JP2014035747A - Semiconductor memory device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a semiconductor memory device capable of storing management information in a memory to be controlled while preventing a situation in which a normal operation is impeded due to the existence of defective blocks, a forced power shutdown, or the like.SOLUTION: Management information includes: first management information that is stored in a first block group R1; and second management information that is stored in a second block group R2. The second management information includes restoration information for restoring appropriate first management information in the first block group R1 when the appropriate first management information is not identified in the first block group R1.

Description

  The present invention relates to a semiconductor memory device.

  A memory system including a memory controller and a memory cell array such as a flash memory has been put into practical use. In a NAND flash memory or the like, there are randomly acquired congenital defective blocks that occur at the time of manufacture, and acquired defective blocks that occur due to normal use after shipment. Since it is necessary to avoid these bad blocks and use them, the position of the bad block is managed by predetermined management information.

  If you try to store management information in the memory cell array that is the memory to be controlled, if the storage block is a congenital defective block, or it was not a congenital defective block at the time of storage, it will later become an acquired defective block. In this case, some or all of the management information data may be destroyed or lost. Also, if the power is forcibly shut down due to a user's mistaken operation while writing new management information to update the management information, some or all of the data in the management information is destroyed or lost. There is a possibility of being. If the management information is destroyed or lost, then the normal operation of the memory system is hindered.

  Therefore, in the memory system according to the background art, by preparing a dedicated non-volatile memory different from the control target memory and storing the management information in the non-volatile memory, destruction of the management information is prevented. . As another method, management information is stored in a non-volatile memory area in the memory controller, or management information is given to the memory controller by setting a DIP switch, thereby preventing destruction of the management information.

  Patent Document 1 listed below discloses a memory system including a controller and a flash memory. Management information including a list of defective blocks is divided into a plurality of divided pieces, and each divided piece is individually stored in the flash memory. A management memory using a volatile memory such as a DRAM is mounted in the controller. Management information is stored in the management memory by reading a plurality of divided pieces from the flash memory to the management memory.

JP 2011-59866 A

  As described above, in the memory system according to the background art, management information is stored in a dedicated non-volatile memory different from the control target memory. Therefore, since it is necessary to add a dedicated non-volatile memory in addition to the control target memory, the manufacturing cost increases and the control becomes complicated.

  The present invention has been made to solve such a problem, and management information is controlled while avoiding a situation in which normal operation is hindered due to the presence of a defective block or forced power shutdown. An object is to obtain a semiconductor memory device that can be stored in a memory.

  A semiconductor memory device according to a first aspect of the present invention includes a storage unit having a plurality of blocks, each of which is a data erasing unit, and a control unit that controls the storage unit, and all of the storage unit has A first block group consisting of a plurality of first predetermined number of blocks, which is assigned to the blocks in a limited manner as blocks for storing management information necessary for the control unit to manage the storage unit; , A second block group other than the first block group, and management information includes first management information stored in the first block group, second management information stored in the second block group, and The second management information includes restoration information for restoring the appropriate first management information in the first block group when the appropriate first management information cannot be specified in the first block group. It is characterized by being included Than is.

  According to the semiconductor memory device of the first aspect, the management information includes the first management information stored in the first block group and the second management information stored in the second block group. The second management information includes restoration information for restoring the appropriate first management information in the first block group when the appropriate first management information cannot be specified in the first block group. . Therefore, even when it is not possible to specify the appropriate first management information in the first block group, the control unit determines the appropriate first management information based on the restoration information included in the second management information. It can be restored within one block group. As a result, it is possible to store the management information in the control target memory while avoiding the situation where the normal operation is hindered due to the presence of a defective block or forced power cut-off.

  The semiconductor memory device according to the second aspect of the present invention is characterized in that, in the semiconductor memory device according to the first aspect, the second management information is information that is rewritten more frequently than the first management information. Is.

  According to the semiconductor memory device according to the second aspect, the second management information is information that is rewritten more frequently than the first management information. By dividing the management information into the first management information and the second management information according to the rewrite frequency, and storing the first management information with a low rewrite frequency in the limited first block group, Data rewriting does not occur frequently in the first block group. As a result, since the occurrence of acquired defective blocks in the first block group is suppressed, the reliability of the apparatus can be improved.

  In the semiconductor memory device according to the third aspect of the present invention, in the semiconductor memory device according to the first or second aspect, the first management information includes, in the first management information, positional information indicating a block storing the second management information. And the second management information includes position information indicating a defective block.

  According to the semiconductor memory device of the third aspect, the first management information includes position information indicating a block storing the second management information, and the second management information includes a position indicating a defective block. Contains information. Therefore, the control unit can acquire the position information indicating the block storing the second management information by reading the first management information from the storage unit, and stores the second management information based on the position information. By reading from the part, it is possible to acquire position information indicating a defective block.

  In the semiconductor memory device according to the fourth aspect of the present invention, particularly in the semiconductor memory device according to the third aspect, the first management information includes position information indicating a defective block in the first block group. The 2 management information includes position information indicating defective blocks in the second block group.

  According to the semiconductor memory device of the fourth aspect, the first management information includes position information indicating a defective block in the first block group, and the second management information includes a defect in the second block group. Position information indicating a block is included. Therefore, when an acquired defective block occurs in the second block group, only the second management information needs to be updated, and the update of the first management information can be omitted.

  The semiconductor memory device according to the fifth aspect of the present invention is the semiconductor memory device according to any one of the first to fourth aspects, in particular, a plurality of second predetermined number of blocks in the second block group. The second management information having the same contents is stored in each of the above.

  According to the semiconductor memory device of the fifth aspect, in the second block group, the second management information having the same contents is stored in each of the plurality of second predetermined number of blocks. Therefore, even if one of the second predetermined number of second management information stored in the second predetermined number of blocks is destroyed or lost, the control unit is not destroyed or lost. The remaining second management information can be acquired from the storage unit. As a result, it is possible to store the management information in the control target memory while avoiding the situation where the normal operation is hindered due to the presence of a defective block or forced power cut-off.

  A semiconductor memory device according to a sixth aspect of the present invention is the semiconductor memory device according to the fifth aspect, in particular, the control unit includes a plurality of third predetermined numbers in which normal second management information is stored. Normal blocks are identified from the second block group, and one appropriate first block is determined by majority from the third predetermined number of second management information stored in the third predetermined number of normal blocks. The second management information is specified, and the valid first management information is restored based on the restoration information included in the valid second management information.

  According to the semiconductor memory device of the sixth aspect, the control unit specifies the third predetermined number of normal blocks from the second block group. Then, from the third predetermined number of second management information stored in the specified third predetermined number of normal blocks, one appropriate second management information is specified by majority vote. Therefore, even when multiple versions of the second management information are stored in the second block group by repeating the update, it is possible to appropriately specify the appropriate second management information.

  A semiconductor memory device according to a seventh aspect of the present invention is the semiconductor memory device according to the sixth aspect, in which the control unit stores the second management information from all the blocks stored therein. The normal block is specified by excluding the abnormal block in which the data error has occurred in the second management information.

  According to the semiconductor memory device of the seventh aspect, the control unit identifies the normal block by excluding the abnormal block from all the blocks storing the second management information. Accordingly, it is possible to avoid a situation in which an abnormal block is specified as a normal block, and it is possible to appropriately specify appropriate second management information.

  The semiconductor memory device according to an eighth aspect of the present invention is the semiconductor memory device according to the sixth or seventh aspect, in particular, the control unit includes three normal blocks among the third predetermined number of normal blocks. , And one appropriate second management information is specified by a majority vote from the three pieces of second management information stored in the three normal blocks.

  According to the semiconductor memory device of the eighth aspect, the control unit specifies three normal blocks from the third predetermined number of normal blocks, and the three stored in the three normal blocks. From the second management information, one appropriate second management information is specified by majority vote. In this way, by limiting the target of the majority decision to the three pieces of second management information, the majority decision can be easily made, and thus the time required for processing can be shortened.

  A semiconductor memory device according to a ninth aspect of the present invention is the semiconductor memory device according to the eighth aspect, in particular, the control unit includes second management information stored in a certain normal block, 2. Majority voting is performed using two pieces of second management information stored in two specific blocks in which position information is described in two pieces of management information.

  According to the semiconductor memory device of the ninth aspect, the control unit includes the second management information stored in a certain normal block and the specific 2 in which the position information is described in the second management information. A majority decision is made using two pieces of second management information stored in each block. Therefore, since the target of majority decision can be limited to the second management information having the same contents written together at the time of update, it is possible to increase the possibility of specifying appropriate second management information by majority decision.

  In the semiconductor memory device according to the tenth aspect of the present invention, in particular, in the semiconductor memory device according to the eighth or ninth aspect, the control unit is appropriate depending on the majority vote for the three specified normal blocks. When the second management information cannot be specified, the other three normal blocks are specified from the third predetermined number of normal blocks, and the three first blocks stored in the other three normal blocks are specified. Among the two management information, one appropriate second management information is specified by majority vote.

  According to the semiconductor memory device of the tenth aspect, when the control unit cannot identify the appropriate second management information by the majority vote for the identified three normal blocks, the other three normal Re-identify the block and make a majority vote. Thereby, it is possible to increase the possibility of specifying appropriate second management information.

