WO2017193967A1

WO2017193967A1 - Data inspection method and device in solid state memory

Info

Publication number: WO2017193967A1
Application number: PCT/CN2017/083988
Authority: WO
Inventors: 吴黎明; 赵万; 黄斌; 杨虓; 王永良
Original assignee: 华为技术有限公司
Priority date: 2016-05-12
Filing date: 2017-05-11
Publication date: 2017-11-16
Also published as: CN107368257A; CN107368257B

Abstract

A data inspection method and device in a solid state memory, relating to the technical field of storage. The method comprises: reading, in a current inspection cycle and according to the time when a memory block mounted in a first linked list group is mounted in the first linked list group, a valid data page stored in the memory block mounted in the first linked list group (201); storing the read valid data page into an enabled memory block, the enabled memory block being a blank memory block used for a data page to be written to (202); mounting the enabled memory block enabled in the inspection cycle to a second linked list group according to the time when the enabled memory block is enabled (203); and when a next inspection cycle of the current inspection cycle starts, using the second linked list group as a first linked list group, and using the first linked list group as a second linked list group (204). On the basis of optimization of data stability, the method can cohere stable data to some memory blocks, and increase the number of memory blocks that can be used for unstable data, so as to increase the redundancy rate, reduce write amplification and SSD wear, and prolong the service life of the SSD.

Description

Data inspection method and device in solid state memory

Technical field

The present invention relates to the field of storage technologies, and in particular, to a data inspection method and apparatus in a solid state memory.

Background technique

Solid State Disk (Solid State Disk), also known as electronic hard disk or solid state electronic disk, is a hard disk implemented by solid-state electronic storage devices, mainly composed of a controller and a memory. The SSD has the same interface specification, physical size, and usage mode as the ordinary hard disk. The special feature is that there is no mechanical structure, and the unique storage medium NAND Flash feature is used to perform the write operation in units of blocks. The page performs an erase operation for the unit. The SSD solid state drive frees the mechanical heads of the mechanical hard disk, the rotating shaft of the disk and the mechanical components such as the control motor. There is no motor to accelerate the rotation process. There are no mechanical moving parts inside, no mechanical failure, no collision, shock and vibration.

NAND Flash is a non-volatile random access storage medium, which is characterized by the fact that data does not disappear after power off. NAND Flash is based on a floating gate transistor design. The purpose of writing "0" or "1" is achieved by charging and discharging to the floating gate in the memory, and the data stored in the current memory cell is obtained by determining the amount of charge in the floating gate.

The existence of various leakage mechanisms makes the characteristics of electrons stored on the floating gate seriously affected. With repeated erasing, the channel insulation layer will be damaged, and the electrons in the floating gate will accelerate the leakage of charges due to the damage of the insulation layer. Stress-induced leakage current (SILC) also reduces the internal threshold voltage; electrons trapped in the insulating layer may be trapped by the floating gate, and the higher the temperature, the more obvious, called the inherent charge loss (ICL), which makes the internal threshold The voltage is reduced; these are all Data Retention errors.

Long-term storage of data in NAND Flash poses a risk of data loss due to Data Retention errors, especially if the host does not update the data for a long time. In order to prevent Data Retention errors from occurring, you need to use the inspection method to refresh the data to prevent the data from being stored for a long time.

At present, the method for implementing the patrol inspection by the SSD may be: according to the mapping table of the storage logical address to the physical address, according to the order of the logical addresses in the mapping table, the valid data in the storage block corresponding to the physical address mapped to the logical address is in turn Read out and write to a new blank block (a blank block is a block that has no data written, such as a block that is reclaimed and erased by the garbage collection module).

The existing SSD inspection methods have at least the following problems:

Since the foregoing patrol method is to patrol the storage blocks in the SSD according to the logical address, the stable data that has not been erased for a long time and the active data that has just been written are migrated to the same storage block during the patrol process. Since the active data in the memory block may be updated soon to become an invalid data page, and the stable data is not updated, it is still a valid data page, so that when the memory block is erased, it is required to be updated. Valid data pages are migrated, which increases write amplification.

Summary of the invention

In order to solve the problem of the prior art, an embodiment of the present invention provides a data inspection method in a solid state memory. And equipment. The technical solution is as follows:

In a first aspect, a data patrol method in a solid state memory is provided, the solid state memory comprising a plurality of memory blocks, each memory block being configured to store a plurality of data pages, wherein the solid chain memory further stores a first linked list a group and a second linked list group, wherein the first linked list group is mounted with a plurality of storage blocks, the second linked list group is for mounting an enabled storage block, and the enabled storage block is a storage enabled to write a data page The method includes: reading, in the current patrol period, the time in the first linked list group according to the time when the storage block mounted in the first linked list group is mounted to the first linked list group a valid data page stored in the mounted memory block; storing the read valid data page to the enabled memory block, the enabled memory block being a blank memory block used to write the data page; The enabled enabled storage block in the cycle is mounted to the second linked list group according to the time when the enabled storage block is enabled, and the second linked list group is taken as the beginning of the next inspection cycle of the current inspection cycle. The first linked list group, the first Group list as the second list set. Through the optimization based on data stability, stable data can be cohesive to a part of the storage block, and the number of storage blocks that can be used for the unstable data is increased, so that the redundancy ratio of the SSD can be significantly improved, thereby achieving the write amplification of the SSD. Reduce the wear and tear on SSD and extend the service life of SSD.

