JPH02156348A - Overflow area control method for plural hash tables - Google Patents

Abstract

PURPOSE:To secure an enough overflow area by providing plural hash table areas and a memory pool to the used in common as the overflow areas of the respective plural hash areas in a memory. CONSTITUTION:In the memory, plural hash table areas 1, 2 and 3 and a memory pool 5, which is used in common as the overflow area of the respective plural hash tables, are provided. When a symbol is registered by a processing means 6 and if the overflow area is needed, the overflow area is secured from the memory pool 5 and registered. When the overflow area is deleted, it is deleted and when the overflow area becomes empty, the area is released. Thus, since a memory area is efficiently used, the comparatively enough overflow area can be secured.

Description

[Detailed Description of the Invention] [Summary] Regarding the overflow area management method for multiple hash tables, a relatively sufficient overflow area is secured when multiple hash tables are used in a computer with a relatively small memory configuration. In order to provide a method for managing the overflow area of multiple hash tables, a memory pool is secured in memory that is commonly used as an overflow area for multiple hash tables, and a processing means is provided. When an overflow area is needed during symbol registration, an overflow area is secured from the memory pool and registered, and when the symbol is to be deleted, it is deleted, and when the overflow area becomes empty, it is released.

[Industrial application field]

For example, the present invention is a reservation system that uses a computer with a relatively small memory configuration to make reservations for multiple movie theaters, and when registering or deleting a reservation person for each movie theater, multiple hash tables are used. This invention relates to a method for managing overflow areas in multiple hash tables, such as when registering and deleting symbols corresponding to reservation users using .

A hash table is a hash table in which when a symbol is registered in a table (hash table), n=1.2, 3. Apply a hash function so that it becomes an integer, find n=f(X), and calculate
1.2, 3 in the f(x) column of the hash table shown in A).
It is to be registered in the row with the corresponding number.

In this way, when searching for the symbol X, by calculating with the hash function, the value of the row of the hash table can be found, making the search easy.

In general, no matter how you choose the hash function, the symbols x, y
, the value is the same as f (x) = f (y), and there may be a conflict even if you try to register it. In such a case,
Since X has already been registered in the hash table, y is registered in the overflow area and linked with a pointer to facilitate the search for y.

Also, for the symbols x, y, and z, f (x) = f (y) =
If f (z), y and z are registered in the overflow area, and y and 2 are further linked with a pointer to facilitate the search for 2.

In recent years, systems using microprocessors may not have sufficient memory due to other constraints, so it is possible to use multiple hash tables to represent symbols on computers with relatively small memory configurations. When registering or deleting, it is necessary to use memory efficiently.

[Conventional technology]

Conventionally, when a plurality of hash tables are used, a sufficient overflow area deemed necessary for each hash table is secured in advance in memory.

In this case, an imbalance may occur in the usage of the overflow area, and the memory area allocated to the overflow area with a small usage amount is wasted.

[Problem to be solved by the invention]

The conventional method includes a wasted memory area and requires a sufficiently large memory area, and there is a problem that a computer with a relatively small memory configuration cannot secure a sufficient overflow area.

SUMMARY OF THE INVENTION An object of the present invention is to provide a method for managing overflow areas of a plurality of hash tables, which can secure a relatively sufficient overflow area when a plurality of hash tables are used in a computer with a relatively small memory configuration.

[Means to solve the problem]

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

As shown in Figure 1, there are multiple hash table areas 1 in memory.
．． 2.3. ...and secure a memory pool 5 that is commonly used as an overflow area for each of the multiple hash tables,
In the processing means 6, when an overflow area is required during symbol registration, an overflow area is secured from the memory pool 5 and registered, when the symbol is to be deleted, it is deleted, and when the overflow area becomes empty, it is released.

[For production]

According to the present invention, a memory pool 5 that is commonly used as an overflow area for each hashish table, which is smaller than the total overflow area in the conventional case where an overflow area is secured for each hashish table in advance, is used as a memory pool. When an overflow area is needed during symbol registration, the processing means 6 secures an overflow area from the memory pool 5 and registers it, deletes it when deleting it, and releases it when the overflow area becomes empty. In this way, the memory area is used efficiently, so a relatively sufficient overflow area can be secured.

[Embodiment] FIG. 2 is a diagram showing memory allocation in an embodiment of the present invention, and FIG.
The figure is a flowchart of the registration program according to the embodiment of the present invention, FIG. A diagram showing the configuration of a computer to which the present invention is applied, and FIG. 7 is a hash table of an embodiment of the present invention.

FIG. 8 is a diagram showing a memory block and unused list, and FIG. 8 is a diagram showing a memory pool management table according to an embodiment of the present invention.

The configuration of a computer with a small memory that uses multiple hash tables is shown in Figure 6. For example, reservations can be registered and deleted for each movie theater A, B, and C. When registering or deleting a reservation person's symbol using multiple hash tables, for example, if a symbol for movie theater A is input, the microprocessor IO selects the hash table to be applied to the symbol for movie theater A. Functional calculation program in ROM
12, perform the calculation, and store the calculation result in the memory 11.
Register or delete from the hash list for movie theater A.

The allocation of hash tables and memory pools on this memory 1 is as shown in Figure 2, and multiple hash tables 1, 2, 3
．． ...and a memory pool 5 that is commonly used as an overflow area for each hash table is allocated.

The size of the memory pool 5 in this case is smaller than the total of the conventional overflow areas, and when an overflow area is required, the memory blocks to be allocated as the overflow area are:
It is smaller than the conventional overflow area, and the memory boolean 5 includes a large number of memory blocks.

The hash table is shown in Figure 7 (A), with a column for the value f (x) obtained by the hash function, a column for registering the symbol (x), a column for indicating information about the symbol, and an overflow area. It consists of a series of overflow pointers that write addresses of overflow areas for linking.

The memory block is shown in Figure 7 (B) and consists of a column for registering the symbol (x), a column for indicating information about the symbol, and a column for overflow pointers for writing the address of the row of the memory block. , each row is called a cell.

Note that the memory block may have a structure of 1 cell so that it can be released immediately upon deletion, but since this increases the frequency of memory block acquisition, it is usually configured to have about 10% of the cells in the hash table.

Therefore, for example, when a memory block is acquired for one hash table as shown in FIG. 7, the symbol A to be registered is
If the value of the hash function of A is 2, it will be registered in the 2nd row of the column of symbol (x), and if the value of the hash function of the symbol BB to be registered is 4, it will be registered in the 4th row of the column of symbol (x). If the value of the hash function of the symbol CC to be registered is 5, it is registered in the 5th row of the symbol (x) column.

Next, when the value of the hash function of the symbol EE to be registered becomes 2, the symbol AA is registered in the second row of the symbol (X) column, and there is a collision, so at this time, the memory pool 5 overflows. Secure memory block No. 1 as an area, register it in the first row of the symbol (x) column of memory block No. 1 shown in FIG. The address of the first line of block N011 is written and linked.

Next, when the value of the hash function of the symbol DD to be registered becomes 5, the symbol CC is registered in the 5th row of the symbol (X) column, and there is a collision, so the memory shown in FIG. 7 (B) Register it in the second row of the symbol (X) column of block No. 1, and set the overflow pointer in the fifth row of the hash table to memory block No. 61.
Write the address on the second line to link.

Next, when the value of the hash function of the symbol FF to be registered becomes 2, the symbol AA is registered in the second row of the symbol (X) column, and the memory block is registered in the second row of the overflow pointer column. Since the address of the first line of N001- is written,
Looking at this first line, the symbol EE is also written here, so it is registered in the third row of the symbol (x) column of memory block N001, and the overflow pointer in the first row is set to the memory block No. 1. Write the address on the third line to link.

When a memory block is acquired, a list of unused cells as shown in FIG. 7(C) is created for each memory block in the memory, and the pointer to the unused list is set to the first row of the unused list. Write the address of the cell, and write the address of the next cell to the overflow pointer of each cell in the unused list, linking them one after another, for example, placing a symbol in the cell in the first row of the memory block. Then, the cell in the first row of the unused list is removed, and the address of the cell in the second row is written in the pointer to the unused list.

Further, a memory pool management table as shown in FIG. 8 is provided on the memory, and an area is provided for writing the NOo of the memory block, the start address of the memory block, and the number of cells in the memory block.

The ROM 12 shown in FIG. 6 stores the registration program, deletion program, and organization program shown in FIGS. 3 to 5 as processing programs, and the microprocessor 10 reads these programs and processes them.

Next, symbol registration will be explained according to the registration program shown in FIG.

In step 1, calculate the hash function of the symbol, in step 2 check whether the symbol is registered in the calculation result row of the hash table, and if it is not registered, register it in the hash table in step 3. .

If it is registered, it is checked in step 4 whether the address has been written to the pointer to the unused list shown in FIG. In step 6, each cell of the memory block is linked with a pointer to the unused list and the list of unused cells, as shown in FIG. 7(C), and in step 7, as shown in FIG. The address of the acquired memory block and the number of cells are registered in the memory pool management table, and the process proceeds to step 8.

In step 8, an unused cell is acquired by the address of the pointer to the unused list in FIG. 7(C), and in step 9
Then, the content is registered in an unused cell of the acquired memory block, and the address of the registered cell is written and linked to the upper hash table or cell overflow pointer.

Next, symbol deletion will be explained according to the deletion program shown in FIG.

In step 1, the hash function of the symbol to be deleted is calculated, and the row value of the hash table is determined.

(2) If the symbol to be deleted is in the corresponding row of the hash table and no cell address is written in the overflow pointer of the row, the contents are deleted in step 2.

■If the symbol to be deleted is in the corresponding row of the hash table and the address of the cell is written in the overflow pointer of the row, in step 3, the contents of the cell first linked with the overflow pointer are Overwrite and delete, step 4
Then, in the memory pool management table of FIG. 8, the number of used cells in the memory block number of the deleted cell is decreased by 1, and the process proceeds to step 5.

In step 5, the remaining number of cells used is approximately 30% of the number used.
If it is below this level, the process ends, and if it is below this level, the memory block is organized using the organizing program described below.

■If the symbol is in the memory block, check whether the address of the next cell has been written to the overflow pointer of the memory block in step 7, and if it has been written, proceed to step 1 8 and start the next link. The contents of the cell are deleted by overwriting the previous cell, and the process proceeds to step 10.

If the next cell has not been written in step 7, the contents of the cell are deleted in step 9 and the process proceeds to step 10.

In step 10, the number of used memory blocks of the deleted cell in the memory pool management table is decreased by 1, and the process proceeds to step 11.

In step 11, the remaining number of cells used is approximately 3 of the number used.
If it is not less than about 0%, the process ends, and if it is less than 0%, the memory block is sorted by the sorting program described below.

Next, the arrangement of memory blocks will be explained according to the memory block arrangement program shown in FIG.

In step 1, overflow pointers in a plurality of hash tables are searched and the process proceeds to step 2.

In step 2, if the linked cell is a sorted memory block, acquire an unused cell from the unused list of multiple memory blocks, copy the contents to it, and Set the address of the cell overflow area to the address of the copied unused list cell,
In step 3, the number of uses of the corresponding memory blotter of the organized cell in the memory pool management table is set to 1:$i.The operation of proceeding to step 4 is repeated until the end of a plurality of hash tables, and the process proceeds to step 5.

If the used number becomes 0 in step 5, the organized memory block is released in step 6 so that it can be used again.

If the number of uses does not reach O in step 5, step 7
Step 8 searches the unused list of multiple memory blocks for block No. 8, and searches for block No.
If O9 is a sorted memory block, use the overflow pointer of the previous cell as the address of the next cell, reduce the number of used blocks in the corresponding block No. 1 in the memory pool management table by 1 in step 9, and proceed to step 10. Steps from step 7 The operations up to IO are repeated until the end of the unused list.

Then the organized memory block will be empty, so step I
I releases the organized memory block so that it can be used again.

In this way, a memory block as an overflow area of the memory pool is dynamically allocated to improve memory usage efficiency, so a relatively sufficient overflow area can be secured.

Note that if you use a computer O3 based on the lTR0N specifications, a task can acquire a memory block by using the system call get-blk, and release it by using tel-blk. Blocks can be acquired and released easily.

〔Effect of the invention〕

As explained in detail above, according to the present invention, in order to allocate an overflow area, a memory pool that is commonly used as an overflow area by a plurality of hash tables is provided, and a memory block is dynamically allocated as the overflow area of the memory pool. Since memory usage efficiency is improved, a relatively sufficient overflow area can be secured when symbols are registered or deleted using multiple hash tables on a computer with a relatively small memory configuration.

[Brief explanation of the drawing]

Figure 1 is a block diagram of the principle of the present invention, Figure 2 is a diagram showing memory allocation in an embodiment of the present invention, and Figure 3 is a diagram showing memory allocation in an embodiment of the present invention.
Figure 4 is a flowchart of a registration program according to an embodiment of the present invention, Figure 4 is a flowchart of a deletion program according to an embodiment of this invention, Figure 5 is a flowchart of a sorting program according to an embodiment of this invention, and Figure 6 is an example of an example. FIG. 7 is a diagram showing the configuration of a computer to which the present invention is applied. FIG. 7 is a diagram showing the Hunsch table, memory block, and unused list according to the embodiment of the present invention. FIG. 8 is a memory pool management table according to the embodiment of the present invention. FIG. In the figure, 1.2.3 is a hash table, 5 is a memory pool, 6 is a processing means, 10 is a microprocessor, 11 is a memory, and 12 is a ROM. I3 indicates a symbol input section. The principle of destruction is the composition of the coscipiputer that will be published. Figure 1.Memorial phone number is 1 table.Figure 1.

Claims (1)

[Claims] When registering and deleting symbols using a plurality of hash tables in a computer with a relatively small memory configuration, a plurality of hash table areas (1, 2, 3, etc.) are stored in the memory. )
A memory pool (5) commonly used as an overflow area for each of the plurality of hash tables is secured, and a processing means (6) is secured.
, when an overflow area is needed when registering a symbol, the overflow area is secured from the memory pool (5) and registered, the overflow area is deleted when the symbol is deleted, and the overflow area is released when the overflow area becomes empty. A method for managing overflow space for multiple hash tables.