JP2009093528A - Memory data management apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the memory corruption resistance of a NAND type flash memory and to store data and backup data while efficiently using the memory. <P>SOLUTION: In storing data in the NAND type flash memory 11, data are stored by using respective pages in order. In this case, identification number (ID number) is assigned to each data to be stored, the identification number is inserted into a header part of each data and a marker is added to the top of the header part. Further, when the same value as the marker exists in each data, conversion for further inserting a marker value is executed so that marker values in the data are continuously (overlappedly) arrayed. Consequently, the top position of the header part can be easily specified by retrieving a marker value on which a bit array is not overlapped and the header part can be arranged on an optional position in the memory. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a memory data management device that manages data writing and data reading to a nonvolatile storage device such as a flash memory (in particular, a NAND flash memory).

  There is a physical limit to the number of times data can be rewritten in the flash memory, and the maximum number of times data is rewritten is usually about 100,000 times. A flash memory cell (memory cell) that has been rewritten further has a high possibility of being destroyed, and once destroyed, data rewriting becomes impossible thereafter.

  There are a plurality of types of flash memories, but in recent years, the integration density of NAND flash memories has increased, and the price has been reduced. In the NAND flash memory, when data is rewritten, new data is written after data is erased in units of blocks. That is, at the time of data rewriting, even in the case of 1-byte data rewriting, it is always necessary to erase in units of blocks (usually several tens of kilobytes).

  Therefore, when data is rewritten many times, if the write position (address) is always fixed, the memory cell at that position will eventually be destroyed. Such destruction of a memory cell at a certain writing position makes the entire block including the memory cell unusable, resulting in a large loss of capacity, efficiency, and reliability.

To solve the above problem,
・ Distribute data writing positions to prevent destruction of memory cells due to concentration of data writing. ・ If damage is detected by an error code, spare blocks are used instead. Conventional techniques such as these are known.

  As a conventional technique, there is known a technique for managing used / unused memory areas and managing the number of rewrites (erase) in units of blocks and storing the management information in an arbitrary location in the flash memory. Yes. For example, in Patent Document 1 below, data cannot be written unless the flash memory is in an initial state. Therefore, in order to perform efficient data rewriting, block and data management information is stored on the flash memory, and this management is performed. A mechanism for separately performing writing processing and erasing processing using information is disclosed.

Further, even when the power supply is interrupted for some reason during the erasing or writing of data, the data is destroyed. In this case, it is desirable that data is restored when power is supplied again. For this purpose, for example, there is a method of always returning to a fixed initial value when power is supplied again, or a method of storing backup data in advance in preparation for destruction of data. . The latter method of saving backup data is to save the last correctly written data without erasing it, and if data corruption occurs when writing new data, the saved data is updated to the latest data. It is a method to treat as.
JP 2000-222292 A

  However, in the conventional method that distributes the number of times of writing for each block by managing in units of blocks, if writing to the same position (offset from the block first byte) in the block occurs intensively, even if the block is distributed Even so, there is a first problem that the lifetime of the block is shortened due to destruction of some memory cells.

  In addition, with the conventional method of storing management information in blocks separate from data, management information needs to be updated each time data is written, and recovery is performed in consideration of the destruction of the block that holds management information. This mechanism is necessary. This is inefficient both in terms of processing amount and memory usage, and there is a second problem that there is still concern about block destruction. That is, according to the technology for storing the management information in the flash memory, it is necessary to update the management information at the same time as the rewriting of data, and the number of times of data rewriting of the memory cell is consumed by the update of the management information. It will end up. Further, since there is a possibility of destruction of the management information itself (or possibility of destruction of the memory cell in which the management information is written), it is necessary to back up the management information in another block.

  In addition, when performing a method of saving backup data, at least the backup data must be written in a separate block in order to ensure that the data can be recovered regardless of the power interruption. Furthermore, it may be necessary to store information indicating in which block (or address) the backup data exists somewhere in the memory. Further, it is necessary to save the backup data without erasing it every time data is written.

