Data storage device aligning partition to boundary of sector when partition offset correlates with offset of write commands

Abstract

A data storage device is disclosed comprising a non-volatile memory comprising a plurality of sectors. A partition map is evaluated that identifies a partition accessed through a plurality of logical block addresses (LBAs), where each LBA maps to a fraction of a sector. A partition offset is determined for the partition relative to a boundary of one of the sectors. N write commands are received each having a write offset relative to a corresponding sector. When the write offset for X of the N write commands matches the partition offset, at least part of the partition is moved to align at least part of the partition to a boundary of one of the sectors.

Description

BACKGROUND

FIG. 1A shows a data storage device 2 (e.g., a disk drive or a solid state drive) comprising control circuitry 4 and a non-volatile memory 6, such as a disk or a flash memory. The non-volatile memory 6 comprises a plurality of memory segments referred to as sectors. To facilitate defect mapping, the sectors are accessed indirectly through logical block addresses (LBAs). In this manner, if a sector degrades, the system/user data can be relocated to a spare sector and the corresponding LBA(s) remapped to the spare sector.

The LBAs of a data storage device may be divided into a number of partitions, wherein each partition stores a file system (e.g., a n-bit FAT file system, or a NT File System) identifying system files and user files. A partition may be bootable meaning that it stores a bootable operating system (OS) such as Windows OS or Mac OS. When a computer system is initially powered on, a master boot record (MBR) is read from the data storage device (typically stored in the first logical data sector). The MBR identifies a partition table that stores a partition map for each partition of the data storage device as illustrated in FIG. 1B. The partition table typically includes up to four primary partition maps identifying up to four primary partitions, wherein a primary partition may be further subdivided into extended partitions as illustrated in FIG. 1C. Each partition map (primary and extended) identifies a starting LBA and ending LBA of the corresponding partition, a file system type field, and a flag indicating whether the partition is bootable as shown in FIG. 1D. When the computer system finds a bootable partition, it uses the file system type field to access the partition's file system in order to read and execute the OS files from the partition (i.e., boot the computer system).

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A shows a prior art data storage device comprising control circuitry and a non-volatile memory comprising a plurality of partitions.

FIG. 1B shows a prior art partition table comprising a number of partition maps.

FIG. 1C shows a prior art partition subdivided into a number of extended partitions.

FIG. 1D shows a prior art partition map identifying a starting LBA and ending LBA for the partition, a file system type field, and a flag indicating whether the partition is bootable.

FIG. 2A shows a data storage device comprising a non-volatile memory and control circuitry according to an embodiment of the present invention.

FIG. 2B shows a flow diagram according to an embodiment of the present invention wherein at least part of a partition is aligned with a sector boundary when a write offset matches a partition offset a predetermined number of times over N write commands.

FIGS. 3A and 3B illustrate an embodiment of the present invention wherein at least part of a partition is aligned by shifting a starting LBA of the partition.

FIGS. 4A and 4B illustrate an embodiment of the present invention wherein at least part of a partition is aligned by remapping at least one LBA from a first sector to a second sector.

FIG. 5 is a flow diagram according to an embodiment of the present invention wherein when the partition map is modified, the procedure for detecting a misaligned partition is reset.

FIG. 6 illustrates an embodiment of the present invention wherein the data storage device comprises a disk drive.

FIG. 7 illustrates an embodiment of the present invention wherein the data storage device comprises a solid state drive.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

FIG. 2A shows a data storage device 8 according to an embodiment of the present invention comprising a non-volatile memory 10 comprising a plurality of sectors. The data storage device 8 further comprises control circuitry 12 operable to execute the flow diagram of FIG. 2B, wherein a partition map is evaluated (step 14) that identifies a partition accessed through a plurality of logical block addresses (LBAs), where each LBA maps to a fraction of a sector. A partition offset is determined for the partition relative to a boundary of one of the sectors (step 16). N write commands are received (step 20) each having a write offset relative to a corresponding sector. When the write offset for X of the N write commands matches the partition offset (step 30), at least part of the partition is moved to align at least part of the partition to a boundary of one of the sectors (step 32).

