:%.+]] to i8*
+  // CK15-DAG: [[VALP0]] = bitcast [[ST]]* [[DECL]] to i8*
+
+  // CK15-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 1
+  // CK15-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 1
+  // CK15-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK15-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK15-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK15-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK15-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK15-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK15-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK15: call void [[KERNEL2:@.+]]([[ST]]* [[DECL]], i[[sz]] {{.+}})
+  ssst.bar<210>(789);
+}
+
+// CK15: define internal void [[KERNEL]]([[ST]]* [[THIS:%.+]], i[[sz]] [[ARG:%.+]])
+// CK15: [[ADDR0:%.+]] = alloca [[ST]]*,
+// CK15: [[ADDR1:%.+]] = alloca i[[sz]],
+// CK15: store [[ST]]* [[THIS]], [[ST]]** [[ADDR0]],
+// CK15: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR1]],
+// CK15: [[REF0:%.+]] = load [[ST]]*, [[ST]]** [[ADDR0]],
+// CK15-64: [[CADDR1:%.+]] = bitcast i[[sz]]* [[ADDR1]] to i32*
+// CK15-64: {{.+}} = load i32,  i32* [[CADDR1]],
+// CK15-32: {{.+}} = load i32, i32* [[ADDR1]],
+// CK15: {{.+}} = getelementptr inbounds [[ST]], [[ST]]* [[REF0]], i32 0, i32 0
+
+// CK15: define internal void [[KERNEL2]]([[ST]]* [[THIS:%.+]], i[[sz]] [[ARG:%.+]])
+// CK15: [[ADDR0:%.+]] = alloca [[ST]]*,
+// CK15: [[ADDR1:%.+]] = alloca i[[sz]],
+// CK15: store [[ST]]* [[THIS]], [[ST]]** [[ADDR0]],
+// CK15: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR1]],
+// CK15: [[REF0:%.+]] = load [[ST]]*, [[ST]]** [[ADDR0]],
+// CK15-64: [[CADDR1:%.+]] = bitcast i[[sz]]* [[ADDR1]] to i32*
+// CK15-64: {{.+}} = load i32,  i32* [[CADDR1]],
+// CK15-32: {{.+}} = load i32, i32* [[ADDR1]],
+// CK15: {{.+}} = getelementptr inbounds [[ST]], [[ST]]* [[REF0]], i32 0, i32 0
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK16 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK16 --check-prefix CK16-64
+// RUN: %clang_cc1 -DCK16 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK16 --check-prefix CK16-64
+// RUN: %clang_cc1 -DCK16 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK16 --check-prefix CK16-32
+// RUN: %clang_cc1 -DCK16 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK16 --check-prefix CK16-32
+#ifdef CK16
+
+// CK16-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types:
+// - OMP_MAP_BYCOPY = 128
+// CK16-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+template<int y>
+int foo(int d) {
+  int res = d;
+  #pragma omp target
+  {
+    res += y;
+  }
+  return res;
+}
+// CK16-LABEL: implicit_maps_templated_function
+void implicit_maps_templated_function (int a){
+  int i = a;
+
+  // CK16: define {{.*}}i32 @{{.+}}foo{{.+}}(i32 {{[^,]+}})
+  // CK16-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK16-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK16-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+
+  // CK16-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK16-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK16-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK16-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK16-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK16-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK16-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK16-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK16-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK16: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  i = foo<543>(i);
+}
+// CK16: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK16: [[ADDR:%.+]] = alloca i[[sz]],
+// CK16: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK16-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK16-64: {{.+}} = load i32, i32* [[CADDR]],
+// CK16-32: {{.+}} = load i32, i32* [[ADDR]],
+
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK17 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK17
+// RUN: %clang_cc1 -DCK17 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK17
+// RUN: %clang_cc1 -DCK17 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK17
+// RUN: %clang_cc1 -DCK17 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK17
+#ifdef CK17
+
+// CK17-DAG: [[ST:%.+]] = type { i32, double }
+// CK17-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} {{16|12}}]
+// Map types: OMP_MAP_TO + OMP_MAP_FROM = 2 + 1
+// CK17-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 3]
+
+class SSS {
+public:
+  int a;
+  double b;
+};
+
+// CK17-LABEL: implicit_maps_struct
+void implicit_maps_struct (int a){
+  SSS s = {a, (double)a};
+
+  // CK17-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK17-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK17-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+  // CK17-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK17-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK17-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK17-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK17-DAG: [[VALBP]] = bitcast [[ST]]* [[DECL:%.+]] to i8*
+  // CK17-DAG: [[VALP]] = bitcast [[ST]]* [[DECL]] to i8*
+
+  // CK17: call void [[KERNEL:@.+]]([[ST]]* [[DECL]])
+  #pragma omp target
+  {
+    s.a += 1;
+    s.b += 1.0;
+  }
+}
+
+// CK17: define internal void [[KERNEL]]([[ST]]* {{.+}}[[ARG:%.+]])
+// CK17: [[ADDR:%.+]] = alloca [[ST]]*,
+// CK17: store [[ST]]* [[ARG]], [[ST]]** [[ADDR]],
+// CK17: [[REF:%.+]] = load [[ST]]*, [[ST]]** [[ADDR]],
+// CK17: {{.+}} = getelementptr inbounds [[ST]], [[ST]]* [[REF]], i32 0, i32 0
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK18 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK18 --check-prefix CK18-64
+// RUN: %clang_cc1 -DCK18 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK18 --check-prefix CK18-64
+// RUN: %clang_cc1 -DCK18 -verify -fopenmp -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK18 --check-prefix CK18-32
+// RUN: %clang_cc1 -DCK18 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK18 --check-prefix CK18-32
+#ifdef CK18
+
+// CK18-DAG: [[SIZES:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 4]
+// Map types:
+// - OMP_MAP_BYCOPY = 128
+// CK18-DAG: [[TYPES:@.+]] = {{.+}}constant [1 x i32] [i32 128]
+
+template<typename T>
+int foo(T d) {
+  #pragma omp target
+  {
+    d += (T)1;
+  }
+  return d;
+}
+// CK18-LABEL: implicit_maps_template_type_capture
+void implicit_maps_template_type_capture (int a){
+  int i = a;
+
+  // CK18: define {{.*}}i32 @{{.+}}foo{{.+}}(i32 {{[^,]+}})
+  // CK18-DAG: call i32 @__tgt_target(i32 {{.+}}, i8* {{.+}}, i32 1, i8** [[BPGEP:%[0-9]+]], i8** [[PGEP:%[0-9]+]], {{.+}}[[SIZES]]{{.+}}, {{.+}}[[TYPES]]{{.+}})
+  // CK18-DAG: [[BPGEP]] = getelementptr inbounds {{.+}}[[BPS:%[^,]+]], i32 0, i32 0
+  // CK18-DAG: [[PGEP]] = getelementptr inbounds {{.+}}[[PS:%[^,]+]], i32 0, i32 0
+
+  // CK18-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BPS]], i32 0, i32 0
+  // CK18-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[PS]], i32 0, i32 0
+  // CK18-DAG: store i8* [[VALBP:%.+]], i8** [[BP1]],
+  // CK18-DAG: store i8* [[VALP:%.+]], i8** [[P1]],
+  // CK18-DAG: [[VALBP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK18-DAG: [[VALP]] = inttoptr i[[sz]] [[VAL:%.+]] to i8*
+  // CK18-DAG: [[VAL]] = load i[[sz]], i[[sz]]* [[ADDR:%.+]],
+  // CK18-64-DAG: [[CADDR:%.+]] = bitcast i[[sz]]* [[ADDR]] to i32*
+  // CK18-64-DAG: store i32 {{.+}}, i32* [[CADDR]],
+
+  // CK18: call void [[KERNEL:@.+]](i[[sz]] [[VAL]])
+  i = foo(i);
+}
+// CK18: define internal void [[KERNEL]](i[[sz]] [[ARG:%.+]])
+// CK18: [[ADDR:%.+]] = alloca i[[sz]],
+// CK18: store i[[sz]] [[ARG]], i[[sz]]* [[ADDR]],
+// CK18-64: [[CADDR:%.+]] = bitcast i64* [[ADDR]] to i32*
+// CK18-64: {{.+}} = load i32, i32* [[CADDR]],
+// CK18-32: {{.+}} = load i32, i32* [[ADDR]],
+
+#endif
+#endif
diff --git a/test/OpenMP/target_map_messages.cpp b/test/OpenMP/target_map_messages.cpp
new file mode 100644
index 000000000000..d61e766c6b4b
--- /dev/null
+++ b/test/OpenMP/target_map_messages.cpp
@@ -0,0 +1,207 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}}
+extern S1 a;
+class S2 {
+  mutable int a;
+public:
+  S2():a(0) { }
+  S2(S2 &s2):a(s2.a) { }
+  static float S2s; // expected-note 4 {{mappable type cannot contain static members}}
+  static const float S2sc; // expected-note 4 {{mappable type cannot contain static members}}
+};
+const float S2::S2sc = 0;
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+public:
+  S3():a(0) { }
+  S3(S3 &s3):a(s3.a) { }
+};
+const S3 c;
+const S3 ca[5];
+extern const int f;
+class S4 {
+  int a;
+  S4();
+  S4(const S4 &s4);
+public:
+  S4(int v):a(v) { }
+};
+class S5 {
+  int a;
+  S5():a(0) {}
+  S5(const S5 &s5):a(s5.a) { }
+public:
+  S5(int v):a(v) { }
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+typedef int from;
+
+template <typename T, int I> // expected-note {{declared here}}
+T tmain(T argc) {
+  const T d = 5;
+  const T da[5] = { 0 };
+  S4 e(4);
+  S5 g(5);
+  T i, t[20];
+  T &j = i;
+  T *k = &j;
+  T x;
+  T y;
+  T to, tofrom, always;
+  const T (&l)[5] = da;
+
+
+#pragma omp target map // expected-error {{expected '(' after 'map'}}
+#pragma omp target map( // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected expression}}
+#pragma omp target map() // expected-error {{expected expression}}
+#pragma omp target map(alloc) // expected-error {{use of undeclared identifier 'alloc'}}
+#pragma omp target map(to argc // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected ',' or ')' in 'map' clause}}
+#pragma omp target map(to:) // expected-error {{expected expression}}
+#pragma omp target map(from: argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+#pragma omp target map(x: y) // expected-error {{incorrect map type, expected one of 'to', 'from', 'tofrom', 'alloc', 'release', or 'delete'}}
+#pragma omp target map(x)
+  foo();
+#pragma omp target map(tofrom: t[:I])
+  foo();
+#pragma omp target map(T: a) // expected-error {{incorrect map type, expected one of 'to', 'from', 'tofrom', 'alloc', 'release', or 'delete'}}
+  foo();
+#pragma omp target map(T) // expected-error {{'T' does not refer to a value}}
+  foo();
+#pragma omp target map(I) // expected-error 2 {{expected variable name, array element or array section}}
+  foo();
+#pragma omp target map(S2::S2s)
+  foo();
+#pragma omp target map(S2::S2sc)
+  foo();
+#pragma omp target map(x)
+  foo();
+#pragma omp target map(to: x)
+  foo();
+#pragma omp target map(to: to)
+  foo();
+#pragma omp target map(to)
+  foo();
+#pragma omp target map(to, x)
+  foo();
+#pragma omp target map(to x) // expected-error {{expected ',' or ')' in 'map' clause}}
+#pragma omp target map(tofrom: argc > 0 ? x : y) // expected-error 2 {{expected variable name, array element or array section}} 
+#pragma omp target map(argc)
+#pragma omp target map(S1) // expected-error {{'S1' does not refer to a value}}
+#pragma omp target map(a, b, c, d, f) // expected-error {{incomplete type 'S1' where a complete type is required}} expected-error 2 {{type 'S2' is not mappable to target}}
+#pragma omp target map(ba) // expected-error 2 {{type 'S2' is not mappable to target}}
+#pragma omp target map(ca)
+#pragma omp target map(da)
+#pragma omp target map(S2::S2s)
+#pragma omp target map(S2::S2sc)
+#pragma omp target map(e, g)
+#pragma omp target map(h) // expected-error {{threadprivate variables are not allowed in map clause}}
+#pragma omp target map(k), map(k) // expected-error 2 {{variable already marked as mapped in current construct}} expected-note 2 {{used here}}
+#pragma omp target map(k), map(k[:5]) // expected-error 2 {{variable already marked as mapped in current construct}} expected-note 2 {{used here}}
+  foo();
+#pragma omp target map(da)
+#pragma omp target map(da[:4])
+  foo();
+#pragma omp target map(k, j, l) // expected-note 4 {{used here}}
+#pragma omp target map(k[:4]) // expected-error 2 {{variable already marked as mapped in current construct}}
+#pragma omp target map(j)
+#pragma omp target map(l[:5]) // expected-error 2 {{variable already marked as mapped in current construct}}
+  foo();
+#pragma omp target map(k[:4], j, l[:5]) // expected-note 4 {{used here}}
+#pragma omp target map(k) // expected-error 2 {{variable already marked as mapped in current construct}}
+#pragma omp target map(j)
+#pragma omp target map(l) // expected-error 2 {{variable already marked as mapped in current construct}}
+  foo();
+
+#pragma omp target map(always, tofrom: x)
+#pragma omp target map(always: x) // expected-error {{missing map type}}
+#pragma omp target map(tofrom, always: x) // expected-error {{incorrect map type modifier, expected 'always'}} expected-error {{incorrect map type, expected one of 'to', 'from', 'tofrom', 'alloc', 'release', or 'delete'}}
+#pragma omp target map(always, tofrom: always, tofrom, x)
+#pragma omp target map(tofrom j) // expected-error {{expected ',' or ')' in 'map' clause}}
+  foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  const int d = 5;
+  const int da[5] = { 0 };
+  S4 e(4);
+  S5 g(5);
+  int i;
+  int &j = i;
+  int *k = &j;
+  int x;
+  int y;
+  int to, tofrom, always;
+  const int (&l)[5] = da;
+#pragma omp target map // expected-error {{expected '(' after 'map'}}
+#pragma omp target map( // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected expression}}
+#pragma omp target map() // expected-error {{expected expression}}
+#pragma omp target map(alloc) // expected-error {{use of undeclared identifier 'alloc'}}
+#pragma omp target map(to argc // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{expected ',' or ')' in 'map' clause}}
+#pragma omp target map(to:) // expected-error {{expected expression}}
+#pragma omp target map(from: argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+#pragma omp target map(x: y) // expected-error {{incorrect map type, expected one of 'to', 'from', 'tofrom', 'alloc', 'release', or 'delete'}}
+#pragma omp target map(x)
+  foo();
+#pragma omp target map(to: x)
+  foo();
+#pragma omp target map(to: to)
+  foo();
+#pragma omp target map(to)
+  foo();
+#pragma omp target map(to, x)
+  foo();
+#pragma omp target map(to x) // expected-error {{expected ',' or ')' in 'map' clause}}
+#pragma omp target map(tofrom: argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
+#pragma omp target map(argc)
+#pragma omp target map(S1) // expected-error {{'S1' does not refer to a value}}
+#pragma omp target map(a, b, c, d, f) // expected-error {{incomplete type 'S1' where a complete type is required}} expected-error 2 {{type 'S2' is not mappable to target}}
+#pragma omp target map(argv[1])
+#pragma omp target map(ba) // expected-error 2 {{type 'S2' is not mappable to target}}
+#pragma omp target map(ca)
+#pragma omp target map(da)
+#pragma omp target map(S2::S2s)
+#pragma omp target map(S2::S2sc)
+#pragma omp target map(e, g)
+#pragma omp target map(h) // expected-error {{threadprivate variables are not allowed in map clause}}
+#pragma omp target map(k), map(k) // expected-error {{variable already marked as mapped in current construct}} expected-note {{used here}}
+#pragma omp target map(k), map(k[:5]) // expected-error {{variable already marked as mapped in current construct}} expected-note {{used here}}
+  foo();
+#pragma omp target map(da)
+#pragma omp target map(da[:4])
+  foo();
+#pragma omp target map(k, j, l) // expected-note 2 {{used here}}
+#pragma omp target map(k[:4]) // expected-error {{variable already marked as mapped in current construct}}
+#pragma omp target map(j)
+#pragma omp target map(l[:5]) // expected-error {{variable already marked as mapped in current construct}}
+  foo();
+#pragma omp target map(k[:4], j, l[:5]) // expected-note 2 {{used here}}
+#pragma omp target map(k) // expected-error {{variable already marked as mapped in current construct}}
+#pragma omp target map(j)
+#pragma omp target map(l) // expected-error {{variable already marked as mapped in current construct}}
+  foo();
+
+#pragma omp target map(always, tofrom: x)
+#pragma omp target map(always: x) // expected-error {{missing map type}}
+#pragma omp target map(tofrom, always: x) // expected-error {{incorrect map type modifier, expected 'always'}} expected-error {{incorrect map type, expected one of 'to', 'from', 'tofrom', 'alloc', 'release', or 'delete'}}
+#pragma omp target map(always, tofrom: always, tofrom, x)
+#pragma omp target map(tofrom j) // expected-error {{expected ',' or ')' in 'map' clause}}
+  foo();
+
+  return tmain<int, 3>(argc)+tmain<from, 4>(argc); // expected-note {{in instantiation of function template specialization 'tmain<int, 3>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<int, 4>' requested here}}
+}
+
diff --git a/test/OpenMP/task_ast_print.cpp b/test/OpenMP/task_ast_print.cpp
index 5fd41038ad29..723139b08183 100644
--- a/test/OpenMP/task_ast_print.cpp
+++ b/test/OpenMP/task_ast_print.cpp
@@ -33,11 +33,12 @@ T tmain(T argc, T *argv) {
   T b = argc, c, d, e, f, g;
   static T a;
   S<T> s;
-#pragma omp task untied depend(in : argc)
+  T arr[argc];
+#pragma omp task untied depend(in : argc, argv[b:argc], arr[:]) if (task : argc > 0)
   a = 2;
-#pragma omp task default(none), private(argc, b) firstprivate(argv) shared(d) if (argc > 0) final(S<T>::TS > 0)
+#pragma omp task default(none), private(argc, b) firstprivate(argv) shared(d) if (argc > 0) final(S<T>::TS > 0) priority(argc)
   foo();
-#pragma omp task if (C) mergeable
+#pragma omp task if (C) mergeable priority(C)
   foo();
   return 0;
 }
@@ -46,31 +47,34 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: int b = argc, c, d, e, f, g;
 // CHECK-NEXT: static int a;
 // CHECK-NEXT: S<int> s;
-// CHECK-NEXT: #pragma omp task untied depend(in : argc)
+// CHECK-NEXT: int arr[argc];
+// CHECK-NEXT: #pragma omp task untied depend(in : argc,argv[b:argc],arr[:]) if(task: argc > 0)
 // CHECK-NEXT: a = 2;
-// CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) final(S<int>::TS > 0)
+// CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) final(S<int>::TS > 0) priority(argc)
 // CHECK-NEXT: foo()
-// CHECK-NEXT: #pragma omp task if(5) mergeable
+// CHECK-NEXT: #pragma omp task if(5) mergeable priority(5)
 // CHECK-NEXT: foo()
 // CHECK: template <typename T = long, int C = 1> long tmain(long argc, long *argv) {
 // CHECK-NEXT: long b = argc, c, d, e, f, g;
 // CHECK-NEXT: static long a;
 // CHECK-NEXT: S<long> s;
-// CHECK-NEXT: #pragma omp task untied depend(in : argc)
+// CHECK-NEXT: long arr[argc];
+// CHECK-NEXT: #pragma omp task untied depend(in : argc,argv[b:argc],arr[:]) if(task: argc > 0)
 // CHECK-NEXT: a = 2;
-// CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) final(S<long>::TS > 0)
+// CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) final(S<long>::TS > 0) priority(argc)
 // CHECK-NEXT: foo()
-// CHECK-NEXT: #pragma omp task if(1) mergeable
+// CHECK-NEXT: #pragma omp task if(1) mergeable priority(1)
 // CHECK-NEXT: foo()
 // CHECK: template <typename T, int C> T tmain(T argc, T *argv) {
 // CHECK-NEXT: T b = argc, c, d, e, f, g;
 // CHECK-NEXT: static T a;
 // CHECK-NEXT: S<T> s;
-// CHECK-NEXT: #pragma omp task untied depend(in : argc)
+// CHECK-NEXT: T arr[argc];
+// CHECK-NEXT: #pragma omp task untied depend(in : argc,argv[b:argc],arr[:]) if(task: argc > 0)
 // CHECK-NEXT: a = 2;
-// CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) final(S<T>::TS > 0)
+// CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) shared(d) if(argc > 0) final(S<T>::TS > 0) priority(argc)
 // CHECK-NEXT: foo()
-// CHECK-NEXT: #pragma omp task if(C) mergeable
+// CHECK-NEXT: #pragma omp task if(C) mergeable priority(C)
 // CHECK-NEXT: foo()
 
