| 摘要 |
PURPOSE: When there is an error on SCSI(Small Computer System Interface) bus, The damage of data in a hard disk is prevented by reconnecting a corresponding command, and retrying that until completed normally, not by returning the error without a reconnecting function. So stability of the hard disk can be improved. CONSTITUTION: When an error on SCSI bus happens, there are four ways to reconnect. First, process the reconnection during an execution of the SCSI command with the error. To do so, use a semaphore way for a unity of an occupation of disk resources, mutual exclusion between tasks, and data security. Second, guarantee the processing of the reconnection after finishing an I/O job. Third, control a script data pointer followed by a reconnection counter. A controlling part of the script data pointer maintains or changes the data pointer according to the reconnection count. Last, check an exception of the reconnection. The reconnection process is managed for all errors, but handled exceptionally if not reconnected even though the SCSI bus is reset. |
