JPH07311710A - Buffer cache managing method for computer - Google Patents

Abstract

PURPOSE:To improve the utilization efficiency of a buffer cache of buffer cache size corresponding to the highest frequency by storing an access request size in a buffer cache management table. CONSTITUTION:The upper-limit value and lower-limit value of access size as use standards of the buffer cache 8 and the number of request headers 6 are set in a buffer cache use definition table 3. Then, the size of a data storage area prepared in a request header 6 is entered as the request header extension size of the buffer cache use definition table 3 on the basis of the lower-limit value of the access size. Namely, data of data size corresponding to the highest frequency are left on the buffer cache 8 preferentially according to the upper- limit value and lower-limit value in the buffer cache use definition table 3, so the utilization efficiency of the buffer cache 8 of buffer cache size corresponding to the highest frequency can be improved when the buffer cache 8 is reused in a random memory device 1.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of managing data transfer between a non-volatile memory of a computer and a buffer cache provided on a random access memory by an operating system.

[0002]

2. Description of the Related Art Conventionally, in a computer, access time to a file on a disk memory which constitutes a non-volatile memory such as a floppy disk or a hard disk has been an important factor affecting the performance of the computer. Therefore, a method is adopted in which a part of the data stored in the disk memory is read into a buffer cache prepared on the random access memory, and subsequent user access is performed via the buffer cache.

There are one to several kinds of buffer cache sizes prepared in the computer. When several kinds of sizes are prepared, the user selects the buffer cache with the most frequently accessed size in the file and connects the disk memory as a file. When the user makes an access request, the free buffer cache allocation method is LRU (L
The most recently used algorithm is generally used; an algorithm that allocates the earliest accessed buffer cache as a free buffer cache to a new access request; the longest unused algorithm).

Further, since data is transferred between the disk memory and the buffer cache, a flash operation (sync operation) for reflecting the data in the buffer cache in the disk memory is required at some timing.
There are two of them, one is a write-through method of writing data to the buffer cache and at the same time writing to the disk memory. The other is a method of transmitting the flash timing of a program that is periodically activated to the operating system by a system call, and the operating system collectively flashes. With this operating system, the importance and order of the data to be flashed is not considered, so if a power failure occurs during flashing, for example, some of the data that makes up the file will be lost, and so the file structure will be configured. It is difficult for the user to predict the damage situation in the file or to control the user so that the damage does not occur until the file is destroyed where the data is lost.

With the advent of the large-capacity semiconductor disk memory, access to a file constructed on the semiconductor disk memory also requires access via the buffer cache, so that the semiconductor disk memory is accessed via the buffer cache. Data overhead at the time of power failure, if you create the overhead of memory-to-memory copy to copy to the user buffer, or if the semiconductor disk memory body has a battery backup function but the buffer cache itself does not have a battery backup function,
Problems such as coherency between the buffer cache and media are not resolved.

[0006] FIG. 11 is a schematic diagram of Maurice J. J. published by Kyoritsu Publishing Co., Ltd. Bach
It is a figure which shows the buffer cache management method described in "Design of UNIX kernel" pages 32-49. In FIG. 11, 7 is a buffer header, 8 is a buffer cache managed by operating a pointer on the buffer header 7, 100 is a free list for managing the usable buffer headers 7, and 101 is the head of the free list 7. And 110 are a free list header that holds the end of the hash list, 110 is a hash list for accessing the buffer cache 8 at high speed, and 111 is a queue header that manages the beginning and end of the hash list 110.

The buffer header 7 has a device number, a block number, a status field and various pointers.
The device number and the block number are for specifying the data position on the disk memory. The status field indicates the usage status of the buffer cache 8 such as whether the buffer cache 8 is being used or whether the buffer cache 8 has been changed. The pointer to the buffer cache 8 in the various pointers indicates the buffer header 7. The buffer header 7 is managed by two management structures, one of which is the free list 100 and the other of which is the hash list 110. The free list 100 and the hash list 110 have a bidirectional list structure. That is, the free list 100 has a front pointer 102 and a rear pointer 103.
The queue of the hash list 110 is also managed by a forward pointer and a backward pointer not shown.