 enum Enum {};
@@ -79,15 +83,16 @@ int main(int argc, char **argv) {
   long x;
   int b = argc, c, d, e, f, g;
   static int a;
+  int arr[10];
 #pragma omp threadprivate(a)
   Enum ee;
 // CHECK: Enum ee;
-#pragma omp task untied mergeable depend(out:argv[1])
-  // CHECK-NEXT: #pragma omp task untied mergeable depend(out : argv[1])
+#pragma omp task untied mergeable depend(out:argv[:a][1], (arr)[0:]) if(task: argc > 0) priority(f)
+  // CHECK-NEXT: #pragma omp task untied mergeable depend(out : argv[:a][1],(arr)[0:]) if(task: argc > 0) priority(f)
   a = 2;
 // CHECK-NEXT: a = 2;
-#pragma omp task default(none), private(argc, b) firstprivate(argv) if (argc > 0) final(a > 0) depend(inout : a)
-  // CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) if(argc > 0) final(a > 0) depend(inout : a)
+#pragma omp task default(none), private(argc, b) firstprivate(argv) if (argc > 0) final(a > 0) depend(inout : a, argv[:argc],arr[:a]) priority(23)
+  // CHECK-NEXT: #pragma omp task default(none) private(argc,b) firstprivate(argv) if(argc > 0) final(a > 0) depend(inout : a,argv[:argc],arr[:a]) priority(23)
   foo();
   // CHECK-NEXT: foo();
   return tmain<int, 5>(b, &b) + tmain<long, 1>(x, &x);
diff --git a/test/OpenMP/task_codegen.cpp b/test/OpenMP/task_codegen.cpp
index 139ac505a9f6..23dc014aad3b 100644
--- a/test/OpenMP/task_codegen.cpp
+++ b/test/OpenMP/task_codegen.cpp
@@ -18,12 +18,13 @@ struct S {
   ~S() {}
 };
 int a;
-// CHECK-LABEL : @main
+// CHECK-LABEL: @main
 int main() {
 // CHECK: [[B:%.+]] = alloca i8
 // CHECK: [[S:%.+]] = alloca [2 x [[STRUCT_S]]]
   char b;
   S s[2];
+  int arr[10][a];
 // CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num([[IDENT_T]]* @{{.+}})
 // CHECK: [[B_REF:%.+]] = getelementptr inbounds [[STRUCT_SHAREDS]], [[STRUCT_SHAREDS]]* [[CAPTURES:%.+]], i32 0, i32 0
 // CHECK: store i8* [[B]], i8** [[B_REF]]
@@ -52,33 +53,49 @@ int main() {
 // CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[SHAREDS_REF]], i8* [[BITCAST]], i64 8, i32 8, i1 false)
 // CHECK: [[DESTRUCTORS_REF_PTR:%.+]] = getelementptr inbounds [[KMP_TASK_T]], [[KMP_TASK_T]]* [[TASK_PTR]], i32 0, i32 3
 // CHECK: store i32 (i32, i8*)* null, i32 (i32, i8*)** [[DESTRUCTORS_REF_PTR]]
-// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
-// CHECK: store i64 ptrtoint (i32* @{{.+}} to i64), i64*
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
-// CHECK: store i64 4, i64*
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
-// CHECK: store i8 1, i8*
-// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 1
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: [[DEP:%.*]] = getelementptr inbounds [4 x [[KMP_DEPEND_INFO]]], [4 x [[KMP_DEPEND_INFO]]]* [[DEPENDENCIES:%.*]], i64 0, i64 0
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 0
+// CHECK: store i64 ptrtoint (i32* @{{.+}} to i64), i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 1
+// CHECK: store i64 4, i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 2
+// CHECK: store i8 1, i8* [[T0]]
+// CHECK: [[DEP:%.*]] = getelementptr inbounds [4 x [[KMP_DEPEND_INFO]]], [4 x [[KMP_DEPEND_INFO]]]* [[DEPENDENCIES]], i64 0, i64 1
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 0
 // CHECK: ptrtoint i8* [[B]] to i64
-// CHECK: store i64 %{{[^,]+}}, i64*
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
-// CHECK: store i64 1, i64*
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
-// CHECK: store i8 1, i8*
-// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 2
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: store i64 %{{[^,]+}}, i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 1
+// CHECK: store i64 1, i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 2
+// CHECK: store i8 1, i8* [[T0]]
+// CHECK: [[DEP:%.*]] = getelementptr inbounds [4 x [[KMP_DEPEND_INFO]]], [4 x [[KMP_DEPEND_INFO]]]* [[DEPENDENCIES]], i64 0, i64 2
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 0
 // CHECK: ptrtoint [2 x [[STRUCT_S]]]* [[S]] to i64
-// CHECK: store i64 %{{[^,]+}}, i64*
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
-// CHECK: store i64 8, i64*
-// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
-// CHECK: store i8 1, i8*
-// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
-// CHECK: bitcast [[KMP_DEPEND_INFO]]* %{{.+}} to i8*
-// CHECK: call i32 @__kmpc_omp_task_with_deps([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]], i32 3, i8* %{{[^,]+}}, i32 0, i8* null)
-#pragma omp task shared(a, s) depend(in : a, b, s)
+// CHECK: store i64 %{{[^,]+}}, i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 1
+// CHECK: store i64 8, i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 2
+// CHECK: store i8 1, i8* [[T0]]
+// CHECK: [[IDX1:%.+]] = mul nsw i64 0, [[A_VAL:%.+]]
+// CHECK: [[START:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[IDX1:%.+]] = mul nsw i64 9, [[A_VAL]]
+// CHECK: [[END:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[END1:%.+]] = getelementptr i32, i32* [[END]], i32 1
+// CHECK: [[START_INT:%.+]] = ptrtoint i32* [[START]] to i64
+// CHECK: [[END_INT:%.+]] = ptrtoint i32* [[END1]] to i64
+// CHECK: [[SIZEOF:%.+]] = sub nuw i64 [[END_INT]], [[START_INT]]
+// CHECK: [[DEP:%.*]] = getelementptr inbounds [4 x [[KMP_DEPEND_INFO]]], [4 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i64 0, i64 3
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* [[DEP]], i32 0, i32 0
+// CHECK: [[T1:%.*]] = ptrtoint i32* [[START]] to i64
+// CHECK: store i64 [[T1]], i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
+// CHECK: store i64 [[SIZEOF]], i64* [[T0]]
+// CHECK: [[T0:%.*]] = getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
+// CHECK: store i8 1, i8* [[T0]]
+// CHECK: [[DEPS:%.*]] = getelementptr inbounds [4 x [[KMP_DEPEND_INFO]]], [4 x [[KMP_DEPEND_INFO]]]* [[DEPENDENCIES]], i32 0, i32 0
+// CHECK: bitcast [[KMP_DEPEND_INFO]]* [[DEPS]] to i8*
+// CHECK: call i32 @__kmpc_omp_task_with_deps([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]], i32 4, i8* %{{[^,]+}}, i32 0, i8* null)
+#pragma omp task shared(a, s) depend(in : a, b, s, arr[:])
   {
     a = 15;
     s[1].a = 10;
@@ -94,34 +111,56 @@ int main() {
 // CHECK: [[ORIG_TASK_PTR:%.+]] = call i8* @__kmpc_omp_task_alloc([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i32 0, i64 32, i64 1,
 // 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: getelementptr inbounds [2 x [[STRUCT_S]]], [2 x [[STRUCT_S]]]* [[S]], i32 0, i64 0
-// CHECK: getelementptr inbounds [1 x [[KMP_DEPEND_INFO]]], [1 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: getelementptr inbounds [2 x [[STRUCT_S]]], [2 x [[STRUCT_S]]]* [[S]], i64 0, i64 0
+// CHECK: getelementptr inbounds [2 x [[KMP_DEPEND_INFO]]], [2 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i64 0, i64 0
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
 // CHECK: ptrtoint [[STRUCT_S]]* %{{.+}} to i64
 // CHECK: store i64 %{{[^,]+}}, i64*
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
 // CHECK: store i64 4, i64*
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
-// CHECK: store i8 2, i8*
-// CHECK: getelementptr inbounds [1 x [[KMP_DEPEND_INFO]]], [1 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: store i8 3, i8*
+// CHECK: [[IDX1:%.+]] = mul nsw i64 4, [[A_VAL]]
+// CHECK: [[START:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[B_VAL:%.+]] = load i8, i8* [[B]]
+// CHECK: [[IDX2:%.+]] = sext i8 [[B_VAL]] to i64
+// CHECK: [[START1:%.+]] = getelementptr inbounds i32, i32* [[START]], i64 [[IDX2]]
+// CHECK: [[IDX1:%.+]] = mul nsw i64 9, [[A_VAL]]
+// CHECK: [[END:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[B_VAL:%.+]] = load i8, i8* [[B]]
+// CHECK: [[IDX2:%.+]] = sext i8 [[B_VAL]] to i64
+// CHECK: [[END1:%.+]] = getelementptr inbounds i32, i32* [[END]], i64 [[IDX2]]
+// CHECK: [[END2:%.+]] = getelementptr i32, i32* [[END1]], i32 1
+// CHECK: [[START_INT:%.+]] = ptrtoint i32* [[START1]] to i64
+// CHECK: [[END_INT:%.+]] = ptrtoint i32* [[END2]] to i64
+// CHECK: [[SIZEOF:%.+]] = sub nuw i64 [[END_INT]], [[START_INT]]
+// CHECK: getelementptr inbounds [2 x [[KMP_DEPEND_INFO]]], [2 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i64 0, i64 1
+// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: ptrtoint i32* [[START1]] to i64
+// CHECK: store i64 %{{[^,]+}}, i64*
+// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
+// CHECK: store i64 [[SIZEOF]], i64*
+// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
+// CHECK: store i8 3, i8*
+// CHECK: getelementptr inbounds [2 x [[KMP_DEPEND_INFO]]], [2 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
 // CHECK: bitcast [[KMP_DEPEND_INFO]]* %{{.+}} to i8*
-// CHECK: call i32 @__kmpc_omp_task_with_deps([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]], i32 1, i8* %{{[^,]+}}, i32 0, i8* null)
-#pragma omp task untied depend(out : s[0])
+// CHECK: call i32 @__kmpc_omp_task_with_deps([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]], i32 2, i8* %{{[^,]+}}, i32 0, i8* null)
+#pragma omp task untied depend(out : s[0], arr[4:][b])
   {
     a = 1;
   }
 // CHECK: [[ORIG_TASK_PTR:%.+]] = call i8* @__kmpc_omp_task_alloc([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i32 3, i64 32, i64 1,
 // 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: getelementptr inbounds [2 x [[KMP_DEPEND_INFO]]], [2 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i64 0, i64 0
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
 // CHECK: store i64 ptrtoint (i32* @{{.+}} to i64), i64*
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
 // CHECK: store i64 4, i64*
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
 // CHECK: store i8 3, i8*
-// CHECK: getelementptr inbounds [2 x [[STRUCT_S]]], [2 x [[STRUCT_S]]]* [[S]], i32 0, i64 1
-// CHECK: getelementptr inbounds [2 x [[KMP_DEPEND_INFO]]], [2 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 1
+// CHECK: getelementptr inbounds [2 x [[STRUCT_S]]], [2 x [[STRUCT_S]]]* [[S]], i64 0, i64 1
+// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i64 0, i64 1
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
 // CHECK: ptrtoint [[STRUCT_S]]* %{{.+}} to i64
 // CHECK: store i64 %{{[^,]+}}, i64*
@@ -129,10 +168,34 @@ int main() {
 // CHECK: store i64 4, i64*
 // CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
 // CHECK: store i8 3, i8*
-// CHECK: getelementptr inbounds [2 x [[KMP_DEPEND_INFO]]], [2 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: [[IDX1:%.+]] = mul nsw i64 0, [[A_VAL]]
+// CHECK: [[START:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[START1:%.+]] = getelementptr inbounds i32, i32* [[START]], i64 3
+// CHECK: [[NEW_A_VAL:%.+]] = load i32, i32* @{{.+}},
+// CHECK: [[NEW_A_VAL_I64:%.+]] = sext i32 [[NEW_A_VAL]] to i64
+// CHECK: [[SUB:%.+]] = add nsw i64 -1, [[NEW_A_VAL_I64]]
+// CHECK: [[IDX1:%.+]] = mul nsw i64 [[SUB]], [[A_VAL]]
+// CHECK: [[END:%.+]] = getelementptr inbounds i32, i32* %{{.+}}, i64 [[IDX1]]
+// CHECK: [[NEW_A_VAL:%.+]] = load i32, i32* @{{.+}},
+// CHECK: [[NEW_A_VAL_I64:%.+]] = sext i32 [[NEW_A_VAL]] to i64
+// CHECK: [[IDX2:%.+]] = sub nsw i64 [[NEW_A_VAL_I64]], 1
+// CHECK: [[END1:%.+]] = getelementptr inbounds i32, i32* [[END]], i64 [[IDX2]]
+// CHECK: [[END2:%.+]] = getelementptr i32, i32* [[END1]], i32 1
+// CHECK: [[START_INT:%.+]] = ptrtoint i32* [[START1]] to i64
+// CHECK: [[END_INT:%.+]] = ptrtoint i32* [[END2]] to i64
+// CHECK: [[SIZEOF:%.+]] = sub nuw i64 [[END_INT]], [[START_INT]]
+// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i64 0, i64 2
+// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 0
+// CHECK: ptrtoint i32* [[START1]] to i64
+// CHECK: store i64 %{{[^,]+}}, i64*
+// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 1
+// CHECK: store i64 [[SIZEOF]], i64*
+// CHECK: getelementptr inbounds [[KMP_DEPEND_INFO]], [[KMP_DEPEND_INFO]]* %{{[^,]+}}, i32 0, i32 2
+// CHECK: store i8 3, i8*
+// CHECK: getelementptr inbounds [3 x [[KMP_DEPEND_INFO]]], [3 x [[KMP_DEPEND_INFO]]]* %{{[^,]+}}, i32 0, i32 0
 // CHECK: bitcast [[KMP_DEPEND_INFO]]* %{{.+}} to i8*
-// CHECK: call i32 @__kmpc_omp_task_with_deps([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]], i32 2, i8* %{{[^,]+}}, i32 0, i8* null)
-#pragma omp task final(true) depend(inout: a, s[1])
+// CHECK: call i32 @__kmpc_omp_task_with_deps([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i8* [[ORIG_TASK_PTR]], i32 3, i8* %{{[^,]+}}, i32 0, i8* null)
+#pragma omp task final(true) depend(inout: a, s[1], arr[:a][3:])
   {
     a = 2;
   }
@@ -157,33 +220,36 @@ int main() {
 // CHECK: [[CMP:%.+]] = icmp ne i8 [[B_VAL]], 0
 // CHECK: [[FINAL:%.+]] = select i1 [[CMP]], i32 2, i32 0
 // CHECK: [[FLAGS:%.+]] = or i32 [[FINAL]], 1
-// CHECK: [[ORIG_TASK_PTR:%.+]] = call i8* @__kmpc_omp_task_alloc([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i32 [[FLAGS]], i64 32, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_T]]{{.*}}*)* [[TASK_ENTRY5:@.+]] to i32 (i32, i8*)*))
+// CHECK: [[ORIG_TASK_PTR:%.+]] = call i8* @__kmpc_omp_task_alloc([[IDENT_T]]* @{{.+}}, i32 [[GTID]], i32 [[FLAGS]], i64 32, i64 8, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_T]]{{.*}}*)* [[TASK_ENTRY5:@.+]] 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 final(b)
+  int c __attribute__((aligned(128)));
+#pragma omp task final(b) shared(c)
   {
     a = 4;
+    c = 5;
   }
   return a;
 }
-// CHECK: define internal i32 [[TASK_ENTRY1]](i32, [[KMP_TASK_T]]{{.*}}*)
+// CHECK: define internal i32 [[TASK_ENTRY1]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
 // CHECK: store i32 15, i32* [[A_PTR:@.+]]
 // CHECK: [[A_VAL:%.+]] = load i32, i32* [[A_PTR]]
 // CHECK: [[A_VAL_I8:%.+]] = trunc i32 [[A_VAL]] to i8
 // CHECK: store i8 [[A_VAL_I8]], i8* %{{.+}}
 // CHECK: store i32 10, i32* %{{.+}}
 
-// CHECK: define internal i32 [[TASK_ENTRY2]](i32, [[KMP_TASK_T]]{{.*}}*)
+// CHECK: define internal i32 [[TASK_ENTRY2]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
 // CHECK: store i32 1, i32* [[A_PTR:@.+]]
 
-// CHECK: define internal i32 [[TASK_ENTRY3]](i32, [[KMP_TASK_T]]{{.*}}*)
+// CHECK: define internal i32 [[TASK_ENTRY3]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
 // CHECK: store i32 2, i32* [[A_PTR:@.+]]
 
-// CHECK: define internal i32 [[TASK_ENTRY4]](i32, [[KMP_TASK_T]]{{.*}}*)
+// CHECK: define internal i32 [[TASK_ENTRY4]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
 // CHECK: store i32 3, i32* [[A_PTR:@.+]]
 
-// CHECK: define internal i32 [[TASK_ENTRY5]](i32, [[KMP_TASK_T]]{{.*}}*)
+// CHECK: define internal i32 [[TASK_ENTRY5]](i32, [[KMP_TASK_T]]{{.*}}* noalias)
 // CHECK: store i32 4, i32* [[A_PTR:@.+]]
+// CHECK: store i32 5, i32* [[C_PTR:%.+]], align 128
 #endif
 
diff --git a/test/OpenMP/task_depend_messages.cpp b/test/OpenMP/task_depend_messages.cpp
index 152350cca93a..39bf4847890c 100644
--- a/test/OpenMP/task_depend_messages.cpp
+++ b/test/OpenMP/task_depend_messages.cpp
@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 -o - -std=c++11 %s
 
 void foo() {
 }
@@ -14,15 +14,18 @@ class vector {
     int operator[](int index) { return 0; }
 };
 
-int main(int argc, char **argv) {
+int main(int argc, char **argv, char *env[]) {
   vector vec;
   typedef float V __attribute__((vector_size(16)));
   V a;
+  auto arr = x; // expected-error {{use of undeclared identifier 'x'}}
 
   #pragma omp task depend // expected-error {{expected '(' after 'depend'}}
   #pragma omp task depend ( // expected-error {{expected 'in', 'out' or 'inout' in OpenMP clause 'depend'}} expected-error {{expected ')'}} expected-note {{to match this '('}} expected-warning {{missing ':' after dependency type - ignoring}}
   #pragma omp task depend () // expected-error {{expected 'in', 'out' or 'inout' in OpenMP clause 'depend'}} expected-warning {{missing ':' after dependency type - ignoring}}
   #pragma omp task depend (argc // expected-error {{expected 'in', 'out' or 'inout' in OpenMP clause 'depend'}} expected-warning {{missing ':' after dependency type - ignoring}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task depend (source : argc) // expected-error {{expected 'in', 'out' or 'inout' in OpenMP clause 'depend'}}
+  #pragma omp task depend (source) // expected-error {{expected expression}} expected-warning {{missing ':' after dependency type - ignoring}}
   #pragma omp task depend (in : argc)) // expected-warning {{extra tokens at the end of '#pragma omp task' are ignored}}
   #pragma omp task depend (out: ) // expected-error {{expected expression}}
   #pragma omp task depend (inout : foobool(argc)), depend (in, argc) // expected-error {{expected variable name, array element or array section}} expected-warning {{missing ':' after dependency type - ignoring}} expected-error {{expected expression}}
@@ -33,6 +36,22 @@ int main(int argc, char **argv) {
   #pragma omp task depend (in : ) // expected-error {{expected expression}}
   #pragma omp task depend (in : main) // expected-error {{expected variable name, array element or array section}}
   #pragma omp task depend(in : a[0]) // expected-error{{expected variable name, array element or array section}}
+  #pragma omp task depend (in : vec[1:2]) // expected-error {{ value is not an array or pointer}}
+  #pragma omp task depend (in : argv[ // expected-error {{expected expression}} expected-error {{expected ']'}} expected-error {{expected ')'}} expected-note {{to match this '['}} expected-note {{to match this '('}}
+  #pragma omp task depend (in : argv[: // expected-error {{expected expression}} expected-error {{expected ']'}} expected-error {{expected ')'}} expected-note {{to match this '['}} expected-note {{to match this '('}}
+  #pragma omp task depend (in : argv[:] // expected-error {{section length is unspecified and cannot be inferred because subscripted value is not an array}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task depend (in : argv[argc: // expected-error {{expected expression}} expected-error {{expected ']'}} expected-error {{expected ')'}} expected-note {{to match this '['}} expected-note {{to match this '('}}
+  #pragma omp task depend (in : argv[argc:argc] // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task depend (in : argv[0:-1]) // expected-error {{section length is evaluated to a negative value -1}}
+  #pragma omp task depend (in : argv[-1:0]) // expected-error {{section lower bound is evaluated to a negative value -1}}
+  #pragma omp task depend (in : argv[:]) // expected-error {{section length is unspecified and cannot be inferred because subscripted value is not an array}}
+  #pragma omp task depend (in : argv[3:4:1]) // expected-error {{expected ']'}} expected-note {{to match this '['}}
+  #pragma omp task depend(in:a[0:1]) // expected-error {{subscripted value is not an array or pointer}}
+  #pragma omp task depend(in:argv[argv[:2]:1]) // expected-error {{OpenMP array section is not allowed here}}
+  #pragma omp task depend(in:argv[0:][:]) // expected-error {{section length is unspecified and cannot be inferred because subscripted value is not an array}}
+  #pragma omp task depend(in:env[0:][:]) // expected-error {{section length is unspecified and cannot be inferred because subscripted value is an array of unknown bound}}
+  #pragma omp task depend(in : argv[ : argc][1 : argc - 1])
+  #pragma omp task depend(in : arr[0])
   foo();
 
   return 0;
diff --git a/test/OpenMP/task_firstprivate_codegen.cpp b/test/OpenMP/task_firstprivate_codegen.cpp
index 89aadbeb6a3f..e2244140d1e1 100644
--- a/test/OpenMP/task_firstprivate_codegen.cpp
+++ b/test/OpenMP/task_firstprivate_codegen.cpp
@@ -26,18 +26,18 @@ volatile double g;
 
 // CHECK-DAG: [[KMP_TASK_T_TY:%.+]] = type { i8*, i32 (i32, i8*)*, i32, i32 (i32, i8*)* }
 // CHECK-DAG: [[S_DOUBLE_TY:%.+]] = type { double }
-// CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { [2 x i32]*, i32*, [2 x [[S_DOUBLE_TY]]]*, [[S_DOUBLE_TY]]* }
-// CHECK-DAG: [[PRIVATES_MAIN_TY:%.+]] = type {{.?}}{ [[S_DOUBLE_TY]], [2 x [[S_DOUBLE_TY]]], i32, [2 x i32]
+// CHECK-DAG: [[PRIVATES_MAIN_TY:%.+]] = type {{.?}}{ [2 x [[S_DOUBLE_TY]]], [[S_DOUBLE_TY]], i32, [2 x i32]
+// CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { [2 x i32]*, i32*, [2 x [[S_DOUBLE_TY]]]*, [[S_DOUBLE_TY]]*, i{{[0-9]+}}* }
 // CHECK-DAG: [[KMP_TASK_MAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [[PRIVATES_MAIN_TY]] }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i32 }
 // CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { [2 x i32]*, i32*, [2 x [[S_INT_TY]]]*, [[S_INT_TY]]* }
-// CHECK-DAG: [[PRIVATES_TMAIN_TY:%.+]] = type { i32, [2 x i32], [2 x [[S_INT_TY]]], [[S_INT_TY]] }
-// CHECK-DAG: [[KMP_TASK_TMAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [[PRIVATES_TMAIN_TY]] }
+// CHECK-DAG: [[PRIVATES_TMAIN_TY:%.+]] = type { i32, [2 x i32], [2 x [[S_INT_TY]]], [[S_INT_TY]], [104 x i8] }
+// CHECK-DAG: [[KMP_TASK_TMAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [{{[0-9]+}} x i8], [[PRIVATES_TMAIN_TY]] }
 template <typename T>
 T tmain() {
   S<T> ttt;
   S<T> test(ttt);
-  T t_var = T();
+  T t_var __attribute__((aligned(128))) = T();
   T vec[] = {1, 2};
   S<T> s_arr[] = {1, 2};
   S<T> var(3);
@@ -50,22 +50,31 @@ T tmain() {
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global double
+  // LAMBDA: [[SIVAR:@.+]] = internal global i{{[0-9]+}} 0,
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{( x86_thiscallcc)?}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 40, i64 8, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+  // LAMBDA: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 48, i64 16, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
 // LAMBDA: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
 // LAMBDA: [[G_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 0
 // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_ADDR_REF]]
 // LAMBDA: [[G_VAL:%.+]] = load volatile double, double* [[G_REF]]
 // LAMBDA: store volatile double [[G_VAL]], double* [[G_PRIVATE_ADDR]]
+
+// LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// LAMBDA: [[SIVAR_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
+// LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_ADDR_REF]]
+// LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF]]
+// LAMBDA: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
+
 // LAMBDA: call i32 @__kmpc_omp_task(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]])
 // LAMBDA: ret
-#pragma omp task firstprivate(g)
+#pragma omp task firstprivate(g, sivar)
   {
     // LAMBDA: define {{.+}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG_PTR:%.+]])
     // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
@@ -75,13 +84,16 @@ int main() {
     // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
 
     // LAMBDA: store double* %{{.+}}, double** %{{.+}},
-    // LAMBDA: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}*)
+    // LAMBDA: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}* noalias)
     g = 1;
+    sivar = 11;
     // LAMBDA: store double 1.0{{.+}}, double* %{{.+}},
+    // LAMBDA: store i{{[0-9]+}} 11, i{{[0-9]+}}* %{{.+}},
     // LAMBDA: call void [[INNER_LAMBDA]](%
     // LAMBDA: ret
     [&]() {
       g = 2;
+      sivar = 22;
     }();
   }
   }();
@@ -92,31 +104,45 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 40, i64 8, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+  // BLOCKS: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 48, i64 16, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
   // BLOCKS: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
   // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
   // BLOCKS: [[G_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 0
   // BLOCKS: [[G_REF:%.+]] = load double*, double** [[G_ADDR_REF]]
   // BLOCKS: [[G_VAL:%.+]] = load volatile double, double* [[G_REF]]
   // BLOCKS: store volatile double [[G_VAL]], double* [[G_PRIVATE_ADDR]]
+
+  // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+  // BLOCKS: [[SIVAR_ADDR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
+  // BLOCKS: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_ADDR_REF]]
+  // BLOCKS: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF]]
+  // BLOCKS: store i{{[0-9]+}} [[SIVAR_VAL]], i{{[0-9]+}}* [[SIVAR_PRIVATE_ADDR]]
   // BLOCKS: call i32 @__kmpc_omp_task(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]])
   // BLOCKS: ret
-#pragma omp task firstprivate(g)
+#pragma omp task firstprivate(g, sivar)
   {
     // BLOCKS: define {{.+}} void {{@.+}}(i8*
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: store double 2.0{{.+}}, double*
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[ISVAR]]{{[[^:word:]]}}
+    // BLOCKS: store i{{[0-9]+}} 22, i{{[0-9]+}}*
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: ret
 
     // BLOCKS: store double* %{{.+}}, double** %{{.+}},
-    // BLOCKS: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}*)
+    // BLOCKS: store i{{[0-9]+}}* %{{.+}}, i{{[0-9]+}}** %{{.+}},
+    // BLOCKS: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}* noalias)
     g = 1;
+    sivar = 11;
     // BLOCKS: store double 1.0{{.+}}, double* %{{.+}},
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: store i{{[0-9]+}} 11, i{{[0-9]+}}* %{{.+}},
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
     ^{
       g = 2;
+      sivar = 22;
     }();
   }
   }();
@@ -128,15 +154,17 @@ int main() {
   int vec[] = {1, 2};
   S<double> s_arr[] = {1, 2};
   S<double> var(3);
-#pragma omp task firstprivate(var, t_var, s_arr, vec, s_arr, var)
+#pragma omp task firstprivate(var, t_var, s_arr, vec, s_arr, var, sivar)
   {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 33;
   }
   return tmain<int>();
 #endif
 }
 
+// CHECK: [[SIVAR:.+]] = internal global i{{[0-9]+}} 0,
 // CHECK: define i{{[0-9]+}} @main()
 // CHECK: alloca [[S_DOUBLE_TY]],
 // CHECK: [[TEST:%.+]] = alloca [[S_DOUBLE_TY]],
@@ -157,13 +185,15 @@ int main() {
 // CHECK: store [2 x [[S_DOUBLE_TY]]]* [[S_ARR_ADDR]], [2 x [[S_DOUBLE_TY]]]** [[S_ARR_REF]],
 // CHECK: [[VAR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 3
 // CHECK: store [[S_DOUBLE_TY]]* [[VAR_ADDR]], [[S_DOUBLE_TY]]** [[VAR_REF]],
+// CHECK: [[SIVAR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* %{{.+}}, i{{[0-9]+}} 0, i{{[0-9]+}} 4
+// CHECK: store i{{[0-9]+}}* [[SIVAR]], i{{[0-9]+}}** [[SIVAR_REF]],
 
 // Allocate task.
 // Returns struct kmp_task_t {
 //         [[KMP_TASK_T]] task_data;
 //         [[KMP_TASK_MAIN_TY]] privates;
 //       };
-// CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID]], i32 1, i64 72, i64 32, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_MAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+// CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID]], i32 1, i64 72, i64 40, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_MAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
 // CHECK: [[RES_KMP_TASK:%.+]] = bitcast i8* [[RES]] to [[KMP_TASK_MAIN_TY]]*
 
 // Fill kmp_task_t->shareds by copying from original capture argument.
@@ -171,7 +201,7 @@ int main() {
 // CHECK: [[SHAREDS_REF_ADDR:%.+]] = getelementptr inbounds [[KMP_TASK_T_TY]], [[KMP_TASK_T_TY]]* [[TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[SHAREDS_REF:%.+]] = load i8*, i8** [[SHAREDS_REF_ADDR]],
 // CHECK: [[CAPTURES_ADDR:%.+]] = bitcast [[CAP_MAIN_TY]]* %{{.+}} to i8*
-// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[SHAREDS_REF]], i8* [[CAPTURES_ADDR]], i64 32, i32 8, i1 false)
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[SHAREDS_REF]], i8* [[CAPTURES_ADDR]], i64 40, i32 8, i1 false)
 
 // Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).
 // Also copy address of private copy to the corresponding shareds reference.
@@ -179,14 +209,8 @@ int main() {
 // CHECK: [[SHAREDS:%.+]] = bitcast i8* [[SHAREDS_REF]] to [[CAP_MAIN_TY]]*
 
 // Constructors for s_arr and var.
-// var;
-// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
-// CHECK: [[VAR_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 3
-// CHECK: [[VAR_REF:%.+]] = load [[S_DOUBLE_TY]]*, [[S_DOUBLE_TY]]** [[VAR_ADDR_REF]],
-// CHECK: call void [[S_DOUBLE_TY_COPY_CONSTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF]], [[S_DOUBLE_TY]]* {{.*}}[[VAR_REF]],
-
 // s_arr;
-// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: [[S_ARR_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 2
 // CHECK: load [2 x [[S_DOUBLE_TY]]]*, [2 x [[S_DOUBLE_TY]]]** [[S_ARR_ADDR_REF]],
 // CHECK: call void [[S_DOUBLE_TY_COPY_CONSTR]]([[S_DOUBLE_TY]]* [[S_ARR_CUR:%[^,]+]],
@@ -195,6 +219,12 @@ int main() {
 // CHECK: icmp eq
 // CHECK: br i1
 
