| 摘要 |
PURPOSE:To simplify the error processing by executing a report immediately, when a fault by which a sort processing cannot be continued is detected by a sort processor, and reporting the contents of an error to a host mechanism after the sort processing is completed, in the case of a fault by which the sort processing can be continued. CONSTITUTION:When some sort processor detects a fault by which a sort processing cannot be continued, the sort processing by sort processors 12 - 14 of all stages are interrupted immediately, and the contents of an error at that time are reported to a host mechanism 15. Also, when a fault by which the sort processing can be continued is detected, the processing of the sort processors 12 - 14 of all stages is stopped after the sort processing is completed, the contents of an error are reported to the host mechanism 15, and after an error analysis is executed, the sort processing is reexecuted. In such a way, even if a kind and a quality of the error increase in proportion to the number of stages of the sort processes 12 - 14, the error processing is executed simply. Also, when a fault exists in a data bus, it is prevented that a bus parity check does not work in the course of collecting incorrect fault information. |