The LRU for managing the allocation of the buffer cache 8 will be described. A buffer cache containing the contents of blocks of the disk memory body, that is, a buffer cache allocated to the disk block 8
Will not be allocated to another disk block until all other buffer caches 8 have been used. Operating system is buffer cache 8
Manages the free list 100 of the buffer cache 8 in order to save the order in which was used. All buffer caches 8 are tied to the free list header 101 when the operating system is booted. When a new free buffer cache is needed, the operating system fetches the buffer cache 8 from the beginning of the free list header 101. However, when the buffer cache 8 allocated to a specific disk block is required, it is taken out from the middle of the free list 100 and used.

If the operating system returns the buffer cache 8 to the free list, the free list 100
To the end of the free list 100, and to the beginning of the free list 100 in special cases such as when an error occurs. In any case, it is not returned to the middle of the free list 100. Therefore, the buffer cache 8 at the head of the free list 100 is older than the one at the rear of the free list 100. If there is no free buffer cache on the free list 100, the user will have to wait until a free buffer cache is created.

When the operating system accesses the buffer cache 8 assigned to a specific disk block, it searches by a combination of a device number and a block number. In this case, rather than searching the entire buffer cache 8, the buffer header 7 is managed by individual queues hashed as a function of device number and block number. Hash list 110 of FIG.
Then, a case where four queues are created by the remainder (mod) obtained by dividing the block number by 4 is shown as an example.

In the read operation by the operating system, when the buffer cache 8 holding the data of the target disk block is not in the queue header 11, a free buffer cache is found from the free list 100 and the status field of the buffer header 7 is found. Is marked as being used, the disk block is once read from the corresponding disk block, and then copied to the user buffer. On the contrary, when the buffer cache 8 holding the data of the target disk block exists on the queue header 111, it is read from the buffer cache 8. After reading, the in-use mark in the status field of the buffer header 7 is erased, and the buffer cache 8 after reading is connected to the end of the free list 100.

Similarly, the write operation by the operating system searches for the target buffer cache 8 on the queue header 111, and if not, finds a free buffer cache from the free list 100 and buffers it from the corresponding disk block. After the data is read into the cache 8, the buffer cache 8 is modified, the status field of the buffer header 7 is marked, and the modified buffer cache 8 is connected to the end of the free list 100. That is, writing to the disk memory is not performed immediately.

The data in the buffer cache 8 is written in the disk memory in the following two cases.
One is a free list for reading or writing 1
When searching for a free buffer cache from above 00, the one marked with a correction in the status field on the buffer header 7 is found. The other is when the operating system periodically searches through all the buffer headers 7 and writes the status field marked as modified.

Therefore, in the case of the write operation, there is a time lag between the correction on the buffer cache 8 and the flash operation for reflecting the correction content on the corresponding disk block, and during the time lag, the power supply is changed. If the operating system becomes inoperable due to a disconnection or a crash, a conflict will occur between the buffer cache 8 and the disk block.

[0015]

The above conventional buffer cache management method has the following problems. (1) Since the buffer cache 8 is used for all file access requests, the buffer cache 8 must be secured even for a large-capacity access size with a low access frequency. The buffer cache 8 with a high access size becomes invalid, and the utilization rate of the buffer cache 8 for requests with the buffer cache size decreases. (2) Since the buffer cache 8 is used for all file access requests, the buffer cache 8 must be secured even when accessing a small file, and the buffer cache 8 is not used efficiently. For example, if the buffer cache size is 1 Kbyte and the file size is 100 bytes. 924 bytes are wasted. (3) Since the buffer cache 8 is collectively managed, the operation of flushing the data in the buffer cache 8 to the disk memory and the operation of not using the buffer cache 8 cannot be controlled for each file. Therefore, it is not possible to prioritize important files and less important files. (4) As the total buffer cache size increases,
The load and flush time of the batch file operation that is activated periodically increases. (5) When the buffer cache 8 is exhausted, it enters a waiting state for an unspecified time.

The present invention has been made to solve the above problems. A first object of the present invention is to improve the utilization efficiency of the buffer cache with the most frequent buffer cache size, and a second object. Is to control the necessity of using the buffer cache, and the third purpose is to reduce data loss when the system is down.
A fourth purpose is to perform a flush operation with a good timing, and a fifth purpose is to reduce the unspecified waiting time when the buffer cache is exhausted.

[0017]

According to a first aspect of the buffer cache management method of the present invention, an access request size is stored in a buffer cache management table, and the access request size is stored as a buffer cache reuse criterion. It is what

According to the second aspect of the buffer cache management method of the present invention, in response to an access request to the small size file, the data in the buffer cache of the small size file is stored in the free area of the buffer cache management table. It is something to save.

