diff --git a/openmp/libomptarget/include/omptarget.h b/openmp/libomptarget/include/omptarget.h
--- a/openmp/libomptarget/include/omptarget.h
+++ b/openmp/libomptarget/include/omptarget.h
@@ -119,12 +119,23 @@
       *EntriesEnd; // End of the table with all the entries (non inclusive)
 };
 
+// clang-format on
+
+struct DeviceTy;
+
 /// This struct contains information exchanged between different asynchronous
 /// operations for device-dependent optimization and potential synchronization
 struct __tgt_async_info {
-  // A pointer to a queue-like structure where offloading operations are issued.
-  // We assume to use this structure to do synchronization. In CUDA backend, it
-  // is CUstream.
+  /// Synchronize the queue with \p Device, if necessary, and resets it to
+  /// indicate no pending actions.
+  ///
+  /// \returns OFFLOAD_FAIL or OFFLOAD_SUCCESS appropriately.
+  int synchronize(DeviceTy &Device);
+
+  /// A pointer to a queue-like structure where offloading operations are
+  /// issued. We assume to use this structure to do synchronization. In CUDA
+  /// backend, it is CUstream. If it is a nullptr there are no pending actions
+  /// that are not synchronized yet.
   void *Queue = nullptr;
 };
 
@@ -135,8 +146,6 @@
   uint64_t Stride;
 };
 
-// clang-format on
-
 #ifdef __cplusplus
 extern "C" {
 #endif
diff --git a/openmp/libomptarget/src/omptarget.cpp b/openmp/libomptarget/src/omptarget.cpp
--- a/openmp/libomptarget/src/omptarget.cpp
+++ b/openmp/libomptarget/src/omptarget.cpp
@@ -19,6 +19,17 @@
 #include <cassert>
 #include <vector>
 
+int __tgt_async_info::synchronize(DeviceTy &Device) {
+  int Result = OFFLOAD_SUCCESS;
+  if (Queue) {
+    // If we have a queue we need to synchronize it now. As it is synchronized
+    // afterwards we can get rid of the queue indicating no pending actions.
+    Result = Device.synchronize(this);
+    Queue = nullptr;
+  }
+  return Result;
+}
+
 /* All begin addresses for partially mapped structs must be 8-aligned in order
  * to ensure proper alignment of members. E.g.
  *
@@ -452,17 +463,6 @@
       : HstPtrBegin(HstPtr), DataSize(Size), ForceDelete(ForceDelete),
         HasCloseModifier(HasCloseModifier) {}
 };
-
-/// Synchronize device
-static int syncDevice(DeviceTy &Device, __tgt_async_info *AsyncInfo) {
-  assert(AsyncInfo && AsyncInfo->Queue && "Invalid AsyncInfo");
-  if (Device.synchronize(AsyncInfo) != OFFLOAD_SUCCESS) {
-    REPORT("Failed to synchronize device.\n");
-    return OFFLOAD_FAIL;
-  }
-
-  return OFFLOAD_SUCCESS;
-}
 } // namespace
 
 /// Internal function to undo the mapping and retrieve the data from the device.
@@ -642,8 +642,8 @@
   // nullptr, there is no data transfer happened because once there is,
   // AsyncInfo->Queue will not be nullptr, so again, we don't need to
   // synchronize.
-  if (AsyncInfo && AsyncInfo->Queue) {
-    Ret = syncDevice(Device, AsyncInfo);
+  if (AsyncInfo) {
+    Ret = AsyncInfo->synchronize(Device);
     if (Ret != OFFLOAD_SUCCESS)
       return OFFLOAD_FAIL;
   }
@@ -1322,11 +1322,13 @@
       REPORT("Failed to process data after launching the kernel.\n");
       return OFFLOAD_FAIL;
     }
-  } else if (AsyncInfo->Queue) {
+  } else {
+    // TODO: We should not synchronize here but on the outer level once we pass
+    // in a reference AsyncInfo object.
     // If ArgNum is zero, but AsyncInfo.Queue is valid, then the kernel doesn't
     // hava any argument, and the device supports async operations, so we need a
     // sync at this point.
-    return syncDevice(Device, AsyncInfo);
+    return AsyncInfo->synchronize(Device);
   }
 
   return OFFLOAD_SUCCESS;