Error detection/correction and fault detection/recovery – Pulse or data error handling – Memory testing
Reexamination Certificate
1998-11-24
2001-09-25
Decady, Albert (Department: 2784)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Memory testing
C711S162000
Reexamination Certificate
active
06295619
ABSTRACT:
FIELD OF THE INVENTION
The present invention pertains to the field of computer storage systems. More particularly, the present invention pertains to a method for error management in a solid state disk drive.
BACKGROUND OF THE INVENTION
The nonvolatility of magnetic disk drives (“hard” drives) is one of the primary reasons for their use as mass storage devices in computer systems. They may be turned on and off without loss of data. Hard drives include one or more circular platters. Data is stored on concentric tracks of magnetic media on the platters. Each track is further divided into sectors. A sector of data is typically 512 bytes in size.
A hard drive controller typically receives sector numbers from a host processor (e.g., central processing unit or CPU). These sector numbers are then translated to a physical location for accessing the hard drive to obtain a sector of data. A given sector number generally always corresponds to the same physical location on the hard drive. In other words a unique one-to-one mapping exists between sector numbers and physical locations within the hard drive.
A typical user file stored on a hard drive occupies many sectors distributed within the disk drive. These sectors can be distributed across multiple tracks and platters. In addition, the sectors used by the user file may not be contiguous sectors within a track.
Hard drives suffer from a number of disadvantages including sensitivity to vibration, dust, humidity, and sudden changes in velocity. Hard drives also consume relatively large amounts of power, which makes them less attractive for computers that are battery powered.
Solid state disks provide an alternative to magnetic hard disk drives. Solid state disks (also “solid state drives”) use semiconductor memories, also referred to as solid state memories, as a storage medium. Semiconductor memories are comparatively more rugged than hard drives and offer the advantage of being much less sensitive to vibration, dust, humidity, and sudden changes in velocity. Semiconductor memories also tend to require less power than a typical hard drive with similar storage capacity.
One type of nonvolatile semiconductor memory is the FLASH electrically erasable programmable read only memory (FLASH memory). FLASH memories can be programmed by the user and once programmed retain their data until erased. FLASH memories are erased by erasing a block of memory at a time. Erasing a FLASH memory tends to require considerable more time than reading or programming operations.
Some prior art solid state disks use FLASH memory. Unlike the hard drive, a sector number provided by the host CPU may not correspond to the same physical address for accessing a location within the solid state drive due to the limitations in erasing and programming a FLASH memory device. In one embodiment, sectors of data are written to available locations instead of a unique location within the FLASH memory in order to defer an erase operation. Thus the sector number becomes a “logical” sector number as opposed to a physical sector number identified with a unique address. This technique requires version control and additional overhead including translation and index tables for locating the physical address within the FLASH memory that corresponds to a logical sector number. Without the translation and index tables the data may not be retrievable even though it is still stored in the nonvolatile memory.
Although the solid state memories are relatively immune to environmental factors that negatively affect hard drives, solid state memories have their own shortcomings. Unlike a magnetic hard drive, semiconductor memories such as FLASH memory suffer charge loss over time. The charge on the floating gate of a FLASH cell may dissipate due to, for example, leakage. This charge loss may cause a change in the interpreted state of the FLASH cell which in turn results in data loss. Typically charge loss is not due to a permanent defect and although the data is lost, the cell is not permanently damaged and can be used for future storage. Thus the charge loss problem can be termed a temporary defect.
Another disadvantage of semiconductor memories is failure due to column shorts. Memory arrays are arranged in rows and columns. Typically the rows are associated with “words” of memory and the columns correspond to bitlines which represent “bits”. If one or more bitlines are electrically shorted together, possibly during the manufacturing process, the result is a column short. A column short permanently affects all shorted bitlines and can essentially render useless the sensing of every bit associated with a shorted bitline for every word in at least one block of the array.
One disadvantage of using logical sector numbers is that if the overhead data such as the translation or index tables become corrupted such as from charge loss or column shorts, the data may not be retrievable because it cannot be located.
Thus improving the error management for both column shorts and charge loss within a solid state disk drive is desirable.
SUMMARY OF THE INVENTION
One object of the present invention is to improve the correctability of memory errors due to charge loss.
A primary logical identifier is used to locate at least one partition of data stored in the memory after receiving a requested partition number identifying a requested partition. A secondary logical identifier is used to compute a value to further identify the at least one partition of data. The value is computed by performing a logical AND operation between the primary logical identifier and the secondary logical identifier. This value is compared with the requested partition number.
Another object is to locate memory errors due to column shorts. A method of detecting and locating bit errors due to a column short in a memory is described. First a written stripe of data is written to the memory, each bit of the written stripe having a known value. The written stripe of data is compared to the read stripe of data. The bit errors are located by locating a position of each inverted known value in the read stripe of data.
Another object is to improve the correctability of memory errors due to a column short using the location of the bits affected by the column short. A method of correcting column short errors for partition data read from a memory is described. A mask is generated to locate one or more bit positions within a word of memory affected by shorted columns. Every word of the partition data is masked to store a first value in the bit positions affected by shorted columns. The masked partition data is validated. If the validation fails, every word of the partition data is masked to store a second value in the bit positions affected by shorted columns.
REFERENCES:
patent: 4876684 (1989-10-01), Guntheroth
patent: 5014273 (1991-05-01), Gagliardo et al.
patent: 5070474 (1991-12-01), Tuma et al.
patent: 5131089 (1992-07-01), Cole
patent: 5200959 (1993-04-01), Gross et al.
patent: 5394362 (1995-02-01), Banks
patent: 5438573 (1995-08-01), Mangan et al.
patent: 5479633 (1995-12-01), Wells et al.
patent: 5535369 (1996-07-01), Wells et al.
patent: 5581723 (1996-12-01), Hasbun et al.
patent: 5586285 (1996-12-01), Hasbun et al.
patent: 5603001 (1997-02-01), Sukegawa et al.
patent: 5640529 (1997-06-01), Hasbun
patent: 5737742 (1998-04-01), Achiwa et al.
patent: 5835955 (1998-11-01), Dornier et al.
patent: 6112017 (2000-08-01), Wise
Hasbun Robert N.
Leemann Daniel H.
Blakely , Sokoloff, Taylor & Zafman LLP
De'cady Albert
Intel Corporation
Lamarre Guy
LandOfFree
Method and apparatus for error management in a solid state... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for error management in a solid state..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for error management in a solid state... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2535485