JPH01320563A - Common memory control method - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Summary] Common memory control method in a redundant system In order to solve the problem that in the conventional check point method, when an error occurs in one system, the data being input at that time becomes invalid. In addition, by utilizing a common memory that can be accessed from both systems and adding a means to confirm the normal completion of data input to the common memory, continuous use of input data is made possible.

[Industrial application field]

The present invention relates to a common memory control method in a duplex system. The present invention is particularly effective when applied to cases where loss of data input and re-input have a large impact on the system.

[Conventional technology]

When an abnormality occurs in the system in terms of both hardware and software, in a redundant system, the system is switched and processing continues by going back to a previously recorded check point. Therefore, if there is input data during this time, it will be necessary to re-input it.

[Problem to be solved by the invention]

In the case of experimental data, etc., it may be highly temporary and cannot be re-inputted, or even if it is possible, it may require a considerable amount of money, which is a major problem.

The present invention was created in view of this point, and is a common memory that provides a common memory in a redundant system and uses the common memory to enable effective use of data even when one system is abnormal. The purpose is to provide a control method.

[Means for Solving the Problem] Even if an abnormality occurs in the system, it is often possible to continue accepting data unless the data input hardware is abnormal. In the case of a duplex system, the present invention utilizes a shared memory that can continue to be accessed from the switched system to take over data even if the systems are switched. However, in this case,
The important point is that it is possible to confirm that the data input has been completed normally only from the common memory itself. The present invention deals with this by providing a reference flag for each data unit in the common memory.

The duplex system of the present invention includes an A-system computer system,
It includes a B-system computer system, a common memory 300, an input/output device 600, and a changeover switch 500.

The A-system computer system has a central processing unit 100A, a memory 200A, and an input/output control device 400A, and the B-system computer system has a central processing unit 100B, a memo 'J
20OB and an input/output control device 400B.

The changeover switch 600 switches the input/output device 600 between the input/output control device 400A and the input/output control device? 2400B. The common memory 300 has a reference flag provided for each predetermined data unit, and when the reference flag is “
When accessed under the state of “0”, the reference flag becomes “1”
configured to be. The common memory 300 is accessible from both the A-system computer system and the B-system computer system.

When the main computer system attempts to perform an input/output operation, the main computer system uses the area allocated as the input/output data area, the area allocated as the startup data storage area, and the area allocated as the startup result report area in the common memory 300. Notify the slave computer system and save the common memory 3
The reference flag of 00 is set to "0'", the startup data is stored in the startup data storage area of the common memory 300, and the startup data is sent to the input/output control device of the own system for startup.

The input/output control device is configured to perform input/output control operations upon receiving an input/output activation instruction, and store an activation result report in the activation result report area of the common memory 300 upon completion of the input/output operation.

When the slave computer system detects an abnormality in the main computer system, it refers to the startup data storage area of the common memory 300, reads the startup result report area, and determines that the startup result indicates normal termination. In this case, the system is configured to continue operating as a duplex system using data in the input/output data area of the common memory 300.

[Embodiment] FIG. 2 is a block diagram of an embodiment of the present invention, FIG. 3 is a block diagram showing an example of the configuration of a common memory control section, and FIG. 4 is a diagram showing an example of the layout of the common memory. In the figure,
100A and 100B are central processing units, 200A and 200
'B is a memory, 300 is a common memory, 310 is a common memory control unit, 311 is a control register, 312 is a microprocessor, 320 is a common memory memory unit, 400^
and 400B are input/output control devices, 500 is a changeover switch,
Reference numeral 600 indicates input/output devices. Also, ■,
■, . . . indicate the order of operation.

The central processing unit 100A, memory 200A, and input/output control device 400A constitute an A-system computer system, and the central processing unit 100B, memory 200B, and input/output control device 400B constitute a B-system computer system. The central processing unit 100A is a central processing bag 'f1100
Similarly, the central processing unit 100B is monitoring the state of the central processing unit 100B.
monitors the status of central processing unit xooA. The common memory 17300 is shared by the A system and the B system.

The common memory 300 includes a control section 310 and a memory section 32.
Consists of 0. Control section 310 includes control registers, a microprocessor, control storage, and common memory clear circuitry. As shown in FIG. 3, control register 311 stores status, commands, and command details. The memory section 320 has a layout as shown in FIG. That is, the memory unit 320 has an input/output data area, an activation data storage area for the input/output control device, and an activation result reporting area for the input/output control device.

Further, a reference flag is provided for each data unit of a predetermined length. The specifications of the reference flag are as follows.

■ When accessed from the system (either reading or writing is fine) while it is in the “0” state, it changes to “1”.

■ There is at least one reference flag per byte. Whether or not to set one for every 2 bytes or more depends on how the system memory is used.

■ “1” is sent from the common memory control unit to each reference flag.
Or “I can confirm 0pa.

■ When the power is turned on, it becomes "0". However, it is also possible to use the function (■) and have the control unit of the common memory perform this operation.

The common memory control unit 310: (a) specifies an address and sets the corresponding reference flag to "0";

(2) Specify a range of addresses and check which of the corresponding reference flags are set to 1.

It has the following functions. Input/output control devices 400A and 400B can access common memory 300. The input/output control devices 400A and 400B transfer the result report to the startup result report area upon completion of the operation.

The address of the activation result report area is specified at the time of activation.

The changeover switch 500 connects the input/output device 600 to the input/output control bag f400A or 400B.

The operation of the embodiment shown in FIG. 2 will be described in the case where system A is the main system, system B is the slave system, and data input is performed.

■ The usage (layout) of the common memory should be understood between A and B systems.

■ When a startup cause occurs in the A system, instruct the common memory control unit to set the reference flag of the area in the common memory to be used to "0".The target is the startup data storage area,
These are a startup result report area and an input/output data area.

■ Write startup data from A system to common memory.

■ Start the input/output control device from the A system. The startup data at this time is the same as the startup data in (3).

■ Data input begins from the input/output control device.

■Here, assume that an abnormality has occurred in the A system.

■'B system takes over the system control authority of A system.

(2) The B system causes the common memory control unit to search for the part of the reference flag "1" in the start-up data storage area known in advance, and obtains the address of "1".

■ Based on the data in ■, make the common memory control unit check the reference flag of the startup result report area.

If it is "0", wait for a certain period of time (preferably store it in the startup data storage area at startup) and then check again.If it is "1", proceed to the process for "1" below.
``Then, the input system of system A has also failed, so we give up on taking over the input data.

"When '1°', check the contents of the activation result, and if it ends normally, proceed to (2); if it ends abnormally, give up taking over the input data.

■ For confirmation, use the reference flag to check whether data of the specified length has been input.

[Phase] The B system instructs the control unit of the common memory to set the related reference flag to "0".

■ Continue processing as a system by referring to manually generated data. Note that the changeover switch is switched after the process (3) is completed.

The above description is about data input, but the same can be done about data output. However, since the data to be sent is in the common memory, even if the output system hardware fails in a failed system, it can be retransmitted on a new system.

〔Effect of the invention〕

As is clear from the above explanation, according to the present invention, even if the input/output activation system fails before the operation is completed, the data can be made valid as long as the data input system does not fail. This is effective when receiving data that is difficult to retransmit, and can improve system reliability.

Another advantage is that the hardware required for this is concentrated in just one component, the common memory, and there is no need to modify various elements piece by piece.

[Brief explanation of the drawing]

FIG. 1 is a diagram of the principle of the present invention, FIG. 2 is a block diagram of an embodiment of the present invention, FIG. 3 is a block diagram of an example of the configuration of a common memory control section, and FIG. 4 is an example of the layout of the common memory. FIG. 1008 and 100B...Central processing unit, 200A and 2
00B...Memory, 300...Common memory, 310
... common memory control section, 311 ... control register, 312 ... microprocessor, 320 ... common memory memory section, 400A and 400B ... input/output control device, 500 ... changeover switch , 600... input/output device. Patent applicant: Pfn Co., Ltd.

Claims (1)

[Claims] An A-based computer system having a central processing unit (100A), a memory (200A), and an input/output control device (400A), a central processing unit (100B), a memory (200B), and an input/output control device. (400B), an input/output device (600), and a switch that connects the input/output device (600) to either the input/output control device (400A) or the input/output control device (400B). A switch (500) is configured such that a reference flag is provided for each data unit of a predetermined length, and the reference flag becomes "1" when accessed while the reference flag is "0". In a redundant system that includes a common memory (300) that can be accessed from both the system computer system and the system B computer system, when the main system computer system attempts to perform an input/output operation, it The area allocated as the data area, the area allocated as the startup data storage area, and the area allocated as the startup result report area is notified to the slave computer system, and the common memory (300
) is set to "0", the startup data is stored in the startup data storage area of the common memory (300), and the startup data is sent to the input/output control device of the own system for startup. The devices (400A, 400B) are configured to perform an input/output control operation upon receiving the input/output activation instruction, and store an activation result report in the activation result report area of the common memory (300) when the input/output operation is completed; When the slave computer system detects an abnormality in the main computer system, it refers to the startup data storage area of the common memory (300) and reads the startup result report area, and if the startup result indicates normal termination. If it is determined, the common memory (30
A common memory control method characterized in that the common memory control method is configured to continue operating as a duplex system using data in the input/output data area of 0).