  Further, when data recovery is performed when the data is destroyed, it is desirable that the data be restored as close to the data as possible before the data is destroyed. In this case, the backup data needs to be stored in a separate block. However, when the backup unit is a block, if the data size is smaller than the block size, it is inefficient in terms of memory usage. There is a third problem.

  In addition, according to the conventional technique, every time new data is saved, backup is performed at the same time. Therefore, there is a fourth problem that the amount of processing, the number of times of writing, and the writing time are increased by the backup processing. .

  In order to solve the above problems, the present invention improves the memory destruction resistance performance of the NAND flash memory and can efficiently use the memory by efficiently managing the data and storing the backup data. An object of the present invention is to provide a simple memory data management device.

In order to achieve the above object, according to the present invention, in a memory data management device for managing the writing / reading of data to / from a NAND flash memory,
In the NAND flash memory, a plurality of divided areas in which the storage area is divided in advance for each predetermined capacity are defined as pages, respectively, and the first divided area from the first divided area to the last divided area of the storage area A continuous page number is assigned up to the last page, and by designating this page number, data is written / read out for each page. It is performed in a batch for each block consisting of a predetermined number of pages, and all the recording areas in the erased block are memories that have a predetermined value determined in advance by the erasing process,
Different from the predetermined value determined in advance by the erasing process at the head of the header part including the identification number assigned to identify each different type of data and the data length information of the data assigned the identification number, and the identification A marker set to a value not used for the number is added, and when the same value as the value of the marker exists in the data, the data is converted so that the value overlaps to generate converted data, Write data generation means for combining the header portion to which the marker is added and the conversion data to generate write data;
Referring to each page of the NAND flash memory, unused area specifying means for specifying an erased page whose data value stored in each page is the predetermined value as an unused area;
Writing means for writing the data for writing from the start position of the writable area, using the unused area specified by the unused area specifying means as the writable area;
The page number is traced back from the written area immediately before the writable area to search for data with a value that matches the marker value, and the data with the value that matches the marker value was not duplicated. In this case, this data is determined as the marker, and a header part specifying means for specifying the position of the marker as the head position of the header
From the header part specified by the header part specifying means, specify the header part including each identification number assigned to each of the different types of data generated by the write data generation means, and identify each identification number. The storage location of the latest written data of the different types of data to which each identification number is assigned is specified by making the first detected written area the final writing area of the different types of data to which the respective identification numbers are assigned. The latest data identification means,
A memory data management device is provided.

  Furthermore, according to the present invention, in addition to the above-described configuration, the write data generation means, when adding the marker to the header portion, the predetermined value determined in advance by the erasing process immediately after the marker. And a specific value different from that of the marker, or when adding the marker to the header part, the value of the head part of the identification number arranged immediately after the marker is set as the value of the marker. There is provided a memory data management device that is set so as not to match.

  According to the present invention, the memory destruction resistance of the NAND flash memory can be improved, and efficient data use can be realized by performing efficient data management and backup data storage.

  Hereinafter, an example of the configuration of a data management device (memory data management device) of a nonvolatile storage device according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of a configuration of a data management device of a nonvolatile storage device according to an embodiment of the present invention.

  As shown in FIG. 1, a data management device of a nonvolatile storage device using a NAND flash memory 11 is, for example, a controller 13 mounted on a CPU (Central Processing Unit) 12. Realized. The CPU 12 is configured to be able to read and write data to and from the RAM 14 such as the NAND flash memory 11 and, for example, an arithmetic RAM (Random Access Memory) included in the nonvolatile storage device.

  FIG. 2 is a diagram showing an example of memory division of the NAND flash memory in the embodiment of the present invention. As shown in FIG. 2, the NAND flash memory 11 stores a program executed by the CPU 12, and a storage space (data storage space) for storing data is secured. Yes. Data written to and read from the NAND flash memory 11 by the CPU 12 is stored in this data storage space.

  FIG. 3 is a diagram showing an example of an address space in which data to be written / read to / from the NAND flash memory is arranged in the embodiment of the present invention. Data written to and read from the NAND flash memory 11 is arranged in an address space as shown in FIG. This corresponds to the address space on the RAM 14. In FIG. 3, a continuous address space is divided into several areas, and IDs are assigned to the divided areas (ID = 1 to 3 in FIG. 3). For example, the data in the area of ID = 2 will be described below as data compressed at the time of writing to the NAND flash memory 11. FIG. 3 shows an example in which IDs are added to areas obtained by dividing a continuous address space. However, each area to which an ID is added need not exist in a continuous address space; Even in a space (the addresses of each other overlap), there is no problem because each area can be distinguished by the ID.

