Reconstructive error recovery procedure (ERP) using reserved buffer

• US 9,590,660 B2
• Filed: 05/22/2015
• Issued: 03/07/2017
• Est. Priority Date: 11/14/2012
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Katagiri, Takashi, Nylander-Hill, Pamela R.
• Primary Examiner(s): Decady, Albert
• Assistant Examiner(s): KABIR, ENAMUL MD
• Application Number:
  US14/720,703
• Publication Number:
  US 20150256205A1
• Time in Patent Office: 655 Days
• US Class Current: 1/1
• CPC Class Codes: G11B 20/1833 by adding special lists or symbols to the coded information G11B20/1806, G11B20/1866 take precedence
• CPC Class Codes: G11B 2020/1853 using a product code which has inner and outer parity symbols
• CPC Class Codes: H03M 13/2906 using block codes H03M13/2957 takes precedence
• CPC Class Codes: H03M 13/2909 Product codes
• CPC Class Codes: H03M 13/293 with erasure setting
• CPC Class Codes: H03M 13/3715 Adaptation to the number of estimated errors or to the channel state

Abstract

According to one embodiment, a method for reading data includes reading a data set repeatedly using different settings until either: a reconstructed data set is sent to a host and/or stored, or a maximum number of rereads has been reached, after each reading of the data set, storing each row to a reserved data buffer that has no errors or errors in the row are correctable using C1-error correction code (ECC) unless a matching row already exists in the reserved data buffer that has fewer corrected errors therein, assembling the data set from the rows stored in the reserved data buffer to form an assembled data set, correcting any remaining errors in the assembled data set using C2-ECC to form the reconstructed data set, and sending the reconstructed data set to the host and/or storing the reconstructed data set.

Claims

1. A system for reading data, the system comprising:

a processor and logic that is integrated with the processor, executable by the processor, or integrated with and executable by the processor, the logic causing the processor to;

read a data set one or more times, the data set comprising a plurality of portions which combine together to wholly form the data set, wherein during each read of the data set, settings are changed in an attempt to better read the data set;

store all portions of the data set to a reserved data buffer read during a first reading of the data set;

determine whether the stored data set is correctable using at least one of C1-error correction code (ECC) and C2-ECC, wherein the stored data set is recovered in response to the data set being correctable using at least one of C1-ECC and C2-ECC;

recover the stored data set in response to a determination that the data set is correctable using at least one of C1-ECC and C2-ECC;

reread portions assumed to be in the data set in response to a determination that the data set is not correctable using at least one of C1-ECC and C2-ECC;

perform a validity check to determine whether a currently read portion is in the data set;

determine that the currently read portion of the data set is correctable using C1-ECC using a syndrome check in response to a determination that the validity check confirms that the portion is from the data set;

store one or more of the portions of the data set to the reserved data buffer after subsequent readings of the data set in response to a determination that the one or more portions are correctable using C1-ECC;

aggregate the one or more portions of the data set in the reserved data buffer to form an assembled data set, wherein each of the one or more portions are correctable using C1-ECC;

determine whether C2-ECC is capable of correcting all errors in the assembled data set;

correct any remaining errors in the assembled data set using C2-ECC to form a reconstructed data set and send the reconstructed data set to a host in response to a determination that C2-ECC is capable of correcting all errors in the assembled data set; and

reread the data set using a different setting in an error recovery procedure (ERP) when an error in the assembled data set is not correctable using C2-ECC.

2. The system as recited in claim 1, wherein the plurality of portions of the data set are C1-encoded rows in an array representing the data set, and wherein the logic configured to determine that the currently read portion of the data set is correctable using C1-ECC using the syndrome check is further configured to:

determine whether a current row read from the data set has been updated after the first reading of the data set; and

determine whether a data set number stored to a header of the current row matches a data set number assigned to the data set.

3. The system as recited in claim 1, wherein when a portion already exists in the reserved data buffer, the existing portion is overwritten when a corresponding portion from the one or more portions that are correctable using C1-ECC has fewer corrected errors therein than the existing portion.

4. The system as recited in claim 1, wherein when the data set is reread, existing portions in the reserved data buffer are overwritten with portions having fewer errors corrected therein, and wherein the logic is further configured to determine whether C2-ECC is capable of correcting all errors in the assembled data set until either the reconstructed data set is sent to the host, or a number of allowable retries is reached and the data set is reported to the host as unreadable.

5. The system as recited in claim 1, wherein the data set is read from a magnetic tape, and wherein at least a portion of the data set is re-read using different settings until all errors in the assembled data set are correctable using C2-ECC or a maximum number of rereads has been reached.

6. The system as recited in claim 5, wherein different settings comprise a change to at least one of:

a tape speed during reading, a data flow correction power, channel parameters, and servo tracking.

7. A method for reading data, the method comprising:

reading a data set one or more times, the data set comprising a plurality of portions which combine together to wholly form the data set, wherein during each read of the data set, settings are changed in an attempt to better read the data set;

storing all portions of the data set to a reserved data buffer read during a first reading of the data set;

determining whether the stored data set is correctable using at least one of C1-error correction code (ECC) and C2-ECC, wherein the stored data set is recovered in response to the data set being correctable using at least one of C1-ECC and C2-ECC;

recovering the stored data set in response to a determination that the data set is correctable using at least one of C1-ECC and C2-ECC;

rereading portions assumed to be in the data set in response to a determination that the data set is not correctable using at least one of C1-ECC and C2-ECC;

performing a validity check to determine whether a currently read portion is in the data set;

determining that the currently read portion of the data set is correctable using C1-ECC using a syndrome check in response to a determination that the validity check confirms that the portion is from the data set;

storing one or more of the portions of the data set to the reserved data buffer after subsequent readings of the data set in response to a determination that the one or more portions are correctable using C1-ECC;

aggregating the one or more portions of the data set in the reserved data buffer to form an assembled data set, wherein each of the one or more portions are correctable using C1-ECC;

determining whether C2-ECC is capable of correcting all errors in the assembled data set;

correcting any remaining errors in the assembled data set using C2-ECC to form a reconstructed data set and sending the reconstructed data set to a host in response to a determination that C2-ECC is capable of correcting all errors in the assembled data set; and

rereading the data set using a different setting in an error recovery procedure (ERP) in response to a determination that an error in the assembled data set is not correctable using C2-ECC.

8. The method as recited in claim 7, wherein the plurality of portions of the data set are C1-encoded rows in an array representing the data set.

9. The method as recited in claim 7, wherein a portion that already exists in the reserved data buffer is overwritten in response to a determination that a corresponding portion from the one or more portions that are correctable using C1-ECC has fewer corrected errors therein than the portion that already exists in the reserved data buffer.

10. The method as recited in claim 7, wherein when the data set is reread, existing portions in the reserved data buffer are overwritten with portions having fewer errors corrected therein, and further comprising determining whether C2-ECC is capable of correcting all errors in the assembled data set until either the reconstructed data set is sent to the host, or a number of allowable retries is reached and the data set is reported to the host as unreadable.

11. The method as recited in claim 7, wherein the data set is read from a magnetic tape, and wherein at least a portion of the data set is re-read using different settings until all errors in the assembled data set are correctable using C2-ECC or a maximum number of rereads has been reached.

12. The method as recited in claim 11, wherein different settings comprise a change to at least one of:

a tape speed during reading, a data flow correction power, channel parameters, and servo tracking.