JP5787095B2 - Method for storing data in non-volatile memory - Google Patents

Description

  The present invention relates to a method for storing data in a nonvolatile memory, and more particularly, to a method for storing data in a memory such as a flash memory that requires erasing before data writing and has a limit on the number of times of erasing.

  The use of rewritable non-volatile memory such as flash memory has expanded as a storage medium for storing data used in electronic devices such as computers or electronic control units (ECU: Electronic Control Unit) for vehicles. Yes.

  In the flash memory which is a nonvolatile memory, the write process for setting the erase state to the write state can be executed in units of addresses or bits on the principle of operation. It is performed in batches (also called blocks) and cannot be executed in address units or bit units.

  Therefore, in a nonvolatile semiconductor memory device, generally, when updating all or a part of data already written in a predetermined memory block, writing to write the updated data into a memory block that has been subjected to batch erase processing After the process is executed, a data update process for executing a batch erase process for batch erasing the update source memory block storing the data before update is executed.

  At this time, a RAM (Random Access Memory) is provided separately from the flash memory, and during the data update process, the data in the memory block is saved in the RAM, which is a data saving area, and the memory block Perform batch erase. A technique for writing data saved in a RAM to the memory block or another memory block is disclosed (see Patent Document 1).

JP 2000-243093 A

  The configuration of Patent Document 1 has a problem that when the data in the memory block is saved in the RAM, the saved data is lost if a situation occurs in which power is not supplied to the RAM due to a power failure or the like. Further, since a RAM is required in addition to the flash memory, there is a problem that the circuit cannot be miniaturized and the manufacturing cost increases.

  Therefore, a configuration using a flash memory instead of the RAM for the data saving area is also conceivable. FIG. 1 shows a configuration example of a nonvolatile semiconductor memory device (hereinafter referred to as ECU) in which the data storage method in the flash memory in this configuration is used. In FIG. 1, only main parts directly related to execution of the data storage method in the flash memory are shown.

  The ECU 1 includes a CPU 2, a ROM 3 that stores programs executed by the CPU 2, a RAM 4 that temporarily stores programs and data, a flash memory 5 that stores various data necessary for the operation of the ECU 1, and an input / output circuit (FIG. 1). In this case, these are connected to each other via a bus 7 so that data transmission is possible.

  Further, the sensor group 10 and the actuator group 11 may be connected to the input / output circuit 6. In such a configuration, the ECU 1 acquires the state of the sensor group 10 and performs drive control of the actuator group 11 based on the state. Data necessary for these controls is stored in the flash memory 5.

  The flash memory 5 includes a plurality (four in FIG. 1) of memory blocks (hereinafter abbreviated as “blocks”) A, B, C, and D, and blocks A to C are used as data storage blocks. Is used as a data saving block. 2 illustrates the block A, the entire area of each block is used as a data storage unit for storing data.

  A method of storing data in the flash memory in the configuration of FIGS. 1 and 2 will be described with reference to FIG. First, Step S200 shows a state in which data is written in all of the data storage units 100 of the block A. In this state, when erasing the old data (A) of the block A, first, the latest data (A) of the block A is written into the data storage unit 400 of the block D as in step S210.

  In general, data in a flash memory is stored in association with a data ID and a data value. When the data already stored in the block A is updated and the updated data is added, the block A stores a plurality of data having the same data ID. However, since the data is written from the top address of the data storage unit 100, it can be determined that the data written at the address closest to the end with the same data ID is the latest data.

  Next, as in step S220, the data storage unit 100 of the block A is erased collectively (that is, all areas are in a state of no data). Next, as in step S230, the latest data (A) of block A saved in block D is written to block A (ie, data storage unit 100). Finally, as in step S240, the block D (that is, the data storage unit 400) is erased collectively.

