JP6148996B2 - Apparatus and method for partitioned read-modify-write operations - Google Patents

Description

  The present disclosure relates to data storage devices, and more particularly to data storage memories that use read-modify-write (RMW) operations, such as shingled magnetic recording (SMR). .

  In one embodiment, the apparatus may include a data storage medium that includes a band of tracks having a plurality of data tracks arranged in a tiled manner with at least one track partially overlapping adjacent tracks and the controller. The controller reads the first read-modify for the first partition so as to virtually divide the band of tracks into at least a first partition and a second partition, each partition including at least one track. It can be configured to perform a write-write operation and to perform a second read-modify-write-write operation for the second partition.

  In another embodiment, the method divides the roof band of the tracks of the data storage medium into at least a first partition and a second partition, each partition including at least one track; Performing a first read-modify-write-write operation for the second and performing a second read-modify-write-write operation for the second partition.

  In yet another embodiment, an apparatus can include a computer readable data storage medium that stores instructions that cause the processor to perform the method when the instructions are executed by the processor. The method divides the roof band of the tracks of the data storage medium into at least a first partition and a second partition, each partition including at least one track, and a first read for the first partition Performing a modify-write-write operation and performing a second read-modify-write-write operation for the second partition.

FIG. 1 is a diagram of an exemplary embodiment of a system designed for partitioned read-modify-write operations. FIG. 2 is a diagram of an exemplary embodiment of a shingled record for use with segmented read-modify-write operations. FIG. 3 is a diagram of another exemplary embodiment of a shingled record for use with segmented read-modify-write operations. FIG. 4 is a diagram of another exemplary embodiment of a system with shingled records for use with segmented read-modify-write operations. FIG. 5 is a flowchart of an exemplary embodiment of a method for partitioned read-modify-write operations. FIG. 6 is another flowchart of an exemplary embodiment of a method of partitioned read-modify-write operations. FIG. 7 is a diagram of another exemplary embodiment of a shingled record for use with partitioned read-modify-write operations. FIG. 8 is a diagram of another exemplary embodiment of a shingled record for use with segmented read-modify-write operations.

  In the following detailed description of the embodiments, reference is made to the accompanying drawings that form a part hereof, and which are shown by way of illustration of specific embodiments in the specification. It is to be understood that other embodiments can be utilized and structural changes can be made without departing from the scope of the present disclosure.

  FIG. 1 depicts an embodiment of a system for partitioned read-modify-write operations, generally designated 100. The system 100 can include a host 102 and a data storage device (DSD) 104. Host 102 may also be referred to as a host system or host computer. The host 102 can be a desktop computer, laptop computer, server, tablet computer, phone, music player, another electronic device, or any combination thereof. Similarly, the DSD 104 can be any of the devices listed above, or any other device that can be used to store or retrieve data, such as a hard disk drive (HDD) or a hybrid disk drive. Host 102 and DSD 104 may be connected by an interface, such as a wired or wireless connection, or by a local area network (LAN) or a wide area network (WAN). In some embodiments, the DSD 104 may be a stand-alone device that is not directly connected to the host 102, or both the host 102 and the DSD 104 may be part of a single unit.

  The DSD 104 can include one or more non-volatile memories. In the depicted embodiment, the DSD 104 can include a rotatable disk memory 106. In other embodiments, the DSD 104 may include additional memory or memory types, including volatile and non-volatile solid state memory. For example, the DSD 104 may be a hybrid HDD that includes both the disk memory 106 and the nonvolatile solid-state memory 108.

  In an embodiment of the system 100, the disk memory 106 may have one or more zones configured to store data on roof tile data tracks using roof magnetic recording (SMR). SMR is a recording system for increasing the data recording density on a disc, whereby at least one track of data partially overlaps adjacent data tracks. SMR will be described in more detail with respect to FIGS.

  SMR is a method of performing a write operation in one radial direction at the disk width, where the tracks partially overlap each other similar to a roof tile. Referring to FIG. 2, when it is assumed that writing is performed in the direction indicated by the arrow in the tile-writing method, when writing is performed on the track N, the adjacent track N-1 is partially Overlap. When writing is performed on track N + 1, adjacent tracks N partially overlap. In contrast to the recording method in which each track is written without the intended overlap by the recording method, SMR results in an increase in track / inch (TPI) characteristics as the recording density in the radial direction of the storage medium. Can do.

