摘要 |
PURPOSE: To provide the method and device for controlling memory with which a write error generated at the memory of a distributed computer is remotely and automatically corrected. CONSTITUTION: The write data of (j) bits and its object address fetched from an NCP (network control unit) are inputted through a system bus 30 to a memory device 33. A control circuit 336 reads out respective data corresponding to the object address from a memory data area 331 and its redundant bit area 332 of (k) bits (k>j) and those data are checked by an error check circuit 333. When any error is detected, the control circuit 336 reports it to the NCP, stops write processing, receives a mode switching signal from the NCP, sets the gate circuit of a switching means 338 and inhibits the output of an error signal. Further, corrected data are fetched from the NCP into a data embedding circuit 335 and the data of object (k) bits are reloaded. Afterwards, the mode is switched to an access mode and the write processing is started again. |