Index: cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
===================================================================
--- cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
+++ cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
@@ -886,6 +886,11 @@
   bool DynamicWithOrderedClause =
       Dynamic && S.getSingleClause(OMPC_ordered) != nullptr;
   SourceLocation Loc = S.getLocStart();
+  // Generate !llvm.loop.parallel metadata for loads and stores for loops with
+  // dynamic/guided scheduling and without ordered clause.
+  LoopStack.setParallel((ScheduleKind == OMPC_SCHEDULE_dynamic ||
+                         ScheduleKind == OMPC_SCHEDULE_guided) &&
+                        !DynamicWithOrderedClause);
   EmitOMPInnerLoop(
       S, LoopScope.requiresCleanups(), S.getCond(/*SeparateIter=*/false),
       S.getInc(),
Index: cfe/trunk/test/OpenMP/for_codegen.cpp
===================================================================
--- cfe/trunk/test/OpenMP/for_codegen.cpp
+++ cfe/trunk/test/OpenMP/for_codegen.cpp
@@ -41,6 +41,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK-NOT: !llvm.mem.parallel_loop_access
     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
@@ -81,6 +82,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK-NOT: !llvm.mem.parallel_loop_access
     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
@@ -129,6 +131,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK-NOT: !llvm.mem.parallel_loop_access
     a[i] = b[i] * c[i] * d[i];
 // CHECK: [[IV1_2:%.+]] = load i32, i32* [[OMP_IV]]{{.*}}
 // CHECK-NEXT: [[ADD1_2:%.+]] = add i32 [[IV1_2]], 1
@@ -180,7 +183,7 @@
 // CHECK-NEXT: store i64 [[CALC_I_2]], i64* [[LC_I:.+]]
 // ... loop body ...
 // End of body: store into a[i]:
-// CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK: store float [[RESULT:%.+]], float* {{%.+}}!llvm.mem.parallel_loop_access
     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
@@ -221,7 +224,7 @@
 // CHECK-NEXT: store i64 [[CALC_I_2]], i64* [[LC_I:.+]]
 // ... loop body ...
 // End of body: store into a[i]:
-// CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK: store float [[RESULT:%.+]], float* {{%.+}}!llvm.mem.parallel_loop_access
     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
@@ -266,6 +269,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK-NOT: !llvm.mem.parallel_loop_access
     a[i] = b[i] * c[i] * d[i];
 // CHECK: [[IV1_2:%.+]] = load i64, i64* [[OMP_IV]]{{.*}}
 // CHECK-NEXT: [[ADD1_2:%.+]] = add nsw i64 [[IV1_2]], 1
@@ -307,6 +311,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// CHECK-NOT: !llvm.mem.parallel_loop_access
     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
Index: cfe/trunk/test/OpenMP/ordered_codegen.cpp
===================================================================
--- cfe/trunk/test/OpenMP/ordered_codegen.cpp
+++ cfe/trunk/test/OpenMP/ordered_codegen.cpp
@@ -39,6 +39,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// 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
@@ -86,6 +87,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// 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
@@ -139,6 +141,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// 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
@@ -189,6 +192,7 @@
 // ... loop body ...
 // End of body: store into a[i]:
 // CHECK: store float [[RESULT:%.+]], float* {{%.+}}
+// 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