  Furthermore, SMR can generate a flow in one direction. Therefore, the constraint that N-1 tracks cannot be written after N tracks have been written should be satisfied. As illustrated in FIG. 3, when the track N-1 is written in the reverse direction of the tile recording direction after writing on the track N, the track N is overlapped or adjacent to track interference (ATI). Due to this, reading may become impossible. Therefore, changing the data recorded on track N-1 after recording track N or the data recorded on track N after recording track N + 1 is simply a non-tile track that can be overwritten at any time. Requires a different writing strategy.

  Turning now to FIG. 4, a diagram of another exemplary embodiment of a system using shingled records for segmented read-modify-write operations is shown. The rotating disk medium 402 may be divided into a plurality of zones (eg, zone 1, zone 2, etc.), and each zone may contain a plurality of data tracks.

  Due to the single writing direction of the SMR, writing a given track N-1 after track N has been written will result in all tile tracks following track N-1 (ie, track N, track N + 1, Track N + 2 etc.) needs to be rewritten. To achieve this realistically, a set of tracks can be grouped into “bands” where the band ends with a guard track. In some embodiments, the guard track can be a non-tile track or a tile track that is not used to record data. If track N-1 needs to be written, track N-1 to the guard track can be rewritten, while tracks in other bands cannot be affected. Accordingly, the track in each zone can be divided into a plurality of bands. As depicted in the example embodiment of FIG. 4, zone 1 can encompass band 0 (B_0) through band k (B_k), while zone 2 includes band k + 1 (B_k + 1), band k + 2 ( B_k + 2) and the like. In an example embodiment, each zone may contain 100 data tracks, and the 100 data tracks may be divided into 10 bands each containing 10 tracks. Each track may be further divided into data sectors associated with a plurality of logical block addresses (LBA), and data may be stored in each LBA.

  In an embodiment, a DSD using SMR may receive a write command to overwrite one or more LBAs corresponding to the data sectors of the roof track in the band. The data to be written can come from the host device, the DSD media cache, or another location. In order to execute a write command, it may be necessary to read multiple tracks of the band, modify the read data based on the write command, and then write the modified data back into the band. This may be referred to as a read-modify-write operation (RMW), or more specifically for a SMR device, a banded read-write operation (BRO). Data read from a track can be temporarily stored in volatile memory, such as a buffer or register, where the data is modified before being written back to the band.

  For example, if the write command includes an LBA associated with the data sector of a track in band 1, the data can be read from band 1 into volatile memory and modified according to the write command. If DSD begins to write modified data back to band 1 and an unexpected power failure occurs, some data may be lost beyond repair. For example, if track N was partially written when a power failure occurred, track N + 1 may not be readable due to adjacent track interference. Since the contents of track N + 1 in volatile memory can be lost as well, there would be no way to recover the data for track N + 1.

  One solution to avoid such problems may be to perform read-modify-write-write (RMWW) instead of RMW operations. Referring to FIG. 5, a flowchart of an exemplary embodiment of a read-modify-write-write operation is shown, designated generally at 500. The method may include, at 502, reading data from band 1 located on the main storage device and storing it in a buffer. The main storage device can be a tiled disk type memory, a non-volatile solid state memory that requires RMW operation, or any other type of memory that can use RMW operation. The buffer can be a volatile solid-state memory such as DRAM or a non-volatile solid-state memory such as flash memory. Once the data is in the buffer, at 504, the data can be modified or merged with new or other data. Data manipulations can be performed including, but not limited to, writing new data, modifying existing data, cleaning a band, or any combination thereof.

  Once any modification is complete, the modified data may be written 506 to a non-volatile memory area available from the buffer. In certain embodiments, this may be a scratchpad area on the disk, a different non-volatile memory such as flash memory, or any combination thereof. Further, as described below, multiple scratch pads can be used, can be of different sizes, and may or may not be on the same storage medium.

  The modified data can also be written back to the main storage device at 508 for the same band from which the original data was read, or it can be written to a different band. Performing RMWW operations adds a layer of redundancy to help protect against data loss. For example, if power loss occurs during writing to the scratchpad, the original data should still be safe in band 1. If power loss occurs during writing to band 1, the modified data should be secured within the scratchpad.