  FIG. 11 shows the transition of the erase count of each block when the data update process of FIG. 10 is repeatedly executed. In the first state (the state before block erasure), the number of times of erasure of each block is 0, but the data update processing of block B whose data storage amount is full is performed, and block D which is a data saving block The number of erasures increases from 0 to 1. In the second state, the data update process of the block B whose data storage amount is full is performed again, and the number of erasures of the block D, which is a data saving block, increases by one from one. Similarly, in the third state, the data update process of the block A in which the data storage amount is full is performed, and the erase count of the block D is further increased by 1 from 2.

  As described above, since the block D which is a data saving block is shared by the data storage blocks (A to C), the number of erasures is the sum of the number of erasures of the data storage blocks. Therefore, the number of erases of the data saving block reaches the limit value (for example, 500 times) in the specification of the flash memory earlier than the number of erases of the data storage block.

  For this reason, even if the number of times of erasing the data storage block does not reach the limit value, the flash memory cannot be used as a whole, and the ECU 1 is likely not to operate normally, and the ECU 1 itself needs to be replaced. The user who uses this ECU must bear the cost.

  Against the background of the above problems, an object of the present invention is to provide a data storage method in a nonvolatile memory capable of extending the usable period of a storage medium.

Means for Solving the Problems and Effects of the Invention

A method for storing data in a non-volatile memory for solving the above-described problem is a method for storing data in a non-volatile memory in a non-volatile semiconductor storage device including the non-volatile memory, and the non-volatile memory has one or more addresses. And a memory control unit that performs batch erasure in units of memory blocks and writing in units of storage areas or bits of a predetermined number of addresses with respect to individual memory blocks of a non-volatile memory. Each of the plurality of memory blocks includes a data storage unit that stores data, and a data management information storage unit that stores data management information including the number of times of erasure of the memory block that has been collectively erased. One of the memory blocks is used to save data stored in other memory blocks. The memory control block is determined in advance, and the memory control unit leaves only the update data to be stored out of the data stored in the data storage unit and erases data other than the update data. In the data update process, after the update data of the data update source memory block that is the memory block to be updated is saved to the data save memory block , the data update source memory block When the number of erasures is not the minimum and there are a plurality of memory blocks with the minimum number of erasures, a memory block having a smaller storage amount in the data storage unit is determined as a new data saving memory block.

  With the above configuration, the data evacuation memory block is not fixed, so the number of erasures of each memory block can be approximately leveled, the usable period of the nonvolatile memory can be extended, and the product life of the semiconductor memory device can be extended. It is possible to reduce the cost burden on the user.

  The memory control unit in the data storage method in the nonvolatile memory according to the present invention saves the update data of the data update source memory block to the data storage unit of the data save memory block in the data update process, and The data update source memory block erase count is saved in the data management information storage section of the data saving memory block, and then the data updating source memory block is batch erased and the data saved in the data saving memory block An update of the erase count of the update source memory block is written in the data management information storage section of the data update source memory block, and then a new data saving memory block is determined based on the erase count of each memory block.

  With the above configuration, the number of erases of each memory block can be accurately grasped.

  Further, the memory control unit in the method for storing data in the nonvolatile memory according to the present invention determines a memory block having the smallest number of erases as a new data saving memory block in the data update process.

  With the above configuration, the number of erases of each memory block can be approximately leveled.

  The memory control unit in the method for storing data in the nonvolatile memory according to the present invention determines the data update source memory block as a new data saving memory block in the data update process.

  With the above configuration, data is not moved from the data saving memory block to another memory block, so there is no need to erase the data saving memory block at once, and the number of times the data saving memory block is erased increases unnecessarily. Can be prevented.

  In addition, the memory control unit in the method for storing data in the nonvolatile memory according to the present invention determines the data saving memory block as the new data saving memory block in the data update process, and at that time, the data saving memory block The storage contents of the data storage unit are written in the data storage unit of the data update source memory block, and the number of times of erasure of the data saving memory block is written in the data management information storage unit of the data update source memory block, and then the data The data evacuation memory block is obtained by performing batch erasure of the evacuation memory block and further updating the number of erasures of the data evacuation memory block stored in the data management information storage unit of the data update source memory block. Write to the data management information storage unit.

