Error detection/correction and fault detection/recovery – Pulse or data error handling – Replacement of memory spare location – portion – or segment
Reexamination Certificate
1998-11-05
2002-08-27
Wong, Peter (Department: 2181)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Replacement of memory spare location, portion, or segment
C714S718000, C714S708000
Reexamination Certificate
active
06442715
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to computer storage devices, and more particularly to methods, circuitry and software for managing defects in computer storage media, such as computer disk drive media.
2. Description of the Related Art
Disk drive media virtually always contains defective areas, where it is impossible to reliably record and read back data. Rather than take a manufacturing yield loss, systems have been created to avoid the recording of data on those defective areas. Typically, a given size media defect becomes more apparent and affects more data as data is required to be recorded more densely. Therefore, defect management schemes have been an essential facilitating component for the increase in disk drive capacity that has been occurring for many years.
Over the years, this function of avoiding defective areas has migrated from the host computer and operating system to the disk drive itself. This migration has been a logical extension of the integration of more functions on the disk drive circuitry. As a result, operating systems no longer have intimate information of such items as the geometry of the drive, the block ordering, skewing, or the defect management techniques employed. Removing this intelligence from the operating system has allowed disk drive vendors to optimize these items for the specific product. Thus, these items are now commonly implemented by a combination of firmware and hardware on the disk drive.
For background understanding, the physical address of a given block of data on a disk drive is composed of three components: cylinder, head, and sector. The cylinder represents the radius of the data, the head represents the disk surface of the data, and the sector represents the rotational position of the data. Collectively, this type of address is often referred to as a “CHS” address. The combination of cylinder and head is often referred to as a “track” address. When the host computer desires to transfer one or more data blocks, it specifies the address of the first block and a block count parameter to the disk drive. Most interfaces allow the data address to take the form of a CHS address, however, this is primarily for historical reasons. A CHS address passed by the host computer implies that the computer has knowledge of the geometry of the drive, the layout of the blocks (e.g., after transferring track X, it will know which track is next), the skewing, and the locations of the defects. As mentioned above, because these functions are now commonly embedded in the disk drive, the disk drive firmware translates the passed CHS address into an actual CHS address before accessing the data. For clarity, the host side CHS address is generally referred to as a “logical CHS” address, and the actual CHS address is referred to as a “physical CHS” address.
Modem disk drive interfaces allow the host to specify a logical block address (LBA). Some interfaces allow either LBA or logical CHS addressing, others allow LBA addressing only. An LBA is simply a block number that allows the host computer to access the data without any real or perceived information on the geometry, skew, layout, or of the location of media defects. Often, a logical block address is a byproduct of the logical CHS address conversion, and therefore it is generally more efficient to simply pass a logical block address from the host computer to the disk drive, if the interface allows either. However, most modem disk drives implement defect management during the address translation process, regardless of whether the address specified by the host computer was a logical CHS or LBA address.
Two major techniques are commonly used to avoid media defects: block slipping and block relocation. It should be noted here that other techniques have also been used over the years, however, these two techniques in various forms are used exclusively in virtually all modem disk drives. Furthermore, virtually most all modem disk drives commonly employ both techniques.
Block slipping is simply the avoidance of defective blocks by jumping over the defective block in the address translation process. Consider the following LBA to physical sector relationship shown in table A.
TABLE A
LBA:
0
1
2
3
4
5
6
7
Sector:
0
1
2
3
4
5
6
7
8
Physical sector 8 is reserved for use if a defective location is discovered, and is not normally accessible by the host computer. If physical sector 4 is determined to be defective, the address translation process can use this information to change the LBA to physical sector relationship and avoid the defect as shown in table B.
TABLE B
LBA:
0
1
2
3
x
4
5
6
7
Sector:
0
1
2
3
4
5
6
7
8
In this example, physical sector 4 is not addressable for host transfers. The address translation process avoids the defective sector while maintaining the correlation between the logical block address and the sequential nature of the data being stored. Therefore, sector 8 will now be addressable as LBA 7. The major advantage to block slipping is that it incurs minimal performance loss on multiple block transfers. In the above example, if all 8 logical blocks were being transferred, the actual transfer time is increased by only 12.5% due to the effect of the defect. However, the actual percentage of the time penalty associated with the defect is generally far less than this, because total access time also includes seek and rotational latency delays, which are unaffected by the defect.
Block slipping does have a major disadvantage. If a physical sector is determined to be defective after customer data has been placed on the drive, slipping the block requires user data to be moved. How much data must be moved is a function of the location of the defect versus the location of the pool. However, as is well known, moving user data is very problematic. For example, power loss during the process could result in corrupted data. Also, moving data can be very time consuming, which can cause host timeout issues. Therefore, because the movement of user data is risky and slow, block slipping is generally used only for defects found in the factory, or if used for defects found in the field, reserved sector pools are placed frequently enough (every track or every cylinder or some other limited range) such that the risk and time associated with the data movement are minimized.
The other defect management technique mentioned above is block relocation. Block relocation is the avoidance of defective blocks by jumping instead to another, out of sequence block, and then returning to resume the transfer. Reference is again made to the LBA to sector relationship of table A, where the physical sector 8 is reserved for use if a defective location is discovered, and is not normally accessible by the host computer. If physical sector 4 is determined to be defective, the address translation process can use this information to change the LBA to physical sector relationship and avoid the defect as shown in table C.
TABLE C
LBA:
0
1
2
3
x
5
6
7
4
Sector:
0
1
2
3
4
5
6
7
8
In this example, the address translation process avoids the defective sector, but does not maintain the correlation between the logical block address and the sequential nature of the data being stored. Accordingly, LBA 4 is out of order. Block relocation therefore relies on one or more pools of unused sectors, into which the logical block address accesses can be redirected as needed to avoid defective sectors. Disk drive vendors commonly place these pools at the end of each track, each cylinder, each zone, or one large pool at the end of the volume. Often, these pools are also common with block slipping pools. For example, consider the case where one sector per track is reserved. If a track has two defects, one of the defects may be slipped, thus consuming the spare sector, and the other defect may be relocated to a nearby track that has no defects.
Another common implementation that shares one reserved pool for slipped and relocated blocks consumes the pool from the earliest address upward for slipped blocks and f
Jorgenson Lisa K.
Slater Steven H.
STMicroelectrics N.V.
Vo Tim
Wong Peter
LandOfFree
Look-ahead reallocation disk drive defect management does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Look-ahead reallocation disk drive defect management, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Look-ahead reallocation disk drive defect management will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2888730