JPH09293025A - Method for controlling memory access and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To guarantee the reliability of data which is read from a non-volatile memory after the lowering of a power source voltage and power interruption even when they occur by copying data management information stored in the non-volatile memory in a volatile memory before the usage of the non-volatile memory is started. SOLUTION: When a power source is turned on, the contents of a physical block flag group which constitutes data management information preserved inside a user data memory (flash memory) 200 and the contents of block mapping information are copied in a work memory 400 as a pre-processing for the usage start of the user data memory 200 and they are stored as the initial value of the physical block state flag group for displaying a latest state and as the initial value of block mapping information for display the latest state. At this time, one more group is copied and preserved in the work memory 400 concerning the contents of the physical block state flag group and stored as the physical block state flag group for display an initial state.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a non-volatile memory having a limited access frequency due to the nature of a memory element (for example
The present invention relates to a memory access control method and device suitable for flash memory, etc., and more particularly to a memory access control method and device that contributes to extending the life of such a non-volatile memory while maintaining data reliability when a power failure occurs.

[0002]

2. Description of the Related Art It is known that a flash memory, which is a non-volatile memory, has a limited number of accesses due to the electrical characteristics of its storage element. In particular, it is used in a page memory method in which the flash memory storage area is formatted into a data storage area consisting of a plurality of physical blocks and a data management information storage area for managing the usage status of the data storage area. Is more frequently accessed in the data management area than in the data storage area,
There is a problem that the memory life is short due to the consumption of the storage elements forming the data management area.

As a measure against such a problem, prior to the use of the flash memory, the contents of the data management information storage area of the flash memory are composed of a work memory (a semiconductor having a normal structure) whose access frequency is practically unlimited. It is usually composed of a volatile memory such as SRAM, DRAM, etc.), and while the memory is in use, the data management information on this work memory is accessed to update it sequentially, An intermittent rewriting method is conceivable in which the contents of the data management information on the flash memory are rewritten with the contents of the data management information on the work memory intermittently after the use is completed or at an appropriate time during the use.

According to the intermittent rewriting method of such data management information, although data writing, reading, erasing, etc. are performed by direct access to the data storage area on the flash memory, the high level of data management information accompanying it. Since the frequent sequential update is performed by an indirect access, that is, via the access to the management information storage area on the work memory, the frequency of access to the data management information storage area on the flash memory is reduced and the life of the flash memory is shortened. Increase.

[0005]

However, in the above-mentioned intermittent rewriting method, although the contents of the data in the flash memory and the contents of the data management information in the work memory are sequentially updated, The content of the data management information in the memory is updated only intermittently.Therefore, if the content of the data management information in the work memory is lost due to power failure or voltage drop of the system power supply voltage during that time, the flash memory In the above, there is a problem in that the content of the data and the content of the data management information are left inconsistent and the flash memory becomes unusable. In particular, this problem becomes remarkable when the flash memory is used as a user data storage memory of a small portable electronic device such as a PDA. That is, in this kind of small portable electronic device, it is difficult to mount a large-capacity battery due to size and weight restrictions, so there is a high possibility that the battery will run out during use, and it will be mistakenly used during use. There is a high possibility that a power failure may occur due to poor contact of the battery due to the impact when dropped. If there is a mismatch between the data and the data management information in the flash memory each time the battery runs out or a power failure occurs, and the reliability of the read data is lost, it cannot be put to practical use as a product.

The present invention has been made in view of the above problems. An object of the present invention is to provide a non-volatile memory having a limited access count due to the nature of a storage element, based on the specified access count. (EN) Provided is a memory access method and device which can be utilized to the maximum extent and can guarantee the reliability of data read from the nonvolatile memory after that even if the power supply voltage drops or a power failure occurs. It is in.

[0007]

The invention according to claim 1 of this application is a method for controlling memory access to a non-volatile memory having a limited number of accesses due to the nature of a storage element, and Prior to the start of use of the non-volatile memory, the data management information stored in the non-volatile memory is copied onto a volatile memory for which the number of accesses is practically unlimited, and used as the data management information for displaying the latest status. Of the physical block status flag group and the block mapping information forming the data management information, a set of physical block status flag groups is separately copied and stored on the volatile memory to display an initial state physical block status flag group. The pre-processing step to determine whether the given memory access request is an erase request or a write request. Request type determining step, an access target physical block searching step for obtaining a physical block mapped to a logical block specified by the memory access request based on the stored block mapping information, and the saved initial state A block use state determination step of determining that the searched access target physical block is in use by referring to a display physical block state flag group, the physical block state flag group for displaying the initial state, and By referring to the physical block state flag group for displaying the latest state, a free physical block search step of searching for a physical block that is in an empty state in the initial state and is also in the latest state, and the request type determination step The memory access request is determined to be an erase request and When the access target physical block searched in the access target physical block search step is determined to be in use in the block usage status determination step, the latest state display physical block status flag of the access target physical block is not used. An erase request execution step of changing, erasing processing for the access target physical block of the non-volatile memory, and separating the block mapping relating to the access target physical block in the block mapping information; and an access request by the request type determination step. Is a write request and the access target physical block searched in the access target physical block searching step is judged to be in use in the block usage status judging step, The physical block status flag for the latest status display of the target physical block is changed to not in use, and the write processing to the relevant physical block in the nonvolatile memory is avoided, and instead, the data related to the write request is changed. A write request execution step of writing to the free physical block searched in the free physical block search step in the nonvolatile memory, and separating the block mapping relating to the physical block to be accessed in the block mapping information and re-establishing the new block mapping In response to the arrival of a predetermined data management information update request, the contents of the data management information in the non-volatile memory are changed to the contents of the data management information for displaying the latest state stored in the volatile memory. Data management information update step to update, In the memory access control method characterized by Bei.

