JP3812933B2 - File system and control method thereof - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a non-volatile semiconductor memory device such as a non-volatile memory including a flash memory, a data restructuring method such as a memory card using the same, a file system employing a data refresh management method, and a control method thereof.
[0002]
[Prior art]
Conventionally, a flash memory is a non-volatile semiconductor memory device having a function of electrically erasing memory cells (referred to as sectors or blocks) in a single chip or in an arbitrary area. As memory cells of this flash memory, for example, there are those called NOR type and NAND type. In any memory cell, the memory cell configuration shown in FIG. 17 is the basic structure of data storage.
[0003]
In FIG. 17, the memory cell 100 has a floating gate type field effect transistor structure, and one element can constitute one bit (one cell). Therefore, it is excellent in high integration. Data is written into the memory cell 100 by applying about 12 V to the control gate electrode G, about 7 V to the drain electrode D, and 0 V to the source electrode S, and generating hot electrons generated near the drain junction at the floating gate electrode. This is done by injecting into FG. By this writing, the threshold voltage Vth viewed from the control gate electrode G of the memory cell 100 is increased (data “0”).
[0004]
Further, the memory cell 100 having the floating gate type field effect transistor structure has a higher value when one element forms a multi-value (a threshold voltage Vth of the memory cell is made to be a power of 2 n at intervals of several hundred mV). Excellent integration.
[0005]
Data writing to the memory cell 100 was generated in the vicinity of the drain junction by setting the source electrode S to 0 V, applying a pulse of about 12 V to the control gate electrode G and about 7 V to the drain electrode D for several microseconds. This is done by injecting hot electrons into the floating gate electrode FG. By this writing, the threshold voltage Vth viewed from the control gate electrode G of the memory cell 100 is increased. The threshold voltage Vth of the memory cell 100 is controlled by changing the voltage of the control gate electrode G, changing the voltage of the drain electrode D, or changing the pulse width.
[0006]
On the other hand, in erasing, the control gate electrode G is grounded, and a positive high voltage (about 12 V) is applied to the source electrode S, thereby generating a high electric field between the floating gate electrode FG and the source electrode S, and thin gate oxidation. This is done by extracting electrons accumulated in the floating gate electrode FG to the source electrode S using a tunneling phenomenon through the film. This erasure is normally performed in block units (for example, 16 Kbytes or 64 Kbytes). By this erasing, the threshold voltage Vth viewed from the control gate electrode G is lowered (data “1”). At this time, since the memory cell 100 does not have a selection transistor, the negative threshold voltage Vth (excessive erasure) is a fatal defect (a defect in which correct data cannot be read when reading data).
[0007]
In data reading, a low voltage of about 0 V is applied to the source electrode S and a low voltage of about 1 V is applied to the drain electrode D, and a voltage of about 5 V is applied to the control gate electrode G. This is performed by utilizing the correspondence between information “1” and “0”. The reason why the voltage to the drain electrode D is lowered is to prevent a parasitic weak write operation (soft write).
[0008]
Further, when reading data in multi-value storage, a low voltage of about 0 V is applied to the source electrode S and a low voltage of about 1 V is applied to the drain electrode D, and the voltage applied to the control gate electrode G is changed. This is performed using the voltage of the control gate electrode G. In this memory cell 100, since data writing is performed on the drain electrode D side and erasing is performed on the source electrode S side, it is desirable to optimize the junction profile individually so as to suit each operation. That is, the source electrode S and the drain electrode D have an asymmetric structure, and the drain junction uses an electric field concentration type profile to increase write efficiency, and the source junction uses an electric field relaxation type profile capable of applying a high voltage. is doing.
[0009]
In the method of applying a high voltage to the source electrode S at the time of erasing, since the breakdown voltage of the source junction must be increased, the source electrode S side is difficult to be miniaturized, and hot holes are generated near the source electrode S. There is a problem that the portion is trapped in the tunnel insulating film and the reliability of the cell is lowered.
[0010]
Therefore, as another example of erasing (negative gate erasing), a negative voltage (about -10 V) is applied to the control gate electrode G, a power supply voltage (about 5 V) is applied to the source electrode S, and erasing is performed by a tunnel current.
[0011]
One advantage of this erasing method is that since the voltage applied to the source electrode S during erasing is low, the junction breakdown voltage on the source electrode S side may be low, so that the gate length of the cell can be shortened. It is. In addition, when the negative gate erase method is used, there is an advantage (referred to as sector erase) that the erase block size can be easily reduced.
[0012]
Further, in the erasing method in which a high electric field is applied to the source electrode S, a band-to-band tunneling current that is a parasitic current caused by the high electric field flows, and the current value becomes several mA across the entire chip. It becomes difficult to use a booster circuit with limited capability. Therefore, conventionally, a high voltage Vpp for erasing has been supplied from the outside of the chip. In the negative gate erasing method, the power supply voltage Vcc (5 V or 3 V) can be supplied to the source electrode S, so that there is an advantage that a single power supply can be made relatively easily.
[0013]
In the method using hot electrons for data writing, a current of about 1 mA flows per cell at the time of writing. Therefore, a flash memory that uses an FN tunnel current as in a conventional EEPROM and reduces the current flowing per cell at the time of data writing. There is also.
[0014]
With the miniaturization of semiconductor processes and the spread of battery-driven portable devices, there is a demand for lower operating power. Therefore, there is an active demand / development of 3V single operation products and 2.7V single operation products instead of 5V single operation.
[0015]
When reading data with a 3V power supply (Vcc) or a 2.7V power supply, in the current flash EEPROM, the control gate line (word line) has a power supply potential (Vcc = 3V) or an internal boost for speeding up / expanding operation margin. About 5V is applied.
[0016]
In such a nonvolatile semiconductor memory device, there are many operation states (writing, block erasing, batch erasing of all chips, status register, and so on) compared to a random access memory (RAM) capable of writing and reading in a short time. Read). If a large number of operating states are made to correspond to combinations of external control signals (/ CE, / WE, / CE, etc.), the control signals in the conventional EPROM and EEPROM become insufficient, and a new control signal needs to be added. As a result, usability deteriorates, and a method of taking a command system without increasing the number of control signal lines has been devised and has become mainstream in the world. In this nonvolatile semiconductor memory device, a command input by a user enters a circuit that recognizes a command called CSM (command state machine), and WSM (write state machine) performs an operation corresponding to the command (erase / erase / state). Write).
[0017]
Some nonvolatile semiconductor memory devices of this type are obtained by dividing the size of erase blocks in a chip unevenly or evenly.
[0018]
There are a write / erase method using a FN tunnel current and a NAND type memory cell system in which 8 or 16 memory cells are connected in series. The VAND type is slower in reading speed than the NOR type, but has an advantage that the memory cell size can be reduced.