In a first possible implementation manner of the first aspect of the present invention, the first linked list group includes multiple linked lists, where each linked list is used to mount a storage block that is enabled in a same time period; The time when the storage block mounted in the first linked list group is mounted to the first linked list, and the valid data page stored in the storage block mounted in the first linked list group is read: The sequence of time periods corresponding to each linked list in the first linked list group reads the valid data pages stored in the storage blocks mounted in each linked list; and the enabled storage blocks enabled in the patrol cycle are enabled according to the storage block. Time-mounting to the second linked list group includes: mounting the enabled storage block to a linked list corresponding to a time period in which the enabled storage block is enabled. Reading the valid data page stored in the storage cache linked in the linked list corresponding to each time period according to the sequence of each time period in the first linked list, and enabling the enabled storage block to be mounted to the enabled storage block is enabled In the linked list corresponding to the time period in which the time is located, it can be ensured that the patrol operation can be performed on the storage block according to the activation time of the storage block during each patrol operation.

In a second possible implementation manner of the first aspect of the present invention, when the valid data page in the storage block mounted by the target linked list in the first linked list group is read, the method further includes: The target linked list establishes a plurality of child linked lists, each of which corresponds to an invalid data page, and the child linked list is sorted according to the number of invalid data pages; and the invalid data in the storage block mounted by the target linked list is determined. The number of pages; the storage block in the target linked list is mounted to the child linked list corresponding to the determined number of invalid data pages according to the determined number of invalid data pages; according to the child linked list in the target linked list The number of valid invalid data pages in the middle reads the valid data pages in the storage blocks mounted in each sub-linked list in order of increasing number. By sequentially reading the valid data pages of each of the plurality of memory blocks in order of the number of invalid pages included in the plurality of memory blocks, the blank memory blocks can be freed as soon as possible for writing Enter new data or use it to relocate data during the inspection process.

In a third possible implementation manner of the first aspect of the present invention, when the read valid data page is stored to the enabled memory block, the method further includes: when the enabled memory block is already full, Gets the valid data page read by the next enabled block store. By storing the valid data page to the enabled memory block, and after the enabled memory block is full, the next enabled memory block can achieve stable data cohesion and reduce write amplification.

In a fourth possible implementation manner of the first aspect of the present invention, the method further includes: after the valid data pages in the storage block are all read, the storage block from the first linked list Remove from the group. Before the inspection, the storage block that performs the recovery operation is removed from the first linked list group, and the data in the storage block can be prevented from being patrolled again during the inspection, thereby further reducing the inspection process. The number of reads and writes.

In a second aspect, a data patrol device in a solid state memory is provided, the device comprising a plurality of functional modules for performing the method of the first aspect described above. In a possible implementation, the apparatus further includes other functional modules for performing the methods described in the various possible implementations of the first aspect described above.

In a third aspect, a data patrol device in a solid state memory is provided, the device comprising:

processor;

a memory for storing processor executable instructions;

The processor is configured to read the first time according to a time when the storage block mounted in the first linked list group is mounted to the first linked list group in a current patrol period. a valid data page stored in a storage block mounted in the linked list group; storing the read valid data page to the enabled storage block, the enabled storage block being a blank storage block used to write a data page; The enabled enabled storage block in the patrol period is mounted to the second linked list group according to the time when the enabled storage block is enabled, and the second linked list is started at the beginning of the next patrol period of the current patrol period As the first linked list group, the first linked list group is used as the second linked list group. Through the optimization based on data stability, stable data can be cohesive to a part of the storage block, and the number of storage blocks that can be used for the unstable data is increased, so that the redundancy ratio of the SSD can be significantly improved, thereby achieving the write amplification of the SSD. Reduce the wear and tear on SSD and extend the service life of SSD.

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present invention. Other drawings may also be obtained from those of ordinary skill in the art in light of the inventive work.