+// var;
+// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// CHECK: [[VAR_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 3
+// CHECK: [[VAR_REF:%.+]] = load [[S_DOUBLE_TY]]*, [[S_DOUBLE_TY]]** [[VAR_ADDR_REF]],
+// CHECK: call void [[S_DOUBLE_TY_COPY_CONSTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF]], [[S_DOUBLE_TY]]* {{.*}}[[VAR_REF]],
+
 // t_var;
 // CHECK: [[PRIVATE_T_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 2
 // CHECK: [[T_VAR_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 1
@@ -207,6 +237,13 @@ int main() {
 // CHECK: [[VEC_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 0
 // CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(
 
+// sivar;
+// CHECK: [[PRIVATE_SIVAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 4
+// CHECK: [[SIVAR_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_MAIN_TY]], [[CAP_MAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 4
+// CHECK: [[SIVAR_REF:%.+]] = load i{{.+}}*, i{{.+}}** [[SIVAR_ADDR_REF]],
+// CHECK: [[SIVAR:%.+]] = load i{{.+}}, i{{.+}}* [[SIVAR_REF]],
+// CHECK: store i32 [[SIVAR]], i32* [[PRIVATE_SIVAR_REF]],
+
 // Provide pointer to destructor function, which will destroy private variables at the end of the task.
 // CHECK: [[DESTRUCTORS_REF:%.+]] = getelementptr inbounds [[KMP_TASK_T_TY]], [[KMP_TASK_T_TY]]* [[TASK]], i{{.+}} 0, i{{.+}} 3
 // CHECK: store i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_MAIN_TY]]*)* [[DESTRUCTORS:@.+]] to i32 (i32, i8*)*), i32 (i32, i8*)** [[DESTRUCTORS_REF]],
@@ -225,61 +262,69 @@ int main() {
 // CHECK: ret
 //
 
-// CHECK: define internal void [[PRIVATES_MAP_FN:@.+]]([[PRIVATES_MAIN_TY]]* noalias, [[S_DOUBLE_TY]]** noalias, i32** noalias, [2 x [[S_DOUBLE_TY]]]** noalias, [2 x i32]** noalias)
+// CHECK: define internal void [[PRIVATES_MAP_FN:@.+]]([[PRIVATES_MAIN_TY]]* noalias, [[S_DOUBLE_TY]]** noalias, i32** noalias, [2 x [[S_DOUBLE_TY]]]** noalias, [2 x i32]** noalias, i32** noalias)
 // CHECK: [[PRIVATES:%.+]] = load [[PRIVATES_MAIN_TY]]*, [[PRIVATES_MAIN_TY]]**
-// CHECK: [[PRIV_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 0
-// CHECK: [[ARG1:%.+]] = load [[S_DOUBLE_TY]]**, [[S_DOUBLE_TY]]*** {{.+}},
-// CHECK: store [[S_DOUBLE_TY]]* [[PRIV_VAR]], [[S_DOUBLE_TY]]** [[ARG1]],
-// CHECK: [[PRIV_S_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 1
+// CHECK: [[PRIV_S_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 0
 // CHECK: [[ARG3:%.+]] = load [2 x [[S_DOUBLE_TY]]]**, [2 x [[S_DOUBLE_TY]]]*** %{{.+}},
 // CHECK: store [2 x [[S_DOUBLE_TY]]]* [[PRIV_S_VAR]], [2 x [[S_DOUBLE_TY]]]** [[ARG3]],
+// CHECK: [[PRIV_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 1
+// CHECK: [[ARG1:%.+]] = load [[S_DOUBLE_TY]]**, [[S_DOUBLE_TY]]*** {{.+}},
+// CHECK: store [[S_DOUBLE_TY]]* [[PRIV_VAR]], [[S_DOUBLE_TY]]** [[ARG1]],
 // CHECK: [[PRIV_T_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 2
 // CHECK: [[ARG2:%.+]] = load i32**, i32*** %{{.+}},
 // CHECK: store i32* [[PRIV_T_VAR]], i32** [[ARG2]],
 // CHECK: [[PRIV_VEC:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 3
 // CHECK: [[ARG4:%.+]] = load [2 x i32]**, [2 x i32]*** %{{.+}},
 // CHECK: store [2 x i32]* [[PRIV_VEC]], [2 x i32]** [[ARG4]],
+// CHECK: [[PRIV_SIVAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 4
+// CHECK: [[ARG5:%.+]] = load i{{[0-9]+}}**, i{{[0-9]+}}*** %{{.+}},
+// CHECK: store i{{[0-9]+}}* [[PRIV_SIVAR]], i{{[0-9]+}}** [[ARG5]],
 // CHECK: ret void
 
-// CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_MAIN_TY]]*)
+// CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_MAIN_TY]]* noalias)
 
 // CHECK: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_DOUBLE_TY]]*,
 // CHECK: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
 // CHECK: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_DOUBLE_TY]]]*,
 // CHECK: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
-// CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_MAIN_TY]]*, [[S_DOUBLE_TY]]**, i32**, [2 x [[S_DOUBLE_TY]]]**, [2 x i32]**)* [[PRIVATES_MAP_FN]] to void (i8*, ...)*), void (i8*, ...)** [[MAP_FN_ADDR:%.+]],
+// CHECK: [[PRIV_SIVAR_ADDR:%.+]] = alloca i32*,
+// CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_MAIN_TY]]*, [[S_DOUBLE_TY]]**, i32**, [2 x [[S_DOUBLE_TY]]]**, [2 x i32]**, i32**)* [[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* %{{.+}}, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]], i32** [[PRIV_T_VAR_ADDR]], [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]])
+
+// CHECK: call void (i8*, ...) [[MAP_FN]](i8* %{{.+}}, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]], i32** [[PRIV_T_VAR_ADDR]], [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]], i32** [[PRIV_SIVAR_ADDR]])
+
 // CHECK: [[PRIV_VAR:%.+]] = load [[S_DOUBLE_TY]]*, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]],
 // CHECK: [[PRIV_T_VAR:%.+]] = load i32*, i32** [[PRIV_T_VAR_ADDR]],
 // CHECK: [[PRIV_S_ARR:%.+]] = load [2 x [[S_DOUBLE_TY]]]*, [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]],
 // CHECK: [[PRIV_VEC:%.+]] = load [2 x i32]*, [2 x i32]** [[PRIV_VEC_ADDR]],
+// CHECK: [[PRIV_SIVAR:%.+]] = load i32*, i32** [[PRIV_SIVAR_ADDR]],
 
 // Privates actually are used.
 // CHECK-DAG: [[PRIV_VAR]]
 // CHECK-DAG: [[PRIV_T_VAR]]
 // CHECK-DAG: [[PRIV_S_ARR]]
 // CHECK-DAG: [[PRIV_VEC]]
+// CHECK-DAG: [[PRIV_SIVAR]]
 
 // CHECK: ret
 
-// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_MAIN_TY]]*)
+// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_MAIN_TY]]* noalias)
 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
-// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
+// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// CHECK: call void [[S_DOUBLE_TY_DESTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF]])
 // CHECK: getelementptr inbounds [2 x [[S_DOUBLE_TY]]], [2 x [[S_DOUBLE_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} 2
 // CHECK: [[PRIVATE_S_ARR_ELEM_REF:%.+]] = getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} -1
 // CHECK: call void [[S_DOUBLE_TY_DESTR]]([[S_DOUBLE_TY]]* [[PRIVATE_S_ARR_ELEM_REF]])
 // CHECK: icmp eq
 // CHECK: br i1
-// CHECK: call void [[S_DOUBLE_TY_DESTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF]])
 // CHECK: ret i32
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
 // CHECK: alloca [[S_INT_TY]],
 // CHECK: [[TEST:%.+]] = alloca [[S_INT_TY]],
-// CHECK: [[T_VAR_ADDR:%.+]] = alloca i32,
+// CHECK: [[T_VAR_ADDR:%.+]] = alloca i32, align 128
 // CHECK: [[VEC_ADDR:%.+]] = alloca [2 x i32],
 // CHECK: [[S_ARR_ADDR:%.+]] = alloca [2 x [[S_INT_TY]]],
 // CHECK: [[VAR_ADDR:%.+]] = alloca [[S_INT_TY]],
@@ -302,7 +347,7 @@ int main() {
 //         [[KMP_TASK_T_TY]] task_data;
 //         [[KMP_TASK_TMAIN_TY]] privates;
 //       };
-// CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID]], i32 1, i64 56, i64 32, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_TMAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+// CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID]], i32 1, i64 256, i64 32, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_TMAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
 // CHECK: [[RES_KMP_TASK:%.+]] = bitcast i8* [[RES]] to [[KMP_TASK_TMAIN_TY]]*
 
 // Fill kmp_task_t->shareds by copying from original capture argument.
@@ -313,15 +358,15 @@ int main() {
 // CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[SHAREDS_REF]], i8* [[CAPTURES_ADDR]], i64 32, i32 8, i1 false)
 
 // Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).
-// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
 // CHECK: [[SHAREDS:%.+]] = bitcast i8* [[SHAREDS_REF]] to [[CAP_TMAIN_TY]]*
 
 // t_var;
 // CHECK: [[PRIVATE_T_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
 // CHECK: [[T_VAR_ADDR_REF:%.+]] = getelementptr inbounds [[CAP_TMAIN_TY]], [[CAP_TMAIN_TY]]* [[SHAREDS]], i{{.+}} 0, i{{.+}} 1
 // CHECK: [[T_VAR_REF:%.+]] = load i{{.+}}*, i{{.+}}** [[T_VAR_ADDR_REF]],
-// CHECK: [[T_VAR:%.+]] = load i{{.+}}, i{{.+}}* [[T_VAR_REF]],
-// CHECK: store i32 [[T_VAR]], i32* [[PRIVATE_T_VAR_REF]],
+// CHECK: [[T_VAR:%.+]] = load i{{.+}}, i{{.+}}* [[T_VAR_REF]], align 128
+// CHECK: store i32 [[T_VAR]], i32* [[PRIVATE_T_VAR_REF]], align 128
 
 // vec;
 // CHECK: [[PRIVATE_VEC_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
@@ -376,7 +421,7 @@ int main() {
 // CHECK: store [[S_INT_TY]]* [[PRIV_VAR]], [[S_INT_TY]]** [[ARG4]],
 // CHECK: ret void
 
-// CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_TMAIN_TY]]*)
+// 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]*,
@@ -398,8 +443,8 @@ int main() {
 
 // CHECK: ret
 
-// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_TMAIN_TY]]*)
-// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_TMAIN_TY]]* noalias)
+// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
 // CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 2
 // CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 3
 // CHECK: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[PRIVATE_VAR_REF]])
diff --git a/test/OpenMP/task_firstprivate_messages.cpp b/test/OpenMP/task_firstprivate_messages.cpp
index 0126d47c50f9..ef5f3856430a 100644
--- a/test/OpenMP/task_firstprivate_messages.cpp
+++ b/test/OpenMP/task_firstprivate_messages.cpp
@@ -70,7 +70,8 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;                                               // expected-note {{'j' defined here}}
+  int &j = i;
+  static int m;
 #pragma omp task firstprivate                               // expected-error {{expected '(' after 'firstprivate'}}
 #pragma omp task firstprivate(                              // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
 #pragma omp task firstprivate()                             // expected-error {{expected expression}}
@@ -92,7 +93,8 @@ int main(int argc, char **argv) {
   foo();
 #pragma omp task shared(i)
 #pragma omp task firstprivate(i)
-#pragma omp task firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp task firstprivate(j)
+#pragma omp task firstprivate(m) // OK
   foo();
 
   return 0;
diff --git a/test/OpenMP/task_if_codegen.cpp b/test/OpenMP/task_if_codegen.cpp
index d08c27f5a408..5992be02a84a 100644
--- a/test/OpenMP/task_if_codegen.cpp
+++ b/test/OpenMP/task_if_codegen.cpp
@@ -21,14 +21,14 @@ int Arg;
 
 // CHECK-LABEL: define void @{{.+}}gtid_test
 void gtid_test() {
-// CHECK:  call void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 1, {{.+}}* [[GTID_TEST_REGION1:@.+]] to void
+// CHECK:  call void {{.+}} @__kmpc_fork_call(%{{.+}}* @{{.+}}, i{{.+}} 0, {{.+}}* [[GTID_TEST_REGION1:@.+]] to void
 #pragma omp parallel
-#pragma omp task if (false)
+#pragma omp task if (task: false)
   gtid_test();
 // CHECK: ret void
 }
 
-// CHECK: define internal void [[GTID_TEST_REGION1]](i32* [[GTID_PARAM:%.+]], i
+// CHECK: define internal void [[GTID_TEST_REGION1]](i32* noalias [[GTID_PARAM:%.+]], i
 // CHECK: store i32* [[GTID_PARAM]], i32** [[GTID_ADDR_REF:%.+]],
 // CHECK: [[GTID_ADDR:%.+]] = load i32*, i32** [[GTID_ADDR_REF]]
 // CHECK: [[GTID:%.+]] = load i32, i32* [[GTID_ADDR]]
@@ -45,13 +45,13 @@ void gtid_test() {
 
 template <typename T>
 int tmain(T Arg) {
-#pragma omp task if (true)
+#pragma omp task if (task: true)
   fn1();
 #pragma omp task if (false)
   fn2();
 #pragma omp task if (Arg)
   fn3();
-#pragma omp task if (Arg) depend(in : Arg)
+#pragma omp task if (task: Arg) depend(in : Arg)
   fn4();
 #pragma omp task if (Arg) depend(out : Arg)
   fn5();
diff --git a/test/OpenMP/task_if_messages.cpp b/test/OpenMP/task_if_messages.cpp
index ae6bb13093d9..f3f56d51673a 100644
--- a/test/OpenMP/task_if_messages.cpp
+++ b/test/OpenMP/task_if_messages.cpp
@@ -22,6 +22,12 @@ int tmain(T argc, S **argv) {
   #pragma omp task if (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp task if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp task if(argc)
+  #pragma omp task if(task : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task if(task : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task if(task : argc)
+  #pragma omp task if(task : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp task'}}
+  #pragma omp task if(task : argc) if (task:argc) // expected-error {{directive '#pragma omp task' cannot contain more than one 'if' clause with 'task' name modifier}}
+  #pragma omp task if(task : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
   foo();
 
   return 0;
@@ -40,6 +46,12 @@ int main(int argc, char **argv) {
   #pragma omp task if (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp task if (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
   #pragma omp task if(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task if(task : // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task if(task : argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task if(task : argc)
+  #pragma omp task if(task : argc) if (for:argc) // expected-error {{directive name modifier 'for' is not allowed for '#pragma omp task'}}
+  #pragma omp task if(task : argc) if (task:argc) // expected-error {{directive '#pragma omp task' cannot contain more than one 'if' clause with 'task' name modifier}}
+  #pragma omp task if(task : argc) if (argc) // expected-error {{no more 'if' clause is allowed}} expected-note {{previous clause with directive name modifier specified here}}
   foo();
 
   return tmain(argc, argv);
diff --git a/test/OpenMP/task_messages.cpp b/test/OpenMP/task_messages.cpp
index ec67998afdce..64bf8a40f025 100644
--- a/test/OpenMP/task_messages.cpp
+++ b/test/OpenMP/task_messages.cpp
@@ -6,7 +6,7 @@ void foo() {
 #pragma omp task // expected-error {{unexpected OpenMP directive '#pragma omp task'}}
 
 class S {
-  S(const S &s) { a = s.a + 12; } // expected-note 6 {{implicitly declared private here}}
+  S(const S &s) { a = s.a + 12; } // expected-note 10 {{implicitly declared private here}}
   int a;
 
 public:
@@ -29,7 +29,7 @@ public:
 template <class T>
 int foo() {
   T a;
-  T &b = a; // expected-note 4 {{'b' defined here}}
+  T &b = a;
   int r;
   S1 s1;
 // expected-error@+1 2 {{call to deleted constructor of 'S1'}}
@@ -53,16 +53,11 @@ int foo() {
 #pragma omp task
 #pragma omp parallel
   ++a;
-// expected-error@+2 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'int &'}}
-// expected-error@+1 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'S &'}}
+// expected-error@+2 {{calling a private constructor of class 'S'}}
 #pragma omp task
-  // expected-note@+1 2 {{used here}}
   ++b;
-// expected-error@+2 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'int &'}}
-// expected-error@+1 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'S &'}}
 #pragma omp task
-// expected-error@+2 {{calling a private constructor of class 'S'}}
-// expected-note@+1 2 {{used here}}
+// expected-error@+1 2 {{calling a private constructor of class 'S'}}
 #pragma omp parallel shared(a, b)
   ++a, ++b;
 // expected-note@+1 3 {{defined as reduction}}
@@ -120,9 +115,9 @@ int foo() {
 
 int main(int argc, char **argv) {
   int a;
-  int &b = a; // expected-note 2 {{'b' defined here}}
+  int &b = a;
   S sa;
-  S &sb = sa; // expected-note 2 {{'sb' defined here}}
+  S &sb = sa;
   int r;
 #pragma omp task { // expected-warning {{extra tokens at the end of '#pragma omp task' are ignored}}
   foo();
@@ -192,13 +187,9 @@ L2:
 #pragma omp task
 #pragma omp parallel
   ++a;
-// expected-error@+1 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'int &'}}
 #pragma omp task
-  // expected-note@+1 {{used here}}
   ++b;
-// expected-error@+1 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'int &'}}
 #pragma omp task
-// expected-note@+1 {{used here}}
 #pragma omp parallel shared(a, b)
   ++a, ++b;
 #pragma omp task default(none)
@@ -218,14 +209,11 @@ L2:
 #pragma omp task
 #pragma omp parallel
   ++sa;
-// expected-error@+1 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'S &'}}
+// expected-error@+2 {{calling a private constructor of class 'S'}}
 #pragma omp task
-  // expected-note@+1 {{used here}}
   ++sb;
-// expected-error@+1 {{predetermined as a firstprivate in a task construct variable cannot be of reference type 'S &'}}
+// expected-error@+2 2 {{calling a private constructor of class 'S'}}
 #pragma omp task
-// expected-error@+2 {{calling a private constructor of class 'S'}}
-// expected-note@+1 {{used here}}
 #pragma omp parallel shared(sa, sb)
   ++sa, ++sb;
 // expected-note@+1 2 {{defined as reduction}}
diff --git a/test/OpenMP/task_priority_messages.cpp b/test/OpenMP/task_priority_messages.cpp
new file mode 100644
index 000000000000..8a5553eb08c5
--- /dev/null
+++ b/test/OpenMP/task_priority_messages.cpp
@@ -0,0 +1,47 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp task priority // expected-error {{expected '(' after 'priority'}}
+  #pragma omp task priority ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority () // expected-error {{expected expression}}
+  #pragma omp task priority (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority (argc)) // expected-warning {{extra tokens at the end of '#pragma omp task' are ignored}}
+  #pragma omp task priority (argc > 0 ? argv[1][0] : argv[2][argc])
+  #pragma omp task priority (foobool(argc)), priority (true) // expected-error {{directive '#pragma omp task' cannot contain more than one 'priority' clause}}
+  #pragma omp task priority (S) // expected-error {{'S' does not refer to a value}}
+  #pragma omp task priority (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority(0)
+  #pragma omp task priority(-1) // expected-error {{argument to 'priority' clause must be a non-negative integer value}}
+  foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp task priority // expected-error {{expected '(' after 'priority'}}
+  #pragma omp task priority ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority () // expected-error {{expected expression}}
+  #pragma omp task priority (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority (argc)) // expected-warning {{extra tokens at the end of '#pragma omp task' are ignored}}
+  #pragma omp task priority (argc > 0 ? argv[1][0] : argv[2][argc])
+  #pragma omp task priority (foobool(argc)), priority (true) // expected-error {{directive '#pragma omp task' cannot contain more than one 'priority' clause}}
+  #pragma omp task priority (S1) // expected-error {{'S1' does not refer to a value}}
+  #pragma omp task priority (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  #pragma omp task priority(0)
+  #pragma omp task priority(-1) // expected-error {{argument to 'priority' clause must be a non-negative integer value}}
+  foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/task_private_codegen.cpp b/test/OpenMP/task_private_codegen.cpp
index b29d0d39a1e2..1455fd11a91d 100644
--- a/test/OpenMP/task_private_codegen.cpp
+++ b/test/OpenMP/task_private_codegen.cpp
@@ -27,16 +27,16 @@ volatile double g;
 // CHECK-DAG: [[KMP_TASK_T_TY:%.+]] = type { i8*, i32 (i32, i8*)*, i32, i32 (i32, i8*)* }
 // CHECK-DAG: [[S_DOUBLE_TY:%.+]] = type { double }
 // CHECK-DAG: [[CAP_MAIN_TY:%.+]] = type { i8 }
-// CHECK-DAG: [[PRIVATES_MAIN_TY:%.+]] = type {{.?}}{ [[S_DOUBLE_TY]], [2 x [[S_DOUBLE_TY]]], i32, [2 x i32]
+// CHECK-DAG: [[PRIVATES_MAIN_TY:%.+]] = type {{.?}}{ [2 x [[S_DOUBLE_TY]]], [[S_DOUBLE_TY]], i32, [2 x i32]
 // CHECK-DAG: [[KMP_TASK_MAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [[PRIVATES_MAIN_TY]] }
 // CHECK-DAG: [[S_INT_TY:%.+]] = type { i32 }
 // CHECK-DAG: [[CAP_TMAIN_TY:%.+]] = type { i8 }
-// CHECK-DAG: [[PRIVATES_TMAIN_TY:%.+]] = type { i32, [2 x i32], [2 x [[S_INT_TY]]], [[S_INT_TY]] }
-// CHECK-DAG: [[KMP_TASK_TMAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [[PRIVATES_TMAIN_TY]] }
+// CHECK-DAG: [[PRIVATES_TMAIN_TY:%.+]] = type { i32, [2 x i32], [2 x [[S_INT_TY]]], [[S_INT_TY]], [104 x i8] }
+// CHECK-DAG: [[KMP_TASK_TMAIN_TY:%.+]] = type { [[KMP_TASK_T_TY]], [{{[0-9]+}} x i8], [[PRIVATES_TMAIN_TY]] }
 template <typename T>
 T tmain() {
   S<T> test;
-  T t_var = T();
+  T t_var __attribute__((aligned(128))) = T();
   T vec[] = {1, 2};
   S<T> s_arr[] = {1, 2};
   S<T> var(3);
@@ -49,18 +49,20 @@ T tmain() {
 }
 
 int main() {
+  static int sivar;
 #ifdef LAMBDA
   // LAMBDA: [[G:@.+]] = global double
   // LAMBDA-LABEL: @main
   // LAMBDA: call{{( x86_thiscallcc)?}} void [[OUTER_LAMBDA:@.+]](
   [&]() {
   // LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
-  // LAMBDA: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 40, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+  // LAMBDA: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 48, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
 // LAMBDA: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
 // LAMBDA: [[G_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
+// LAMBDA: [[SIVAR_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
 // LAMBDA: call i32 @__kmpc_omp_task(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]])
 // LAMBDA: ret
-#pragma omp task private(g)
+#pragma omp task private(g, sivar)
   {
     // LAMBDA: define {{.+}} void [[INNER_LAMBDA:@.+]](%{{.+}}* [[ARG_PTR:%.+]])
     // LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
@@ -68,14 +70,20 @@ int main() {
     // LAMBDA: [[G_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
     // LAMBDA: [[G_REF:%.+]] = load double*, double** [[G_PTR_REF]]
     // LAMBDA: store double 2.0{{.+}}, double* [[G_REF]]
+    // LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+    // LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
+    // LAMBDA: store i{{[0-9]+}} 3, i{{[0-9]+}}* [[SIVAR_REF]]
 
-    // LAMBDA: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}*)
+    // LAMBDA: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}* noalias)
     g = 1;
+    sivar = 2;
     // LAMBDA: store double 1.0{{.+}}, double* %{{.+}},
+    // LAMBDA: store i{{[0-9]+}} 2, i{{[0-9]+}}* %{{.+}},
     // LAMBDA: call void [[INNER_LAMBDA]](%
     // LAMBDA: ret
     [&]() {
       g = 2;
+      sivar = 3;
     }();
   }
   }();
@@ -86,26 +94,34 @@ int main() {
   // BLOCKS: call void {{%.+}}(i8
   ^{
   // BLOCKS: define{{.*}} internal{{.*}} void {{.+}}(i8*
-  // BLOCKS: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 40, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+  // BLOCKS: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc(%{{[^ ]+}} @{{[^,]+}}, i32 %{{[^,]+}}, i32 1, i64 48, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, %{{[^*]+}}*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
   // BLOCKS: [[PRIVATES:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* %{{.+}}, i{{.+}} 0, i{{.+}} 1
   // BLOCKS: [[G_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
+  // BLOCKS: [[SIVAR_PRIVATE_ADDR:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
   // BLOCKS: call i32 @__kmpc_omp_task(%{{.+}}* @{{.+}}, i32 %{{.+}}, i8* [[RES]])
   // BLOCKS: ret
-#pragma omp task private(g)
+#pragma omp task private(g, sivar)
   {
     // BLOCKS: define {{.+}} void {{@.+}}(i8*
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
     // BLOCKS: store double 2.0{{.+}}, double*
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
+    // BLOCKS: store i{{[0-9]+}} 4, i{{[0-9]+}}*
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: ret
 
-    // BLOCKS: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}*)
+    // BLOCKS: define internal i32 [[TASK_ENTRY]](i32, %{{.+}}* noalias)
     g = 1;
+    sivar = 3;
     // BLOCKS: store double 1.0{{.+}}, double* %{{.+}},
     // BLOCKS-NOT: [[G]]{{[[^:word:]]}}
+    // BLOCKS: store i{{[0-9]+}} 3, i{{[0-9]+}}* %{{.+}},
+    // BLOCKS-NOT: [[SIVAR]]{{[[^:word:]]}}
     // BLOCKS: call void {{%.+}}(i8
     ^{
       g = 2;
+      sivar = 4;
     }();
   }
   }();
@@ -116,11 +132,14 @@ int main() {
   int vec[] = {1, 2};
   S<double> s_arr[] = {1, 2};
   S<double> var(3);
-#pragma omp task private(var, t_var, s_arr, vec, s_arr, var)
+#pragma omp task private(var, t_var, s_arr, vec, s_arr, var, sivar)
   {
     vec[0] = t_var;
     s_arr[0] = var;
+    sivar = 8;
   }
+#pragma omp task
+  g+=1;
   return tmain<int>();
 #endif
 }
@@ -152,12 +171,8 @@ int main() {
 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
 
