JPS5832422B2 - Micro Shindan Houshiki - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a fault diagnosis method using a microprogram in a multiplexed data processing device.

In online systems that require high reliability, systems are multiplexed so that in the event of a single failure, services can be continuously provided without stopping the system's functions.

In recent years, data processing devices have come to be controlled by microprograms, and with this, so-called microdiagnosis, in which failures are diagnosed using microprograms, is being adopted.

According to this micro-diagnosis, it is possible to diagnose microprograms step by step, and in general, it is possible to perform detailed diagnosis, and it is expected to pinpoint fault locations with high accuracy. Therefore, when a fault occurs, it can be repaired quickly. This meets the requirements for high reliability of online systems.

The main microdiagnostic methods for conventional data processing equipment are: (1) An external writing device is provided to write a diagnostic microprogram to the control memory, and when a failure occurs, the diagnostic microprogram is written to the control memory and then executed. method to do.

(2) A resident microdiagnosis method in which a diagnostic microprogram is stored in the control memory in advance.

(3) A method in which a control memory package for microdiagnosis is added externally or partially replaced at the time of diagnosis.The method in (1) requires an external writing device to be used only for writing the diagnostic microprogram. Although failures rarely occur in method (2), it is necessary to reserve a resident area in the control memory for microdiagnosis, and the control memory cannot be used effectively.

The method (3) requires that a control memory package for diagnosis be prepared for each system, and that a person insert the package one by one at the time of diagnosis.

As described above, all of the conventional methods have major drawbacks.

An object of the present invention is to provide a microdiagnosis method that enables online microdiagnosis using a multiplexed data processing device.

Another object of the present invention is to provide a microdiagnosis method that allows a microdiagnosis program to be written from main memory to control memory without having an external writing device.

Still another object of the present invention is to reduce the capacity of the diagnostic resident microprogram in the control memory by storing the microdiagnostic program in the main memory.

Another object of the present invention is to provide a micro-diagnosis method that makes it possible to improve the self-diagnosis performance of an abnormal data processing device and to increase the throughput of online services of a normal data processing device.

According to the micro-diagnosis method of the present invention, when a failure occurs in multiplexed data processing devices, after identifying normal data processing devices and abnormal data processing devices, (1) a normal central control device; (2) When the abnormal central controller receives an instruction to execute this diagnosis, it executes multiple words of the diagnostic microprogram from the specified main memory address. (3) The abnormal central control unit executes the written multi-word diagnostic microprogram and determines the diagnostic result.

This judgment is made by receiving the notification at the normal central control unit each time the diagnosis is completed, reading the above diagnosis result, and then determining the diagnosis result at the abnormal central control unit. Read with device.

(4) Furthermore, if necessary, the test counter is incremented by one step each time the diagnostic microprogram is executed by the abnormal central controller, and the central controller becomes normal only when an abnormality is detected by judging the diagnostic results. reads the judgment result and test counter.

Next, an embodiment of the microdiagnosis method according to the present invention will be described with reference to the drawings regarding a duplex data processing device.

First, a method of activating the diagnosis of an abnormal central control unit from a normal central control unit will be explained with reference to FIGS. 1 to 4.

After a failure occurs and the normal data processing device and the abnormal data processing device are identified, the central control device of the normal data processing device issues a diagnostic command to the abnormal central control device.

The format is shown in Figure 1: instruction operation code f (this becomes a diagnostic instruction), index register designation X2, pace register designation b2,
It consists of display stand d2, and from this diagnostic command,
The abnormality central control unit calculates from which address in the main memory the diagnostic microprogram should be written into the control memory.

This operation is the sum a2 of the contents of register X2, the contents of register b2, and the displacement d2.

In this execution, as shown in Fig. 2, a diagnosis start instruction is read from the main memory (I), the address a2 from which the diagnosis microprogram is to be read from the main memory is computed L(X), and the result a2 is the register BRo). This shift L (P) and the contents of address a2 are written to bits O) to 31 of the normal shift register (AO), and the remaining upper 32 to 45 bits of this normal shift register are written by a microinstruction. (A
1).

As a result, as shown in Figure 3, the shift register address a2, parity p, etc. are written, and when a transfer start is input to the last bit 44, this bit causes the contents of this shift register to be transferred to the abnormal center. Control is performed to serially transfer the data to the shift register of the control device.

When the contents are received from the shift register on the normal side, the diagnostic microprogram is written from the main memory to the control memory on the abnormal side.

That is, as shown in FIG. 4, the contents of the shift register from the normal side are received by the shift register 5.

Of the contents of the shift register 5, the control information 51 is immediately passed through the selector 9 and decoded by the decoder 10.

The fixed address 30 is set in the control memory address 6 by this decode output, and the diagnostic resident microprogram 12 stored in the fixed address 30 in the control memory 7 is read out to the microinstruction register 8, and the control by this resident program is performed. Move.

By executing this diagnostic resident microprogram 12,
It is checked whether there is any failure in the read/write control relationship with respect to the main memory 11.

After this check is completed, writing of the diagnostic microprogram stored in the main memory 11 to the control memory 7 is started.

That is, among the diagnostic activation information sent from the normal side and stored in the shift register 5, the address information a2 for the main memory 11, that is, the lower 32 bits in FIG. 3, is transferred to the operand address register 2, and a read request is issued. After that, the diagnostic microprogram stored at address a2 of the main memory 11 is read into the buffer register 1, and then transferred to the shift register 5.

Next, the contents of the operand address register 2 are incremented by 1 by the circuit 3 to create the address of the diagnostic microinstruction itself to be brought from the main memory 11 next.

Regarding the format of the data read into the shift register 5, as shown in FIG. 5, the microprogram is written in the DATA section, and the address of the control memory .gamma. to which this microprogram is written is written in the ADR section.

Next, the main memory 11 is read out, and the ADR section 5 subcommittee 3 memory address register 6 in the current report stored in the shift register 5 is set, and the DATA section 52 in the shift register 5 is set.
The data is written to the lower part of the unit side 2 memory 7.

(One word in the control memory 7 is twice as long as the DATA section 52 side 2.

) Similarly, the diagnostic microprogram is read from the primary address of the main memory 11 and written into the upper part of the control memory 7.

The above operation is repeated for the number of words in the diagnostic microprogram writing area 13 of the control memory 7.

In this way, the writing operation of the diagnostic microprogram from the main memory 11 to the control memory γ is completed.

In FIGS. 4 and 7, thick lines with arrows are data lines, and thin lines with arrows are control lines.

Next, the diagnostic execution and completion report of the diagnostic microprogram written in the control memory 7 will be explained with reference to FIG.

When writing of the specified number of words into the diagnostic microprogram area 13 is completed, control is transferred to the written microprogram and microdiagnosis is started.

The method of micro-diagnosis in the central control unit 32 on the abnormal side is that the correct data is contained in the micro-instruction itself and is stored in the micro-instruction register 8, the operation result is placed in the register 26, and the comparison circuit 24 compares the contents of the register 26 with the micro-instruction. A comparison is made with the contents of the register 8, and if a mismatch occurs, the diagnostic result display section 25 is set.

In addition, when a diagnostic microprogram with a specified number of words is executed and one test is completed, the test counter 23 is incremented by 1 by the last microprogram, the normal side central control unit 3i is notified by a program interrupt, and the abnormal side data is The processing device then returns to the stopped state until it is activated by the next diagnostic command from the normal central control device 31.

The diagnosis result display section 25 and the test counter 23 are connected to the information line 2.
7 to the test circuit 22 of the normal central control unit 31.

FIG. 6 shows the control flow and data flow of the registers of the abnormal central controller from when the normal central controller receives the diagnostic activation to the end of the diagnosis.

Finally, the normal central control device 31 determines the diagnosis result of the abnormal central control device 32.

In FIG. 7, the normal central controller 31 learns of the end of the diagnosis by a program interrupt (not shown) from the abnormal central controller 32, and the diagnostic result display section 25 and test counter 23 are connected to the test circuit by the information line 27. 22, and as a result of this interrupt, the diagnostic microprogram is read in the normal central control unit 31 from its control memory 7 into the microinstruction register 8, and the test field of the microprogram is connected to the test circuit 22, which circuit At 22, it is determined whether the condition is satisfied or not, and as a result, the next microprogram address is determined by the control memory address selector 21.

With the above operations, the normal central control unit can read the diagnosis result of the abnormal central control unit, that is, whether the test was completed normally or not, and the test number at that time.

In the above, the abnormal central control unit notifies the normal central control unit of the completion of each test circuit, and the abnormal central control unit stops operating until it receives the next diagnostic command.
It is also possible to not notify the end of the test every time.

In other words, when the address of the main memory where the diagnostic microprogram is stored is notified from the normal central control unit by a diagnostic command, the abnormal central control unit executes the microdiagnosis, and if there is an abnormality, it notifies the normal central control unit. However, if it is normal, it is possible to write the diagnostic microprogram for the next test into the control memory from consecutive addresses in main memory and execute microdiagnosis without waiting for the next diagnostic command from the normal side. .

Also, in order to reduce the amount of hardware, instead of the counter 23 that counts the test numbers, soft counting may be performed using a program on the normal central control unit side.

As explained above, the microdiagnosis method of the present invention has the following effects.

(1) By simply storing a small-capacity diagnostic resident microprogram that checks the write/read functions of the main memory, the control memory can be used effectively, and the central control unit can function normally without the need for an external writing device. The abnormal side performs self-diagnosis in response to a diagnostic command from the normal side. Therefore, the normal side requires only a small amount of control for diagnosis and is able to carry out its original processing, thereby making it possible to perform online micro-diagnosis.

(2) When the abnormality central control unit determines the test result and automatically performs the next test if it is normal, the burden on the normal data processing unit is further reduced and online processing is performed. Improved ability can be measured.

[Brief explanation of drawings]

Figure 1 is a diagram showing the command format of the diagnostic command, Figure 2 is a time chart of the diagnostic command, and Figure 3 is a diagram showing the control report and data clearing information sent to the abnormality central control unit in order to start diagnosis in the abnormality data processing device. Figure 4 is a block diagram showing the configuration of the shift register of the normal central control unit for sending information, Figure 4 is a block diagram showing the configuration for writing the diagnostic microprogram from the main memory to the control memory, and Figure 5 is the block diagram showing the configuration for writing the diagnostic microprogram from the main memory to the control memory. A diagram showing the format configuration of the diagnostic microprogram that is being used.
FIG. 6 is a diagram showing the control flow and the movement of data in each register, and FIG. 7 is a block diagram showing a configuration for reading the diagnostic results of the abnormal central controller on the normal central controller side. 6: Control memory address register, 7: Control memory, 8
: Microinstruction register, 11: Main memory, 22: Test circuit, 23: Test counter, 24: Comparison circuit, 25:
Diagnosis result display section.

Claims (1)

[Claims] 1. In a multiplexed data processing device, a means for a normal central controller to start a microprogram of an abnormal central controller, and a means for causing a microprogram to be stored in the main memory of the abnormal central controller by the activated microprogram. means for writing a plurality of words of a diagnostic macro program into the control memory; means for executing the diagnostic micro program; comparing means for determining the result of the diagnosis; A micro-diagnosis method comprising a reading means.