[0019]
As described above, normally, as shown in FIG. 18, a binary value (1 bit) is stored in one memory cell. There are also attempts to record multi-values such as 4-level (2 bits), 8-level (3-bit), and even 16-level (4-bit) in one memory cell.
[0020]
For example, when binary (1 bit) is stored in one memory cell, the data “1” is set to the threshold voltage Vth = 2.5V of the memory cell, and the data “0” is set to the threshold voltage Vth = 6 of the memory cell. .5v, the difference in threshold voltage Vth between data is about 4V. However, when storing, for example, 4 values (2 bits) in one memory cell, it is necessary to further store 2 values between 4V.
[0021]
That is, the data “11” is the memory cell threshold voltage Vth = 2 V, the data “10” is the memory cell threshold voltage Vth = 3.5 V, the data “01” is the memory cell threshold voltage Vth = 5 v, and the data If “00” is defined as the threshold voltage Vth of the memory cell = 6.5 v, the difference in threshold voltage Vth between data is reduced to about 1.5V.
[0022]
Thus, as more values are stored in one memory cell, the potential difference between one storage state and another storage state in one memory cell becomes smaller (data reading errors are more likely to occur). There is a drawback, but there is also an advantage that a large-capacity memory can be manufactured at low cost.
[0023]
Utilizing such non-volatile characteristics of flash memory, it has been proposed to use it for data storage (file system) applications such as a magnetic disk represented by a hard disk. That is, the data handling unit is made as small as several hundred bytes like a hard disk, and the data is managed efficiently. However, when a file system is constructed with a flash memory, there are the following problems (1) to (3).
[0024]
(1) First, in this type of nonvolatile semiconductor memory device, the data read operation is about 20 microseconds (μSec) and the erase operation is several times compared to the high data read speed of about 100 nanoseconds (nSec). Even slower, 100 milliseconds (mSec).
[0025]
(2) Next, in the flash memory, data is rewritten in units of blocks, and data cannot be rewritten in units of 1 byte.
[0026]
(3) Further, the data retention period can be about 10 years or more. However, as the number of block erasures increases, the deterioration of the oxide film progresses and the leakage current increases, so the data retention period tends to be shortened. is there. Therefore, the number of times of erasure is limited to about 100,000 times for each block, and rewriting occurs intensively in one block, and if the number of times of erasure of one block exceeds 100,000 times, the reliability of the entire flash memory chip is increased. The characteristics deteriorate significantly.
[0027]
In order to use the flash memory for data storage (file storage), the following countermeasures are taken against the problems (1) to (3). First, a status bit indicating an erased or valid state of data is added to the data. That is, in practice, the erase operation is not performed, and only the writing to the status bit indicates that the data state has changed. As described above, the data erasure is replaced by the writing of the status bit, so that the data erasure is apparently speeded up by 1000 times or more, and the actual erasing operation does not need to be frequently performed.
[0028]
However, when data in the block is not actually erased and only the status bits are erased, that is, invalid data increases, there is no free space for writing new data. When there is no more area for writing data, valid data is copied to a previously erased spare block. When this copying is completed, the block of the copy source that has a large amount of erased data in the block is erased and newly set as a spare block. This operation is called reconstruction (reclaim).
[0029]
This reconstruction operation is schematically shown in FIGS. 19 (1) to 19 (3). That is, there are an erase block 1 and an erase block 2, where the erase block 1 represents a data storage block and the block 2 represents a spare block. It is assumed that the spare block has been erased in advance (all bit values are “1”). Each erase block has five data handling units (hereinafter referred to as sectors), and each sector has a status bit indicating the status. In order to simplify this explanation, here, the status bit is 1 bit, and when the value is “1”, the sector data is valid, and when the value is “0”, the sector data is invalid. In FIG. 19A, the erase block 1 has data written in all the sectors A to E. Among them, the sector B and the sector D are invalid data with the status bit being “0”. Since there are no more sectors to which data can be written in this block, a reconstruction operation is performed. In the reconstruction, as shown in FIG. 19 (2), the contents of the sector in which the status bit is “1” and the data is valid are copied to the spare block (block 2). A sector that has not been copied in the spare block is in an erased state, and data can be written to this sector. As described above, when valid data is copied to the spare block, the original data storage block (block 1) is erased and used as a new spare block (FIG. 19 (3)).
[0030]
Japanese Utility Model Publication No. 6-48051 “IC card” uses a memory that can be erased only in units of one chip, a semiconductor memory that consists of a plurality of chips, and a semiconductor memory that can be erased in units of bytes, and can be erased only in units of one chip. 1 shows an example of an IC card in which information on a file written in a simple memory is written in a semiconductor memory that can be erased in byte units.
[0031]
In Japanese Laid-Open Patent Publication No. 6-223591 “Flash EEPROM Array Clean-Up Method”, the number of times of rewriting is larger than the number of other erase blocks in the block to be subjected to the conventional rebuilding operation because the number of erase times is averaged. Either a small number of blocks are selected, or an erase block with a large number of erased (invalid) data is selected in order to facilitate data writing.
[0032]
[Problems to be solved by the invention]
However, in the above conventional configuration, in order to lower the bit unit price, in recent years, a plurality of bits (2 bits, 3 bits, and 4 bits) have been stored in one memory cell. This is called multi-leveling. In the case of multi-leveling, the more the bit is stored in one memory cell, the smaller the potential difference of the threshold voltage Vth between data must be reduced. Data tends to deteriorate, for example, data cannot be retained for 10 years, and there are problems in terms of reliability in data retention and product yield.
[0033]
The present invention solves the above-described conventional problem, and refreshes a block that is judged to have deteriorated data as a target block for the reconstruction operation, thereby reducing the retained data even if the multi-value is increased. An object of the present invention is to provide a file system and a control method therefor that can be prevented and can improve the reliability of data retention and improve the product yield.
[0034]
[Means for Solving the Problems]
The file system of the present invention has a configuration in which an erase block is logically divided into smaller storage area units than a non-volatile semiconductor memory device configured by a plurality of erase blocks and capable of erasing data in units of blocks. In a file system in which a file is managed using state information of storage area units and a block is reconstructed to release an invalid storage area unit included in the erase block, the date when the block was reconstructed for each erase block is displayed. Reconstruction date storage means for storing in a predetermined storage area, and reconstruction control means for preferentially rebuilding a block for which a certain period has elapsed from the date stored in the reconstruction date storage means Yes, and the above objective is achieved. Also, the file system control method of the present invention is a storage system in which an erase block is logically divided into smaller blocks than a non-volatile semiconductor memory device constituted by a plurality of erase blocks and capable of erasing data in units of blocks. In the file system control method for managing files using area unit and erase block status information, the block rebuild process that releases invalid storage area units contained in erase blocks is rebuilt for each erase block. The date is stored in a predetermined storage area, and a block for which a certain period has elapsed from that date is preferentially processed, whereby the above object is achieved.
