Detecting and correcting hard errors in a memory array

• US 9,189,326 B2
• Filed: 10/08/2013
• Issued: 11/17/2015
• Est. Priority Date: 10/08/2013
• Status: Active Grant

• Current Assignee: Advanced Micro Devices, Inc.
• Original Assignee: Advanced Micro Devices, Inc.
• Inventors: Kalamatianos, John, John, Johnsy Kanjirapallil, Gelinas, Robert, Sridharan, Vilas K., Nevius, Phillip E.
• Primary Examiner(s): Alphonse, Fritz
• Application Number:
  US14/048,830
• Publication Number:
  US 20150100848A1
• Time in Patent Office: 770 Days
• Field of Search: 714/763, 714/752, 714/773, 714/710, 714/766, 714/764
• US Class Current: 1/1
• CPC Class Codes: G06F 11/1008 in individual solid state devices G06F11/1004 takes precedence
• CPC Class Codes: G11C 11/41 forming static cells with positive feedback, i.e. cells not needing refreshing or charge regeneration, e.g. bistable multivibrator or Schmitt trigger
• CPC Class Codes: G11C 2029/0409 Online test
• 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
• CPC Class Codes: G11C 29/78 using programmable devices

Abstract

Hard errors in the memory array can be detected and corrected in real-time using reusable entries in an error status buffer. Data may be rewritten to a portion of a memory array and a register in response to a first error in data read from the portion of the memory array. The rewritten data may then be written from the register to an entry of an error status buffer in response to the rewritten data read from the register differing from the rewritten data read from the portion of the memory array.

Claims

1. A method comprising:

rewriting data to a portion of a memory array and to a register in response to a first error in data read from the portion of the memory array;

reading the rewritten data from the portion of the memory array and from the register; and

writing the rewritten data from the register to an entry of an error status buffer in response to the rewritten data read from the register differing from the rewritten data read from the portion of the memory array.

2. The method of claim 1, further comprising:

detecting the first error in the data read from the portion of the memory array using at least one of a parity bit or an error correcting code.

3. The method of claim 1, wherein each row of the memory array comprises a plurality of subsets of memory elements, the method further comprising:

modifying an error status associated with the entry of the error status buffer to indicate that the error is a hard error that persisted after rewriting the data to the portion of the memory array; and

storing a fault index that indicates a row of the memory array that includes the portion and a fault group index that indicates a subset of the row of the memory array that includes the portion.

4. The method of claim 3, wherein writing the rewritten data to the entry of the error status buffer comprises writing the rewritten data corresponding to the subset of the row of the memory array to the entry of the error status buffer.

5. The method of claim 4, further comprising:

multiplexing data read from the entry of the error status buffer into data read from the row of the memory array using the fault group index as a selection signal for a plurality of multiplexers associated with the plurality of subsets of the row of the memory array.

6. The method of claim 5, further comprising:

bypassing the multiplexing of data read from the entry of the error status buffer into data read from the row of the memory array concurrently with comparing the rewritten data read from the register to the rewritten data read from the portion of the memory array to determine whether the rewritten data read from the register differs from the rewritten data read from the portion of the memory array.

7. The method of claim 1, further comprising:

restoring at least one of the error status buffer and an array status register following power gating of a power domain that includes the memory array using at least one of stored values and recomputed values of said at least one of the error status buffer and the array status register.

8. The method of claim 1, further comprising:

writing new data to the memory and the register in response to an address of a write request hitting the entry in the error status buffer and then comparing the new data written to the memory with the corresponding new data written to the register; and

filling the entry in the error status buffer with the new data written to the register in response to the new data written to the memory being different than the new data written to the register or invalidating the entry in the error status buffer in response to the new data written to the memory being the same as the new data written to the register.

9. The method of claim 1, further comprising:

detecting at least one second error in data read from the portion of the memory array, rewriting data to the portion of the memory array and the register in response to detecting said at least one second error, and invalidating the entry of the error status buffer in response to data read from the portion of the memory array being the same as data read from the register.

10. An integrated circuit comprising:

a memory array;

a register;

an error status buffer; and

repair logic to rewrite data to a portion of the memory array and the register in response to a first error in data read from the portion of the memory array, and wherein the rewritten data is written from the register to an entry of the error status buffer in response to the rewritten data read from the register differing from the rewritten data read from the portion of the memory array.

11. The integrated circuit of claim 10, wherein the portion of the memory array is associated with at least one of a parity bit or an error correcting code, and wherein the first error in the data read from the portion of the memory array is detected using at least one of the parity bit or the error correcting code.

12. The integrated circuit of claim 10, wherein the entry of the error status buffer comprises information indicating an error status that indicates whether the error is a hard error, and wherein the memory array comprises a plurality of rows, and wherein each row of the plurality of rows comprises a plurality of subsets of memory elements, and wherein the entry in the error status buffer comprises a fault index that indicates a row of the memory array that includes the portion and a fault group index that indicates a subset of the row of the memory array that includes the portion.

13. The integrated circuit of claim 12, wherein the rewritten data corresponding to the subset of the row of the memory array is written to the entry of the error status buffer in response to the rewritten data read from the register differing from the rewritten data read from the portion of the memory array.

14. The integrated circuit of claim 13, further comprising:

a plurality of multiplexers associated with the plurality of subsets of the row of the memory array, and wherein the multiplexes multiplex data read from the entry of the error status buffer into data read from the row of the memory array using the fault group index as a selection signal.

15. The integrated circuit of claim 14, further comprising:

a comparator to compare the rewritten data read from the register to the rewritten data read from the portion of the memory array to determine whether the rewritten data read from the register differs from the rewritten data read from the portion of the memory array, and wherein multiplexing data read from the entry of the error status buffer into data read from the row of the memory array is bypassed concurrently with comparing the rewritten data read from the register to the rewritten data read from the portion of the memory array.

16. The integrated circuit of claim 10, wherein the repair logic is to restore at least one of the error status buffer and an array status register following power gating of a power domain that includes the memory array using at least one of stored values and recomputed values of said at least one of the error status buffer and the array status register.

17. The integrated circuit of claim 10, wherein data is rewritten to the portion of the memory array and the register in response to detecting at least one second error in data read from the portion of the memory array, and wherein the entry of the error status buffer is invalidated in response to data read from the portion of the memory array being the same as data read from the register.

18. A non-transitory computer readable medium embodying a set of executable instructions, the set of executable instructions to manipulate at least one processor to:

rewrite data to a portion of a memory array and to a register in response to a first error in data read from the portion of the memory array;

read the rewritten data from the portion of the memory array and from the register; and

write the rewritten data from the register to an entry of an error status buffer in response to the rewritten data read from the register differing from the rewritten data read from the portion of the memory array.

19. The non-transitory computer readable medium set forth in claim 18, comprising a set of executable instructions to manipulate said at least one processor to modify an error status associated with the entry of the error status buffer to indicate that the error is a hard error that persisted after rewriting the data to the portion of the memory array and store a fault index that indicates a row of the memory array that includes the portion and a fault group index that indicates a subset of the row of the memory array that includes the portion.

20. The non-transitory computer readable medium set forth in claim 18, comprising a set of executable instructions to manipulate said at least one processor to detect at least one second error in data read from the portion of the memory array, rewrite data to the portion of the memory array and the register in response to detecting said at least one second error, and invalidate the entry of the error status buffer in response to data read from the portion of the memory array being the same as data read from the register.