  The RMWW method may rely on a buffer for storing data so that the data is transmitted from the main storage medium to the scratchpad. The problem can occur when the buffer is not large enough to hold all of the data in the band to be transmitted. These problems can occur because changes to any piece of data on a band can require reading or writing to the entire band. For example, when a piece of 2 kilobytes on the 100 megabyte band is changed, the entire 100 megabyte may need to be read into or written from the buffer. The band size may need to be reduced, or the buffer size may need to be increased to avoid buffer capacity issues.

  A solution to the buffer sizing problem may be to divide the band into the read sections of the selected track for the purpose of performing RMWW operations. Each partition can contain multiple tracks, and the track fragments can be fixed in size or dynamic and can be specified with increasing LBA order across the band. When data manipulation needs to be performed, only the partition, not the entire band, may need to be modified, saving time and reducing the buffer size.

  The disadvantage of dividing the band into sections is that there is no physical guard track in the band. Without a guard track, damage to track N can occur when track N-1 is modified. To overcome this problem, a logical guard track can be implemented in place of the physical guard track at the end of the partition, allowing control and flexibility when writing a layer of tile data. The logical guard track can be at least one track in width and can be moved into a band area accessible to the user. The logical guard track can be floating, i.e. it can move as a segmentation progression. In addition, one or more logical guard tracks can be used in the band.

  The method of partitioning the read-modify-write operation is when data in a partition (section 1) including N tracks and starting with track A can be read from the main storage to the buffer from track A to N. Can work. In various embodiments, track A may be the first physically located track in the band, or the RMW operation is for such a portion from the end of the band rather than what needs to be rewritten. So it can be done only in a different track in the band. For example, if all merges are in the last three tracks of the band, their writing should not interfere with previously occurring tracks (ie, the last track overlaps the next of the last track) If the direction of overlap occurs), only those three tracks may need to be rewritten. However, if the data is to be changed in the first track of the band, all tracks in that band should need to be rewritten in the RMW operation.

  After the selected data is in the buffer, data manipulation can be performed to modify the data, such as merging any new data with the existing data. Once the data operation is complete, the modified data including tracks A through N can be written to the first scratch pad (spA). The modified tracks A through N-1, now in spA, can then be written to the respective bands in the main store. Writing to track N-1 can corrupt track N, so track N is a logical on the main store because the data from track N is in the buffer and can now be stored in spA. It can be considered a guard track. In some situations, when a read operation is performed during this process, the data in the scratchpad is most recently updated, such as when the data on the main storage medium, logical guard track, track N can be corrupted. Such a read can be directed to the respective scratchpad, as it can be modified or simply a valid copy.

  Further, the second partition may be selected for tracks that occur next that have not yet been modified as part of the current RMWW operation, and may include tracks N + 1 through M in the given example. . Data from the second partition may be read from the main store into a buffer and modified, and the buffer may also still contain track N. The second segment can then be written to spB, and track N can also be available in the buffer, resulting in spB encompassing tracks N through M. After the second partition is stored in spB, the modified tracks N through M-1 can then be written to the main storage. At this point, tracks A through M-1 on the main storage may contain valid data, and M may be a logical guard track. This process can be repeated multiple times until all segments in the band have undergone an RMWW operation, thus resulting in completion of the RMWW operation for the entire band, because the data capacity of the band performs the RMWW operation. May be useful when it is larger than the size of the associated buffer used.

  Referring to FIG. 6, a flowchart of an exemplary embodiment of a method for partitioning a read-modify-write operation is shown, generally designated 600. In some examples, the method 600 allows the system to perform an RMW operation for data stored in a main store that has a data size that is larger than can be stored in an associated buffer. Can make it possible. An example of such a system is shown in the above figures for a disk having a tile track band. The method 600 allows each band that can contain more data than can be stored in the associated buffer space to be divided into sections that contain one or more tracks, and While 600 can perform read-modify-write-write (RMWW) operations for each partition, it can implement a logical guard band.

  In some embodiments, a band may be divided into multiple virtual partitions for the purpose of performing RMWW operations on all bands. In method 600, at 602, the first segment may be read and may include the first track, track X through track Y, which may include N tracks. The amount of data allocated in tracks X through Y, including X and Y, should not exceed the size of the associated buffer. Furthermore, the amount of data allocated in tracks X through Y, including X and Y, should not exceed the size of the first scratch pad (spA). Data manipulation, such as modifying data from the selected track with any new data received from the host, may be performed on the data in the buffer at 604.

