TWI376599B - Method and nonvolatile memory system of managing file allocation table information - Google Patents

Description

1376599 IX. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present invention relates to non-volatile memory systems, and to methods of operating non-volatile memory systems. [Prior Art] A modular, portable, non-volatile memory device is available that can be easily connected to and disconnected from a host device such as a digital camera, a digital audio recorder, and a personal computer. For example, flash memory cards are rewritable in a memory system that allows erasing and overwriting a memory address = as a system or user. The establishment of a low-cost memory technology that can be programmed (〇τρ) or a few times programmable (FTP) allows for the use of new modules for the device, which is similar to a film in a camera, five people. It is used to simply take and print photos, while eliminating the low cost card of the one. Ideally, we should be able to use this low-cost card in a standard flash camera, but these technologies may not be compatible, because the portable D〇SFAT12/16 standard system requires a system structure that can be erased^ (10) or the custom memory system of the memory technology update can solve this problem, but this may limit the total available market to a new camera, or may require the user to use the body Upgrade its existing camera.秋遐 [Abstract] In the operation of a non-volatile memory, the address of the address range of the host memory is allocated to the shovel shovel ★ σ t Zuo Yi benefit from the address The other end allocates clusters. Even if it is not possible to rewrite or erase the material, 124767.doc 1376599 allows the host to transmit an update to previously written data by such a distribution of a controller. The distribution by the host is recorded in a file allocation table (FAT). The assignment by the controller is not recorded in the fat but instead is recorded in the volatile memory. When a host requests FAT information, the controller corrects the fat from the non-volatile memory based on the record of the volatile memory.

In a method of managing data of a non-volatile memory, one of the hosts stores data in the non-volatile memory, and the host records the allocation of the stored data in a file also stored in the non-volatile memory. Allocation Table. A portion of the slot allocation table is purchased from the non-volatile memory array K from a request from the host for a portion of the file allocation table stored in the non-volatile memory array. Subsequently, the portion of the knives matching table is repaired according to a record indicating that the memory controller is configured to store at least one memory location received from the host, and the file is The corrected portion of the matching table is transmitted to the host.

:: method t, maintaining a cluster of images in a volatile memory:: mapping: each of the plurality of clusters - the allocating - allocating / de-allocating port receives a master for the allocation information about the plurality of clusters = 'From the cluster map determines whether the plural clusters have been divided

The pass will generate the distribution information from the cluster map and pass the device back to the host. L The assigned industry Cui β generates a FAT section to show that the 无 兮 隼 Μ Μ Μ Μ Μ Μ Μ Μ Μ Μ Μ 。 。 。 。 。 。 。 。 。 。 。 。 。 。 The right child set mapping indication. Six people's memory read "i" one allocated cluster, the allocation information will be taken from the non-volume item. 124767.doc 1376599 [Embodiment]

Certain embodiments described herein may be used to enable one or fewer programmable memory devices to function in conjunction with existing consumer electronic devices (eg, in conjunction with flash (an erasable, reprogrammable non-volatile memory) The operator does not require a firmware update to provide backward compatibility while minimizing user impact. Thus, these embodiments are a convenient way to bridge one or less programmable memory and existing consumer electronics devices with flash card slots. These specific embodiments also allow for the design of future consumer electronic devices that do not require updating firmware to include a file system for customizing one or less programmable memory. The specific embodiments described may also be applied to multiple programmable non-volatile memories. It is not required to utilize the technique in a memory that can be programmed once.

Referring now to the drawings, Figure 1 is a block diagram of a host device 5 connected to a memory device. Both the host device 5 and the memory device 1 include electronic connectors that are paired with one another to electronically couple the host device 5 to the memory device 1 through the interface 15. As used herein, the term "lightly," means direct-coupled or indirectly coupled through one or more intervening components. The host device 5 can take the form of a consumer electronic device, such as a digital- or static camera, a personal digital assistant, a cellular phone, a digital audio player, or a personal computer (eg, such a USB device) Two computers in which the II/writer or pcMciA card adapter is read. In this embodiment, the host device 5 contains, for example, the multi-write standard system 7 of the OS-FAT slot system. , the other controller of the reverse compatible controller', for example, can be used together with a controller that can be used together with the volatile memory! 24767.άο〇1376599. The memory device ίο can be used in a modular, compact, handheld The form of the unit, such as a memory card or strip. The memory device 10 includes a controller 20 and a write-only memory array 30. The configurable host 5 and the memory device 10 interface 15' Any portable storage medium for multimedia, secure digital, memory stick, compact flash memory, smart media, XD, USB, HS-MMC, or many portable storage media available. Controls are detailed below. Device 20 allows the memory device 1 to be backward compatible with a host device using a multi-write standard system. Thus, in this context, sometimes the controller 20 will be referred to as a "reverse compatible controller". Or "bcc". An embodiment of the controller 20 utilizes an ASIC of a finite state machine combined with standard combinational logic. However, the controller 20 can be implemented in a variety of other forms, such as but not limited to having A microcontroller or a microprocessor of the firmware. Further, although the controller 2 is separated from the memory array 30 in this embodiment, a controller and a memory array can be integrated into a single The cost is saved in the die. In this embodiment, the controller 20 is designed to be very similar to the controller for rewritable memory technology. An example of the difference is that the controller 20 does not require any wear leveling or Associated with other logic that erases volatile memory. Although the design shown herein may include such additional steps and functions, it may not be optimal cost. The write-only memory array 30 includes a plurality of Write-once-only programmable memory cell. A field-programmable write-only memory cell system manufactured in an initial, non-programmable digital state can be manufactured at a time after the switching 124767.doc -9 1376599

To a replacement, stylized digital memory unit. For example, the original, unprogrammed digit state can be identified as the logical 丨 (or logical 0) state, and the stylized digit state can be identified as the logical 〇 (or logical 〇 state) because the memory cells are A write-only type, once switched to a stylized digit state (eg, the logic 0 state), an original, unprogrammed digit state of a storage location (eg, the logic 1 state) cannot be restored. A programmable (ie, write-only) memory array, the memory array 30 can be in the form of a small number of programmable (FTp) memory arrays, which can be written more than once but not as much as one. The memory array has a memory array multiple times. At the same time, the memory device 1 can contain an additional memory array (single write, less programmable, or multiple writes).

The write-once memory array 30 can take any suitable form, such as a solid state memory device (ie, a memory device that responds to electronic read and write signals), a magnetic storage device (eg, a hard disk drive). Or an optical storage device (such as a CD or DVD). When having the form of a solid state memory cell, the δ-resonant cell in the memory array 3 can be organized in a two- or three-dimensional manner. In one embodiment, the memory array 3 is a three-dimensional array, such as an array of U.S. Patent No. 6,034,882 to Johnson et al. In some other embodiments, an erasable volatile memory, such as an erasable flash memory, is used. To illustrate these specific embodiments, the DOS FAT file system (e.g., used in a Windows operating system) will be used as an example of a multi-write file system. It should be noted that these specific embodiments can be positioned with other types of multiple write files 124767.doc -10- 1376599. A common approach is to use an entity-to-logical address table that is read or calculated at startup and then stored in the controller's volatile memory. In the OPT memory, the host allocates updated data to the cluster that was previously allocated to the old data. Therefore, the same logical address range is occupied, but extra space must be occupied in the physical memory. This difference in the FAT quotation between the range of logical addresses that are not available and the range of physical addresses that are actually available in the memory will cause problems. For example, the host will attempt to write to a full memory.

In an entity-addressed memory, data from the host is stored according to the address provided by the host. Therefore, the address provided by the host is regarded as a physical address, not just a logical address, and does not require any physical-to-logical address translation table H as long as the material is not stored in the address provided by the host. There may be some exceptions in the system. In general, some kind of program is needed to track such information. One solution is to use a "Sideband area" (or, an area of the overhead area) that is invisible to the host. An example of this solution is given in the US patent application. In the specific embodiment, (4) the memory array is organized in units of 528 bytes, each section stores a section, but the host is a 512-bit tuple. The segment transmits the data, leaving extra space for the two groups. These additional 16-bytes are called sidebands and can be used by the controller to store, but not limited to, Ecc data, data and address verification flags. Additional information marked with the remapping indicator. If the party/reconciliation has remapped the page, the controller 20 uses the sideband to store a physical address. If the page is written to the first time No remapping metrics are required because the real and logical addresses will be the same. When the host 5 attempts to correct the data, the controller will write the new data to a new location and will The physical address of this location is stored in the sideband of the old location. In this case, a host does not know whether a memory system is logically addressed or physically addressed, and the specific embodiments described herein can be applied to physical addressed memory or logically addressed memory. This description refers to logically addressed memory. But it should be understood that, in general, it can be applied - specific predetermined logic to the entity

Mapping, processing a physical addressed memory as a logically addressed memory. In a logical addressing or entity; t-site system, the host updates or de-allocates the data in the OTP memory, and the host can (through FAT) see more free space than is actually available in the physical memory. In order to solve the problem of using the memory of the interpretation controller (4), a simple method is to reserve the memory space for use by the controller before the time of selling the memory device 10. This type of space is not available to this host. The controller 20 will use this amount of static memory space until no more reserved space is left.

stop. At this point, the controller 20 can send a card-card error, allow for any writes that are not one-to-one mapping, or many other operations. We can see that this method, called static allocation, would be desirable because of its limitations—the number and size of operations that the user can perform. In an alternative dynamic allocation scheme shown in Figure 3, both the host and the controller allocate clusters from the same logical address space 3〇1. In one example, as shown in Figure 3, a host allocates clusters 'from the top of the available address space 3' and the controller allocates clusters from the bottom of the available address space. Therefore, the host allocation space 303 is used for new data, and the controller allocates *124767.doc 13· 1376599 between 305 for updated material. In this way, there is no predetermined limit on the division of the available space between the host allocation space and the controller allocation space. However, the space usage of the controller must be reported to the host in some way. In one arrangement, the space used by the controller is communicated to the host via FAT information transmitted to the host. In this way, when the available physical space in the memory is reduced, the logical space available to the host as shown in the FAT is also reduced.

In the example of FIG. 4A, the controller 2 notifies the host 5 that the space has been replayed by writing the FAT table 40 7 stored in the non-volatile suffix.

Shoot. An example of a memory system using such a system is provided in U.S. Patent Application Publication No. 2006/0047920. When additional memory is required, just as the host 5 will allocate a cluster to the archive, the controller 2 can allocate a new cluster for its own use. That is, this scheme dynamically updates the FAT table 4〇7 when additional space is needed for re-mapping. The benefit of this embodiment is that the number of remapping/file corrections that can be performed by the host 5 and the user is not limited. The controller 20 will allow it to be remapped until the card is fully full. In this particular embodiment, when most of the hosts are allocating memory from the top down, the controller 20 allocates memory from the bottom up (as shown in Figure 2, starting from the logical space). This allocation scheme will allow the controller 2 and the host 5 to independently allocate memory. This also allows the controller 2 to recognize that when the two segments collide, the card 10 is full. When the card is full, the controller 2 can choose to set its permanent and temporary write (four) register, send It is impossible for the host 5 to write to the card 10 again. In addition, when the card is full, the card is full * When the card is full, the controller will generally use some other system to prevent additional writers. 124767.doc 1376599 does not attempt to write a full card because The fat indicates that the logical address space is full (if the host caches the FAT data, the cache FAT will not display a full logical address space). A no-cache system will know how much memory is left, and the cached system will read the FAT table at startup to see how much remapping space is used before that particular job phase.

In a logically addressed memory, the use of physical memory space does not have to follow the same pattern followed by the use of logical space. 4B through 4D illustrate how the physical memory space is used for data storage when the controller and host store data and allocate logical address space as shown in FIG. 4A.

Figure 4B shows the physical memory 4〇9 in an initial state in which a portion 411 of the physical memory is used to store control data and the remaining physical memory 409 is unprogrammed. The control data can include data generated by the controller for managing data in the array of memory, such as data used by the controller for logical entity mapping. As usual, such control data is seen by the host. In this example, the data (including MBR and FAT) supplied by the host is regarded as the host data. In this example, (iv) the data is generated during an initial formatting operation. η...n The core memory of the host data in 409. The host data is written below the control data - available entity location 413. After writing the host material, at least a portion of the control data must be updated to reflect the newly written host data. This updated control data is written to the available lower-physical location 415 after the host profile. This updated control data causes a portion 417 of the previously stored control data to be invalidated. 124767.doc •15- ^376599 Figure 4D shows the physical memory 409 after the host transmits the updated data to replace a portion of the previously programmed host data. For example, the host can transmit the same logic as the pre-stored saved segment. Several sections of the host data of the address. The updated host profile is stored in the next available entity location 419 and then the control profile associated with the updated host profile is stored in the next available entity location 421. The host material has been updated to invalidate a portion 423 of the original host material. The control data has been updated to invalidate a portion 425 of the first updated control material. Therefore, part of the physical memory space is occupied by the obsolete data sections 417, 423, 425, and the scrapped data sections contain the obsolete host data and the obsolete control data. The controller can allocate logic space to interpret this loss of physical space and invalidate host data and control data. Should be succinct, static and dynamic remapping are not mutually exclusive. We can choose to implement both in an early morning product. For example, a user may wish to delete or correct an archive even if the card is full and no additional files can be added. Additional data corrections are allowed by pre-allocating a set amount of memory to the right and not being used under normal operation. For example, the card 1 〇 can have a static remapping of 500 KB set aside until the card 1 全 is full due to the host and controller assignment collisions. At this point, the controller 2 can allow additional data for the host to be written to the static allocation area until the desired operation is complete. Later, when the card 1 is substantially full for most uses, a smart filter can be used to allow certain user operations such as deletion and renaming to occur in the static area, while other operations will result in an error. β 124767.doc -16-

Figure 5 shows an alternative to Figure 4. In the scenario t shown in FIG. 5, as before, the host 427 allocates a cluster from the top, and the controller 429 allocates a cluster from the bottom 'however', this will only be recorded by the cluster of the host 4" Stored in the non-volatile memory U3^

in. Therefore, the replica of fatfa stored in the non-volatile memory 433 shows the distribution clusters 0 to 2, and the display clusters 3 to 9 can be used for additional data storage. The distribution of clusters 7 through 9 of the controller 429 is not recorded in the fat 431, but is recorded in a record 435 stored in a volatile memory 437. In this case, the volatile memory 437 is a controller 429. A static random access memory (SRAM), however, other volatile memory can be used. In this case, the record shows the cluster allocated by the host and the controller.

Both clusters are allocated. The records in this example only record the status of a cluster as assigned or deallocated. No additional information is stored in the record. In contrast, the FAT not only records whether or not a cluster is allocated, but also provides additional information such as a cluster below a chain or an end of file indicator. For each cluster, the record 435 can have a single bit (to indicate allocation/de-allocation) and can be viewed as a one-bit map or a cluster map. When the host 427 requests FAT information (generally by transmitting a read request for one or more FAT segments), the controller 429 provides FAT information, which is stored in the FAT 431 of the memory 433. Both the FAT information and the allocation information stored in the record 435 of the volatile memory 437. The request portion of the FAT information is read from the non-volatile memory 433 and transmitted to an editor 439 in the controller 429. The record 435 is also read from the volatile memory 437 and is used by the editor 439 to edit or correct the FAT information from the 124767.doc • 17-1376599 non-volatile memory 43 3 . In addition to being assigned to the FAT 431 by the host 42 7 , the editor 439 changes the FAT information to reflect the assignment of the controller 429. The allocation by the controller 429 can be indicated by indicating the allocated cluster in any suitable manner consistent with the FAT scheme of the host 427. For example, the clusters can be marked as bad clusters, reserved clusters, or used clusters. The corrected FAT information 441 is then transmitted to the host 427. In this way, the host 427 is notified of the non-volatile memory.

The unavailable space' is available by the remapping operation of the controller 429. This prevents the host 427 from attempting to allocate a cluster that the controller 429 has used. In one embodiment, a cluster that has been assigned to the controller is assigned to a file created by the controller. The file may have a name such as "not used to indicate that it is not a host file. This file is displayed by a directory entry that is provided when the host requests to store the directory information of the 5 memory. A directory item that can be modified in the fat information can be added to a directory that is stored in the memory. This allows - the user sees how much capacity is unusable due to controller operation. In particular, this allows a user to interpret any significant difference between the capacity of a memory card and the amount of user data stored prior to reaching a full condition. The edit can be formed to become a dedicated circuit in the controller, or can be implemented by some combination of a lemma or a hard body. When the memory system is first turned on, a record will initially be generated. A controller can scan the memory to construct the initial record. The set of f assigned by the host can be simply (4) by reading the FAT. The set of allocations by the controller can be obtained by scanning the § memory itself or by scanning a management structure for indicating the use of the concealed use of 124767.doc • 18· 1376599 • ‘. When the controller implements a remapping operation, the controller uses an extra cluster and reacts this usage to the record. Initially, the controller can use some space exclusive to the controller. Later, when such exclusive space is fully utilized, the controller can use the logical space that is also used by the host, so the controller also begins recording the use of the workstation in the record. In general, when the amount of scrapped material reaches a limit, the controller begins to use the logic 2 available to the host. The & limit is determined by the amount of spare capacity available. In the middle, the total physical capacity of the memory minus the logical combination of the memory (as seen by the host) minus the control data space and some operational space. In some cases, the controller may indicate the use of the logical space in the - record before using a corresponding portion of the real =. For example, as long as a controller has been scheduled to require some kind of operation in a certain space in the physical memory, the controller can reserve this space and make it unavailable to the host by indicating that it is unusable in the record. . The allocation of the host by % can also be reflected in the record, but this is not necessary because the allocation of the controller is recorded in the FATf. In contrast to the example in which both the host and the controller are recorded in the FAT, in this example, the number of writes to the non-volatile memory has been reduced. Since there are fewer (4) writes to the non-volatile memory, less space is needed to store the fat. ▲ Figure 6 shows an example of a host non-sequentially allocating a cluster. In general, the host is in a sequence of unwritten memory allocation clusters. 'Rather, this is not always the case. As long as a host allocates clusters non-sequentially, 124767.doc 1376599 is the same. Preferably, when the host allocates the cluster column, the control stolen distribution cluster will not generate a full card. In this example, an initially allocated cluster map 651 is obtained from FAT 653 stored in the memory (also

The source generates the cluster map. This map 651 indicates the cluster that is non-sequentially allocated by the host. The initial allocated cluster map 651' is then modified to reflect the cluster allocated by the controller, and a complete cluster map 655 is provided, the controller assigns the cluster from the bottom of the address range to the area of the host allocation cluster, and This area is also stretched in the area of the host allocation cluster. Therefore, in this scheme, the memory full condition is not generated when the host allocation cluster meets the controller knife cluster. Instead, the controller seeks to be used under the remapping operation - the available clusters... the memory king condition can occur only if it meets some other condition, such as allocating all the clusters in the address range. When the host requests FAT information, the FAT information is read from the non-volatile memory and corrected before being transmitted to the host. Figure 6 shows a first map 651 as assigned by the cluster indicated by FAT 653 stored in the non-volatile memory. Figure 6 shows a second mapping 655 as assigned to the cluster indicated by the host. The second map contains the clusters allocated by the controller and the clusters allocated by the host. In this scenario, it is not necessary for the controller to serially distribute the cluster from one end of the address range. A different approach to clustering and controller allocation clustering is possible with a hybrid host. Figure 7 shows a first example of a host operation that results in a change in record. The FAT 761 shows that clusters 0 through 2 are occupied by profile A before the host operates. Both FAT 76 1 and record 763 of Figure 7 show logical space instead of 124767.doc -20- 1376599

The use of physical space. The record 763 also shows that clusters 〇 to 2 are allocated. No other clusters are shown as assigned. The host then assigns data to clusters 0 through 2. The data is written to a new location in the physical memory, which is recorded in an entity-to-logical address translation table. Update the FAT 761 to display the data assigned to the cluster 〇 to 2. Therefore, the allocated space indicated by the FAT 761 is the same as before. However, this entity A uses more space in the memory, because the data B is written to a new location, and the data A still occupies another location. To indicate this usage, the record 763 is updated to show the use of clusters 7 through 9. Therefore, the record indicates the use of a six cluster set, which corresponds to the amount of physical memory used. If the host asks for a FAT message of 9 to 9 at this time, the FAT information returned to the host will indicate that the clusters are 2 and 7 to 9 are unavailable.

Figure 8 shows a second example of a host operation that results in a change in record. Before this host operation, the FAT 871 displays the cluster 〇 to 2 used by the data. The record 873 also shows that these clusters have been allocated. The host then unassigns the clusters 0 through 2. For example, the host may attempt to delete the data occupying the cluster (1) by, for example, marking the cluster 〇 to 2 as available, and by removing the files assigned to the cluster 〇 to 2 from the directory (in the case of shortening a file) In 'can mark the cluster as unavailable' and update the directory entry with the new length). This causes the cluster to be 2 to store additional data in the fat(7). However, the operation of the host makes the space in the physical memory no longer available. In order to reverse: the used space in the physical memory is updated to the record, so if the (4) host requests the bundle 9 then the clusters 7 to 9 will be displayed as unavailable. In the alternative example 124767.doc 21 1376599, even after the cluster 〇 to 2 is deallocated by the stomach host and after (10) is deallocated, the record indicates that the cluster 分配 ' - z is subsequently assigned, when the host requests the cluster 〇 to 2 When FAT information is displayed, it indicates that it is to be assigned. In the case of a peptide, this may cause a problem because the host may have a separation: a record (such as a cached part of FAT), a basin knife, a knife) a buckle does not de-allocate the cluster to 2 〇

The above example means that the controller uses the memory space to store the host data. However, in other situations, the controller can use the memory space for other uses, and use the recorded space as explained to explain the used space in the memory. Erasable or OTP memory can be used. The embedded application running on a memory card in the _ example can use the U between the non-volatile ticks. It can be traced to the use of the space in the application-non-volatile memory and communicated to the host via a cluster map in the volatile memory.

In another embodiment, a cluster mapping reaction in a volatile memory can be used - a distribution of hosts operating according to a different interface. U.S. Patent Application Serial No. 11/196,869 discloses a memory system in which the two host interfaces share the same physical area in the memory. A first host maintains a FAT indicating the use of its logical address space, and a second host uses a different management system that does not require FAT. However, the FAT is modified to display the use of logical space in the FAT of the second host, the logical space corresponding to the space in the physical memory used by the second host to store data. In an alternate configuration, instead of modifying the FAT to react to activity by the second host, a cluster map or similar record can be maintained to show such usage. When the first host requests 124AT.doc -22- 1376599, the FAT information can be read and corrected according to the record to reflect the use of the memory by both the first and second hosts. In an example not shown in Figure 9, the records maintained in the volatile memory can be used to reduce the number of FAT reads stored in the non-volatile memory. When the host requests - one or more sectors 891 of a particular logical address range, the memory controller can check the record of the range to determine if any of the clusters in the range have been allocated. If there is no distribution cluster 1 in the range, the controller can generate FAT information by establishing one or more magic segment states to reflect this. Therefore, there is no need to access any non-volatile memory: if the record indicates a cluster in the address range of the allocation request, then the sell 894 is stored in the non-volatile memory to obtain full coffee information. It can be corrected according to the record before being transmitted to the host_. 895 is read from the non-volatile memory, and the record of the distribution of the error by the paint controller is treated as a file for maintenance. In the above-mentioned formula, only (4) is maintained in the volatile memory, and is not stored in the non-volatile memory. However, in some cases, this file can be used as a power source to cut off the part of the routine or otherwise stored separately in the non-volatile § memory. When initially, this can be reduced in the volatile material. The time required to generate a record in the hidden system of the hidden system. II = Example ''The cluster allocated by the host and the controller: Loss of use, :: By comparing the FAT The allocation of indications and the failure of the actual memory usage are discriminated during the analysis period. The sufficiency space is allocated in units of clusters. The cluster size is generally 124767.doc -23· The memory system used by the end, and possibly In the case of the memory system, a cluster consists of 32 segments of data. As long as the non-volatile memory system can be erased, the size of a block (minimum erase unit) can also be 32 regions. In other examples, a block may contain two or more clusters. All patents, patent applications, articles, books, specifications, other publications, documents, and things referred to herein are hereby incorporated by reference in their entirety as Place Use of it. In the event of any inconsistency or conflict between the definition or use of a term between any public disclosure, document or thing and the body of this document, the definition or use of that term in this document shall prevail. Although the various aspects of the invention have been described in terms of some preferred embodiments of the invention, it is understood that the invention is in the scope of the appended claims. Displaying a host device in communication with a memory device including a write-once memory array. Figure 2 shows a memory map of the write-once memory array. Figure 3 shows a memory map containing the Write the use of the space within the memory array, where the host allocates the cluster from one end, and the controller allocates the cluster from the other end of the address range. Figure 4A shows a logical address space of a write-only memory. The memory map includes a file allocation table (FAT), and the file allocation table reflects both the allocation by the host and the allocation by the controller. The physical address of the memory in Figure 4A before storing the host data is 124767.doc • 24· 1376599 A mapping of the space β Figure 4C shows the memory of Figure 4B after the host data is updated and the control data is updated to deer the host. A map of the physical address space of the intercepted entity. The figure shows a mapping of the physical address space of the memory of Figure 4C after receiving and storing an update to the host data, and storing the updated control. Figure 5 shows - Alternative to the embodiment 'where stored in non-volatile memory

The -FAT reaction in the body is allocated by the host, and a record response stored in the volatile memory is not the information of the fat and the record by the allocation of the controller, in response to the FAT information. Host request. Figure 6 shows an embodiment in which a host non-sequentially allocates a space in a hidden object, and a controller allocates space in a manner that accommodates the allocated space.

Figure 7 shows an embodiment in which a host allocates updated data reflected in a record of one of the non-volatile memories. Figure 8 shows an example in which a host de-allocates a number of clusters and updates the record to indicate the use of an equal number of clusters. Figure 9 shows a flow chart of a technique for recording the number of non-volatile memory readings using one of the volatile memories. [Main component symbol description] 5 Host device 7 Multi-write slot system 10 Memory device 124767.doc •25- 1376599

15 Interface 20, 429 Controller 30 Write-only memory array 50 Memory map 301 Logical address space 303 Host allocation space 305 Controller allocation space 407, 431, 653 File allocation table in non-volatile memory 409 Physical memory Body 411 Part of the physical memory 413, 415, 419, 421 Physical location 417, 425 Part of the control data 423 Part of the host data 427 Host 433 Non-volatile memory 435, 763, 873 Record 437 Volatile memory 439 Editor 441 Corrected file allocation table information 651, 655 cluster map 761, 871 Tan file allocation table 124767.doc -26-

Claims (1)

Patent application scope: A method for managing a non-volatile memory array material used by a host to store KI ^ 仔 ' ' 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该 该A file allocation table in a volume array, the method comprising: receiving a request from the host for a portion of the file allocation table stored in the non-volatile memory array; reading the (four) allocation table from the non-volatile memory array The portion of the file allocation table is modified according to a record, and the memory controller is not used by the memory controller to store at least the address of the data received from the host; and the slot allocation table has been modified. Partially transferred to the host. 2. If the request item, 土. 10,000 goes, where the record is maintained in a volatile memory, OJL JL '' Yang Fei is maintained in the non-volatile memory array. 3. If the request jg 1 + +, A method ', where the non-volatile memory array is not erasable. 4. The method of claim 1 wherein the record is a cluster map comprising a plurality of clusters each of which is an allocation/deallocation state of the cluster. 5. The method of claim 4 ^ ^ 1 wherein the host responds to the allocation of data to a first month 1J has been divided into gp f隹 the most set and the use of the address by the memory controller occurs. 6 • If the method is requested, it should be used by the memory controller to de-allocate the previously allocated user | # /, '. 124767.doc 1376599 A method of managing data in a non-volatile memory array, comprising: receiving a first portion of one of data from a host, the first portion of the data being assigned by the host to a previously assigned one a first cluster; storing the first portion of the data in the non-volatile memory array; receiving distribution information from the host, the distribution information linking the data The first portion and the first cluster, the allocation information indicates that a second cluster is available for storing additional data; storing the allocation information in the non-volatile memory array; and subsequently, responding to a host of the allocation information Requesting, reading the allocation resource m from the non-volatile brother memory array, and correcting the & allocation information indicating that both the first cluster and the second cluster are unavailable for storage of additional data. The method of data, wherein one of the records is recorded in the non-volatile management, and the non-volatile memory host distributes the data in the cluster, and the method includes: generating a cluster map, Instructing the allocation/de-allocation state of each cluster of the plurality of clusters; maintaining the cluster mapping in the -volatile memory; responding to receiving a host for the complex information about the plural gray, the singularity, and the money From the cluster map, the county determines whether the county allocates the plurality of clusters. If the plurality of clusters are not allocated, the generated information is transmitted back to the host. 124767.doc -2^ 9. In the method of claim 8, the at least one of the cows is assigned to the plurality of clusters. The set reads the allocation 1 from the non-volatile memory. • As in the method of claim 9, the allocation is made, and the record is corrected. The record is read by u to provide the corrected allocation information to the host. The method of item 0, which causes the at least one cluster of the plurality of clusters to be corrected. The knife is assigned to 2' species management in a non-volatile $, accompanied by a method of non-volatile (four) material, the (four) Φ a ~ 彡 has been assigned to the bedding stored in the non-volatile memory - Logical address space, the method includes: recording, by the allocation of the address space of the host, an auditory system structure stored in the array of the initial memory; the address space of the host is not used by the host The distribution record is stored in - swaying! • Producing a "hidden array" rather than a record in the non-volatile memory array; and responding to a host request for allocation of data, reviewing the record, and providing a response to the host based on the contents of the record. 13. The method of claim 2, wherein the allocation of the address space not by the host is used by a memory controller to update the allocation of the address space of the control data. 14. The method of claim 2, wherein the allocation by the address space of the host does not record its allocation in the file system structure by the allocation of the address space of the other host. 15. The method of claim 12, wherein the location of the address space of an embedded application within the non-volatile memory system is allocated by the address space of the host device 124767.doc 1376599. 16. The method of claim 12, wherein the response comprises assigning a particular broadcast to an indication of the address space to be allocated by the record. 1 7. A non-volatile memory system, comprising: a non-volatile memory array storing data from a host, and storing to indicate that the data is allocated by the host to a logical address space At least a first portion of a file system structure; a controller that manages data in the non-volatile memory array, the controller assigning at least a second portion of the logical address space, not in the file system structure Indicating an allocation of the second portion; and a volatile suffix that records the allocation of the second portion of the logical address space of the controller. ...the sinusoidal memory system of claim 17 wherein the non-volatile memory array is not erasable. 19. The non-volatile memory system of claim 17, wherein the volatile memory and the controller are formed on a common die. 20. The non-volatile memory system of claim 17, wherein the second portion of the logical address space is data transmitted by the controller to the host to update previously stored in the non-volatile memory array Information in the middle. 21 - The non-volatile memory memory system of claim 17 - the second portion of the space is controlled by the control memory system, further comprising the non-embedded application 'the logical addresser is assigned to the internal Information stored in the embedded application program 124767.doc. 22. The non-volatile memory system of claim 7, further comprising an editor in the controller, the editor correcting the file m structure when reading from the non-volatile memory array, and then The modified system structure has been transferred to the host. 23. The non-volatile memory system of claim 22, wherein the editor modifies the structure based on information recorded in the volatile memory. 4. The non-volatile memory system of claim 17, wherein the non-volatile memory system is embedded in a removable memory card. 2. The non-volatile memory system of claim 17, wherein the controller checks the volatile memory for a host request from the non-volatile memory system for the allocation information of the logical address. And if the volatile memory indicates that the logical address is not allocated, the controller returns the following one finger*: the logical address is not separated, and the allocation information is read from the non-volatile memory array. . 26. A non-volatile memory system, comprising: a non-volatile memory array storing data from a host and storing a file system structure for indicating allocation of data by the host, and '- a memory controller' that manages data in the non-volatile memory array, the memory controller including an editor that corrects from the non-volatile memory based on a record stored in a volatile memory The file system structure of the volume array, the memory controller responds to the host request 124767.doc 1376599 to transfer the modified file system structure to the host. 27. The non-volatile memory system of claim 26, wherein the record indicates an allocation of logical space by the controller. 28. The non-volatile memory system of claim 26, wherein the non-volatile memory array is non-erasable. 29. The non-volatile memory system of claim 26, wherein the non-volatile system is replaceable in a memory card according to an interface interfacing with a host. 124767.doc