The invention according to claim 2 of this application is an apparatus for controlling memory access to a non-volatile memory having a limited number of accesses due to the nature of a memory element, and using the non-volatile memory. Prior to the start
Data management information stored in the non-volatile memory,
Data management information for displaying the latest status is obtained by copying on a volatile memory in which the number of accesses is not practically limited. Further, among the physical block status flag group and the block mapping information that constitute the data management information, physical Preprocessing means for separately storing one set of block state flag groups on the volatile memory to form a physical block state flag group for initial state display, and the given memory access request is either an erase request or a write request. And an access target physical block search unit for determining a physical block mapped to a logical block specified by the memory access request, based on the stored block mapping information, Referring to the stored physical block state flag group for initial state display, Referring to the block use state determination means for determining that the searched access target physical block is in use, the initial state display physical block state flag group and the latest state display physical block state flag group And a free physical block search means for searching a physical block that is empty in the initial state and free even in the latest state, and the memory access request is determined to be an erase request by the request type determination means and the access When the access target physical block searched by the target physical block search means is determined to be in use by the block usage status determination means, the latest state display physical block status flag of the access target physical block is changed to unused. , For the physical block to be accessed in the nonvolatile memory To avoid processed,
In addition, in the block mapping information, the erase request execution unit that separates the block mapping relating to the physical block to be accessed, and the request type determination unit determine that the access request is a write request, and the physical block search unit to be accessed searches the access request. When the access target physical block is determined to be in use by the block use state determination means, the latest state display physical block state flag of the access target physical block is changed to not in use, and the access in the nonvolatile memory is performed. Avoid writing to the target physical block,
Instead, the data related to the write request is written in the free physical block searched by the free physical block search means in the nonvolatile memory, and the block mapping relating to the physical block to be accessed is separated in the block mapping information. In response to the arrival of a predetermined data management information update request, the write request executing means for newly re-establishing the block mapping,
Data management information updating means for updating the content of the data management information in the non-volatile memory with the content of the data management information for displaying the latest status stored in the volatile memory. It is in the access control device.

Here, in the invention of claim 1 or claim 2, the "property of the memory element" is a physical or chemical property or the like that causes the life of the element to be shortened as the number of accesses increases. There are, in particular, those caused by various causes such as the charge accumulation form based on the semiconductor structure and the resistance of the wiring material.

As the activation condition of the "pre-processing step", various forms such as power-on, arrival of an activation command from an application or BIOS, etc. may be adopted. From the nature of the present invention, it is sufficient that the “data management information” includes at least a physical block status flag group and block mapping information, but the scope of the present invention is other than that. It does not proactively exclude the inclusion of information. In addition, the physical block status flag group for initial status display is intentionally stored and saved in a volatile memory (so-called system memory) with unlimited access count.
By reading the contents of this stored flag group,
This is so that the initial state can be immediately confirmed without actually reading the contents of the nonvolatile memory whose access frequency is limited each time. In other words, if the non-volatile memory is accessed each time to confirm the initial state, the life of the non-volatile memory (which is remarkable in the case of the flash memory, for example) can be shortened by the access for reading the non-volatile memory. It is because The "physical block status flag group" is a set of individual status flags that indicate whether each physical block in the data storage area of the non-volatile memory is in a used state or an unused state. I mean. Also, "initial state"
Means the state at the start of use of the memory, or the state at the time of updating the latest data management information after the start of use.

Further, the meaning of "discrimination of access request type" is to anticipate that this may cause a possibility of data modification in any physical block in the data storage area of the non-volatile memory (for example, flash memory). It is in. It is to be understood by those skilled in the art that "is either" includes an action of recognizing whether the request is an erase request or a write request from the relationship with execution of an erase request or write request described later. Then it will be easily understood.

[Search for physical block to be accessed]
The block mapping information used in the above is information that defines the coupling relationship between the logical block specified by the access request and the physical block in the non-volatile memory, and can be configured by, for example, a table that defines the relationship between the two. . This connection relationship can be arbitrarily mapped or unmapped (separated).

Further, as described above, the "block use state determination" is performed by referring to the physical block state flag group for initial state display stored in the volatile memory, so that the contents of the non-volatile memory can be determined. There is no fear of memory consumption by referring to.

Further, the meaning of "search for free physical block" is to search for a physical block that is in an empty state both in the initial state and in the latest state. It will be easily understood by those skilled in the art that it will have a function of automatically selecting one of them.

The meaning of "erase request execution" means that when the physical block mapped to the logical block related to the erase request is actually in use, the data of the physical block is not actually erased, Virtually erasing only on the physical block status flag on the volatile memory. As a result, even if an erasing request arrives, inconsistency does not occur between the data and the data management information on the nonvolatile memory. Moreover, at that time, the mapping between the logical block and the physical block is separated, so that the read request for the logical block does not occur and the data that should have been erased is not read by mistake.

