Minor note: enbd_clr_kernel_queue: shouldn't "schedule_timeout(1);" be preceded by a "set_current_state(TASK_INTERRUPTIBLE);" (io_request_lock is my current suspect). Regards Anders