| 摘要 |
<p>Each of object-oriented device drivers is described in terms of a multi-thread object which can allocate a message processing thread and an interrupt processing thread to be exclusively used for each interrupt. The device driver, when executing a processing based on a message received from another object, assigns the processing to the message processing thread. When an event has occurred requesting an interrupt to the device driver, the device driver executes the interrupt processing corresponding to the event in the interrupt processing thread corresponding to the event. In case that the corresponding processing thread is busy due to execution of another interrupt processing in response to an earlier interrupt when the event has occurred, the interrupt processing corresponding to the event is held without being executed regardless of the state of the message processing thread, and is executed only after completion of the interrupt processing for the earlier interrupt that is under execution in the interrupt processing thread. Interrupt latency is shortened without affecting advantages brought about by the use of the object-oriented device drivers. <IMAGE></p> |