1 is a schematic structural diagram of a solid state memory according to an embodiment of the present invention;

2A is a flowchart of a data inspection method in a solid state memory according to an embodiment of the present invention;

2B is a schematic diagram of a first linked list set according to an embodiment of the present invention;

2C is a schematic diagram of a target linked list according to an embodiment of the present invention;

FIG. 3 is a block diagram of a data patrol device in a solid state memory according to an embodiment of the present invention.

detailed description

The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

1 is a schematic structural diagram of a solid state memory according to an embodiment of the present invention, where the solid state memory includes a plurality of memory blocks, each of the memory blocks is configured to store a plurality of data pages, and the first linked list is further stored in the solid state memory. And a second linked list group, the first linked list group is mounted with a plurality of storage blocks, the second linked list group is used for mounting an enabled storage block, and the enabled storage block is a storage block enabled to write a data page .

Referring to Figure 1, the solid state memory is mainly composed of a controller, a memory, and a memory. The memory is used to provide a data storage service, and the controller allocates a data processing path for the data write operation, so that the host or the server accesses the data stored in the solid-state memory, thereby performing processes such as data reading, writing, and deleting. The memory is a solid state storage array, and the solid state storage array may be a flash array or a DRAM (Dynamic Random Access Memory) array, which is a core component of a solid state hard disk with hardware write protection function for emitting solid state. A request for reading, writing data, and garbage collection of a memory in which program instructions are stored, and the controller executes program instructions in the memory to process data in the memory.

Since data is stored in NAND Flash for a long time, there is a risk of data loss caused by Data Retention errors, especially in the case where the host does not update the data for a long time. In order to prevent DataRetention errors from occurring, you need to use the inspection method to refresh the data to prevent the data from being stored for a long time.

In order to avoid an error in the stored data due to electronic leakage of the storage medium, the SSD periodically performs a patrol and carry operation on the stored data to solve the above problem. Since the service life of the SSD is related to the rewritable number of the SSD, during the patrol process, the erase operation is repeatedly performed on the memory block in the SSD, thus affecting the service life of the SSD. In order to minimize the impact of the inspection operation on the life of the SSD, the present invention provides a data inspection method, which is used to store different time periods by introducing a linked list group. The storage block of the data store performs the patrol operation in the order of mounting to achieve the purpose of stable data cohesion, thereby reducing the write amplification, reducing the number of erasing and writing to the SSD, and thereby reducing the patrol during the subsequent patrol process. The impact of inspection operations on the life of the SSD.

FIG. 2A is a flowchart of a data patrol method in a solid state memory according to an embodiment of the present invention. The solid state memory includes a plurality of memory blocks, each of which is used to store a plurality of data pages, and the solid state memory further includes Storing a first linked list group and a second linked list group, the first linked list group is mounted with a plurality of storage blocks, the second linked list group is for mounting an enabled storage block, and the enabled storage block is enabled to be written The storage block of the data page; see Figure 2A, the method includes:

201. During the current patrol period, according to the time when the storage block mounted in the first linked list group is mounted to the first linked list group, the storage stored in the storage block mounted in the first linked list group is read. Valid data page.

The first linked list group is stored in the solid-state memory, and the first linked list group includes a plurality of linked lists, each linked list is used for mounting a storage block whose enabled time is in the same time period; the enabled time may be a storage block The time at which the data is written may also be the time when the storage block performs the erasing operation; the number of linked lists included in the first linked list group is determined by the span of the patrol period and the mounting time period in the first linked list group. For the determination of the span of the mounting time period, it may be set to any fixed value. The specific value and the determining method of the spanning period span are not limited in the embodiment of the present invention.

It should be noted that each storage block in all the storage blocks of the SSD corresponds to an index value, and the corresponding storage block can be uniquely determined according to each index value; for example, all storage blocks in the SSD form a one-dimensional array, and each The logical position of the memory block in the one-dimensional array is the index value of the corresponding memory block. In order to implement the mounting operation of the storage blocks in the SSD, each storage block corresponds to a structure, and the structure corresponding to each storage block includes a precursor node pointer and a subsequent node pointer. After the storage block is mounted to the linked list, Obtaining an index value of a previous storage block in the linked list of the mounted storage block, storing the index value at a precursor pointer of the structure of the mounted storage block, and storing the mounted storage block The index value of the block is stored at the backdrive pointer of the previous memory block. For example, three storage blocks a, b, and c are sequentially mounted under a certain linked list, and the index value of the a storage block is stored in the precursor node pointer of the corresponding structure of the b storage block, and the corresponding structure of the b storage block The index value of the c memory block is stored in the subsequent node pointer of the body.

It should be noted that, when the storage block mounted in the first linked list group is mounted to the first linked list group, the valid data page stored in the storage block mounted in the first linked list group is read. Previously, the storage block is mounted to the linked list corresponding to the corresponding time period of the first linked list group according to the order of the activation time of each storage block from first to last.

The number of the linked list included in the first linked list group is determined according to the patrol period and the span of the mounting time period, so that the storage blocks are sequentially mounted in the order of the activation time to the corresponding linked list of the first linked list group, which is convenient for The data in the storage block mounted under the first linked list group is inspected and managed.