 // Constructors for s_arr and var.
-// var;
-// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
-// CHECK: call void [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF:%.+]])
-
 // a_arr;
-// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 // CHECK: getelementptr inbounds [2 x [[S_DOUBLE_TY]]], [2 x [[S_DOUBLE_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} 2
 // CHECK: call void [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]* [[S_ARR_CUR:%.+]])
@@ -165,12 +180,17 @@ int main() {
 // CHECK: icmp eq
 // CHECK: br i1
 
+// var;
+// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// CHECK: call void [[S_DOUBLE_TY_DEF_CONSTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF:%.+]])
+
 // Provide pointer to destructor function, which will destroy private variables at the end of the task.
 // CHECK: [[DESTRUCTORS_REF:%.+]] = getelementptr inbounds [[KMP_TASK_T_TY]], [[KMP_TASK_T_TY]]* [[TASK]], i{{.+}} 0, i{{.+}} 3
 // CHECK: store i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_MAIN_TY]]*)* [[DESTRUCTORS:@.+]] to i32 (i32, i8*)*), i32 (i32, i8*)** [[DESTRUCTORS_REF]],
 
 // Start task.
 // CHECK: call i32 @__kmpc_omp_task([[LOC]], i32 [[GTID]], i8* [[RES]])
+// CHECK: call i32 @__kmpc_omp_task([[LOC]], i32 [[GTID]], i8*
 
 // CHECK: = call i{{.+}} [[TMAIN_INT:@.+]]()
 
@@ -183,14 +203,14 @@ int main() {
 // CHECK: ret
 //
 
-// CHECK: define internal void [[PRIVATES_MAP_FN:@.+]]([[PRIVATES_MAIN_TY]]* noalias, [[S_DOUBLE_TY]]** noalias, i32** noalias, [2 x [[S_DOUBLE_TY]]]** noalias, [2 x i32]** noalias)
+// CHECK: define internal void [[PRIVATES_MAP_FN:@.+]]([[PRIVATES_MAIN_TY]]* noalias, [[S_DOUBLE_TY]]** noalias, i32** noalias, [2 x [[S_DOUBLE_TY]]]** noalias, [2 x i32]** noalias, i32** noalias)
 // CHECK: [[PRIVATES:%.+]] = load [[PRIVATES_MAIN_TY]]*, [[PRIVATES_MAIN_TY]]**
-// CHECK: [[PRIV_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 0
-// CHECK: [[ARG1:%.+]] = load [[S_DOUBLE_TY]]**, [[S_DOUBLE_TY]]*** {{.+}},
-// CHECK: store [[S_DOUBLE_TY]]* [[PRIV_VAR]], [[S_DOUBLE_TY]]** [[ARG1]],
-// CHECK: [[PRIV_S_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 1
+// CHECK: [[PRIV_S_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 0
 // CHECK: [[ARG3:%.+]] = load [2 x [[S_DOUBLE_TY]]]**, [2 x [[S_DOUBLE_TY]]]*** %{{.+}},
 // CHECK: store [2 x [[S_DOUBLE_TY]]]* [[PRIV_S_VAR]], [2 x [[S_DOUBLE_TY]]]** [[ARG3]],
+// CHECK: [[PRIV_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 1
+// CHECK: [[ARG1:%.+]] = load [[S_DOUBLE_TY]]**, [[S_DOUBLE_TY]]*** {{.+}},
+// CHECK: store [[S_DOUBLE_TY]]* [[PRIV_VAR]], [[S_DOUBLE_TY]]** [[ARG1]],
 // CHECK: [[PRIV_T_VAR:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i32 0, i32 2
 // CHECK: [[ARG2:%.+]] = load i32**, i32*** %{{.+}},
 // CHECK: store i32* [[PRIV_T_VAR]], i32** [[ARG2]],
@@ -199,39 +219,42 @@ int main() {
 // CHECK: store [2 x i32]* [[PRIV_VEC]], [2 x i32]** [[ARG4]],
 // CHECK: ret void
 
-// CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_MAIN_TY]]*)
+// CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_MAIN_TY]]* noalias)
 
 // CHECK: [[PRIV_VAR_ADDR:%.+]] = alloca [[S_DOUBLE_TY]]*,
 // CHECK: [[PRIV_T_VAR_ADDR:%.+]] = alloca i32*,
 // CHECK: [[PRIV_S_ARR_ADDR:%.+]] = alloca [2 x [[S_DOUBLE_TY]]]*,
 // CHECK: [[PRIV_VEC_ADDR:%.+]] = alloca [2 x i32]*,
-// CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_MAIN_TY]]*, [[S_DOUBLE_TY]]**, i32**, [2 x [[S_DOUBLE_TY]]]**, [2 x i32]**)* [[PRIVATES_MAP_FN]] to void (i8*, ...)*), void (i8*, ...)** [[MAP_FN_ADDR:%.+]],
+// CHECK: [[PRIV_SIVAR_ADDR:%.+]] = alloca i32*,
+// CHECK: store void (i8*, ...)* bitcast (void ([[PRIVATES_MAIN_TY]]*, [[S_DOUBLE_TY]]**, i32**, [2 x [[S_DOUBLE_TY]]]**, [2 x i32]**, i32**)* [[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* %{{.+}}, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]], i32** [[PRIV_T_VAR_ADDR]], [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]])
+// CHECK: call void (i8*, ...) [[MAP_FN]](i8* %{{.+}}, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]], i32** [[PRIV_T_VAR_ADDR]], [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]], [2 x i32]** [[PRIV_VEC_ADDR]], i32** [[PRIV_SIVAR_ADDR]])
 // CHECK: [[PRIV_VAR:%.+]] = load [[S_DOUBLE_TY]]*, [[S_DOUBLE_TY]]** [[PRIV_VAR_ADDR]],
 // CHECK: [[PRIV_T_VAR:%.+]] = load i32*, i32** [[PRIV_T_VAR_ADDR]],
 // CHECK: [[PRIV_S_ARR:%.+]] = load [2 x [[S_DOUBLE_TY]]]*, [2 x [[S_DOUBLE_TY]]]** [[PRIV_S_ARR_ADDR]],
 // CHECK: [[PRIV_VEC:%.+]] = load [2 x i32]*, [2 x i32]** [[PRIV_VEC_ADDR]],
+// CHECK: [[PRIV_SIVAR:%.+]] = load i32*, i32** [[PRIV_SIVAR_ADDR]],
 
 // Privates actually are used.
 // CHECK-DAG: [[PRIV_VAR]]
 // CHECK-DAG: [[PRIV_T_VAR]]
 // CHECK-DAG: [[PRIV_S_ARR]]
 // CHECK-DAG: [[PRIV_VEC]]
+// CHECK_DAG: [[PRIV_SIVAR]]
 
 // CHECK: ret
 
-// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_MAIN_TY]]*)
+// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_MAIN_TY]]* noalias)
 // CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_MAIN_TY]], [[KMP_TASK_MAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
-// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
-// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 0
+// CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_MAIN_TY]], [[PRIVATES_MAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 1
+// CHECK: call void [[S_DOUBLE_TY_DESTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF]])
 // CHECK: getelementptr inbounds [2 x [[S_DOUBLE_TY]]], [2 x [[S_DOUBLE_TY]]]* [[PRIVATE_S_ARR_REF]], i{{.+}} 0, i{{.+}} 0
 // CHECK: getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} 2
 // CHECK: [[PRIVATE_S_ARR_ELEM_REF:%.+]] = getelementptr inbounds [[S_DOUBLE_TY]], [[S_DOUBLE_TY]]* %{{.+}}, i{{.+}} -1
 // CHECK: call void [[S_DOUBLE_TY_DESTR]]([[S_DOUBLE_TY]]* [[PRIVATE_S_ARR_ELEM_REF]])
 // CHECK: icmp eq
 // CHECK: br i1
-// CHECK: call void [[S_DOUBLE_TY_DESTR]]([[S_DOUBLE_TY]]* [[PRIVATE_VAR_REF]])
 // CHECK: ret i32
 
 // CHECK: define {{.*}} i{{[0-9]+}} [[TMAIN_INT]]()
@@ -252,13 +275,13 @@ int main() {
 //         [[KMP_TASK_T_TY]] task_data;
 //         [[KMP_TASK_TMAIN_TY]] privates;
 //       };
-// CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID]], i32 1, i64 56, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_TMAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
+// CHECK: [[RES:%.+]] = call i8* @__kmpc_omp_task_alloc([[LOC]], i32 [[GTID]], i32 1, i64 256, i64 1, i32 (i32, i8*)* bitcast (i32 (i32, [[KMP_TASK_TMAIN_TY]]*)* [[TASK_ENTRY:@[^ ]+]] to i32 (i32, i8*)*))
 // CHECK: [[RES_KMP_TASK:%.+]] = bitcast i8* [[RES]] to [[KMP_TASK_TMAIN_TY]]*
 
 // CHECK: [[TASK:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
 
 // Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).
-// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
 
 // Constructors for s_arr and var.
 // a_arr;
@@ -306,7 +329,7 @@ int main() {
 // CHECK: store [[S_INT_TY]]* [[PRIV_VAR]], [[S_INT_TY]]** [[ARG4]],
 // CHECK: ret void
 
-// CHECK: define internal i32 [[TASK_ENTRY]](i32, [[KMP_TASK_TMAIN_TY]]*)
+// 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]*,
@@ -328,8 +351,8 @@ int main() {
 
 // CHECK: ret
 
-// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_TMAIN_TY]]*)
-// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
+// CHECK: define internal i32 [[DESTRUCTORS]](i32, [[KMP_TASK_TMAIN_TY]]* noalias)
+// CHECK: [[PRIVATES:%.+]] = getelementptr inbounds [[KMP_TASK_TMAIN_TY]], [[KMP_TASK_TMAIN_TY]]* [[RES_KMP_TASK:%.+]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
 // CHECK: [[PRIVATE_S_ARR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 2
 // CHECK: [[PRIVATE_VAR_REF:%.+]] = getelementptr inbounds [[PRIVATES_TMAIN_TY]], [[PRIVATES_TMAIN_TY]]* [[PRIVATES]], i{{.+}} 0, i{{.+}} 3
 // CHECK: call void [[S_INT_TY_DESTR]]([[S_INT_TY]]* [[PRIVATE_VAR_REF]])
diff --git a/test/OpenMP/task_private_messages.cpp b/test/OpenMP/task_private_messages.cpp
index c2c1f519d0c1..6bce1352b13e 100644
--- a/test/OpenMP/task_private_messages.cpp
+++ b/test/OpenMP/task_private_messages.cpp
@@ -14,7 +14,7 @@ class S2 {
 
 public:
   S2() : a(0) {}
-  static float S2s;
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
 };
 const S2 b;
 const S2 ba[5];
@@ -64,7 +64,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i;                                          // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp task private                               // expected-error {{expected '(' after 'private'}}
 #pragma omp task private(                              // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
 #pragma omp task private()                             // expected-error {{expected expression}}
@@ -78,7 +78,7 @@ int main(int argc, char **argv) {
 #pragma omp task private(ba)
 #pragma omp task private(ca)           // expected-error {{shared variable cannot be private}}
 #pragma omp task private(da)           // expected-error {{shared variable cannot be private}}
-#pragma omp task private(S2::S2s)
+#pragma omp task private(S2::S2s)      // expected-error {{shared variable cannot be private}}
 #pragma omp task private(e, g)         // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
 #pragma omp task private(threadvar, B::x)    // expected-error 2 {{threadprivate or thread local variable cannot be private}}
 #pragma omp task shared(i), private(i) // expected-error {{shared variable cannot be private}} expected-note {{defined as shared}}
@@ -86,13 +86,16 @@ int main(int argc, char **argv) {
 #pragma omp task firstprivate(i) private(i) // expected-error {{firstprivate variable cannot be private}} expected-note {{defined as firstprivate}}
   foo();
 #pragma omp task private(i)
-#pragma omp task private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type 'int &'}}
+#pragma omp task private(j)
   foo();
 #pragma omp task firstprivate(i)
   for (int k = 0; k < 10; ++k) {
 #pragma omp task private(i)
     foo();
   }
+  static int m;
+#pragma omp task private(m) // OK
+    foo();
 
