| 摘要 |
An apparatus is disclosed which detects the existence of an error, in a computer system, corrects the error, and takes steps to ensure that the error will never again re-occur. The error resides in the integrity of data stored in a memory. When the data is read from memory and found to be erroneous, the data is corrected and stored partly in a spare portion of the memory; however, the identity of the data part stored in the spare portion of the memory is recorded in a separate memory of much smaller size. During a subsequent read of the data from the memory, the separate memory is simultaneously consulted to determine the identity of the data part stored in the spare portion of the memory. The corrected data part is retrieved from the spare portion of the memory at once during a first read of said memory.
<??>When data is read from memory (10), it enters an error correction matrix (12). The error correction matrix determines the existence of a single bit error or a double bit error. The error correction matrix of capable of correcting a single bit error, but it is not capable of correcting a double bit error. If a double bit error is detected, the data is transmitted to a processor (22) which, with the aid of a stored program (18), determines the specific bits in error associated with the double bit error. The specific bits are corrected by the processor. A reference table (16), known as a bit steering array, is updated to record the identity of one of the bits in error associated with the double bit error. The bit steering array is consulted (14), and the data is stored into memory (10) once again. Thereby, the one of the bits originally in error is stored in a spare memory chip, the remaining bits, including the other erroneous bit, are stored in their original locations in memory. The double bit error has been reduced to a single bit error. |