[0035]
Preferably, the reconstruction control means in the file system of the present invention is a reconstruction elapsed time calculation means for calculating the elapsed time from the most recent block reconstruction time for each block based on the date, and First calculation result storage means for storing calculation results in a storage area that can be accessed at high speed in a table format, and firstly reconstructing a block after a certain period of time based on the calculation results of elapsed time Restructuring processing means. Preferably, in the file system control method of the present invention, the calculation result obtained by calculating the elapsed time from the most recent block reconstruction for each block is displayed in a table format in a storage area accessible at high speed. Based on the calculation result of the elapsed time, a block for which a certain period has elapsed is preferentially set as a reconstruction target block.
[0036]
Further preferably, the reconstruction control means in the file system of the present invention includes an erase count counting means for counting the erase count of the block, and a second calculation for storing the calculation result in a tabular form in a storage area accessible at high speed. The result storage means and the second reconstruction processing means for preferentially rebuilding at a predetermined short period based on the date as the number of times of erasure stored in the second calculation result storage means increases. Preferably, in the file system control method of the present invention, a block with a larger number of deletions is preferentially set as a target block to be reconstructed with a predetermined short period based on the date.
[0037]
Further preferably, the reconstruction control means in the file system of the present invention includes an erase count counting means for counting the number of block erases, and a time since the most recent block rebuild for each block based on the date. Reconstructed elapsed time calculation means for calculating time, third calculation result storage means for storing the erase count and elapsed time in a table format in a storage area accessible at high speed, and calculating parameters from the erase count and elapsed time Parameter calculation means, parameter storage means for storing the calculation results calculated by the parameter calculation means in a table format in a storage area that can be accessed at high speed, and on the basis of this parameter, a block after a certain period of time is given priority. And a third reconstruction processing means for reconstructing. Preferably, in the file system control method of the present invention, the calculation result calculated from the number of block erasures and the elapsed time from the most recent block reconstruction for each block based on the date, The data is stored in a table format in a storage area that can be accessed at high speed. Based on the calculation result, a block after a certain period of time is preferentially set as a target block for reconstruction.
[0038]
Further preferably, the reconstruction control means in the file system of the present invention performs the reconstruction operation at least one of when the power is turned on, when returning from the system reset, and when the reconstruction operation is not performed for a certain period. Give priority.
[0039]
Further preferably, the reconstruction control means in the file system of the present invention has memory cell characteristic deterioration detecting means for detecting that the characteristics of the memory cell have deteriorated, and the memory cell characteristic deterioration detecting means When the deterioration is detected, the reconstruction operation is preferentially performed on the storage area unit and the erase block including the memory cell in which the deterioration is detected.
[0040]
  Further preferably, the number of erasuresAs a calculation result obtained from the reconstruction dateIt is represented by a multiplier of 10, and the multiplier is used in the calculation formula.Preferably, a multivalue is stored in a memory cell of the nonvolatile semiconductor memory device.
[0041]
Further preferably, the reconstruction date storage means in the file system of the present invention stores the reconstruction date in the status information recording area having a number exceeding the storage area unit number that actually exists.
[0042]
Further preferably, the reconstruction control means in the file system according to the present invention obtains the elapsed time from the time of reconstruction, the number of erasures, and the obtained information in the status information recording area having a number exceeding the storage area unit number that actually exists. At least one of the parameters is stored.
[0043]
Further preferably, in the file system of the present invention, an unusable physical block number is stored in a status information recording area having a number exceeding the storage area unit number that actually exists.
[0044]
Further preferably, in the file system of the present invention, the unusable physical block number is stored in a storage area accessible at high speed.
[0045]
The operation of the above configuration will be described below. As a block to be reconstructed, data deterioration, that is, a block in which a state in which the oxide film of the memory cell has deteriorated and the reliability of the stored data has occurred as described above is taken into consideration. Specifically, compared to other erase blocks, not only blocks with more erased (invalid) data are selected, but also data degradation has progressed (or data degradation may have progressed). The block is also a target block for reconstruction. More specifically, the date (year and month) at the time of building the block is written in a part of the block. In general, as the number of times of rewriting increases, the generation of an oxide film trap and the deterioration of the oxide film progress, and the target of the reconstruction operation is performed after a certain period of time. Alternatively, the block erase count and the date (year / month) at the time of block construction are written in a part of the block. In general, as the number of rewrites increases, the generation of oxide traps and the deterioration of oxide films progress and the leakage current increases, so the data deterioration tends to progress. (Preferentially) is the target of the reconstruction operation. That is, valid data in a block that is considered to have deteriorated due to a large number of rewrites is transferred to the spare block side having a smaller number of rewrites.
[0046]
Further, in the case of a nonvolatile memory device having a refresh function (or refresh command), the threshold voltage Vth of the memory cell has changed more than a specified value as a result of execution of the refresh necessity determination, and the memory cell is refreshed. If it is determined that it is necessary, the block is preferentially subjected to the reconstruction operation. As an example of the refresh function, there is JP-A-7-37397. The refresh function referred to here is a reference voltage serving as a reference for determining whether data held in a memory cell is “0” or “1”. It is shown that the read margin is checked by changing, and it is determined that the memory cell with a small margin is deteriorating, and an excessive write operation is performed to recover the threshold voltage Vth.
[0047]
As a block to be reconstructed, a block with a smaller number of rewrites than other erase blocks is selected, or a block with many erased (invalid) data is selected. In order to lower the bit unit price, in recent years, multi-valued storage is performed in which a plurality of bits (2 bits, 3 bits, or 4 bits) are stored in one memory cell. The more bits are stored in the memory cell, the smaller the potential difference between the data, so the data retention time degrades (cannot be retained for 10 years), and reliability and yield problems occur. In addition, since a block whose rewrite date (year and month) is old (a block for which data deterioration is a concern) is a target of the reconstruction operation, in particular, a nonvolatile semiconductor memory device manufactured by a miniaturization process or one Degradation of data retention characteristics (cannot be retained for 10 years) of multilevel cells that store multiple bits (2 bits, 3 bits, or 4 bits) in a memory cell is eliminated, eliminating problems of reliability and yield
[0048]
Specifically, each erase block has an area for storing the date (year / month) of rebuilding the block, and blocks for which a certain period has passed are subject to rebuilding. (Cannot be held for 10 years) is resolved, and the problem of yield is also resolved.
[0049]
In addition, each erase block has an area for storing the date when the block was reconstructed, and the calculation result of the elapsed time for each block is stored as a table on a high-speed memory such as RAM, and a certain period of time has elapsed. Since the reconstructed block is to be reconstructed, the deterioration of the data retention characteristic (cannot be retained for 10 years) is resolved, the problem of yield is solved, and high-speed search is possible.
[0050]
In addition, each erase block has an area for storing the number of block erases and the year and month when the block is reconstructed, and as the number of block erases increases, the data is reconstructed at a constant short cycle. Degradation (cannot be held for 10 years) is eliminated and yield problems are eliminated.