  At 606, a check can be made to determine if track Y is a physical guard track that can also be considered the last track in the band, in which case no logical guard track can be required. . In some of the examples described herein, the last track of a band can be assumed to be a physical guard track, but one can define the last track of a band as a non-guard track. It is not possible to include a physical guard track as part of a defined band, in which case slight variations to the examples given here can be achieved as well.

  Continuing, if Y is not a physical guard track, at 608, data on tracks X through Y can be stored from the buffer into spA. If Y is a physical guard track, at 610, data from track X to Y-1 may be stored in spA. This is because the data on the physical guard track, if any, may be unreadable or unusable, so the physical guard track will store the address mapped storage location. Cannot be included. In future iterations of the process, the selected data can be stored in spB.

  Once the selected data is stored in spA or spB, tracks X through Y-1 can be stored at 612 in the main storage device. In a roof storage system, writing to Y-1 can corrupt data on track Y, resulting in track Y being corrupted, thus making a copy of track Y's data spA or spB (or so) Holding on to both) can create a logical guard track while partitioning bands while it is desired. Failure to write data from track Y to spA or spB (or another location) before writing data to track Y-1 in the corresponding band may result in loss of data on track Y.

  Once the current partition is written to the main data storage device, the method 600 may determine if there are further partitions for the selected band that have not yet undergone a partitioned RMWW operation. If all of the sections corresponding to the selected band have been written, and there are no further sections of the selected band to write, i.e. all bands or applicable sections of bands have undergone an RMWW operation, Processing may end at 616. If another segment of the track is needed, at 618, X can be incremented by Zx and Y can be incremented by Zy. In some embodiments, Zx can be equal to Y−X + 1 and Zy can be greater than or equal to Zx. If Zx is equal to YX + 1, the newly started track in the next segment should be one higher than the logical guard track, and the data from the previous logical guard track, Y-1, is Can be in a buffer, or previously stored in spA or spB.

  For example, a segment containing six tracks may begin with track 100 and end with track 105 having a logical guard track on track 105. This may result in Zx = 6, thus moving the first track of the next segment, track X, from track 100 to track 106. If Zy = Zx, the last track in the segment, track Y, is also shifted by 6 tracks and moves from track 105 to track 111. Since the partition size will remain the same throughout each iteration of the process, it may be desirable to set Zx = Zy. However, in some embodiments, it may be advantageous to increase the track size piecewise during some or all of the processing iterations, in which case a value greater than or less than Zx. May need to be set to Zy. For example, consider the case where a segment containing six tracks starts at track 100 and ends at track 105 and is incremented as follows. That is, X = X + Zx and Y = Y + Zy, where Zx = Y−X + 1 and Zy = Y−X + 2. Under such an incrementing scheme, the first track in the segment can be moved from track 100 to track 106, and the last track in the segment can be moved from track 105 to track 112, resulting in The section contains seven tracks instead of six. Therefore, Zx and Zy may change the value for each iteration based on system choice, which may correspond to maximum efficiency in the buffer or scratchpad area.

  After X and Y are incremented by Zx and Zy, respectively, it may be necessary to determine when track Y is a physical guard track if a logical guard track cannot be required. The process can be repeated until a physical guard track is reached or until the system completes the RMWW operation for all of the selected bands. Therefore, the scratch pads spA and spB can alternate throughout the process, regardless of how many sections are present. There is no rule to limit the number of scratch pads to two, but more scratch pads can be used if necessary.

  FIG. 7 is an exemplary embodiment of a method for partitioning read-modify-write operations. An exemplary process may begin at track t_begin and end at t_last. For purposes of this example, t_last is the last track or physical guard track of the selected band. If the band is to undergo an RMW operation, the data can be read at 702 from the track t_begin to t_N into a buffer where data operations can be performed. The modified data corresponding to tracks t_begin through t_N may also be stored at 702 on the first scratch pad (spA). This ensures at 706 that when track t_N-1 is overwritten, a copy of the data for track N can be corrupted on the main store, but stored in the scratchpad. The data in track N can now be considered as a logical guard track. Next, tracks t_N + 1 through t_R can be read from the main storage to the buffer at 708, modified, and stored in spB. The data can then be written to the main storage, which can result in tracks t_begin through t_R-1 being stored in the main data storage medium as part of the current RMW operation, hence , Track t_R becomes a logical guard track.