  In the present invention, as will be described later, the size of the data to be written (the data subsequent to the header information) is included in the header information added to the data. For this reason, the size of data in the address space and the size of data written to the flash memory do not necessarily have to match. In addition, the size of data may change each time data is written. Therefore, in the present invention, the data size can be reduced by using the data compression technique.

  The controller 13 of the CPU 12 can generate memory management information and data management information by analyzing the contents of the NAND flash memory 11 when the system is started, for example. Then, the controller 13 of the CPU 12 reads the data with each ID added onto the RAM 14 and at this time checks the checksum. As a result of the checksum check, when it is detected that the data is destroyed, old data corresponding to the destroyed data is newly read onto the RAM 14 as backup data. Data that has been compressed (for example, data with ID = 2) is decompressed at the same time as being read into the RAM 14, and as a result, the decompressed data is placed on the RAM 14.

  As described above, by adopting a configuration in which the data stored in the NAND flash memory 11 is read onto the RAM 14, the data read processing related to the NAND flash memory 11 needs to be performed only once at the time of system startup or the like. . That is, since the data stored in the NAND flash memory 11 is expanded on the RAM 14 when the system is activated, it is not necessary to read the data from the NAND flash memory 11 each time. The RAM 14 has a faster data writing / reading processing speed than the NAND flash memory 11. Therefore, according to the present invention, the processing speed for reading data is faster than when data is read from the NAND flash memory 11 in response to a data read request.

  On the other hand, when updating the data of the NAND flash memory 11, first, the data of the RAM 14 expanded on the RAM 14 is updated. At the time of this update, the data in the NAND flash memory 11 is not updated. The data in the NAND flash memory 11 is updated at another arbitrary timing by writing the data in the RAM 14 updated up to that timing. As a result, a plurality of data writes to the NAND flash memory 11 that have been conventionally required can be combined into one data write, and the number of data writes to the NAND flash memory 11 can be reduced. Become.

  When using an operating system that supports multi-threading, a write-only thread is prepared. Then, using this writing thread, it is checked whether or not the data to which each ID written in the RAM 14 is rewritten at a certain timing (for example, every second), and the data of an arbitrary ID is rewritten. If it is detected, data is written into the NAND flash memory 11. The write-only thread always checks whether there is an unused area with a certain capacity or more. If the unused area is insufficient, a process for increasing the unused area is performed.

  Next, data identification numbers (ID numbers) referred to when data is written / read in the embodiment of the present invention will be described.

  As described above, the data stored in the NAND flash memory 11 is arranged in the same address space, and the address space is divided into several areas, and each divided area is unique. The identification number (ID number) is assigned. Data writing / reading is performed in units of ID numbers.

  FIG. 4 is a diagram showing an example of the structure of data actually written in the NAND flash memory in the embodiment of the present invention. Note that the length of this data may exceed the length of one page of the NAND flash memory.

  FIG. 4A is a diagram illustrating an example of a structure in which a marker (index) is added to the head of the header information portion of data in the embodiment of the present invention. In FIG. 4 (a), the header information part has 2 bytes as the ID number part which is the description area of the ID number, 2 bytes as the size part which is the area to describe the data size of one data to which the ID is added, The checksum insertion part, which is an area for describing the checksum for checking the validity of the data part, has 2 bytes, and the size of 1 byte for enabling the information to be correctly searched in the memory The marker is added to the beginning. The header information part of 7 bytes in total as described above is added to the head of the data.

  However, the marker value must satisfy the following conditions. That is, in the header information part, the first 1 byte (1 byte immediately after the marker) of the 2-byte ID arranged next to the marker must have a value different from the marker value. Since all the bits in the unused area of the type flash memory are set to 1, it is 0xFF in byte units, but the marker value must be different from this value (0xFF). In order to ensure that the former condition is satisfied, a specific value (for example, a value such as 0x00) different from the marker value may be inserted immediately after the marker as shown in FIG. In this case, the length of the header information part is 8 bytes in total.