[0051]
In addition, each erase block has an area for storing the number of block erases and the year and month when the block is reconstructed, and the calculation result calculated from the number of block erases and the elapsed time for each block is stored on a high-speed memory such as a RAM. Since a block that has been stored for a certain period of time and whose fixed period has passed is targeted for reconstruction, the deterioration of data retention characteristics (cannot be retained for 10 years) is eliminated, and the problem of yield is eliminated.
[0052]
Further, when the power is turned on, when the system is reset from the system reset, or when the rebuilding operation is not performed for a certain period of time, the execution of the rebuilding operation is started preferentially, thereby degrading the data retention characteristics (10 years). Can't hold money), and the yield problem is solved.
[0053]
Furthermore, since the calculation result (a parameter whose amount changes depending on the conditions of the elapsed time and the erase count) obtained from the data rewrite count and the reconstruction date is stored, the storage capacity of the RAM can be reduced.
[0054]
Furthermore, since the sector status information recording area having a number exceeding the actually existing sector number is used, the storage area can be effectively used without being wasted.
[0055]
Further, since the unusable physical block number is stored, a flash memory including a defective block (data cannot be rewritten) can be used, and the product yield is increased and the cost is reduced. In addition, since unusable physical block numbers are stored on the RAM, high-speed search is possible.
[0056]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments 1 and 2 of the file system of the present invention will be described below with reference to the drawings.
(Embodiment 1)
FIG. 1 is a block diagram showing a system configuration of a file system according to the first embodiment of the present invention.
[0057]
In FIG. 1, a flash memory file system 1 (hereinafter referred to as a file system 1) includes a file system control unit 2, a file system memory unit 3, a flash memory control unit 4, and a flash memory unit 5 for storing data. The file is managed using the state information of the erase block and the storage area unit (sector), and a block restructuring operation for releasing invalid sectors included in the erase block is performed.
[0058]
The file system control unit 2 is requested by an application (or operating system OS) 6 and controls various data processes including a reconstruction process performed by the file system 1. As shown in FIG. 2, the file system control unit 2 stores the date of rebuilding the block for each erase block in the file system memory unit 3 having a predetermined storage area as shown in FIG. The storage means 21 and the rebuild control means 22 for preferentially rebuilding blocks for which a certain period has elapsed from the date stored in the rebuild date storage means 21 are provided. The rebuild date is stored in the file system memory unit 3 and then stored in a predetermined area (sector # 128 of the block control sector) of the flash memory unit 5 via the flash memory control unit 4.
[0059]
The reconstruction control means 22 is a reconstruction elapsed time calculation means 221 that calculates the elapsed time from the most recent block reconstruction for each block based on the date, and a storage that can access the calculation result at high speed. First calculation result storage means 222 for storing the area in a table format, and first reconstruction processing means 223 for preferentially reconstructing a block after a certain period of time based on the calculation result of elapsed time Is.
[0060]
The file system memory unit 3 has a storage area that can be accessed at high speed, and is used by the file system 1 to store control information such as status information (FIGS. 4 to 6) described later in the storage area. is there. As the control information, in addition to the state information, for example, the number of erasures and the reconstruction date, the elapsed time from this date, the parameter calculation result, etc. are stored in a table format.
[0061]
The flash memory control unit 4 is instructed by the file system control unit 2 to perform normal data reading, data writing and data erasing to the flash memory unit 5, and further data reading, data writing and data erasing by a reconstruction process, etc. It controls various data processing.
[0062]
The flash memory unit 5 stores data, and is a non-volatile semiconductor memory device configured by a plurality of erase blocks and capable of erasing data in units of blocks. Here, the erase block has a logically smaller data storage. It is divided into sectors which are area units. Hereinafter, it demonstrates in detail using FIG.
[0063]
FIG. 3 is a data configuration diagram stored in the flash memory unit 5 of FIG. As shown in FIG. 3, each erase block of the flash memory is divided into sectors that are data handling units, each erase block has a size of 64 kilobytes, and each sector has a size of 512 bytes. In this case, each erase block has 128 sectors (the block control sector and the data area of sector # 1 to sector # 127 in FIG. 3).
[0064]
Here, any one erase block among a plurality of erase blocks in the flash memory is assumed to be a spare block for reconstruction of the file system 1 (data erased and formatted block).
[0065]
The plurality of erase blocks have 0, 1, 2,. . . And physical block number. The numbers assigned as sector # 1 to sector # 127 are referred to as physical sector numbers. The physical block number and the physical sector number are uniquely determined by hardware, and in principle have fixed values. The logical block number and the logical sector number are numbers provided for efficient management of the file system, and their values can change.
[0066]
Of the sectors in the erase block, one sector existing at the head is reserved in advance as a sector for storing control information of the block (hereinafter referred to as a block control sector).
[0067]
In this block control sector, 2 bytes of status information of the corresponding erase block itself, 2 bytes of status information of the sector # 1 in the erase block, and 2 bytes of status information of the sector # 127 are arranged from the head. 128 pieces = 256 bytes of information. The remaining 256 bytes (status information after sector # 128 in FIG. 2) stores the reconstruction date (date) and the like.
[0068]
The status information of the corresponding erase block itself includes (1) 11111111b: unused, (2) 11111110b: data transfer in progress as the logical block number of 0 to 255 to which the file stem control unit 2 assigns and the current block status. (3) 11111100b: During original block erasure, (4) 111111000b: data present, (5) 11110000b: block full.
[0069]
Similarly to the state information of the corresponding erase block itself, the state information of the sector in the corresponding erase block includes the logical sector numbers 0 to 4095 assigned by the file system control unit 2 and the current sector state as (I) 1111b: Not used, (II) 1110b: Data writing in progress, (III) 1100b: Data writing complete, (IV) 1000b: Data valid, (V) 0000b: Data invalid.
[0070]
As described above, the most recent reconstruction date / time (or year / month) of the corresponding block is stored in the portion (sector # 128) exceeding the sector # 127 status information in FIG. For example, in order to store the reconstruction date and time, for example, “1999111220” is set to represent 20:00 on November 12, 1999. In order to store the reconstruction date, for example, “199911” represents November 1999. Alternatively, after 2000, for example, “00011” represents November 2000, and “00112” represents December 2001, for example. For further simplification, after 2000, for example, “0011” may represent November 2000, and “0112” may represent December 2001, for example.
[0071]
FIG. 4 is a data configuration diagram (hereinafter referred to as a block information table) regarding each erase block stored in the file system memory unit 3 (RAM) of FIG. In FIG. 4, the block information includes a physical block number, a logical block number, and a block state as one unit. Specifically, “physical block number = 0, logical block number = 0, block state = This indicates that there is an erase block with “data present” and an erase block with “physical block number = 1, logical block number = 1, block state = data full (block full)”.