In the embodiment of FIG. 2B, an X counter and a Y counter are initialized to zero (step 18). Each time a write command is received from a host (step 20), the Y counter is incremented (step 22), and each time the write offset of the write command matches the partition offset of the partition (step 24), the X counter is incremented (step 26). After executing N write commands (Y=N at step 28), the X counter is evaluated, wherein if X/N is greater than a threshold (step 30), at least part of the partition is moved to align at least part of the partition to a boundary of one of the sectors (step 32). In the embodiment of FIG. 2B, if X/N is not greater than the threshold (step 30), then the flow diagram is repeated after resetting the X and Y counters (step 18).

In the embodiments of the present invention, when the LBAs of a write command do not align with the boundaries of the sectors in the non-volatile memory 10, the control circuitry 12 will read the fractional sector, modify the data using the new write data, and then write the modified data back to the sector. This is illustrated in FIG. 3A which shows a number of LBAs each mapped to a sector through a physical block address (PBA). In the example of FIG. 3A, each sector of the non-volatile memory stores four host blocks (HBs) such that there are four LBAs mapped to each PBA of a sector. However, any suitable fractional relationship may be employed between the number of LBAs mapped to a PBA of a sector, such as mapping eight LBAs to a PBA of a sector. If the host issues a write command starting at LBA 6 and having a length of eight LBAs, the control circuitry 12 will read PBA_—1, modify the HBs corresponding to LBAs 6 and 7, and rewrite the modified data to PBA_—1. A similar read-modify-write is required for the last sector of the write command. In this example, the write offset for the write command is two, meaning that the beginning of the write command is offset two LBAs from the beginning of the first sector. In the example of FIG. 3A, four LBAs map into a PBA, and therefore the write offset may be one of zero, one, two and three. In the embodiment where eight LBAs map into a PBA, the write offset may be one of zero, one, two, three, four, five, six and seven.

In the example of FIG. 3A, a partition is defined having a starting LBA of two and a corresponding partition offset of two meaning that the beginning of the partition is offset two LBAs from the beginning of the first sector of the partition. Since the partition offset is two LBAs, an assumption could be made that shifting the partition by two LBAs in order to align the partition with the boundary of a sector will reduce the number of read-modify-write operations and therefore improve performance. However, this assumption could be incorrect leading to even worse performance if the write offsets of the actual write commands do not match the partition offset. Accordingly in the embodiments of the present invention, the write offset of the write commands are evaluated over time to verify that the write offsets for a predetermined number of write commands actually match the partition offset before aligning at least part of the partition to a sector boundary. In the example of FIG. 3A, the write offset for N write commands are evaluated, and at least part of the partition is aligned with a sector (shifted by two host blocks) when the write offset for X of the N write commands matches the partition offset (i.e., when the write offset equals two for X of the N write commands).

Any suitable technique may be employed to move at least part of a partition to align at least part of the partition to a sector boundary. In an embodiment illustrated in FIG. 3B, moving at least part of the partition comprises shifting a starting LBA of the partition. In the example of FIG. 3B, the starting LBA of the partition is changed from LBA two to LBA four so that the beginning of the partition aligns with PBA_—1. In this embodiment, the partition map (e.g., stored in the MBR) is also modified to reflect the change to the staring LBA of the partition, and in one embodiment, the length of the partition may be modified. For example, if the partitions are defined using consecutive LBAs, the length of a partition may be reduced (e.g., by two LBAs in the example of FIG. 3B) to account for the shift in the starting LBA of the partition. Although the example of FIG. 3B shows the starting LBA of the partition shifting right by two LBAs, in another embodiment the starting LBA may be shifted left by the partition offset (and the length of the preceding partition reduced).

