JP5991239B2 - Nonvolatile semiconductor memory write control method and microcomputer - Google Patents

Description

  The present invention relates to a write control method for a nonvolatile semiconductor memory and a microcomputer including the nonvolatile semiconductor memory.

  In recent years, a microcomputer incorporating a flash memory which is a kind of nonvolatile semiconductor memory has been used. For example, the flash memory can be used for storing a user program used by a CPU (Central Processing Unit) of the microcomputer, Used for writing and storing various data. Here, the flash memory can be written for each word, but the erasure needs to be erased collectively in units of a block composed of a plurality of words.

  When the number of block erase increases, the life of the flash memory is shortened. Therefore, a technique has been proposed in which erasure is not performed as much as possible when data stored in the flash memory is updated. For example, Patent Document 1 discloses a technique for managing data in units of IDs and adding only updated data to reduce the number of erasures. In addition, there is a method in which data that can be updated and data that is not updated are managed separately to efficiently use an area in the block and reduce the number of erasures.

JP 2004-164493 A

However, it is also assumed that there is a difference in the update frequency between multiple types of data that may be updated. For example, if data with a low update frequency is included in one block, the area in which data with a high update frequency can be written is reduced by that amount, which may increase the number of block erases.
The present invention has been made in view of the above circumstances, and an object thereof is to provide a nonvolatile semiconductor memory write control method and a microcomputer that can be efficiently rewritten according to the update frequency of each data. It is in.

According to the data write control method of the nonvolatile semiconductor memory according to claim 1 or 2 , when the update frequency over a predetermined period is measured for each of a plurality of types of data (first step), based on the measurement result A plurality of types of data are divided into a low update frequency group and a high update frequency group (second step). Thereafter, the updated data is written by dividing the low frequency block for writing data belonging to the low update frequency group and the high frequency block for writing data belonging to the high update frequency group (third step).

As a result, for the high-frequency block, updated data can be written in all areas of the block, and the number of times of erasure is reduced. Also, the low frequency block is erased less frequently because the usable area is not filled with data with high update frequency. Therefore, it is possible to extend the life of the nonvolatile semiconductor memory by avoiding an increase in the number of block erasures by mixing data with different update frequencies in one block and reducing the number of block erasures as a whole. Can do.
According to the first aspect of the present invention, when the number of blocks into which data can be written becomes 2 blocks or less, the first to third steps are not executed and the plurality of types of data are updated respectively. Write to the block.
According to the invention of claim 2, in the second step, when the number of types of data belonging to one of the low update frequency group and the high update frequency group exceeds a predetermined threshold, If the plurality of types of data are updated without executing the third step, a part of the number of types of data belonging to the low update frequency group and a part of the number of types of data belonging to the high update frequency group are obtained. A fourth step of writing to the common block is executed.

The flowchart which is 1st Embodiment and shows the control content of CPU. The flowchart which shows the update frequency measurement process of step S3 Diagram explaining data write management using multiple blocks The figure which shows the processing image of normal management of step S9 The figure which shows the change of the block erasing frequency of this embodiment and the prior art according to the ratio change of the amount of used data and block capacity. (A) is a functional block diagram showing the configuration of the microcomputer, (b) is a diagram showing a write data management table in the RAM FIG. 4 is a diagram corresponding to FIG. 4 when the number of data in a group is extremely large in the second embodiment. 1 equivalent diagram The flowchart of the threshold value determination process which is 3rd Embodiment and is performed in step S5 FIG. 3 equivalent view showing the fourth embodiment

(First embodiment)
As shown in FIG. 6A, the microcomputer 1 includes a CPU 2, a flash memory (FLASH) for program storage 3, a flash memory 4 for data storage (nonvolatile semiconductor memory), a RAM (Random Access Memory) 5, A peripheral circuit 6 is provided. The CPU 2 has the function of a memory controller that controls data writing and rewriting (updating), erasing of blocks, switching of blocks to be written, and the like by reading program data from the flash memory 3 and executing the program data. Yes.
The RAM 5 is a main memory of the CPU 2, and is used as a work area for temporarily storing data, calculation results of the CPU 2, and the like when data is written or rewritten. The flash memory 4 includes a memory cell divided into a plurality of blocks as a storage unit.

  Next, the operation of the present embodiment will be described with reference to FIGS. The CPU 2 uses, for example, a plurality of types of data that are updated according to application control conditions. Each is managed with an ID. Then, when any data is updated, the updated data is sequentially written in the empty area of the flash memory 4. When data is written to all of one block, if there is no other free area, the written block is erased and the block is written. The capacity of one block is, for example, about 2 kbytes.

  As shown in FIG. 1, first, the CPU 2 sequentially writes data in one block of the flash memory 4 from the updated data regardless of the type of data as “initial management” (S1). At that time, if the number of writable blocks in the flash memory 4 is “3” or more (S2; YES), the update frequency is measured for each data (S3). Then, until the specified time T1 (predetermined time) has elapsed (S4; NO), the process returns to step S1, and the above processing is repeatedly executed.

  If the number of writable blocks in step S2 is less than “3” (NO), the process returns to step S1 to continue the initial management. In general, a flash memory is subjected to a memory test for checking whether data can be normally written even during use. The memory test is performed by, for example, whether or not the written data value can be read as it is (verify), and subsequent use (write) is prohibited for the block that is determined not to be normally written. The number of writable blocks may vary. Steps S3 and S4 correspond to the first step and update frequency measuring means.

  That is, Data1, Data2,... Shown in FIG. 3 indicate data IDs (types). Data of the same number is of the same type (here, 6 types), and data written later is updated data. As shown in FIG. 3 (1) → (2) (circled numbers in FIG. 3), when updated data is first written in block 1 and data is written in all areas of block 1, the next block is different. Write updated data to. If the blocks 2 and 3 have already been used, the block 2 is erased before writing. When data is written in all areas of the block 2, the block 3 is erased and writing is performed. Such processing is sequentially repeated (FIG. 3 (3) → (4)).

  As shown in FIG. 2, when the type of data to be measured is determined by the pointer N (S11, N = 0 to 5 in the example of FIG. 3), it is determined whether or not the data (N) is updated. (S12). If it has been updated (YES), the corresponding update frequency counter is incremented (S13). Then, the pointer N is incremented until the measurement of all data is completed (S14; NO) (S15), and the process returns to step S12 to continue the measurement.

  FIG. 6B is an example of a management table when the update frequency is measured using, for example, the flash memory 4. Each data ID, data body, data value checksum, and update count are shown. It consists of a counter. This management table may be arranged in the RAM 5 (in this case, it is not necessary to store the data body).

  Reference is again made to FIG. If the specified time elapses during the initial management (S4; YES), the update frequency of each data at that time is referred to, for example, the data is classified into two groups according to the level with respect to a predetermined threshold To do. For example, those with an update frequency less than the threshold are set as a low update frequency group, and those with an update frequency exceeding the threshold are set as a high update frequency group (S5, second step, update frequency determination means). Then, when the next write target (written) block is erased (S6), the write destination when any one of the low update frequency group and the high update frequency group is subsequently updated is set to the erased block. Set (S7, third step, block management means).

  At this time, if there are “3” or more usable blocks in the flash memory 4 (S8; YES), management is performed in two blocks as “normal management” thereafter (S9, third step, block management means). “Management by two blocks” means that management is divided into two parts, a low frequency block for writing data belonging to the low update frequency group and a high frequency block for writing data belonging to the high update frequency group.

  FIGS. 3 (5) → (6) are processing images of steps S6 and S7. For example, the update frequency of Data1 to 3 is high (high update frequency group) and the update frequency of Data4 to 6 is low (low update frequency group). Then, the block 2 is erased, and if any of the data 1 to 3 is updated thereafter, the updated data is written into the block 2 (high frequency block). Then, the data 4 to 6 updated thereafter are written in the remaining area of the block 1 (low frequency block).

  FIG. 4 is a processing image of “normal management” (S9). When all the areas of block 2 are used by repeating the update of Data 1 to 3 from the state of FIG. (High-frequency block) is erased, and updated Data 1 to 3 are written thereafter (upper row (7) in FIG. 4). After that, when all the areas of the block 3 are used in the same manner, the block 2 is erased and the updated Data1 to 3 are written thereafter (upper part (8) in FIG. 4).

  In addition, if data 4 to 6 are repeatedly updated from the state of FIG. 3 (6), if all areas of block 1 are used first, then block 3 (low frequency block) is erased and updated thereafter. Data 4 to 6 are written (lower row (7) in FIG. 4). After that, when all the areas of the block 3 are used in the same manner, the block 1 is erased and the updated data 1 to 3 are written (lower part (8) in FIG. 4).

  Then, as shown in FIG. 1, until the prescribed time T2 (predetermined time) elapses after the transition to the normal management in step S9 (S10; NO), the routine returns to step S8 and the normal management is continued, but the prescribed time T2 (YES), the process returns to step S1 and re-executes from the initial management. The specified time T2 is set to a time longer than the specified time T1 in step S4 (for example, several tens of times T1). That is, the update frequency of each data may change depending on the control conditions of the application, etc. Therefore, when the normal management is continued to some extent, the update frequency is measured again, and the low update frequency group and the high update frequency group The data belonging to each is sorted again.

Here, as shown in FIG. 5, when the maximum number of data that can be stored in one block is M and the number of data to be used (number of IDs) is U, the updated data as in the prior art is put into one block. If written without distinction, the number of block erase times E1 until all data is updated is
E1 = U / (MU) (1)
(Provided that M> U is a condition). That is, how many times the U-type data can be updated before erasing the block once is represented by (MU) / U, and therefore, the expression (1) is the reciprocal thereof.

On the other hand, if the data is divided into a high update frequency group and a low update frequency group and managed as two blocks in use as in this embodiment, the number of block erase times E2 until all data is updated is
E2 = 2 × U / (2M−U) (2)
It becomes. Therefore, the number Ed obtained by reducing the number of block erases is:
Ed = E1-E2 = U / (MU) -2 * U / (2MU)
^{= U 2 / {(M-} U) (2M-U)} ≧ 0 ... (3)
Therefore, the number of erasures is surely reduced.
Here, as a specific numerical example, when calculating with M = 100 and U = 30,
E1 = 30 / (100-30) ≈0.43
E2 = 2 × 30 / (2 × 100−30) ≈0.35
Therefore, E1> E2.

  FIG. 5 is a graph showing the number of conventional erasures when M = 100 and the number of erasures according to this embodiment. In the conventional case, the number of erasures rapidly increases as the ratio of the used data amount to the block capacity approaches 60% to 100%. On the other hand, in the case of the present embodiment, it can be seen that the number of erasures does not increase so much even when the ratio increases, and the difference between the two increases greatly.

  As described above, according to the present embodiment, when the CPU 2 measures the update frequency over the specified time T1 for each of a plurality of types of data written to the flash memory 4, a plurality of types of data are determined based on the measurement result. Divide the data into a low update frequency group and a high update frequency group. Thereafter, the updated data is written by dividing the low-frequency block for writing data belonging to the low update frequency group and the high-frequency block for writing data belonging to the high update frequency group.

  As a result, for the high-frequency block, updated data can be written in all areas of the block, and the number of times of erasure is reduced. Also, the low frequency block is erased less frequently because the usable area is not filled with data with high update frequency. Therefore, it is possible to avoid an increase in the number of block erasures due to a mixture of data having different update frequencies in one block, and to extend the life of the flash memory 4 by reducing the number of block erases as a whole. it can.

  In addition, since the CPU 2 re-executes the processing from the step S1 when the specified time T2 has elapsed since the start of the normal management in the step S9, the CPU 2 is low even when the update frequency of each data changes according to the subsequent situation. An update frequency group and a high update frequency group can be appropriately determined. Furthermore, the CPU 2 does not perform normal management when the number of blocks into which data in the flash memory 4 can be written is 2 blocks or less, and writes data to a common block when a plurality of types of data are updated. If the write data cannot be managed separately for the high-frequency block, writing can be performed as usual.

(Second Embodiment)
7 and 8 show the second embodiment. The same parts as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted. Hereinafter, different parts will be described. In the second embodiment, as a result of separating each data into a low update frequency group and a high update frequency group in step S5, the size of one data group is extremely large as shown in FIG. Occupy the majority.

  In this case, when step S6 is executed, “YES” is determined in the subsequent step S16 by exceeding a predetermined upper limit (threshold, for example, 90% of the capacity of one block). Then, when each data group is divided into ½ (part) of each data group size, new write destinations of ½ of the low update frequency group and ½ of the high update frequency group are The block erased in step S6 is set (S17, fourth step, block management means). Then, the process proceeds to step S8.

  In the case shown in FIGS. 7B, 5, and 6, data 1 to 6 belong to the high update frequency group, and data 7 and 8 belong to the low update frequency group. If Data1-6 occupy most of the block capacity, then updated Data1-3 and Data7 are written to the erased block 2. When the remaining Data 4 to 6 and Data 8 are updated, if there is an empty area in block 1, it is written in block 1 as it is. If the number of data is an odd number, adjustment is made at (1/2 ± 1).

  Then, when writing is performed in the entire area of the block 1 prior to the block 2, the updated data 4 to 6 and data 8 are written by erasing the block 3 (not shown). Therefore, the “normal management” executed in step S9 after step S17 has a management form as shown in FIGS.

  As described above, according to the second embodiment, the CPU 2 does not execute step S7 when the number of types of data belonging to one of the low update frequency group and the high update frequency group exceeds a predetermined threshold value. When each data is updated, 1/2 of the number of data IDs belonging to the low update frequency group and 1/2 of the number of data IDs belonging to the high update frequency group are written in a common block. Therefore, even when there is a bias in the number of data belonging to each group, an increase in the number of block erasures can be suppressed.

(Third embodiment)
In the third embodiment, the CPU 2 determines a threshold for separating the low update frequency group and the high update frequency group based on the update frequency measurement result. FIG. 9 is executed in step S5, but the CPU 2 refers to the update frequency of each data stored in the management table as shown in FIG. 6B (S21), and determines the minimum value and The maximum value is determined (S22). Then, a threshold value is set between the minimum value and the maximum value (S23). Here, as a method of setting the threshold value, for example, an intermediate value between the minimum value and the maximum value or a weighted average of all the update frequency numbers can be considered.

  As described above, according to the third embodiment, the CPU 2 determines the threshold value for dividing the plurality of types of data into the low update frequency group and the high update frequency group in step S5, and the update frequency measurement result in step S3. Therefore, the threshold value can be appropriately determined according to the actual update frequency of each data.

(Fourth embodiment)
In the fourth embodiment, as shown in FIG. 10, when there are four or more blocks that can be written to the flash memory 4, one of them (for example, the block 4) is fixed data that is not likely to be updated. Used as a storage area. Then, the same block management as in the first to third embodiments is performed using the blocks 1 to 3.
As described above, when the CPU 2 also handles fixed data, the block in which the fixed data is stored and the block in which the updated data are stored are managed separately, so that the increase in the number of block erasures can be suppressed. .

The present invention is not limited to the embodiments described above or shown in the drawings, and the following modifications or expansions are possible.
The number of blocks of the flash memory may be “5” or more.
When the specified time T2 has elapsed, the process from step S1 may be re-executed as necessary.
In the second embodiment, when the size of one data group occupies most of the capacity of one block, the number of data types for writing the next block belongs to the low update frequency group and the high update frequency group. It is not limited to ½ of the data, but may be a part of each data group.

  In the drawings, 1 is a microcomputer (update frequency measuring means, update frequency determining means, block management means), 2 is a CPU, and 4 is a flash memory (nonvolatile semiconductor memory).

Claims (12)

In a method of writing a plurality of types of updated data in a vacant area with respect to a nonvolatile memory (4) having a plurality of blocks of a predetermined size, which is a unit capable of erasing data collectively,
A first step of measuring the update frequency over a predetermined period for each of the plurality of types of data;
A second step of dividing the plurality of types of data into a low update frequency group and a high update frequency group based on a result of the measurement;
A low frequency block to write data belonging to a low update frequency groups, separately and frequently block to write data belonging to a high update frequency group, Ri Do and a third step of writing the updated data,
When the number of blocks into which data can be written is 2 blocks or less, the first to third steps are not executed,
A data writing control method for a nonvolatile semiconductor memory, wherein writing to a common block is performed when each of the plurality of types of data is updated . In a method of writing a plurality of types of updated data in a vacant area with respect to a nonvolatile memory (4) having a plurality of blocks of a predetermined size, which is a unit capable of erasing data collectively,
A first step of measuring the update frequency over a predetermined period for each of the plurality of types of data;
A second step of dividing the plurality of types of data into a low update frequency group and a high update frequency group based on a result of the measurement;
The third step of writing the updated data by dividing the low frequency block for writing data belonging to the low update frequency group and the high frequency block for writing data belonging to the high update frequency group,
In the second step, when the number of types of data belonging to one of the low update frequency group and the high update frequency group exceeds a predetermined threshold, the third step is not executed,
When each of the plurality of types of data is updated,
A nonvolatile semiconductor memory characterized by executing a fourth step of writing a part of the number of types of data belonging to the low update frequency group and a part of the number of types of data belonging to the high update frequency group to a common block Data write control method. 3. The data write control method for a nonvolatile semiconductor memory according to claim 1 , wherein the first to third steps are re-executed when a predetermined time elapses after the execution of the third step is started . In the second step, when the number of types of data belonging to one of the low update frequency group and the high update frequency group exceeds a predetermined threshold, the third step is not executed,
When each of the plurality of types of data is updated,
Low update frequency part of the number types of data belonging to the group and the high update frequency data belonging to group a part of the number of kinds and executes a fourth step of writing to a common block claim 1 or 4. A data write control method for a nonvolatile semiconductor memory according to 3 .   In the second step, a threshold for dividing the plurality of types of data into a low update frequency group and a high update frequency group is determined based on the update frequency measurement result in the first step. The data write control method for a nonvolatile semiconductor memory according to claim 1. If there are 4 or more blocks in which data can be written, use one of them as a storage area for fixed data that is not updated,
6. The data write control method for a nonvolatile semiconductor memory according to claim 1, wherein the first to third steps are executed using the remaining three blocks. A microcomputer (1) having a plurality of blocks of a predetermined size, which is a unit capable of erasing data collectively, and having a nonvolatile memory (4) in which a plurality of types of updated data are written and stored in an empty area ) And
For each of the plurality of types of data, update frequency measuring means for measuring the update frequency over a certain period,
Based on the measurement result, the plurality of types of data is divided into a low update frequency group and a high update frequency group, an update frequency determination means,
Thereafter, a low frequency block for writing data belonging to the low update frequency group and a high frequency block for writing data belonging to the high update frequency group, and a writing block management means for writing updated data are provided ,
When the number of blocks in which data can be written is 2 blocks or less, the update frequency measurement means, the update frequency determination means, and the block management means are not functioned,
A microcomputer that writes to a common block when each of the plurality of types of data is updated . A microcomputer (1) having a plurality of blocks of a predetermined size, which is a unit capable of erasing data collectively, and having a nonvolatile memory (4) in which a plurality of types of updated data are written and stored in an empty area ) And
For each of the plurality of types of data, update frequency measuring means for measuring the update frequency over a certain period,
Based on the measurement result, the plurality of types of data is divided into a low update frequency group and a high update frequency group, an update frequency determination means,
Thereafter, a low frequency block for writing data belonging to the low update frequency group and a high frequency block for writing data belonging to the high update frequency group, and a writing block management means for writing updated data are provided,
In the state where the number of types of data belonging to any one of the low update frequency group and the high update frequency group exceeds a predetermined threshold according to the determination result of the update frequency determination unit, the block management unit When a plurality of types of data are updated, a part of the number of types of data belonging to the low update frequency group and a part of the number of types of data belonging to the high update frequency group are written in a common block. A microcomputer. 9. The microcomputer according to claim 7, wherein the update frequency measuring means restarts the measurement of the update frequency when a predetermined time elapses after the block management by the block management means is executed . In the state where the number of types of data belonging to any one of the low update frequency group and the high update frequency group exceeds a predetermined threshold according to the determination result of the update frequency determination unit, the block management unit When a plurality of types of data are updated, a part of the number of types of data belonging to the low update frequency group and a part of the number of types of data belonging to the high update frequency group are written in a common block. The microcomputer according to claim 7 or 9 . The update frequency determination means, wherein said plurality of types of data, characterized in that the threshold value for bisecting into low update frequency group and the high update frequency group is determined based on the prior SL update frequency measurement result Item 11. The microcomputer according to any one of Items 7 to 10. If there are 4 or more blocks in which data can be written, use one of them as a storage area for fixed data that is not updated,
The microcomputer according to any one of claims 7 to 11, wherein the remaining three blocks are used to cause the update frequency measuring means, the update frequency determining means, and the block management means to function.

Images