Further, the meaning of "write request execution" means that when the physical block mapped to the logical block related to the write request is actually in use, the data of the physical block is not actually overwritten and erased. , While virtually overwriting and erasing only on the physical block status flag on the volatile memory,
It is to write to a physical block that is vacant in both the initial state and the latest state. As a result, even if a write request arrives, there is no inconsistency between the data and the data management information on the nonvolatile memory. Moreover, at that time, the mapping between the logical block and the physical block up to that point is separated, but a new mapping is set between the actually written physical block and the logical block. When the read request for the block arrives, the overwritten data is normally read, and the data that should have been overwritten and erased is not erroneously read.

Further, the meaning of "data management information update" is to match the contents of the data management information of the non-volatile memory with the latest state of the data storage area. Here, the data management information update request may be given from an application program, or means (eg, BIO) hierarchically lower than the application program.
You may make it give from S etc.). Regarding the timing of giving the data management information update request, it may be given cyclically at a fixed cycle, given at any timing from an application program or BIOS, or given at any timing by a command from an operator. Various aspects can be taken.
Furthermore, it is conceivable that the application cycle of the data management information update request may be appropriately determined according to the usage mode (eg, writing frequency or erasing frequency) of the electronic device to which the request is applied, the expected power failure frequency, and the like.

It should be understood by those skilled in the art that, in the description of the above claims, the case where an erase request or a write request for the newly mapped logical block after the next time arrives as a result of the execution of the write request described above. Not stipulated. In addition, as a reminder, in this case, it is clear that the state of the data storage area in the initial state is not changed, so it is considered that the normal erasing process or writing process may be performed. The same is true when a write request for a new logical block that has not been mapped has arrived, and in this case as well, the newly mapped physical block changes the state of the data storage area in the initial state. Since it is clear that there is no such a case, it is considered that the normal writing process should be performed. Furthermore, even if a read request for any of the logical blocks arrives, it is clear that the state of the data storage area in the initial state does not change, and therefore it is considered that the normal write processing should be performed. .

According to the first or second aspect of the invention, the write process, the read process, and the erase process can be performed on the logical block without any trouble, while the previous process can be performed on the physical block. The contents at the time of updating the data management status are retained as they are until the next data management status update, and the consistency between the data and the data management information is maintained. Although the stored data will be lost after that, the data already existing at the time of the previous update can be read normally, and the reliability of the read data from the non-volatile memory is guaranteed while minimizing the loss of data. it can. Moreover, while the data management information in the volatile memory (work memory) is updated sequentially, the data management information in the non-volatile memory (flash memory, etc.) is updated only intermittently. It is possible to make maximum use of the non-volatile memory having a limited access count under the specified access count.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. As described above, the effect of the memory access control method and device according to the present invention is considered to be remarkable when a flash memory is used as a user data storage memory of a small portable electronic device such as a PDA. . In other words, in this type of small portable electronic device, it is difficult to mount a large capacity battery due to size and weight restrictions, and there is a high possibility that the battery will run out during use, and it will accidentally fall during use. There is also a high possibility that a power failure will occur due to poor battery contact due to shock or the like. If there is a mismatch between the data and the data management information in the flash memory each time the battery runs out or a power failure occurs, and the reliability of the read data is lost, it cannot be put to practical use as a product. The application of the present invention solves this problem as follows.

The hardware configuration of a small portable electronic device to which the present invention is applied will be schematically described with reference to the block diagram of FIG. As shown in the figure, this type of small portable electronic device has a man-machine interface section (in recent examples, a pen-touch type input pad is often used for exchanging data with a user. 100)
A user data memory 200 for storing user data such as a telephone directory, a schedule, etc., a system program memory 300 for storing a system program for realizing various operations of the device, and a system program memory A work memory 400 used as a work area at the time of execution, and a CPU 500 for realizing the operation function of the device are roughly configured, and these device components are well-known such as an address bus, a data bus, System bus 60 consisting of control bus, etc.
Connected to each other by 0. Since the basic configuration and operation of this type of stored program control device are well known to those skilled in the art, further description will be omitted.

As the user data memory 200, a rewritable flash memory, which is a non-volatile memory whose number of accesses is limited due to the nature of the storage element, is used, and even if the power switch is turned off by this. The user data is saved. On the other hand, the work memory 400 is a rewritable normal semiconductor structure DRA that is a volatile element.
M and SRAM are used. Therefore, in the state where the power switch is turned on and some operation is performed, if a power failure occurs due to a battery connection failure or a battery connection failure due to a drop impact, the work memory 4
The stored contents of 00 may be lost.

As shown in FIG. 6, in the storage area of the user data memory 200, physical block groups B01 to B0 are included.
A storage area forming n and a physical block status flag group F
The storage area for storing 01 to F0n and the storage area for storing the block mapping information MAP0 are formatted in advance.

The individual physical blocks forming the physical block groups B01 to B0n are for storing the individual logical blocks forming the user data, and their sizes are the same, for example, and the logical blocks of the logical blocks are the same. The size is the same. Then, the respective logical blocks of the user data that sequentially arrive with the access request are physical block groups B01 to B0, as will be described later in detail.
It will be allocated and stored in any of n.