FIGS. 4A and 4B illustrate an alternative embodiment for moving at least part of a partition in order to align at least part of the partition to a sector boundary. In this embodiment, the control circuitry 12 reserves at least one pad sector at the end of each partition (e.g., PBA_—25 in FIG. 4A). The mapping of LBAs to PBAs is shifted by the partition offset so that the starting LBA of the partition (LBA two in FIG. 4A) aligns with the sector boundary as shown in FIG. 4B. The last two LBAs of the partition (98 and 99) are mapped to the first two host blocks of the pad sector at the end of the partition. In this embodiment, it is not necessary to modify the partition map (e.g., stored in the MBR) since the starting LBA (and length) of the partition remain the same. In another embodiment, the control circuitry 12 may reserve a pad sector at the beginning of each partition so that the LBA to PBA mapping may be shifted left by the partition offset.

In the embodiment of FIGS. 3A and 3B and in the embodiment of FIGS. 4A and 4B, the host blocks of previously written sectors are moved so that the mapping of LBA to HB remains intact. To avoid corrupting the host data while shifting the HBs, in one embodiment the control circuitry 12 may unmount the partition so that the host does not attempt to access the partition during the shifting procedure. Other embodiments may allow the host to access the partition during the shifting operation while taking other precautions, such as making a copy of HBs that are being shifted. In one embodiment, the beginning of a partition may store critical data, such as operating system data and file system data that if corrupted may render the data storage device inoperable. In this embodiment, the control circuitry 12 may shift a part of the partition that extends past a certain number of LBAs from the beginning of the partition. That is, the control circuitry 12 may choose not to shift the beginning of a partition to avoid the possibility of corrupting critical data during the shift operation.

FIG. 5 is a flow diagram according to an embodiment of the present invention which expands on the flow diagram of FIG. 2B. In this embodiment, a host may modify the partition map(s) of the data storage device (e.g., by writing to the MBR). Accordingly, the control circuitry 12 monitors the partition map(s) and when the partition map(s) is modified, the flow diagram of FIG. 5 is restarted by initializing the X and Y counters to zero at step 18. The control circuitry 12 may detect that a partition map is modified (step 34) during the process of evaluating the N write commands prior to shifting the partition. Alternatively, the control circuitry 12 may detect that a partition map is modified (step 36) after the partition has already been shifted (or while shifting the partition at step 32). In either case, the control circuitry 12 resets the X and Y counters so that the procedure of correlating the write offsets with the partition offset is restarted.

In the embodiments of the present invention, the control circuitry 12 may evaluate the partition map at step 14 of FIG. 2B using any suitable technique. In one embodiment, the control circuitry 12 may read a MBR from the non-volatile memory which is typically stored in the first sector (sector zero), where the MBR comprises a partition table that includes one or more partition maps. In another embodiment, the control circuitry 12 may receive a partition map from a host as part of a configuration command. For example, a host may send a command to configure the data storage device with desired partition map(s) using a suitable configuration command, or the host may notify the data storage device of the partition map configuration to enable the control circuitry to perform the internal partition alignment procedure described above. In yet another embodiment, the control circuitry 12 may monitor an access pattern of the non-volatile memory by the host in order to infer the content of the partition map(s), including the partition offset(s). In the flow diagram of FIG. 5, the control circuitry 12 may detect that a partition map has been modified by detecting a write to the MBR (a write to sector zero), or receiving a configuration or notification command from the host, or by detecting a change in the access pattern to the non-volatile memory, or using any other suitable technique.

The flow diagrams disclosed herein may be carried out by a microprocessor executing code segments of a program stored on a computer readable medium. Any suitable computer readable storage medium capable of storing code segments may be employed, such as the data storage device being aligned, or a separate data storage device, such as a disk drive, or a flash memory, or a CD.

The embodiments of the present invention may be used to align a partition of any suitable data storage device. FIG. 6 shows a data storage device in the form of a disk drive comprising a head 38 actuated over a disk 40 having a plurality of tracks 42. The disk drive further comprises control circuitry 44 for receiving access commands from a host (write/read commands) and for generating a control signal applied to a voice coil motor (VCM) to rotate an actuator arm about a pivot in order to position the head 38 radially over the disk 40 to access a target track. Each track is divided into a number of sectors, wherein each sector is capable of storing multiple host blocks.