  In addition, regarding the data arranged in the data part, if the same value as the marker is included in the data, it is the same as the marker immediately after the bit array having the same value as the marker included in the data. Insert a value before writing it to memory. In other words, if the data contains the same value as the marker, the same value as the marker in the state of being stored in the memory is a bit array state that is continuously duplicated (the same value as the marker is repeated twice). In a state where it is arranged). The position of the header information part of each data is specified by this marker, and by examining each data in the memory, it becomes possible to distinguish the header information part, the data part, and the unused area. Since the position of this marker does not depend on the page, the ID can be arranged at an arbitrary position without fixing the ID at the top of each page, and data can be packed and written immediately after the data written immediately before. become able to. Therefore, the memory can be used efficiently. In the data portion, the portion having the same value as the marker value has a duplicate bit arrangement in the state stored in the memory, but when actually read, the original bit arrangement is restored. Need to be returned.

  In addition, even if the same value as the marker is included in the data part, the same value as the marker included in the data part is the bit arrangement state in which the same value as the marker is continuously repeated (the same value as the marker). By writing to the memory in a state where the ID is continuously arranged twice, the same value is repeated depending on the setting of a specific value (for example, 0x00) inserted immediately after the head portion of the ID or the marker. This makes it possible to distinguish from a marker in the header information portion that is not generated, and to reliably identify the marker present in the header information portion and the same value as the marker present in the data portion.

  In addition, when a certain area on the memory is in an initialized state or an erased state (that is, when the area is unused and in a writable state), the header information portion of the area has a specific information The ID number of the value (0xFF) is set. This is because the value of the NAND flash memory 11 is set to 0xFF immediately after initialization or immediately after erasure. Thereby, it is possible to determine whether each page is unused or in use by examining the value of the ID number of each area. Note that the value of the marker or a specific value (for example, 0x00) inserted immediately after the marker needs to be set to a value different from 0xFF indicating the unused area.

  Next, the operation of the data management device (memory data management device) of the nonvolatile memory device according to the embodiment of the present invention will be described. First, with reference to FIG. 5, a description will be given of processing at the time of starting the data management device of the nonvolatile storage device according to the embodiment of the present invention. FIG. 5 is a flowchart showing an example of processing at the time of starting the data management device of the nonvolatile storage device according to the embodiment of the present invention.

  First, the data management device clears the memory management information at startup (when power is turned on) (step S101), and then sequentially reads each page in the memory until all pages have been read (step S102). Repeat the process of detecting markers. If the same value as the marker continues immediately after the marker value found by the search, the value exists in the data part, and the value is not processed as it is, and the marker continues. Search for. On the other hand, if the header information portion marker is found, the header information portion added to the data in the NAND flash memory 11 is read (step S103), and the header information portion is analyzed to generate memory management information (step S104). ).

  Here, the memory management information generated in step S104 will be described with reference to FIG. FIG. 6 is a diagram showing an example of memory management information managed by the data management device of the nonvolatile storage device according to the embodiment of the present invention. Whether the memory management information is unused or erased (writable state) or already written (writable state) for all areas included in the data storage space of the NAND flash memory 11 It is information for managing.

  As described above, in the present invention, it is possible to determine whether an information area is unused or in use by examining the value of the ID number in the header information portion corresponding to each information area. Therefore, for example, as shown in FIG. 6, the value of the ID number part of the header information part is extracted, and the information area at that time and the information of the data position in the information area are retained on the RAM 14. Configure memory management information.

  Then, memory management information is generated based on the information (for example, ID number) in the header information section read out in step S103 (step S104), the page counter is incremented by one (step S105), and the process related to step S102 is performed again. By returning to step S, all bytes included in all areas of all pages are sequentially analyzed, and memory management information including information on all areas is generated.

  When reading of all pages is completed and memory management information including information on all areas is generated, the data management apparatus, based on the memory management information, the area included in the data storage space of the NAND flash memory 11 Is used / unused (step S106), an unused area is searched, the corresponding continuous area is set as a writable area, and the start position of the unused continuous area is set as a write start position. (Step S107).