  The process can be easily extended to accommodate different sized bands. In this embodiment, processing continues to the end of the band. At 714, data from tracks t_last-x + 1 to t_last-1 is read into the buffer and modified. Since t_last-x should also be still in the buffer, tracks t_last-x through t_last-1 can be stored in the scratchpad spLAST. The scratch pad spLAST can be spA, spB, or a completely different scratch pad. Tracks t_last-x through t_last-1 can then be written into corresponding tracks t_last-x through t_last-1 on the main data storage medium. When processing reaches the last segment of the band, a physical guard track may be present, so a logical guard track may not be necessary. In the example provided, the last track t_last may be a physical guard band and need not be stored on any scratchpad.

  Referring to FIG. 8, a graphical representation of a method for partitioning read-modify-write operations is shown, generally designated 800. In the particular example shown, the system may write two sections of data, each section consisting of up to six tracks. However, additional partitions or different partition sizes can be written as needed. In the example shown, tracks in the first section may include tracks 100-105, while tracks in the second section may include tracks 106-111.

  The method 800 may include two passes that indicate the processing that can be performed during two iterations or a read-modify-write operation where each pass is partitioned. In the first pass (path A) 802, tracks 100 through 105 may be read from the main data storage device to the buffer at 806. Method 800 may perform modifications, merging, or other data manipulation while the data is in the buffer. Next, at 808, data corresponding to tracks 100-105 may be stored in spA. Then, at 812, tracks 100-104 can be written to the main data storage device. Track 105 can then be a logical guard track, since a corresponding copy of the data can be in the buffer, and the copy can be stored in spA, while the corresponding track 105 on the main data storage device is Can be damaged due to the writing of the track 104.

  Moving the second path (path B) 804, the second segment may include track 106 to track 111 and may be read from the main data storage device at 814. Data corresponding to tracks 106 (or 105, because it should still be in the buffer) through 111 can be stored in 818 at 818. Then, at 820, tracks 105 through 110 can be written to corresponding tracks in the main data storage device.

  Recall that valid data from the track 105 from the main data storage device can still be in the buffer and still in the spA. Therefore, at 820, when data tracks 105 through 110 are written to the corresponding tracks on the main data storage device, track 105 will be overwritten with valid data, indicating that it is a logical guard track. The track 111 is then made a logical guard track.

  In some embodiments, there may be multiple paths, such as a third path (path C) (not shown), a fourth path (path D) (not shown), for example. The scratch pad can be used one or more times during processing. For example, spA can be used in path A, path C, and the like. However, the size of spA need not be the same for each path and can depend on the partition size. If a partition in path C includes 8 tracks, the size of spA in path A may be different from path C.

  According to various embodiments, the methods described herein may be implemented as one or more software programs running on a computing device, such as a computer processor or controller. According to another embodiment, the methods described herein may be implemented as one or more software programs running on a computing device such as a personal computer using a disk drive. Dedicated hardware implementations, including but not limited to application specific integrated circuits, programmable logic arrays, and other hardware devices are similarly implemented to implement the methods described herein. Can be built. Further, the methods and processes described herein may be implemented as a computer-readable data storage medium or device that includes instructions that cause the processor to perform the methods and processes when the instructions are executed.

  The examples of embodiments described herein are intended to give a general understanding of the structure of the various embodiments. The examples are not intended to serve as a complete description of all of the elements and features of apparatus and systems that utilize the structures or methods described herein. Many other embodiments will be apparent to those of skill in the art upon reviewing the disclosure. For example, the RMW operation partition described herein may be useful for data storage devices or storage media that use RMW operations or comparable operations, such as, for example, a program-erase cycle for solid state memory. Other embodiments may be utilized and derived from the disclosure so that structural and logical substitutions and changes may be made without departing from the scope of the disclosure. Moreover, while specific embodiments have been illustrated and described herein, any subsequent arrangement designed to achieve the same or similar purpose can be substituted for the specific embodiments shown. Should be understood.

  This disclosure is intended to cover any and all subsequent adaptations or variations of various embodiments. Combinations of the above embodiments, and other embodiments not specifically described herein, will be apparent to those of skill in the art upon reviewing the description. Further, the examples are merely representative and may not be drawn to scale. Certain parts within the scope of the examples may be exaggerated while other parts may be reduced. Accordingly, the disclosure and drawings are to be regarded as illustrative rather than restrictive.