FIG. 7 shows a data storage device in the form of a solid state drive comprising a plurality of non-volatile semiconductor memories 46A, 46B, etc., such as flash memories, and control circuitry 48 for accessing the non-volatile semiconductor memories. In one embodiment, a sector of the solid state drive corresponds to a page of memory in a memory array, wherein each page stores a plurality of host blocks. A hybrid data storage device may also be employed comprising components of a disk drive shown in FIG. 6 combined with the non-volatile semiconductor memories shown in FIG. 7.

Claims

1. A method of aligning a partition of a data storage device to a boundary of a sector, the data storage device comprising a non-volatile memory comprising a plurality of sectors, the method comprising: (a) evaluating a partition map that identifies a partition accessed through a plurality of logical block addresses (LBAs), and each LBA maps to a fraction of a sector;(b) determining a partition offset for the partition relative to a boundary of one of the sectors;(c) receiving N write commands each having a write offset relative to a corresponding sector; and(d) when the write offset for X of the N write commands matches the partition offset, moving at least part of the partition to align at least part of the partition to a boundary of one of the sectors, wherein X and N are positive integers.

2. The method as recited in claim 1, wherein moving at least part of the partition comprises shifting at least one LBA of the partition.

3. The method as recited in claim 2, wherein moving at least part of the partition comprises shifting a starting LBA of the partition.

4. The method as recited in claim 1, wherein moving at least part of the partition comprises remapping at least one LBA from a first sector to a second sector.

5. The method as recited in claim 1, further comprising repeating steps (a) through (d) when the partition map is modified.

6. The method as recited in claim 1, wherein each LBA maps to one-fourth of a sector such that the partition offset may be one of zero, one, two and three.

7. The method as recited in claim 1, wherein each LBA maps to one-eighth of a sector such that the partition offset may be one of zero, one, two, three, four, five, six, and seven.

8. The method as recited in claim 1, further comprising reading the partition map from the non-volatile memory.

9. The method as recited in claim 1, further comprising receiving the partition map from a host.

10. The method as recited in claim 1, further comprising inferring the partition map based on an access pattern of the non-volatile memory.

11. A data storage device comprising: a non-volatile memory comprising a plurality of sectors; andcontrol circuitry operable to aligning a partition of a data storage device to a boundary of a sector by:(a) evaluating a partition map that identifies a partition accessed through a plurality of logical block addresses (LBAs), and each LBA maps to a fraction of a sector;(b) determining a partition offset for the partition relative to a boundary of one of the sectors;(c) receiving N write commands each having a write offset relative to a corresponding sector; and(d) when the write offset for X of the N write commands matches the partition offset, moving at least part of the partition to align at least part of the partition to a boundary of one of the sectors, wherein X and N are positive integers.

12. The data storage device as recited in claim 11, wherein moving at least part of the partition comprises shifting at least one LBA of the partition.

13. The data storage device as recited in claim 12, wherein moving at least part of the partition comprises shifting a starting LBA of the partition.

14. The data storage device as recited in claim 11, wherein moving at least part of the partition comprises remapping at least one LBA from a first sector to a second sector.

15. The data storage device as recited in claim 11, wherein the control circuitry is further operable to repeat steps (a) through (d) when the partition map is modified.

16. The data storage device as recited in claim 11, wherein each LBA maps to one-fourth of a sector such that the partition offset may be one of zero, one, two and three.

17. The data storage device as recited in claim 11, wherein each LBA maps to one-eighth of a sector such that the partition offset may be one of zero, one, two, three, four, five, six, and seven.

18. The data storage device as recited in claim 11, wherein the control circuitry is further operable to read the partition map from the non-volatile memory.

19. The data storage device as recited in claim 11, wherein the control circuitry is further operable to receive the partition map from a host.

20. The data storage device as recited in claim 11, wherein the control circuitry is further operable to infer the partition map based on an access pattern of the non-volatile memory.