  Subsequently, the data management apparatus repeatedly performs processing for generating data management information for managing the position and data size where the latest data is written until the data for all ID numbers has been searched (step S108). ). In this processing, the data management device specifies each header information part by searching for a marker included in the header information part, going back from the byte immediately before the write start position, and each ID included in each header information part. The latest information part (latest information area) of the number is searched (step S109).

  Then, the checksum included in the checksum part of the latest information part found as a result of the search is extracted, and the checksum of the data existing in the data part of the latest information part is calculated, and these checksums match. It is inspected whether or not to perform (step S110). If the checksum in the header information part matches the checksum obtained as a result of the calculation as a result of this check (“Yes” in step S110), the data in the latest information part is correct data (not destroyed). The data management information is updated by adding a set of the ID number of the latest information part, the address for specifying the latest information part, and the data size to the data management information (step S111).

  On the other hand, if the checksum in the header information part does not match the checksum obtained as a result of the calculation in step S110 (“No” in step S110), the latest data is destroyed. The data recovery process is performed. In this restoration process, the backup data having the same ID number is searched further back from the latest position (step S112), and the validity of the data by the checksum is again checked for the found backup data. Note that backup data is searched again until valid data (data that has not been destroyed) is found. Then, after performing data restoration with the effective backup data as the latest information part (latest position), the memory management information and the data management information are updated corresponding to this data restoration. Then, after updating the data management information of all the ID numbers, it waits for a data write request (step S113).

  Here, generation processing of data management information will be described with reference to FIGS. FIG. 7 is a diagram showing an example of the last write page and the backup page of the NAND flash memory according to the embodiment of the present invention. FIG. 8 is a diagram showing an example of data management information managed by the data management device of the nonvolatile storage device according to the embodiment of the present invention. Here, for the sake of simplicity, it is assumed that the marker is always located at the top of the page (that is, the header information part including the marker and the total length of the data match the number of bytes in one page). However, the present invention is not limited to this assumption, and is also applicable to more general conditions such as when the information part (information area) including the header information part and the data part does not match the page size. Is possible.

  For example, when the final write data area of the NAND flash memory 11 (assuming the page unit in FIG. 7 as described above) is in the state of FIG. 7 at the time of startup, the final write data with the ID number A is “ID: A Last written page ”. Further, the information area found at the beginning of the marker from the “ID: A last written page” becomes the area of the last written data of each ID. In FIG. 7, “ID: B last write page” that is the last write data with ID number B and “ID: C last write page” that is the last write data with ID number C are specified.

  Also, for example, when it is determined by the checksum of “ID: A last written page” that the data in “ID: A last written page” has been destroyed, the ID number goes back up the information area. A is searched for the backup data of A, and as a result, “ID: A backup page” is found. FIG. 7 also shows “ID: B backup page” which is backup data with ID number B. If the data of “ID: B last written page” is destroyed, this “ Data recovery is performed using “ID: B backup page”.

  As a result, as shown in FIG. 7, for each ID number, a data management having a set of the ID number of the latest information part (final write data), the address of the latest information part (final write data), and the data size. Information is created. The data management information is information for managing in which information area the latest data is included for each ID number. Thereafter, the data management information is sequentially updated when new writing is performed.

  As described above, the data management device of the nonvolatile memory device according to the embodiment of the present invention generates the memory management information and the data management information by analyzing the information written in the NAND flash memory 11 Since these pieces of information can be held on the RAM 14, it is not necessary to secure an area for holding these pieces of information on the NAND flash memory 11. As a result, the memory management information and the data management information are generated, for example, every time the system is started. Therefore, the memory management information and the data management information are not held in the NAND flash memory 11, and the NAND type As the flash memory 11 deteriorates, the memory management information and the data management information itself are not destroyed.

  When data rewrite occurs in the data write request waiting state in step S113, the data management device compares the writable area size on the NAND flash memory 11 with the size of the new data to be written. Whether or not writing is possible is determined.

