JPH03148733A - Shared memory area allocating and processing system - Google Patents

Abstract

PURPOSE:To decrease the allocation processing overhead of a shared memory area by possessing an intrinsic area in a required size on a shared memory and allocating the shared memory area from the inside of the intrinsic area to an idle area. CONSTITUTION:When the execution of a program is started by an arithmetic unit 11, an area possession part 14 possesses the prescribed area from the idle memory area of a shared memory 13 and defines the possessed area as an intrinsic area 17 according to a method for accessing an exclusive control flag 5 and idle area information 6 on the shared memory 13. An allocation control part 10 holds the leading address of the intrinsic area 17 possessed by the area possession part 14 and allocation control information 18 in a main memory 12 and when the control part is called by an area allocation request, the allocation control information 18 are referred and the area in the required size is segmented from the intrinsic area 17. Thus, since the area allocation of the shared memory 13 can be executed only by accessing the main memory 12 in the program, the overhead of the area allocation can be decreased.

Description

[Detailed Description of the Invention] [Summary] Regarding the area allocation of the shared storage device of a multiprocessor system consisting of an arithmetic unit having a shared storage device and a separate main storage device, the present invention relates to area allocation of the shared storage device. , aims at a shared storage area allocation processing method that can reduce shared storage area allocation processing overhead, and has a plurality of arithmetic units and main storage devices, and a shared storage device, and each of the arithmetic units has a separate main memory. and the common shared storage device, and executes an area acquisition unit and an allocation management unit, and the area acquisition unit acquires the shared storage device at a predetermined time when the execution of a program starts in the processing device. The allocation management unit acquires a predetermined area from the free storage area of the storage area and makes it a unique area, the allocation management unit holds allocation management information of the unique area in the main storage, and the program executed by the arithmetic unit stores it in the shared storage. The area requested above is allocated from the specific area according to the allocation management information.

[Industrial application field]

The present invention relates to a processing method for allocating a shared storage area to a shared storage device in a multiprocessor system consisting of an arithmetic unit having a shared storage device and a separate main storage device.

[Conventional technology]

FIG. 2 is a block diagram showing an example of the configuration of a multiprocessor system.

The plurality of processing devices 1 are connected to individual main storage devices 2, and each processing device W1 executes a program separately loaded into the main storage device 2 and stored in the main storage device 2 or the shared storage device 3. access data and perform necessary system management and application processing.

The shared storage device 3 can be accessed by all the computing devices 2, and contains areas for commonly used management data and work data for business processing.

With this system configuration, for example, a program is loaded into the main storage device 2 of a plurality of required processing devices, and a task executed by each processing device 1 executes the program based on management information provided in the shared storage device 3. By dividing tasks, one program can be divided into appropriate configurations and executed in parallel by multiple tasks.

In such a case, the storage area required by the program executed on each computing device 1 is generally stored on the shared storage device 3 because the processing result needs to be referenced by the program executed on the other computing device 1. Therefore, when there is a request for a storage area, the area acquisition unit 4 using the management program executed by the processing unit 1 acquires an area of the requested size from the free space on the shared storage device.

To acquire this area, for example, as shown in FIG. 3, first, in processing step 20, the exclusive control flag 5 provided for exclusive control on the shared storage device 3 is accessed, and a lock is attempted to be acquired using a known method. If the lock cannot be acquired, repeated attempts are made and waits (processing steps 21 and 22). After suppressing contention with other arithmetic units 1 by acquiring the lock, in processing step 23, the size of the free area, the area start address, etc. An area to be acquired from the empty area is determined by referring to appropriate free area information 6 indicating the area, and the area is secured by updating the free area information 6 so that the area is occupied, and then processing step 24 The exclusive control flag 5 is restored to release the lock, and the process ends.

[Problem to be solved by the invention]

The shared storage device 3 requires a relatively large capacity, for example 4 gigabytes, in order to be able to store a large amount of data, and is configured with relatively low-speed storage elements due to economical requirements, and is not capable of transferring blocks of large amounts of data. However, access to exclusive control flags, management information, etc. as mentioned above is relatively inefficient, and when the above-mentioned area acquisition processing occurs frequently, the overhead of the processing unit Wl increases. The problem is that it significantly increases

An object of the present invention is to provide a shared storage area allocation processing method that can reduce accesses to coexisting storage devices and reduce shared storage area allocation processing overhead.

[Means to solve the problem]

FIG. 1 is a block diagram showing the configuration of the present invention.

The figure shows the configuration of a shared storage area allocation processing method, which includes a plurality of arithmetic units 11, a main storage device 12, and a shared storage device t.
13, and each arithmetic unit 11 has an individual main storage device 1.
2 and a common shared storage device 13, and executes the area acquisition unit 14 and the allocation management unit 10, and the area acquisition unit 14
At a predetermined time when the execution of the program is started on the arithmetic unit 11, for example, a shared memory is written as in the past? ! ! By accessing the exclusive control flag 5 and free space information 6 on the device, a predetermined area is acquired from the free storage area of the shared storage device 13 and set as the unique area 17, and the allocation management unit 10 allocates the unique area. The management information 18 is held in the main storage device 12, and a program executed by the arithmetic unit 11 is stored in the shared storage device 1.
8 is allocated from the unique area 17 according to the allocation management information 17.

[For production]

By using the above processing method, for example, when a program starts executing, it acquires a unique area of the size required by the program on the shared storage device, and when the program starts executing, By allocating from within the area, the process of acquiring and releasing areas of the shared storage device during program execution can be processed in a relatively short time by only accessing the main storage device.

【Example〕

The area acquisition unit 14 and the allocation management unit 10 in FIG. 1 are each provided as a subroutine called from an application program, and a compiler that generates a program to be executed in this system, for example, at the beginning of a unit program executed as one task, A subroutine call instruction is inserted so that an area acquisition unit 14 is called to acquire a specific area of a size necessary for the program, and when the task is finished, the area acquisition unit 14 is called again to release the specific area.

When each continuation within the task acquires/releases its own local area, a subroutine call instruction is inserted to call the allocation management unit IO and cut out an area of the required size from the unique area 17.

When the area acquisition unit 14 is called from such a program, in order to acquire an area of the requested size from the free area of the shared storage device W13, it executes an exclusive control operation according to the process flow shown in FIG. 3, for example. Access the exclusive control flag 5 to try to acquire a lock, and if it is not possible to acquire a lock, repeat the trial and wait, acquire the lock, thereby suppressing contention with other processing units l, and then refer to the free space information 6. Then, the area to be acquired is determined, the free area information 6 is updated, and then the flag 5 is restored to release the lock, and the acquisition is completed.

Thereafter, allocation management information 18 is generated on the main storage device 12, the acquired area is set to -unallocated state, and the address of the acquired area is passed to the caller.

Also, when an area release request is made, the free area information row of the shared storage device 13 is accessed exclusively in the same way as when acquiring the area, and the free area information 6 is updated so that the acquired area becomes a free area. do.

The allocation management unit IO holds the start address of the unique area 17 acquired by the area acquisition unit 14 and allocation management information 18.

The allocation management information 18 may be known, for example, bitmap information for indicating the size and allocation status of the unique area 17. When called with an area allocation request, the allocation management unit 10 refers to the allocation management information 18. , extracts an area of the requested size from the unallocated area, updates the area to the allocated state, returns the start address of the extracted area to the request source, and in the case of a release request, moves from the specified start address to the specified one. Regarding the size area, the allocation management information 18 is updated to indicate unallocated shoes.

In the above example, the unique area 17 is taken for each task, but it is also possible to share the unique area among the tasks executed by the processing unit o, and take the shared unique area at the start of a job step, for example. Yes, in that case, the area acquisition unit 14 executes the acquisition process only when the unique area has not yet been acquired. Note that although the area acquisition unit 14 and the allocation management unit 1O are implemented as subroutines of the application program,
It may be executed as a system management service, or if it is performed by system management, for example, at the time of initial program loading, a unique area for each computing device 11 is taken, and the task to be executed by each computing device 11 is It can also be assigned.

As described above, when acquiring an area directly from the shared storage device 13, processing time is required as in the conventional case, but the processing can be reduced to, for example, only at the beginning of the program.
Since areas in the shared storage device can be allocated within the program by only accessing the main storage device 12, the overhead of area allocation can be reduced.

- [Effects of the Invention] As is clear from the above description, according to the present invention, in a multiprocessor system consisting of an arithmetic unit having a shared storage device and an individual main storage device, the area allocation processing overhead of the shared storage device is This has a significant industrial effect of improving the processing efficiency of the system.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of the present invention, FIG. 2 is a block diagram showing an example of a conventional configuration, and FIG. 3 is a flowchart of area acquisition processing. In the figure, 1.11 is an arithmetic unit, 2.12 is a main storage device, and 3.
13 is a shared storage device, 4.14 is an area acquisition unit, 5 is an exclusive control flag, 6 is free area information, IO is an allocation management unit,
17 is a unique area, 18 is allocation management information, 20
-24 indicate processing steps. Patent applicant: Director of the Agency of Industrial Science and Technology Ken Sugiura II9rfi
a-news ILZ#J Figure 1: Broken diagram showing Leno's uninvented configuration

Claims (1)

[Claims] It has a plurality of arithmetic units (11), a main memory device (12), and a shared memory device (13), and each of the arithmetic units (11) has a separate main memory device (12). )
and the common shared storage device (13) to execute an area acquisition unit (14) and an allocation management unit (10), and the area acquisition unit (14) is connected to the common shared storage device (13). the shared storage device (13) at a predetermined time to start execution of the program;
A specific area is acquired from the free storage area of
7), the allocation management unit (10) holds allocation management information (18) of the unique area (17) in the main storage device (12), and the program executed by the arithmetic unit (11) A shared storage area allocation processing method characterized in that a requested area on a shared storage device (13) is allocated from the specific area according to the allocation management information.