  With the above configuration, the data update process can be performed without increasing the number of times of erasing the memory block that is neither the data update source memory block nor the data saving memory block.

  The memory control unit in the method for storing data in the nonvolatile memory according to the present invention determines a memory block that is neither the data update source memory block nor the data save memory block as a new data save memory block in the data update process. At that time, the storage contents of the data storage unit of the new data saving memory block are written into the data storage unit of the data updating source memory block, and the number of erasures of the new data saving memory block is set to the data updating source memory block. The new data saving memory block is erased at a time, and the new data saving memory block stored in the data management information storing part of the data update source memory block is further erased. The new data save memory block is updated with the erase count. Write to the data management information storage unit.

  With the above configuration, a memory block that is neither a data update source memory block nor a data saving memory block can be used as a new data saving memory block, which leads to equalization of the number of erasures of each memory block.

  The data management information storage unit in the method for storing data in the nonvolatile memory of the present invention stores a data flag for identifying the storage contents of the data storage unit of the memory block, and the memory control unit performs data update processing. The update data is saved to the data save memory block, and the data flag stored in the data update source memory block is also saved to the data management information storage unit of the data save memory block and saved to the data save memory block. When writing the updated data to the data storage unit of another memory block, the data flag saved to the data saving memory block is also written to the data management information storage unit of the other memory block.

  With the above configuration, even if the storage content of the data storage unit moves to another memory block, it is possible to refer to desired data from the external device.

1 is a diagram illustrating a system configuration example of a nonvolatile semiconductor memory device. The figure which shows the structural example of the memory block by a prior art. The figure which shows the structural example of the memory block of this invention. The flowchart explaining a data update process. The figure which shows the state transition of each block in FIG. The figure explaining the state transition of each block following FIG. The figure explaining the state transition of each block following FIG. FIG. 5 is a diagram for explaining another example of a method for determining a new data saving block in FIG. 4. The figure which shows the state transition of each block in the data update process of this invention. The figure explaining the data storage method by a prior art. The figure which shows the state transition of each block in the data update process of a prior art.

  Hereinafter, a method for storing data in a nonvolatile memory according to the present invention will be described with reference to the drawings. Since the nonvolatile semiconductor memory device using the data storage method in the nonvolatile memory is the same as that of the ECU 1 in FIG. 1, only the differences from the conventional configuration will be described here. The CPU 2 corresponds to the memory control unit of the present invention.

  Unlike the conventional configuration, each of the blocks A to D of the flash memory 5 also serves as a data storage block and a data saving block. The memory block has one or more addresses. Further, the CPU 2 performs batch erasure in units of memory blocks and writing in units of storage areas or bits in a predetermined number of addresses with respect to individual memory blocks.

  FIG. 3 illustrates the configuration of a memory block (that is, a block) of the flash memory 5 according to the present invention by exemplifying the block A. The block A does not use the entire storage area as the data storage unit 100 described above, but uses a part thereof as the data management information storage unit 110 that stores data management information. The data management information storage unit 110 includes, as data management information, the number of times of erasing the own block (“erase number” of the present invention) 111, the number of times of erasure 112 of the migration source block that is the block whose data has been transferred to the own block, and the CPU 2. The data flag 113 which can be referred to as an index for identifying the storage contents of the data storage unit of the memory block is stored.

  A data update process according to the present invention will be described with reference to FIGS. This process is included in the program stored in the ROM 3 and is executed by the CPU 2. 4 shows the flow of processing, and FIGS. 5 to 7 show the state of each block corresponding to each step of FIG.

