mei: reduce flow control only for completed messages

This fixes bug when only first chunk of a large message split
by hbuf_max_len is written to the hardware.
All the consequent chunks will not get a new credit.

A regression introduced by the commit
0ef319c93cebff9f82bdd0cdbb298f2dd00acda8
mei: streamline write complete flow function

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/misc/mei/interrupt.c

@@ -356,13 +356,14 @@ static int mei_irq_thread_write_complete(struct mei_device *dev, s32 *slots,
 		return -ENODEV;
 	}
 
-	if (mei_cl_flow_ctrl_reduce(cl))
-		return -ENODEV;
 
 	cl->status = 0;
 	cb->buf_idx += mei_hdr.length;
-	if (mei_hdr.msg_complete)
+	if (mei_hdr.msg_complete) {
+		if (mei_cl_flow_ctrl_reduce(cl))
+			return -ENODEV;
 		list_move_tail(&cb->list, &dev->write_waiting_list.list);
+	}
 
 	return 0;
 }