Claims (20)

A data storage medium comprising a band of tracks having a plurality of data tracks arranged in a tiled manner with at least one track partially overlapping adjacent tracks;
A controller,
So as to virtually divide the band of the tracks into at least a first section and a second section, each section including at least one track;
Configured to perform a first read-modify-write-write operation for the first partition, and to perform a second read-modify-write-write operation for the second partition A device comprising a controller. The controller,
So as to read the first section into the buffer;
So as to modify the data of the first partition in the buffer;
Writing the first section to a first scratchpad, and physically adjacent tracks corresponding to the first section on the data storage medium with less than all of the first sections The apparatus of claim 1, comprising the controller further configured to write. The controller,
So as to read the second section into the buffer;
So as to modify the data of the second partition in the buffer;
Writing the second section including at least the last track of the first section to a second scratchpad, and on a physically adjacent track corresponding to the second section on the data storage medium The apparatus of claim 1 or 2, further comprising the controller, further configured to write less than all of the second section. 3. The apparatus of claim 2 , wherein the portion less than all of the first section is the first section minus the last adjacent track of the first section. 4. The apparatus of claim 3 , wherein the portion less than all of the second segment is the second segment minus the last adjacent track of the second segment. The controller,
Repeat the read-modify-write-write operation for any additional partition until all tracks in the band have undergone a corresponding read-modify-write-write operation, and within the segment To stop performing read-modify-write-write operations for the band if the last track is a physical guard band that separates the band from other tracks not in the band, The apparatus of claim 1, further comprising the controller. A first scratchpad;
A second scratchpad;
Writing the first segment to the first scratchpad during the first read-modify-write-write operation, and the second segment during the second read-modify-write-write operation. The apparatus of claim 1, further comprising the controller further configured to write two sections to the second scratchpad.   Alternately writing to the first scratchpad and writing to the second scratchpad for further read-modify-write-write operations for additional segmentation of the band. The apparatus of claim 7, further comprising the controller configured. The first section includes a first range of physically adjacent tracks of the band;
The second section includes a second range of physically adjacent tracks of the band;
9. The apparatus of any one of the preceding claims, further comprising a first track of the second section being adjacent to a last track of the first section. Dividing the roof band of the tracks of the data storage medium into at least a first section and a second section, each section including at least one track;
Performing a first read-modify-write-write operation for the first partition;
Performing a second read-modify-write-write operation for the second partition.   The method of claim 10, further comprising a plurality of roof tile bands of tracks having a plurality of overlapping data tracks arranged such that at least one track partially overlaps an adjacent track.   12. A method according to claim 10 or 11, further comprising implementing a logical guard band between the first read-modify-write-write operation and the second read-modify-write-write operation. . Performing a read-modify-write-write (RMWW) operation for any additional segment until all tracks in the roof band have undergone a corresponding read-modify-write-write operation;
If the last track in the segment is a physical guard band that separates the roof tile band from other tracks not in the band, further stopping performing an RMWW operation for the band; The method of claim 10.   14. The method of claim 13, further comprising implementing a logical guard band that floats between sections corresponding to two consecutive read-modify-write-write operations.   Implementing a floating logical guard band and moving the floating logical guard band between the last segment processed through the RMWW operation and the next partition to be processed through the RMWW operation 14. The method of claim 13, further comprising: Writing the first segment to a first scratch pad during the first read-modify-write-write operation;
16. The method of any one of claims 10-15, further comprising writing the second section to a second scratchpad during the second read-modify-write-write operation.   Alternate writing to the first and second scratchpads for alternate iterations of additional read-modify-write-write operations corresponding to additional sections of the roof band The method of claim 16 further comprising: When the instruction is executed by the processor,
Dividing the roof band of the tracks of the data storage medium into at least a first section and a second section, each section including at least one track;
Performing a first read-modify-write-write operation for the first partition;
An apparatus comprising a computer readable data storage medium storing the instructions, causing the processor to perform a method comprising performing a second read-modify-write-write operation for the second partition. . When the instruction is executed by the processor,
After the second read-modify-write-write operation is completed, a logical guard band that is not a physical guard band is transferred to the first read-modify-write-write operation and the second read-modify-write. 19. The apparatus of claim 18, comprising a computer readable data storage medium storing the instructions that causes the processor to perform the method further comprising implementing during a write operation. When the instruction is executed by the processor,
Writing the first segment to a first scratch pad during the first read-modify-write-write operation;
Storing the instructions for causing the processor to perform the method further comprising: writing the second partition to a second scratchpad during the second read-modify-write-write operation. The apparatus of claim 18, comprising a computer readable data storage medium.