  In the initial state of FIG. 4 or FIG. 5 (step S10), the blocks A to C are storage blocks, and the block D is a save block (data save memory block of the present invention). In addition, the number of erasures (111, 211, 311 and 411) of the own block of each block is α, β, γ and δ, respectively. In addition, the number of erasures (112, 212, 312, 412) of the migration source block of each block is all “no data (ie, data is erased)”. The data flags (113, 213, 313, and 413) of each block are A, B, C, and D, respectively, and the actual address and data content of each block match.

  First, for example, when the data storage amount of the data storage unit 100 of the block A becomes full, the block A is set as the migration source block (S10). Next, the latest data (update data of the present invention) (A), the number of times the block has been erased (α), and the data flag (A), which are stored contents of the data management information storage unit 110 of the block A, are saved. The data is copied to the corresponding storage unit (400 or 410) of the block for use (block D in the example of FIG. 5) (S11). The erase count (α) of the block A's own block is stored in the erase count 412 of the transfer source block of the block D. Next, the block A data is erased at once (S12). At this time, all stored contents of the data storage unit 100 and the data management information storage unit 110 are erased.

  Next, erasure of the own block in the data management information storage unit 110 of block A is made by adding 1 to the erase count (412) of the migration source block (that is, block A) copied to block D (α + 1). Copy the number of times (111) (S13).

  At this time, the latest data of the block A and the information of the data management unit are transferred to the block D. When the CPU 2 tries to refer to the latest data (A), first, the data flag of each block is referred to. The latest data (A) can be read from the block D in which is “A”.

  Next, a new data saving block (sometimes abbreviated as “new saving block”) which is the next saving block is determined from the blocks A to D (S14). For example, among the blocks, the block with the smallest erase count (111, 211, 311 and 411) is set as a new save block.

  Here, when the number of times of erasure of the block A which is the migration source block is the smallest (S15: Yes), the block A is determined as a new saving block (S16). Then, this process ends with each block maintaining the current state. At this time, the state of each block is as shown in step S13 of FIG.

  On the other hand, when the number of erasures of the block A that is the migration source block is not the smallest, that is, when a block other than the migration source block is determined as a new save block (S15: No), the block becomes a new save block. Before determining, it is determined whether or not it is a storage block. Hereinafter, the state of each block will be described with reference to FIG. In the example of FIG. 6, this corresponds to the case where the block B or the block C is determined as a new saving block, but here it is assumed that the block B is determined as a new saving block.

  Since the block B determined as the new saving block is a storage block before the determination (S17: Yes), the latest data (B) of the block B, the erase count (β) of the own block, and the data flag (B) are stored. The data is copied to block A, which is the previous migration source block (S18). The erase count (β) of the block B's own block is stored in the erase count 112 of the migration source block of the block A. Here, when the CPU 2 tries to refer to the latest data (B), the latest data (B) is read from the block A whose data flag is “B”.

  Next, the data in block B, which is a new saving block, is erased all at once (S19). At this time, all stored contents of the data storage unit 200 and the data management unit 210 are erased.

  Next, the value obtained by adding 1 to the content (β) of the erase count 112 of the migration source block (ie, block B) of block A is copied to the erase count 211 of the block B's own block (S20). Then, this process ends.

  Hereinafter, the state of each block will be described with reference to FIG. When the new save block is not a storage block before the determination, that is, when the new save block is determined to be the old save block D as shown in FIG. 7 (S17: No), the block is stored in the block D. The latest data (A), the number of erasures of the current block (δ), and the data flag (A) are copied to the block A which is the previous transfer source block (S21). The erase count (δ) of the block D's own block is stored in the erase count 112 of the transfer source block of the block A. That is, the name of the block matches the stored content.

  Next, the data of block D is erased at once (S22). At this time, all stored contents of the data storage unit 400 and the data management unit 410 are erased.

  Next, the value obtained by adding 1 to the content (δ) of the erase count 112 of the migration source block of block A is copied to the erase count 411 of the own block of block D (S23). Then, this process ends.

