Storage error management
AssigneeFusion-Io, Inc.
InventorsEvan Orme, James G. Peterson
Life of this patent
Estimating the term…
Abstract
A storage controller identifies a storage division comprising unreadable data. In response, the storage controller relocates readable data (if any) from the identified storage division. The storage controller may be further configured to associate the unreadable data with a logical address, and to record persistent metadata to indicate that the logical address is mapped to unreadable, corrupt data. The storage controller may clear the corruption indicator from the logical address. The logical address may be cleared in response to determining that the unreadable data is invalid and/or available from another source.
Claims20
A method, comprising:
- identifying a logical address that is associated with unreadable data on a storage unit within an erase division of a storage medium;
- writing persistent metadata to the storage medium to record that the identified logical address is associated with the unreadable data;
- and
- relocating readable data from storage units within the erase division in response to identifying the logical address associated with the unreadable data.
An apparatus, comprising:
- a translation layer that maps logical addresses of a logical address space to storage addresses of a non-volatile storage medium;
- a reliability monitor that detects a read error pertaining to a storage address, wherein the storage address is mapped to a particular logical address by the translation layer;
- and
- a corruption manager configured to record persistent metadata to flag corrupt logical addresses, wherein the corruption manager determines whether to flag the particular logical address in response to detecting the read error.
A system, comprising:
- a storage medium comprising a plurality of storage divisions;
- a storage controller that detects a read failure pertaining to data stored at a storage address within a particular one of the storage divisions;
- and
- a corruption manager that appends an entry to a storage log on the storage medium that applies a corrupt data designation to the logical identifier, wherein the storage controller is configured to return an error in response to requests to read logical identifiers having a corrupt data designation.
Dependencies are inferred from the claim text.