According to a third aspect of the buffer cache management method of the present invention, a large-capacity access request is accessed by bypassing the buffer cache.

In the buffer cache management method according to the fourth aspect of the present invention, whether or not to use the buffer cache for each file can be set.

In the buffer cache management method according to the fifth aspect of the present invention, the flush operation sequence is performed for each write operation.

According to a sixth aspect of the buffer cache management method of the present invention, a batch flash operation by the operating system or a flash operation designated by the user is selected.

The buffer cache management method according to the seventh aspect of the present invention is to stop the use of the buffer cache when the buffer cache is exhausted.

[0024]

According to the buffer cache management method of the first invention,
In the reuse of the buffer cache, the data having the most frequent data size remains on the buffer cache, and the efficiency of using the buffer cache at the most frequent buffer cache size is improved.

According to the buffer cache management method of the second invention, the buffer cache can be prevented from being fragmented by a small amount of data, and the utilization efficiency of the buffer cache with the most frequent buffer cache size can be improved.

The buffer cache management method of the third invention prevents the buffer cache with the highest request frequency from being replaced by an access request with a large capacity size, and improves the utilization efficiency of the buffer cache with the buffer cache size with the highest frequency. To do.

According to the buffer cache management method of the fourth invention, it is possible to control the necessity of using the buffer cache.

According to the buffer cache management method of the fifth aspect of the present invention, the influence of data loss is reduced when the system goes down due to power interruption during the flash operation.

In the buffer cache management method of the sixth aspect of the present invention, the user can control the timing of the flash operation, the load on the batch flash operation is reduced, and the user can explicitly issue the flash instruction when the write operation is completed. Therefore, a flash operation with good timing can be provided.

The buffer cache management method of the seventh invention eliminates the unspecified waiting time when the buffer cache is exhausted.

[0031]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, each embodiment of the present invention will be described with reference to FIGS.
In the following description, the same parts as those in the above-mentioned conventional example are designated by the same reference numerals. Example 1 (corresponding to claim 1). 1 is a block diagram showing a computer used in the buffer cache management method of the first embodiment, FIG. 2 is a block diagram showing a table used in the first embodiment, and FIG. 3 is a flowchart showing an initial operation of the first embodiment.
FIG. 4 is a flowchart showing the reading operation of the first embodiment,
FIG. 4 is a flowchart showing the write operation of the first embodiment.

In FIG. 1, reference numeral 1 denotes a random memory device having a random access memory, in which tables described below are arranged. Reference numeral 2 is a non-volatile memory device having a disk memory such as a hard disk, which manages data by block number as a block of a fixed size, that is, a block for each disk block 9. Reference numeral 3 is a buffer cache use definition table for the user to define the use standard of the buffer cache 8. 4 is a file definition table, which is at the start of use of the file (at the time of opening)
The buffer cache 8 is controlled for each file. A request header 6 controls the buffer cache 8 depending on the access size. Reference numeral 5 denotes a request header management table for managing the request header 6, which holds pointers (large size, small size, free list) for composing three types of bidirectional lists. Reference numeral 7 is a buffer header. Relationship between the buffer cache 8 and the disk block 9, the free list 100 and the hash list 110
The usage of is the same as the conventional method.

Initialization of various tables at the time of rising in the first embodiment will be described with reference to FIGS. 2 and 3. User uses buffer cache definition table 3
Then, the upper limit value and the lower limit value of the access size and the number of request headers 6 as the reference for using the buffer cache 8 are set (step 300). As the request header extension size in the buffer cache use definition table 3, the size of the data storage area prepared in the request header 6 is entered based on the lower limit value of the access size. (Step 30
1).

Then, as many request headers 6 as the number of request headers 6 in the buffer cache usage definition table 3 are prepared (step 302). At this time, the size of the data storage area in the request header 6 is prepared according to the size set in the request header extension size in the buffer cache use definition table 3, but the size of the data storage area is the minimum required for reading and writing by the device. Round up to an integer multiple of the size (disk block size).