Physical block status flag group F01 to F0n
The individual physical block status flags that make up each of the physical block groups have a one-to-one correspondence with the individual physical blocks that make up the physical block groups B01 to B0n, and each physical block is in use or not in use. Is displayed.
Since the physical block state flag groups F01 to F0n are non-volatile, the physical block group B01 to F0n immediately after the power is turned on.
The use status of B0n is confirmed by referring to the contents of the physical block status flag groups F01 to F0n. The block mapping information MAP0 is information indicating a connection relationship between physical blocks and logical blocks, that is, which logical block is stored in each physical block forming the physical block groups B01 to B0n. Since this block mapping information MAP0 is also non-volatile, the physical block groups B01 to B0n immediately after the power is turned on.
The storage block confirmation (which logical block is stored) is performed by referring to the block mapping information MAP0.

In the following description, the physical block status flag groups F01 to F0n and the block mapping information MAP0 are collectively referred to as "data management information".

The reference (reading) and updating (reading and writing) of the data management information are performed by batch reading or writing processing for a series of storage areas in which the data management information is stored, because of necessity of bit processing and the like. Therefore, the physical block status flag groups F01 to F
0n and the access frequency to the data management information storage area in which the block mapping information MAP0 is stored are significantly higher than the access frequency to the data storage areas configuring the physical block groups B01 to B0n. Therefore, if the flash memory that constitutes the user data memory 200 is directly accessed each time to refer to and update the data management information, the data management information storage area is heavily consumed, and the data management information is stored before the data storage area. Only the data management area reaches its life, and as a result, the life of the entire flash memory is shortened.

Therefore, according to the present invention, as described below, the data management information in the user data memory 200 is copied to the work memory 400, and in order to refer to and update the data management information sequentially, While accessing the data management information in the copied work memory 400,
By adopting the method of intermittently updating the data management information in the user data memory 200 with the content of the data management information in the work memory 400 at an appropriate timing, the problem of the life reduction of the flash memory is solved. ing. In addition, while erasing or rewriting the contents of the data storage area of the user data memory 200 while accessing the data management information in the work memory 400, between the data and the data management information in the user data memory 200. Even if an unexpected power failure occurs, the user data memory 200 can be continuously used even if an unexpected power failure occurs. are doing.

That is, as shown in the general flow chart of FIG. 2 showing the software configuration (configuration of the system program) of the whole of the present invention, when the power is turned on for the use of this device, the user data memory 200 is displayed.
As a pre-process prior to the start of use of the physical block status flag group F01 to which the data management information stored in the user data memory (flash memory) 200 is configured.
The contents of F0n and the block mapping information MAP0 are as shown in FIG.
The data is copied in and stored as the initial values of the physical block status flag groups F11 to F1n for displaying the latest status and as the initial values of the block mapping information MAP1 for displaying the latest status (steps 201 and 202). At this time, with respect to the contents of the physical block state flag groups F01 to F0n, another set is further copied and stored in the work memory 400 and stored as the physical block state flag groups F21 to F2n for displaying the initial state ( Step 20
3). The physical block state flag groups F21 to F2n for displaying the initial state are provided to avoid consumption of the user data memory 200 due to direct access to the user data memory (flash memory) 200 for confirmation of the initial state. This is because.

Thereafter, the present apparatus enters a state of waiting for the arrival of the access request and the arrival of the data management information update request (steps 204NO, 205NO). In this state, if any access request arrives (YES in step 204), the request type is determined (step 206), and the determination result is an erase request, a write request,
Depending on whether it is a read request, an erase request execution process (step 207), a write request execution process (step 208), or a read request execution process (step 20).
Either of 9) is executed. Thereby, while maintaining the consistency between the data in the user memory 200 and the data management information, the data management information in the work memory 400 is sequentially accessed to erase, write, and
The reading process will be performed. During this period, for example, when a data management information update request issued by the application program arrives (YES in step 205), FIG.
As shown in (b), the data management information (physical block status flag groups F11 to F1n and block mapping information MAP1) in the work memory 400, which has been sequentially updated, is stored in the user data memory 200. Data management information (physical block status flag group F01-
F0n and the block mapping information MAP0) are rewritten (steps 210 and 211), thereby completing the actual updating process of the data management information in the user data memory 200. After that, the contents of the physical block status flag groups F01 to F0n immediately after the update are copied again into the work memory 400 (step 203), and a new physical block status flag group F21 for initial state display is displayed.
Stored as F2n.

According to the configuration of the present invention shown in the general flow chart of FIG. 2, the data management information in the work memory 400 is sequentially updated each time data is accessed, whereas the user data memory (flash memory) is updated. Since the data management information in the memory 200 is updated only intermittently in response to the data management information update request, the access frequency to the flash memory configuring the user data memory 200 is reduced by that much, and the extent of wear is reduced. Is also reduced, which contributes to an increase in the life of the flash memory.

Next, erase processing, write processing, and
The contents of each process will be described in detail while paying attention to the fact that the consistency between the data and the data management information is maintained in the user data memory 200 while the read process is normally performed.

Details of the erase request execution process (step 207) will be described with reference to the flowchart of FIG. In the figure, when the process is started, the physical block B0k mapped to the logical block designated by the erase request is searched based on the block mapping information MAP1 in the work memory 400 (step 301). here,
When the physical block B0k mapped to the logical block designated by the erase request is not found (NO in step 302), a predetermined error process is performed. On the other hand, when such a physical block B0k is found (YES in step 302), then the physical block state flag group F21 to F21 for displaying the initial state in the work memory 400 is displayed.
By referring to F2n, the usage status of the searched physical block B0k in the initial state (the state when the latest data management information was updated) is checked (step 3).
03).

