JP4766781B2 - Semiconductor memory device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a nonvolatile semiconductor memory device using a flash memory or the like that is read / written / erased in units of blocks.
[0002]
[Prior art]
Non-volatile semiconductor memories that are read, written, and erased on a sector basis include, for example, M5M29F25611VP manufactured by Mitsubishi Electric and HN29W25611 manufactured by Hitachi, Ltd. These are flash memories that are read, written and erased in units of 2 kbyte sectors. A conventional storage device using flash memory converts a sector address (logical address) written by the host computer into a physical address of the flash memory, and further reads and writes sector data to and from the flash memory by a command issued by the host computer. .
[0003]
In the flash memory, there are an initial defective sector within a specified value and a defective sector that cannot be used after the start of use. For such bad sectors that occur initially and later, a process of converting the logical address of the host computer into the physical address of the alternative sector is required. On the other hand, it is also known that even if a write / erase error has occurred once, there is a sector that operates normally when the write / erase is performed again. Such sector errors are considered to be accidental errors.
[0004]
[Problems to be solved by the invention]
Conventionally, when a write / erase error occurs, it is unconditionally regarded as a bad sector, and an accidental error sector is also converted to an alternative sector. For this reason, the accidental error sector that can operate normally is not used and is wasted. A sufficient reserve sector must be secured even at the start of use, and the usable storage capacity of the user to be secured in advance is suppressed.
[0005]
An object of the present invention is to provide a highly reliable and low-cost semiconductor memory device that can use a storage capacity without waste by reusing an accidental error sector.
[0006]
[Means for Solving the Problems]
The semiconductor memory device of the present invention supports a plurality of user sectors for recording data, an address indicating the position of the user sector, and an error count value indicating the number of recording errors in the user sector for each user sector. A non-volatile memory cell having a user sector table defined, and for an error sector in which a recording error has occurred among the user sectors, based on an error count value of the error sector defined in the user sector table Thus, a controller for determining whether or not to record data later is provided, whereby the above object is achieved.
[0007]
The nonvolatile memory cell further includes a spare sector that records data in place of the error sector when a recording error occurs, and the controller later stores the error sector determined to record data later. It may be changed to a spare sector to be used.
[0008]
The non-volatile memory cell further includes a spare sector table that defines, for each spare sector, an address indicating the position of the spare sector and an error count value indicating the number of recording errors in the spare sector in association with each other. It may be.
[0009]
If the error count value of the error sector is smaller than a preset value, the controller determines that data will be recorded later in the error sector, and if larger, data is recorded later in the error sector. You may decide not to.
[0010]
In the spare sector table, the order of spare sectors to be used is added, and when the controller determines that data is to be recorded later in the error sector, data in the earliest spare sector is based on the order. And the spare sector changed from the error sector may be assigned the earliest order.
[0011]
If the ratio between the error count value of the error sector and the number of times of recording in the error sector until a recording error occurs is smaller than a preset value, the controller transfers data to the error sector later. If it is determined that data is to be recorded, and if it is large, it may be determined that data will not be recorded later in the error sector.
[0012]
In the spare sector table, the order in which spare sectors are used is added, and when the controller determines that data is to be recorded later in the error sector, the spare sector table is assigned the earliest spare sector based on the order. Data may be recorded, and the spare sector changed from the error sector may be given the last order.
[0013]
The controller may shift the order of spare sectors following the earliest spare sector forward.
[0014]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments 1 and 2 of the present invention will be described below with reference to the accompanying drawings.
[0015]
(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of a semiconductor memory device 1 having nonvolatile memory cells 15 according to the present invention. The semiconductor memory device 1 is, for example, a flash memory, and is a card type semiconductor memory device 1 (for example, an ATA card). The card-type semiconductor storage device 1 is used as an external storage medium for recording digital information such as character information, image information, and music information by a digital device.
[0016]
Referring to FIG. 1, card type semiconductor memory device 1 is required for accessing an internal controller 10 for controlling the internal operation of semiconductor memory device 1, ROM and RAM 11 for storing an operation program of internal controller 10, and flash memory. Control sequencer 12 for generating a control signal at a predetermined timing and controlling access to the flash memory, and a control command for accessing the flash memory by a terminal (not shown) equipped with a card. A plurality of registers 13 (parameter register 13-1, command register 13-2, status register 13-3), a data buffer 14 for temporarily storing data to be read or written, and a large capacity A flash memory 15 comprising a memory cell array for recording data; It is provided.
[0017]
In order for a terminal (not shown) to record data in the flash memory 15 of the card-type semiconductor storage device 1 and to read data from the flash memory 15, the terminal uses an address via a card control signal and a card address bus. Each register is accessed by data for register selection to be sent to the decoder 16 and data to be sent to the register 13 and the data buffer 14 via the card data bus, and data is read and written. Specifically, when the terminal sets a parameter in the parameter register 13-1 and writes a command in the command register 13-2, the internal controller 10 reads the parameter register 13-1 and the command register 13-2, and sends a command. A predetermined process is performed. The status register 13-3 is read to determine whether the command processing has ended normally or has ended in error. Data in the flash memory 15 is accessed via the data buffer 14.
[0018]
As will be described later, the flash memory 15 includes, in addition to user sectors, defective sectors and sectors storing data necessary for various controls. Since the terminal refers to the user sector with a logical address, the internal controller 10 performs logical-physical address conversion.
[0019]
FIG. 2A shows a memory map of the flash memory 15. The flash memory 15 is large and includes a user data recording area 15-1, a user sector address conversion table 15-2, and a spare sector address conversion table 15-3. The user data recording area 15-1 is an area for recording data desired by the user. However, in the user data recording area 15-1, in addition to a user sector storing data, a defective sector that cannot record data from the beginning or that cannot record data after the start of use, and a defective sector occur. There is also a spare sector (described later) which is a feature of the present invention and is used as an alternative to the above. These sectors are mixed at random physical address positions. A logical address is assigned to the user sector, which is indicated by a number in [] in the example of FIG. Similarly, a logical number is assigned to the spare sector.
[0020]
FIG. 2B shows a user sector address conversion table 15-2 according to the present invention. In the user sector address conversion table 15-2, physical addresses are stored in order of logical addresses so that the physical address USR_PA [k] of the user sector [k] corresponding to the offset parameter k which is a logical address can be obtained. Further, the table 15-2 stores the error count value USR_ER [k] of the user sector [k] in addition to the physical address. The error count value USR_ER [k] is a value representing the number of times that a write / erase error (recording error) has occurred in the sector in the past. Conventionally, a sector in which a recording error has occurred is not used afterwards. However, even if such a sector is used, if it is written / erased again, it will be reused if it operates normally. The number of errors is held as the value USR_ER [k]. For example, a sector with an error count of 0 (user sector with an offset parameter k = 2) has not caused an error so far, and a sector with an error count of 3 (user sector with an offset parameter k = 1) has already been 3 times. Indicates that an error has occurred. In the following description, a sector that normally operates when a write / erase is performed again even if a recording error has occurred once is referred to as an accidental error sector.
[0021]
FIG. 2C shows the spare sector address conversion table 15-3. In the spare sector address conversion table 15-3, physical addresses are stored in order of logical addresses so that the physical address RSV_PA [i] of the spare sector [i] corresponding to the offset parameter i which is a logical address can be obtained. The “reserved sector” is a sector for substitution in which user data is stored instead when a recording error occurs in the user sector. How to substitute will be described in detail with reference to FIG. Further, in addition to the physical address, the error count value RSV_ER [i] of the sector is stored. The spare sector error count value RSV_ER [i] takes a value of 0 or more. As will be described later, since the spare sector functions as a user sector when data is recorded, it will be registered in the user sector address conversion table 15-2 in the future. Therefore, the spare sector must also hold the error count value RSV_ER [i].
[0022]
Since the address conversion tables 15-2 and 15-3 are stored in the flash memory cell array 15, information can be retained even after the power is turned off, and the retained information can be electrically erased.
[0023]
Next, processing for determining whether or not an error has occurred and reusing the error sector will be described. The process described below is mainly based on the control of the internal controller 10 (FIG. 1). FIG. 3 shows a flowchart of the rewrite process when a recording error occurs in the first embodiment. First, the logical address k of the user sector address where the error has occurred and the head sector number i of the spare sector are acquired (step S301). Next, it is determined whether or not the head number i of the spare sector exceeds a predetermined upper limit value, and the presence or absence of the spare sector is checked (step S302). If the start sector number i of the spare sector exceeds the reserved number of spare sectors, the spare sector is used up, and no further spare sector can be assigned. Therefore, it ends in error. On the other hand, if the head number i of the spare sector does not exceed the predetermined upper limit value, the error count value USR_ER is calculated from the user sector address k with reference to the user sector address conversion table 15-2 ((b) of FIG. 2). [k] is acquired (step S303). Then, based on the error count value USR_ER [k], it is determined whether or not the error is a random error (step S304).
[0024]
The determination of whether or not it is a random error will be described with two specific examples. As a first example, whether or not it is an accidental error is determined by whether or not the error count value USR_ER [k] has reached a specified number of times, that is, whether or not USR_ER [k] <the specified number of times. A sector in which errors have occurred more than the specified number of times is regarded as a bad sector and is not used thereafter. Although how many times the specified number of times is set can be determined as appropriate, it is, for example, 10 times.
[0025]
As a second example, the error count USR_ER [k] divided by the number of rewrites (recording error occurrence rate: USR_ER [k] / (number of rewrites))) reaches the specified value as to whether or not it is an accidental error. It is determined by whether or not it exists. That is, if the recording error occurrence rate is larger than the specified value, the sector is regarded as a bad sector and is not used thereafter. The degree to which the specified value is set can be determined as appropriate. For example, it is 0.20 (a value that causes one error for five writings).
[0026]
If the error is less than or equal to the specified number in the first example, or if the recording error occurrence rate is less than or equal to the specified value in the second example, it is regarded as an accidental error and the process proceeds to the next process. The next process is a process for reusing an error sector as a spare sector and replacing data recording with the spare sector. First, 1 is added to the error count value USR_ER [k] (step S305). Then, the physical address USR_PA [k] and error count value USR_ER [k] of the error sector are temporarily saved in the temporary variable temp (step S306). This is for later registration as a spare sector. Thereafter, the values of the physical address RSV_PA [i] and error count RSV_ER [i] of the spare sector number i are stored in the positions where USR_PA [k] and USR_ER [k] of the user sector address conversion table 15-2 are stored ( Write to offset k) (step S307). This process means that a spare sector is assigned as a user sector to be written. As a result, the replacement from the accidental error sector to the spare sector is completed at this point. Next, the sector in which the accidental error has occurred is registered as a spare sector so that it can be used again. Specifically, in the spare sector address conversion table 15-3, the physicality of the accidental error sector that was saved at the position (offset i) where RSV_PA [i] and RSV_ER [i] used for substitution were stored. The address USR_PA [k] and the error count value USR_ER [k] are written (step S308). By this process, the accidental error sector is registered as a spare sector to be used next. In other words, this process means that when it is determined that it is an accidental error, the random error sector and the spare sector to be used next are replaced. Thereafter, the process proceeds to step S311.
[0027]
On the other hand, as a result of the processing in step S304, if the sector is not determined to be an accidental error sector (that is, determined to be a bad sector), the sector is not used again, so the user sector address conversion table 15-2 is rewritten. Do. Specifically, the spare sector address RSV_PA [i] and the error count value RSV_ER [i] are assigned as the physical address USR_PA [k] and error count value USR_ER [k] of the user sector to be written (step S309). . As a result, the physical address indicated by USR_PA [k] immediately before rewriting is no longer registered in the user sector address conversion table 15-2, and the sector at the address is discarded. As a result of the processing in step S309, the spare sector of number i has been used as a user sector, so the head number i of an unused spare sector is incremented by 1 (step S310), and the process proceeds to step S311. The head number i of the unused spare sector is stored in the nonvolatile memory as a parameter.
[0028]
Next, the writing process that could not be normally completed due to an error is performed again on the user sector value USR_PA [k] replaced by the spare sector (step S311), and it is determined whether or not the writing process was successful (step S312). If the writing has been completed normally, the process ends. If the writing has not been completed normally, the process returns to step S302 again, and the processing from step S302 is performed again.
[0029]
According to the present embodiment, a sector in which a recording error has occurred and it is determined as an accidental error is replaced with a spare sector, and itself is registered as a spare sector to be used next. Whether the sector in which the recording error has occurred is an accidental error sector is determined based on the cumulative error occurrence count. A bad sector can be excluded because it always results in an error no matter how many times it is written. By reusing the accidental error sector, it is possible to prevent unnecessary consumption of the error sector and secure a storage capacity. Therefore, the lifetime of the apparatus can be extended and the cost can be reduced. Although an error occurs in the accidental error sector, since it has the same performance as a normal user sector at the time of re-recording, the reliability is high.
[0030]
(Embodiment 2)
In the second embodiment, an invention will be described in which error sectors are reused and error sectors are discarded by a method different from that of the first embodiment. In the first embodiment, the accidental error sector is registered as a spare sector to be used next. In the second embodiment, the random error sector is registered at the end of the spare sector.
[0031]
Also in the second embodiment, the configuration of the semiconductor memory device 1 of FIG. 1 and the data structure of the flash memory cell array 15 (FIG. 1) of FIG. 2 are used. However, since description of FIG. 1 and FIG. 2 was already performed in Embodiment 1, it abbreviate | omits below.
[0032]
FIG. 4 shows a flowchart of the rewrite process when a recording error occurs in the second embodiment. With reference to FIG. 4, a process for determining whether or not an error has occurred and reusing an error sector will be described. The process described below is mainly based on the control of the internal controller 10 (FIG. 1). First, the logical address k of the user sector address where the recording error has occurred and the number N of unused spare sectors are acquired (step S401). Next, it is determined whether or not the number of unused spare sectors N is greater than or equal to a predetermined one, and the presence or absence of spare sectors is checked (step S402). If the number of unused spare sectors N is 0, the spare sectors are used up, and no more spare sectors can be allocated. Therefore, it ends in error. On the other hand, when the number of unused spare sectors N is 1 or more, the error count value USR_ER [k] is obtained from the user sector address k with reference to the user sector address conversion table 15-2 ((b) in FIG. 2). (Step S403). Then, based on the error count value USR_ER [k], it is determined whether or not the error is a random error (step S404).
[0033]
The determination of whether or not it is an accidental error can be made based on either of the two examples specifically described in the first embodiment. That is, it may be performed based on whether or not the error count value USR_ER [k] has reached a specified number of times, or whether or not the recording error occurrence rate has reached a specified value. These explanations are omitted because they are the first embodiment.
[0034]
When it is determined as an accidental error, the error sector is reused as a spare sector, and the process proceeds to a process for substituting data recording with the spare sector. First, 1 is added to the error count USR_ER [k] (step S405). Then, the physical address USR_PA [k] and the error count USR_ER [k] are temporarily saved in the temporary variable temp (step S406). Then, the physical address RSV_PA [0] and the error count RSV_ER [0] of the spare sector number 0 are set to the positions (offset k) where USR_PA [k] and USR_ER [k] of the user sector address conversion table 15-2 are stored. ) (Step S407). This process means that a spare sector is assigned as a user sector to be written. The reason “spare sector number 0” is that the spare sector always uses the 0th. As a result, the replacement to the spare sector is completed at this point.
[0035]
Next, the spare sector address conversion table is updated in order to register the sector causing the accidental error as a spare sector so that it can be used again. Since the spare sector always uses the 0th sector, after the spare sector is used, the spare sector is shifted one by one so that the next spare sector becomes the 0th sector (step S408). Further, the sector in which the accidental error has occurred is registered at the end of the spare sector so that it can be used again (step S409). Specifically, the physical address USR_PA [k] and the error count USR_ER [k] saved in step S406 are written in the offset parameter (N-1) of the spare sector address conversion table. By this processing, the accidental error sector is registered at the end of the spare sector, and can be prepared for future use. Thereafter, the process proceeds to step S413.
[0036]
On the other hand, if it is not determined as an accidental error sector (ie, it is determined as a bad sector) as a result of the processing in step S404, the sector is not used again, so the user sector address conversion table 15-2 (FIG. 2). ). Specifically, the spare sector address RSV_PA [0] and error count value RSV_ER [0] are assigned as the physical address USR_PA [k] and error count value USR_ER [k] of the user sector to be written (step S410). . As a result, the physical address indicated by USR_PA [k] immediately before rewriting is no longer registered in the user sector address conversion table 15-2 (FIG. 2), and the sector at the address is discarded. As a result of the processing in step S410, the spare sector number 0 has been used as a user sector, so that the spare sector number is shifted forward by one so that the next spare sector becomes 0th (step S411). . Since the number of unused spare sectors N is decreased by one, N−1 is set (step S412). The number of unused spare sectors N is stored as a parameter on the nonvolatile memory.
[0037]
Next, the writing process that could not be normally completed due to a recording error is performed again on the user sector value USR_PA [k] replaced by the spare sector (step S413), and it is determined whether the writing process was successful (step S414). . If the writing has been completed normally, the process ends. If the writing has not been completed normally, the process returns to step S402 again, and the processing from step S402 is performed again.
[0038]
According to the present embodiment, a sector in which a recording error has occurred and it is determined as an accidental error is replaced with a spare sector, and itself is registered at the end of the spare sector for future use. Whether the sector in which the recording error has occurred is an accidental error sector is determined based on the cumulative error occurrence count. A bad sector can be excluded because it always results in an error no matter how many times it is written. By reusing the accidental error sector, it is possible to prevent unnecessary consumption of the sector and to secure a storage capacity. Therefore, the lifetime of the apparatus is extended and cost reduction is realized. As described in the first embodiment, the reliability of the accidental error sector is high.
[0039]
The processing flow of the semiconductor memory device 1 (FIG. 1) described with reference to FIGS. 3 and 4 is also realized as a program that operates in this way. Such a program is executed by the internal controller 10 (FIG. 1).
[0040]
【The invention's effect】
Whether or not data is to be recorded later is determined based on the error count value of the error sector defined in the user sector table for an error sector in which a recording error has occurred among user sectors. As a result, even in a sector in which a recording error has occurred, data is reused for recording, so that it is possible to prevent unnecessary consumption of the sector and secure a storage capacity. Further, it is possible to extend the life of the apparatus with high reliability and to reduce the cost.
[0041]
A spare sector is provided as an alternative sector for recording data when a recording error occurs. When data is recorded later, the error sector is changed to a spare sector. As a result, when a recording error occurs later, data can be recorded in the error sector, and the error sector can be used effectively.
[0042]
A spare sector table defining an error count value indicating the number of recording errors in the spare sector is also provided for the spare sector. As a result, even if data is recorded in the spare sector later, it can function as a user sector by registering the error count value in the user sector table.
[0043]
In order to determine whether to record data later, the accumulated error count value is compared with a preset value. Thereby, when it is smaller than the preset value, the error sector can be effectively reused.
[0044]
The spare sector changed from the error sector is assigned the earliest order. Therefore, the error sector can be effectively reused.
[0045]
In order to determine whether to record data later, the ratio between the error count value of the error sector and the number of times of recording in the error sector until a recording error occurs is compared with a preset value. Thereby, when it is smaller than the preset value, the error sector can be effectively reused.
[0046]
The spare sector changed from the error sector is given the last order. Therefore, the error sector can be effectively reused.
[0047]
The order of spare sectors following the earliest spare sector is shifted forward. As a result, the use order of spare sectors can be managed.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a semiconductor memory device of the present invention having a nonvolatile memory cell.
FIG. 2A is a diagram showing a memory map of a flash memory. (B) is a figure which shows the user sector address conversion table by this invention. (C) is a diagram showing a spare sector address conversion table.
FIG. 3 is a flowchart of a rewrite process when a write error occurs in the first embodiment.
FIG. 4 is a flowchart of a rewrite process when a write error occurs in the second embodiment.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Semiconductor memory device, 10 Internal controller, 15 Flash memory cell array, 15-1 User data recording area, 15-2 User sector address conversion table, 15-3 Reserve sector address conversion table

Claims (7)

A plurality of user sectors for recording data;
A non-volatile memory cell having, for each user sector, an address indicating the position of the user sector, and a user sector table that defines an error count value indicating the number of recording errors in the user sector. A controller for determining whether to record data later, based on an error count value of the error sector defined in the user sector table, for an error sector in which a recording error has occurred among the user sectors ,
The nonvolatile memory cell further includes a spare sector for recording data instead of the error sector when a recording error occurs,
The semiconductor memory device , wherein the controller changes the error sector, which is determined to record data later, to a spare sector to be used later . The non-volatile memory cell further includes a spare sector table that defines, for each spare sector, an address indicating the position of the spare sector and an error count value indicating the number of recording errors in the spare sector in association with each other. The semiconductor memory device according to claim 1 . If the error count value of the error sector is smaller than a preset value, the controller determines that data will be recorded later in the error sector, and if larger, data is recorded later in the error sector. The semiconductor memory device according to claim 2 , wherein it is determined not to be performed. In the spare sector table, the order of spare sectors to be used is attached,
If it is determined that data is to be recorded later in the error sector, the controller records data in the earliest spare sector based on the order, and the spare sector changed from the error sector has the earliest order. The semiconductor memory device according to claim 3 , wherein If the ratio between the error count value of the error sector and the number of times of recording in the error sector until a recording error occurs is smaller than a preset value, the controller transfers data to the error sector later. The semiconductor memory device according to claim 2 , wherein it is determined to be recorded, and when it is large, it is determined that data is not recorded later in the error sector. The spare sector table has an order in which spare sectors are used,
If the controller determines that data is to be recorded later in the error sector, the controller records data in the earliest spare sector based on the order, and the spare sector changed from the error sector has the last order. The semiconductor memory device according to claim 5 , wherein The semiconductor memory device according to claim 6 , wherein the controller shifts the order of spare sectors subsequent to the earliest spare sector forward.

Images