Subsequently, the request header 6 prepared on the random access memory of the random memory device 2 is connected to the free list pointer in the request header management table 5 (see FIG. 1) (step 303). The request headers 6 are connected to each other by using the pointer on the request header connection list in the request header 6. The unused buffer cache is connected via the request header 6. That is, the request size of the first request header 6 connected to the free list next pointer in the request header management table 5 is set to the total buffer cache size, and all the empty buffer headers 7 existing in the random access memory are requested headers. Connect all forward pointers to buffer header 7 in 6 (step 304). By the above operation, in order to obtain an unused buffer cache when an access request is generated, the computer has a buffer cache 8 connected to the request header 6 connected to the head of the free list next pointer on the request header management table 5. Will be used. The case where the request header 6 has no space will be described later. The request header 6 is acquired from the pointer before the free list of the request header management table 5.

Next, the read operation and the write operation will be described with reference to FIG. When the reading operation starts, one request header 6 is acquired (step 401). The empty request header is obtained from the pointer before the free list of the request header 6 of the request header management table 5. Then, the request size and the in-use flag to the status flag are set as a necessary description in the request header 6 (step 402). Then, the request header 6 is connected (step 403).
That is, the connection to the request header management table 5 is made based on the buffer cache size prepared in the random access memory. If the request size is larger than the buffer cache size, the size increases in order from the large size next pointer in the request header management table 5 toward the previous pointer, that is, the request size increases as it approaches the previous pointer. The request header 6 is connected to. If the requested size is smaller than the buffer cache size, it is connected to the small size pointer.
In this case, the request header 6 is connected so that the size decreases from the small size pointer next pointer to the previous pointer. Then, the necessary buffer cache 8 is prepared. That is, the size of the request header 6 is checked to see if the buffer cache 8 is present in the first request header 6 connected to the free list next pointer of the request header management table 5 (step 404). The status flag of the request header 6 is set to "in use" (step 405), the required number is subtracted from the request size by the required number of blocks (step 406), and the required number is separated (step 407). Until the end, when used, the request size of the request header 6 is set to "0".

In step 404, if the buffer cache 8 does not exist or the required number is insufficient,
According to the access request size (step 450), if the access request size of the request header management table 5 is larger than the buffer cache size prepared in the random access memory of the random memory device 2, the large size pointer previous pointer, that is, the buffer held The larger number of headers 7 is searched (step 451). If the access request size is smaller than the buffer cache size prepared in the random access memory, the small size pointer previous pointer is searched (step 452). At this time, the status header in the request header 6 is searched for one that is in use or has not been modified.

As a result, the in-use flag is set in the status flag of the corresponding request header 6 (step 453),
The required buffer cache 8 is subtracted from the size in the request header 6 (step 454), and the buffer cache 8 is separated from the request header 6 (step 45).
5) If the number of buffer headers 7 connected to the corresponding request header 6 is surplus, as in the case of the above initialization, the buffer header 7 is added to the head of the free list next pointer of the request header management table 5 and connected (step). 457). If there is not enough, the next request header 6 is retrieved from step 451 or step 452 to secure the buffer header 7 of the required size.

The secured buffer header 7 is the request header 6
Connect to the forward pointer to the buffer header 7 inside (step 408). To connect the buffer headers 7 to each other, a pointer to the next buffer header of the buffer header 7 in FIG. 2 is used. Upon completion of reading from the disk memory (step 409), the in-use flag of the request header 6 is reset.

Next, the write operation will be described. To obtain the request header 6 and free buffer cache required for access, the above steps 401 to 408 and step 4 are performed.
This is the same as the read operation shown in 50 to 457. Therefore, after the read operation to the buffer cache 8 in step 409 is completed, the buffer cache 8 is modified, and the modified flag is set in the status flag in the request header 6 (step 410). The modified flab is turned off (step 411) and the in-use flab is reset (step 412).

In short, according to the first embodiment, the data of the most frequent data size is preferentially left on the buffer cache 8 due to the upper limit value and the lower limit value of the buffer cache use definition table 3, so that the random memory device is used. In the reuse of the buffer cache 8 in 2, the utilization efficiency of the buffer cache 8 with the most frequent buffer cache size can be improved.

In the first embodiment, an example in which the buffer header 7 and the request header 6 are separated is shown and described, but the same effect can be obtained even if both tables are configured as the same table.

Embodiment 2 (corresponding to claim 2). The second embodiment is characterized in that the buffer cache 8 is not used for a file access request having a size smaller than the buffer cache size, but the free area of the buffer cache management table in which the access request size is stored is used. Therefore, the components of the computer used in the buffer cache management method of the second embodiment will be described using the reference numerals of FIG.