   return 0;
 }
diff --git a/test/OpenMP/taskgroup_codegen.cpp b/test/OpenMP/taskgroup_codegen.cpp
index c04e4c723b8d..d1bc2aafc7b4 100644
--- a/test/OpenMP/taskgroup_codegen.cpp
+++ b/test/OpenMP/taskgroup_codegen.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -verify -fopenmp -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
-// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -gline-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp -fexceptions -fcxx-exceptions -debug-info-kind=line-tables-only -x c++ -emit-llvm %s -o - | FileCheck %s --check-prefix=TERM_DEBUG
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
 #ifndef HEADER
diff --git a/test/OpenMP/taskloop_ast_print.cpp b/test/OpenMP/taskloop_ast_print.cpp
new file mode 100644
index 000000000000..06164ab168d9
--- /dev/null
+++ b/test/OpenMP/taskloop_ast_print.cpp
@@ -0,0 +1,74 @@
+// RUN: %clang_cc1 -verify -fopenmp -ast-print %s | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -std=c++11 -include-pch %t -fsyntax-only -verify %s -ast-print | FileCheck %s
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void foo() {}
+
+template <class T, int N>
+T tmain(T argc) {
+  T b = argc, c, d, e, f, g;
+  static T a;
+// CHECK: static T a;
+#pragma omp taskloop if(taskloop: argc > N) default(shared) untied priority(N) grainsize(N)
+  // CHECK-NEXT: #pragma omp taskloop if(taskloop: argc > N) default(shared) untied priority(N) grainsize(N)
+  for (int i = 0; i < 2; ++i)
+    a = 2;
+// CHECK-NEXT: for (int i = 0; i < 2; ++i)
+// CHECK-NEXT: a = 2;
+#pragma omp parallel
+#pragma omp taskloop private(argc, b), firstprivate(c, d), lastprivate(d, f) collapse(N) shared(g) if (c) final(d) mergeable priority(f) nogroup num_tasks(N)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
+            foo();
+  // CHECK-NEXT: #pragma omp parallel
+  // CHECK-NEXT: #pragma omp taskloop private(argc,b) firstprivate(c,d) lastprivate(d,f) collapse(N) shared(g) if(c) final(d) mergeable priority(f) nogroup num_tasks(N)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: foo();
+  return T();
+}
+
+int main(int argc, char **argv) {
+  int b = argc, c, d, e, f, g;
+  static int a;
+// CHECK: static int a;
+#pragma omp taskloop if(taskloop: a) default(none) shared(a) final(b) priority(5) nogroup num_tasks(argc)
+  // CHECK-NEXT: #pragma omp taskloop if(taskloop: a) default(none) shared(a) final(b) priority(5) nogroup num_tasks(argc)
+  for (int i = 0; i < 2; ++i)
+    a = 2;
+// CHECK-NEXT: for (int i = 0; i < 2; ++i)
+// CHECK-NEXT: a = 2;
+#pragma omp parallel
+#pragma omp taskloop private(argc, b), firstprivate(argv, c), lastprivate(d, f) collapse(2) shared(g) if(argc) mergeable priority(argc) grainsize(argc)
+  for (int i = 0; i < 10; ++i)
+    for (int j = 0; j < 10; ++j)
+      foo();
+  // CHECK-NEXT: #pragma omp parallel
+  // CHECK-NEXT: #pragma omp taskloop private(argc,b) firstprivate(argv,c) lastprivate(d,f) collapse(2) shared(g) if(argc) mergeable priority(argc) grainsize(argc)
+  // CHECK-NEXT: for (int i = 0; i < 10; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
+  // CHECK-NEXT: foo();
+  return (tmain<int, 5>(argc) + tmain<char, 1>(argv[0][0]));
+}
+
+#endif
diff --git a/test/OpenMP/taskloop_collapse_messages.cpp b/test/OpenMP/taskloop_collapse_messages.cpp
new file mode 100644
index 000000000000..f33da11f5ed9
--- /dev/null
+++ b/test/OpenMP/taskloop_collapse_messages.cpp
@@ -0,0 +1,83 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp taskloop collapse // expected-error {{expected '(' after 'collapse'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp taskloop collapse (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  #pragma omp taskloop collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp taskloop', but found only 1}}
+  // expected-error@+3 2 {{directive '#pragma omp taskloop' cannot contain more than one 'collapse' clause}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop collapse (foobool(argc)), collapse (true), collapse (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop collapse (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse (1)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop collapse (2) // expected-note {{as specified in 'collapse' clause}}
+  foo(); // expected-error {{expected 2 for loops after '#pragma omp taskloop'}}
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop collapse // expected-error {{expected '(' after 'collapse'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop collapse ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop collapse () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop collapse (4 // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-note {{as specified in 'collapse' clause}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4]; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+  #pragma omp taskloop collapse (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}  expected-note {{as specified in 'collapse' clause}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4]; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+  #pragma omp taskloop collapse (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp taskloop' cannot contain more than one 'collapse' clause}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  #pragma omp taskloop collapse (foobool(argc)), collapse (true), collapse (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop collapse (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp taskloop collapse (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp taskloop' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp taskloop collapse(collapse(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  #pragma omp taskloop collapse (2) // expected-note {{as specified in 'collapse' clause}}
+  foo(); // expected-error {{expected 2 for loops after '#pragma omp taskloop'}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
+  return tmain<int, char, 1, 0>(argc, argv);
+}
+
diff --git a/test/OpenMP/taskloop_final_messages.cpp b/test/OpenMP/taskloop_final_messages.cpp
new file mode 100644
index 000000000000..a2d7cdae9673
--- /dev/null
+++ b/test/OpenMP/taskloop_final_messages.cpp
@@ -0,0 +1,90 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+#pragma omp taskloop final // expected-error {{expected '(' after 'final'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final() // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc > 0 ? argv[1] : argv[2])
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(foobool(argc)), final(true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'final' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc)
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+#pragma omp taskloop final // expected-error {{expected '(' after 'final'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final() // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc > 0 ? argv[1] : argv[2])
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(foobool(argc)), final(true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'final' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop final(if (tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_firstprivate_messages.cpp b/test/OpenMP/taskloop_firstprivate_messages.cpp
new file mode 100644
index 000000000000..e2e87e4697a5
--- /dev/null
+++ b/test/OpenMP/taskloop_firstprivate_messages.cpp
@@ -0,0 +1,313 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+  S2(const S2 &s2) : a(s2.a) {}
+  static float S2s;
+  static const float S2sc;
+};
+const float S2::S2sc = 0;
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+  S3 &operator=(const S3 &s3);
+
+public:
+  S3() : a(0) {} // expected-note 2 {{candidate constructor not viable: requires 0 arguments, but 1 was provided}}
+  S3(S3 &s3) : a(s3.a) {} // expected-note 2 {{candidate constructor not viable: 1st argument ('const S3') would lose const qualifier}}
+};
+const S3 c;
+const S3 ca[5];
+extern const int f;
+class S4 {
+  int a;
+  S4();
+  S4(const S4 &s4); // expected-note 2 {{implicitly declared private here}}
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5(const S5 &s5) : a(s5.a) {} // expected-note 4 {{implicitly declared private here}}
+
+public:
+  S5() : a(0) {}
+  S5(int v) : a(v) {}
+};
+class S6 {
+  int a;
+  S6() : a(0) {}
+
+public:
+  S6(const S6 &s6) : a(s6.a) {}
+  S6(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(int argc, char **argv) {
+  I e(4);
+  C g(5);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop firstprivate // expected-error {{expected '(' after 'firstprivate'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(a, b) // expected-error {{firstprivate variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(h) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop firstprivate(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop firstprivate(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(g) firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel private(i)
+#pragma omp taskloop firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be firstprivate, predetermined as private}}
+    foo();
+#pragma omp parallel reduction(+ : i)
+#pragma omp taskloop firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be firstprivate, predetermined as private}}
+    foo();
+  return 0;
+}
+
+void bar(S4 a[2]) {
+#pragma omp parallel
+#pragma omp taskloop firstprivate(a)
+  for (int i = 0; i < 2; ++i)
+    foo();
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace B {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  const int d = 5;
+  const int da[5] = {0};
+  S4 e(4);
+  S5 g(5);
+  S3 m;
+  S6 n(2);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop firstprivate // expected-error {{expected '(' after 'firstprivate'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate() // expected-error {{expected expression}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argc)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(a, b, c, d, f) // expected-error {{firstprivate variable with incomplete type 'S1'}} expected-error {{no matching constructor for initialization of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(argv[1]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(2 * 2) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(ba) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(ca) // expected-error {{no matching constructor for initialization of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(da) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+  int xa;
+#pragma omp parallel
+#pragma omp taskloop firstprivate(xa) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(S2::S2s) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(S2::S2sc) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop safelen(5) // expected-error {{unexpected OpenMP clause 'safelen' in directive '#pragma omp taskloop'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(m) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(h) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop private(xa), firstprivate(xa) // expected-error {{private variable cannot be firstprivate}} expected-note {{defined as private}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i)    // expected-error {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be firstprivate, predetermined as private}}
+    foo();
+#pragma omp parallel shared(xa)
+#pragma omp taskloop firstprivate(xa) // OK: may be firstprivate
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(j)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(g) firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(n) firstprivate(n) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop firstprivate(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel private(i)
+#pragma omp taskloop firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be firstprivate, predetermined as private}}
+    foo();
+#pragma omp parallel reduction(+ : i)
+#pragma omp taskloop firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be firstprivate, predetermined as private}}
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(B::x) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+  static int si;
+#pragma omp taskloop firstprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
+
+  return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
+}
+
diff --git a/test/OpenMP/taskloop_grainsize_messages.cpp b/test/OpenMP/taskloop_grainsize_messages.cpp
new file mode 100644
index 000000000000..047f925b8250
--- /dev/null
+++ b/test/OpenMP/taskloop_grainsize_messages.cpp
@@ -0,0 +1,99 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp taskloop grainsize // expected-error {{expected '(' after 'grainsize'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (foobool(argc)), grainsize (true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'grainsize' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(0) // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(-1) // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(argc) num_tasks(argc) // expected-error {{'num_tasks' and 'grainsize' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'grainsize' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop grainsize // expected-error {{expected '(' after 'grainsize'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (foobool(argc)), grainsize (true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'grainsize' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(0)  // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(-1) // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop grainsize(argc) num_tasks(argc) // expected-error {{'num_tasks' and 'grainsize' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'grainsize' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_lastprivate_messages.cpp b/test/OpenMP/taskloop_lastprivate_messages.cpp
new file mode 100644
index 000000000000..e4364448159c
--- /dev/null
+++ b/test/OpenMP/taskloop_lastprivate_messages.cpp
@@ -0,0 +1,287 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+  S2(S2 &s2) : a(s2.a) {}
+  const S2 &operator =(const S2&) const;
+  S2 &operator =(const S2&);
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
+  static const float S2sc;
+};
+const float S2::S2sc = 0; // expected-note {{static data member is predetermined as shared}}
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+  S3 &operator=(const S3 &s3); // expected-note 2 {{implicitly declared private here}}
+
+public:
+  S3() : a(0) {}
+  S3(S3 &s3) : a(s3.a) {}
+};
+const S3 c;         // expected-note {{global variable is predetermined as shared}}
+const S3 ca[5];     // expected-note {{global variable is predetermined as shared}}
+extern const int f; // expected-note {{global variable is predetermined as shared}}
+class S4 {
+  int a;
+  S4();             // expected-note 3 {{implicitly declared private here}}
+  S4(const S4 &s4);
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5() : a(0) {} // expected-note {{implicitly declared private here}}
+
+public:
+  S5(const S5 &s5) : a(s5.a) {}
+  S5(int v) : a(v) {}
+};
+class S6 {
+  int a;
+  S6() : a(0) {}
+
+public:
+  S6(const S6 &s6) : a(s6.a) {}
+  S6(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(int argc, char **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop lastprivate // expected-error {{expected '(' after 'lastprivate'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(a, b) // expected-error {{lastprivate variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(e, g) // expected-error 2 {{calling a private constructor of class 'S4'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(h) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop lastprivate(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop lastprivate(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  return 0;
+}
+
+void bar(S4 a[2]) {
+#pragma omp parallel
+#pragma omp taskloop lastprivate(a)
+  for (int i = 0; i < 2; ++i)
+    foo();
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace B {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  const int d = 5;       // expected-note {{constant variable is predetermined as shared}}
+  const int da[5] = {0}; // expected-note {{constant variable is predetermined as shared}}
+  S4 e(4);
+  S5 g(5);
+  S3 m;
+  S6 n(2);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop lastprivate // expected-error {{expected '(' after 'lastprivate'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate() // expected-error {{expected expression}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argc)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(a, b, c, d, f) // expected-error {{lastprivate variable with incomplete type 'S1'}} expected-error 3 {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(argv[1]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(2 * 2) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(ba)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(ca) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(da) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+  int xa;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(xa) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(S2::S2s) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(S2::S2sc) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop safelen(5) // expected-error {{unexpected OpenMP clause 'safelen' in directive '#pragma omp taskloop'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(m) // expected-error {{'operator=' is a private member of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(h) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(B::x) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop private(xa), lastprivate(xa) // expected-error {{private variable cannot be lastprivate}} expected-note {{defined as private}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(i)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel private(xa)
+#pragma omp taskloop lastprivate(xa)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel reduction(+ : xa)
+#pragma omp taskloop lastprivate(xa)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(j)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop firstprivate(m) lastprivate(m) // expected-error {{'operator=' is a private member of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop lastprivate(n) firstprivate(n) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+  static int si;
+#pragma omp taskloop lastprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
+  return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
+}
diff --git a/test/OpenMP/taskloop_loop_messages.cpp b/test/OpenMP/taskloop_loop_messages.cpp
new file mode 100644
index 000000000000..02518e572f3a
--- /dev/null
+++ b/test/OpenMP/taskloop_loop_messages.cpp
@@ -0,0 +1,738 @@
+// RUN: %clang_cc1 -fsyntax-only -fopenmp -x c++ -std=c++11 -fexceptions -fcxx-exceptions -verify %s
+
+class S {
+  int a;
+  S() : a(0) {}
+
+public:
+  S(int v) : a(v) {}
+  S(const S &s) : a(s.a) {}
+};
+
+static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
+#pragma omp threadprivate(sii)
+static int globalii;
+
+// Currently, we cannot use "0" for global register variables.
+// register int reg0 __asm__("0");
+int reg0;
+
+int test_iteration_spaces() {
+  const int N = 100;
+  float a[N], b[N], c[N];
+  int ii, jj, kk;
+  float fii;
+  double dii;
+  register int reg; // expected-warning {{'register' storage class specifier is deprecated}}
+#pragma omp parallel
+#pragma omp taskloop
+  for (int i = 0; i < 10; i += 1) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop
+  for (char i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop
+  for (char i = 0; i < 10; i += '\1') {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop
+  for (long long i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+// expected-error@+2 {{expression must have integral or unscoped enumeration type, not 'double'}}
+#pragma omp taskloop
+  for (long long i = 0; i < 10; i += 1.5) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop
+  for (long long i = 0; i < 'z'; i += 1u) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or random access iterator type}}
+#pragma omp taskloop
+  for (float fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or random access iterator type}}
+#pragma omp taskloop
+  for (double fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (int &ref = ii; ref < 10; ref++) {
+  }
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (int i; i < 10; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (int i = 0, j = 0; i < 10; ++i)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-warning@+3 {{expression result unused}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (ii + 1; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (c[ii] = 0; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// Ok to skip parenthesises.
+#pragma omp taskloop
+  for (((ii)) = 0; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop
+  for (int i = 0; i; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+3 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'i'}}
+#pragma omp taskloop
+  for (int i = 0; jj < kk; ii++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop
+  for (int i = 0; !!i; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop
+  for (int i = 0; i != 1; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop
+  for (int i = 0;; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// Ok.
+#pragma omp taskloop
+  for (int i = 11; i > 10; i--)
+    c[i] = a[i];
+
+#pragma omp parallel
+// Ok.
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i)
+    c[i] = a[i];
+
+#pragma omp parallel
+// Ok.
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ++jj)
+    c[ii] = a[jj];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ++++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// Ok but undefined behavior (in general, cannot check that incr
+// is really loop-invariant).
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii = ii + ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{expression must have integral or unscoped enumeration type, not 'float'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii = ii + 1.0f)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// Ok - step was converted to integer type.
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii = ii + (int)1.1f)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; jj = ii + 2)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-warning@+3 {{relational comparison result unused}}
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii<10; jj> kk + 2)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10;)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-warning@+3 {{expression result unused}}
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; !ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii ? ++ii : ++jj)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii = ii < 10)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii = ii + 0)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; ii = ii + (int)(0.8 - 0.45))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; (ii) < 10; ii -= 25)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; (ii < 10); ii -= 0)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; ii > 10; (ii += 0))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; ii < 10; (ii) = (1 - 1) + (ii))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for ((ii = 0); ii > 10; (ii -= 0))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (ii = 0; (ii < 10); (ii -= 0))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+2  {{defined as firstprivate}}
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be firstprivate, predetermined as private}}
+#pragma omp taskloop firstprivate(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{unexpected OpenMP clause 'linear' in directive '#pragma omp taskloop'}}
+// expected-note@+1 {{defined as linear}}
+#pragma omp taskloop linear(ii)
+// expected-error@+1 {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be linear, predetermined as private}}
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+#pragma omp taskloop private(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+#pragma omp taskloop lastprivate(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+  {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp taskloop' directive may not be threadprivate or thread local, predetermined as private}}
+#pragma omp taskloop
+    for (sii = 0; sii < 10; sii += 1)
+      c[sii] = a[sii];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop
+    for (reg0 = 0; reg0 < 10; reg0 += 1)
+      c[reg0] = a[reg0];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop
+    for (reg = 0; reg < 10; reg += 1)
+      c[reg] = a[reg];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop
+    for (globalii = 0; globalii < 10; globalii += 1)
+      c[globalii] = a[globalii];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop collapse(2)
+    for (ii = 0; ii < 10; ii += 1)
+    for (globalii = 0; globalii < 10; globalii += 1)
+      c[globalii] += a[globalii] + ii;
+  }
+
+#pragma omp parallel
+// expected-error@+2 {{statement after '#pragma omp taskloop' must be a for loop}}
+#pragma omp taskloop
+  for (auto &item : a) {
+    item = item + 1;
+  }
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'i' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (unsigned i = 9; i < 10; i--) {
+    c[i] = a[i] + b[i];
+  }
+
+  int(*lb)[4] = nullptr;
+#pragma omp parallel
+#pragma omp taskloop
+  for (int(*p)[4] = lb; p < lb + 8; ++p) {
+  }
+
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (int a{0}; a < 10; ++a) {
+  }
+
+  return 0;
+}
+
+// Iterators allowed in openmp for-loops.
+namespace std {
+struct random_access_iterator_tag {};
+template <class Iter>
+struct iterator_traits {
+  typedef typename Iter::difference_type difference_type;
+  typedef typename Iter::iterator_category iterator_category;
+};
+template <class Iter>
+typename iterator_traits<Iter>::difference_type
+distance(Iter first, Iter last) { return first - last; }
+}
+class Iter0 {
+public:
+  Iter0() {}
+  Iter0(const Iter0 &) {}
+  Iter0 operator++() { return *this; }
+  Iter0 operator--() { return *this; }
+  bool operator<(Iter0 a) { return true; }
+};
+// expected-note@+2 {{candidate function not viable: no known conversion from 'GoodIter' to 'Iter0' for 1st argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'Iter0' for 1st argument}}
+int operator-(Iter0 a, Iter0 b) { return 0; }
+class Iter1 {
+public:
+  Iter1(float f = 0.0f, double d = 0.0) {}
+  Iter1(const Iter1 &) {}
+  Iter1 operator++() { return *this; }
+  Iter1 operator--() { return *this; }
+  bool operator<(Iter1 a) { return true; }
+  bool operator>=(Iter1 a) { return false; }
+};
+class GoodIter {
+public:
+  GoodIter() {}
+  GoodIter(const GoodIter &) {}
+  GoodIter(int fst, int snd) {}
+  GoodIter &operator=(const GoodIter &that) { return *this; }
+  GoodIter &operator=(const Iter0 &that) { return *this; }
+  GoodIter &operator+=(int x) { return *this; }
+  GoodIter &operator-=(int x) { return *this; }
+  explicit GoodIter(void *) {}
+  GoodIter operator++() { return *this; }
+  GoodIter operator--() { return *this; }
+  bool operator!() { return true; }
+  bool operator<(GoodIter a) { return true; }
+  bool operator<=(GoodIter a) { return true; }
+  bool operator>=(GoodIter a) { return false; }
+  typedef int difference_type;
+  typedef std::random_access_iterator_tag iterator_category;
+};
+// expected-note@+2 {{candidate function not viable: no known conversion from 'Iter0' to 'GoodIter' for 2nd argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'GoodIter' for 1st argument}}
+int operator-(GoodIter a, GoodIter b) { return 0; }
+// expected-note@+1 3 {{candidate function not viable: requires single argument 'a', but 2 arguments were provided}}
+GoodIter operator-(GoodIter a) { return a; }
+// expected-note@+2 {{candidate function not viable: no known conversion from 'Iter0' to 'int' for 2nd argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'GoodIter' for 1st argument}}
+GoodIter operator-(GoodIter a, int v) { return GoodIter(); }
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter0' to 'GoodIter' for 1st argument}}
+GoodIter operator+(GoodIter a, int v) { return GoodIter(); }
+// expected-note@+2 {{candidate function not viable: no known conversion from 'GoodIter' to 'int' for 1st argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'int' for 1st argument}}
+GoodIter operator-(int v, GoodIter a) { return GoodIter(); }
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter0' to 'int' for 1st argument}}
+GoodIter operator+(int v, GoodIter a) { return GoodIter(); }
+
+int test_with_random_access_iterator() {
+  GoodIter begin, end;
+  Iter0 begin0, end0;
+#pragma omp parallel
+#pragma omp taskloop
+  for (GoodIter I = begin; I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (GoodIter &I = begin; I < end; ++I)
+    ++I;
+#pragma omp parallel
+#pragma omp taskloop
+  for (GoodIter I = begin; I >= end; --I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (GoodIter I(begin); I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (GoodIter I(nullptr); I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (GoodIter I(0); I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (GoodIter I(1, 2); I < end; ++I)
+    ++I;
+#pragma omp parallel
+#pragma omp taskloop
+  for (begin = GoodIter(0); begin < end; ++begin)
+    ++begin;
+// expected-error@+4 {{invalid operands to binary expression ('GoodIter' and 'Iter0')}}
+// expected-error@+3 {{could not calculate number of iterations calling 'operator-' with upper and lower loop bounds}}
+#pragma omp parallel
+#pragma omp taskloop
+  for (begin = begin0; begin < end; ++begin)
+    ++begin;
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (++begin; begin < end; ++begin)
+    ++begin;
+#pragma omp parallel
+#pragma omp taskloop
+  for (begin = end; begin < end; ++begin)
+    ++begin;
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'I'}}
+#pragma omp taskloop
+  for (GoodIter I = begin; I - I; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'I'}}
+#pragma omp taskloop
+  for (GoodIter I = begin; begin < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'I'}}
+#pragma omp taskloop
+  for (GoodIter I = begin; !I; ++I)
+    ++I;
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (GoodIter I = begin; I >= end; I = I + 1)
+    ++I;
+#pragma omp parallel
+#pragma omp taskloop
+  for (GoodIter I = begin; I >= end; I = I - 1)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'I'}}
+#pragma omp taskloop
+  for (GoodIter I = begin; I >= end; I = -I)
+    ++I;
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (GoodIter I = begin; I >= end; I = 2 + I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'I'}}
+#pragma omp taskloop
+  for (GoodIter I = begin; I >= end; I = 2 - I)
+    ++I;
+// In the following example, we cannot update the loop variable using '+='
+// expected-error@+3 {{invalid operands to binary expression ('Iter0' and 'int')}}
+#pragma omp parallel
+#pragma omp taskloop
+  for (Iter0 I = begin0; I < end0; ++I)
+    ++I;
+#pragma omp parallel
+// Initializer is constructor without params.
+// expected-error@+3 {{invalid operands to binary expression ('Iter0' and 'int')}}
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (Iter0 I; I < end0; ++I)
+    ++I;
+  Iter1 begin1, end1;
+// expected-error@+4 {{invalid operands to binary expression ('Iter1' and 'Iter1')}}
+// expected-error@+3 {{could not calculate number of iterations calling 'operator-' with upper and lower loop bounds}}
+#pragma omp parallel
+#pragma omp taskloop
+  for (Iter1 I = begin1; I < end1; ++I)
+    ++I;
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (Iter1 I = begin1; I >= end1; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+5 {{invalid operands to binary expression ('Iter1' and 'float')}}
+// expected-error@+4 {{could not calculate number of iterations calling 'operator-' with upper and lower loop bounds}}
+// Initializer is constructor with all default params.
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop
+  for (Iter1 I; I < end1; ++I) {
+  }
+  return 0;
+}
+
+template <typename IT, int ST>
+class TC {
+public:
+  int dotest_lt(IT begin, IT end) {
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+    for (IT I = begin; I < end; I = I + ST) {
+      ++I;
+    }
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+    for (IT I = begin; I <= end; I += ST) {
+      ++I;
+    }
+#pragma omp parallel
+#pragma omp taskloop
+    for (IT I = begin; I < end; ++I) {
+      ++I;
+    }
+  }
+
+  static IT step() {
+    return IT(ST);
+  }
+};
+template <typename IT, int ST = 0>
+int dotest_gt(IT begin, IT end) {
+#pragma omp parallel
+// expected-note@+3 2 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (IT I = begin; I >= end; I = I + ST) {
+    ++I;
+  }
+#pragma omp parallel
+// expected-note@+3 2 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (IT I = begin; I >= end; I += ST) {
+    ++I;
+  }
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop
+  for (IT I = begin; I >= end; ++I) {
+    ++I;
+  }
+
+#pragma omp parallel
+#pragma omp taskloop
+  for (IT I = begin; I < end; I += TC<int, ST>::step()) {
+    ++I;
+  }
+}
+
+void test_with_template() {
+  GoodIter begin, end;
+  TC<GoodIter, 100> t1;
+  TC<GoodIter, -100> t2;
+  t1.dotest_lt(begin, end);
+  t2.dotest_lt(begin, end);         // expected-note {{in instantiation of member function 'TC<GoodIter, -100>::dotest_lt' requested here}}
+  dotest_gt(begin, end);            // expected-note {{in instantiation of function template specialization 'dotest_gt<GoodIter, 0>' requested here}}
+  dotest_gt<unsigned, -10>(0, 100); // expected-note {{in instantiation of function template specialization 'dotest_gt<unsigned int, -10>' requested here}}
+}
+
+void test_loop_break() {
+  const int N = 100;
+  float a[N], b[N], c[N];
+#pragma omp parallel
+#pragma omp taskloop
+  for (int i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+    for (int j = 0; j < 10; ++j) {
+      if (a[i] > b[j])
+        break; // OK in nested loop
+    }
+    switch (i) {
+    case 1:
+      b[i]++;
+      break;
+    default:
+      break;
+    }
+    if (c[i] > 10)
+      break; // expected-error {{'break' statement cannot be used in OpenMP for loop}}
+
+    if (c[i] > 11)
+      break; // expected-error {{'break' statement cannot be used in OpenMP for loop}}
+  }
+
+#pragma omp parallel
+#pragma omp taskloop
+  for (int i = 0; i < 10; i++) {
+    for (int j = 0; j < 10; j++) {
+      c[i] = a[i] + b[i];
+      if (c[i] > 10) {
+        if (c[i] < 20) {
+          break; // OK
+        }
+      }
+    }
+  }
+}
+
+void test_loop_eh() {
+  const int N = 100;
+  float a[N], b[N], c[N];
+#pragma omp parallel
+#pragma omp taskloop
+  for (int i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+    try {
+      for (int j = 0; j < 10; ++j) {
+        if (a[i] > b[j])
+          throw a[i];
+      }
+      throw a[i];
+    } catch (float f) {
+      if (f > 0.1)
+        throw a[i];
+      return; // expected-error {{cannot return from OpenMP region}}
+    }
+    switch (i) {
+    case 1:
+      b[i]++;
+      break;
+    default:
+      break;
+    }
+    for (int j = 0; j < 10; j++) {
+      if (c[i] > 10)
+        throw c[i];
+    }
+  }
+  if (c[9] > 10)
+    throw c[9]; // OK
+
+#pragma omp parallel
+#pragma omp taskloop
+  for (int i = 0; i < 10; ++i) {
+    struct S {
+      void g() { throw 0; }
+    };
+  }
+}
+
+void test_loop_firstprivate_lastprivate() {
+  S s(4);
+#pragma omp parallel
+#pragma omp taskloop lastprivate(s) firstprivate(s)
+  for (int i = 0; i < 16; ++i)
+    ;
+}
+
diff --git a/test/OpenMP/taskloop_misc_messages.c b/test/OpenMP/taskloop_misc_messages.c
new file mode 100644
index 000000000000..f9e084e0edd1
--- /dev/null
+++ b/test/OpenMP/taskloop_misc_messages.c
@@ -0,0 +1,373 @@
+// RUN: %clang_cc1 -fsyntax-only -fopenmp -triple x86_64-unknown-unknown -verify %s
+
+// expected-error@+1 {{unexpected OpenMP directive '#pragma omp taskloop'}}
+#pragma omp taskloop
+
+// expected-error@+1 {{unexpected OpenMP directive '#pragma omp taskloop'}}
+#pragma omp taskloop foo
+
+void test_no_clause() {
+  int i;
+#pragma omp taskloop
+  for (i = 0; i < 16; ++i)
+    ;
+
+// expected-error@+2 {{statement after '#pragma omp taskloop' must be a for loop}}
+#pragma omp taskloop
+  ++i;
+}
+
+void test_branch_protected_scope() {
+  int i = 0;
+L1:
+  ++i;
+
+  int x[24];
+
+#pragma omp parallel
+#pragma omp taskloop
+  for (i = 0; i < 16; ++i) {
+    if (i == 5)
+      goto L1; // expected-error {{use of undeclared label 'L1'}}
+    else if (i == 6)
+      return; // expected-error {{cannot return from OpenMP region}}
+    else if (i == 7)
+      goto L2;
+    else if (i == 8) {
+    L2:
+      x[i]++;
+    }
+  }
+
+  if (x[0] == 0)
+    goto L2; // expected-error {{use of undeclared label 'L2'}}
+  else if (x[1] == 1)
+    goto L1;
+}
+
+void test_invalid_clause() {
+  int i;
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+#pragma omp taskloop foo bar
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{directive '#pragma omp taskloop' cannot contain more than one 'nogroup' clause}}
+#pragma omp taskloop nogroup nogroup
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_non_identifiers() {
+  int i, x;
+
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+#pragma omp taskloop;
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-warning@+3 {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+// expected-error@+2 {{unexpected OpenMP clause 'linear' in directive '#pragma omp taskloop'}}
+#pragma omp parallel
+#pragma omp taskloop linear(x);
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+#pragma omp taskloop private(x);
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+#pragma omp taskloop, private(x);
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+extern int foo();
+
+void test_collapse() {
+  int i;
+#pragma omp parallel
+// expected-error@+1 {{expected '('}}
+#pragma omp taskloop collapse
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop collapse(
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop collapse()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop collapse(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}  expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop collapse(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+// expected-error@+1 {{expected '('}}
+#pragma omp taskloop collapse 4)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4,
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4, )
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+// expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4 4)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4, , 4)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+#pragma omp taskloop collapse(4)
+  for (int i1 = 0; i1 < 16; ++i1)
+    for (int i2 = 0; i2 < 16; ++i2)
+      for (int i3 = 0; i3 < 16; ++i3)
+        for (int i4 = 0; i4 < 16; ++i4)
+          foo();
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop collapse(4, 8)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop', but found only 1}}
+#pragma omp parallel
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp taskloop collapse(2.5)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp taskloop collapse(foo())
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
+#pragma omp taskloop collapse(-5)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
+#pragma omp taskloop collapse(0)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
+#pragma omp taskloop collapse(5 - 5)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_private() {
+  int i;
+#pragma omp parallel
+// expected-error@+2 {{expected expression}}
+// expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop private(
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop private(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop private(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop private()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop private(int)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected variable name}}
+#pragma omp taskloop private(0)
+  for (i = 0; i < 16; ++i)
+    ;
+
+  int x, y, z;
+#pragma omp parallel
+#pragma omp taskloop private(x)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop private(x, y)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop private(x, y, z)
+  for (i = 0; i < 16; ++i) {
+    x = y * i + z;
+  }
+}
+
+void test_lastprivate() {
+  int i;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop lastprivate(
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop lastprivate(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop lastprivate(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop lastprivate()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop lastprivate(int)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected variable name}}
+#pragma omp taskloop lastprivate(0)
+  for (i = 0; i < 16; ++i)
+    ;
+
+  int x, y, z;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(x)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(x, y)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(x, y, z)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_firstprivate() {
+  int i;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop firstprivate(
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop firstprivate(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop firstprivate(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop firstprivate()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop firstprivate(int)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected variable name}}
+#pragma omp taskloop firstprivate(0)
+  for (i = 0; i < 16; ++i)
+    ;
+
+  int x, y, z;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(x) firstprivate(x)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(x, y) firstprivate(x, y)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop lastprivate(x, y, z) firstprivate(x, y, z)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_loop_messages() {
+  float a[100], b[100], c[100];
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or pointer type}}
+#pragma omp taskloop
+  for (float fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or pointer type}}
+#pragma omp taskloop
+  for (double fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+
+  // expected-warning@+2 {{OpenMP loop iteration variable cannot have more than 64 bits size and will be narrowed}}
+  #pragma omp taskloop
+  for (__int128 ii = 0; ii < 10; ii++) {
+    c[ii] = a[ii] + b[ii];
+  }
+}
+
diff --git a/test/OpenMP/taskloop_num_tasks_messages.cpp b/test/OpenMP/taskloop_num_tasks_messages.cpp
new file mode 100644
index 000000000000..94d74eb41eb0
--- /dev/null
+++ b/test/OpenMP/taskloop_num_tasks_messages.cpp
@@ -0,0 +1,99 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp taskloop num_tasks // expected-error {{expected '(' after 'num_tasks'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (foobool(argc)), num_tasks (true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'num_tasks' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(0) // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(-1) // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(argc) grainsize(argc) // expected-error {{'grainsize' and 'num_tasks' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'num_tasks' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop num_tasks // expected-error {{expected '(' after 'num_tasks'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (foobool(argc)), num_tasks (true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'num_tasks' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(0)  // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(-1) // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop num_tasks(argc) grainsize(argc) // expected-error {{'grainsize' and 'num_tasks' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'num_tasks' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_priority_messages.cpp b/test/OpenMP/taskloop_priority_messages.cpp
new file mode 100644
index 000000000000..6c1584570eb9
--- /dev/null
+++ b/test/OpenMP/taskloop_priority_messages.cpp
@@ -0,0 +1,93 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp taskloop priority // expected-error {{expected '(' after 'priority'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (foobool(argc)), priority (true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'priority' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority(0)
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority(-1) // expected-error {{argument to 'priority' clause must be a non-negative integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop priority // expected-error {{expected '(' after 'priority'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (foobool(argc)), priority (true) // expected-error {{directive '#pragma omp taskloop' cannot contain more than one 'priority' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority(0)
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop priority(-1) // expected-error {{argument to 'priority' clause must be a non-negative integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_private_messages.cpp b/test/OpenMP/taskloop_private_messages.cpp
new file mode 100644
index 000000000000..3d00d3f252b3
--- /dev/null
+++ b/test/OpenMP/taskloop_private_messages.cpp
@@ -0,0 +1,195 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+};
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+
+public:
+  S3() : a(0) {}
+};
+const S3 ca[5];
+class S4 {
+  int a;
+  S4(); // expected-note {{implicitly declared private here}}
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5() : a(0) {} // expected-note {{implicitly declared private here}}
+
+public:
+  S5(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(I argc, C **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+#pragma omp taskloop private // expected-error {{expected '(' after 'private'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(a, b) // expected-error {{private variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(e, g)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(h) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop shared(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop private(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop private(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  return 0;
+}
+
+void bar(S4 a[2]) {
+#pragma omp parallel
+#pragma omp taskloop private(a)
+  for (int i = 0; i < 2; ++i)
+    foo();
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace B {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  S4 e(4);
+  S5 g(5);
+  int i;
+  int &j = i;
+#pragma omp taskloop private // expected-error {{expected '(' after 'private'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(a, b) // expected-error {{private variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(h) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(B::x) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop shared(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int i;
+#pragma omp taskloop private(i)
+    for (int k = 0; k < argc; ++k)
+      ++k;
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop private(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop private(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  static int si;
+#pragma omp taskloop private(si) // OK
+  for(int k = 0; k < argc; ++k)
+    si = k + 1;
+
+  return 0;
+}
+
diff --git a/test/OpenMP/taskloop_simd_aligned_messages.cpp b/test/OpenMP/taskloop_simd_aligned_messages.cpp
new file mode 100644
index 000000000000..b62af044c60f
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_aligned_messages.cpp
@@ -0,0 +1,202 @@
+// RUN: %clang_cc1 -x c++ -std=c++11 -verify -fopenmp %s
+
+struct B {
+  static int ib[20]; // expected-note 0 {{'B::ib' declared here}}
+  static constexpr int bfoo() { return 8; }
+};
+namespace X {
+  B x; // expected-note {{'x' defined here}}
+};
+constexpr int bfoo() { return 4; }
+
+int **z;
+const int C1 = 1;
+const int C2 = 2;
+void test_aligned_colons(int *&rp)
+{
+  int *B = 0;
+  #pragma omp taskloop simd aligned(B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'}}
+  #pragma omp taskloop simd aligned(B::ib:B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd aligned(B:B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'?}}
+  #pragma omp taskloop simd aligned(z:B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd aligned(B:B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+2 {{integral constant expression must have integral or unscoped enumeration type, not 'int **'}}
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'B'}}
+  #pragma omp taskloop simd aligned(X::x : ::z)
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{integral constant expression must have integral or unscoped enumeration type, not 'B'}}
+  #pragma omp taskloop simd aligned(B,rp,::z: X::x)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd aligned(::z)
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{expected variable name}}
+  #pragma omp taskloop simd aligned(B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-warning@+1 {{aligned clause will be ignored because the requested alignment is not a power of 2}}
+  #pragma omp taskloop simd aligned(B::ib,B:C1+C2)
+  for (int i = 0; i < 10; ++i) ;
+}
+
+// expected-note@+1 {{'num' defined here}}
+template<int L, class T, class N> T test_template(T* arr, N num) {
+  N i;
+  T sum = (T)0;
+  T ind2 = - num * L;
+  // Negative number is passed as L.
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd aligned(arr:L)
+  for (i = 0; i < num; ++i) {
+    T cur = arr[(int)ind2];
+    ind2 += L;
+    sum += cur;
+  }
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'int'}}
+  #pragma omp taskloop simd aligned(num:4)
+  for (i = 0; i < num; ++i);
+  return T();
+}
+
+template<int LEN> int test_warn() {
+  int *ind2 = 0;
+  // expected-error@+1 {{argument to 'aligned' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd aligned(ind2:LEN)
+  for (int i = 0; i < 100; i++) {
+    ind2 += LEN;
+  }
+  return 0;
+}
+
+struct S1; // expected-note 2 {{declared here}}
+extern S1 a; // expected-note {{'a' declared here}}
+class S2 {
+  mutable int a;
+public:
+  S2():a(0) { }
+};
+const S2 b; // expected-note 1 {{'b' defined here}}
+const S2 ba[5];
+class S3 {
+  int a;
+public:
+  S3():a(0) { }
+};
+const S3 ca[5];
+class S4 {
+  int a;
+  S4();
+public:
+  S4(int v):a(v) { }
+};
+class S5 {
+  int a;
+  S5():a(0) {}
+public:
+  S5(int v):a(v) { }
+};
+
+S3 h; // expected-note 2 {{'h' defined here}}
+#pragma omp threadprivate(h)
+
+template<class I, class C> int foomain(I argc, C **argv) {
+  I e(argc);
+  I g(argc);
+  int i; // expected-note {{declared here}} expected-note {{'i' defined here}}
+  // expected-note@+2 {{declared here}}
+  // expected-note@+1 {{reference to 'i' is not a constant expression}}
+  int &j = i;
+  #pragma omp taskloop simd aligned // expected-error {{expected '(' after 'aligned'}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned () // expected-error {{expected expression}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argc : 5) // expected-warning {{aligned clause will be ignored because the requested alignment is not a power of 2}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (S1) // expected-error {{'S1' does not refer to a value}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argv[1]) // expected-error {{expected variable name}}
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned(e, g)
+  for (I k = 0; k < argc; ++k) ++k;
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'S3'}}
+  #pragma omp taskloop simd aligned(h)
+  for (I k = 0; k < argc; ++k) ++k;
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'int'}}
+  #pragma omp taskloop simd aligned(i)
+  for (I k = 0; k < argc; ++k) ++k;
+  #pragma omp parallel
+  {
+    int *v = 0;
+    I i;
+    #pragma omp taskloop simd aligned(v:16)
+    for (I k = 0; k < argc; ++k) { i = k; v += 2; }
+  }
+  float *f;
+  #pragma omp taskloop simd aligned(f)
+  for (I k = 0; k < argc; ++k) ++k;
+  int v = 0;
+  // expected-note@+2 {{initializer of 'j' is not a constant expression}}
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd aligned(f:j)
+  for (I k = 0; k < argc; ++k) { ++k; v += j; }
+  #pragma omp taskloop simd aligned(f)
+  for (I k = 0; k < argc; ++k) ++k;
+  return 0;
+}
+
+// expected-note@+1 2 {{'argc' defined here}}
+int main(int argc, char **argv) {
+  double darr[100];
+  // expected-note@+1 {{in instantiation of function template specialization 'test_template<-4, double, int>' requested here}}
+  test_template<-4>(darr, 4);
+  test_warn<4>(); // ok
+  // expected-note@+1 {{in instantiation of function template specialization 'test_warn<0>' requested here}}
+  test_warn<0>();
+
+  int i;
+  int &j = i;
+  #pragma omp taskloop simd aligned // expected-error {{expected '(' after 'aligned'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned () // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argv // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'int'}}
+  #pragma omp taskloop simd aligned (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'int'}}
+  #pragma omp taskloop simd aligned (argc)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'S1'}}
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'S2'}}
+  #pragma omp taskloop simd aligned (a, b) 
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd aligned (argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+1 {{argument of aligned clause should be array, pointer, reference to array or reference to pointer, not 'S3'}}
+  #pragma omp taskloop simd aligned(h)
+  for (int k = 0; k < argc; ++k) ++k;
+  int *pargc = &argc;
+  foomain<int*,char>(pargc,argv);
+  return 0;
+}
+
diff --git a/test/OpenMP/taskloop_simd_ast_print.cpp b/test/OpenMP/taskloop_simd_ast_print.cpp
new file mode 100644
index 000000000000..3e3b7e4bec2a
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_ast_print.cpp
@@ -0,0 +1,75 @@
+// RUN: %clang_cc1 -verify -fopenmp -ast-print %s | FileCheck %s
+// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -std=c++11 -include-pch %t -fsyntax-only -verify %s -ast-print | FileCheck %s
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void foo() {}
+
+template <class T, int N>
+T tmain(T argc) {
+  T b = argc, c, d, e, f, g;
+  T *ptr;
+  static T a;
+// CHECK: static T a;
+#pragma omp taskloop simd if(taskloop: argc > N) default(shared) untied priority(N) safelen(N) linear(c) aligned(ptr) grainsize(N)
+  // CHECK-NEXT: #pragma omp taskloop simd if(taskloop: argc > N) default(shared) untied priority(N) safelen(N) linear(c) aligned(ptr) grainsize(N)
+  for (int i = 0; i < 2; ++i)
+    a = 2;
+// CHECK-NEXT: for (int i = 0; i < 2; ++i)
+// CHECK-NEXT: a = 2;
+#pragma omp parallel
+#pragma omp taskloop simd private(argc, b), firstprivate(c, d), lastprivate(d, f) collapse(N) shared(g) if (c) final(d) mergeable priority(f) simdlen(N) nogroup num_tasks(N)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
+  for (int i = 0; i < 2; ++i)
+    for (int j = 0; j < 2; ++j)
+      for (int j = 0; j < 2; ++j)
+        for (int j = 0; j < 2; ++j)
+          for (int j = 0; j < 2; ++j)
+            foo();
+  // CHECK-NEXT: #pragma omp parallel
+  // CHECK-NEXT: #pragma omp taskloop simd private(argc,b) firstprivate(c,d) lastprivate(d,f) collapse(N) shared(g) if(c) final(d) mergeable priority(f) simdlen(N) nogroup num_tasks(N)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int i = 0; i < 2; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: for (int j = 0; j < 2; ++j)
+  // CHECK-NEXT: foo();
+  return T();
+}
+
+int main(int argc, char **argv) {
+  int b = argc, c, d, e, f, g;
+  static int a;
+// CHECK: static int a;
+#pragma omp taskloop simd if(taskloop: a) default(none) shared(a) final(b) priority(5) safelen(8) linear(b), aligned(argv) nogroup num_tasks(argc)
+  // CHECK-NEXT: #pragma omp taskloop simd if(taskloop: a) default(none) shared(a) final(b) priority(5) safelen(8) linear(b) aligned(argv) nogroup num_tasks(argc)
+  for (int i = 0; i < 2; ++i)
+    a = 2;
+// CHECK-NEXT: for (int i = 0; i < 2; ++i)
+// CHECK-NEXT: a = 2;
+#pragma omp parallel
+#pragma omp taskloop simd private(argc, b), firstprivate(argv, c), lastprivate(d, f) collapse(2) shared(g) if(argc) mergeable priority(argc) simdlen(16) grainsize(argc)
+  for (int i = 0; i < 10; ++i)
+    for (int j = 0; j < 10; ++j)
+      foo();
+  // CHECK-NEXT: #pragma omp parallel
+  // CHECK-NEXT: #pragma omp taskloop simd private(argc,b) firstprivate(argv,c) lastprivate(d,f) collapse(2) shared(g) if(argc) mergeable priority(argc) simdlen(16) grainsize(argc)
+  // CHECK-NEXT: for (int i = 0; i < 10; ++i)
+  // CHECK-NEXT: for (int j = 0; j < 10; ++j)
+  // CHECK-NEXT: foo();
+  return (tmain<int, 5>(argc) + tmain<char, 1>(argv[0][0]));
+}
+
+#endif
diff --git a/test/OpenMP/taskloop_simd_collapse_messages.cpp b/test/OpenMP/taskloop_simd_collapse_messages.cpp
new file mode 100644
index 000000000000..d178c0834d62
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_collapse_messages.cpp
@@ -0,0 +1,83 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp taskloop simd collapse // expected-error {{expected '(' after 'collapse'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp taskloop simd collapse (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd collapse (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse (1)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse ((ST > 0) ? 1 + ST : 2) // expected-note 2 {{as specified in 'collapse' clause}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST]; // expected-error 2 {{expected 2 for loops after '#pragma omp taskloop simd', but found only 1}}
+  // expected-error@+3 2 {{directive '#pragma omp taskloop simd' cannot contain more than one 'collapse' clause}}
+  // expected-error@+2 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd collapse (foobool(argc)), collapse (true), collapse (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd collapse (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse (1)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse (N) // expected-error {{argument to 'collapse' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd collapse (2) // expected-note {{as specified in 'collapse' clause}}
+  foo(); // expected-error {{expected 2 for loops after '#pragma omp taskloop simd'}}
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop simd collapse // expected-error {{expected '(' after 'collapse'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd collapse ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd collapse () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd collapse (4 // expected-error {{expected ')'}} expected-note {{to match this '('}} expected-note {{as specified in 'collapse' clause}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4]; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+  #pragma omp taskloop simd collapse (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}  expected-note {{as specified in 'collapse' clause}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4]; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+  #pragma omp taskloop simd collapse (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp taskloop simd' cannot contain more than one 'collapse' clause}}
+  // expected-error@+1 2 {{argument to 'collapse' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd collapse (foobool(argc)), collapse (true), collapse (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd collapse (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd collapse (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp taskloop simd' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp taskloop simd collapse(collapse(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  #pragma omp taskloop simd collapse (2) // expected-note {{as specified in 'collapse' clause}}
+  foo(); // expected-error {{expected 2 for loops after '#pragma omp taskloop simd'}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 1, 0>' requested here}}
+  return tmain<int, char, 1, 0>(argc, argv);
+}
+
diff --git a/test/OpenMP/taskloop_simd_final_messages.cpp b/test/OpenMP/taskloop_simd_final_messages.cpp
new file mode 100644
index 000000000000..c6580fbaa9b4
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_final_messages.cpp
@@ -0,0 +1,90 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+#pragma omp taskloop simd final // expected-error {{expected '(' after 'final'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final() // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc > 0 ? argv[1] : argv[2])
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(foobool(argc)), final(true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'final' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc)
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+#pragma omp taskloop simd final // expected-error {{expected '(' after 'final'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final() // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc > 0 ? argv[1] : argv[2])
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(foobool(argc)), final(true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'final' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argv[1] = 2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+#pragma omp taskloop simd final(if (tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_simd_firstprivate_messages.cpp b/test/OpenMP/taskloop_simd_firstprivate_messages.cpp
new file mode 100644
index 000000000000..83946695204b
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_firstprivate_messages.cpp
@@ -0,0 +1,313 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+  S2(const S2 &s2) : a(s2.a) {}
+  static float S2s;
+  static const float S2sc;
+};
+const float S2::S2sc = 0;
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+  S3 &operator=(const S3 &s3);
+
+public:
+  S3() : a(0) {} // expected-note 2 {{candidate constructor not viable: requires 0 arguments, but 1 was provided}}
+  S3(S3 &s3) : a(s3.a) {} // expected-note 2 {{candidate constructor not viable: 1st argument ('const S3') would lose const qualifier}}