Here, if it is not in use, the data stored in the physical block B0k has been accumulated since the last update of the data management information and did not exist at the last update of the data management information. Even if the data management information cannot be updated due to a power failure after the data is erased, the data (B01 to B0n) and the data management information (on the user data memory 200) and the data management information ( There is no possibility of inconsistency between F01 to F0n and MAP0). Therefore, when it is determined that the physical block B0k is not being used (step 304), the data of the physical block B0k is erased (step 305), and the physical block state flag F1k for displaying the latest state corresponding to the physical block B0k is not set. It is changed during use (step 306), the mapping of the physical block B0k in the block mapping information MAP1 is separated to make the physical block inaccessible (step 307), and then the process ends. By doing so, even if the data management information cannot be updated due to a power failure, the data that did not originally exist at the time of the previous update is erased. Therefore, the user data memory 200
There is no risk of inconsistency between the data (B01 to B0n) and the data management information (F01 to F0n, MAP0). After that, when a data management information update request arrives, the contents of the data management information (F11 to F1n, MAP1) on the work memory match the data contents (B01 to B0n) on the data storage area. Therefore, it is possible to update the data management information (F01 to F0n, MAP0) in the user data memory 200 while making the most of the data sequential update results up to that point.

On the other hand, if it is in use, it means that the data stored in the physical block B0k already existed at the time of the last update of the data management information, and after this is erased. If it is impossible to update the data management information due to a power failure, there is an inconsistency between the data (B01 to B0n) and the data management information (F01 to F0n, MAP0) on the user data memory 200. May occur. Therefore, when it is determined that the physical block B0k is in use (YES in step 304), the physical block state flag F1k for displaying the latest state corresponding to the physical block B0k is not used without erasing the data. Change to (Step 30
6) Further, after the mapping of the physical block B0k in the block mapping information MAP1 is separated (step 307), the processing is ended. In this way, if a power failure occurs and the data management information (F01-F01)
0n, MAP0) cannot be updated, the data existing at the time of the previous update is not erased, and therefore the data (B01 to B0n) and the data management information (F01 to F0) on the user data memory 200.
There is no possibility of inconsistency with the (n, MAP0). On the other hand, although the data of the physical block B0k is stored without being erased, since the physical block is accessed based on the block mapping information MAP1 in the work memory 400, the data that should have been erased is erroneously read. There is no danger of being lost. Incidentally, when a data management information update request arrives thereafter, the mapping information MAP0 relating to the physical block B0k is also separated on the user data memory 200, and the physical block status flag F0k is not used. After that, when a write request for a new logical block arrives, the physical block B0k can be used as a candidate for a free physical block for storing the new block, although the previous data is stored. .

Next, a write request execution process (step 2)
08) will be described in detail with reference to the flowchart of FIG. In the figure, when the process is started, the physical block B0k mapped to the logical block designated by the write request is searched based on the block mapping information MAP1 in the work memory 400 (step 4).
01). If such a physical block is found here (YES in step 402), then the work memory 40
Physical block status flag group F2 for displaying initial status in 0
By referring to 1 to F2n, the usage status of the searched physical block B0k in the initial state (the state when the latest data management information was updated) is checked (step 403).

Here, if it is not in use, it means that the data stored in the physical block B0k did not exist at the time of the last update of the data management information. Even if the data management information (F01 to F0n, MAP0) cannot be updated due to a power failure, the data (B01 to B0n) and the data management information (F0) are stored in the user data memory 200.
01-F0n, MAP0) does not cause a mismatch. Therefore, if it is determined that the physical block B0k is not being used (NO in step 404), the data of the physical block B0k is overwritten with new data (step 40).
5) Then, the physical block status flag F1k for displaying the latest status corresponding to the physical block B0k is changed to “in use” (step 406), and the process is ended. In this way, if a power failure occurs and the data management information (F01 to F0
(n, MAP0) cannot be updated, the data that originally did not exist at the time of the previous update was overwritten and erased. Therefore, the data (B01 to B0n) and the data management information (on the user data memory 200) are deleted. F01 ~
There is no risk of inconsistency between F0n and MAP0).
On the other hand, when a data management information update request arrives thereafter, the data management information (F11 to F1
Since the contents of (n, MAP1) match the data contents (B01 to B0n) in the data storage area, the data management information (F01 ~ F0
n, MAP0) can be updated.