[0072]
FIG. 5 is a data configuration diagram (hereinafter referred to as sector information table 1) regarding each sector stored in the file system memory unit 3 (RAM) of FIG. In FIG. 5, the physical block number, the physical sector number, the logical sector number, and information with the sector state as one unit are included. Specifically, “physical block number = 0, physical sector number = 0, Physical sector with logical sector number = 100, sector state = data valid, and physical sector with physical block number = 0, physical sector number = 1, physical sector = 10, sector state = data invalid, etc. It shows that
[0073]
FIG. 6 is a data configuration diagram (hereinafter referred to as sector information table 2) regarding the number of sector states for each erase block stored in the file system memory unit 3 (RAM) of FIG. In FIG. 6, the information includes the physical block number, the number of unused sectors, the number of valid data sectors, the number of invalid data sectors, and the reconstruction date as one unit. = 0, the number of unused sectors = 100, the number of valid data sectors = 20, the number of invalid data sectors = 7, and the reconstructed year and month 10010 (October 2000) ", and the" physical block number = 1, The number of unused sectors = 0, the number of valid data sectors = 50, the number of invalid data sectors = 77, and the erase block “reconstruction date 00112 (December 2001)” exist.
[0074]
The above-described block information table in FIG. 4, sector information table 1 in FIG. 5 and sector information table 2 in FIG. 6 are obtained from the flash memory unit 5 from the block control sector and the block control sector in FIG. The status information after sector # 128 is read out and created in the file system memory unit 3 (RAM). Even when the block information table, the sector information table 1 and the sector information table 2 are not created in the RAM unit, the file system 1 can be used, but the speed of data exchange with the flash memory unit 5 There is an advantage that it is faster than the exchange.
[0075]
The operation of the above configuration will be described below.
[0076]
FIG. 7 is a flowchart relating to the reading process of the control data stored in the file system memory unit 5 of FIG.
[0077]
As shown in FIG. 7, when a data read request stored in the flash memory unit 5 is generated, the data read request is sent from the application (or operating system OS) 6 in FIG. 1 to the file system control unit 2. A logical sector number to be read is given, and the file system control unit 2 reads from the sector information table 1 stored in the file system memory unit 3 (RAM) the sector state of the sector storing the data to be read is “data” The logical sector number that is “valid” is searched, and the physical storage location (physical sector number) of the corresponding sector in the flash memory unit 5 is acquired (step S121). The flash memory control unit 4 performs sector reading of the acquired physical storage location (step S122). Further, the file system control unit 2 determines whether there is still data to be read (step S123). If there is still data to be read, the process returns to step S121, and if there is no data to be read, the process ends. To do.
[0078]
Next, when a data write request to the flash memory unit 5 occurs, the data write request is given by the application (or operating system OS) 6 of FIG. The file system control unit 2 acquires the physical block number having the largest number of unused sectors from the sector information table 2 stored in the file system memory unit 3 (RAM), and has the physical block number been acquired? Check if. When a writable block cannot be obtained, the file system 1 is reconstructed to secure a write area.
[0079]
As described above, when a data write request to the flash memory unit 5 is generated, but a sector in which data can be written cannot be obtained, that is, when there are many invalid sectors and there are no unused sectors, the application 6 (operating system OS) generates a request for rebuilding the file system 1 and the file system control unit 2 controls the rebuilding operation. Hereinafter, control of this reconstruction operation will be described.
[0080]
FIGS. 8 and 9 are flowcharts relating to the rebuilding process of the file system 1 in FIG. 1, and FIGS. 10 and 11 are data configuration diagrams showing data changes (operations) during the rebuilding process in FIGS. is there. A method for constructing the file system 1 using the table having the state information related to the blocks and sectors as shown in FIGS. 4 to 6 and the state information similar to the data structure shown in FIG. It is shown in Japanese Patent Laid-Open No. 11-254974. Here, typical FIG. 12 (corresponding to FIG. 3 of the present invention) and FIG. 13 (corresponding to FIG. 6 of the present invention) are shown as reference examples. In the present invention, it can be understood by comparing with FIG. 12, but “state information after sector # 128” is added as shown in FIG. 3, and it can be understood by comparing with FIG. 13, but as shown in FIG. "Rebuild date" has been added. Here, this added point (characteristic part of the present invention) will be mainly described, and details such as how to use the block state information and the sector state information from the application 6 (or operating system OS) will be described in Japanese Patent Application No. 11. The detailed description is omitted here as it conforms to 2549743.
[0081]
In FIG. 10, a block to be reconstructed is a block # 0, and a spare block is a block #n. First, a physical block number whose reconstructed date has passed for a certain period is obtained from the sector information table 2, and then a logical block number corresponding to this physical block number is obtained from the block information table (step 151; A1, A2). However, if no data has passed for a certain period, the physical block number and logical block number with the largest number of data invalid sectors are acquired. Hereinafter, these block numbers are referred to as a reconstruction target physical block number and a reconstruction target logical block number.
[0082]
Next, the spare block, that is, the logical block number of the block reserved in the unused state (erased state) for the reconstruction of the file system 1 is updated to the logical block number to be reconstructed (A3), The block state is updated from “unused” to “data transfer in progress” (A4). Since the logical block number is “11111111b” as will be described later, the spare block can be specified by searching the block information table. Similarly, in the block information table, the logical block number of the spare block is updated to the reconstruction target logical block number (A5), and the block status is updated from “unused” to “data transfer in progress” (A6).
[0083]
Further, the valid sector data included in the block having the physical block number to be reconstructed is copied to the spare block. Here, in order to efficiently perform the copying operation, the sector information table 1 is searched for the sector having the data block number of the reconstruction target physical block number, and the physical sector number is acquired. (Step 153, A7).
[0084]
When the sector status is “data valid”, the sector status information in the block control sector having the physical sector number acquired in A7 stored in the block having the physical block number to be reconstructed and the data in the data area are stored. The sector information having the same physical sector in the spare block and the data area are copied (A8, A9).
[0085]
Further, the sector status information table 1 is updated. That is, the logical sector number and sector state ("data valid") corresponding to the physical sector number acquired in A7 having the physical block number to be reconstructed from the sector information table 1 are acquired in A7 each having the physical block number of the spare block. Are copied to the logical sector number and sector state corresponding to the same physical sector number (A10, A11).
[0086]
As described above, the data of one sector whose sector state is “data valid” and information associated with the sector are copied from the reconstruction target block to the spare block. In order to copy all the sectors in the reconstruction target block whose data state is “data valid”, whether or not there are still remaining sectors in the block of the reconstruction target physical block number from the sector state information table 1 Confirmation is made (step 156), and if there are remaining sectors, the process returns to step 153 and the above processing is executed until there are no remaining sectors.
[0087]
Next, the sector information table 2 is updated, the reconstruction target block is erased, and the state information associated therewith is updated. That is, the number of data effective sectors of the physical block to be reconstructed in the sector information table 2 is copied to the number of data effective sectors of the spare block (A12), and the date of this reconstruction operation is entered in the reconstruction date. (Step 157, A13).
[0088]
As the number of unused sectors, a value obtained by subtracting the number of valid data sectors from the total number of sectors included in the block is entered (A14).
[0089]
The number of invalid data sectors is “0” immediately after reconstruction. Next, in advance of erasure of the reconstruction target block, the block state in the block state information in the block control sector stored in the spare block is updated from “data transferring” to “original block erasing” (A15), Similarly, the block state of the corresponding block information table is updated from “data transfer in progress” to “original block erasure in progress” (steps 158 and A16).
[0090]
Also, the reconstruction date is entered in the status information after sector # 128 stored in the spare block, as in A13 (A17).
[0091]
Here, the erase operation of the reconstruction target block is actually performed (step 159). When the erase operation is completed, the block state in the block state information in the block control sector stored in the spare block is updated from “original block erased” to “data present” (A18), and the spare block in the block information table is updated. Similarly, the block state is updated from “deleting original block” to “data present” (step 160, A19).
[0092]
Since all data in the block becomes “1” by the erasing operation, the block having the logical block number “11111111b” in the block state information stored in a certain block is defined as a spare block. By the operation, the reconfiguration target block automatically becomes a spare block, and the block information in the block control sector automatically becomes “unused; 11111111b”.
[0093]
In addition, each block number item erased in step 159 of the block information table (FIG. 4), sector information table 1 (FIG. 5), and sector information table 2 (FIG. 6) automatically becomes a spare block. Since no reference is made in writing or reading data to the flash memory unit 5, no update is necessary. The reconstruction operation is thus completed.
[0094]
In the first embodiment, the reconstruction process of the file system 1 has been described as erasing one erase block. However, this series of reconstruction processes can be executed a plurality of times.
[0095]
When the file system 1 is composed of a plurality of flash memories, an arbitrary one erase block is secured in advance as a spare block for reconstruction of the file system 1 for each flash memory. Instead, a configuration may be adopted in which one spare block is secured for a plurality of flash memories. In this case, there is an advantage that the number of reserved blocks to be secured can be reduced.
(Embodiment 2)
In the first embodiment, the case where the latest reconstruction date (for example, year and month) of the corresponding block is stored after the sector # 128 state information in FIG. 3 and the reconstruction process is performed using the reconstruction date is shown. In the second embodiment, after the sector # 128 status information in FIG. 3, in addition to the most recent reconstruction date (for example, year and month) of the corresponding block, the number of erasures of the corresponding block is also stored, and the reconstruction process is performed using these This is the case.
[0096]
FIG. 14 is a block diagram illustrating a functional configuration of the file system control unit of the file system according to the second embodiment of the present invention. In FIG. 14, the file system control unit 2A has a reconstruction date storage unit 21A and a reconstruction control unit 22A. The reconstruction control unit 22A includes an erase number counting unit 221A that counts the number of block erasures, a reconstruction elapsed time calculation unit 222A that calculates an elapsed time from the most recent block reconstruction date for each block, Third calculation result storage means 223A for storing the erase count and elapsed time in a table format in the file system memory section 3 having a storage area accessible at high speed, and a predetermined parameter (elapsed time and Parameter calculation means 224A for calculating the amount of change according to each condition of the number of erasures), and the calculation result calculated by this parameter calculation means 224A in the file system memory section 3 having a storage area accessible at high speed The parameter storage means 225A for storing the data and the third reconstruction processing for performing the reconstruction processing based on the parameters And it has a means 226A.
[0097]
FIG. 15 is a configuration diagram of data relating to each sector state for each erase block (another example of the sector information table 2) stored in the file system memory unit 3 of FIG. In FIG. 15, there is table information in which a physical block number, the number of unused sectors, the number of valid data sectors, the number of invalid data sectors, the reconstruction date and time, and the number of erasures are set as one unit. FIG. 15 shows an erase block in which, for example, physical block number = 0, unused sector number 100, data valid sector number = 20, and data invalid sector number = 7, as in FIG. 6 described in the first embodiment. Further, for example, physical block number = 1, number of unused sectors = 0, number of valid data sectors = 50, and number of invalid data sectors = 77 are shown. Further, in FIG. 15, in addition to the reconstruction date of FIG. 6, the storage area is increased, but the number of erasures is also stored. The number of erasures is 10,000, which represents 10,000 times.
[0098]
  Focusing on the fact that the memory capacity can be reduced by storing the number of erasures (or the number of rewrites) and the calculation results (parameters whose amount changes according to the elapsed time and the number of erasures) calculated from the reconstruction date. If the date is October 2001, the rebuild date of the block is December 1999, and the number of deletions is 2102, then only the year is considered,
  2001-1999 = 2 years
  The number of erasures is represented by a multiplier of 10, 2102= 2 × 10 3 is stored as a calculation result on the RAM (file system memory unit 3), and 2 years, 3rd power, and about 2 are stored as parameters as “0232” as follows. This makes it possible to represent the year and number of times with three digits.
[0099]
As a method for determining whether or not to perform the rebuilding process, if it is within one year from the time of the latest rebuilding process, the number of times of erasure is 70000 times or more, that is, 0047 or more is the target of the rebuilding. If it is one year from the time, the number of erasures is 50000 times or more, that is, “0145” or more is the object of reconstruction, and if it is 2 years from the latest reconstruction process, the number of erasures is 20000 times or more, If “0242” or more is the target of reconstruction, and if it is 3 years from the most recent reconstruction process, the number of deletions is 7000 times or more, that is, “0337” or more is the target of reconstruction, and the latest reconstruction process 4 years from the time of deletion, the number of erasures is 5000 times or more, that is, “0435” or more is the target of reconstruction, and if it is 5 years from the most recent reconstruction process, the number of erasures is 2000 times or more, that is, “ 0532 "or more of reconstruction If it is 6 years from the time of the most recent reconstruction process, the number of deletions is 700 times or more, that is, “0627” or more is the target of reconstruction, and if it is 7 years from the most recent reconstruction process, If the number of times is 500 times or more, that is, “0725” or more is the target of reconstruction, and if it is 8 years from the most recent rebuilding process, the number of erasures is 200 times or more, that is, “0822” or more If it is 9 years from the most recent reconstruction process, the number of deletions is 70 times or more, that is, “0917” or more is the object of reconstruction, and if it is 10 years or more from the most recent reconstruction process, The number of times is 50 times or more, that is, 1015 or more is the object of reconstruction.
[0100]
When a data write request to the flash memory unit 5 is generated, but a sector in which data can be written cannot be obtained, a request to reconstruct the file system 1 is generated. From the sector information table 2, first, the physical block number and logical block number whose calculation results of the number of erasures and the elapsed time have exceeded the above-mentioned fixed values (parameters) necessary for data guarantee are obtained (step 151). If no fixed period has passed, the physical block number and logical block number with the largest number of data invalid sectors are acquired (step 151).
[0101]
Conventionally, as a block to be reconstructed, a block with a smaller number of rewrites than other erase blocks is selected, or a block with many erased (invalid) data is selected. In order to lower the bit unit price, in recent years, a plurality of bits (2 bits, 3 bits, and 4 bits) have been stored in one memory cell (referred to as multi-valued). However, in the case of multi-value, the more the bits are stored in one memory cell, the smaller the potential difference of the threshold voltage Vth between the data, so that the retained data deteriorates and the data retention time is 10 years. It cannot be held. This causes problems of data retention reliability and product yield. On the other hand, according to the present invention represented by the first and second embodiments, a block in which data has deteriorated or a block whose rewrite date (year / month) is old (a block in which data deterioration is a concern) is reproduced. Since it is a target of the construction operation, in particular, a non-volatile semiconductor memory device manufactured by a miniaturization process and a multi-value storage that stores a plurality of bits (2 bits, 3 bits, and 4 bits) in one memory cell Deterioration of the data retention characteristics of the cell (cannot be retained for 10 years) can be eliminated, and the above problems of data retention reliability and product yield can be eliminated.
[0102]
As described above, the effect will be further described. As the reconstruction operation of the first embodiment,
(A) If a predetermined time has elapsed since the previous restructuring operation, data deterioration has occurred, so that the data is rebuilt in order to refresh the data. As a result, the deterioration of the retained data can be prevented even when the number of values is increased, and the reliability of data retention and the yield can be improved.
[0103]
Furthermore, as a more detailed correspondence, as the reconstruction operation of the second embodiment,
(B) Since the data retention characteristic deteriorates as the number of rewrites increases, the data is refreshed when a shorter period of time elapses from the previous restructuring operation as the number of rewrites increases. Then, rebuild. As a specific example, reconstruction is performed when the calculation result of the number of rewrites and the elapsed time exceeds a certain value. As a result, the deterioration of the retained data can be prevented even when the number of values is increased, and the reliability of data retention and the yield can be improved.
[0104]
In the first embodiment, the most recent reconstruction date of the corresponding block is stored after the sector # 128 state information in FIG. 3, and in the second embodiment, after the sector # 128 state information in FIG. In addition to the most recent reconstruction date of the block, the number of erasures of the corresponding block is stored. However, the present invention is not limited to this. Even if the reconstruction process is performed using the stored data, the deterioration of the retained data can be prevented even when the multi-value is used, and the effect of the present invention can be improved in the reliability and the yield for retaining the data. Play. In this case, the reconstruction control means includes an erase count counting means for counting the number of block erases, a second calculation result storage means for storing the calculation results in a table format in a storage area accessible at high speed, 2 The second calculation processing unit may be configured to preferentially reconstruct at a predetermined short period based on the date as the number of erases stored in the calculation result storage unit increases.
[0105]
Further, as an example other than the first and second embodiments, in the case of a flash memory having a refresh command, a memory cell is deteriorated and a block determined to be refreshed in order to recover the threshold voltage Vth of the memory cell. In order to make it the object of the next reconstruction operation, the above-mentioned sector information table 2 of FIG. 15 is recorded at a place where the calculation result (parameter calculation result) of the rewrite count (erasure count) and the elapsed time is recorded. It is also possible to record a value greater than a certain value (for example, 9999) and preferentially become the target of reconstruction at the next reconstruction.
[0106]
Further, as an example other than the first and second embodiments, a defective block that cannot be rewritten or a block whose erase count (or rewrite count) exceeds a specified value (for example, the rewrite count guaranteed by the device specification) is shown in FIG. As shown, the block status is prohibited (blocks not used for data recording). Then, the use prohibition block number is recorded in the state information area after sector # 128 in FIG.
[0107]
As another example, the nonvolatile semiconductor memory device of the present invention may be provided with a power supply voltage detection circuit that detects power-on, and the above-described reconstruction operation may be activated when power-on is detected. As yet another example, the above-described reconstruction operation may be started when returning from a system reset.
[0108]
【The invention's effect】
As described above, according to the present invention, each erase block has an area for storing the date (year / month) of rebuilding the block, and a block that has passed a certain period from the most recent rebuilding date is reconstructed. Since it is an object of construction, it is possible to eliminate deterioration of data retention characteristics (cannot be retained for 10 years), and to solve the problem of product yield.
[0109]
In addition, each storage block has an area for storing the date (year / month) of rebuilding the block, and a storage area where the calculation result of the elapsed time from the most recent rebuilding date for each block can be accessed at high speed Stored in a table format (table) above, and blocks whose fixed period has passed since the most recent reconstruction date are subject to reconstruction, so high-speed search is possible and data retention characteristics deteriorate (cannot be retained for 10 years) ), And the product yield problem can be solved.
[0110]
Furthermore, each erase block has an area for storing the number of block erases and the date (year and month) of rebuilding the block. Deterioration of retention characteristics (cannot be retained for 10 years) can be eliminated, and data retention reliability and product yield problems can be resolved.
[0111]
In addition, each erase block has an area for storing the number of block erases and the date (year and month) when the block was reconstructed, and the calculation result calculated from the number of block erases and the elapsed time for each block can be calculated at high speed. Data is stored in an accessible storage area as a table (table), and blocks whose fixed period has passed since the most recent reconstruction date are subject to reconstruction. This can eliminate the problems of data retention reliability and product yield. In this case, there is an advantage that data on a storage area accessible at high speed can be reduced.
[0112]
Furthermore, if the rebuilding operation is started preferentially when the power is turned on, when returning from system reset, or when the rebuilding operation is not performed for a certain period, the data retention characteristics deteriorate (cannot be retained for 10 years) And the problems of data retention reliability and product yield can be solved.
[0113]
Furthermore, when the memory cell deteriorates and recovers its deteriorated characteristics, that is, when a refresh operation is required, the reconstruction operation is performed, so that the deterioration of data retention characteristics (cannot be maintained for 10 years) can be eliminated, and the product yield can be improved. The problem can be solved.
[0114]
Furthermore, since the calculation result obtained from the number of rewrites (the number of erasures) and the reconstruction date is stored, the storage capacity can be reduced.
[0115]
Furthermore, since the sector status information recording area having a number exceeding the sector number that actually exists is used for recording the number of rewrites (erasure count) and the reconstruction date, the storage area can be used effectively.
[0116]
Furthermore, since the unusable physical block number is stored, a flash memory including a defective block that cannot be rewritten can be used, and the product yield is increased and the cost is reduced.
[0117]
Furthermore, since the unusable physical block number is stored in a storage area that can be accessed at high speed, a high-speed search can be performed.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a system configuration of a file system in Embodiment 1 of the present invention.
2 is a block diagram illustrating a functional configuration of a file system control unit in FIG. 1; FIG.
FIG. 3 is a data configuration diagram stored in the flash memory unit of FIG. 1;
4 is a data configuration diagram related to a block information table stored in the file system memory unit of FIG. 1. FIG.
5 is a data configuration diagram related to the sector information table 1 stored in the file system memory unit of FIG. 1. FIG.
6 is a data configuration diagram related to the sector information table 2 stored in the file system memory unit of FIG. 1;
7 is a flowchart regarding control data read processing stored in the file system memory unit of FIG. 1; FIG.
FIG. 8 is a flowchart related to the file system reconstruction process (part 1) of FIG. 1;
FIG. 9 is a flowchart related to the file system restructuring process (part 2) of FIG. 1;
10 is a data configuration diagram showing data change (part 1) during the reconstruction process of FIGS. 8 and 9. FIG.
11 is a data configuration diagram showing data change (part 2) during the reconstruction process of FIGS. 8 and 9. FIG.
FIG. 12 is a data configuration diagram stored in a conventional flash memory unit.
13 is a data configuration diagram related to the sector information table 2 stored in the file system memory unit of FIG. 12;
FIG. 14 is a block diagram showing a functional configuration of a file system control unit of a file system in Embodiment 2 of the present invention.
FIG. 15 is a data configuration diagram relating to each sector state for each erase block, stored in the file system memory unit of FIG. 1;
FIG. 16 is a data configuration diagram related to a block information table stored in the file system memory unit of the present invention.
FIG. 17 is a configuration diagram of a memory cell having a floating gate type field effect transistor structure.
FIG. 18 is a diagram showing a threshold voltage distribution when binary (1 bit) is stored in one memory cell;
FIGS. 19 (1) to (3) are diagrams schematically showing conventional reconstruction operations in order.
[Explanation of symbols]
1 File system
2,2A File system controller
21, 21A Reconstruction date storage means
22,22A Reconstruction control means
221, 222A Reconstruction elapsed time calculation means
222 First calculation result storage means
223 first reconstruction processing means
221A Erase count counting means
223A Third calculation result storage means
224A parameter calculation means
225A parameter storage means
226A Third reconstruction processing means
3 File system memory
4 Flash memory controller
5 Flash memory part
6 Application (or operating system OS)

Claims (16)

For a non-volatile semiconductor memory device composed of a plurality of erase blocks and capable of erasing data in units of blocks, the erase blocks are logically divided into smaller storage areas, and the states of the erase blocks and storage areas In a file system in which a file is managed using information and a block is reconstructed to release an invalid storage area unit included in the erase block,
Reconstruction date storage means for storing in a predetermined storage area the date that the block was reconstructed for each erase block, and blocks that have passed a certain period of time from the date stored in the reconstruction date storage means are preferentially reconstructed. A file system having reconstruction control means for performing a construction process.   The reconstruction control means includes a reconstruction elapsed time calculation means for calculating an elapsed time from the most recent block reconstruction performed for each block based on the date, and a storage that can access the calculation result at high speed. A first calculation result storing means for storing the data in a tabular form in the area; and a first reconstruction processing means for preferentially reconstructing a block after a certain period of time based on the calculation result of the elapsed time. Item 1. The file system according to Item 1.   The reconstruction control means includes an erase count counting means for counting the number of erases of the block, a second calculation result storage means for storing the calculation results in a table format in a storage area accessible at high speed, and the second calculation 2. The file system according to claim 1, further comprising: a second reconstruction processing unit that preferentially reconstructs a predetermined short period based on the date as the number of times of erasure stored in the result storage unit increases.   The rebuild control means includes an erase count counting means for counting the erase count of the block, and a rebuild elapsed time for calculating an elapsed time from the most recent block rebuild for each block based on the date. A calculation means, a third calculation result storage means for storing the erase count and elapsed time in a tabular form in a storage area accessible at high speed, a parameter calculation means for calculating a parameter from the erase count and elapsed time, Parameter storage means for storing the calculation results calculated by the parameter calculation means in a table format in a storage area accessible at high speed, and a third method for preferentially reconstructing blocks after a certain period of time based on the parameters. The file system according to claim 1, further comprising a reconstruction processing unit.   The rebuild control means preferentially performs the rebuild operation when power is turned on, when returning from a system reset, or at least when the rebuild operation is not performed for a certain period of time. A file system according to any one of the above.   A memory cell characteristic deterioration detecting unit for detecting that the characteristic of the memory cell has deteriorated, and the memory cell characteristic deterioration detecting unit includes a memory cell in which the deterioration is detected when the memory cell characteristic deterioration detecting unit detects the deterioration of the memory cell characteristic; 6. The file system according to claim 1, wherein the rebuilding operation is preferentially performed on the storage area unit and the erase block. The file system according to claim 3 or 4, wherein the file system is expressed by a multiplier of 10 as a calculation result obtained from the number of times of erasure and the reconstruction date, and the multiplier is used in the calculation formula.   The file system according to claim 1, wherein the reconstruction date storage unit stores the reconstruction date in a state information recording area having a number exceeding a storage area unit number that actually exists.   The reconstruction control means stores in the status information recording area having a number exceeding the storage area unit number that actually exists, at least one of the elapsed time from the time of the reconstruction, the number of erasures, and the parameters obtained therefrom. The file system according to any one of claims 1 to 8.   The file system according to any one of claims 1 to 9, wherein an unusable physical block number is stored in a status information recording area having a number exceeding a storage area unit number that actually exists.   The file system according to claim 1, wherein an unusable physical block number is stored in a storage area accessible at high speed. The file system according to claim 1, wherein multiple values are stored in a memory cell of the nonvolatile semiconductor memory device. For a non-volatile semiconductor memory device composed of a plurality of erase blocks and capable of erasing data in block units, the erase block is logically divided into smaller storage areas and the erase block status information. In a file system control method for managing files,
The reconstruction process of the block that releases the invalid storage area unit included in the erase block stores the date of rebuilding the block for each erase block in a predetermined storage area, and starts a certain period from the date. A file system control method that preferentially processes blocks that have passed. The calculation result of calculating the elapsed time from the most recent block reconstruction for each block is stored in a table format in a storage area accessible at high speed, and based on the calculation result of the elapsed time, The file system control method according to claim 13, wherein a block for which a certain period has elapsed is preferentially reconstructed. 14. The file system control method according to claim 13 , wherein a block having a larger number of erases is preferentially reconstructed at a predetermined short period based on the date and preferentially. Store the calculation results calculated from the number of block erasures and the elapsed time from the most recent block reconstruction for each block based on the date, in a high-speed accessible storage area in table format, The file system control method according to claim 13, wherein a block whose fixed period has elapsed is preferentially reconstructed based on the calculation result.

Images