+};
+const S3 c;
+const S3 ca[5];
+extern const int f;
+class S4 {
+  int a;
+  S4();
+  S4(const S4 &s4); // expected-note 2 {{implicitly declared private here}}
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5(const S5 &s5) : a(s5.a) {} // expected-note 4 {{implicitly declared private here}}
+
+public:
+  S5() : a(0) {}
+  S5(int v) : a(v) {}
+};
+class S6 {
+  int a;
+  S6() : a(0) {}
+
+public:
+  S6(const S6 &s6) : a(s6.a) {}
+  S6(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(int argc, char **argv) {
+  I e(4);
+  C g(5);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate // expected-error {{expected '(' after 'firstprivate'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(a, b) // expected-error {{firstprivate variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(h) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop simd firstprivate(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop simd firstprivate(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(g) firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel private(i)
+#pragma omp taskloop simd firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be firstprivate, predetermined as linear}}
+    foo();
+#pragma omp parallel reduction(+ : i)
+#pragma omp taskloop simd firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be firstprivate, predetermined as linear}}
+    foo();
+  return 0;
+}
+
+void bar(S4 a[2]) {
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(a)
+  for (int i = 0; i < 2; ++i)
+    foo();
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace B {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  const int d = 5;
+  const int da[5] = {0};
+  S4 e(4);
+  S5 g(5);
+  S3 m;
+  S6 n(2);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate // expected-error {{expected '(' after 'firstprivate'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate() // expected-error {{expected expression}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argc)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(a, b, c, d, f) // expected-error {{firstprivate variable with incomplete type 'S1'}} expected-error {{no matching constructor for initialization of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(argv[1]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(2 * 2) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(ba) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(ca) // expected-error {{no matching constructor for initialization of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(da) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+  int xa;
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(xa) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(S2::S2s) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(S2::S2sc) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd safelen(5)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(m) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(h) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd private(xa), firstprivate(xa) // expected-error {{private variable cannot be firstprivate}} expected-note {{defined as private}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i)    // expected-error {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be firstprivate, predetermined as linear}}
+    foo();
+#pragma omp parallel shared(xa)
+#pragma omp taskloop simd firstprivate(xa) // OK: may be firstprivate
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(j)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(g) firstprivate(g) // expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(n) firstprivate(n) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop simd firstprivate(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel private(i)
+#pragma omp taskloop simd firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be firstprivate, predetermined as linear}}
+    foo();
+#pragma omp parallel reduction(+ : i)
+#pragma omp taskloop simd firstprivate(i) // expected-note {{defined as firstprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be firstprivate, predetermined as linear}}
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(B::x) // expected-error {{threadprivate or thread local variable cannot be firstprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+  static int si;
+#pragma omp taskloop simd firstprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
+
+  return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
+}
+
diff --git a/test/OpenMP/taskloop_simd_grainsize_messages.cpp b/test/OpenMP/taskloop_simd_grainsize_messages.cpp
new file mode 100644
index 000000000000..45ccb31824fa
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_grainsize_messages.cpp
@@ -0,0 +1,99 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp taskloop simd grainsize // expected-error {{expected '(' after 'grainsize'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (foobool(argc)), grainsize (true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'grainsize' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(0) // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(-1) // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(argc) num_tasks(argc) // expected-error {{'num_tasks' and 'grainsize' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'grainsize' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop simd grainsize // expected-error {{expected '(' after 'grainsize'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (foobool(argc)), grainsize (true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'grainsize' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(0)  // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(-1) // expected-error {{argument to 'grainsize' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd grainsize(argc) num_tasks(argc) // expected-error {{'num_tasks' and 'grainsize' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'grainsize' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_simd_lastprivate_messages.cpp b/test/OpenMP/taskloop_simd_lastprivate_messages.cpp
new file mode 100644
index 000000000000..ed1bdf5a6e3e
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_lastprivate_messages.cpp
@@ -0,0 +1,287 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+  S2(S2 &s2) : a(s2.a) {}
+  const S2 &operator =(const S2&) const;
+  S2 &operator =(const S2&);
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
+  static const float S2sc;
+};
+const float S2::S2sc = 0; // expected-note {{static data member is predetermined as shared}}
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+  S3 &operator=(const S3 &s3); // expected-note 2 {{implicitly declared private here}}
+
+public:
+  S3() : a(0) {}
+  S3(S3 &s3) : a(s3.a) {}
+};
+const S3 c;         // expected-note {{global variable is predetermined as shared}}
+const S3 ca[5];     // expected-note {{global variable is predetermined as shared}}
+extern const int f; // expected-note {{global variable is predetermined as shared}}
+class S4 {
+  int a;
+  S4();             // expected-note 3 {{implicitly declared private here}}
+  S4(const S4 &s4);
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5() : a(0) {} // expected-note {{implicitly declared private here}}
+
+public:
+  S5(const S5 &s5) : a(s5.a) {}
+  S5(int v) : a(v) {}
+};
+class S6 {
+  int a;
+  S6() : a(0) {}
+
+public:
+  S6(const S6 &s6) : a(s6.a) {}
+  S6(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(int argc, char **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(a, b) // expected-error {{lastprivate variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(e, g) // expected-error 2 {{calling a private constructor of class 'S4'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(h) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop simd lastprivate(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop simd lastprivate(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  return 0;
+}
+
+void bar(S4 a[2]) {
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(a)
+  for (int i = 0; i < 2; ++i)
+    foo();
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace B {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  const int d = 5;       // expected-note {{constant variable is predetermined as shared}}
+  const int da[5] = {0}; // expected-note {{constant variable is predetermined as shared}}
+  S4 e(4);
+  S5 g(5);
+  S3 m;
+  S6 n(2);
+  int i;
+  int &j = i;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate // expected-error {{expected '(' after 'lastprivate'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate() // expected-error {{expected expression}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argc)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(S1) // expected-error {{'S1' does not refer to a value}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(a, b, c, d, f) // expected-error {{lastprivate variable with incomplete type 'S1'}} expected-error 3 {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(argv[1]) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(2 * 2) // expected-error {{expected variable name}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(ba)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(ca) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(da) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+  int xa;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(xa) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(S2::S2s) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(S2::S2sc) // expected-error {{shared variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd safelen(5)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(m) // expected-error {{'operator=' is a private member of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(h) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(B::x) // expected-error {{threadprivate or thread local variable cannot be lastprivate}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd private(xa), lastprivate(xa) // expected-error {{private variable cannot be lastprivate}} expected-note {{defined as private}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(i) // expected-note {{defined as lastprivate}}
+  for (i = 0; i < argc; ++i) // expected-error {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be lastprivate, predetermined as linear}}
+    foo();
+#pragma omp parallel private(xa)
+#pragma omp taskloop simd lastprivate(xa)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel reduction(+ : xa)
+#pragma omp taskloop simd lastprivate(xa)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(j)
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd firstprivate(m) lastprivate(m) // expected-error {{'operator=' is a private member of 'S3'}}
+  for (i = 0; i < argc; ++i)
+    foo();
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(n) firstprivate(n) // OK
+  for (i = 0; i < argc; ++i)
+    foo();
+  static int si;
+#pragma omp taskloop simd lastprivate(si) // OK
+  for (i = 0; i < argc; ++i)
+    si = i + 1;
+  return foomain<S4, S5>(argc, argv); // expected-note {{in instantiation of function template specialization 'foomain<S4, S5>' requested here}}
+}
diff --git a/test/OpenMP/taskloop_simd_linear_messages.cpp b/test/OpenMP/taskloop_simd_linear_messages.cpp
new file mode 100644
index 000000000000..c9a82b9dc148
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_linear_messages.cpp
@@ -0,0 +1,249 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+namespace X {
+  int x;
+};
+
+struct B {
+  static int ib; // expected-note {{'B::ib' declared here}}
+  static int bfoo() { return 8; }
+};
+
+int bfoo() { return 4; }
+
+int z;
+const int C1 = 1;
+const int C2 = 2;
+void test_linear_colons()
+{
+  int B = 0;
+  #pragma omp taskloop simd linear(B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'}}
+  #pragma omp taskloop simd linear(B::ib:B:bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{use of undeclared identifier 'ib'; did you mean 'B::ib'}}
+  #pragma omp taskloop simd linear(B:ib)
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{unexpected ':' in nested name specifier; did you mean '::'?}}
+  #pragma omp taskloop simd linear(z:B:ib)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd linear(B:B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd linear(X::x : ::z)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd linear(B,::z, X::x)
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd linear(::z)
+  for (int i = 0; i < 10; ++i) ;
+  // expected-error@+1 {{expected variable name}}
+  #pragma omp taskloop simd linear(B::bfoo())
+  for (int i = 0; i < 10; ++i) ;
+  #pragma omp taskloop simd linear(B::ib,B:C1+C2)
+  for (int i = 0; i < 10; ++i) ;
+}
+
+template<int L, class T, class N> T test_template(T* arr, N num) {
+  N i;
+  T sum = (T)0;
+  T ind2 = - num * L; // expected-note {{'ind2' defined here}}
+  // expected-error@+1 {{argument of a linear clause should be of integral or pointer type}}
+#pragma omp taskloop simd linear(ind2:L)
+  for (i = 0; i < num; ++i) {
+    T cur = arr[(int)ind2];
+    ind2 += L;
+    sum += cur;
+  }
+  return T();
+}
+
+template<int LEN> int test_warn() {
+  int ind2 = 0;
+  // expected-warning@+1 {{zero linear step (ind2 should probably be const)}}
+  #pragma omp taskloop simd linear(ind2:LEN)
+  for (int i = 0; i < 100; i++) {
+    ind2 += LEN;
+  }
+  return ind2;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+public:
+  S2():a(0) { }
+};
+const S2 b; // expected-note 2 {{'b' defined here}}
+const S2 ba[5];
+class S3 {
+  int a;
+public:
+  S3():a(0) { }
+};
+const S3 ca[5];
+class S4 {
+  int a;
+  S4();
+public:
+  S4(int v):a(v) { }
+};
+class S5 {
+  int a;
+  S5():a(0) {}
+public:
+  S5(int v):a(v) { }
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template<class I, class C> int foomain(I argc, C **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+  #pragma omp taskloop simd linear // expected-error {{expected '(' after 'linear'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (val // expected-error {{use of undeclared identifier 'val'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (uval( // expected-error {{expected expression}} expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (ref() // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (foo() // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear () // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (val argc // expected-error {{use of undeclared identifier 'val'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (val(argc, // expected-error {{expected expression}} expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc : 5)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{linear variable with incomplete type 'S1'}}
+  // expected-error@+1 {{const-qualified variable cannot be linear}}
+  #pragma omp taskloop simd linear (val(a, b):B::ib)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(ref(e, g)) // expected-error 2 {{variable of non-reference type 'int' can be used only with 'val' modifier, but used with 'ref'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(h) // expected-error {{threadprivate or thread local variable cannot be linear}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(uval(i)) // expected-error {{variable of non-reference type 'int' can be used only with 'val' modifier, but used with 'uval'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp parallel
+  {
+    int v = 0;
+    int i;
+    #pragma omp taskloop simd linear(v:i)
+    for (int k = 0; k < argc; ++k) { i = k; v += i; }
+  }
+  #pragma omp taskloop simd linear(ref(j))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(uval(j))
+  for (int k = 0; k < argc; ++k) ++k;
+  int v = 0;
+  #pragma omp taskloop simd linear(v:j)
+  for (int k = 0; k < argc; ++k) { ++k; v += j; }
+  #pragma omp taskloop simd linear(i)
+  for (int k = 0; k < argc; ++k) ++k;
+  return 0;
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace C {
+using A::x;
+}
+
+void linear_modifiers(int argc) {
+  int &f = argc;
+  #pragma omp taskloop simd linear(f)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(val(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(uval(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(ref(f))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(foo(f)) // expected-error {{expected one of 'ref', val' or 'uval' modifiers}}
+  for (int k = 0; k < argc; ++k) ++k;
+}
+
+int f;
+int main(int argc, char **argv) {
+  double darr[100];
+  // expected-note@+1 {{in instantiation of function template specialization 'test_template<-4, double, int>' requested here}}
+  test_template<-4>(darr, 4);
+  // expected-note@+1 {{in instantiation of function template specialization 'test_warn<0>' requested here}}
+  test_warn<0>();
+
+  S4 e(4); // expected-note {{'e' defined here}}
+  S5 g(5); // expected-note {{'g' defined here}}
+  int i;
+  int &j = i;
+  #pragma omp taskloop simd linear(f) linear(f) // expected-error {{linear variable cannot be linear}} expected-note {{defined as linear}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear // expected-error {{expected '(' after 'linear'}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear () // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (val // expected-error {{use of undeclared identifier 'val'}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (ref()) // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (foo()) // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argc)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{linear variable with incomplete type 'S1'}}
+  // expected-error@+1 {{const-qualified variable cannot be linear}}
+  #pragma omp taskloop simd linear(a, b)
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear (argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k) ++k;
+  // expected-error@+2 {{argument of a linear clause should be of integral or pointer type, not 'S4'}}
+  // expected-error@+1 {{argument of a linear clause should be of integral or pointer type, not 'S5'}}
+  #pragma omp taskloop simd linear(val(e, g))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(h, C::x) // expected-error 2 {{threadprivate or thread local variable cannot be linear}}
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp parallel
+  {
+    int i;
+    #pragma omp taskloop simd linear(val(i))
+    for (int k = 0; k < argc; ++k) ++k;
+    #pragma omp taskloop simd linear(uval(i) : 4) // expected-error {{variable of non-reference type 'int' can be used only with 'val' modifier, but used with 'uval'}}
+    for (int k = 0; k < argc; ++k) { ++k; i += 4; }
+  }
+  #pragma omp taskloop simd linear(ref(j))
+  for (int k = 0; k < argc; ++k) ++k;
+  #pragma omp taskloop simd linear(i)
+  for (int k = 0; k < argc; ++k) ++k;
+
+  foomain<int,char>(argc,argv); // expected-note {{in instantiation of function template specialization 'foomain<int, char>' requested here}}
+  return 0;
+}
+
diff --git a/test/OpenMP/taskloop_simd_loop_messages.cpp b/test/OpenMP/taskloop_simd_loop_messages.cpp
new file mode 100644
index 000000000000..47318721102f
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_loop_messages.cpp
@@ -0,0 +1,739 @@
+// RUN: %clang_cc1 -fsyntax-only -fopenmp -x c++ -std=c++11 -fexceptions -fcxx-exceptions -verify %s
+
+class S {
+  int a;
+  S() : a(0) {}
+
+public:
+  S(int v) : a(v) {}
+  S(const S &s) : a(s.a) {}
+};
+
+static int sii;
+// expected-note@+1 {{defined as threadprivate or thread local}}
+#pragma omp threadprivate(sii)
+static int globalii;
+
+// Currently, we cannot use "0" for global register variables.
+// register int reg0 __asm__("0");
+int reg0;
+
+int test_iteration_spaces() {
+  const int N = 100;
+  float a[N], b[N], c[N];
+  int ii, jj, kk;
+  float fii;
+  double dii;
+  register int reg; // expected-warning {{'register' storage class specifier is deprecated}}
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (int i = 0; i < 10; i += 1) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (char i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (char i = 0; i < 10; i += '\1') {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (long long i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+// expected-error@+2 {{expression must have integral or unscoped enumeration type, not 'double'}}
+#pragma omp taskloop simd
+  for (long long i = 0; i < 10; i += 1.5) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (long long i = 0; i < 'z'; i += 1u) {
+    c[i] = a[i] + b[i];
+  }
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or random access iterator type}}
+#pragma omp taskloop simd
+  for (float fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or random access iterator type}}
+#pragma omp taskloop simd
+  for (double fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (int &ref = ii; ref < 10; ref++) {
+  }
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (int i; i < 10; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (int i = 0, j = 0; i < 10; ++i)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-warning@+3 {{expression result unused}}
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (ii + 1; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (c[ii] = 0; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// Ok to skip parenthesises.
+#pragma omp taskloop simd
+  for (((ii)) = 0; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop simd
+  for (int i = 0; i; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+3 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'i'}}
+#pragma omp taskloop simd
+  for (int i = 0; jj < kk; ii++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop simd
+  for (int i = 0; !!i; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop simd
+  for (int i = 0; i != 1; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'i'}}
+#pragma omp taskloop simd
+  for (int i = 0;; i++)
+    c[i] = a[i];
+
+#pragma omp parallel
+// Ok.
+#pragma omp taskloop simd
+  for (int i = 11; i > 10; i--)
+    c[i] = a[i];
+
+#pragma omp parallel
+// Ok.
+#pragma omp taskloop simd
+  for (int i = 0; i < 10; ++i)
+    c[i] = a[i];
+
+#pragma omp parallel
+// Ok.
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ++jj)
+    c[ii] = a[jj];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ++++ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// Ok but undefined behavior (in general, cannot check that incr
+// is really loop-invariant).
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii = ii + ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{expression must have integral or unscoped enumeration type, not 'float'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii = ii + 1.0f)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// Ok - step was converted to integer type.
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii = ii + (int)1.1f)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; jj = ii + 2)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-warning@+3 {{relational comparison result unused}}
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii<10; jj> kk + 2)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10;)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-warning@+3 {{expression result unused}}
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; !ii)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii ? ++ii : ++jj)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'ii'}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii = ii < 10)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii = ii + 0)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; ii = ii + (int)(0.8 - 0.45))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; (ii) < 10; ii -= 25)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; (ii < 10); ii -= 0)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; ii > 10; (ii += 0))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; ii < 10; (ii) = (1 - 1) + (ii))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for ((ii = 0); ii > 10; (ii -= 0))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'ii' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (ii = 0; (ii < 10); (ii -= 0))
+    c[ii] = a[ii];
+
+#pragma omp parallel
+// expected-note@+2  {{defined as firstprivate}}
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be firstprivate, predetermined as linear}}
+#pragma omp taskloop simd firstprivate(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+#pragma omp taskloop simd linear(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+// expected-note@+3 {{defined as private}}
+// expected-error@+3 {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be private, predetermined as linear}}
+#pragma omp parallel
+#pragma omp taskloop simd private(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+// expected-note@+3 {{defined as lastprivate}}
+// expected-error@+3 {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be lastprivate, predetermined as linear}}
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(ii)
+  for (ii = 0; ii < 10; ii++)
+    c[ii] = a[ii];
+
+#pragma omp parallel
+  {
+// expected-error@+2 {{loop iteration variable in the associated loop of 'omp taskloop simd' directive may not be threadprivate or thread local, predetermined as linear}}
+#pragma omp taskloop simd
+    for (sii = 0; sii < 10; sii += 1)
+      c[sii] = a[sii];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop simd
+    for (reg0 = 0; reg0 < 10; reg0 += 1)
+      c[reg0] = a[reg0];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop simd
+    for (reg = 0; reg < 10; reg += 1)
+      c[reg] = a[reg];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop simd
+    for (globalii = 0; globalii < 10; globalii += 1)
+      c[globalii] = a[globalii];
+  }
+
+#pragma omp parallel
+  {
+#pragma omp taskloop simd collapse(2)
+    for (ii = 0; ii < 10; ii += 1)
+    for (globalii = 0; globalii < 10; globalii += 1)
+      c[globalii] += a[globalii] + ii;
+  }
+
+#pragma omp parallel
+// expected-error@+2 {{statement after '#pragma omp taskloop simd' must be a for loop}}
+#pragma omp taskloop simd
+  for (auto &item : a) {
+    item = item + 1;
+  }
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'i' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (unsigned i = 9; i < 10; i--) {
+    c[i] = a[i] + b[i];
+  }
+
+  int(*lb)[4] = nullptr;
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (int(*p)[4] = lb; p < lb + 8; ++p) {
+  }
+
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (int a{0}; a < 10; ++a) {
+  }
+
+  return 0;
+}
+
+// Iterators allowed in openmp for-loops.
+namespace std {
+struct random_access_iterator_tag {};
+template <class Iter>
+struct iterator_traits {
+  typedef typename Iter::difference_type difference_type;
+  typedef typename Iter::iterator_category iterator_category;
+};
+template <class Iter>
+typename iterator_traits<Iter>::difference_type
+distance(Iter first, Iter last) { return first - last; }
+}
+class Iter0 {
+public:
+  Iter0() {}
+  Iter0(const Iter0 &) {}
+  Iter0 operator++() { return *this; }
+  Iter0 operator--() { return *this; }
+  bool operator<(Iter0 a) { return true; }
+};
+// expected-note@+2 {{candidate function not viable: no known conversion from 'GoodIter' to 'Iter0' for 1st argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'Iter0' for 1st argument}}
+int operator-(Iter0 a, Iter0 b) { return 0; }
+class Iter1 {
+public:
+  Iter1(float f = 0.0f, double d = 0.0) {}
+  Iter1(const Iter1 &) {}
+  Iter1 operator++() { return *this; }
+  Iter1 operator--() { return *this; }
+  bool operator<(Iter1 a) { return true; }
+  bool operator>=(Iter1 a) { return false; }
+};
+class GoodIter {
+public:
+  GoodIter() {}
+  GoodIter(const GoodIter &) {}
+  GoodIter(int fst, int snd) {}
+  GoodIter &operator=(const GoodIter &that) { return *this; }
+  GoodIter &operator=(const Iter0 &that) { return *this; }
+  GoodIter &operator+=(int x) { return *this; }
+  GoodIter &operator-=(int x) { return *this; }
+  explicit GoodIter(void *) {}
+  GoodIter operator++() { return *this; }
+  GoodIter operator--() { return *this; }
+  bool operator!() { return true; }
+  bool operator<(GoodIter a) { return true; }
+  bool operator<=(GoodIter a) { return true; }
+  bool operator>=(GoodIter a) { return false; }
+  typedef int difference_type;
+  typedef std::random_access_iterator_tag iterator_category;
+};
+// expected-note@+2 {{candidate function not viable: no known conversion from 'Iter0' to 'GoodIter' for 2nd argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'GoodIter' for 1st argument}}
+int operator-(GoodIter a, GoodIter b) { return 0; }
+// expected-note@+1 3 {{candidate function not viable: requires single argument 'a', but 2 arguments were provided}}
+GoodIter operator-(GoodIter a) { return a; }
+// expected-note@+2 {{candidate function not viable: no known conversion from 'Iter0' to 'int' for 2nd argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'GoodIter' for 1st argument}}
+GoodIter operator-(GoodIter a, int v) { return GoodIter(); }
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter0' to 'GoodIter' for 1st argument}}
+GoodIter operator+(GoodIter a, int v) { return GoodIter(); }
+// expected-note@+2 {{candidate function not viable: no known conversion from 'GoodIter' to 'int' for 1st argument}}
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter1' to 'int' for 1st argument}}
+GoodIter operator-(int v, GoodIter a) { return GoodIter(); }
+// expected-note@+1 2 {{candidate function not viable: no known conversion from 'Iter0' to 'int' for 1st argument}}
+GoodIter operator+(int v, GoodIter a) { return GoodIter(); }
+
+int test_with_random_access_iterator() {
+  GoodIter begin, end;
+  Iter0 begin0, end0;
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (GoodIter &I = begin; I < end; ++I)
+    ++I;
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I >= end; --I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (GoodIter I(begin); I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (GoodIter I(nullptr); I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (GoodIter I(0); I < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (GoodIter I(1, 2); I < end; ++I)
+    ++I;
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (begin = GoodIter(0); begin < end; ++begin)
+    ++begin;
+// expected-error@+4 {{invalid operands to binary expression ('GoodIter' and 'Iter0')}}
+// expected-error@+3 {{could not calculate number of iterations calling 'operator-' with upper and lower loop bounds}}
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (begin = begin0; begin < end; ++begin)
+    ++begin;
+#pragma omp parallel
+// expected-error@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (++begin; begin < end; ++begin)
+    ++begin;
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (begin = end; begin < end; ++begin)
+    ++begin;
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'I'}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I - I; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'I'}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; begin < end; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{condition of OpenMP for loop must be a relational comparison ('<', '<=', '>', or '>=') of loop variable 'I'}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; !I; ++I)
+    ++I;
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I >= end; I = I + 1)
+    ++I;
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I >= end; I = I - 1)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'I'}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I >= end; I = -I)
+    ++I;
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I >= end; I = 2 + I)
+    ++I;
+#pragma omp parallel
+// expected-error@+2 {{increment clause of OpenMP for loop must perform simple addition or subtraction on loop variable 'I'}}
+#pragma omp taskloop simd
+  for (GoodIter I = begin; I >= end; I = 2 - I)
+    ++I;
+// In the following example, we cannot update the loop variable using '+='
+// expected-error@+3 {{invalid operands to binary expression ('Iter0' and 'int')}}
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (Iter0 I = begin0; I < end0; ++I)
+    ++I;
+#pragma omp parallel
+// Initializer is constructor without params.
+// expected-error@+3 {{invalid operands to binary expression ('Iter0' and 'int')}}
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (Iter0 I; I < end0; ++I)
+    ++I;
+  Iter1 begin1, end1;
+// expected-error@+4 {{invalid operands to binary expression ('Iter1' and 'Iter1')}}
+// expected-error@+3 {{could not calculate number of iterations calling 'operator-' with upper and lower loop bounds}}
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (Iter1 I = begin1; I < end1; ++I)
+    ++I;
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (Iter1 I = begin1; I >= end1; ++I)
+    ++I;
+#pragma omp parallel
+// expected-error@+5 {{invalid operands to binary expression ('Iter1' and 'float')}}
+// expected-error@+4 {{could not calculate number of iterations calling 'operator-' with upper and lower loop bounds}}
+// Initializer is constructor with all default params.
+// expected-warning@+2 {{initialization clause of OpenMP for loop is not in canonical form ('var = init' or 'T var = init')}}
+#pragma omp taskloop simd
+  for (Iter1 I; I < end1; ++I) {
+  }
+  return 0;
+}
+
+template <typename IT, int ST>
+class TC {
+public:
+  int dotest_lt(IT begin, IT end) {
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+    for (IT I = begin; I < end; I = I + ST) {
+      ++I;
+    }
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be positive due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to increase on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+    for (IT I = begin; I <= end; I += ST) {
+      ++I;
+    }
+#pragma omp parallel
+#pragma omp taskloop simd
+    for (IT I = begin; I < end; ++I) {
+      ++I;
+    }
+  }
+
+  static IT step() {
+    return IT(ST);
+  }
+};
+template <typename IT, int ST = 0>
+int dotest_gt(IT begin, IT end) {
+#pragma omp parallel
+// expected-note@+3 2 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (IT I = begin; I >= end; I = I + ST) {
+    ++I;
+  }
+#pragma omp parallel
+// expected-note@+3 2 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (IT I = begin; I >= end; I += ST) {
+    ++I;
+  }
+
+#pragma omp parallel
+// expected-note@+3 {{loop step is expected to be negative due to this condition}}
+// expected-error@+2 {{increment expression must cause 'I' to decrease on each iteration of OpenMP for loop}}
+#pragma omp taskloop simd
+  for (IT I = begin; I >= end; ++I) {
+    ++I;
+  }
+
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (IT I = begin; I < end; I += TC<int, ST>::step()) {
+    ++I;
+  }
+}
+
+void test_with_template() {
+  GoodIter begin, end;
+  TC<GoodIter, 100> t1;
+  TC<GoodIter, -100> t2;
+  t1.dotest_lt(begin, end);
+  t2.dotest_lt(begin, end);         // expected-note {{in instantiation of member function 'TC<GoodIter, -100>::dotest_lt' requested here}}
+  dotest_gt(begin, end);            // expected-note {{in instantiation of function template specialization 'dotest_gt<GoodIter, 0>' requested here}}
+  dotest_gt<unsigned, -10>(0, 100); // expected-note {{in instantiation of function template specialization 'dotest_gt<unsigned int, -10>' requested here}}
+}
+
+void test_loop_break() {
+  const int N = 100;
+  float a[N], b[N], c[N];
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (int i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+    for (int j = 0; j < 10; ++j) {
+      if (a[i] > b[j])
+        break; // OK in nested loop
+    }
+    switch (i) {
+    case 1:
+      b[i]++;
+      break;
+    default:
+      break;
+    }
+    if (c[i] > 10)
+      break; // expected-error {{'break' statement cannot be used in OpenMP for loop}}
+
+    if (c[i] > 11)
+      break; // expected-error {{'break' statement cannot be used in OpenMP for loop}}
+  }
+
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (int i = 0; i < 10; i++) {
+    for (int j = 0; j < 10; j++) {
+      c[i] = a[i] + b[i];
+      if (c[i] > 10) {
+        if (c[i] < 20) {
+          break; // OK
+        }
+      }
+    }
+  }
+}
+
+void test_loop_eh() {
+  const int N = 100;
+  float a[N], b[N], c[N];
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (int i = 0; i < 10; i++) {
+    c[i] = a[i] + b[i];
+    try { // expected-error {{'try' statement cannot be used in OpenMP simd region}}
+      for (int j = 0; j < 10; ++j) {
+        if (a[i] > b[j])
+          throw a[i]; // expected-error {{'throw' statement cannot be used in OpenMP simd region}}
+      }
+      throw a[i]; // expected-error {{'throw' statement cannot be used in OpenMP simd region}}
+    } catch (float f) {
+      if (f > 0.1)
+        throw a[i]; // expected-error {{'throw' statement cannot be used in OpenMP simd region}}
+      return; // expected-error {{cannot return from OpenMP region}}
+    }
+    switch (i) {
+    case 1:
+      b[i]++;
+      break;
+    default:
+      break;
+    }
+    for (int j = 0; j < 10; j++) {
+      if (c[i] > 10)
+        throw c[i]; // expected-error {{'throw' statement cannot be used in OpenMP simd region}}
+    }
+  }
+  if (c[9] > 10)
+    throw c[9]; // OK
+
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (int i = 0; i < 10; ++i) {
+    struct S {
+      void g() { throw 0; }
+    };
+  }
+}
+
+void test_loop_firstprivate_lastprivate() {
+  S s(4);
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(s) firstprivate(s)
+  for (int i = 0; i < 16; ++i)
+    ;
+}
+
diff --git a/test/OpenMP/taskloop_simd_misc_messages.c b/test/OpenMP/taskloop_simd_misc_messages.c
new file mode 100644
index 000000000000..61c7b107d496
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_misc_messages.c
@@ -0,0 +1,372 @@
+// RUN: %clang_cc1 -fsyntax-only -fopenmp -triple x86_64-unknown-unknown -verify %s
+
+// expected-error@+1 {{unexpected OpenMP directive '#pragma omp taskloop simd'}}
+#pragma omp taskloop simd
+
+// expected-error@+1 {{unexpected OpenMP directive '#pragma omp taskloop simd'}}
+#pragma omp taskloop simd foo
+
+void test_no_clause() {
+  int i;
+#pragma omp taskloop simd
+  for (i = 0; i < 16; ++i)
+    ;
+
+// expected-error@+2 {{statement after '#pragma omp taskloop simd' must be a for loop}}
+#pragma omp taskloop simd
+  ++i;
+}
+
+void test_branch_protected_scope() {
+  int i = 0;
+L1:
+  ++i;
+
+  int x[24];
+
+#pragma omp parallel
+#pragma omp taskloop simd
+  for (i = 0; i < 16; ++i) {
+    if (i == 5)
+      goto L1; // expected-error {{use of undeclared label 'L1'}}
+    else if (i == 6)
+      return; // expected-error {{cannot return from OpenMP region}}
+    else if (i == 7)
+      goto L2;
+    else if (i == 8) {
+    L2:
+      x[i]++;
+    }
+  }
+
+  if (x[0] == 0)
+    goto L2; // expected-error {{use of undeclared label 'L2'}}
+  else if (x[1] == 1)
+    goto L1;
+}
+
+void test_invalid_clause() {
+  int i;
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+#pragma omp taskloop simd foo bar
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-error@+1 {{directive '#pragma omp taskloop simd' cannot contain more than one 'nogroup' clause}}
+#pragma omp taskloop simd nogroup nogroup
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_non_identifiers() {
+  int i, x;
+
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+#pragma omp taskloop simd;
+  for (i = 0; i < 16; ++i)
+    ;
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+#pragma omp parallel
+#pragma omp taskloop simd linear(x);
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+#pragma omp taskloop simd private(x);
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-warning@+1 {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+#pragma omp taskloop simd, private(x);
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+extern int foo();
+
+void test_collapse() {
+  int i;
+#pragma omp parallel
+// expected-error@+1 {{expected '('}}
+#pragma omp taskloop simd collapse
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop simd collapse(
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd collapse()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}} expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop simd collapse(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}  expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop simd collapse(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-warning@+2 {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+// expected-error@+1 {{expected '('}}
+#pragma omp taskloop simd collapse 4)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4,
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4, )
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+// expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4 4)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4, , 4)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+#pragma omp taskloop simd collapse(4)
+  for (int i1 = 0; i1 < 16; ++i1)
+    for (int i2 = 0; i2 < 16; ++i2)
+      for (int i3 = 0; i3 < 16; ++i3)
+        for (int i4 = 0; i4 < 16; ++i4)
+          foo();
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}}
+// expected-note@+1 {{to match this '('}} expected-note@+1 {{as specified in 'collapse' clause}}
+#pragma omp taskloop simd collapse(4, 8)
+  for (i = 0; i < 16; ++i)
+    ; // expected-error {{expected 4 for loops after '#pragma omp taskloop simd', but found only 1}}
+#pragma omp parallel
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp taskloop simd collapse(2.5)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expression is not an integer constant expression}}
+#pragma omp taskloop simd collapse(foo())
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
+#pragma omp taskloop simd collapse(-5)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
+#pragma omp taskloop simd collapse(0)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{argument to 'collapse' clause must be a strictly positive integer value}}
+#pragma omp taskloop simd collapse(5 - 5)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_private() {
+  int i;
+#pragma omp parallel
+// expected-error@+2 {{expected expression}}
+// expected-error@+1 {{expected ')'}} expected-note@+1 {{to match this '('}}
+#pragma omp taskloop simd private(
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop simd private(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop simd private(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd private()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd private(int)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected variable name}}
+#pragma omp taskloop simd private(0)
+  for (i = 0; i < 16; ++i)
+    ;
+
+  int x, y, z;
+#pragma omp parallel
+#pragma omp taskloop simd private(x)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop simd private(x, y)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop simd private(x, y, z)
+  for (i = 0; i < 16; ++i) {
+    x = y * i + z;
+  }
+}
+
+void test_lastprivate() {
+  int i;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd lastprivate(
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop simd lastprivate(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop simd lastprivate(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd lastprivate()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd lastprivate(int)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected variable name}}
+#pragma omp taskloop simd lastprivate(0)
+  for (i = 0; i < 16; ++i)
+    ;
+
+  int x, y, z;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(x)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(x, y)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(x, y, z)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_firstprivate() {
+  int i;
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd firstprivate(
+  for (i = 0; i < 16; ++i)
+    ;
+
+#pragma omp parallel
+// expected-error@+2 {{expected ')'}} expected-note@+2 {{to match this '('}}
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop simd firstprivate(,
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 2 {{expected expression}}
+#pragma omp taskloop simd firstprivate(, )
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd firstprivate()
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected expression}}
+#pragma omp taskloop simd firstprivate(int)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+// expected-error@+1 {{expected variable name}}
+#pragma omp taskloop simd firstprivate(0)
+  for (i = 0; i < 16; ++i)
+    ;
+
+  int x, y, z;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(x) firstprivate(x)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(x, y) firstprivate(x, y)
+  for (i = 0; i < 16; ++i)
+    ;
+#pragma omp parallel
+#pragma omp taskloop simd lastprivate(x, y, z) firstprivate(x, y, z)
+  for (i = 0; i < 16; ++i)
+    ;
+}
+
+void test_loop_messages() {
+  float a[100], b[100], c[100];
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or pointer type}}
+#pragma omp taskloop simd
+  for (float fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+#pragma omp parallel
+// expected-error@+2 {{variable must be of integer or pointer type}}
+#pragma omp taskloop simd
+  for (double fi = 0; fi < 10.0; fi++) {
+    c[(int)fi] = a[(int)fi] + b[(int)fi];
+  }
+
+  // expected-warning@+2 {{OpenMP loop iteration variable cannot have more than 64 bits size and will be narrowed}}
+  #pragma omp taskloop simd
+  for (__int128 ii = 0; ii < 10; ii++) {
+    c[ii] = a[ii] + b[ii];
+  }
+}
+
diff --git a/test/OpenMP/taskloop_simd_num_tasks_messages.cpp b/test/OpenMP/taskloop_simd_num_tasks_messages.cpp
new file mode 100644
index 000000000000..f26ee79a8740
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_num_tasks_messages.cpp
@@ -0,0 +1,99 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp taskloop simd num_tasks // expected-error {{expected '(' after 'num_tasks'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (foobool(argc)), num_tasks (true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'num_tasks' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(0) // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(-1) // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(argc) grainsize(argc) // expected-error {{'grainsize' and 'num_tasks' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'num_tasks' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop simd num_tasks // expected-error {{expected '(' after 'num_tasks'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (foobool(argc)), num_tasks (true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'num_tasks' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(0)  // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(-1) // expected-error {{argument to 'num_tasks' clause must be a strictly positive integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd num_tasks(argc) grainsize(argc) // expected-error {{'grainsize' and 'num_tasks' clause are mutually exclusive and may not appear on the same directive}} expected-note {{'num_tasks' clause is specified here}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_simd_priority_messages.cpp b/test/OpenMP/taskloop_simd_priority_messages.cpp
new file mode 100644
index 000000000000..a94798f821a6
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_priority_messages.cpp
@@ -0,0 +1,93 @@
+// RUN: %clang_cc1 -verify -fopenmp -ferror-limit 100 %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, class S> // expected-note {{declared here}}
+int tmain(T argc, S **argv) {
+  #pragma omp taskloop simd priority // expected-error {{expected '(' after 'priority'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (foobool(argc)), priority (true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'priority' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority(0)
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority(-1) // expected-error {{argument to 'priority' clause must be a non-negative integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop simd priority // expected-error {{expected '(' after 'priority'}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority () // expected-error {{expected expression}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc > 0 ? argv[1][0] : argv[2][argc])
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (foobool(argc)), priority (true) // expected-error {{directive '#pragma omp taskloop simd' cannot contain more than one 'priority' clause}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (argc argc) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority (1 0) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority(if(tmain(argc, argv) // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority(0)
+  for (int i = 0; i < 10; ++i)
+    foo();
+  #pragma omp taskloop simd priority(-1) // expected-error {{argument to 'priority' clause must be a non-negative integer value}}
+  for (int i = 0; i < 10; ++i)
+    foo();
+
+  return tmain(argc, argv);
+}
diff --git a/test/OpenMP/taskloop_simd_private_messages.cpp b/test/OpenMP/taskloop_simd_private_messages.cpp
new file mode 100644
index 000000000000..4a9b08a7e9e2
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_private_messages.cpp
@@ -0,0 +1,195 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}} expected-note 2 {{forward declaration of 'S1'}}
+extern S1 a;
+class S2 {
+  mutable int a;
+
+public:
+  S2() : a(0) {}
+};
+const S2 b;
+const S2 ba[5];
+class S3 {
+  int a;
+
+public:
+  S3() : a(0) {}
+};
+const S3 ca[5];
+class S4 {
+  int a;
+  S4(); // expected-note {{implicitly declared private here}}
+
+public:
+  S4(int v) : a(v) {}
+};
+class S5 {
+  int a;
+  S5() : a(0) {} // expected-note {{implicitly declared private here}}
+
+public:
+  S5(int v) : a(v) {}
+};
+
+S3 h;
+#pragma omp threadprivate(h) // expected-note 2 {{defined as threadprivate or thread local}}
+
+template <class I, class C>
+int foomain(I argc, C **argv) {
+  I e(4);
+  I g(5);
+  int i;
+  int &j = i;
+#pragma omp taskloop simd private // expected-error {{expected '(' after 'private'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(a, b) // expected-error {{private variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(e, g)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(h) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd shared(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int v = 0;
+    int i;
+#pragma omp taskloop simd private(i)
+    for (int k = 0; k < argc; ++k) {
+      i = k;
+      v += i;
+    }
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop simd private(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  return 0;
+}
+
+void bar(S4 a[2]) {
+#pragma omp parallel
+#pragma omp taskloop simd private(a)
+  for (int i = 0; i < 2; ++i)
+    foo();
+}
+
+namespace A {
+double x;
+#pragma omp threadprivate(x) // expected-note {{defined as threadprivate or thread local}}
+}
+namespace B {
+using A::x;
+}
+
+int main(int argc, char **argv) {
+  S4 e(4);
+  S5 g(5);
+  int i;
+  int &j = i;
+#pragma omp taskloop simd private // expected-error {{expected '(' after 'private'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private() // expected-error {{expected expression}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argc)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(S1) // expected-error {{'S1' does not refer to a value}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(a, b) // expected-error {{private variable with incomplete type 'S1'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(argv[1]) // expected-error {{expected variable name}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(h) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(B::x) // expected-error {{threadprivate or thread local variable cannot be private}}
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd shared(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp parallel
+  {
+    int i;
+#pragma omp taskloop simd private(i)
+    for (int k = 0; k < argc; ++k)
+      ++k;
+  }
+#pragma omp parallel shared(i)
+#pragma omp parallel private(i)
+#pragma omp taskloop simd private(j)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+#pragma omp taskloop simd private(i)
+  for (int k = 0; k < argc; ++k)
+    ++k;
+  static int si;
+#pragma omp taskloop simd private(si) // OK
+  for(int k = 0; k < argc; ++k)
+    si = k + 1;
+
+  return 0;
+}
+
diff --git a/test/OpenMP/taskloop_simd_safelen_messages.cpp b/test/OpenMP/taskloop_simd_safelen_messages.cpp
new file mode 100644
index 000000000000..3182c8af80fa
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_safelen_messages.cpp
@@ -0,0 +1,79 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp taskloop simd safelen // expected-error {{expected '(' after 'safelen'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp taskloop simd safelen (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 {{argument to 'safelen' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd safelen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen (1)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen ((ST > 0) ? 1 + ST : 2)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 2 {{directive '#pragma omp taskloop simd' cannot contain more than one 'safelen' clause}}
+  // expected-error@+2 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd safelen (foobool(argc)), safelen (true), safelen (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd safelen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen (4)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd safelen (N) // expected-error {{argument to 'safelen' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop simd safelen // expected-error {{expected '(' after 'safelen'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd safelen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd safelen () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd safelen (4 // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd safelen (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd safelen (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp taskloop simd' cannot contain more than one 'safelen' clause}}
+  // expected-error@+1 2 {{argument to 'safelen' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd safelen (foobool(argc)), safelen (true), safelen (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd safelen (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd safelen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp taskloop simd' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp taskloop simd safelen(safelen(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 12, 4>' requested here}}
+  return tmain<int, char, 12, 4>(argc, argv);
+}
+
diff --git a/test/OpenMP/taskloop_simd_simdlen_messages.cpp b/test/OpenMP/taskloop_simd_simdlen_messages.cpp
new file mode 100644
index 000000000000..ba3f20e23cfe
--- /dev/null
+++ b/test/OpenMP/taskloop_simd_simdlen_messages.cpp
@@ -0,0 +1,79 @@
+// RUN: %clang_cc1 -verify -fopenmp %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note {{declared here}}
+
+template <class T, typename S, int N, int ST> // expected-note {{declared here}}
+T tmain(T argc, S **argv) { //expected-note 2 {{declared here}}
+  #pragma omp taskloop simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen () // expected-error {{expected expression}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 {{expected ')'}} expected-note@+3 {{to match this '('}}
+  // expected-error@+2 2 {{expression is not an integral constant expression}}
+  // expected-note@+1 2 {{read of non-const variable 'argc' is not allowed in a constant expression}}
+  #pragma omp taskloop simd simdlen (argc 
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd simdlen (ST // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen (1)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen ((ST > 0) ? 1 + ST : 2)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+3 2 {{directive '#pragma omp taskloop simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+2 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd simdlen (foobool(argc)), simdlen (true), simdlen (-5)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen (S) // expected-error {{'S' does not refer to a value}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  // expected-error@+1 2 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen (4)
+  for (int i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  #pragma omp taskloop simd simdlen (N) // expected-error {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  for (T i = ST; i < N; i++) argv[0][i] = argv[0][i] - argv[0][i-ST];
+  return argc;
+}
+
+int main(int argc, char **argv) {
+  #pragma omp taskloop simd simdlen // expected-error {{expected '(' after 'simdlen'}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd simdlen ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd simdlen () // expected-error {{expected expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd simdlen (4 // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd simdlen (2+2)) // expected-warning {{extra tokens at the end of '#pragma omp taskloop simd' are ignored}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd simdlen (foobool(1) > 0 ? 1 : 2) // expected-error {{expression is not an integral constant expression}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{expression is not an integral constant expression}}
+  // expected-error@+2 2 {{directive '#pragma omp taskloop simd' cannot contain more than one 'simdlen' clause}}
+  // expected-error@+1 2 {{argument to 'simdlen' clause must be a strictly positive integer value}}
+  #pragma omp taskloop simd simdlen (foobool(argc)), simdlen (true), simdlen (-5) 
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  #pragma omp taskloop simd simdlen (S1) // expected-error {{'S1' does not refer to a value}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+1 {{expression is not an integral constant expression}}
+  #pragma omp taskloop simd simdlen (argv[1]=2) // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  for (int i = 4; i < 12; i++) argv[0][i] = argv[0][i] - argv[0][i-4];
+  // expected-error@+3 {{statement after '#pragma omp taskloop simd' must be a for loop}}
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, -1, -2>' requested here}}
+  #pragma omp taskloop simd simdlen(simdlen(tmain<int, char, -1, -2>(argc, argv) // expected-error 2 {{expected ')'}} expected-note 2 {{to match this '('}}
+  foo();
+  // expected-note@+1 {{in instantiation of function template specialization 'tmain<int, char, 12, 4>' requested here}}
+  return tmain<int, char, 12, 4>(argc, argv);
+}
+
diff --git a/test/OpenMP/teams_ast_print.cpp b/test/OpenMP/teams_ast_print.cpp
index 6cea914c8626..292586ae530d 100644
--- a/test/OpenMP/teams_ast_print.cpp
+++ b/test/OpenMP/teams_ast_print.cpp
@@ -37,7 +37,7 @@ T tmain(T argc, T *argv) {
 #pragma omp teams
   a=2;
 #pragma omp target
-#pragma omp teams default(none), private(argc,b) firstprivate(argv) shared (d) reduction(+:c) reduction(max:e)
+#pragma omp teams default(none), private(argc,b) firstprivate(argv) shared (d) reduction(+:c) reduction(max:e) num_teams(C) thread_limit(d*C)
   foo();
 #pragma omp target
 #pragma omp teams reduction(^:e, f) reduction(&& : g)
@@ -53,7 +53,7 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: #pragma omp teams
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: #pragma omp target
-// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) shared(d) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) shared(d) reduction(+: c) reduction(max: e) num_teams(5) thread_limit(d * 5)
 // CHECK-NEXT: foo()
 // CHECK-NEXT: #pragma omp target
 // CHECK-NEXT: #pragma omp teams reduction(^: e,f) reduction(&&: g)
@@ -66,7 +66,7 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: #pragma omp teams
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: #pragma omp target
-// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) shared(d) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) shared(d) reduction(+: c) reduction(max: e) num_teams(1) thread_limit(d * 1)
 // CHECK-NEXT: foo()
 // CHECK-NEXT: #pragma omp target
 // CHECK-NEXT: #pragma omp teams reduction(^: e,f) reduction(&&: g)
@@ -79,7 +79,7 @@ T tmain(T argc, T *argv) {
 // CHECK-NEXT: #pragma omp teams
 // CHECK-NEXT: a = 2;
 // CHECK-NEXT: #pragma omp target
-// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) shared(d) reduction(+: c) reduction(max: e)
+// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) shared(d) reduction(+: c) reduction(max: e) num_teams(C) thread_limit(d * C)
 // CHECK-NEXT: foo()
 // CHECK-NEXT: #pragma omp target
 // CHECK-NEXT: #pragma omp teams reduction(^: e,f) reduction(&&: g)
@@ -101,9 +101,9 @@ int main (int argc, char **argv) {
   a=2;
 // CHECK-NEXT: a = 2;
 #pragma omp target
-#pragma omp teams default(none), private(argc,b) firstprivate(argv) reduction(| : c, d) reduction(* : e)
+#pragma omp teams default(none), private(argc,b) num_teams(f) firstprivate(argv) reduction(| : c, d) reduction(* : e) thread_limit(f+g)
 // CHECK-NEXT: #pragma omp target
-// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) firstprivate(argv) reduction(|: c,d) reduction(*: e)
+// CHECK-NEXT: #pragma omp teams default(none) private(argc,b) num_teams(f) firstprivate(argv) reduction(|: c,d) reduction(*: e) thread_limit(f + g)
   foo();
 // CHECK-NEXT: foo();
   return tmain<int, 5>(b, &b) + tmain<long, 1>(x, &x);
diff --git a/test/OpenMP/teams_firstprivate_messages.cpp b/test/OpenMP/teams_firstprivate_messages.cpp
index ac5197749bb4..ba12fde86b5c 100644
--- a/test/OpenMP/teams_firstprivate_messages.cpp
+++ b/test/OpenMP/teams_firstprivate_messages.cpp
@@ -63,7 +63,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
 #pragma omp target
 #pragma omp teams firstprivate // expected-error {{expected '(' after 'firstprivate'}}
   foo();
@@ -125,7 +125,11 @@ int main(int argc, char **argv) {
 #pragma omp teams firstprivate(i)
   foo();
 #pragma omp target
-#pragma omp teams firstprivate(j) // expected-error {{arguments of OpenMP clause 'firstprivate' cannot be of reference type}}
+#pragma omp teams firstprivate(j)
+  foo();
+  static int m;
+#pragma omp target
+#pragma omp teams firstprivate(m) // OK
   foo();
 