On the other hand, if it is in use, it means that the data stored in the physical block B0k already existed at the time of the previous update of the data management information, and after overwriting and erasing this data. If it is impossible to update the data management information due to a power failure, the data (B01 to B0) on the user data memory 200 will be stored.
n) and the data management information (F01 to F0n, MAP0) may be inconsistent. Therefore, when it is determined that the physical block B0k is in use (YES in step 404), the data in the physical block B0k is not overwritten and erased.
First, the physical block state flag F1k for displaying the latest state corresponding to the physical block B0k is changed to “not in use” (step 407), and the mapping of the physical block B0k in the block mapping information MAP1 is separated to remove the physical block. Access is disabled (step 408). Next, by referring to the contents of the physical block state flag groups F11 to F1n for displaying the latest state and the physical block state flag groups F21 to F2n for displaying the initial state, the physical block state flag group is currently in the unused state and is not used even in the initial state. The physical block in the state is searched as an empty physical block B0j (step 409),
The block mapping information MAP1 is updated so that the logical block associated with the write request is mapped to this free physical block B0j (step 410), and then data is written to this free physical block B0j (step 411). ), The corresponding physical block state flag F1j in the latest state display physical block state flag groups F10 to F1n is changed to “in use” (step 4).
12), the process ends. By doing so, even if the data management information cannot be updated due to a power failure, the data (B01) stored in the user data memory 200 is not deleted because the data existing at the previous update is not deleted. ~ B0n) and data management information (F01 ~)
There is no risk of inconsistency between F0n and MAP0).
On the other hand, although the data of the physical block B0k is stored without being erased, since the access to the physical blocks B01 to B0n is performed based on the block mapping information MAP1 in the work memory 400, the data that should have been erased. Of course, there is no possibility of being erroneously read, and the data of the logical block which should have been originally written in the physical block B0k can be normally read from the newly mapped physical block B0j. still,
After that, when a data management information update request arrives,
Also in the block mapping information MAP0 on the user data memory 200, mapping separation for the physical block B0k and new mapping for the physical block B0j are performed, and the physical block status flag F0k is not used. When a write request for a new logical block arrives, the physical block B0k can be used as a free physical block candidate for storing a new block although the previous data is stored.

If the physical block corresponding to the logical block specified by the write request cannot be found (NO in step 402), it is recognized that the write request is for a new logical block, and in this case, the latest status is displayed. Physical block state flag groups F11 to F1n for display and physical block state flag groups F21 to F21 for initial state display
By referring to the contents of F2n, the physical block that is currently in the unused state and is also in the unused state in the initial state is searched as a free physical block B0j (step 413), and the logical block associated with the write request is searched for. The block mapping information MAP1 is updated so as to map to the empty physical block B0j (step 414), and then data is written to this empty physical block B0j (step 405) to display the latest state physical block state. The corresponding physical block state flag F1j in the flag groups F10 to F1n is changed to “in use” (step 406), and the process is ended.

Next, a read request execution process (step 2)
09) will also be briefly described with reference to the flowchart of FIG. In the figure, when the process is started, the physical block B0k mapped to the logical block designated by the read request is searched based on the block mapping information MAP1 in the work memory 400 (step 501). Here, if the physical block B0k mapped to the logical block designated by the read request is not found (NO in step 502), a predetermined error process is performed. On the other hand, when such a physical block B0k is found (YES in step 502), data reading from the physical block B0k is performed (step 503). In this way, the physical block group B01
Data read from B0n is all in work memory 4
Since it is performed based on the contents of the block mapping information MAP1 on 00, the erase request execution process (step 207) and the write request execution process (step 20) described above.
8) Even if the physical block B0k is left without being erased or overwritten by the above, as long as mapping is canceled in the block mapping information MAP1 on the work memory 400, data from such physical block B0k is incorrect. Will not be read out.

In the present invention described above, unusable physical blocks left without being erased or overwritten by the erase request execution process (step 207) and the write request execution process (step 208) described above. Due to the presence of B0k, the number of physical blocks that can actually be used is somewhat reduced. However, according to the current products, the capacity of this type of flash memory is as large as 1M to 2M bytes. In general, the amount of user data handled by electronic devices for use in electronic equipment is sufficiently smaller than that, and even for unusable physical blocks, they are restored to the usable state each time the data management information is updated. As long as the update period of is properly designed, it is considered that there is no problem in practice.

[0042]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to FIGS.
This will be described with reference to FIG. In this embodiment, the user data memory 200 has five physical blocks B01 to B01.
05 is set, and the memory 200
There are five physical block status flags F01 to F05
In addition, block mapping information MAP0 that can be mapped to five physical blocks is set. Further, in the work memory 400, five physical block state flags F11 to F15 for displaying the latest state and block mapping information MAP1 which can be mapped to five physical blocks are set. Furthermore, the work memory 40
Five physical block state flags F21 to F25 for displaying the initial state (state at power-on or last data management information update) are set in 0.

As commonly shown in FIGS. 7 to 9, in the initial state of this embodiment, three physical blocks B01, B02, B04 are in use (indicated by hatching in the drawings), The two physical blocks B03 and B05 are not in use (shown in white in the figure). Further, logical blocks B11, B12, B13 are mapped to the physical blocks B01, B02, B04, respectively,
Correspondingly, in the block mapping information MAP0, physical block B01 and logical block B11, physical block B02 and logical block B12, physical block B
The mapping is set so that 01 and the logical block B11 are connected. On the other hand, physical block B03,
The mapping of B05 is canceled so that it is not connected to any logical block.

Preprocessing (steps 201, 202, 20)
The state immediately after 3) is shown in FIG. As shown in the figure, immediately after the pre-processing, the user data memory 20
The contents of the physical blocks B01 to B05 in 0 and the contents of the data management information (F01 to F05, MAP0) are of course completely identical. In addition, the user data memory 20
The contents of the block mapping information MAP0 in 0 and the contents of the block mapping information MAP1 in the work memory 400 also completely match. Further, physical block status flags F01 to F01 to
The physical block status flags F11 to F15 for displaying the latest status and the physical block status flags F21 to F25 for displaying the initial status in the work memory 400 also completely match.

