runtime: fix a gpgpu event and thread local gpgpu handling bug.

When pending a command queue, we need to record the whole gpgpu
structure not just the batch buffer. For the following reason:

1. We need to keep those private buffer, for example those printf buffers.
2. We need to make sure this gpgpu will not be reused by other enqueuement.

v2:
Don't try to flush all user event attached to the queue.
Just need to flush the current event when doing command queue flush.

Signed-off-by: Zhigang Gong <zhigang.gong@intel.com>
Reviewed-by: "Yang, Rong R" <rong.r.yang@intel.com>

1 files changed, 4 insertions, 0 deletions

diff --git a/src/cl_command_queue.h b/src/cl_command_queue.h
index b79d63ad..bd70f255 100644
--- a/src/cl_command_queue.h
+++ b/src/cl_command_queue.h
@@ -41,6 +41,7 @@ struct _cl_command_queue {
   cl_int    wait_events_num;           /* Number of Non-complete user events */
   cl_int    wait_events_size;          /* The size of array that wait_events point to */
   cl_event  last_event;                /* The last event in the queue, for enqueue mark used */
+  cl_event  current_event;             /* Current event. */
   cl_command_queue_properties  props;  /* Queue properties */
   cl_command_queue prev, next;         /* We chain the command queues together */
   void *thread_data;                   /* Used to store thread context data */
@@ -82,6 +83,9 @@ cl_int cl_command_queue_set_fulsim_buffer(cl_command_queue, cl_mem);
 /* Flush for the command queue */
 extern cl_int cl_command_queue_flush(cl_command_queue);
 
+/* Flush for the specified gpgpu */
+extern void cl_command_queue_flush_gpgpu(cl_command_queue, cl_gpgpu);
+
 /* Wait for the completion of the command queue */
 extern cl_int cl_command_queue_finish(cl_command_queue);