JPH0962534A - Self-diagnostic method for electronic computer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the reliability of electronic computer system which obtains fault detection ability during the operating of electronic computer by performing self-diagnosis while the electronic computer is in the idle state. SOLUTION: The idle state during the operation of electronic computer is detected and the self-diagnosis of hardware of electronic computer is executed. Namely, when HD access processing 20, keyboard input processing 21 and communication processing 22 are performed at fixed time intervals by a certain program but there is no IO access for fixed time after the respective processing, it is detected by a timer 13 inside an OI access monitor logic 6 and most preferential interruption is generated. In this most preferential interruption processing, first of all, a diagnostic object is selected. Next, the information of a section to be the diagnostic object is saved. Further, the diagnosis of the diagnostic object is performed. Then, the result of this diagnosis is discriminated and when there is no error, the recovery of state is performed but when there is any error, error processing is performed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a self-diagnosis technique for electronic computers, and more particularly to a self-diagnosis technique effective for use in a server-class computer equipped with a large capacity memory and a large number of IO devices.

[0002]

2. Description of the Related Art In a general electronic computer system, a diagnostic program stored in a ROM or a diagnostic program loaded by a program stored in a ROM performs initial diagnosis and initial setting of hardware only when power is turned on. I am doing it. This self-diagnosis is performed only when the power of the device is turned on or when the device is reset. When the computer is operating, the occurrence of memory parity is detected by hardware, or IO device access is performed. A method of processing an error detected in association with a hardware access operation of software is generally performed, in which an operating software monitors an error status that sometimes occurs in an IO device.

[0003]

In a computer that constructs a large-scale system, the computer is often energized at all times except when specific maintenance is performed. Therefore, the frequency with which the initial diagnosis of the hardware executed when the power is turned on is also reduced, and the chance of finding a hardware failure decreases. In addition, while the computer is operating, a method is adopted to detect an error that occurs when the software attempts to perform a certain process, but this method allows the software to perform the operation that it tried to do at the time when the error was detected. In some cases, the process cannot be continued. This results in a decrease in the reliability of the electronic computer.

In an electronic computer that constitutes a large-scale system, many computer resources such as memory and IO are installed, and all of these hardware are diagnosed by a diagnostic program when the power is turned on as described above. Has a problem that it takes a lot of time to start up an electronic computer and is not easy to use.

[0005]

An idle state during operation of an electronic computer is detected, and a self-diagnosis of the hardware of the electronic computer is carried out.

[0006] It is desirable that the computer be started up without self-diagnosis of the hardware of the computer when the computer is powered on.

[0007]

[Function] Since the hardware self-diagnosis is performed by detecting the idle state during the operation of the electronic computer, the frequency of the hardware self-diagnosis can be guaranteed even when the electronic computer is operated continuously for a long time. Therefore, the reliability of the electronic computer can be improved.

Further, since the hardware self-diagnosis is not performed when the power of the electronic computer is turned on, the startup time of the electronic computer can be shortened.

[0009]

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows an example of a hardware configuration of an electronic computer having a highest priority hardware interrupt, a memory for saving data, and an IO device monitoring logic, and shows a configuration of a computer system according to an embodiment of the present invention. It is a block diagram shown.

In FIG. 1, 1 is a CPU, 2 is a memory controller, 3 is a main memory, 4 is a CPU bus, and the CPU 1 is connected to a memory controller 2 and a main memory 3 via a CPU bus 4. Further, 7 is a system bus, 8, 9 and 10 are IO controllers connected to the system bus 7, and 5 is a bridge connecting the CPU bus 4 and the system bus 7. That is, CP
The U bus 4 is connected to the system bus 7 via the bridge 5, and each IO controller 8 on the system bus 7
9 and 10 have a structure accessible from the CPU 1. Reference numeral 6 denotes an IO access monitoring logic, which is hardware having a function of monitoring access of the CPU 1 to the IO controller. The IO access monitoring logic 6 has one or more timers 13 therein, and the IO access monitoring logic 6 keeps 1
When the CPU does not access the corresponding specific IO as indicated by 2, the CPU 1 has a function of generating the highest priority interrupt. The value of the timer 13 and the IO controller to be monitored for IO access can be programmed by software, and the IO access monitoring condition that is the optimum condition for the electronic computer system can be set.

The conventional self-diagnosis is performed by the CPU 1 executing the contents of the ROM 14 when the power of the electronic computer is turned on. When the CPU 1 is reset when the power is turned on, the CPU 1 starts execution from a specific address. The address first executed by the CPU 1 is hardware-assigned to the ROM 14,
A program for verifying the contents of the program of the ROM 14 itself and a program for testing the main memory 3 and the IO controllers 8, 9 and 10 are written in advance in the ROM 14 so that the hardware configuring the electronic computer when the power is turned on is written. The self-diagnosis of the wear was done at once. According to this method, if the number of components or the amount of hardware configuring the electronic computer increases as described above, the time required for diagnosis increases accordingly, and a hardware failure that occurs during the operation of the electronic computer is detected. There was a problem that I could not.

