mpl: add MPL_gpu_stream_synchronize

Useful for GPU stream based MPI extensions.

src/mpl/include/mpl_gpu.h

@@ -61,7 +61,7 @@ typedef enum {
     MPL_GPU_COPY_H2D,
     MPL_GPU_COPY_D2D_INCOMING,  /* copy from remote to local */
     MPL_GPU_COPY_D2D_OUTGOING,  /* copy from local to remote */
-    MPL_GPU_COPY_DIRECTION_NONE,  /* copy in any direction and to/from any buffer type */
+    MPL_GPU_COPY_DIRECTION_NONE,        /* copy in any direction and to/from any buffer type */
 } MPL_gpu_copy_direction_t;
 
 #define MPL_GPU_COPY_DIRECTION_TYPES 4
@@ -152,6 +152,7 @@ int MPL_gpu_launch_hostfn(MPL_gpu_stream_t stream, MPL_gpu_hostfn fn, void *data
 bool MPL_gpu_stream_is_valid(MPL_gpu_stream_t stream);
 void MPL_gpu_enqueue_trigger(MPL_gpu_event_t * var, MPL_gpu_stream_t stream);
 void MPL_gpu_enqueue_wait(MPL_gpu_event_t * var, MPL_gpu_stream_t stream);
+int MPL_gpu_stream_synchronize(MPL_gpu_stream_t stream);
 
 /* the synchronization event has the similar semantics as completion counter,
  * init to a count, then each completion decrement it by 1. */

src/mpl/src/gpu/mpl_gpu_cuda.c

@@ -589,3 +589,10 @@ bool MPL_gpu_stream_is_valid(MPL_gpu_stream_t stream)
     result = cudaStreamQuery(stream);
     return (result != cudaErrorInvalidResourceHandle);
 }
+
+int MPL_gpu_stream_synchronize(cudaStream_t stream)
+{
+    cudaError_t result;
+    result = cudaStreamSynchronize(stream);
+    return result;
+}

src/mpl/src/gpu/mpl_gpu_fallback.c

@@ -160,6 +160,11 @@ int MPL_gpu_launch_hostfn(int stream, MPL_gpu_hostfn fn, void *data)
     return -1;
 }
 
+int MPL_gpu_stream_synchronize(int stream)
+{
+    return -1;
+}
+
 bool MPL_gpu_stream_is_valid(MPL_gpu_stream_t stream)
 {
     return false;

src/mpl/src/gpu/mpl_gpu_hip.c

@@ -530,6 +530,13 @@ int MPL_gpu_launch_hostfn(hipStream_t stream, MPL_gpu_hostfn fn, void *data)
     return result;
 }
 
+int MPL_gpu_stream_synchronize(hipStream_t stream)
+{
+    hipError_t result;
+    result = hipStreamSynchronize(stream);
+    return result;
+}
+
 /* ---- */
 bool MPL_gpu_stream_is_valid(MPL_gpu_stream_t stream)
 {

src/mpl/src/gpu/mpl_gpu_ze.c

@@ -45,7 +45,7 @@ static int gpu_initialized = 0;
 static uint32_t device_count;   /* Counts all local devices, does not include subdevices */
 static uint32_t local_ze_device_count;  /* Counts all local devices and subdevices */
 static uint32_t global_ze_device_count; /* Counts all global devices and subdevices */
-static int max_dev_id;  /* Does not include subdevices */
+static int max_dev_id;          /* Does not include subdevices */
 static int max_subdev_id;
 static char **device_list = NULL;
 static int *engine_conversion = NULL;
@@ -2460,6 +2460,11 @@ int MPL_gpu_launch_hostfn(int stream, MPL_gpu_hostfn fn, void *data)
     return -1;
 }
 
+int MPL_gpu_stream_synchronize(int stream)
+{
+    return -1;
+}
+
 bool MPL_gpu_stream_is_valid(MPL_gpu_stream_t stream)
 {
     return false;