That is, FIG. 5 is a flowchart showing the buffer cache management method of the second embodiment. As shown in FIG. 5, in the second embodiment, the access request file size of the user is compared with the lower limit value in the buffer cache usage definition table 3 (step 501), and if it is smaller, the data stored in the request header 6 is stored. The area is treated as the buffer cache 8 and the data is stored in the data storage area. As a result, the buffer cache 8 is not subdivided by a small amount of data. If it exceeds, that is, the access request between the upper limit value and the lower limit value (step 501) follows the flowchart shown in FIG. In the buffer cache management method according to the second embodiment, when the access request file size is smaller than the lower limit value in the buffer cache usage definition table 3, after storing the data in the data storage area of the request header 6, the request header 6 is secured. (Step 502). That is, one request header 6 is acquired from the pointer before the free list in the request header management table 5, the access size is entered in the request size of the request header 6, the busy flag is written in the status flag, and the forward header to the buffer header 7 is written. NULL is entered in the pointer (step 503).

After that, the request header 6 is connected (step 504). This connection operation includes a connection operation to the request header management table 5 and a connection operation to the hash queue related to the disk memory device. The connection to the request header management table 5 is made based on the buffer cache size prepared in the random access memory of the random memory device 2. That is, it is connected to the small pointer. In this case, the small size pointer is connected so that the size decreases from the next pointer to the previous pointer. On the other hand, the connection to the hash queue does not use the buffer header 7, but the pointer to the hash queue in the request header 6 is connected. That is, the buffer cache access of the corresponding device block in the buffer cache access can be searched from the queue header 111 (see FIG. 11).

Subsequently, the extended area of the request header 6 is read (step 505), it is judged whether the access is a read or a write (step 506), and if the access is a write, the status flag in the request header 6 is corrected after writing the data. Is set (step 50
7) Write scheduling is performed (step 50).
8). Then, upon completion of reading and writing, the busy flag in the request header 6 is turned off (step 509).

Embodiment 3 (corresponding to claim 3). The third embodiment is characterized in that input / output is performed without using the buffer cache 8 for an access request of a large size in which replacement of the buffer cache 8 occurs. The components of the computer used in the buffer cache management method will be described using the reference numerals in FIG.

That is, FIG. 6 is a flowchart showing the buffer cache management method of the third embodiment. As shown in FIG. 6, in the third embodiment, it is determined whether the user's access request file size exceeds the upper limit value in the buffer cache usage definition table 3 (step 601). If it exceeds, the buffer cache 8 is not used and the user buffer is directly accessed. When it does not exceed, that is, the access request between the upper limit value and the lower limit value follows the flowchart shown in FIG. 4 (step 601). In the buffer cache management method of the third embodiment, the request header 6 is secured when the access request file size exceeds the upper limit value in the buffer cache usage definition table 3 (step 602). That is, one request header 6 is acquired from the pointer before the free list of the request header management table 5, and the busy flag is written in the status flag to this request header 6 (step 604).

After that, the request header 6 is connected (step 604). This connection connects to the small pointer of the request header management table 5. This means that the requested size is 4
If it is 500 bytes and the minimum access size of the corresponding disk memory device is 1 Kbyte, it is possible to read 4 Kbytes directly into the user area, but after reading the last 404 bytes into some buffer, This is because it is necessary to copy to the user buffer. However, if the requested access size is an integral multiple of the access size of the disk memory device, acquisition, connection and return of the request buffer are unnecessary.

Subsequently, it is judged whether the access is a read or a write (step 605). If it is a read, an integral multiple of the device access size of the request access size is directly read into the user area (step 606), and the remaining part is the request header. After the data is read into the extension area 6 (step 607), the required number of bytes is copied to the user area (step 608). On the contrary, if the access is writing, an integral multiple of the device access size of the request access size is directly written to the user area or the disk memory (step 611), and the remaining portion is from the disk area corresponding to the extended area of the request header After being read once (step 612), correction is performed on the extended area and then writing to the disk memory is performed (step 613). Then, the request header 6 used for writing and reading turns off the in-use flag (step 613) and returns it to the free list of the request header management table 5 (step 610). As a result, this Example 2
According to the above, the most frequently used buffer cache 8 is not replaced by the large-capacity access request size.

Embodiment 4 (corresponding to claim 4). The fourth embodiment is characterized in that whether or not to use the buffer cache is selected according to the usage pattern of the file as shown in FIG. 7, so that the computer used in the buffer cache management method of the fourth embodiment is The components will be described using the reference numerals in FIG.

