JPS624743B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for easily modifying diagnostic data used when diagnosing a failure of an information processing apparatus using a functional diagnosis method.

In recent years, the use of information processing devices has expanded in all fields, and the social impact when such devices break down has become increasingly large. For this reason, failure diagnosis technology that quickly points out the location of a failure in a device is becoming increasingly important.

One of the failure diagnosis methods for information processing devices is called functional diagnosis. This is a diagnostic method that monitors the input/output signal status of each part when there is a hardware failure or not, while operating the hardware functions in the same way as when running a normal program, and uses the results to determine the location of the failure. This has the advantage that the amount of diagnostic control hardware added to the device can be reduced.

This function diagnosis method requires input conditions A for sufficiently testing the functions of the hardware, and B, such as a correspondence list of failure groups that can be detected based on the input conditions. Generally, A is called a test input, B is called a diagnostic dictionary, and the combination of A and B is called diagnostic data. Specifically, in order to perform a functional diagnosis, it is necessary to set the status of the device to be diagnosed, set the operation command, advance the clock, read the status of the device to be diagnosed (hereinafter referred to as scan out), and confirm that the read condition is normal. If the condition is different from the (fault-free) state, a series of procedures are required, such as indexing the fault dictionary. Among these procedures, ~ is called test input A, and generally a diagnostic designer specifies each condition. ~ is the diagnostic dictionary B, which consists of the number of clocks at the time of scan out, an address indicating the scan out point, a logical value (normal value) in a normal state, and a fault dictionary. It is determined by hardware or software simulation using the procedures in the fault dictionary as input.

By the way, the method for creating the above-mentioned diagnostic dictionary includes:
(a) A hardware simulation method in which a test input is applied to a normal equipment to be diagnosed with a simulated fault inserted, and a diagnostic dictionary is created from the response state at that time, and (b) A circuit of the equipment to be diagnosed in a computer system. There is a software simulation method in which a diagnostic dictionary is created by expanding the information, performing fault simulation on the test input, and finding the fault groups detected by the test input.For large equipment, method (b) is used. It is common to use Creating a diagnostic dictionary using this method (b) is called failure simulation.

Generally, creating a diagnostic dictionary by fault simulation requires an enormous amount of computer processing time.
Experience shows that the central processing unit of a very large computer (approximately 15
It takes about 5 hours per block, or about 500 hours in total, to divide a CPU with 10,000 gates into about 100 circuit blocks (adders, multipliers, etc.) and create a diagnostic dictionary for each circuit block by fault simulation. Requires calculation time. Once a diagnostic dictionary has been created using this method, if a hardware change occurs in the information processing device to be diagnosed, conventionally, the diagnostic dictionary for the circuit block (often multiple) related to the change will fail again. Recreated by simulation. However, doing so would require a large amount of man-hours to correct the diagnostic data, which would delay delivery to the user, and during this period there would be times when the user did not have the diagnostic program.
A problem arises in that the MTTR (Mean Time to Repair) decreases.

However, when the inventors examined in detail the details of the hardware changes seen in past examples, they found that
It has been found that although the diagnostic dictionary is changed due to hardware changes, there are many cases in which the fault dictionary in the diagnostic dictionary can be used as is without being changed. For example, if a circuit block is added and the number of clocks to reach circuit block i is changed, if the added circuit block is independent, the scan out in the diagnostic dictionary for circuit block i is changed. Only the number of clocks is changed, and the contents of other diagnostic dictionaries remain unchanged. More generalizing, among the elements that make up existing diagnostic dictionaries, the number of scan out clocks, scan
This means that there are cases where it is possible to modify the diagnostic dictionary by simply modifying one or more of the out address and the normal value without performing fault simulation and modifying the fault dictionary.

The present invention is based on the above-mentioned findings, and is characterized in that diagnostic data consisting of a diagnostic dictionary and test input created in advance by fault simulation is used when diagnosing a fault using a functional diagnostic method for an information processing device. In the method of modifying the information processing device in accordance with hardware changes, the fault dictionary among the elements constituting the diagnostic data is used as is, and other scan-out clock numbers, scan-out addresses, and normal values are processed through fault simulation. The point is that you can directly change it without changing it. This will be explained in detail below with reference to the drawings.

FIG. 1 is a schematic explanatory diagram of the method of the present invention, in which the left side shows existing diagnostic data for block number i (corresponding to circuit block i) to be corrected, and the right side shows new diagnostic data after the correction. As mentioned above, the diagnostic data consists of the test input A and the diagnostic dictionary B, and the diagnostic dictionary B further includes the number of scan out clocks n, m, the number of scan out addresses a, b, and the normal value. It is configured with x, y, and fault dictionary α as the minimum unit. still,
Number of clocks m, scan address b, normal value y,
... is used when fault diagnosis is performed under two or more conditions. The present invention is applied to the case where the fault dictionary α of the diagnostic dictionary B does not need to be changed. The solid arrows in the figure indicate data that can be used as-is as new diagnostic data without requiring any changes. On the other hand, the dashed arrows are data that must be corrected due to hardware changes, but these are scan out clock numbers n, m, scan out addresses a, b, normal values. It's just x, y... New diagnostic data
n′, m′…… is the scan out after correction
The number of clocks, a′, b′……… are the corrected scan out addresses, and x′, y′……… are the corrected normal values. Since these corrections can be made without failure simulation, the processing time is only about a few minutes per circuit block.

FIG. 2 is a processing flow showing one embodiment of the present invention, in which a is an overall diagram and b is a detailed diagram of the modification processing program. Figure 2a shows how existing diagnostic data with a designated block number is read from an existing diagnostic data file, it is run through a change processing program according to various changes, and the modified new diagnostic data is stored in the new diagnostic data file. This is what is shown. Figure b shows the flow of processing for changing diagnostic data while determining whether the specified change is the number of clocks, a scan-out address, or a normal value.

When an error is discovered at one point (scanout point) in circuit block i, the fault dictionary α indicates all systems that may cause an error at that point as a group of conductor numbers. It is created in advance using a failure simulation. Of course, depending on the details of the hardware change, it may be necessary to change this fault dictionary according to the fault simulation. The invention is limited to unnecessary hardware changes.

As described above, according to the present invention, the correction of the diagnostic dictionary, which conventionally required several hours through failure simulation, can be shortened to only a few minutes, and the practical effects are enormous.

[Brief explanation of the drawing]

FIG. 1 is a schematic explanatory diagram of the present invention, and FIG. 2 is a flowchart showing an embodiment of the present invention. In the figure, A is a test input, B is a diagnostic dictionary, α is a fault dictionary, n is the number of scan-out clocks, a is a scan-out address, and x is a normal value.

Claims (1)

[Claims] 1. A method for modifying diagnostic data, which is used when diagnosing an information processing device using a functional diagnosis method, and is made up of a diagnostic dictionary and test inputs created in advance through a failure simulation, in accordance with a change in the hardware of the information processing device. , a method for correcting diagnostic data characterized in that among the elements constituting the diagnostic data, the fault dictionary is used as is, and other scan-out clock numbers, scan-out addresses, and normal values are directly changed without going through fault simulation. .