   return 0;
diff --git a/test/OpenMP/teams_num_teams_messages.cpp b/test/OpenMP/teams_num_teams_messages.cpp
new file mode 100644
index 000000000000..01871127bda5
--- /dev/null
+++ b/test/OpenMP/teams_num_teams_messages.cpp
@@ -0,0 +1,111 @@
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 100 -o - %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}}
+
+template <typename T, int C> // expected-note {{declared here}}
+T tmain(T argc) {
+  char **a;
+#pragma omp target
+#pragma omp teams num_teams(C)
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(T) // expected-error {{'T' does not refer to a value}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams // expected-error {{expected '(' after 'num_teams'}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams() // expected-error {{expected expression}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(argc)) // expected-warning {{extra tokens at the end of '#pragma omp teams' are ignored}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(argc > 0 ? a[1] : a[2]) // expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(argc + argc)
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(argc), num_teams (argc+1) // expected-error {{directive '#pragma omp teams' cannot contain more than one 'num_teams' clause}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(S1) // expected-error {{'S1' does not refer to a value}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(-2) // expected-error {{argument to 'num_teams' clause must be a strictly positive integer value}}
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(-10u)
+  foo();
+#pragma omp target
+#pragma omp teams num_teams(3.14) // expected-error 2 {{expression must have integral or unscoped enumeration type, not 'double'}}
+  foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+#pragma omp target
+#pragma omp teams num_teams // expected-error {{expected '(' after 'num_teams'}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams () // expected-error {{expected expression}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (argc)) // expected-warning {{extra tokens at the end of '#pragma omp teams' are ignored}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (argc > 0 ? argv[1] : argv[2]) // expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (argc + argc)
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (argc), num_teams (argc+1) // expected-error {{directive '#pragma omp teams' cannot contain more than one 'num_teams' clause}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (S1) // expected-error {{'S1' does not refer to a value}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (-2) // expected-error {{argument to 'num_teams' clause must be a strictly positive integer value}}
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (-10u)
+  foo();
+
+#pragma omp target
+#pragma omp teams num_teams (3.14) // expected-error {{expression must have integral or unscoped enumeration type, not 'double'}}
+  foo();
+
+  return tmain<int, 10>(argc); // expected-note {{in instantiation of function template specialization 'tmain<int, 10>' requested here}}
+}
diff --git a/test/OpenMP/teams_private_messages.cpp b/test/OpenMP/teams_private_messages.cpp
index f50060ff29b3..344ef8daf996 100644
--- a/test/OpenMP/teams_private_messages.cpp
+++ b/test/OpenMP/teams_private_messages.cpp
@@ -13,7 +13,7 @@ class S2 {
   mutable int a;
 public:
   S2():a(0) { }
-  static float S2s;
+  static float S2s; // expected-note {{static data member is predetermined as shared}}
 };
 const S2 b;
 const S2 ba[5];
@@ -55,7 +55,7 @@ int main(int argc, char **argv) {
   S4 e(4);
   S5 g(5);
   int i;
-  int &j = i; // expected-note {{'j' defined here}}
+  int &j = i;
   #pragma omp target
   #pragma omp teams private // expected-error {{expected '(' after 'private'}}
   foo();
@@ -96,7 +96,7 @@ int main(int argc, char **argv) {
   #pragma omp teams private(da) // expected-error {{shared variable cannot be private}}
   foo();
   #pragma omp target
-  #pragma omp teams private(S2::S2s)
+  #pragma omp teams private(S2::S2s) // expected-error {{shared variable cannot be private}}
   foo();
   #pragma omp target
   #pragma omp teams private(e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{calling a private constructor of class 'S5'}}
@@ -114,7 +114,7 @@ int main(int argc, char **argv) {
   #pragma omp teams private(i)
   foo();
   #pragma omp target
-  #pragma omp teams private(j) // expected-error {{arguments of OpenMP clause 'private' cannot be of reference type 'int &'}}
+  #pragma omp teams private(j)
   foo();
   #pragma omp target
   #pragma omp teams firstprivate(i)
@@ -122,6 +122,10 @@ int main(int argc, char **argv) {
     #pragma omp parallel private(i)
     foo();
   }
+  static int m;
+  #pragma omp target
+  #pragma omp teams private(m) // OK
+  foo();
 