The state immediately after the erase request execution process (step 207) is shown in FIG. Now, assume that an erase request for the logical block B12 has arrived following the preprocessing. At this time, referring to the block mapping information MAP1 for displaying the latest state in FIG. 7, the logical block B12
Is mapped to the physical block B02 (see the position indicated by arrow A in FIG. 7). Further, referring to the physical block state flag F22 for displaying the initial state in FIG. 7, it can be seen that the physical block B02 was in use in the initial state (see the arrow B portion in FIG. 7). In this case, the logical block B12 stored in the physical block B02 is not erased (see the location indicated by arrow A in FIG. 8). Further, the physical block state flag F12 for displaying the latest state is changed during non-use (see the position of arrow B in FIG. 8). Further, in the block mapping information MAP1 for displaying the latest state, the physical block B
02 and the logical block B12 are released from the mapping (refer to the arrow C in FIG. 8). Therefore, the data (B01 to B05) in the user data memory 200 and the data management information (F01 to
(F05, MAP0) consistency is maintained, a power failure occurs, and the stored information (F11 to F15,
Even if (F21 to F25, MAP1) disappears, it does not hinder the access (erasing, writing, reading) to the user data memory 200. On the other hand, even if a read request for the logical block B12 arrives in an attempt to erroneously read the data that should have been erased, the mapping for the logical block B12 in the block mapping information MAP1 for displaying the latest state has already been canceled. It is determined that there is an error, and the contents of the logical block B12 stored in the physical block B02 will not be erroneously read. When a data management information update request arrives, the physical block status flag group F0 in the user data memory 200 is set based on the contents of the physical block status flag groups F11 to F15 for displaying the latest status and the block mapping information MAP1.
1 to F05 and the contents of the block mapping information MAP0 are rewritten, and thereafter, the contents of the physical block state flag groups F21 to F25 for displaying the initial state are changed to the physical block state flag groups F01 to F0 of the user data memory 200.
After being updated with the content of 5, the content of the physical block status flag F22 for displaying the initial status is changed to "not in use", so that the data stored in the physical block B02 (logical block B12). Is judged to be invalid anymore, the physical block B02 is treated in the same manner as an empty state, and any data can be overwritten.

Write request execution process (step 208)
The state immediately after is shown in FIG. Now, tentatively, following the pretreatment,
It is assumed that a write request for the logical block B12 has arrived. At this time, referring to the block mapping information MAP1 for displaying the latest state in FIG. 7, it can be seen that the logical block B12 is mapped to the physical block B02 (see the location indicated by arrow A in FIG. 7). Further, referring to the physical block state flag F22 for displaying the initial state in FIG. 7, it can be seen that the physical block B02 was in use in the initial state (see the arrow B portion in FIG. 7). In this case, the logical block B12 stored in the physical block B02
Is not overwritten (see the arrow A in FIG. 9).
Also, the physical block status flag F12 for displaying the latest status
Is changed during non-use (see the position of arrow B in FIG. 9). Furthermore, the latest state display block mapping information MAP1
Then, the mapping between the physical block B02 and the logical block B12 is canceled (see the arrow C in FIG. 9). In addition, the physical block status flag group F11 for displaying the latest status
-F15 and the physical block state flag groups F21 to F25 for displaying the initial state, the physical block B03 is obtained as the physical block that is in the unused state in the latest state and is also in the unused state in the initial state. Then, the data corresponding to the logical block B12 is written to the obtained free physical block B03 (see the position of arrow D in FIG. 9). Along with this, the content of the physical block status flag F13 for displaying the latest status is changed to "in use", and at the same time, in the block mapping information MAP1 for displaying the latest status, the physical block B03 and the logical block B12 are linked to each other. Mapping will be set up (see the arrow F in FIG. 9). Therefore, although the overwrite request for the physical block B02 has arrived, the data (B01 to B05) and the data management information (F01 to F01) in the user data memory 200 are received.
05, MAP 0). for that reason,
Information stored in the work memory 400 (F11
~ F15, F21 to F25, MAP1) disappears,
Access to the user data memory 200 (erasing, writing, reading) is not hindered. On the other hand, in an attempt to read overwritten data, the logical block B1
Even if a read request for 2 arrives, the logical block B1 of the block mapping information MAP1 for displaying the latest status is displayed.
Since 2 is mapped to physical block B03,
The latest contents of the logical block B12 stored in the physical block B03 are normally read, and the contents of the logical block B12 stored in the physical block B02 that should have been overwritten should not be read by mistake. When a data management information update request arrives, the user data memory 200 is updated with the contents of the physical block status flag groups F11 to F15 for displaying the latest status and the block mapping information MAP1.
Contents of the physical block status flag groups F01 to F05 and the block mapping information MAP0 in
After that, the physical block status flag group F for initial status display
After the contents of 21 to F25 are updated with the contents of the physical block status flag groups F01 to F05 of the user data memory 200, the contents of the physical block status flag F22 for displaying the initial status are changed to unused. As a result, the data (logical block B12) stored in the physical block B02 is determined to be invalid anymore, the physical block B02 is treated in the same manner as the empty state, and the overwrite of arbitrary data is allowed. .