  FIG. 8 shows another example of step S14 in FIG. When the number of times of erasure of the own block of each memory block (111, 211, 311, 411) is compared, the number of times of erasure of the own block of the migration source block (that is, the block subjected to batch erasure) is minimum (S141: Yes) ), The migration source block is determined as a new saving block (S148).

  On the other hand, when the erase count of the own block of the migration source block is not the minimum (S141: No), it is checked whether there are a plurality of blocks having the minimum erase count of the own block among other blocks. As a result, when there are not a plurality of blocks with the smallest number of erases of the own block (S142: No), the block with the smallest number of erases of the own block is determined as a candidate block that is a candidate for a new saving block (S147). ).

On the other hand, when there are a plurality of blocks in which the number of times of erasure of the own block is the minimum (S142: Yes), for example, a block having a higher priority is determined as a candidate block based on a predetermined priority as exemplified below ( S143).
-The priority is set higher for blocks with less recent data. Since the amount of data movement is small, the processing time can be shortened. In this configuration, when the number of times of erasure of the migration source memory block is not minimum and there are a plurality of memory blocks with the minimum number of times of erasure, a memory block having a smaller storage amount in the data storage unit is assigned as a new data saving memory block. It will be decided.

In the flash memory 5, the priority is set higher for blocks located closer to the migration source block. Since the data movement distance is short, the processing time can be shortened.
A higher priority is assigned to a block whose block name and data flag (113, 213, 313, 413) can be matched. That is, the data flag written to the block can be the same as the block name. Alternatively, when data is moved from the block, the name of the block to be moved and the data flag of the data to be moved can be made the same.

  Next, the difference between the determined number of erasures of the own block of the candidate block and the number of erasures of the own block of the migration source block is calculated (S144), and the difference is a prescribed value (for example, a fixed value such as 2 or For example, it is checked whether or not the value is smaller than a predetermined ratio of the number of writable times of the memory block, such as 1%.

  When the value is below the specified value (S145: Yes), the migration source block is determined as a new saving block (S148). On the other hand, when the value exceeds the specified value (S145: No), the candidate block is determined as a new saving block (S146).

  In the above configuration, when the current number of times of erasure of the migration source memory block is not the minimum, when the difference between the number of erasures and the memory block with the smallest number of erasures is less than a predetermined value, the migration source memory block Is determined as a new data saving memory block, and when the value exceeds a predetermined value, the memory block having the smallest erase count is determined as the new data saving memory block. For example, if the difference in the number of times of erasure is about several times, the difference is reduced by the subsequent data update process, so there is no need to increase the number of times of erasure here, and the processing time of the data update process can be shortened. The effect that leads to the saving of the number of block erases can be expected.

  FIG. 9 shows the state transition of each block when the process of FIG. 4 is executed a predetermined number of times (for example, 500 times). In the first state, the erase count of all blocks is zero. In this state, when the data storage amount of the block B becomes full, the latest data (including the data flag) of the block B is copied to the block D that is the saving block at this time. Then, the data in block B is erased at once. The number of erases is 0 except for block B, but the number of erases of block B is 1, and the number of erases of block D is minimum, or the difference in the number of erases between block B and block D is less than the specified value. The data of block B is still copied to block D, and block B is set as a new saving block (second state).

  In the second state, when the data storage amount of the block D becomes full, the latest data (including the data flag) of the block D is transferred to the block B which is a saving block. Then, the data in block D is erased at once. At this time, the block A with the smallest erase count is set as a new save block (third state). Then, the latest data (including the data flag) of the block A is copied to the block D, and the data of the block A is erased collectively. After block A becomes a new saving block, data (latest data) is newly written in each of block B to block D, and the data storage amount of block C is full.

  In the state where the above process is repeated and the 501st process is performed (block A data is saved in block D and block A is used as a new save block), the number of erases of a specific block is different from FIG. It can avoid protruding and becoming large.