  The semiconductor memory device according to the eleventh aspect of the present invention is the semiconductor memory device according to any one of the fifth to tenth aspects, particularly when the control unit needs to update the second management information. Is characterized in that the updated second management information is written in each of the second predetermined number of blocks in which the second management information before the update is stored.

  According to the semiconductor memory device of the eleventh aspect, when it becomes necessary to update the second management information, the control unit stores the second predetermined number of second management information before the update. The updated second management information is written in each block. Thereby, even after the update, a second predetermined number of second management information having the same contents can be stored in the second block group.

  A semiconductor memory device according to a twelfth aspect of the present invention is the semiconductor memory device according to the eleventh aspect, in particular, the control unit receives version information having a different value every time the second management information is updated, It is characterized in that it is included and written.

  According to the semiconductor memory device in the twelfth aspect, the control unit writes the version information having a different value for each update included in the second management information. Therefore, a majority decision can be made using the version information, and as a result, it is possible to improve the accuracy of appropriately specifying appropriate second management information.

  A semiconductor memory device according to a thirteenth aspect of the present invention is the semiconductor memory device according to the twelfth aspect, particularly when the control unit writes the second management information in a certain block as an acquired defective block. In this case, the version information value included in the second management information is not used in subsequent updates.

  According to the semiconductor memory device of the thirteenth aspect, when the second management information is written in a certain block, the control unit includes the acquired second management information in the case where the block becomes an acquired defective block. The version information value is not used for subsequent updates. Therefore, even if the counter value for generating the version information value makes a round, the second management information stored in the acquired defective block is erroneously specified as valid second management information by majority vote. It is possible to avoid this situation.

  A semiconductor memory device according to a fourteenth aspect of the present invention is the semiconductor memory device according to the eleventh aspect, in which the control unit includes time information related to the update time of the second management information in the second management information. It is characterized by writing.

  According to the semiconductor memory device in the fourteenth aspect, the control unit writes the time information related to the update time of the second management information in the second management information. Therefore, the latest second management information can be specified based on the time information, and as a result, it is possible to improve the accuracy of appropriately specifying appropriate second management information.

  The semiconductor memory device according to the fifteenth aspect of the present invention is the semiconductor memory device according to any one of the eleventh to fourteenth aspects, in particular, the control unit sets the error detection code of the second management information to the second It is characterized in that it is written in the management information.

  According to the semiconductor memory device in the fifteenth aspect, the control unit writes the error detection code of the second management information by including it in the second management information. Therefore, a majority decision can be made using the error detection code, and as a result, it is possible to improve the accuracy of appropriately identifying the appropriate second management information. In addition, it is possible to determine whether or not a data error has occurred using an error detection code. As a result, when a normal block is specified from the second block group, an abnormality in which a data error has occurred Blocks can be excluded.

  A semiconductor memory device according to a sixteenth aspect of the present invention is the semiconductor memory device according to any one of the first to fifteenth aspects, and in particular, the control unit is one of the second predetermined number of blocks. The second management information to be written in the block is written by including position information indicating other blocks in the second predetermined number of blocks.

  According to the semiconductor memory device in the sixteenth aspect, the control unit adds the second management information to be written to one block in the second predetermined number of blocks to the other management information in the second predetermined number of blocks. Write including position information indicating the block. Therefore, since the target of majority decision can be limited to the second management information having the same contents written together at the time of update, it is possible to increase the possibility of specifying appropriate second management information by majority decision.

  According to the present invention, a semiconductor storage device capable of storing management information in a control target memory while avoiding a situation in which normal operation is hindered due to the presence of a defective block or forced power-off Can be obtained.

1 is a diagram showing a simplified configuration of a semiconductor memory device according to a first embodiment of the present invention. It is a figure which extracts and shows a part of memory space of control object memory. It is a figure which shows the kind of control object memory assumed to be used. It is a figure which shows a 1st block group. It is a figure which shows a 2nd block group. It is a figure which shows the allocation of the page in a block. It is a figure which shows the allocation of the page in a block. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the other example of a 1st block group. It is a figure which shows the other example of a 1st block group. It is a figure which shows the decompression | restoration process of 1st management information. It is a figure which shows the example of an update of 2nd management information. It is a figure which shows the example of an update of 2nd management information. It is a figure which shows a 1st block group. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the acquisition process of management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the update process of 1st management information in order. It is a figure which shows the other example of a 1st block group. It is a figure which shows the other example of a 1st block group. FIG. 10 is a diagram showing a simplified configuration of a semiconductor memory device according to Modification 1; It is a figure which shows the 1st example of the 2nd management information piece stored in the control object memory. It is a figure which shows the 2nd example of the 2nd management information piece stored in the control object memory.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, the element which attached | subjected the same code | symbol in different drawing shall show the same or corresponding element.

<Embodiment 1>
FIG. 1 is a diagram showing a simplified configuration of a semiconductor memory device 1 according to the first embodiment of the present invention. The semiconductor storage device 1 includes a memory controller 2 (control unit) and a control target memory 3 (storage unit) controlled by the memory controller 2. The control target memory 3 is a non-volatile memory, and is composed of, for example, a NAND flash memory.

  FIG. 2 is a diagram illustrating a part of the memory space of the control target memory 3. The memory space is divided into a plurality of blocks B1 to BN. A block is a unit for erasing data. All the blocks B1 to BN included in the control target memory 3 are classified into a first block group R1 and a second block group R2.

  The first block group R1 is a set of blocks for storing first management information to be described later, and a specific plurality of blocks limited among all the blocks B1 to BN are assigned as the first block group R1. It has been. In the example of the present embodiment, a total of 10 blocks B1 to B10 are allocated as the first block group R1.

  The second block group R2 is a set of blocks other than the first block group R1. Arbitrary data used by the user is stored in the second block group R2. The second block group R2 stores second management information to be described later.

  In the semiconductor memory device 1 according to the present embodiment, the first management information having the same content is stored in three blocks selected from the ten blocks B1 to B10 belonging to the first block group R1. However, the values of “10” and “3” are examples, and are not limited to these values. Further, the second management information having the same contents is stored in three blocks arbitrarily selected from all the blocks belonging to the second block group R2. However, the value “3” is an example, and is not limited to this value.

  The management information is information necessary for the memory controller 2 to manage the control target memory 3. In the example of the present embodiment, the management information is divided into first management information and second management information.

  The first management information includes position information indicating the locations of the three blocks storing the second management information. As the position information, for example, a block address is used.

  The first management information includes an error detection code. The error detection code is used to determine whether or not a data error has occurred in the first management information. Further, the error detection code is used for majority decision (details will be described later) for specifying appropriate first management information. As an error detection code, a checksum, CRC (Cyclic Redundancy Check), message digest, or the like can be used. In the example of the present embodiment, CRC is used as the error detection code.

  The first management information includes unique ID information (referred to as “first management information ID” in this specification). The first management information ID indicates the version of the first management information, and the value of the first management information ID is updated every time the first management information is updated. In the example of the present embodiment, an integer value incremented by “1” each time the first management information is updated is used as the first management information ID. The first management information ID is used for a majority decision (details will be described later) for specifying appropriate first management information.

  The first management information includes position information indicating the location of the defective block when there is a defective block in the first block group R1. The defective blocks include congenital defective blocks that occur when the control target memory 3 is manufactured and acquired defective blocks that occur due to normal use after shipment.

  When the first management information is updated, the updated first management information is written in a block different from the block storing the first management information before the update in the first block group R1. When the corresponding block becomes an acquired defective block during writing, the updated first management information is written to another block in the first block group R1.

  The second management information includes position information indicating the location of the defective block when there is a defective block in the second block group R2. The second management information includes various information necessary for the memory controller 2 to manage the control target memory 3, such as the location of the general-purpose area and the access speed to the control target memory 3.

  The second management information includes unique ID information (referred to as “second management information ID” in this specification). The second management information ID indicates the version of the second management information, and the value of the second management information ID is updated every time the second management information is updated. In the example of the present embodiment, an integer value incremented by “1” each time the second management information is updated is used as the second management information ID. The second management information ID is used for a majority vote (details will be described later) for specifying appropriate second management information.

  The number of bits of the second management information ID value is secured to a sufficiently large size so as not to overflow practically. As a result, the second management information having the largest second management information ID value is replaced with the latest second management information. Can be determined. Further, when updating the second management information, the absolute time itself representing the update time or a value calculated from the absolute time is used as the second management information ID value, so that the second management information can be selected from the plurality of second management information. The latest second management information can be easily identified.

  Further, the second management information includes position information (referred to as “group address” in this specification) indicating the other two blocks in which the second management information having the same content as the second management information is stored.

  The second management information includes an error detection code. In the example of the present embodiment, the entire CRC value including the group address and the CRC value excluding the group address are included in the second management information. The entire CRC value including the group address is used to determine whether or not a data error has occurred in the second management information. Further, the CRC value excluding the group address is used for majority voting for specifying appropriate second management information.

  The second management information includes restoration information for restoring the appropriate first management information in the first block group R1 when the appropriate first management information cannot be specified in the first block group R1. included. Among various information included in the corresponding first management information, the first management information ID and the position information of the defective block in the first block group R1 are included in the restoration information. These pieces of information may be copied in plain text and included in the restoration information, or compressed information may be included in the restoration information.

  When the second management information is updated, the updated second management information is overwritten on the same block as the block storing the second management information before the update in the second block group R2. When the corresponding block becomes an acquired defective block during writing, the updated second management information is written in another arbitrary block (excluding the defective block) in the second block group R2. Since the first management information includes the position information of the block storing the second management information, it is necessary to update the first management information when the block storing the second management information is changed. There is.

  From the nature of the information included in each management information, it can be said that the second management information has a higher update frequency (rewrite frequency) than the first management information. For example, when an acquired bad block occurs in the user data area in the second block group R2, the second management information needs to be updated, but the first management information does not need to be updated. Since the number of blocks in the second block group R2 is much larger than the number of blocks in the first block group R1, the probability that an acquired defective block will occur in the second block group R2 is acquired in the first block group R1. It is sufficiently higher than the probability that a bad block will occur. Therefore, the situation of updating the second management information without updating the first management information occurs more frequently than the opposite situation.

  FIG. 3 is a diagram illustrating the types of the control target memory 3 that are assumed to be used. In the present embodiment, it is assumed that a plurality of types (four types in this example) of memories having different page sizes, block sizes, total block numbers, read cycles, and innate defect flag positions are assumed. Each memory is given unique ID information (referred to as “memory ID” in this specification) for identifying the type of memory. The memory ID includes page size and block size information.

  Hereinafter, (1) management information storage example, (2) management information acquisition process, and (3) management information update process will be described in order with respect to the semiconductor memory device 1 according to the present embodiment.

(1) Storage Example of Management Information FIG. 4 is a diagram illustrating the first block group R1. In this example, blocks B1 and B5 are congenital defective blocks, and blocks B2 and B4 are invalid blocks (details will be described later).

  Blocks B3, B6 to B9 store first management information. The numbers in parentheses attached to the first management information indicate the first management information ID value (left side) and the CRC value (right side). For convenience of explanation, it is assumed that the CRC value when the first management information is written has the same value as the first management information ID value. For a normal block in which no data error has occurred, the CRC value matches the first management information ID value, while an abnormal block in which a data error has occurred due to an acquired bad block or the like With respect to, the CRC value does not match the first management information ID value. In this example, regarding the first management information (2, 1) stored in the block B6, the CRC value and the first management information ID value do not match. Therefore, the block B6 is an acquired defective block.

  In the example shown in FIG. 4, the latest first management information is the first management information (4, 4). Here, the first management information (4, 4) is stored only in the two blocks B8 and B9. The reason is that after the data in the block B10 is erased, the first management information (4, 4) is stored. This is because it is assumed that the power supply is forcibly cut off by the user before 4) is written. Therefore, the block B10 is an erased block.

  Further, “BAD” written in the block B3 indicates that the block B3 has become an acquired defective block when the first management information is written in the block B3. However, regarding the first management information (1, 1) stored in the block B3, the first management information ID value matches the CRC value. Therefore, it cannot be determined that the block B3 is an acquired defect block by comparing the first management information ID value and the CRC value.

  When the first management information is written into the block and the block becomes an acquired defect block, the memory controller 2 holds the first management information ID value attached to the first management information, Thereafter, when the first management information is updated, the first management information ID value may not be used. Thus, when the first management information ID value is generated using the cyclic counter, the old first management information stored in the acquired defective block and the new first management information updated thereafter Thus, a situation in which the first management information ID values coincide by chance is avoided.

  FIG. 5 is a diagram illustrating the second block group R2. In this example, the second management information is stored in blocks B100, B150, B200, B250, B300, and B350. The second management information having the same contents is stored in the first block set including the blocks B100, B200, and B300, and the second management information having the same contents is stored in the second block set including the blocks B150, B250, and B350. The usage mode is assumed. The numbers in parentheses attached to the second management information indicate the second management information ID value (left side) and the entire CRC value (right side). Numbers in angle brackets attached to the second management information indicate group addresses. For example, when the second management information (3, 3) is written in the block B150, it means that the second management information (3, 3) having the same contents is written in the blocks B250, B350. Similarly to the above, the CRC value of the entire area when writing the second management information is assumed to be the same value as the second management information ID value.

  In the example shown in FIG. 5, the latest second management information is the second management information (4, 4), and the second newest second management information is the second management information (3, 3). The second management information (3, 3) is stored in the three blocks B150, B250, B350. The latest second management information (4, 4) is stored in only one block of the block B100. The reason is that after writing the second management information (4, 4) to the block B100, This is because it is assumed that the power is forcibly shut off by the user before erasing the data in the blocks B200, 300 to write the second management information (4, 4) in the blocks B200, 300.

  Similarly to the above, when the second management information is written into the block, if the block becomes an acquired defective block, the memory controller 2 sets the second management information ID value attached to the second management information. The second management information ID value may not be used when the second management information is updated thereafter. Accordingly, when the second management information ID value is generated using the cyclic counter, the old second management information stored in the acquired defective block and the new second management information updated thereafter Thus, a situation in which the second management information ID values coincide by chance is avoided.

  FIG. 6 is a diagram illustrating allocation of pages in a block with respect to each of the blocks B1 to B10. One block is divided into a plurality of pages. A page is a unit for writing and reading data. In this example, page P0 is a page to which a congenital defect flag is added, page PN is a page to which an invalid flag is added, and page PM is a page to which first management information is written. The first management information may be stored on a plurality of pages in the block.

  When a certain block is a congenital defective block, a congenital defective flag is added to page P0 of the block. The page to which the congenital defect flag is added differs depending on the type of the control target memory 3 as shown in FIG.

  When data of a certain block is invalid, an invalid flag is added to the page PN of that block. The page to which the invalid flag is added is common regardless of the type of the control target memory 3.

  FIG. 7 is a diagram illustrating allocation of pages in a block with respect to the block in which the second management information is stored. In the example of the present embodiment, the second management information is stored in the first three pages (pages P0 to P2) of the block.

  The second management information ID having the same content is stored at the head of each page P0 to P2. Thereby, the memory controller 2 can search for a block in which the second management information is stored from all the blocks belonging to the second block group R2.

  The page P0 stores restoration information among various pieces of information included in the second management information.

  Of the various information included in the second management information, the page P2 stores a CRC value of a part excluding the group address, two group addresses, and a CRC value of the entire area including the group address. Yes.

  Of the various information included in the second management information, the page P1 stores other management information that is not stored in the pages P0 and P2.

(2) Management Information Acquisition Processing FIGS. 8 to 13 are diagrams sequentially illustrating management information acquisition processing executed by the memory controller 2 when the semiconductor memory device 1 is activated. It is assumed that the first block group R1 is in the state shown in FIG.

  First, the memory controller 2 acquires a memory ID from the control target memory 3 by issuing a read command requesting the memory ID to the control target memory 3. Since the memory ID includes page size and block size information, an arbitrary page in the control target memory 3 can be accessed by acquiring the memory ID.

<Step of searching for congenital defective blocks>
Next, referring to FIG. 8, the memory controller 2 searches for all congenital defective blocks for all the blocks B1 to B10 belonging to the first block group R1. Specifically, for each of the blocks B1 to B10, it is determined whether or not each block is a congenital defective block by accessing the storage page for congenital defective flags (page P0 shown in FIG. 6) To do.

  At this time, since the type of the control target memory 3 is not known, the storage page of the congenital defect flag cannot be uniquely specified. Therefore, the memory controller 2 sequentially accesses the storage pages of the congenital defect flags for all types of memories that are assumed to be used. In the example shown in FIG. 3, the memory controller 2 sequentially accesses the 0th byte of the 0th page, the 2048th byte of the 0th page, the 0th byte of the 128th page, and the 8192th byte of the 256th page. Thus, it is determined whether or not a congenital defect flag is added to any storage page. In addition, the access speed to the control target memory 3 at that time is set to the slowest speed at which all types of memories assumed to be used can be normally operated. In the example shown in FIG. 3, access is performed in the slowest read cycle of 50 ns among the four types of memories.

  Note that table information that associates the memory ID with the memory type is created in advance and stored in the memory controller 2, and the memory controller 2 obtains the memory ID from the control target memory 3 and then refers to the table information. The memory type may be specified. In this case, it is possible to determine whether or not a congenital defect flag is added by accessing a storage page corresponding to the specified memory type at an access speed corresponding to the specified memory type.

  As a result of the search, if a congenital defective block exists in the blocks B1 to B10, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 8, blocks B1 and B5 that are congenital defective blocks are excluded from candidates. The blocks B2 to B4 and B6 to B10 that remain without being excluded in this step are referred to as “good blocks” in this specification.

<Invalid block search step>
Next, referring to FIG. 9, the memory controller 2 searches the invalid blocks B2 to B4 and B6 to B10 for invalid blocks. Specifically, with respect to each good block, it is determined whether or not each good block is an invalid block by accessing an invalid flag storage page (page PN shown in FIG. 6).

  Similarly to the above, the access speed to the control target memory 3 is set to the slowest speed at which all types of memories assumed to be used can be normally operated. However, when the memory type can be specified based on the memory ID, the access may be performed at an appropriate access speed corresponding to the specified memory type.

  If there is an invalid block in the good block as a result of the search, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 9, blocks B2 and B4 that are invalid blocks are excluded from the candidates. The blocks B3, B6 to B10 that remain without being excluded in this step are referred to as “effective blocks” in this specification.

<Abnormal block search step>
Next, referring to FIG. 10, the memory controller 2 searches the valid blocks B3 and B6 to B10 for a block in which a data error has occurred. Specifically, for each valid block, the first management information storage page (page PM shown in FIG. 6) is accessed to read the first management information, and the CRC calculation is performed on the read first management information. . Then, the CRC value obtained by the calculation is compared with the CRC value included in the read first management information. If both CRC values match, the memory controller 2 determines that the block is a block in which no data error has occurred (referred to herein as a “normal block”), while both CRC values are If they do not match, the block is determined to be a block in which a data error has occurred (referred to herein as an “abnormal block”).
・ Acquired bad block at the time of writing or erasing ・ It was not completed normally due to the power being forcibly cut off at the time of writing or erasing ・ Both CRC values because no data was written in the erased state If they do not match, the block is determined to be an abnormal block.

  Similarly to the above, the access speed to the control target memory 3 is set to the slowest speed at which all types of memories assumed to be used can be normally operated. However, when the memory type can be specified based on the memory ID, the access may be performed at an appropriate access speed corresponding to the specified memory type.

  As a result of the search, when an abnormal block exists in the effective block, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 10, blocks B6 and B10 that are abnormal blocks are excluded from the candidates.

<Appropriate first management information specifying step>
Next, the memory controller 2 arbitrarily selects four normal blocks from the remaining normal blocks. For example, four normal blocks are selected in ascending order of block addresses. In this example, since the remaining normal blocks are the four blocks B3, B7 to B9, all normal blocks are selected. If the number of normal blocks remaining at this time is two or less, it is determined that the control target memory 3 is faulty because the majority decision described later cannot be performed. If there are three normal blocks remaining at this time, a majority decision described later is performed only once.

  Next, referring to FIG. 11, the memory controller 2 arbitrarily selects three normal blocks from the selected four normal blocks B3, B7 to B9. In this example, three normal blocks B3, B7, B8 are selected in ascending order of block address. Then, the first majority decision is performed by comparing the first management information ID value and the CRC value included in the first management information stored in each of the normal blocks B3, B7, and B8. When two or more of the three pieces of first management information match the first management information ID value and the CRC value, the memory controller 2 determines that the first management information is valid. In this example, the first management information ID value and the CRC value of the three pieces of first management information are (1, 1) (3, 3) (4, 4), respectively, which are not consistent with each other. 1 It is determined that there is no management information.

  Next, referring to FIG. 12, the memory controller 2 arbitrarily selects the other three normal blocks from the selected four normal blocks B3, B7 to B9. In this example, three normal blocks B7 to B9 are selected in descending order of block address. Then, the second majority decision is performed by comparing the first management information ID value and the CRC value included in the first management information stored in each of the normal blocks B7 to B9. In this example, the first management information ID value and CRC value of the three pieces of first management information are (3, 3) (4, 4) (4, 4), and (4, 4) is two or more. Match. Therefore, it is determined that the first management information stored in the blocks B8 and B9 is appropriate first management information.

  In addition, when the appropriate first management information can be specified by the first majority decision, the second majority decision is omitted. In addition, when the appropriate first management information cannot be specified even by the second majority decision, the memory controller 2 determines that the control target memory 3 is temporarily faulty. For the control target memory 3 tentatively determined to be a failure, an appropriate first management information restoration process (details will be described later) is executed. If the appropriate first management information cannot be restored even after executing the restoration process, the control target memory 3 is definitely determined to be a failure. If five or more normal blocks remain, the selection of four normal blocks may be redone and the majority decision may be made again.

<Second management information acquisition step>
Next, referring to FIG. 13, the memory controller 2 determines one arbitrary first management information from the two or more specified appropriate first management information. For example, the block B8 having the smallest block address is selected from the blocks B8 and B9, and the first management information (4, 4) stored in the block B8 is determined as appropriate first management information.

  Next, the memory controller 2 determines the second block group R2 based on the position information (position information indicating the location of the block storing the second management information) included in the determined valid first management information. The second management information is acquired by reading the second management information from the list.

(3) Management information update process When the storage block of the second management information becomes an acquired defect block, the storage block of the second management information is changed to another block in the second block group R2. There is a need. Here, since the first management information includes the location information of the storage block of the second management information, when the storage block of the second management information is changed, it is necessary to update the contents of the first management information. There is. In addition, since the first management information includes the position information of the defective block in the first block group R1, when an acquired defective block occurs in the first block group R1, the first management information The content needs to be updated. Hereinafter, a procedure for updating the first management information will be described.

  14 to 18 are diagrams sequentially illustrating the update process of the first management information executed by the memory controller 2.

  Referring to FIG. 14, the currently effective first management information (that is, the first management information before update) is the first management information (4, 4) stored in blocks B8 and B9. Further, the first management information that is a candidate for majority decision for specifying the appropriate first management information at the time of activation is the first management information stored in the blocks B7 to B9. Hereinafter, the blocks B7 to B9 are referred to as “majority target blocks”.

  Next, referring to FIG. 15, the memory controller 2 secures an empty block in which the updated first management information is to be stored from the first block group R1. Specifically, from the blocks B1 to B10, congenital bad blocks, acquired bad blocks, and majority-target blocks are excluded, and three empty blocks are secured. Since the first management information includes the position information of the defective block in the first block group R1, by referring to the currently effective first management information (4, 4), the first management information in the first block group R1 Congenital bad blocks and acquired bad blocks can be identified. In this example, blocks B2 and B4 that are invalid blocks and a block B10 that is an erased block are secured as empty blocks. If the number of free blocks that can be secured is two or less, it is determined that the control target memory 3 is faulty.

  Next, referring to FIG. 16, the memory controller 2 erases the data of any one of the three empty blocks, and then updates the first management information (5, 5) after the update. Write to block. In this example, after the data in the block B10 is erased, the first management information (5, 5) is written into the block B10. When erasing the data of the empty block or writing the first management information (5, 5), if the corresponding block becomes an acquired defective block, the memory controller 2 stores the first management information and The location information of the defective block included in the second management information is updated, and the free block securing process is repeated.

  Next, referring to FIG. 17, the memory controller 2 adds an invalid flag to any one of the three majority-target blocks. In this example, an invalid flag is added to the block B7. If the corresponding block becomes an acquired defective block when the invalid flag is added, the memory controller 2 updates the position information of the defective block included in the first management information and the second management information. At the same time, the process starts from the process of securing the free block.

  Next, referring to FIG. 18, after erasing the data of the next empty block B2 among the three empty blocks, the memory controller 2 writes the first management information (5, 5) to the block B2. Thereafter, an invalid flag is added to the next block B8 among the three majority target blocks. Next, the memory controller 2 erases the data of the last empty block B4 among the three empty blocks, and then writes the first management information (5, 5) to the block B4. Thereafter, an invalid flag is added to the last block B9 among the three majority target blocks. As a result, the first management information (5, 5) having the same contents is stored in the three empty blocks B2, B4, B10.

  Note that data may be erased instead of adding an invalid flag to the majority decision target blocks B7 to B9. In this case, the writing of the first management information to one empty block and the erasure of data for one majority target block are alternately repeated three times.

<Specific Example 1 of Embodiment 1>
FIG. 19 is a diagram illustrating another example of the first block group R1. In this example, blocks B5 and B9 are congenital defective blocks, and blocks B1, B2 and B10 are invalid blocks. Blocks B4 and B8 are acquired defective blocks. However, regarding the blocks B4 and B8, the management information ID value matches the CRC value. The blocks B3, B6, and B7 store the first management information (6, 6).

  The management information acquisition process for the first block group R1 shown in FIG. 19 will be briefly described. First, blocks B5 and B9 that are congenital defective blocks are excluded. Next, blocks B1, B2, and B10 that are invalid blocks are excluded. Next, an abnormal block is searched, but since no abnormal block exists in this example, no block is excluded as an abnormal block. Next, blocks B3, B4, B6, and B7 are selected as four normal blocks.

  Next, the first majority decision is performed for the first three normal blocks B3, B4, and B6. As a result, the first management information (6, 6) stored in the blocks B3 and B6 is valid. It is specified as the first management information.

  Next, based on the first management information (6, 6) stored in one of the blocks B3, B6, the second management information is read out from the second block group R2.

<Specific Example 2 of Embodiment 1>
FIG. 20 is a diagram illustrating another example of the first block group R1. In this example, blocks B5 and B9 are congenital defective blocks, and blocks B1, B2 and B10 are invalid blocks. Blocks B4 and B6 are acquired defective blocks. However, regarding the blocks B4 and B6, the management information ID value matches the CRC value. The blocks B3, B7, B8 store the first management information (7, 7).

  The management information acquisition process for the first block group R1 shown in FIG. 20 will be briefly described. First, blocks B5 and B9 that are congenital defective blocks are excluded. Next, blocks B1, B2, and B10 that are invalid blocks are excluded. Next, an abnormal block is searched, but since no abnormal block exists in this example, no block is excluded as an abnormal block. Next, blocks B3, B4, B6, and B7 are selected as four normal blocks.

  Next, the first majority vote for the first three normal blocks B3, B4, and B6 is performed. However, since the majority vote is not established, it is determined that there is no valid first management information.

  Next, a second majority vote is performed for the last three normal blocks B4, B6, and B7. However, since the majority vote is not established, it is determined that there is no valid first management information. As a result, it is determined that the control target memory 3 is temporarily faulty.

<Restoration processing of valid first management information>
In the following, a description will be given of an appropriate first management information restoration process that is executed for the control target memory 3 that is tentatively determined to be a failure.

  Referring to FIG. 5, the memory controller 2 stores all second management information from the second block group R2 by performing a search using the second management information ID for pages P0 to P2. Identify the blocks. In this example, blocks B100, B150, B200, B250, B300, and B350 are specified.

  Similarly to the above, the access speed to the control target memory 3 is set to the slowest speed at which all types of memories assumed to be used can be normally operated. However, when the memory type can be specified based on the memory ID, the access may be performed at an appropriate access speed corresponding to the specified memory type. Further, when the restoration process is performed using an external host computer or the like instead of the memory controller 2, the type of memory is clear, and therefore, access may be performed at an appropriate access speed corresponding to the type of memory.

  Next, the memory controller 2 searches the identified block B100, B150, B200, B250, B300, B350 for a block in which a data error has occurred. Specifically, for each block, the entire area of the second management information is read, and the CRC calculation is performed on the entire area of the read second management information. Then, the CRC value obtained by the calculation is compared with the CRC values of the entire area included in the read second management information. If both CRC values match, the memory controller 2 determines that the block is a normal block, and if both CRC values do not match, determines that the block is an abnormal block.

  Similarly to the above, the access speed to the control target memory 3 is set to the slowest speed at which all types of memories assumed to be used can be normally operated. However, when the memory type can be specified based on the memory ID, or when restoration processing is performed using a host computer or the like, the access may be performed at an appropriate access speed corresponding to the memory type.

  If there is an abnormal block as a result of the search, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 5, the block B200 that is an abnormal block is excluded from the candidates.

  Next, the memory controller 2 selects one normal block in which the latest second management information is stored from the remaining normal blocks B100, B150, B250, B300, and B350. Specifically, by comparing the second management information ID values included in the second management information stored in the normal blocks B100, B150, B250, B300, and B350, the second management information ID value is obtained. The largest second management information is specified as the latest second management information. In the example shown in FIG. 5, since the second management information (4, 4) is the latest second management information, the block B100 storing the second management information (4, 4) is selected. Become. If there are a plurality of normal blocks in which the latest second management information is stored, an arbitrary normal block (for example, a normal block having the smallest block address) is selected. Further, when the update time or the like of the second management information is used as the second management information ID value, the second management information having the newest update time or the like (that is, the second management information having the largest second management information ID value). ) Can be specified as the latest second management information.

  Next, the memory controller 2 refers to the group address <200, 300> included in the second management information stored in the normal block B100, thereby making a majority decision for the blocks B100, B200, B300. Do. Specifically, the second management information ID value and the CRC value (the CRC value excluding the group address) included in the second management information stored in each of the blocks B100, B200, and B300 are compared with each other. By doing so, the first majority vote is performed. In this example, since the block B200 is not a normal block, it is excluded from the majority decision candidates, and the second management information ID value and CRC value of the two second management information stored in the blocks B100 and B300 are respectively Since (4, 4) and (1, 1) do not match, the first majority vote is not established.

  Next, the memory controller 2 selects one normal block in which the second new management information is stored secondly from the normal blocks B150, B250, B300, and B350 excluding the normal block B100. Specifically, the second management information ID value is the largest by comparing the second management information ID values included in the second management information stored in the normal blocks B150, B250, B300, and B350. The second management information is specified as the second new second management information. In the example shown in FIG. 5, since the second management information (3, 3) is the second newest second management information, any one normal block (in this example, the blocks B150, B250, B350) The normal block B150) having the smallest block address is selected.

  Next, the memory controller 2 refers to the group address <250, 350> included in the second management information stored in the normal block B150 to make a majority decision for the blocks B150, B250, B350. . Specifically, the second management information ID value and the CRC value (the CRC value excluding the group address) included in the second management information stored in each block B150, B250, and B350 are compared with each other. By doing so, the second majority vote is performed. In this example, the second management information ID values and CRC values of the three pieces of second management information stored in the blocks B150, B250, and B350 are all (3, 3), and two or more coincide with each other. Therefore, it is determined that the second management information stored in the blocks B150, B250, and B350 is valid second management information.

  In addition, even if all the normal blocks B100, B150, B250, B300, and B350 are selected, if appropriate second management information cannot be specified, it is determined that the control target memory 3 is faulty.

  Next, the memory controller 2 determines one arbitrary second management information from the specified two or more pieces of appropriate second management information. For example, the block B150 having the smallest block address is selected from the blocks B150, B250, and B350, and the second management information (3, 3) stored in the block B150 is determined as valid second management information.

  Next, the memory controller 2 restores valid first management information in the first block group R1 based on restoration information included in the determined valid second management information.

  FIG. 21 is a diagram showing the first management information restoration process on the assumption of the example shown in FIG. In this example, the latest first management information (7, 7) is restored as valid first management information. Specifically, the memory controller 2 erases data from the first three normal blocks or invalid blocks (in this example, blocks B1 to B3) in the first block group R1, and then stores them in these blocks B1 to B3. First management information (7, 7) is written.

  In this example, since the first management information (7, 7) is already stored in the block B3, the processing for the block B3 may be omitted. Further, the first management information (7, 7) may be written to all normal blocks, not limited to the three normal blocks from the top. Alternatively, the first management information (7, 7) may be written to the three normal blocks from the top, and only data erasure may be performed on the remaining normal blocks.

  The management information acquisition process for the first block group R1 shown in FIG. 21 will be briefly described. First, blocks B5 and B9 that are congenital defective blocks are excluded. Next, block B10 which is an invalid block is excluded. Next, an abnormal block is searched, but since no abnormal block exists in this example, no block is excluded as an abnormal block. Next, blocks B1 to B4 are selected as four normal blocks.

  Next, a majority vote is performed on the first three normal blocks B1 to B3, and as a result, the first management information (7, 7) stored in the blocks B1 to B3 is the valid first management information. Identified.

  Next, the second management information is read out from the second block group R2 based on the first management information (7, 7) stored in any of the blocks B1 to B3.

  In the above description, by including a group address in the second management information, three pieces of second management information for majority decision are specified based on the group address. Not limited to this example, three pieces of second management information for majority decision may be specified by selecting three normal blocks in ascending order of block address, large order, or randomly.

<Example of updating second management information>
22 and 23 are diagrams illustrating examples of updating the second management information. The configuration of the second block group R2 before the update is assumed to be in the state shown in FIG. In the example shown in FIG. 5, the latest second management information is the second management information (4, 4) stored in the block B100.

  When there is a need to update the second management information due to a new acquired bad block occurring in the second block group R2, the memory controller 2 includes the blocks B100, B200, and B300. The updated second management information (5, 5) is overwritten on one block set or a second block set composed of blocks B150, B250, and B350. In the example shown in FIG. 22, the updated second management information (5, 5) is overwritten on the first block set. However, since it is found that the block B200 is an acquired defective block, the second management information is transferred to another arbitrary block (block B210 in this example) in the second block group R2 instead of the block B200. (5, 5) is written.

  Further, for example, when the block B300 becomes an acquired defect block while the second management information (5, 5) is being written to the block B300, the memory controller 2 performs the second control as shown in FIG. Another arbitrary block (block B310 in this example) in the block group R2 is selected, and the updated second management information (6, 6) is written in the blocks B100, B210, B310. This update includes updating the group address and the bad block address. In this case, since the storage block of the second management information is changed, the first management information needs to be updated.

<Embodiment 2>
Hereinafter, the semiconductor memory device 1 according to the second embodiment of the present invention will be described focusing on differences from the first embodiment.

  Similar to the first embodiment, the first management information includes position information indicating a storage block for the second management information. The first management information includes a first management information ID. Further, the first management information includes position information indicating a defective block in the first block group R1.

  In the present embodiment, in addition to the above, the first management information includes position information (group address) indicating the other two blocks in which the first management information having the same content as the first management information is stored. .

  Further, the first management information includes an error detection code. In the example of the present embodiment, the CRC value of the entire area including the group address and the CRC value of the part excluding the group address are included in the first management information. The entire CRC value including the group address is used to determine whether or not a data error has occurred in the first management information. The CRC value excluding the group address is used for the majority vote for specifying the appropriate first management information.

  Hereinafter, (1) management information storage example, (2) management information acquisition process, and (3) management information update process will be described in order with respect to the semiconductor memory device 1 according to the present embodiment.

(1) Storage Example of Management Information FIG. 24 is a diagram illustrating the first block group R1. In this example, blocks B1 and B5 are congenital bad blocks, blocks B2 and B4 are invalid blocks, and block B10 is an erased block.

  Blocks B3, B6 to B9 store first management information. The numbers in parentheses attached to the first management information indicate the first management information ID value and the CRC value of the entire area. Numbers in angle brackets attached to the first management information indicate group addresses. For example, when the first management information (3, 3) is written in the block B7, it means that the first management information (3, 3) having the same contents is written in the blocks B2, B4.

(2) Management Information Acquisition Processing FIGS. 25 to 30 are diagrams sequentially illustrating management information acquisition processing executed by the memory controller 2 when the semiconductor memory device 1 is activated. It is assumed that the first block group R1 is in the state shown in FIG.

  First, the memory controller 2 acquires a memory ID from the control target memory 3 by issuing a read command requesting the memory ID to the control target memory 3.

<Step of searching for congenital defective blocks>
Next, with reference to FIG. 25, the memory controller 2 searches for all congenital defective blocks for all the blocks B1 to B10 belonging to the first block group R1.

  As a result of the search, if a congenital defective block exists in the blocks B1 to B10, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 25, blocks B1 and B5 that are congenital defective blocks are excluded from candidates.

<Invalid block search step>
Next, referring to FIG. 26, the memory controller 2 searches the invalid blocks for the good blocks B2 to B4 and B6 to B10.

  If there is an invalid block in the good block as a result of the search, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 26, invalid blocks B2 and B4 are excluded from candidates.

<Abnormal block search step>
Next, with reference to FIG. 27, the memory controller 2 searches the valid blocks B3, B6 to B10 for a block in which a data error has occurred. Specifically, for each valid block, the first management information storage page is accessed to read the first management information, and a CRC calculation is performed on the entire area of the first management information including the group address. And the CRC value of the whole area calculated | required by the calculation and the CRC value of the whole area contained in the read 1st management information are compared. If both CRC values match, the memory controller 2 determines that the block is a normal block, and if both CRC values do not match, determines that the block is an abnormal block.

  As a result of the search, when an abnormal block exists in the effective block, the memory controller 2 excludes the block from the acquisition target block candidates. In the example shown in FIG. 27, blocks B6 and B10 that are abnormal blocks are excluded from the candidates.

<Appropriate first management information specifying step>
Next, referring to FIG. 28, the memory controller 2 arbitrarily selects one normal block from the remaining normal blocks B3, B7 to B9. For example, the normal block B3 having the smallest block address is selected. Then, by referring to the group address <4, 6> included in the first management information stored in the normal block B3, a majority decision for the blocks B3, B4, B6 is made. Specifically, the first management information ID value and CRC value (CRC value excluding the group address) included in the first management information stored in each block B3, B4, B6 are compared with each other. By doing so, the first majority vote is performed. In this example, since the blocks B4 and B6 are not normal blocks and are excluded from the majority candidates, the first majority vote is not established.

  Next, referring to FIG. 29, the memory controller 2 arbitrarily selects the next normal block from the normal blocks B3, B7 to B9. For example, the normal block B7 having the next smallest block address is selected. Then, a majority decision is made for the blocks B2, B4, and B7 by referring to the group address <2, 4> included in the first management information stored in the normal block B7. Specifically, the first management information ID value and CRC value (CRC value excluding the group address) included in the first management information stored in each block B2, B4, B7 are compared with each other. By doing so, the second majority vote is performed. In this example, since the blocks B2 and B4 are not normal blocks and are excluded from the majority candidates, the second majority vote is not established.

  Next, referring to FIG. 30, the memory controller 2 arbitrarily selects the next normal block from the normal blocks B3, B7 to B9. For example, the normal block B8 having the next smallest block address is selected. Then, by referring to the group address <9, 10> included in the first management information stored in the normal block B8, a majority decision for the blocks B8 to B10 is performed. Specifically, the first management information ID value and the CRC value (the CRC value excluding the group address) included in the first management information stored in each of the blocks B8 to B10 are compared with each other. The third majority vote is made. In this example, since the block B10 is not a normal block, it is excluded from the majority decision candidates, but the first management information ID value and CRC value of the two first management information stored in the blocks B8 and B9 are both (4, 4), and two or more match. Therefore, it is determined that the first management information stored in the blocks B8 and B9 is appropriate first management information.

  Note that if the appropriate first management information cannot be specified even if all the normal blocks B3, B7 to B9 are selected, the memory controller 2 temporarily sets the control target memory 3 in the same manner as in the first embodiment. It is determined that there is a failure. For the control target memory 3 that has been provisionally determined to be faulty, an appropriate first management information restoration process is executed. The restoration process may be executed using an external host computer or the like. If the appropriate first management information cannot be restored even after executing the restoration process, the control target memory 3 is definitely determined to be a failure.

<Second management information acquisition step>
Next, the memory controller 2 determines one arbitrary first management information from the two or more specified appropriate first management information. For example, the block B8 having the smallest block address is selected from the blocks B8 and B9, and the first management information (4, 4) stored in the block B8 is determined as appropriate first management information.

  Next, the memory controller 2 acquires the second management information by reading the second management information from the second block group R2 based on the position information included in the determined valid first management information. To do.

(3) Management Information Update Processing FIGS. 31 to 35 are diagrams sequentially illustrating first management information update processing executed by the memory controller 2.

  Referring to FIG. 31, the first management information currently valid is the first management information (4, 4) stored in blocks B8 and B9. Further, the first management information that is a candidate for majority decision for specifying appropriate first management information at the time of activation is the first management information stored in the majority decision target blocks B8 to B10.

  Next, referring to FIG. 32, the memory controller 2 secures an empty block from the first block group R1. Specifically, from the blocks B1 to B10, congenital bad blocks, acquired bad blocks, and majority-target blocks are excluded, and three empty blocks are secured. In this example, blocks B2, B4 and B7 are secured as empty blocks.

  Next, referring to FIG. 33, the memory controller 2 erases the data of any one of the three empty blocks, and then sends the updated first management information including the group address to that block. Write to. In this example, after the data in the block B2 is erased, the first management information (5, 5) <4, 7> including the group address <4, 7> indicating the blocks B4, B7 is written into the block B2. .

  Next, referring to FIG. 34, the memory controller 2 adds an invalid flag to any one of the three majority-target blocks. In this example, an invalid flag is added to the block B8.

  Next, referring to FIG. 35, after erasing the data of the next empty block B4 among the three empty blocks, the memory controller 2 deletes the first management information (5, 5) including the group address <2, 7>. 5) Write <2,7> in its block B4. Thereafter, an invalid flag is added to the next block B9 among the three majority target blocks. Next, the memory controller 2 erases the data of the last empty block B7 among the three empty blocks, and then first management information (5, 5) <2, 4 including the group address <2, 4>. > Is written in the block B7. Thereafter, an invalid flag is added to the last block B10 among the three majority target blocks. Thereby, the first management information (5, 5) having the same contents is stored in the three empty blocks B2, B4, B7.

<Specific Example 1 of Embodiment 2>
FIG. 36 is a diagram illustrating another example of the first block group R1. In this example, blocks B5 and B9 are congenital defective blocks, and blocks B1, B2 and B10 are invalid blocks. Blocks B4 and B8 are acquired defective blocks. However, regarding the blocks B4 and B8, the first management information ID value matches the CRC value. Also, the first management information (6, 6) <6, 7> is stored in the block B3, and the first management information (6, 6) <3, 7> is stored in the block B6. The block B7 stores first management information (6, 6) <3, 6>.

  The management information acquisition process for the first block group R1 shown in FIG. 36 will be briefly described. First, blocks B5 and B9 that are congenital defective blocks are excluded. Next, blocks B1, B2, and B10 that are invalid blocks are excluded. Next, an abnormal block is searched, but since no abnormal block exists in this example, no block is excluded as an abnormal block.

  Next, the first normal block B3 is selected, and a majority vote is performed on the normal block B3 and the normal blocks B6 and B7 indicated by the group address <6, 7>. As a result, the first management information (6, 6) stored in the blocks B3, B6, B7 is specified as valid first management information.

  Next, based on the first management information (6, 6) stored in any of the blocks B3, B6, B7, the second management information is read out from the second block group R2.

<Specific Example 2 of Embodiment 2>
FIG. 37 is a diagram illustrating another example of the first block group R1. In this example, blocks B5 and B9 are congenital defective blocks, and blocks B1, B2 and B10 are invalid blocks. Blocks B4 and B6 are acquired defective blocks. However, regarding the blocks B4 and B6, the first management information ID value matches the CRC value. Also, the first management information (7, 7) <7, 8> is stored in the block B3, and the first management information (7, 7) <3, 8> is stored in the block B7. The block B8 stores the first management information (7, 7) <3, 7>.

  The management information acquisition process for the first block group R1 shown in FIG. 37 will be briefly described. First, blocks B5 and B9 that are congenital defective blocks are excluded. Next, blocks B1, B2, and B10 that are invalid blocks are excluded. Next, an abnormal block is searched, but since no abnormal block exists in this example, no block is excluded as an abnormal block.

  Next, the first normal block B3 is selected, and a majority vote is performed on the normal block B3 and the normal blocks B7 and B8 indicated by the group address <7,8>. As a result, the first management information (7, 7) stored in the blocks B3, B7, B8 is specified as valid first management information.

  Next, based on the first management information (7, 7) stored in any of the blocks B3, B7, B8, the second management information is read out from the second block group R2.

  In specific example 2 (FIG. 20) of the first embodiment, which is the same case, the appropriate first management information could not be specified. On the other hand, in the example of FIG. 37, the identification of the appropriate first management information has succeeded by making a majority decision based on the group address.

<Modification 1>
FIG. 38 is a diagram showing a simplified configuration of the semiconductor memory device 1 according to the first modification. The semiconductor memory device 1 includes a memory controller 2 and a plurality of control target memories (four control target memories 3A to 3D in the example of FIG. 38) controlled by the memory controller 2.

  In such a system configuration, the second management information having a large data size may be divided into a plurality of pieces, and each divided piece may be distributed and stored in a plurality of control target memories. In the example shown in FIG. 38, each second management information is divided into three second management information pieces 121-123. The first management information 11 is stored in the control target memory 3A, the second management information piece 121 is stored in the control target memory 3B, the second management information piece 122 is stored in the control target memory 3C, and the second management information piece 123 is stored. Is stored in the control target memory 3D. The first management information 11 includes position information indicating the location of each of the second management information pieces 121 to 123 for each second management information. The first management information 11 includes information related to the number of divisions of the second management information (that is, the number of second management information pieces corresponding to one piece of second management information). When reading the second management information, the second management information pieces 121 to 123 are read in parallel from the control target memories 3B to 3D. In addition to the first management information 11, a second management information piece may be stored in the control target memory 3A in which the first management information 11 is stored. Further, the first management information 11 may be stored not only in the control target memory 3A but also in other control target memories 3B to 3D.

  FIG. 39 is a diagram illustrating a first example of the second management information pieces stored in the control target memories 3B to 3D. Three second management information pieces X11, X12, X13 constituting the second management information X1 are stored in the blocks b, h, k (block addresses b, h, k) of the control target memories 3B, 3C, 3D, respectively. Has been. Further, the three second management information pieces X21, X22, X23 constituting the second management information X2 are stored in the blocks d, g, o (block addresses d, g, o) of the control target memories 3B, 3C, 3D. Each is stored. Further, the three second management information pieces X31, X32, X33 constituting the second management information X3 are stored in the blocks e, i, m (block addresses e, i, m) of the control target memories 3B, 3C, 3D. Each is stored. Also, the three second management information pieces Y11, Y12, Y13 constituting the second management information Y1 are stored in the blocks a, f, n (block addresses a, f, n) of the control target memories 3B, 3C, 3D. Each is stored. Further, the three second management information pieces Z11, Z12, Z13 constituting the second management information Z1 are stored in the blocks c, j, l (block addresses c, j, l) of the control target memories 3B, 3C, 3D. Each is stored.

  The second management information X1, X2, and X3 are second management information having the same contents corresponding to each other. In addition, alphabets in square brackets in the figure indicate block addresses (hereinafter referred to as “inter-memory group addresses”) in which the other two divided second management information pieces are stored. For example, the second management information X1 is divided into three second management information pieces X11, X12, and X13, and the second management information pieces X11, X12, and X13 are stored in the blocks b, h, and k, respectively. The second management information piece X11 includes an inter-memory group address “h, k”, the second management information piece X12 includes an inter-memory group address “b, k”, and the second management information piece X13 includes “b, k”. h ”inter-memory group address.

  In the following, a description will be given of the appropriate first management information restoration process. In the following example, the case where the restoration process is performed by the memory controller 2 will be described. However, the restoration process can also be performed using an external host computer or the like as described above.

  First, the memory controller 2 identifies blocks a, b, c, d, and e storing the second management information piece from the control target memory 3B.

  Next, the memory controller 2 performs processing for excluding abnormal blocks from the blocks a, b, c, d, and e. However, in this example, it is assumed that no abnormal block exists in the blocks a, b, c, d, and e.

  Next, the memory controller 2 arbitrarily selects one block from the blocks a, b, c, d, and e. For example, the block a having the smallest block address is selected. Next, after reading the second management information piece Y11 from the block a, the second management information piece Y11 is read from the blocks f and n by referring to the inter-memory group address “f, n” included in the second management information piece Y11. The management information pieces Y12 and Y13 are read out. Next, the second management information Y1 is restored by combining the second management information Y11, Y12, Y13. The memory controller 2 determines whether or not the blocks f and n are also abnormal blocks, and does not restore the second management information Y1 when at least one of the blocks f and n is an abnormal block. However, in this example, the blocks f and n are not abnormal blocks. The same applies to the other blocks g to m, o.

  Next, the memory controller 2 reads the second management information pieces X11 and Z11 from the blocks b and c by referring to the group address <b, c> included in the second management information Y1. Next, by referring to the inter-memory group address “h, k” included in the second management information piece X11, the second management information pieces X12, X13 are read from the blocks h, k, respectively. Next, the second management information X1 is restored by combining the second management information pieces X11, X12, and X13. Further, by referring to the inter-memory group address “j, l” included in the second management information piece Z11, the second management information pieces Z12 and Z13 are read from the blocks j and l, respectively. Next, the second management information Z1 is restored by combining the second management information pieces Z11, Z12, and Z13.

  Next, the memory controller 2 makes a majority decision on the restored second management information Y1, X1, and Z1, but this majority decision is not established in the example of FIG.

  Next, the memory controller 2 arbitrarily selects the next block from the blocks a, b, c, d, and e. For example, the block b having the next smallest block address is selected. Next, after reading the second management information piece X11 from the block b, the second management information piece X11 is read from the blocks h and k by referring to the inter-memory group address “h, k” included in the second management information piece X11. The management information pieces X12 and X13 are read out. Next, the second management information X1 is restored by combining the second management information X11, X12, and X13. Note that since the second management information X1 has already been restored in the above step, the restoration process of the second management information X1 in this step may be omitted.

  Next, the memory controller 2 reads out the second management information pieces X21 and X31 from the blocks d and e by referring to the group address <d, e> included in the second management information X1. Next, by referring to the inter-memory group address “g, o” included in the second management information piece X21, the second management information pieces X22 and X23 are read from the blocks g and o, respectively. Next, the second management information X2 is restored by combining the second management information pieces X21, X22, and X23. Further, by referring to the inter-memory group address “i, m” included in the second management information piece X31, the second management information pieces X32 and X33 are read from the blocks i and m, respectively. Next, the second management information X3 is restored by combining the second management information pieces X31, X32, and X33.

  Next, the memory controller 2 makes a majority decision on the restored second management information X1, X2, and X3. In the example of FIG. 39, since this majority vote is established, the memory controller 2 determines the second management information X1, X2, and X3 as appropriate second management information.

  Next, the memory controller 2 determines one arbitrary second management information from the appropriate second management information X1, X2, and X3, and the restoration information included in the determined valid second management information. Based on the above, the appropriate first management information is restored in the control target memory 3A.

  FIG. 39 shows an example in which the group address is included in the second management information piece stored in the control target memory 3B. However, the present invention is not limited to this example, and the group addresses are not limited to the control target memories 3B to 3B. It may be included in the second management information piece stored in at least one of the 3Ds. For example, the group address <b, c> may be included in the second management information piece Y12 or the second management information piece Y13 instead of the second management information piece Y11.

  FIG. 40 is a diagram illustrating a second example of the second management information piece stored in the control target memories 3B to 3D. In the example of FIG. 40, the second management information piece does not include an inter-memory group address, and instead, each second management information piece includes a group address. For example, the second management information pieces X11, X21, and X31 include a group address <d, e> <b, e> <b, d>, whereby the second management information pieces having the same contents are stored. A block in the stored control target memory 3B can be specified. Each second management information piece includes an ID value that can identify the second management information before division. For example, the ID values “X11”, “X12”, and “X13” are included in the second management information pieces X11, X12, and X13, respectively, so that the second management information before the division is the second management information X1. It can be specified.

  In the following, a description will be given of the appropriate first management information restoration process. In the following example, the case where the restoration process is performed by the memory controller 2 will be described. However, the restoration process can also be performed using an external host computer or the like as described above.

  First, the memory controller 2 identifies blocks a, b, c, d, and e storing the second management information piece from the control target memory 3B.

  Next, the memory controller 2 performs processing for excluding abnormal blocks from the blocks a, b, c, d, and e. However, in this example, it is assumed that no abnormal block exists in the blocks a, b, c, d, and e.

  Next, the memory controller 2 arbitrarily selects one block from the blocks a, b, c, d, and e. For example, the block a having the smallest block address is selected. Next, after the second management information piece Y11 is read from the block a, the second management information is read from the blocks b and c by referring to the group address <b, c> included in the second management information piece Y11. The pieces X11 and Z11 are read out respectively.

  Next, the memory controller 2 makes a majority decision for the second management information pieces Y11, X11, and Z11, but this majority decision is not established in the example of FIG.

  Next, the memory controller 2 arbitrarily selects the next block from the blocks a, b, c, d, and e. For example, the block b having the next smallest block address is selected. Next, after reading the second management information piece X11 from the block b, the second management information piece is read from the blocks d and e by referring to the group address <d, e> included in the second management information piece X11. The pieces X21 and X31 are read out.

  Next, the memory controller 2 makes a majority decision for the second management information pieces X11, X21, and X31. In the example of FIG. 40, since this majority vote is established, the memory controller 2 determines that the second management information pieces X11, X21, and X31 are valid second management information pieces.

  The memory controller 2 performs the same processing as the control target memory 3B for the control target memories 3C and 3D. As a result, regarding the control target memory 3C, the second management information pieces X12, X22, and X32 are determined to be valid second management information pieces, and regarding the control target memory 3D, the second management information pieces X13, X23, and X33 are determined. It is determined as a valid second management information piece.

  Next, the memory controller 2 combines the corresponding second management information pieces based on the ID values included in the second management information pieces X11 to X13, X21 to X23, and X31 to X33. The appropriate second management information X1, X2, X3 is restored.

  Next, the memory controller 2 determines one arbitrary second management information from the appropriate second management information X1, X2, and X3, and the restoration information included in the determined valid second management information. Based on the above, the appropriate first management information is restored in the control target memory 3A.

<Modification 2>
In the first and second embodiments, the example in which the management information is divided into the first management information and the second management information has been described. However, the management information may be divided into three or more. For example, when dividing into three, information with low rewriting frequency is included in the first management information, information with medium rewriting frequency is included in the second management information, and information with high rewriting frequency is included in the third management information. Include in. The first management information is stored in a first block group with a small number of allocated blocks, the second management information is stored in a second block group with a large number of allocated blocks, and the third management information is all blocks Are stored in the third block group other than the first block group and the second block group (the number of blocks is the largest).

<Modification 3>
In the first and second embodiments, the blocks allocated as the first block group R1 are limited to the blocks B1 to B10. Therefore, when there is a congenital defective block in the blocks B1 to B10, the number of blocks that can actually be used in the first block group R1 decreases according to the number of congenital defective blocks. Therefore, congenital defective blocks may be excluded when assigning the first block group. For example, when the blocks B5 and B9 are congenital defective blocks, the blocks B1 to B4, B6 to B8, and B10 to B12 are assigned as the first block group.

<Summary>
According to the semiconductor memory device 1 according to the first and second embodiments, the management information includes the first management information stored in the first block group R1 and the second management stored in the second block group R2. Information. The second management information includes restoration information for restoring the appropriate first management information in the first block group R1 when the appropriate first management information cannot be specified in the first block group R1. included. Therefore, even when the appropriate first management information cannot be specified in the first block group R1, the memory controller 2 determines the appropriate first management information based on the restoration information included in the second management information. Can be restored in the first block group R1. As a result, it is possible to store the management information in the control target memory 3 while avoiding a situation in which the normal operation is hindered due to the presence of a defective block, forced power shutdown, or the like.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the second management information is information that is rewritten more frequently than the first management information. By dividing the management information into the first management information and the second management information according to the rewrite frequency, and storing the first management information with a low rewrite frequency in the limited first block group R1 In the first block group R1, data rewriting does not occur frequently. As a result, since the occurrence of acquired defective blocks in the first block group R1 is suppressed, the reliability of the apparatus can be improved.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the first management information includes the position information indicating the block storing the second management information, and the second management information , Position information indicating a defective block is included. Therefore, the memory controller 2 can acquire the position information indicating the block storing the second management information by reading the first management information from the control target memory 3, and the second management based on the position information. By reading the information from the control target memory 3, it is possible to acquire position information indicating a defective block.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the first management information includes the position information indicating the defective block in the first block group R1, and the second management information includes Position information indicating a defective block in the second block group R2 is included. Therefore, when an acquired defective block occurs in the second block group R2, only the second management information needs to be updated, and the update of the first management information can be omitted.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the second management information having the same content is stored in each of the second predetermined number of blocks in the second block group R2. Has been. Therefore, even if one of the second predetermined number of second management information stored in the second predetermined number of blocks is destroyed or lost, the memory controller 2 is destroyed or lost. No remaining second management information can be acquired from the control target memory 3. As a result, it is possible to store the management information in the control target memory 3 while avoiding a situation in which the normal operation is hindered due to the presence of a defective block, forced power shutdown, or the like.

  Moreover, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 specifies the third predetermined number of normal blocks from the second block group R2. Then, from the third predetermined number of second management information stored in the specified third predetermined number of normal blocks, one appropriate second management information is specified by majority vote. Accordingly, it is possible to appropriately specify appropriate second management information even when a plurality of versions of second management information are stored in the second block group R2 by repeating the update.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 removes the abnormal block from all the blocks storing the second management information, thereby removing the normal block. Identify. Accordingly, it is possible to avoid a situation in which an abnormal block is specified as a normal block, and it is possible to appropriately specify appropriate second management information.

  In addition, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 identifies three normal blocks from the third predetermined number of normal blocks, and the three normal blocks From the three pieces of second management information stored in, one appropriate second management information is specified by majority vote. In this way, by limiting the target of the majority decision to the three pieces of second management information, the majority decision can be easily made, and thus the time required for processing can be shortened.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 includes the second management information stored in a certain normal block and the location information in the second management information. A majority vote is performed using two pieces of second management information stored in two specific blocks described. Therefore, since the target of majority decision can be limited to the second management information having the same contents written together at the time of update, it is possible to increase the possibility of specifying appropriate second management information by majority decision.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 cannot specify appropriate second management information by majority decision for the specified three normal blocks. Then, the other three normal blocks are specified again and a majority decision is made. Thereby, it is possible to increase the possibility of specifying appropriate second management information.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, when the memory controller 2 needs to update the second management information, the second management information before the update is stored. The updated second management information is written in each of the second predetermined number of blocks. Thereby, even after the update, a second predetermined number of second management information having the same contents can be stored in the second block group R2.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 writes the version information having a different value for each update in the second management information. Therefore, a majority decision can be made using the version information, and as a result, it is possible to improve the accuracy of appropriately specifying appropriate second management information.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, when the block becomes an acquired defect block when the second management information is written in a certain block, the memory controller 2 The value of the version information included in the second management information is not used for subsequent updates. Therefore, even if the counter value for generating the version information value makes a round, the second management information stored in the acquired defective block is erroneously specified as valid second management information by majority vote. It is possible to avoid this situation.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 writes the time information related to the update time of the second management information in the second management information. Therefore, the latest second management information can be specified based on the time information, and as a result, it is possible to improve the accuracy of appropriately specifying appropriate second management information.

  Moreover, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 writes the error detection code of the second management information in the second management information. Therefore, a majority decision can be made using the error detection code, and as a result, it is possible to improve the accuracy of appropriately identifying the appropriate second management information. In addition, it is possible to determine whether or not a data error has occurred using an error detection code. As a result, when a normal block is specified from the second block group, an abnormality in which a data error has occurred Blocks can be excluded.

  Further, according to the semiconductor memory device 1 according to the first and second embodiments, the memory controller 2 stores the second predetermined number of pieces of the second management information written in one block among the second predetermined number of blocks. Including the position information indicating other blocks in the current block is written. Therefore, since the target of majority decision can be limited to the second management information having the same contents written together at the time of update, it is possible to increase the possibility of specifying appropriate second management information by majority decision.

DESCRIPTION OF SYMBOLS 1 Semiconductor memory device 2 Memory controller 3 Memory to be controlled

Claims (16)

A storage unit having a plurality of blocks, each of which is a data erasure unit;
A control unit for controlling the storage unit;
With
In all blocks of the storage unit,
A first block group consisting of a plurality of first predetermined number of blocks, limitedly assigned as blocks for storing management information necessary for the control unit to manage the storage unit;
A second block group other than the first block group;
Contains
Management information
First management information stored in the first block group;
Second management information stored in the second block group;
Including
The second management information includes restoration information for restoring the appropriate first management information in the first block group when the appropriate first management information cannot be specified in the first block group. Storage device.   The semiconductor memory device according to claim 1, wherein the second management information is information that is rewritten more frequently than the first management information. The first management information includes position information indicating a block storing the second management information,
The semiconductor memory device according to claim 1, wherein the second management information includes position information indicating a defective block. The first management information includes position information indicating a defective block in the first block group,
The semiconductor memory device according to claim 3, wherein the second management information includes position information indicating a defective block in the second block group.   5. The semiconductor memory device according to claim 1, wherein second management information having the same content is stored in each of a plurality of second predetermined number of blocks in the second block group. . The controller is
A plurality of third predetermined number of normal blocks in which normal second management information is stored are identified from the second block group,
From the third predetermined number of second management information stored in the third predetermined number of normal blocks, one valid second management information is specified by majority vote,
6. The semiconductor memory device according to claim 5, wherein the valid first management information is restored based on the restoration information included in the valid second management information.   The control unit identifies a normal block by excluding abnormal blocks in which data errors have occurred in the stored second management information from all blocks storing the second management information The semiconductor memory device according to claim 6. The controller is
3 normal blocks are identified from the third predetermined number of normal blocks,
8. The semiconductor memory device according to claim 6, wherein one appropriate second management information is specified by majority vote from the three second management information stored in the three normal blocks. The controller is
Second management information stored in one normal block;
Two pieces of second management information stored in two specific blocks in which position information is described in the second management information;
The semiconductor memory device according to claim 8, wherein a majority vote is performed by using. When the control unit cannot identify appropriate second management information by a majority vote for the specified three normal blocks,
Identify the other three normal blocks from the third predetermined number of normal blocks,
10. The semiconductor memory device according to claim 8, wherein one valid second management information is specified by majority vote from among the three second management information stored in the other three normal blocks. 11. .   When the control unit needs to update the second management information, the second management information after the update is stored in each of the second predetermined number of blocks in which the second management information before the update is stored. The semiconductor memory device according to claim 5, wherein each is written.   The semiconductor memory device according to claim 11, wherein the control unit writes version information having a different value every time the second management information is updated, included in the second management information.   When the second management information is written in a certain block and the block becomes an acquired defect block, the control unit does not use the version information value included in the second management information at the time of the subsequent update. The semiconductor memory device according to claim 12.   The semiconductor memory device according to claim 11, wherein the control unit writes time information related to an update time of the second management information in the second management information.   The semiconductor memory device according to claim 11, wherein the control unit writes the error detection code of the second management information so as to be included in the second management information. The control unit writes the second management information written in one block of the second predetermined number of blocks, including position information indicating other blocks in the second predetermined number of blocks. The semiconductor memory device according to any one of 11 to 15.

Images