That is, FIG. 7 is a flowchart showing the buffer cache management method of the fourth embodiment. As shown in FIG. 7, in the fourth embodiment, when the user starts using the file (open operation), the user selects and specifies whether or not the subsequent access to the file is performed using the buffer cache 8. In subsequent read and write accesses, the buffer cache availability flag of the file definition table 4 is checked (step 700). When the buffer cache availability flag is set, the access via the buffer cache 8 is not performed. That is, step 60 in the third embodiment described above.
This is the same as not using the buffer cache 8 at the time of a large size access request shown in 2 and after. When the buffer cache availability flag is off, the flow chart shown in FIG. 4 is followed (step 701).
As a result, according to the fourth embodiment, the necessity of using the buffer cache can be controlled.

Embodiment 5 (corresponding to claim 5). The fifth embodiment is characterized in that, as shown in FIG. 8, a flash operation that reflects the update content to the disk memory of the non-volatile memory device 1 as the secondary recording medium of the buffer cache 8 is performed for each write operation. Therefore, the components of the computer used in the buffer cache management method of the fifth embodiment will be described using the reference numerals of FIG.

That is, FIG. 8 is a flowchart showing the buffer cache management method of the fifth embodiment. As shown in FIG. 8, in the fifth embodiment, the flush operation of the buffer cache 8 retrieves the request header 6 from the large-sized next pointer and the small-sized next pointer in the request header management table 5 to find the connected request header 6. (Step 80
0), it is determined whether the status flag in the request header 6 is set with the modified flag (step 801). If the modified flag is not set, the process is not performed (step 802). If the modified flag is set, the flush operation is performed to forward the buffer header 7 into the request header 6. The buffer caches 8 connected to the pointers are collectively written in the disk memory (step 803), and at the end of writing, the correction flag is turned off from the status flag of the corresponding request header 6 (step 804). As a result, according to the fifth embodiment, it is possible to locally reduce the influence of data loss when the system is down due to power failure during the flash operation.

Embodiment 6 (corresponding to claim 6). In the sixth embodiment, as shown in FIG. 9, the operator specifies a flash operation that reflects the update contents of the buffer cache 8 to the disk memory of the non-volatile memory device 1 automatically for each file or an operation flash. 1 is used for the buffer cache management method of the sixth embodiment, and the description will be given using the reference numerals of FIG.

That is, FIG. 9 is a flowchart showing the buffer cache management method of the sixth embodiment. As shown in FIG. 9, in the sixth embodiment, when the user starts the use of the file (open operation), the user selects and specifies the modified flush operation of the buffer cache 8 caused by the subsequent write operation to the file. To do. If the flash is performed in a batch, the sync flag in the file definition table 4 is reset to "0", and if not necessary, it is set to "1". That is, at the time of file access, the sync flag of the file definition table 4 is referenced (step 900), and
If the sync flag is set to "1", the sync header flag in the request header 6 is set to "0" after the request header 6 is secured by the write operation of the first embodiment (step 902). As a result, according to the sixth embodiment, the user can control the timing of the flash operation,
The load on the batch flash operation in the computer is reduced, and the user can instruct the flash operation when the write operation is completed.

As a result of the operation of the sixth embodiment, the flag check set in the request header 6 in the buffer cache flush operation according to the fifth embodiment is checked.
That is, when the flag is set to "1", the buffer cache 8 managed by the request header 6 is not collectively flushed. In this case, for example, a method in which the user explicitly indicates the flash timing and the file to the operating system can be considered.

Embodiment 7 (corresponding to claim 7). Since the seventh embodiment is characterized in that the use of the buffer cache 8 is stopped when the buffer cache 8 is exhausted as shown in FIG. 10, the configuration of the computer used in the buffer cache management method of the seventh embodiment is as follows. The elements will be described using the reference numerals of FIG.

That is, FIG. 10 is a flowchart showing the buffer cache management method of the seventh embodiment. As shown in FIG. 10, in the seventh embodiment, the exhaustion of the free buffer cache 8 is the first request header of the request header free list next pointer connected to the request header management table 5 as described in the first embodiment. This occurs when there is no buffer header 7 connected to 6 or when the busy flags of all request headers 6 connected to the large size pointer and the small size pointer are set. The processing at that time is to perform the processing from step 1002 onward, and the processing is the same as the processing when the buffer cache 8 of the sixth embodiment is not used. If the free buffer cache 8 is found, the same processing as in the first embodiment is performed (step 1001). As a result, according to the seventh embodiment, the unspecified waiting time when the buffer cache 8 is exhausted can be eliminated.

