Low-overhead mechanism to detect address faults in ECC-protected memories

• US 10,319,461 B2
• Filed: 06/29/2016
• Issued: 06/11/2019
• Est. Priority Date: 06/29/2016
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Kwon, Kon-Woo, Kozhikkottu, Vivek, Somasekhar, Dinesh
• Primary Examiner(s): Tu, Christine T.
• Application Number:
  US15/197,590
• Publication Number:
  US 20180004597A1
• Time in Patent Office: 1,077 Days
• Field of Search: 714768, 714767, 714799, 714800, 714805, 714807, 714764, 714763, 714758, 36523006, 36523001, 711200
• CPC Class Codes: G06F 11/1048 using arrangements adapted for a specific error detection or correction feature
• CPC Class Codes: G06F 11/1068 in sector programmable memories, e.g. flash disk G06F11/1072 takes precedence
• CPC Class Codes: G11C 29/024 in decoders
• CPC Class Codes: G11C 29/42 using error correcting codes [ECC] or parity check
• CPC Class Codes: G11C 29/52 Protection of memory contents; Detection of errors in memory contents

Abstract

Embodiments are generally directed to a low-overhead mechanism to detect address faults in ECC-protected memories. An embodiment of an apparatus includes a memory array; an error correction code (ECC) encoder for the memory array to encode ECC values based on a data value and a respective address value for the data value; and an ECC decoder for the memory array to decode ECC values that are based on data values and respective addresses for the data values; wherein the apparatus is to detect and correct an error in an address value based on an ECC value, address value, and data value stored in the memory.

Claims

1. An apparatus comprising:

a memory array;

an error correction code (ECC) encoder circuit for the memory array to encode, during a write operation of a data value into the memory array, an ECC value based on the data value and a write address value for the data value;

an ECC decoder circuit for the memory array to receive, during a read operation of the data value, a read ECC value and a read data value from the memory array and to receive a read address value for the data value;

wherein the apparatus is to detect and correct an error in the write or read address values based on the read ECC value, the read address value, and the read data value;

wherein the apparatus is to roll back stored data values of all or a portion of the memory array to an earlier state upon detecting the error in the read or write address values.

2. The apparatus of claim 1, further comprising a first bus for address values, a second bus for data values, and a third bus for ECC values, the first, second and third busses to be coupled to the memory array, wherein the ECC encoder circuit is to receive address values for the first bus and data values for the second bus.

3. The apparatus of claim 1, wherein the ECC decoder circuit is to generate an error location vector to indicate an address bit position of each detected error.

4. The apparatus of claim 3, wherein the apparatus is to generate an address fault signal based on a value of the error location vector.

5. The apparatus of claim 1, further comprising an address scrubber to read data values and their corresponding ECC values from the memory array at a plurality of read address values and to detect errors in the data values'"'"' corresponding read or write address values based on the read address values, the data values and their corresponding ECC values.

6. The apparatus of claim 1, wherein the apparatus is to perform memory diagnostics in response to one or more detected errors in read or write address values.

7. The apparatus of claim 1, wherein a size of the ECC value for the data value and the write address value is a same size as another ECC value for the data value without the write address value.

8. The apparatus of claim 1, wherein the apparatus is a memory module.

9. A method comprising:

receiving a command to read from a memory array at an address value;

reading a data value and an error correction code (ECC) value from the memory array based on the address value, wherein, during a previous write of the data value, the ECC value was generated based on the data value and its write address value;

performing a decoding operation using the address value, the data value and the ECC value;

identifying one or more errors in the address value or write address value based on the decoding operation; and

,

rolling back stored data values of all or a portion of the memory array to an earlier state upon detecting the one or more errors.

10. The method of claim 9, wherein the identifying one or more errors includes generating an error location vector to indicate a location of each detected error in the address or write address value.

11. The method of claim 10, further comprising generating an address fault signal based on a value of the error location vector.

12. The method of claim 9, wherein the command to read the data value is part of a scrubbing operation for the memory array.

13. The method of claim 12, wherein the scrubbing operation includes reading values stored at each address of the memory array.

14. The method of claim 9, further comprising performing memory diagnostics in response to the identifying of one or more errors in the address value or write address value.

15. A computing system comprising:

one or more processors for processing of data;

a memory apparatus for storage of data, the memory apparatus including a memory array;

a wireless transmitter or receiver and one or more antennae for transmission or reception of data; and

,

the computing system further comprising error correction code (ECC) logic, including;

an ECC encoder to encode an ECC value based on a data value and a respective write address value for the writing of the data value into the memory array,

an ECC decoder to receive a read ECC value and read data value from the memory array and to also receive a read address value;

wherein the ECC logic is to detect and correct an error in the read or write address values based on the read ECC value, the read address value, and the read data value;

wherein the apparatus is to roll back stored data values of all or a portion of the memory array to an earlier state upon detecting the error in the read or write address values.

16. The system of claim 15, further comprising a first bus for address values and a second bus for data values, the first and second busses to be coupled to the memory array, wherein the ECC encoder circuit is to receive address values for the first bus and data values for the second bus.

17. The system of claim 15, further comprising an address scrubber to read data values from a plurality of addresses of the memory array and to detect errors in the addresses based on ECC values, the addresses, and the data values.

18. One or more non-transitory computer-readable storage mediums having stored thereon data representing sequences of instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

receiving a command to read from a memory array at an address value;

reading a data value and an error correction code (ECC) value from the memory array based on the address value, wherein, during a previous write of the data value, the ECC value was generated based on the data value and its write address value;

performing a decoding operation using the address value, the data value and the ECC value; and

,

identifying one or more errors in the address value or write address value based on the decoding operation; and

,

rolling back stored data values of all or a portion of the memory array to an earlier state upon detecting the one or more errors.

19. The storage mediums of claim 18, further comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:

reading data values from a plurality of addresses of the memory array and detecting errors in the addresses based on ECC values, the addresses, and the data values.