   return 0;
 }
diff --git a/test/OpenMP/teams_reduction_messages.cpp b/test/OpenMP/teams_reduction_messages.cpp
index 93c4b141591f..87d03485c17b 100644
--- a/test/OpenMP/teams_reduction_messages.cpp
+++ b/test/OpenMP/teams_reduction_messages.cpp
@@ -1,4 +1,6 @@
 // RUN: %clang_cc1 -verify -fopenmp -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++98 -o - %s
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -o - %s
 
 void foo() {
 }
@@ -16,7 +18,7 @@ class S2 {
 public:
   S2() : a(0) {}
   S2(S2 &s2) : a(s2.a) {}
-  static float S2s;
+  static float S2s; // expected-note 2 {{static data member is predetermined as shared}}
   static const float S2sc;
 };
 const float S2::S2sc = 0; // expected-note 2 {{'S2sc' defined here}}
@@ -26,6 +28,7 @@ class S3 {
   int a;
 
 public:
+  int b;
   S3() : a(0) {}
   S3(const S3 &s3) : a(s3.a) {}
   S3 operator+(const S3 &arg1) { return arg1; }
@@ -54,6 +57,9 @@ public:
   S5(int v) : a(v) {}
 };
 class S6 { // expected-note 2 {{candidate function (the implicit copy assignment operator) not viable: no known conversion from 'int' to 'const S6' for 1st argument}}
+#if __cplusplus >= 201103L // C++11 or later
+// expected-note@-2 2 {{candidate function (the implicit move assignment operator) not viable}}
+#endif
   int a;
 
 public:
@@ -103,7 +109,7 @@ T tmain(T argc) {
 #pragma omp teams reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}} expected-error {{invalid operands to binary expression ('float' and 'float')}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name}}
+#pragma omp teams reduction(|| : argc ? i : argc) // expected-error 2 {{expected variable name, array element or array section}}
   foo();
 #pragma omp target
 #pragma omp teams reduction(foo : argc) //expected-error {{incorrect reduction identifier, expected one of '+', '-', '*', '&', '|', '^', '&&', '||', 'min' or 'max'}}
@@ -115,31 +121,31 @@ T tmain(T argc) {
 #pragma omp teams reduction(^ : T) // expected-error {{'T' does not refer to a value}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 3 {{const-qualified variable cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
+#pragma omp teams reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 3 {{const-qualified list item cannot be reduction}} expected-error 3 {{'operator+' is a private member of 'S2'}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified variable cannot be reduction}}
+#pragma omp teams reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 3 {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(max : qa[1]) // expected-error 2 {{expected variable name}}
+#pragma omp teams reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp teams reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp teams reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}} expected-error {{a reduction variable with array type 'const float [5]'}}
+#pragma omp teams reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}} expected-error {{a reduction list item with array type 'const float [5]'}}
   foo();
 #pragma omp target
 #pragma omp teams reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(&& : S2::S2s)
+#pragma omp teams reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp teams reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp target
 #pragma omp teams reduction(+ : h, k) // expected-error {{threadprivate or thread local variable cannot be reduction}}
@@ -158,7 +164,7 @@ T tmain(T argc) {
 #pragma omp teams reduction(+ : p), reduction(+ : p) // expected-error 3 {{variable can appear only once in OpenMP 'reduction' clause}} expected-note 3 {{previously referenced here}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(+ : r) // expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp teams reduction(+ : r) // expected-error 2 {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp parallel shared(i)
 #pragma omp parallel reduction(min : i)
@@ -231,7 +237,7 @@ int main(int argc, char **argv) {
 #pragma omp teams reduction(| : argc, // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name}}
+#pragma omp teams reduction(|| : argc > 0 ? argv[1] : argv[2]) // expected-error {{expected variable name, array element or array section}}
   foo();
 #pragma omp target
 #pragma omp teams reduction(~ : argc) // expected-error {{expected unqualified-id}}
@@ -243,31 +249,31 @@ int main(int argc, char **argv) {
 #pragma omp teams reduction(^ : S1) // expected-error {{'S1' does not refer to a value}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(+ : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{const-qualified variable cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
+#pragma omp teams reduction(+ : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{const-qualified list item cannot be reduction}} expected-error {{'operator+' is a private member of 'S2'}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(min : a, b, c, d, f) // expected-error {{reduction variable with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified variable cannot be reduction}}
+#pragma omp teams reduction(min : a, b, c, d, f) // expected-error {{a reduction list item with incomplete type 'S1'}} expected-error 2 {{arguments of OpenMP clause 'reduction' for 'min' or 'max' must be of arithmetic type}} expected-error 2 {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(max : argv[1]) // expected-error {{expected variable name}}
+#pragma omp teams reduction(max : h.b) // expected-error {{expected variable name, array element or array section}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(+ : ba) // expected-error {{a reduction variable with array type 'const S2 [5]'}}
+#pragma omp teams reduction(+ : ba) // expected-error {{a reduction list item with array type 'const S2 [5]'}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(* : ca) // expected-error {{a reduction variable with array type 'const S3 [5]'}}
+#pragma omp teams reduction(* : ca) // expected-error {{a reduction list item with array type 'const S3 [5]'}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(- : da) // expected-error {{a reduction variable with array type 'const int [5]'}}
+#pragma omp teams reduction(- : da) // expected-error {{a reduction list item with array type 'const int [5]'}}
   foo();
 #pragma omp target
 #pragma omp teams reduction(^ : fl) // expected-error {{invalid operands to binary expression ('float' and 'float')}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(&& : S2::S2s)
+#pragma omp teams reduction(&& : S2::S2s) // expected-error {{shared variable cannot be reduction}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(&& : S2::S2sc) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp teams reduction(&& : S2::S2sc) // expected-error {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp target
 #pragma omp teams reduction(& : e, g) // expected-error {{calling a private constructor of class 'S4'}} expected-error {{invalid operands to binary expression ('S4' and 'S4')}} expected-error {{calling a private constructor of class 'S5'}} expected-error {{invalid operands to binary expression ('S5' and 'S5')}}
@@ -289,7 +295,7 @@ int main(int argc, char **argv) {
 #pragma omp teams reduction(+ : p), reduction(+ : p) // expected-error {{variable can appear only once in OpenMP 'reduction' clause}} expected-note {{previously referenced here}}
   foo();
 #pragma omp target
-#pragma omp teams reduction(+ : r) // expected-error {{const-qualified variable cannot be reduction}}
+#pragma omp teams reduction(+ : r) // expected-error {{const-qualified list item cannot be reduction}}
   foo();
 #pragma omp parallel shared(i)
 #pragma omp parallel reduction(min : i)
@@ -310,6 +316,10 @@ int main(int argc, char **argv) {
 #pragma omp target
 #pragma omp teams reduction(+ : fl)
     foo();
+  static int m;
+#pragma omp target
+#pragma omp teams reduction(+ : m) // OK
+  foo();
 
   return tmain(argc) + tmain(fl); // expected-note {{in instantiation of function template specialization 'tmain<int>' requested here}} expected-note {{in instantiation of function template specialization 'tmain<float>' requested here}}
 }
diff --git a/test/OpenMP/teams_thread_limit_messages.cpp b/test/OpenMP/teams_thread_limit_messages.cpp
new file mode 100644
index 000000000000..1cb147c60781
--- /dev/null
+++ b/test/OpenMP/teams_thread_limit_messages.cpp
@@ -0,0 +1,111 @@
+// RUN: %clang_cc1 -verify -fopenmp -std=c++11 -ferror-limit 100 -o - %s
+
+void foo() {
+}
+
+bool foobool(int argc) {
+  return argc;
+}
+
+struct S1; // expected-note 2 {{declared here}}
+
+template <typename T, int C> // expected-note {{declared here}}
+T tmain(T argc) {
+  char **a;
+#pragma omp target
+#pragma omp teams thread_limit(C)
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(T) // expected-error {{'T' does not refer to a value}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit // expected-error {{expected '(' after 'thread_limit'}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit() // expected-error {{expected expression}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(argc)) // expected-warning {{extra tokens at the end of '#pragma omp teams' are ignored}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(argc > 0 ? a[1] : a[2]) // expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(argc + argc)
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(argc), thread_limit (argc+1) // expected-error {{directive '#pragma omp teams' cannot contain more than one 'thread_limit' clause}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(S1) // expected-error {{'S1' does not refer to a value}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(-2) // expected-error {{argument to 'thread_limit' clause must be a strictly positive integer value}}
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(-10u)
+  foo();
+#pragma omp target
+#pragma omp teams thread_limit(3.14) // expected-error 2 {{expression must have integral or unscoped enumeration type, not 'double'}}
+  foo();
+
+  return 0;
+}
+
+int main(int argc, char **argv) {
+#pragma omp target
+#pragma omp teams thread_limit // expected-error {{expected '(' after 'thread_limit'}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit ( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit () // expected-error {{expected expression}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (argc // expected-error {{expected ')'}} expected-note {{to match this '('}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (argc)) // expected-warning {{extra tokens at the end of '#pragma omp teams' are ignored}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (argc > 0 ? argv[1] : argv[2]) // expected-error {{expression must have integral or unscoped enumeration type, not 'char *'}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (argc + argc)
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (argc), thread_limit (argc+1) // expected-error {{directive '#pragma omp teams' cannot contain more than one 'thread_limit' clause}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (S1) // expected-error {{'S1' does not refer to a value}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (-2) // expected-error {{argument to 'thread_limit' clause must be a strictly positive integer value}}
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (-10u)
+  foo();
+
+#pragma omp target
+#pragma omp teams thread_limit (3.14) // expected-error {{expression must have integral or unscoped enumeration type, not 'double'}}
+  foo();
+
+  return tmain<int, 10>(argc); // expected-note {{in instantiation of function template specialization 'tmain<int, 10>' requested here}}
+}
diff --git a/test/OpenMP/threadprivate_codegen.cpp b/test/OpenMP/threadprivate_codegen.cpp
index 7176e0499273..53d7ef708385 100644
--- a/test/OpenMP/threadprivate_codegen.cpp
+++ b/test/OpenMP/threadprivate_codegen.cpp
@@ -1,10 +1,10 @@
 // RUN: %clang_cc1 -verify -fopenmp -fnoopenmp-use-tls -DBODY -triple x86_64-unknown-unknown -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s
 // RUN: %clang_cc1 -fopenmp -fnoopenmp-use-tls -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -fnoopenmp-use-tls -DBODY -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-DEBUG %s
+// RUN: %clang_cc1 -fopenmp -fnoopenmp-use-tls -DBODY -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-DEBUG %s
 
 // RUN: %clang_cc1 -verify -fopenmp -DBODY -triple x86_64-unknown-unknown -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK-TLS
 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
-// RUN: %clang_cc1 -fopenmp -DBODY -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -g -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-TLS %s
+// RUN: %clang_cc1 -fopenmp -DBODY -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix=CHECK-TLS %s
 
 // expected-no-diagnostics
 // REQUIRES: x86-registered-target
@@ -177,11 +177,11 @@ struct S5 {
 // CHECK-TLS-DAG:  [[ST_S4_ST_GUARD:@_ZGVN2STI2S4E2stE]] = linkonce_odr thread_local global i64 0
 // CHECK-TLS-DAG:  @__tls_guard = internal thread_local global i8 0
 // CHECK-TLS-DAG:  @__dso_handle = external global i8
-// CHECK-TLS-DAG:  [[GS1_TLS_INIT:@_ZTHL3gs1]] = internal alias void ()* @__tls_init
-// CHECK-TLS-DAG:  [[ARR_X_TLS_INIT:@_ZTH5arr_x]] = alias void ()* @__tls_init
-// CHECK-TLS-DAG:  [[ST_INT_ST_TLS_INIT:@_ZTHN2STIiE2stE]] = linkonce_odr alias void ()* @__tls_init
-// CHECK-TLS-DAG:  [[ST_FLOAT_ST_TLS_INIT:@_ZTHN2STIfE2stE]] = linkonce_odr alias void ()* @__tls_init
-// CHECK-TLS-DAG:  [[ST_S4_ST_TLS_INIT:@_ZTHN2STI2S4E2stE]] = linkonce_odr alias void ()* @__tls_init
+// CHECK-TLS-DAG:  [[GS1_TLS_INIT:@_ZTHL3gs1]] = internal alias void (), void ()* @__tls_init
+// CHECK-TLS-DAG:  [[ARR_X_TLS_INIT:@_ZTH5arr_x]] = alias void (), void ()* @__tls_init
+// CHECK-TLS-DAG:  [[ST_INT_ST_TLS_INIT:@_ZTHN2STIiE2stE]] = linkonce_odr alias void (), void ()* @__tls_init
+// CHECK-TLS-DAG:  [[ST_FLOAT_ST_TLS_INIT:@_ZTHN2STIfE2stE]] = linkonce_odr alias void (), void ()* @__tls_init
+// CHECK-TLS-DAG:  [[ST_S4_ST_TLS_INIT:@_ZTHN2STI2S4E2stE]] = linkonce_odr alias void (), void ()* @__tls_init
 
 struct Static {
   static S3 s;
@@ -939,11 +939,11 @@ int foobar() {
 // CHECK-TLS:      define internal void @__tls_init()
 // CHECK-TLS:      [[GRD:%.*]] = load i8, i8* @__tls_guard
 // CHECK-TLS-NEXT: [[IS_INIT:%.*]] = icmp eq i8 [[GRD]], 0
-// CHECK-TLS-NEXT: store i8 1, i8* @__tls_guard
 // CHECK-TLS-NEXT: br i1 [[IS_INIT]], label %[[INIT_LABEL:[^,]+]], label %[[DONE_LABEL:[^,]+]]{{.*}}
 // CHECK-TLS:      [[INIT_LABEL]]
+// CHECK-TLS-NEXT: store i8 1, i8* @__tls_guard
 // CHECK-TLS:      call void [[GS1_CXX_INIT]]
-// CHECK-TLS-NOT   call void [[GS2_CXX_INIT]]
+// CHECK-TLS-NOT:  call void [[GS2_CXX_INIT]]
 // CHECK-TLS:      call void [[ARR_X_CXX_INIT]]
 // CHECK-TLS:      call void [[ST_S4_ST_CXX_INIT]]
 // CHECK-TLS:      [[DONE_LABEL]]
diff --git a/test/OpenMP/threadprivate_messages.cpp b/test/OpenMP/threadprivate_messages.cpp
index 39a4431f0c5b..f71d58bc52a4 100644
--- a/test/OpenMP/threadprivate_messages.cpp
+++ b/test/OpenMP/threadprivate_messages.cpp
@@ -96,7 +96,10 @@ class TempClass {
 static __thread int t; // expected-note {{'t' defined here}}
 #pragma omp threadprivate (t) // expected-error {{variable 't' cannot be threadprivate because it is thread-local}}
 
-register int reg0 __asm__("0"); // expected-note {{'reg0' defined here}}
+// Register "0" is currently an invalid register for global register variables.
+// Use "esp" instead of "0".
+// register int reg0 __asm__("0");
+register int reg0 __asm__("esp"); // expected-note {{'reg0' defined here}}
 #pragma omp threadprivate (reg0) // expected-error {{variable 'reg0' cannot be threadprivate because it is a global named register variable}}
 
 int o; // expected-note {{candidate found by name lookup is 'o'}}