[0060]

According to the first aspect of the present invention, the access request size is stored in the buffer cache management table, and this access request size is used as a reference when reusing the buffer cache. In the reuse, the data with the most frequent data size remains preferentially in the buffer cache, and the effect of using the buffer cache with the most frequent buffer cache size is improved.

According to the second invention, the data in the buffer cache of the small size file is stored in the free area of the buffer cache management table in response to the access request to the small size file. There is an effect that the cache can be prevented from being fragmented by a small amount of data, and the use efficiency of the buffer cache with the most frequent buffer cache size can be improved.

According to the third aspect of the present invention, since a large-capacity access request is accessed by bypassing the buffer cache, the buffer cache with the highest request frequency in the computer is inserted by the large-capacity access request. It is possible to prevent the replacement and to improve the utilization efficiency of the buffer cache with the most frequent buffer cache size.

According to the fourth aspect of the present invention, whether or not to use the buffer cache is selected for each file. Therefore, the necessity of using the buffer cache can be controlled. For example, for a highly reliable file, the buffer cache can be used. This has the effect of enabling fine control such as not using the cache.

According to the fifth aspect of the invention, since the flush operation of the buffer cache is performed every write operation, the influence of data loss is localized at the time of system down due to power interruption during the flush operation. There is an effect.

According to the sixth aspect of the invention, since the user specifies the batch flash operation or the file timing, the user can control the flash operation timing, and the load on the batch flash operation can be reduced. Moreover, there is an effect that a flash operation with good timing can be provided.

According to the seventh invention, since the use of the buffer cache is stopped when the buffer cache is exhausted, it is possible to eliminate the unspecified waiting time when the buffer cache is exhausted.

[Brief description of drawings]

FIG. 1 is a configuration diagram illustrating a computer used in a first embodiment.

FIG. 2 is a configuration diagram showing a table according to the first embodiment.

FIG. 3 is a flowchart showing an initial operation of the first embodiment.

FIG. 4 is a flowchart showing an access operation according to the first embodiment.

FIG. 5 is a flowchart showing a second embodiment.

FIG. 6 is a flowchart showing a third embodiment.

FIG. 7 is a flowchart showing a fourth embodiment.

FIG. 8 is a flowchart showing a fifth embodiment.

FIG. 9 is a flowchart showing a sixth embodiment.

FIG. 10 is a flowchart showing an example 7.

FIG. 11 is a configuration diagram showing a conventional example.

[Explanation of symbols]

1 random memory device, 2 non-volatile memory device, 3
Buffer cache usage definition table, 4 file definition table, 5 request header management table, 6 request header, 7 buffer header, 8 buffer cache, 9 disk blocks, 100 free list, 10
1 Free list header, 102 Free list forward pointer, 103 Free list backward pointer, 110 Queue list, 111 Queue header.

Claims (7)

[Claims] 1. A buffer cache management method for managing data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system, wherein an access request size is stored in a buffer cache management table. A method for managing a buffer cache of a computer characterized by the above. 2. A buffer cache management method for managing data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system, wherein a file access request having a size smaller than the buffer cache size is requested. A buffer cache management method for a computer, which does not use the buffer cache, but uses a free area of a buffer cache management table in which the access request size is stored. 3. A buffer cache management method for managing data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system in a large size in which replacement of the buffer cache occurs. A buffer cache management method for a computer, which performs input / output without using a buffer cache for an access request. 4. A buffer cache management method for managing data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system. A method for managing a buffer cache of a computer, characterized by selecting whether or not. 5. A buffer cache management method for managing a data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system, the update contents of the buffer cache to a secondary recording medium. A buffer cache management method for a computer, characterized in that a flash operation that reflects is performed for each write operation. 6. A buffer cache management method for managing data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system, wherein a batch flash operation by the operating system or a flash specified by a user is performed. A method for managing a buffer cache of a computer, which is characterized by selecting whether to operate. 7. A buffer cache management method for managing data transfer between a non-volatile memory of a computer and a buffer cache prepared on a random access memory by an operating system, wherein use of the buffer cache is stopped when the buffer cache is exhausted. A method for managing a buffer cache of a computer characterized by: