Granted Patent B2

US 8,812,901 · App. 13/241,711 · Granted Aug 19, 2014

Inventor: Robert L. Sheffield, Jr. (Longmont, CO)

Assignee: LSI Corporation

Abstract

Claims (66)

1. A method operable within a storage device, the storage device adapted to couple with a Redundant Array of Independent Drives (RAID) storage controller, the method comprising:

detecting a failure of the storage device;

automatically reconfiguring the storage device into a recovery mode of operation in response to sensing the failure;

responsive to entering the recovery mode, performing the steps of:

receiving a write request to store supplied data at a write logical block address (LBA);

marking the write LBA as storing stale data in response to receipt of the write request;

receiving a read request requesting return of identified data stored at a read LBA;

determining whether the read LBA is marked as storing stale data responsive to receipt of the read request;

completing the read request by returning the identified data stored at the read LBA in response to a determination that the read LBA is not storing stale data; and

failing the read request in response to a determination that the read LBA is marked as storing stale data.

2. The method of claim 1 further comprising:

completing the write request by returning a failure status for the write request without storing the supplied data at the write LBA.

3. The method of claim 1

wherein the step of failing the read request further comprises returning a status code indicating that the identified data is stale.

4. The method of claim 1

wherein the step of marking further comprises writing uncorrectable data at the write LBA, and

wherein the step of determining further comprises:

attempting to read the identified data stored at the read LBA; and

detecting the presence of uncorrectable data stored at the read LBA as indicative of the read LBA being marked as storing stale data.

5. The method of claim 4

wherein the step of writing uncorrectable data further comprises:

completing the write request as though it is a SCSI WRITE_LONG operation having WR_UNCOR and COR_DIS parameters bits set.

6. The method of claim 1

wherein the step of marking further comprises adding information identifying the write LBA to a log memory of the storage device,

wherein the step of determining further comprises:

detecting that the read LBA matches information previously added to the log memory indicative that the read LBA stores stale data.

7. The method of claim 6

wherein the log memory is a non-volatile memory.

8. Apparatus in a storage device, the apparatus comprising:

a failure detector adapted to sense failure of the storage device and adapted to reconfigure the storage device into a recovery mode of operation;

a recovery mode write request processor adapted to process a write request received by the storage device while the storage device is in recovery mode, the write request identifying a write logical block address (LBA) to be written with supplied data, the recovery write request processor adapted to mark the write LBA as storing stale data in response to receipt of the write request; and

a recovery mode read request processor adapted to process a read request received by the storage device while the storage device is in recovery mode, the read request identifying a read LBA to be read, the recovery read processor adapted to determine whether the read LBA is marked as storing stale data, the recovery read processor further adapted to complete the read request by returning requested data in response to a determination that the read LBA is not marked as storing stale data, the recovery read request processor further adapted to complete the read request with a status indicating stale data in response to a determination that the read LBA is marked as storing stale data.

9. The apparatus of claim 8

wherein the recovery mode write processor is further adapted to mark the write LBA as storing stale data by writing uncorrectable data at the write LBA, and

wherein the recovery mode read processor is further adapted determine that the read LBA is marked as storing stale data by attempting to read the data stored at the read LBA and detecting the presence of uncorrectable data stored at the read LBA in response to the attempted read of the data.

10. The apparatus of claim 9

wherein the recovery mode write request processor is further adapted to write uncorrectable data by processing the write request as though it is a SCSI WRITE_LONG operation having WR_UNCOR and COR_DIS parameter bits set.

11. The apparatus of claim 8 further comprising:

a log memory coupled with the recovery mode write request processor and coupled with the recovery mode read request processor,

wherein the recovery mode write request processor is further adapted to mark the write LBA as storing stale data by adding information identifying the write LBA to the log memory, and

wherein the recovery mode read request processor is further adapted to determine that the read LBA is marked as storing stale data by detecting that the read LBA matches information previously added in the log memory indicative that the read LBA stores stale data.

12. The apparatus of claim 11

wherein the log memory is a non-volatile memory device.

13. A Redundant Array of Independent Drives (RAID) storage system comprising:

a plurality of storage devices including a first device and a second device, wherein the first device is configured as part of a RAID logical volume comprising multiple storage devices of the plurality of storage devices; and

a RAID storage controller coupled with the plurality of storage devices and coupled to receive read requests and write requests directed to access the RAID logical volume from host systems,

wherein the first device is adapted to sense its failure and to generate a failure signal indicative of the failure,

wherein the first device is further adapted to configure the first device into a recovery mode of operation responsive to sensing the failure,

wherein the RAID storage controller is adapted to receive the failure signal from the first device indicating possible failure,

wherein the RAID storage controller is further adapted to initiate a sequence of read requests to the first device to copy data from the first device to the second device in response to receipt of the failure signal while continuing to process read requests and write requests from the host systems,

wherein the first device further comprises:

a recovery mode write request processor adapted to process a write request received from the RAID storage controller while the first device is in recovery mode, the write request identifying a write logical block address (LBA) to be written, the recovery write request processor adapted to mark the write LBA as storing stale data in response to receipt of the write request; and

a recovery mode read request processor adapted to process a read request received from the RAID storage controller while the first device is in recovery mode, the read request identifying a read LBA to be read, the recovery read processor adapted to determine whether the read LBA is marked as storing stale data, the recovery read processor further adapted to complete the read request by returning requested data to the RAID storage controller in response to a determination that the read LBA does not store stale data, the recovery read request processor further adapted to complete the read request by returning to the RAID storage controller a stale data status indicating stale data in response to a determination that the read LBA is marked as storing stale data, and

wherein the RAID storage controller is further adapted to rebuild data stored at the write LBA on the second device using data from corresponding blocks of other storage devices of the RAID logical volume in response to receipt of the stale data status.

14. The storage system of claim 13

wherein the recovery mode write processor is further adapted to mark the write LBA as storing stale data by writing uncorrectable data at the write LBA, and

wherein the recovery mode read processor is further adapted to determine that the read LBA is marked as storing stale data by attempting to read the data stored at the read LBA and detecting the presence of uncorrectable data stored at the read LBA in response to the attempted read of the data.

15. The storage system of claim 14

wherein the recovery mode write request processor is further adapted to write uncorrectable data by processing the write request as though it is a SCSI WRITE_LONG operation having WR_UNCOR and COR_DIS parameters bits set.

16. The storage system of claim 13

wherein the first device further comprises:

a log memory coupled with the recovery mode write request processor and coupled with the recovery mode read request processor,

wherein the recovery mode write request processor is further adapted to mark the write LBA as storing stale data by adding information identifying the write LBA to the log memory, and

wherein the recovery mode read request processor is further adapted to determine that the read LBA stores stale data by detecting that the read LBA matches information previously added in the log memory indicative that the read LBA stores stale data.

17. The storage system of claim 16

wherein the log memory is a non-volatile memory device within the first device.

Assignments (9)

Continuity (1)