# US 2001/0025359 A1 · Error detection and correction method in a computer system and main memory controller of the same

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2001/0025359 A1
- Document ID: US20010025359A1
- Status: Published application
- Kind: A1
- Application number: 09/805,169
- Filed: 2001-03-14
- Published: 2001-09-27
- Earliest claimed: 2000-03-14
- Assignees: Hitachi, Ltd.
- Inventors: Tsuyoshi Tanaka
- Web page: https://patentgrep.com/patent/US20010025359A1

## CPC

- G06F 11/10: Adding special bits or symbols to the coded information, e.g. parity check, casting out 9's or 11's
- H03M 13/15: Cyclic codes, i.e. cyclic shifts of codewords produce other codewords, e.g. codes defined by a generator polynomial, Bose-Chaudhuri-Hocquenghem [BCH] codes (H03M13/17 takes precedence)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H03M: CODING; DECODING; CODE CONVERSION IN GENERAL (using fluidic means F15C4/00; optical analogue/digital converters G02F7/00; coding, decoding or code conversion, specially adapted for particular applications, see the relevant subclasses, e.g. G01D, G01R, G06F, G06T, G09G, G10L, G11B, G11C, H04B, H04L, H04M, H04N; ciphering or deciphering for cryptography or other purposes involving the need for secrecy G09C)

## Abstract

According to an error detection and correction method to be implemented in a computer system, when an error is detected in data to be written in a memory, fault information is appended to the data without an increase in the number of bits constituting the data, and the resultant data is stored in the memory. An error control code represented by a SEC-DEC code is adopted for encoding and decoding. Data is encoded into a shortened code. At this time, specific bit positions associated with column vectors deleted from a parity check matrix defined in the error control code are allocated to fault information. Thus, a word to be actually stored in the memory is composed of check bits produced from data to be written and fault information, and information bits constituting the data to be written. Decoding is performed on the assumption that the fault information represents 0s. When data having fault information appended thereto is decoded, the fault information is reproduced through error detection and correction.

## Claims

### Claim 1

An error detection and correction method for one of a memory and a communication line employing a c-bit error correcting/d-bit error detecting code (d>c), said error detection and correction method comprising steps of:
producing check bits according to fault information and object data, the fault information showing one of presence and absence of fault detected during encoding of the object data to be transferred to one of the memory and the communication line, and showing cause of the fault;
encoding the object data and the check bits according to the c-bit error correcting/d-bit error detecting code to produce an encoded word;
transferring the encoded word to one of the memory and the communication line; and
detecting and correcting an error in data received from one of the memory and the communication line according to a bit pattern of a syndrome produced from the received data, and thus reproducing the fault information.

### Claim 2 (depends on 1)

An error correction and detection method according to claim 1, wherein an encoded word produced at said encoding step is a shortened code assigning a specific bit positions which is associated with a column vector deleted from a parity check matrix defined by a c-bit error correcting/d-bit error detecting code, to a bit for the fault information, and, thereby, the encoded word is composed of the object data and the check bits.

### Claim 3

A main memory controller connected to a processor through a processor bus, connected to an I/O unit through an I/O bus, and designed to control writing or reading of a main memory, said main memory controller comprising:
a first error detection circuit for detecting an error in data on said processor bus;
a second error detection circuit for detecting an error in data on said I/O bus;
an encoder circuit for producing check bits using object data, which is transferred over one of said processor bus and I/O bus and written in said main memory, and fault information which, when one of said first and second error detection circuit detects an uncorrectable error in the object data to be written, enables identification of the detection circuit, and for producing a word composed of the object data to be written and the check bits as a word to be written in said main memory;
a decoder circuit for producing a syndrome for data read from said main memory, and detecting and correcting an error according to the bit pattern of the syndrome; and
a fault information detection table for detecting whether the syndrome produced by said decoder circuit exhibits a specific bit pattern so as to identify the source of an uncorrectable error having occurred before encoding is performed by said encoder circuit.

### Claim 4 (depends on 3)

A main memory controller according to claim 3, wherein said main memory controller is connected to another main memory controller via a crossbar switch, and further comprises a third error detection circuit for detecting an error in data transferred via said crossbar switch; and said third error detection circuit is included in the detection circuits to be identified based on the fault information.

### Claim 5 (depends on 3)

A main memory controller according to claim 3, wherein as a coding system implemented in said encoder circuit, a SEC-DED code is adopted to produce a shortened code; specific bit positions among a plurality of bit positions which are associated with column vectors deleted from a parity check matrix defined in the SEC-DED code are allocated to the fault information indicating that an uncorrectable error has been detected in data transferred over one of said processor bus and I/O bus; said decoder circuit decodes received data on the assumption that the fault information represents bits of 0s; when data transferred over one of said processor bus and I/O bus and struck with an error is decoded, a one-bit error indicated with the fault information is detected and corrected; and when a corrected word having a bit of 1 contained in fault information thereof is struck with another one-bit error, the error is detected.

### Claim 6 (depends on 3)

A main memory controller according to claim 3, wherein as a coding system implemented in said encoder circuit, a SbEC-DbED code according to which an error is controlled in units of a b-bit byte (b>2) is adopted to produce a shortened word; specific bit positions among a plurality of bit positions which are associated with column vectors deleted from a parity check matrix defined in the SbEC-DbED code are allocated to fault information indicating that an uncorrectable error has been detected in data transferred over one of said processor bus and I/O bus; said decoder circuit decodes data on the assumption that the fault information represents bits of 0s; when data transferred over one of said processor bus and I/O bus and struck with an error is decoded, a one-byte error indicated with the fault information is detected and corrected; and when the corrected word having a bit of 1 contained in fault information thereof is struck with another one-byte error, the error is detected.

### Claim 7 (depends on 3)

A main memory controller according to claim 3, further comprising a register in which information of an error detected by one of said decoder circuit and error-stricken data is held as a log, and a monitor used to indicate the contents of the register.

### Claim 8

A computer-usable storage medium in which a program having the following steps is stored:
producing check bits according to one of presence and absence of an error sensed during encoding of object data to be transferred to a memory or over a communication line, fault information indicating cause of the error, and the object data;
encoding the object data and the check bits according to a c-bit error correcting/d-bit error detecting code (where d>c) to produce a word;
transferring the word to one of the memory and the communication line; and
detecting and correcting an error in data received from one of the memory and the communication line according to the bit pattern of a syndrome produced from the received data, and thus reproducing the fault information.