  When data can be written, the data management device adds header information to the data, starts writing from the write start position, and when the data writing is completed normally, the data management information and the memory management information Update. As described above, according to the present invention, the process of erasing the previous data is not performed at the time of data writing, so that it is not necessary to perform the process of copying the backup data to another location (another information area) and storing it.

  On the other hand, when data cannot be written, the following processing is performed in order to secure an area where data can be written on the NAND flash memory 11. First, it is checked whether or not the data of the next block continuous to the writable area (the block subsequent to the writable area) can be erased.

  FIG. 9 is a diagram illustrating an example of a method of performing block erase after the data management device of the nonvolatile storage device according to the embodiment of the present invention backs up data in the non-erasable area. In order to simplify the description also in FIG. 9, the header information part including the marker and the total length of the data are illustrated so as to coincide with the number of bytes of one page. It is also applicable to.

  As shown in FIG. 9, when data that needs to be backed up (erasable area) is included in this block (erase candidate block), the data is first copied to a writable area. Make a backup. Thereafter, the data in the erase candidate block is erased. The non-erasable area is the latest information area specified by one of the ID numbers stored in the data management information.

  However, in the method of performing block erasure after backing up the data in the non-erasable area shown in FIG. 9, there is a problem if the data size of the ID number required for backup exceeds the size of the writable area. Occurs. That is, in this case, since the data size of the non-erasable area exceeds the data size of the writable area, the data in the non-erasable area cannot be directly backed up to the writable area, and the writable area can be obtained by block erase. The data in the non-erasable area can be copied only after the image is enlarged.

The following three patterns A to C exist as patterns that exist when such a problem occurs.
A. When only one data needs to be backed up. When there is one data that needs to be backed up, but the writable area needs to be expanded to 2 blocks or more. When there are multiple pieces of data that need to be backed up

  Hereinafter, backup processing A to C in each of the above patterns A to C will be described with reference to FIGS. 10 to 12, for the sake of simplicity of explanation, the header information part including the marker and the total length of the data are illustrated so as to match the number of bytes of one page. It is applicable also in the conditions which have.

  First, the backup process A in the case of the above pattern A will be described. FIG. 10 is a diagram schematically showing an example of backup processing A of the data management device of the nonvolatile storage device according to the embodiment of the present invention.

  In FIG. 10, (1) when backup of the non-erasable area of ID: A is required, the data size of the non-erasable area is compared with the data size of the writable area. (2) If the data size of the writable area is insufficient and the data in the non-erasable area cannot be directly copied, (3) First, the non-erasable area is stored on another memory such as the RAM 14. Temporarily save data. Then, (4) block erasure is performed on the block (erase candidate block) in which the non-erasable area exists, the area including the writable area and the erase candidate block is expanded to the writable area, and (5) the RAM 14 Copy the data in the non-erasable area to the writable area.

  Next, the backup process B in the case of the above pattern B will be described. FIG. 11 is a diagram schematically illustrating an example of the backup process B of the data management device of the nonvolatile storage device according to the embodiment of the present invention.

  11 is similar to the backup processing A shown in FIG. 10, (1) when backup of the non-erasable area of ID: A is required, the data size of the non-erasable area and the writable area Compare with the data size of. (2) If the data size of the writable area is insufficient and the data in the non-erasable area cannot be directly copied, (3) First, the non-erasable area is stored on another memory such as the RAM 14. Temporarily save data. Then, (4) block erasure is performed on the block (erase candidate block) where the non-erasable area exists, and the area including the writable area and the erase candidate block is expanded to the writable area.

  However, even if the writable area is enlarged, if (5) the data size of the writable area is still insufficient and the data in the non-erasable area cannot be directly copied, (6) erased in the RAM 14 While the non-erasable area is stored, the next block is erased to further expand the writable area, and then data in the non-erasable area is copied to the writable area.

  Next, the backup process C in the case of the above pattern C will be described. FIG. 12 is a diagram schematically showing an example of backup processing C of the data management device of the nonvolatile storage device according to the embodiment of the present invention.

  In the backup process C, when there is a plurality of data that needs to be backed up, the data that can be copied to the writable area is copied in order. On the other hand, if the size of the writable area is insufficient, the data that could not be copied The data is copied after the block is erased to erase the block and the writable area is expanded.