[0047]

As is apparent from the above description, according to the present invention, a non-volatile memory having a limited access count due to the nature of a memory element can be utilized to the maximum extent under the specified access count. Even if the power supply voltage drops or a power failure occurs, the reliability of the data read from the nonvolatile memory thereafter is not impaired.

[Brief description of drawings]

FIG. 1 is a block diagram schematically showing a hardware configuration of a small portable electronic device to which the present invention is applied.

FIG. 2 is a general flowchart showing a software configuration of a small portable electronic device to which the present invention is applied.

FIG. 3 is a flowchart showing details of an erase request execution process in the general flowchart.

FIG. 4 is a flowchart showing details of a write request execution process in the general flowchart.

FIG. 5 is a flowchart showing details of a read request execution process in the general flowchart.

FIG. 6 is a memory map for explaining a data flow associated with execution of preprocessing and data management information update processing.

FIG. 7 is a memory map showing the contents of a work memory and a user data memory immediately after execution of preprocessing.

FIG. 8 is a memory map showing contents of a work memory and a user data memory immediately after execution of an erase request.

FIG. 9 is a memory map showing the contents of a work memory and a user data memory immediately after execution of a write request.

[Explanation of symbols]

 100 Man-Machine Interface 200 User Data Memory 300 System Program Memory 400 Work Memory 500 CPU 600 System Bus

Claims (2)

[Claims] 1. A method for controlling memory access to a non-volatile memory having a limited number of accesses due to the nature of a storage element, the method being stored in the non-volatile memory prior to starting use of the non-volatile memory. The data management information is copied to a volatile memory whose number of accesses is practically unlimited to obtain data management information for displaying the latest status, and a physical block status flag group and block mapping information that make up the data management information. Among them, a pre-processing step of separately storing a set of physical block status flags on the volatile memory to form a physical block status flag group for initial status display, and a given memory access request is an erase request. Or an access request type determination step of determining whether the block is a write request; Referring to an access target physical block search step for obtaining a physical block mapped to a logical block specified by the memory access request based on the mapping information, and the stored physical block state flag group for initial state display. A block use state determination step of determining that the searched access target physical block is in use, a physical block state flag group for storing the initial state, and a physical block state flag group for displaying the latest state Referring to, a free physical block search step of searching for a physical block that is empty in the initial state and free in the latest state, and the request type determination step determines that the memory access request is an erase request. And by the access target physical block search step When the searched access target physical block is determined to be in use by the block use status determination step, the latest status display physical block status flag of the access target physical block is changed to unused, and The access request is determined to be a write request by an erase request execution step that avoids the erase process for the access target physical block and separates the block mapping relating to the access target physical block in the block mapping information, and the request type determination step. When the access target physical block searched in the access target physical block searching step is determined to be in use in the block usage status determining step, the physical block for displaying the latest status of the access target physical block is displayed. The state flag is changed to not in use to avoid the write processing to the physical block to be accessed in the non-volatile memory, and instead, the data related to the write request is searched for the free physical block in the non-volatile memory. A write request execution step of writing to a free physical block searched in step and separating the block mapping relating to the physical block to be accessed in the block mapping information and re-establishing a new block mapping, and a predetermined data management information update request arrives. In response to this, the data management information updating step of updating the content of the data management information in the non-volatile memory with the content of the data management information for displaying the latest status stored in the volatile memory, Memory access control method characterized by Law. 2. A device for controlling memory access to a non-volatile memory having a limited access count due to the nature of a memory element, the device being stored in the non-volatile memory prior to starting use of the non-volatile memory. The data management information is copied to a volatile memory whose number of accesses is practically unlimited to obtain data management information for displaying the latest status, and a physical block status flag group and block mapping information that make up the data management information. Among them, preprocessing means for copying and storing a set of physical block status flags on the volatile memory as a physical block status flag group for initial status display, and a given memory access request is an erase request. Access request type determining means for determining whether the request is a write request or the stored block map. Referring to the access target physical block search means for obtaining the physical block mapped to the logical block specified by the memory access request, and the stored physical block state flag group for initial state display, A block use state determination means for determining that the searched access target physical block is in use, a physical block state flag group for displaying the initial state and a physical block state flag group for displaying the latest state. Referring to a free physical block search unit that searches for a physical block that is in an empty state in the initial state and is also in the latest state, and the request type determination unit determines that the memory access request is an erase request, and The access target physical block searched by the access target physical block searching means. Tsu when the click is determined to be in use by the block use state determining means changes the physical block status flag for the latest status of the access target physical block in the non-use,
An erase request execution unit that avoids an erasing process for the access target physical block of the non-volatile memory and separates a block mapping relating to the access target physical block in the block mapping information, and an access request is a write request by the request type determination unit. And the access target physical block searched by the access target physical block search means is determined to be in use by the block usage status determination means, the physical block state for displaying the latest status of the access target physical block The flag is changed to not in use to avoid the write processing to the physical block to be accessed in the non-volatile memory, and instead, the data related to the write request is searched for the free physical block in the non-volatile memory. A write request executing means for writing to a free physical block searched by the means and for separating the block mapping related to the physical block to be accessed in the block mapping information and re-establishing a new block mapping, and a predetermined data management information update request have arrived. Data management information updating means for updating the content of the data management information in the non-volatile memory with the content of the data management information for displaying the latest status stored in the volatile memory. A memory access control device comprising: