JPH06259339A - Semiconductor memory and control method - Google Patents

Abstract

PURPOSE:To decrease the amount of data to be saved from a semiconductor memory to a backup device in case of a power source disconnection or power failure. CONSTITUTION:When the power source is turned ON, data on the backup device are loaded to the semiconductor memory. An update table has update flags corresponding to the respective blocks of the semiconductor memory. An ON update flag indicates that data need to be save and an OFF update flag indicate that the data need not to be saved. When a write access request is sent, a host interface system processing means saves update data so that the number of update blocks whose update is not reflected on the backup device in the semiconductor memory is less than a prescribed value (n) on condition that the sum of the number of ON update flags and the number of blocks of write data exceeds the prescribed value (n). A sequential backup system processing means saves the update blocks of the semiconductor memory unless an access request is processed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention In a semiconductor memory device having a semiconductor memory and a backup device, the amount of data to be saved from the semiconductor memory to the backup device can be reduced when the power is cut off or a power failure occurs. The present invention relates to a semiconductor memory device and a control method.

[0002]

2. Description of the Related Art FIG. 9 is a diagram for explaining a conventional semiconductor memory device. In the figure, 1 is a controller, 9 is a backup device, 10 is a semiconductor memory, 11 is a semiconductor memory device, and 12 is a host. Also,,
, And ′ indicate the data flow, respectively.

The semiconductor memory device 11 includes a controller 1,
The backup device 9 and the semiconductor memory 10 are included. The controller 1 is composed of a processor, for example, and controls the entire semiconductor memory device. The backup device 9 is, for example, a magnetic disk device. The semiconductor memory 10 is divided into a plurality of blocks and managed. The size of the semiconductor memory 10 is 128 MB, for example, and the size of one block is 256, for example.
It is a byte. The block means the semiconductor memory device 1
It is also a unit of data transfer between 1 and the host 12.

FIG. 9A is a diagram for explaining the first conventional example. When the power of the semiconductor memory device is turned on, the data of the backup device 9 passes through the path of the semiconductor memory 11
Loaded on. Host 1 after loading
Data is exchanged through a path between the semiconductor memory 10 and the semiconductor memory 10. When the power is turned off or a power failure occurs, all the data stored in the semiconductor memory 10 is saved in the backup device 9 via the path. When a power failure occurs, the power required for this evacuation operation and system close processing is supplied from the secondary power supply device (secondary battery).

In the first conventional example shown in FIG. 9 (a), all the data stored in the semiconductor memory 10 is saved in the backup device 9 when the power is turned off. It has a drawback that it is necessary to prepare a secondary power supply device.

The second conventional example shown in FIG. 9 (b) is designed to eliminate the drawbacks of the first conventional example. The rectangle with diagonal lines indicates update data (update block). Second
In the conventional example, the update data stored in the semiconductor memory 10 is sequentially saved to the backup device 9 via the path of 'while the access request from the host is not processed. Then, when the power is turned off, the update data that is not reflected in the backup device among the update data stored in the semiconductor memory is saved in the backup device. According to the second conventional example, most of the update data stored in the semiconductor memory 10 is already reflected in the backup device 9 when the power is cut off or a power failure occurs. Semiconductor memory 10 at the time of occurrence
Therefore, the amount of data to be saved in the backup device 9 can be reduced.

The second conventional example contributes to shortening the system closing time and reducing the capacity of the secondary power supply device. However, if write access from the host 12 is frequently performed, a large amount of updated data that is not reflected in the backup device 9 will accumulate in the semiconductor memory 10, and a large amount of data will be saved in the backup device 9 when the power is turned off or a power failure occurs. I have to do it.

Further, in the conventional semiconductor memory device, data is always backed up to the same storage area of the backup device, so if the data cannot be properly backed up from the semiconductor memory to the backup device, the data will be lost. There is a drawback that it will end up.

[0009]

SUMMARY OF THE INVENTION The present invention was created in view of this point, and even in the case where write access from the host is frequently performed, when the power is turned off or a power failure occurs. It is an object of the present invention to provide a semiconductor memory device capable of reducing the amount of data to be saved from a semiconductor memory to a backup device. Also,
Another object of the present invention is to provide a method of controlling a semiconductor memory device that can minimize the loss of data.

[0010]

FIG. 1 is a diagram for explaining the principle of the present invention. According to another aspect of the semiconductor memory device of the present invention, a semiconductor memory divided into a plurality of blocks for management, a backup device, a power-on processing means, a host interface processing means, a sequential backup system processing means, a power-off processing. And an update table and a controller having an update table control means, wherein the update table has an update flag for each block of the semiconductor memory, and the update table control means is a block of the semiconductor memory. When the write data sent from the host is written to, the corresponding update flag is turned on,
When the same data as the data in the block of the semiconductor memory is written in the backup device, the corresponding update flag is turned off, and the power-on processing means is started at the time of power-on to transfer the data of the backup device to the semiconductor memory. The host interface processing means is operated when an access request is sent from the host, and when the access request is a write access request, the write data is sent. The number X of update flags that have been turned on before being read and the number N of blocks of the write data are added, and if the addition result exceeds the specified value n, it is reflected in the backup device of the update blocks in the semiconductor memory. Back up the update block data so that the number of unupdated update blocks is less than or equal to the specified value n. And the sequential backup system processing unit performs processing on condition that the host interface system processing unit is not processing, and corresponds to the ON update flag in the update table. The semiconductor memory block data is written to the backup device, and the power-off processing unit is activated when the power is turned off, and writes the data of the semiconductor memory block corresponding to the ON update flag in the update table. It is characterized in that it is configured to perform a writing process to a backup device.

A method of controlling a semiconductor memory device according to claim 2 is
In a semiconductor memory device in which data is loaded from the backup device to the semiconductor memory when the power is turned on, and the same contents as the contents of the semiconductor memory are stored in the backup device when the power is turned off. divided and a plurality of equal division area and capacity of the semiconductor memory _{(a 1, a 2, ...} ) to generate the data in the semiconductor memory from the divided areas of the backup device that stores latest data (a _i) at the time of power-on Loading,
The update data for the semiconductor memory is backed up to the next divided area (A _{i + 1} ) of the backup device.

[0012]

The operation of the semiconductor memory device according to claim 1 will be described. When the semiconductor memory device is powered on, the power-on processing means operates to load data from the backup device into the semiconductor memory. The semiconductor memory is divided into a plurality of blocks and managed. The update table has an update flag corresponding to each block of the semiconductor memory. The update table control means turns on an update flag corresponding to when write data from the host is written to a certain block of the semiconductor memory, and corresponds to when data of a certain block of the semiconductor memory is written to the backup device. Turn off the update flag.

When an access request is sent from the host after the loading is completed, the host interface processing means operates to check whether the access request is read or write. In the case of writing, the number X of update flags that are ON before the write data is sent and the block number N of the write data are added, and whether the addition result is larger than the specified value n or not. To find out. In the case of a large value, the update data is written to the backup device so that the number of update blocks in the update blocks of the semiconductor memory that are not reflected in the backup device is equal to or less than the specified value n. Naturally, the write data from the host is written in the semiconductor memory.

The sequential backup processing means operates when the host interface processing means is not operating. The sequential backup system processing means performs a process of writing the data of the block of the semiconductor memory corresponding to the ON update flag in the update table to the backup device. The power-off time processing means starts an operation at the time of power-off and writes the data of the block of the semiconductor memory corresponding to the ON update flag into the backup device.

The operation of the semiconductor memory control method according to the second aspect will be described. It is assumed that the latest data storage area is the area A ₁ . When the power of the semiconductor memory device is turned on, data is loaded from the area A ₁ into the semiconductor memory. When loading from the area A ₁ , the next area A ₂ is set as the backup target area, and the backup of the updated data is performed on the area A ₂ . When the power is turned off or a power failure occurs, the contents of the semiconductor memory are the same as the contents of the area A ₂ . In the example shown, the next time the power is turned on,
Data is loaded from the area A ₂ into the semiconductor memory, and the area A ₁ is set as the backup target area.

[0016]

2 is a block diagram of an embodiment of a semiconductor memory device of the present invention. In the figure, 1 is a controller,
2 is a main body I / F control LSI, 3 is a microprogram storage area, 4 is a work area, 5 is a disk control LSI,
6 is a processor, 7 is a DMA, 8 is a buffer, 9 is a backup device, and 10 is a semiconductor memory.

The controller 1 is an LS for controlling the main body I / F.
I2, microprogram storage area 3, work area 4, disk control LSI 5, processor 6, DMA 7, buffer 8 and the like. Main body I / F control LSI2
Controls the exchange of data between the controller 1 and the host. A microprogram is stored in the microprogram storage area 3. The work area 4 has information indicating a disk area to be loaded,
The number X of update blocks, a prescribed value n, an update table, a pointer to the update table, an update history table, etc. are stored. The number of updated blocks means the number of updated blocks in the semiconductor memory that have not been updated by the backup device.

The disk control LSI 5 controls data exchange between the controller 1 and the backup device 9. The processor 6 executes the microprogram in the microprogram storage area 3. Body I / F
The control LSI 2, the disk control LSI 5, the DMA 7, etc. are controlled by the processor 6.

The DMA 7 is an L for data transfer between the main body I / F control LSI 2 and the semiconductor memory 10 and a disk control.
It controls data transfer between the SI 5 and the semiconductor memory 10. The DMA 7 is an LS for controlling the main body I / F.
L for data transfer and disk control between I2 and buffer 8
It also controls the data transfer between SI5 and buffer 8. The buffer 8 has a size of 64 KB, for example. Buffer 8
Is used when writing data from the host to the backup device 9 without passing through the semiconductor memory 10 or when sending data from the backup device 9 to the host without passing through the semiconductor memory 10.

The backup device 9 is, for example, a magnetic disk device. In the illustrated example, the area of the backup device 9 is divided into an area A and an area B. The area A has the same size as the semiconductor memory 10, and the area B also has the same size as the semiconductor memory 10. The semiconductor memory 10 is divided into a plurality of blocks and managed. The size of the semiconductor memory 10 is 128 MB, for example.
The block size is 256 bytes, for example. In addition,
A block is a unit of data transfer between a host and a semiconductor memory device. Data transfer between the semiconductor memory 10 and the backup device 9 is also performed in block units.

FIG. 3 is a diagram for explaining the update table.
The update table has an update flag corresponding to each block. The update flag of 0 indicates that it is not necessary to save the data of the corresponding block to the backup device, and the update flag of 1 indicates that it is necessary to save the data of the corresponding block to the backup device.
The update flag is set to 1 when data is written to the corresponding block, and is set to 0 when data of the corresponding block is written to the backup device. A pointer is provided for the update table.

FIG. 4 is a diagram for explaining the processing when the semiconductor memory device is started up. When the power is turned on, the host notifies the controller 1 of the system close time and the capacity of the secondary power supply device. Upon receiving this notification, the controller 1 calculates and stores the specified value n. In the present invention,
As will be described later, the number of updated blocks stored in the semiconductor memory 10 that are not reflected in the backup device 9 is always set to a specified value n or less. After storing the specified value n, the controller 1 determines whether the latest data storage area is the area A or the area B, and loads the data in the area into the semiconductor memory 10. When the loading is completed, the semiconductor memory device goes into a normal operating state.

A calculation example of the specified value n will be described. It is assumed that the system close time notified from the host = 5 minutes and the secondary power supply capacity = 8 minutes. Upon receiving such a notification, the controller 1 determines that the backup needs to be completed within 8-5 = 3 minutes. If the evacuation time for one block is 0.5 seconds, the number of blocks that can be backed up within 3 minutes is 3 × 60 ÷ 0.5 = 360 blocks. The controller automatically sets the specified value n. In this case, n may be any value as long as it is 360 or less,
The larger is, the lower the probability that the host will wait,
Set the maximum value.

FIG. 5 is a diagram for explaining the processing of the host interface system. When an access request is sent from the host, the processor of the controller executes the host interface system microprogram stored in the microprogram storage area. By executing this microprogram, the following processing is performed. Note that, for simplicity of explanation, it is assumed that only the area A exists in the backup device. In step S1,
Determine whether the access request is read or write. In the case of reading, the process proceeds to step S5, and in the case of writing, the process proceeds to step S2. In step S2, old X + N is calculated and the result is set to new X. However, X is the number of blocks among the updated blocks in the semiconductor memory whose update is not reflected in the backup device, that is, the total number of update flags of 1. N indicates the number of updated blocks this time.

In step S3, it is checked whether new X> n. If Yes, go to step S4, No
In the case of, it progresses to step S6. In step S4,
All of (new X-n) or new X is written to the backup device 9 in the through mode, and the process is completed. When finished, notify the host of the end. The write data of N blocks sent from the host can be written in the backup device 9 as it is. It is natural to write the write data sent from the host into the semiconductor memory. In step S5, the semiconductor memory 10
To read the data, send this data to the host, and end. In step S6, the data sent from the host is written in the semiconductor memory, and the process is ended.

FIG. 6 is a diagram for explaining the processing of the sequential backup system. The sequential backup system is activated when the trigger condition is satisfied. The value of X or time may be used as the trigger condition. When using the value of X, the condition of X ≠ 0 may be used, or a certain set value m (0
≦ m <n) may be determined in advance and X may exceed m. Further, time may be used as a trigger regardless of the value of X. For example, the search logic may be run once every three minutes. Furthermore, it is possible to trigger both the value of X and the time.

The host interface processing has a higher priority than the sequential backup processing. That is, when an access request is sent from the host during the processing of the sequential backup system, the controller interrupts the processing of the sequential backup system, starts the processing of the host interface system, and After the processing of (1) is completed, the processing of the backup system is restarted sequentially. Even if the trigger condition for starting the processing of the sequential backup system is satisfied during the processing of the host interface system, the start of the processing of the sequential backup system is delayed until the processing of the host interface system is completed.

When the trigger condition is satisfied, the controller sequentially starts the processing of the backup system. The processing of the sequential backup system is performed by executing the microprogram for the sequential backup system in the microprogram storage area. First, it is checked whether the update flag designated by the pointer of the update table is 0 or 1, and if the update flag is 1, the data of the corresponding block is backed up in the backup device and the processing is terminated. The update flag is 0
In this case, the pointer is incremented by 1, and it is checked whether the update flag designated by the updated pointer is 0 or 1.

When the power is turned off or a power failure occurs, all the data in the block of the semiconductor memory 10 corresponding to the update flag 1 of the update table (indicating the need to save) is saved in the backup device 9. There are roughly two ways to turn off the power:
There are different types of methods. System Close The operator's command issues a command to end all operations, and the software that receives it closes the system and requests the power control hardware to disconnect it. Environmental abnormalities Hardware detects environmental abnormalities such as power outages and temperature abnormalities,
Turn off the power a few minutes after notifying the software.

From the standpoint of the semiconductor memory device,
Both, software (OS) is supposed to send a command to the semiconductor memory device. The command in this case is a command having the meaning of "save all existing update blocks to the backup device".

On the other hand, the semiconductor memory device (control MPU) always manages the presence / absence of an update block, and if there is an update block, the hardware signal is turned on. This hardware signal can be detected by the power supply control hardware, and the power supply control hardware finally determines whether or not the power supply may be turned off based on the presence or absence of this signal.

For example, it cannot be said that there is no software down or software hang-up when the system is closed or during software processing of an abnormal environment. In such a state, the semiconductor memory device is powered off without the above command being issued, and data is lost. The power control hardware can determine whether or not such a state has occurred by looking at the above-mentioned hardware signal. In this case, the power control hardware operates another hardware signal and directly outputs the signal to the semiconductor memory device. Specifically, a notification equivalent to the command "save all existing update blocks to the backup device" is issued.

FIG. 7 and FIG. 8 are diagrams for explaining high reliability by dividing the area of the backup disk. The area of the backup disk is divided according to the capacity of the semiconductor memory. For example, the capacity of the semiconductor memory is 100M
In the case of B and the disk capacity is 300 MB, the area of the backup disk is divided into three, and the areas obtained by the division are referred to as A, B and C. For the sake of simplicity, the following description will be made with two divisions (two areas A and B).

At the beginning of using the semiconductor memory device, the contents of the backup device such as a disk are blank and all are Nu.
ll data is stored. This is equivalent to a state where the same data is stored in the areas A and B without repairing. When the semiconductor memory device is powered on in this state, data is first loaded from the area A.
This state is shown in FIG. 7 (a).

After the loading is completed, the data in the area A is expanded in the semiconductor memory, and when the host issues a read access request, the data in the semiconductor memory is sent to the host, and when the host issues a write access request,
Write data from the host is written in the semiconductor memory. Assuming that the block updated at the time of writing is the first block, A-1 (A is the disk area number) is stored as the update history. Similarly, the subsequent write access points that have been updated in the even _{A-x 2, A-x} 3, ...
Memorize as. On the other hand, sequential backup system and through mode backup (process of step S4 in FIG. 5)
Backups the backup target area as area B. Similarly, when the semiconductor memory device is powered off, the area B is also targeted for backup. After this operation, the semiconductor memory device is powered off. This state is shown in FIG. 7 (b).

When the power is turned on next time, loading is started from the area B. Then, before starting the normal operation, the update history of one time before (in this case, the above A-1, A-x ₂ , ...)
The contents of are sequentially backed up. At this time, the update history of the previous time is cleared once the backup is completed. This state is shown in FIG. 8 (c).

After that, when the normal operation is started and a write access request is sent, B-x ₁ B-
x _2, continue to create the update history data to say ... and. In the sequential backup system and the through mode backup, the backup target area is backed up as the area A, and the backup when the power of the semiconductor memory device is turned off is also backed up in the area A. After this operation, the semiconductor memory device is powered off. This state is shown in FIG. 8 (e).

When the power is turned on next time, the disk is divided into areas by loading from the area A and always referring to the previous update history to perform sequential backup, through mode backup, and backup at power off. However, the backup time can be shortened when the power is cut off or a power failure occurs. Note that in the case of being divided into two, the contents of the update history of the previous one is targeted for sequential backup before entering the normal operation, but in the case of being divided into three, the contents of the update history of the previous one are recorded. The contents and the contents of the update history two times before are sequentially backed up.

[0039]

As is apparent from the above description, according to the present invention, the number of update blocks among the update blocks in the semiconductor memory whose updates are not reflected in the backup device is always set to the specified value n or less. Therefore, it is possible to reduce the number of update blocks to be backed up when the power is cut off or a power failure occurs, and it is possible to reduce the capacity of the secondary power supply device.

Further, according to the present invention, since the inside of the disk is used evenly, the life of the medium can be extended. Even if the device or system goes down for some reason, the data before the complete state is always saved,
Easy to recover. If the complete data of the previous time does not exist, only the data halfway updated by sequential backup or the like remains, so that recovery is difficult. If the setting allows the previous data to be destroyed, when a medium error is detected while backing up to the divided area, the previous divided area is changed to the latest area. As a result, the evacuation process can be performed normally. It is possible to achieve such effects.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a block diagram of an embodiment of a semiconductor memory device of the present invention.

FIG. 3 is a diagram illustrating an update table.

FIG. 4 is a diagram showing a process when the semiconductor memory device is started up.

FIG. 5 is a diagram showing processing of a host interface system.

FIG. 6 is a diagram showing processing of a sequential backup system.

FIG. 7 is a diagram for explaining how to improve reliability by dividing an area of a backup disk.

FIG. 8 is a diagram for explaining high reliability (continuation) due to area division of a backup disk.

FIG. 9 is a diagram illustrating a conventional semiconductor memory device.

[Explanation of symbols]

 1 Controller 2 Main Body I / F Control LSI 3 Micro Program Storage Area 4 Work Area 5 Disk Control LSI 6 Processor 7 DMA 8 Buffer 9 Backup Device 10 Semiconductor Memory

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Yuihide Komatsu 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Hitomi Monzen, 98, Unokenu, Unoku-cho, Kawakita-gun, Ishikawa Prefecture In PFU Co., Ltd. (72) Inventor Yoshiro Hirai 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited

Claims (2)

[Claims] 1. A semiconductor memory managed by being divided into a plurality of blocks, a backup device, a power-on processing means, a host interface processing means, a sequential backup system processing means, a power-off processing means, and an update table. And a controller having update table control means, wherein the update table has an update flag corresponding to each block of the semiconductor memory, and the update table control means sends the block to the semiconductor memory from the host. When the write data is written, the corresponding update flag is turned on, and when the same data as the block of the semiconductor memory is written to the backup device, the corresponding update flag is turned off. The power-on processing means is started when the power is turned on, and the backup device The host interface processing means is operated when an access request is sent from the host, and when the access request is a write access request, the write operation is performed. When the number X of ON update flags before the data is sent and the block number N of the write data are added, and the addition result exceeds the specified value n, among the update blocks in the semiconductor memory, The processing for writing the data of the update block to the backup device is performed so that the number of update blocks not reflected in the backup device becomes equal to or less than the specified value n. The sequential backup system processing means is a host interface system process. Processing is performed on condition that the means is not processing, and it corresponds to the ON update flag in the update table Is configured to perform the process of writing the data of the block of the semiconductor memory to the backup device, and the power-off processing means is activated at the time of the power-off, and the data of the block of the semiconductor memory corresponding to the ON update flag in the update table. A semiconductor memory device, which is configured to perform a process of writing data to a backup device. 2. A semiconductor memory device in which data is loaded from a backup device into a semiconductor memory when the power is turned on, and the same contents as the contents of the semiconductor memory are stored in the backup device when the power is turned off. by dividing the storage area of the device, the divided region of the plurality of equal division area and capacity of the semiconductor memory _{(a 1, a 2, ...} ) to generate a backup device for storing latest data on power-up (a _i) A method for controlling a semiconductor memory device, comprising: loading data from a memory to a semiconductor memory and backing up updated data for the semiconductor memory to a next divided area (A _{i + 1} ) of the backup device.