  In FIG. 12, a backup process C related to two data (ID: A non-erasable area and ID: B non-erasable area) will be described. Here, the non-erasable area of ID: A has a larger data size than the initial writable area, and the non-erasable area of ID: B has a smaller data size than the initial writable area.

  In FIG. 12, (1) the data size of the non-erasable area of ID: A that needs to be backed up is compared with the data size of the writable area. (2) When the data size of the writable area is insufficient and the data in the non-erasable area cannot be directly copied, (3) the ID: A cannot be erased on another memory such as the RAM 14 The data in the area is temporarily saved and an erasure candidate block is determined.

  In the erasure candidate block, there is data in the non-erasable area of ID: B that needs to be backed up, and subsequently, backup processing of data in the non-erasable area of ID: B is started. (4) The data size of the non-erasable area of ID: B that needs to be backed up is compared with the data size of the writable area, and (5) the data of the non-erasable area of ID: B is made the writable area. make a copy.

  Then, (6) block erasure of the erasure candidate block is performed and the writable area is expanded to the entire erasure candidate block, and (7) data in the non-erasable area of ID: A is copied from the RAM 14 to the writable area. .

  In the above process, the backup processes A to C are described independently. However, by combining these backup processes A to C and repeating the backup process according to the backup processes A to C, any data can be obtained. It is possible to realize the backup process.

  Next, the transition of the data recording state of the NAND flash memory 11 will be described with reference to FIG. FIG. 13 is a diagram schematically showing an example of the transition of the data recording state of the NAND flash memory in the embodiment of the present invention. FIG. 13 shows the structure of the NAND flash memory 11 in units of blocks.

  Data is written to and read from the NAND flash memory 11 for each series of data in units of pages. In writing data, the entire NAND flash memory 11 is used in a ring shape, instead of repeating data erasing / writing only to a specific block every time a write request is made in order to equalize the number of erasures for each block. Then, data is written in each block in order. That is, in the present invention, the position next to the final position of the entire data storage space of the NAND flash memory 11 is used as the head position of the entire data storage space.

  In FIG. 13, for the sake of simplicity, the marker is always located at the top of the page (that is, the header information part including the marker, the total length of the data matches the number of bytes in one page). Although the state is assumed, the present invention is not limited to this assumption, and is a more general condition such as when the information part (information area) including the header information part and the data part does not match the page size. It is also applicable to.

  First, an area in which unused or erased pages are continuous is set as a writable area, and when the nonvolatile storage device is activated, data writing is started from the start position of the writable area. For example, as shown in FIG. 13A, when all areas are activated in an unused state, as shown in FIG. 13B, the first page (first position) of the first block is displayed. Data is written in order starting from (). When the writing of data in the entire area of the block being written is completed, the writing of data to the next consecutive unused block is started as shown in FIG. Similarly, data is sequentially written to the next block.

  Further, for example, when the writable area is insufficient in the state shown in FIG. 13D, data erasure determination is performed on a block other than the block currently being written, and if data erasure is possible, block erasure is performed. To enlarge the writable area as shown in FIG. Then, when data is written up to the last page of the last block as shown in FIG. 13 (f), it returns to the first page, and again as shown in FIG. 13 (g), the first block again. Data is written from the first page. Further, block erasure processing is performed at any time at a determined timing, and for example, the writable area is expanded as shown in FIG.

  Also, at the next startup of the nonvolatile storage device, data is written from the byte next to the previous last write position. For example, when the process ends in the state shown in FIG. 13G, data is written again from the next byte at the last write position of the first block at the next start-up, and continuous data write and block erase are performed again in page order. Repeat the determination.

  As described above, by using the blocks in the order shown in FIG. 13, the entire memory area of the NAND flash memory 11 (the entire area of the data storage space) is equally used to write data. And the number of data write operations can be distributed. Also, by using the blocks in the order shown in FIG. 13, it is not necessary to erase the write data up to the previous time when starting to write data. Data can be saved. In addition, by analyzing the writing position of the NAND flash memory 11 at the time of system startup, it is not necessary to store management information in another area (separate block) of the NAND flash memory 11, There is no need to back up.

  INDUSTRIAL APPLICABILITY The present invention has an effect of improving the memory destruction resistance of a NAND flash memory and efficiently storing data using the memory and backup data. It is applicable to a technique for managing data writing and data reading to and from a non-volatile storage device.

It is a figure which shows an example of a structure of the data management apparatus of the non-volatile storage device in embodiment of this invention. In the embodiment of the present invention, it is a diagram showing an example of memory division of the NAND flash memory. In the embodiment of the present invention, it is a diagram showing an example of an address space in which data to be written to and read from a NAND flash memory is arranged. (A) is a figure which shows an example of the data structure which has a header information part containing a marker in embodiment of this invention, (b) is a specific immediately after a marker in embodiment of this invention. It is a figure which shows an example of the data structure which has the header information part in which the value (value different from a marker) was inserted. It is a flowchart which shows an example of the process at the time of starting of the data management apparatus of the non-volatile storage device in embodiment of this invention. It is a figure which shows an example of the memory management information managed by the data management apparatus of the non-volatile storage device in the embodiment of the present invention. It is a figure which shows an example of the last write page and backup page of NAND type flash memory in embodiment of this invention. It is a figure which shows an example of the data management information managed by the data management apparatus of the non-volatile storage device in the embodiment of the present invention. It is a figure which shows an example of the method of performing block erase, after the data management apparatus of the non-volatile storage device in embodiment of this invention backs up the page of a non-erasable area. It is a figure which shows typically an example of the backup process A of the data management apparatus of the non-volatile storage device in embodiment of this invention. It is a figure which shows typically an example of the backup process B of the data management apparatus of the non-volatile storage device in embodiment of this invention. It is a figure which shows typically an example of the backup process C of the data management apparatus of the non-volatile storage device in embodiment of this invention. In an embodiment of the invention, it is a figure showing typically an example of transition of a data recording state of NAND type flash memory.

Explanation of symbols

11 NAND flash memory 12 CPU
13 Controller 14 Operation RAM

Claims (2)

In a memory data management device that manages writing / reading of data to / from a NAND flash memory,
In the NAND flash memory, a plurality of divided areas in which the storage area is divided in advance for each predetermined capacity are defined as pages, respectively, and the first divided area from the first divided area to the last divided area of the storage area A continuous page number is assigned up to the last page, and by designating this page number, data is written / read out for each page. It is performed in a batch for each block consisting of a predetermined number of pages, and all the recording areas in the erased block are memories that have a predetermined value determined in advance by the erasing process,
Different from the predetermined value determined in advance by the erasing process at the head of the header part including the identification number assigned to identify each different type of data and the data length information of the data assigned the identification number, and the identification A marker set to a value not used for the number is added, and when the same value as the marker value exists in the data, the data is converted so that the values overlap, and converted data is generated, Write data generation means for combining the header portion to which the marker is added and the conversion data to generate write data;
Referring to each page of the NAND flash memory, unused area specifying means for specifying an erased page whose data value stored in each page is the predetermined value as an unused area;
Writing means for writing the data for writing from the start position of the writable area, using the unused area specified by the unused area specifying means as the writable area;
The page number is traced back from the written area immediately before the writable area to search for data with a value that matches the marker value, and the data with the value that matches the marker value was not duplicated. In this case, this data is determined as the marker, and a header part specifying means for specifying the position of the marker as the head position of the header
From the header part specified by the header part specifying means, specify the header part including each identification number assigned to each of the different types of data generated by the write data generation means, and identify each identification number. The storage location of the latest written data of the different types of data to which each identification number is assigned is specified by making the first detected written area the final writing area of the different types of data to which the respective identification numbers are assigned. The latest data identification means,
A memory data management device comprising:   The write data generating means, when adding the marker to the header portion, arranges the predetermined value determined in advance by the erasing process and a specific value different from the marker immediately after the marker, or 2. The memory data management according to claim 1, wherein when adding the marker to the header portion, a value of a head portion of the identification number arranged immediately after the marker is set so as not to coincide with the value of the marker. apparatus.

Images