FIG. 2 is a flow chart showing an example of a program which operates on the hardware shown in FIG. This is because, after the electronic computer has performed a specific process, there is no IO access for a certain period of time, then a highest priority interrupt occurs, the hardware self-diagnosis is executed within the highest priority interrupt process, and again This is an example of returning to the processing of.

Normally, processing is performed on an electronic computer according to a given program. For example, the HD access process 20, keyboard input process 21,
It is assumed that the communication process 22 is performed at regular time intervals. After these processes, when there is no IO access for a certain time (step 23), the timer 13 in the IO access monitoring logic 6 detects this and generates the highest priority interrupt. This interruption occurs unmotivated by the processing indicated by 20, 21, 22 and the operation of the OS.
Therefore, when the highest priority interrupt occurs, all operating states at that time are saved and the process shifts to the highest priority interrupt processing.

The program for the highest priority interrupt processing is usually R
It is stored in the OM. In the highest-priority interrupt processing, the diagnosis target is first selected (step 24). Here, the diagnosis to be performed in this interrupt processing is determined. For example, when diagnosing a memory, the target address is determined. Although there are various methods of making this determination, a method in which the memory space is divided into spaces of 1 KB each and the divided memory spaces are diagnosed one by one each time the highest priority interrupt occurs can be considered. At this time, the information may be recorded in the nonvolatile RAM 15 in order to remember the memory space to be diagnosed next. The same applies to the diagnosis of the IO controller, with the highest priority interrupt processing 8, 9,
It is conceivable to record information in the nonvolatile RAM 15 as to which of the IO controllers shown by 10 is to be diagnosed.

Next, the information of the portion to be diagnosed is saved (step 25). This is to prepare for the case where the target state changes due to the diagnosis. For example, when diagnosing the memory, the contents of the memory are saved to another area. As the save area at this time, an area not normally used by the OS or application program is assigned. For example, RO
R, which is not normally used, is located at the same address as M
AM and non-volatile RAM can be used as such areas. Next, the diagnosis target is diagnosed (step 26). This processing is, for example, a memory read / write test if the diagnosis target is a memory, and an IO register read / write test if the diagnosis target is an IO controller. The result of this diagnosis is judged (step 27), and if there is no error, the state is restored (step 29), but if there is an error, error processing is carried out (step 28). This error processing includes, for example, notifying higher-level software that an error has occurred, storing error information in a non-volatile RAM, degeneration processing of an error occurrence site, and the like. When the processing up to this point is completed, the processing of this highest priority interrupt is completed, so the state saved in step 25 is restored and the interrupt processing returns. If there is a process instructed by the program as the next process, that process is executed, and if there is no such process, the highest priority interrupt is generated again after a fixed time, and the same self-diagnosis operation is performed.

Further, when all self-diagnosis processing is executed by the highest priority interrupt as in the present invention, and only the minimum necessary diagnosis for operating some OSs is executed at the time of power-on, and other than that. A method of implementing the part with the highest priority interrupt is also possible.

[0018]

According to the present invention, it is possible to detect a fault during the operation of the electronic computer by performing self-diagnosis while the electronic computer is in the idle state, and it is possible to improve the reliability of the electronic computer system. Further, the start-up time of the electronic computer can be shortened by simplifying the self-diagnosis when the electronic computer is powered on.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a computer system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an example of a program that operates on hardware according to an embodiment of the present invention.

[Explanation of symbols]

1 ... CPU, 2 ... Memory controller, 3 ... Main memory, 4 ... CPU bus, 5 ... Bridge, 6 ... IO access monitoring logic, 7 ... System bus, 8, 9, 10 ... IO
Controller, 11 ... Highest priority interrupt, 12 ... CPU IO access operation, 13 ... Timer, 14 ... ROM, 15
... nonvolatile RAM.

Claims (3)

[Claims] 1. A self-diagnosis method for a computer, wherein an idle state during operation of the computer is detected, and a self-diagnosis of hardware of the computer is carried out. 2. A computer according to claim 1, wherein the hardware of said computer is self-diagnosed when the power is turned on, and then the hardware of said computer is self-diagnosed during operation. Self-diagnosis method. 3. The self-diagnosis method for an electronic computer according to claim 1, wherein the electronic computer is started up without performing self-diagnosis of the hardware of the electronic computer when the power of the electronic computer is turned on.