For example, when a patrol period is 7 natural days, and the span time period of the first linked list group spans one hour, that is, the storage blocks enabled in the same hour are mounted under the same linked list, then the first The number of linked lists included in a linked list is 7*24. It should be noted that if any storage block is not enabled within an hour, the storage block may not be mounted under the linked list in the first linked list group corresponding to the period of time.

By mounting the storage block to the first linked list group according to the activation time of each storage block from the first to the last, the subsequent storage of the stored data can be performed according to the first linked list group. The mounting order of the storage blocks inspects the mounted storage blocks in the first linked list group.

According to the time when the storage block mounted in the first linked list group is mounted to the first linked list, reading the valid data page stored in the storage block mounted in the first linked list group includes: according to the first linked list Sequential reading of each time period in the group A valid data page stored in a storage block mounted in a linked list corresponding to each time period. That is, according to the span of the mounting time period, the storage block mounted under the first linked list of the first linked list group is inspected during the first time period of the patrol period, and the second in the patrol period. The time period patrols the storage blocks mounted in the second linked list of the first linked list group, and so on, until the patrol operation of all the storage blocks mounted under the first linked list group is completed.

As shown in FIG. 2B, the figure is a schematic diagram of the first linked list group, wherein the linked list A, B, C, and D respectively represent a linked list for mounting a storage block corresponding to each of the mounting time periods in the first linked list group. , a1, a2, a3, and a4 represent storage blocks sequentially mounted in the A linked list in the first linked list group. Similarly, b1, b2, and b3 indicate storages sequentially mounted under the B linked list in the first linked list group. The blocks, c1, c2, c3, and c4 represent memory blocks sequentially mounted under the C linked list in the first linked list group, and d1 and d2 represent memory blocks sequentially mounted under the D linked list in the first linked list group.

In FIG. 2B, according to the write time of the memory block mounted in the first linked list group, the valid data pages in the storage block in the first linked list group are read in order from first to last, that is, Read the data in the storage block mounted under the A, B, C, and D lists. Specifically, for the storage block mounted under the A-chain table, the data in the a1 storage block is enabled before the a2, a2 storage block. The data is enabled before a3, and the data in the a3 memory block is enabled before a4. Therefore, the mount order is a1, a2, a3, and a4. When reading the data in the storage block under the linked list, The valid data pages of the a1, a2, a3, and a4 memory blocks are also sequentially read in the order of the activation time.

When the valid data page in the storage block mounted by the target linked list in the first linked list is read, the method further includes: creating a plurality of sub-linked lists for the target linked list, and each sub-linked list corresponds to the number of invalid data pages And the child linked list is sorted according to the number of invalid data pages; determining the number of invalid data pages in the storage block mounted by the target linked list; storing the target linked list according to the determined number of invalid data pages The block is mounted to the child linked list corresponding to the determined number of invalid data pages; and the number of corresponding invalid data pages in the child linked list in the target linked list is read in the order of the number of invalid data pages in the descending order. A valid data page in the mounted memory block.

For example, when each storage block contains 128 pages, the target linked list includes 129 sub-linked lists, that is, the number of valid data pages in each storage block may be 0-128, and there are 129 cases. A memory block that does not contain a valid data page is mounted under the first child linked list in the target linked list, and a memory block containing one invalid data page is mounted under the second child linked list of the target linked list, and contains 128 invalid data pages. The storage block is mounted under the 129th sub-list of the target linked list, and the target linked list can be represented by FIG. 2C. In FIG. 2C, if there is no corresponding amount of invalid data in the storage block that needs to be sorted according to the number of invalid data pages. The storage block of the page does not mount the storage block under the corresponding sub-chain list of the target linked list. For the case where the storage block is mounted in the target linked list, the valid data pages in the corresponding storage block are sequentially traversed and recovered in order, and the storage block that does not contain the valid data page is also mounted on the first list of the target linked list. The memory block under the sub-chain table, because it does not contain the memory block of the valid data page, can be used as a blank memory block only when it is directly erased, and does not need to perform effective data relocation operation. Therefore, when the memory block is recovered, The storage blocks under the first sub-chain list of the target linked list are preferentially processed, and then the other storage blocks are sequentially processed according to the mounting order of the target linked list.

For example, in FIG. 2C, the numbers in the boxes indicate sub-chain lists for mounting the memory blocks containing the corresponding number of valid data pages, and the memory blocks a1, a2, a3, and a4 not including the valid data pages are mounted on the Under the first sub-chain list of the target linked list, that is, the sub-chain list labeled 0, the storage blocks b1, b2, and b3 including one valid data page are mounted under the second sub-chain list of the target linked list, that is, labeled 1. The child linked list, and so on, the memory blocks n1, n2, n3, and n4 including 128 valid data pages are mounted under the 129th child linked list of the target linked list, that is, the child linked list labeled 128.

It should be noted that the number of invalid data pages mounted under any sub-linked list of the target linked list is the same. The plurality of storage blocks may sequentially read the valid data pages in each storage block according to the mounting order, or may randomly read the valid data pages of each of the plurality of storage blocks without following the mounting order. The data of each of the plurality of storage blocks may be read in other orders, which is not limited by the embodiment of the present invention.

When a plurality of storage blocks are mounted in a linked list in the first linked list group, the plurality of storage blocks are sequentially read according to the order of the number of invalid data pages included in the plurality of storage blocks. The valid data pages of each memory block can free up blank memory blocks for writing new data or for relocating data during the inspection process.

202. Store the read valid data page to an enabled memory block, which is a blank memory block used to write a data page.

The enabled storage block may be used to store data that is relocated during the patrol process, and may also be used to store newly written data, which is not limited in this embodiment of the present invention. By storing the valid data page in the first linked list group read during the patrol process to the enabled storage block, and because the first linked list group is to mount the storage block according to the enable time, according to the first linked list group The mounting sequence is patrolled, and the data in the storage block with the earlier activation time can be stored in the same enabled storage block to achieve stable data cohesion. When the storage block is patrolled again, the storage block can be patrolled. All the stable data stored in the storage block is migrated. If the stable data in the storage block is dispersed in a plurality of storage blocks, the stable data in the plurality of storage blocks needs to be migrated, so the stable data is cohesive. By going to a memory block, it is possible to avoid the division writing and erasing operations on the data, thereby reducing the write amplification and prolonging the service life of the SSD. Among them, the stable data refers to the data in the SSD stored data that has not been overwritten or deleted for a long time, and can be regarded as stable data, and vice versa.

203. Mount the enabled enabled storage block in the patrol cycle to the second linked list group according to the time when the enabled storage block is enabled.

The specific method of mounting the enabled storage block in the patrol period to the second linked list group according to the time when the storage block is enabled is: mounting the enabled storage block to the time period corresponding to the time when the enabled storage block is enabled. In the linked list.

When the enabled storage block is not full, the valid data page in the next storage block is continuously read according to the mounting order of the storage blocks in the first linked list group, and the valid data page is written into the enabled storage block. Until the fill block is filled. After the enabled storage block is full, the valid data page read by the next enabled storage block storage is obtained, and the enabled storage storage block is mounted under the corresponding linked list of the second linked list.

When the enabled storage block is enabled, the enabled storage block is mounted under the second linked list group according to the enabled time of the enabled storage block, and the second linked list group is used for mounting according to the storage block's enable time. It should be noted that the enable time of the enabled memory block may be the time when the write operation is performed on the enabled memory block, or the time when the erase operation is performed on the enabled memory block, where the erase operation refers to The erase operation performed before this write operation is performed.

Maintaining two linked list groups of the first linked list group and the second linked list group in the SSD, when performing a patrol operation on the data in the storage block in the first linked list group during the current patrol period, The storage blocks that perform the write operation in the current patrol period are sequentially mounted to the second linked list group in the order of activation time; in the next patrol period of the current patrol period, in the second linked list group When the data in the mounted storage block performs the patrol operation, the storage blocks that perform the write operation in the patrol cycle are sequentially mounted in the first linked list group in the order of activation time.

It should be noted that, since performing a patrol operation on data in any storage block refers to a process of reading valid data in the storage block from the storage block and writing it into another storage block; When the data in the storage block in the first linked list group performs the patrol operation, the storage block that performs the write operation in the current patrol period is the enabled storage block, which includes writing the data according to the write request of the application. The storage block also includes the first write from the first during the inspection A storage block of data that is relocated by a storage block mounted in a linked list group.

If two linked list groups for managing the patrol process are maintained in the SSD, including the first linked list group and the second linked list group, the storage blocks mounted in the two linked list groups can be followed during the patrol process. The mount order performs the patrol operation. The process of performing the patrol inspection by the data patrol method may be: if a patrol period is 7 natural days, and the span of the first linked list group and the second linked list group spans one hour, the first The linked list group and the second linked list group respectively comprise 7*24 linked lists, and the storage blocks of the written data are sequentially mounted to the corresponding linked list of the first linked list group according to the activation time order of the storage blocks, for example, The storage block enabled in the first hour is mounted in the first linked list of the first linked list group, and the storage block enabled in the second hour is mounted in the second linked list of the first linked list group to This type of push is mounted to a block of memory under the same linked list with an enable interval of no more than one hour.

That is, during the first patrol period, the patrol operation is not performed, and only the storage block of the newly written data is subjected to the mount operation, and in the second patrol period, the first linked list group is hanged. The data in the storage block performs a patrol operation, and the storage block of the newly written data is mounted in the second linked list group, and the data in the newly written storage block includes valid data and basis for relocation during the patrol inspection process. The application's write requests the newly written data.

When performing the patrol operation on the first linked list group, performing valid data in the storage block of the first linked list of the first linked list group in the first mounting time period of the second linked list group Relocating, when the storage block is mounted to the second linked list of the second linked list group in the second mounting time period of the second linked list group, the second linked list of the first linked list group is mounted The valid data in the storage block is moved, and so on.

After the second patrol period ends, the data in the storage block of the first linked list group is patrolled, and the valid data in the storage block mounted under the first linked list group is relocated, and the storage blocks are After being recycled or erased, it is mounted under the erase block linked list group or the blank block linked list group to wait for the data to be erased or rewritten. In the third patrol period, the data in the storage block of the second linked list group is patrolled, and the storage block of the newly written data is mounted in the first linked list group.

By mounting the storage block to the corresponding linked list of the second linked list group according to the activation time, the purpose of stable data cohesion can be achieved, and when the storage block is patrolled again, all the stored in the storage block can be stored. Stable data is migrated. If the stable data in the storage block is dispersed in multiple storage blocks, the stable data in the multiple storage blocks needs to be migrated, so that the stable data is cohesively integrated into one storage block, which can be avoided. The data is divided and erased, and the SSD write amplification can be reduced in the process of patrolling the data in the storage block in the first linked list group, thereby prolonging the service life of the SSD.

In another embodiment of the present invention, during the operation of the SSD, when any of the storage blocks mounted in the first linked list group is reclaimed before being patrolled, the storage block is from the first linked list group. Excluding, if the valid data in the memory block is relocated, but the erase operation is not performed, the memory block is removed from the first linked list, and the memory block is mounted in the erase block linked list group, if the memory block An erase operation is performed to mount the memory block to the blank block linked list group.

By removing the memory block that performs the recovery operation from the first linked list group before the patrol inspection, it is possible to prevent the data in the storage block from being patrolled again during the patrol process, thereby further reducing the patrol process. The number of reads and writes.

During the preset time period of the patrol cycle, the patrol operation of all the storage blocks mounted in the first linked list group is performed. The preset time period can be set to any time period, which is not limited by the embodiment of the present invention. For example, if a patrol period is 7 natural days, and the span of the first linked list group spans one hour, the first linked list group includes 7*24 linked lists, and the second linked list group The mounting time period corresponding to the 6th/th +1th linked list is used to patrol the data in the storage blocks of the first to seventh linked lists of the first linked list, and the seven linked lists are logged. The data in the storage block is patrolled within one hour, and the 6th to 24th linked list in the second linked list group corresponds to the 8th to 14th linked list of the first linked list. The data in the storage block is patrolled, and the data in the storage block under the seven linked lists is patrolled within one hour, and so on, and a mount time span can be determined according to the patrol period. The number of linked lists that need to complete the inspection.

That is, on the last day of the patrol cycle, the patrol process of all the data in the storage block mounted in the first linked list group is completed. In the next patrol cycle, the storage blocks that perform the write operation in the first 6 days are sequentially mounted in the order of the write time to the corresponding linked list of the second linked list group, and on the seventh day, according to the above method, the first linked list group is Mount the data in the storage block to perform the patrol operation.

In different usage environments of the SSD, the newly enabled storage blocks are mounted to the corresponding linked list group according to the writing time or the erasing time, and data is carried on the data of different linked list groups in different inspection periods. Inspections can achieve the cohesion of stable data and reduce the purpose of write amplification.

204. At the beginning of the next patrol period of the current patrol period, the second linked list group is used as the first linked list group, and the first linked list group is used as the second linked list group, and step 201 is repeatedly performed.

In the next inspection cycle, the same cycle as the storage block in the first linked list group is patrolled, and the next cycle of the inspection operation is performed, and so on.

By reading the valid data in the storage block in the first linked list group according to the activation time from the first to the last, and storing the read valid data to the blank storage block, the patrol process can be performed during the patrol process. Separate stable data and unstable data into different storage blocks, which can greatly reduce the capacity occupied by stable data and increase the redundant space corresponding to unstable data, thereby effectively reducing write amplification, reducing wear and tear, and improving Performance and stability, extending the service life; further, by removing the storage block that performs the recovery operation from the first linked list group before the inspection, it is possible to avoid data in the storage block again during the inspection process. The inspection can further reduce the number of reading and writing in the inspection process; by performing the recycling operation on multiple storage blocks to be recycled in order of the number of invalid pages included, the number of invalid pages can be maximized. The storage blocks are recycled faster, which in turn optimizes garbage collection.

For example, an SSD with a capacity of 100 GB has an actual physical capacity of 128 GB. If 100 GB of valid data has been stored in the SSD, the overall redundancy ratio is (128-100)/100=28%, in the 100 GB of valid data. 50GB of data is stable data, rarely modified or deleted, another 50GB of data is unsteady data, storage time is short, and often modified or deleted, it should be noted that 128GB of actual capacity is involved in reading and writing. process.

When the patrol operation is performed, the memory block corresponding to the 50 GB stable data contains a part of the invalid page, that is, the 50 GB stable data actually occupies a storage block of about (50/100)*128=64 GB, after the patrol operation. The valid data is moved to the blank storage block, and the storage block occupied by the valid data is released. The inspection process is a storage block that coheres the stable data into a new storage block and occupies the original valid data. The process of releasing invalid data contained in it.

After the patrol operation, the original 100GB of valid data in the SSD, 50GB of stable data is cohesive to a memory block close to 50GB, and the available capacity of 50GB of unstable data is 128-50=78GB, redundancy ratio For (78-50)/50=56%, the redundancy ratio is twice the redundancy ratio of the original whole disk. Since the disk is equivalent to such a logical disk at this time, the logical capacity is 50 GB and the physical capacity is 78 GB. With a capacity of 50GB, the actual redundancy ratio of the disk is 56%, so the disk has more storage space for storing valid data during garbage collection, which can effectively reduce the amount of data moved during garbage collection, thereby reducing the SSD. Write amplification to extend the life of the SSD.

Further, 50 GB of unsteady data in the 100 GB of valid data is also cohesive into a new memory block due to Unstable data storage time is short and often modified or deleted. Therefore, invalid pages in memory blocks storing unstable data will become invalid data faster. By garbage collection, priority is given to reclaiming memory blocks containing many invalid pages. It is possible to enable memory blocks containing a large number of invalid pages to be recycled faster.

Through the optimization based on data stability, stable data can be cohesive to a part of the storage block, and the number of storage blocks that can be used for the unstable data is increased, so that the redundancy ratio of the SSD can be significantly improved by each of the plurality of storage blocks to be recycled. The storage blocks contain the order of invalid pages from most to less, and the plurality of to-be-recovered storage blocks are sequentially recovered, so that the available blank storage blocks can be obtained quickly, thereby optimizing the effect of garbage collection, thereby reducing the write of the SSD. Zoom in to reduce the wear and tear on the SSD and extend the life of the SSD.

FIG. 3 is a block diagram of a data patrol device in a solid state memory according to an embodiment of the present invention. Referring to FIG. 3, the apparatus includes a reading module 301, a storage module 302, and a processing module 303.

The reading module 301 is configured to read the time in the first linked list group according to the time when the storage block mounted in the first linked list group is mounted to the first linked list group in the current patrol period. a valid data page stored in the stored memory block;

a storage module 302, configured to store the read valid data page to the enabled storage block;

The processing module 303 is configured to mount the enabled enabled storage block in the patrol period to the second linked list group according to the time when the enabled storage block is enabled, and at the beginning of the next patrol period of the current patrol period, The second linked list group is used as the first linked list group, and the first linked list group is used as the second linked list group.

In a first possible implementation manner provided by the present invention, the first linked list group includes multiple linked lists, and each linked list is used to mount a storage block that is enabled in a same time period;

The reading module 301 is configured to read valid data pages stored in the storage blocks mounted in the linked lists according to the time period corresponding to each linked list in the first linked list;

The processing module 303 is configured to mount the enabled storage block to a linked list corresponding to a time period in which the enabled storage block is enabled.

In a second possible implementation manner provided by the present invention, the device further includes:

a linked list establishing module, configured to create a plurality of sub-linked lists for the target linked list, each sub-linked list corresponding to an invalid data page, and the sub-linked list is sorted according to the number of invalid data pages;

a determining module, configured to determine the number of invalid data pages in the storage block mounted by the target linked list;

a mounting module, configured to mount the storage block in the target linked list to the sub-linked list corresponding to the determined number of invalid data pages according to the determined number of invalid data pages;

The reading module 301 is further configured to read valid data pages in the storage blocks mounted in each sub-chain list according to the number of corresponding invalid data pages in the sub-chain table in the target linked list. .

In a third possible implementation manner provided by the present invention, the device further includes:

Obtaining a module, after the enabled storage block is full, acquiring a valid data page read by the next enabled storage block storage.

In a fourth possible implementation manner provided by the present invention, the device further includes:

And an excerpt module, configured to remove the memory block from the first linked list group after the valid data pages in the storage block are all read.

It should be noted that the data patrol device provided in the above embodiment is only exemplified by the division of the above functional modules when performing data patrol inspection. In actual applications, the functions may be assigned different functions according to needs. The module can be completed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the data patrol device and the data patrol method are provided in the same embodiment. The specific implementation process is described in detail in the method embodiment, and details are not described herein again.

A person skilled in the art may understand that all or part of the steps of implementing the above embodiments may be completed by hardware, or may be instructed by a program to execute related hardware, and the program may be stored in a computer readable storage medium. The storage medium mentioned may be a read only memory, a magnetic disk or an optical disk or the like.

The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalents, improvements, etc., which are within the spirit and scope of the present invention, should be included in the protection of the present invention. Within the scope.

Claims

A data patrol method in a solid-state memory, the solid-state memory comprising a plurality of memory blocks, each memory block for storing a plurality of data pages, wherein the solid-state memory further stores a first linked list group and a second linked list group The first linked list group is mounted with a plurality of storage blocks, the second linked list group is configured to mount an enabled storage block, and the enabled storage block is a storage block that is enabled to write a data page; The method includes:

During the current patrol period, according to the time when the storage block mounted in the first linked list group is mounted to the first linked list group, the storage block mounted in the first linked list group is read. a valid data page stored; storing the read valid data page to the enabled memory block;

And enabling the enabled enabled storage block in the patrol period to be mounted to the second linked list group according to the enabled time, and at the beginning of the next patrol period of the current patrol period, the second linked list group is taken as The first linked list group is the first linked list group as the second linked list group.
The method according to claim 1, wherein the first linked list group comprises a plurality of linked lists, each linked list for mounting a storage block whose enabled time is within a same time period;

And reading, according to a time when the storage block mounted in the first linked list group is mounted to the first linked list, reading a valid data page stored in the storage block mounted in the first linked list group includes :

And reading valid data pages stored in the storage blocks mounted in each linked list according to the time period corresponding to each linked list in the first linked list group;

The mounting the enabled storage block enabled in the patrol period to the second linked list group according to the enabled time includes:

The enabled storage block is mounted in a linked list corresponding to a time period in which the enabled storage block is enabled.
The method according to claim 2, wherein when the valid data page in the storage block mounted by the target linked list in the first linked list is read, the method further comprises:

Establishing a plurality of sub-linked lists for the target linked list, each sub-linked list corresponding to an invalid data page, and the sub-linked list is sorted according to the number of invalid data pages;

Determining the number of invalid data pages in the storage block mounted by the target linked list;

Storing the storage blocks in the target linked list to the child linked list corresponding to the determined number of invalid data pages according to the determined number of invalid data pages;

The valid data pages in the storage blocks mounted in the respective child linked lists are read in order from the number of corresponding invalid data pages in the child linked list in the target linked list.
The method of claim 1, wherein when the read valid data page is stored to the enabled memory block, the method further comprises:

After the enabled memory block is full, the valid data page read by the next enabled memory block is fetched.
The method of claim 1 wherein the valid data page in said memory block is fully After the reading, the memory block is removed from the first linked list.
A data patrol device in a solid-state memory, the solid-state memory comprising a plurality of memory blocks, each memory block for storing a plurality of data pages, wherein the solid-state memory further stores a first linked list group and a second linked list group The first linked list group is mounted with a plurality of storage blocks, the second linked list group is configured to mount an enabled storage block, and the enabled storage block is a storage block that is enabled to write a data page; The device includes:

a reading module, configured to read, in the current patrol period, the time in the first linked list group according to the time when the storage block mounted in the first linked list group is mounted to the first linked list group A valid data page stored in the mounted memory block;

a storage module, configured to store the read valid data page to the enabled storage block, where the enabled storage block is a blank storage block used to write a data page;

a processing module, configured to mount the enabled enabled storage block in the patrol period to the second linked list group according to a time when the enabled storage block is enabled, at the beginning of the next patrol period of the current patrol period, The second linked list group is used as the first linked list group, and the first linked list group is used as the second linked list group.
The apparatus according to claim 6, wherein the first linked list group comprises a plurality of linked lists, each linked list for mounting a storage block whose enabled time is within a same time period;

The reading module is configured to read valid data pages stored in the storage blocks mounted in each linked list according to the time period corresponding to each linked list in the first linked list;

The processing module is configured to mount the enabled storage block to a linked list corresponding to a time period in which the enabled storage block is enabled.
The device according to claim 7, wherein the device further comprises:

a linked list establishing module, configured to create a plurality of sub-linked lists for the target linked list to be read, each sub-linked list corresponding to an invalid data page, and the sub-linked list is sorted according to the number of invalid data pages;

a determining module, configured to determine the number of invalid data pages in the storage block mounted by the target linked list;

a mounting module, configured to mount the storage block in the target linked list to a sub-chain list corresponding to the determined number of invalid data pages according to the determined number of invalid data pages;

The reading module is further configured to read valid data pages in the storage blocks mounted in each sub-chain list according to the order of the number of corresponding invalid data pages in the sub-chain table in the target linked list.
The device according to claim 6, wherein the device further comprises:

Obtaining a module, after the enabled storage block is full, acquiring a valid data page read by the next enabled storage block storage.
The device according to claim 6, wherein the device further comprises:

And an excerpt module, configured to remove the memory block from the first linked list group after the valid data pages in the storage block are all read.