  In addition to the flash memory, it includes a plurality of memory blocks having one or more addresses. For each memory block, batch erasure in units of memory blocks, writing in units of storage areas or bits in a predetermined number of addresses Applicable to storage media to be performed.

  Although the embodiments of the present invention have been described above, these are merely examples, and the present invention is not limited to these embodiments, and the knowledge of those skilled in the art can be used without departing from the spirit of the claims. Various modifications based on this are possible.

1 ECU (nonvolatile semiconductor memory device)
2 CPU (memory control unit)
5 Flash memory A to D Memory block (block, data storage block, data saving block)
100 Data storage unit 110 Data management information storage unit 111 Number of erases of own block 112 Number of erases of source block 113 Data flag

Claims (5)

A method for storing data in a non-volatile memory in a non-volatile semiconductor memory device comprising a non-volatile memory,
The nonvolatile memory includes a plurality of memory blocks having one or more addresses,
A memory control unit that performs batch erasure in units of memory blocks, writing in units of storage areas of a predetermined number of addresses or bits in units of each of the memory blocks of the nonvolatile memory,
Each of the plurality of memory blocks includes a data storage unit that stores data, and a data management information storage unit that stores data management information including the number of times of erasure of the memory block that has been collectively erased. ,
One of the plurality of memory blocks is predetermined as a data saving memory block for saving data stored in another memory block,
The memory control unit performs only a data update process for erasing data other than the update data, leaving only the update data to be stored among the data stored in the data storage unit,
In the data update process, after the update data of the data update source memory block, which is a memory block that is a target of data update, is saved to the data save memory block ,
When the number of erasures of the data update source memory block is not minimum and there are a plurality of memory blocks with the minimum number of erasures, a memory block having a smaller storage amount in the data storage unit is used as a new data saving memory block. A method of storing data in a non-volatile memory, characterized by determining. In the data update process, the memory control unit
The update data of the data update source memory block is saved to the data storage unit of the data save memory block, and the number of erasures of the data update source memory block is set to the data management information of the data save memory block. After saving to the storage unit, the data update source memory block is collectively erased,
Writing the updated number of erasures of the data update source memory block saved in the data save memory block into the data management information storage unit of the data update source memory block;
2. The method of storing data in the nonvolatile memory according to claim 1 , wherein the new data saving memory block is determined thereafter . When the memory controller determines the data saving memory block as the new data saving memory block in the data update process,
The storage contents of the data storage unit of the data saving memory block are written to the data storage unit of the data updating source memory block, and the number of erasures of the data saving memory block is stored in the data management information storage of the data updating source memory block After writing to the block, erase the data saving memory block at once,
Furthermore, the updated data erasure count of the data saving memory block stored in the data management information storage unit of the data update source memory block is written to the data management information storage unit of the data saving memory block. A method for storing data in a nonvolatile memory according to claim 1 or 2. In the data update process, the memory control unit determines a memory block that is neither the data update source memory block nor the data save memory block as the new data save memory block.
The storage contents of the data storage unit of the new data saving memory block are written to the data storage unit of the data updating source memory block, and the number of erasures of the new data saving memory block is set to the data of the data updating source memory block. After writing to the management information storage unit, perform batch erasure of the new data saving memory block,
Further, an updated number of erasures of the new data saving memory block stored in the data management information storage unit of the data update source memory block is updated in the data management information storage unit of the new data saving memory block. The method for storing data in the nonvolatile memory according to any one of claims 1 to 3, wherein writing is performed . The data management information storage unit stores a data flag for identifying the storage contents of the data storage unit of the memory block,
In the data update process, the memory control unit saves the update data to the data save memory block, and the data flag stored in the data update source memory block also includes the data flag of the data save memory block. Save to the data management information storage unit,
When writing the update data saved in the data saving memory block to the data storage unit of another memory block, the data flag saved in the data saving memory block is also stored in the data management information of the other memory block. The method for storing data in the nonvolatile memory according to claim 1 , wherein the data is written in the unit .

Images