JPH11110309A - Memory diagnosing method of computer system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the rising time of an operating system from the power supply application or reboot start of a computer system till the operation start of the operating system. SOLUTION: This method reads the range of an initial diagnosis (101) from nonvolatile memory such as an EEPROM from the power supply application or reboot start of a computer system till the operation start of an operating system to perform an initial diagnosis of memory (102) and starts up the operating system (104). Then, it starts a job in a memory area in which the operating system is already diagnosed (111). Next, the operating system accesses additional diagnostic processing of a residual diagnosis uncompleted area (115) at a step at which it decides as appropriate during job execution (112) and uses an area where diagnosis is completed for the job (118). The processing is repeated (113) until it finishes diagnosing the entire memory areas mounted on the computer system.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a method for diagnosing memory in a computer system.

[0002]

2. Description of the Related Art As disclosed in Japanese Patent Application Laid-Open No. 4-256033, it is used in a computer system having a multiprocessor configuration, and equalizes the load of memory diagnosis that each processor bears, thereby shortening the startup time of an operating system. There is a memory diagnosis method.

A case where the present method is applied to a process of starting an operating system in a certain computer system will be described with reference to FIG. FIG. 3A shows an operation procedure of the present method, and FIG. 3B shows a memory map of a computer system used in the description of the present method.

First, after turning on the power of the computer system, the amount of memory installed in the system is calculated (30).
0). Next, the calculated memory capacity is divided into the number of mounted processors (301). Here, it is assumed that the number of processors mounted on the system is n, the divided memory areas are M1, M2,..., Mn, respectively, and the memory capacity mounted on the system is Mmax. Next, a processor for diagnosing each memory area is allocated (302). Here, the assigned processor is CP
U1, CPU2,..., CPUn. Next, the assigned processors start their own diagnosis (303).

[0005] The memory diagnosis time in this case is (Mmax / n) ×, where t is the diagnosis time per unit memory capacity.
t. On the other hand, when all the memory diagnoses are performed with a single processor without using this method, the result is Mmax × t, so the effect of the number of processors can be expected by the number of processors and the diagnostic time can be reduced.

[0006]

As described above, in the prior art, the diagnosis time of the memory is determined by the number of processors mounted on the computer system.
Since all memory areas are diagnosed at one time, there is a problem that the diagnosis time is proportional to the memory capacity mounted on the computer system. Therefore, in the case of an emergency reboot due to a processor failure or the like, the recovery time of the operating system becomes longer than before the failure even in an emergency due to the decrease in the number of processors. Further, it is assumed that an online business is performed by a bank or the like in which a plurality of computer systems using the conventional technology are used to share the business with each other. Generally, when the same processor type and the same number of processors are used, the processing capacity of the computer system is proportional to the installed memory capacity. Normal. Therefore, computer systems having different memory capacities are mixed. Here, when the power is simultaneously turned on to all the computer systems and the business is started in a state where all the operating systems have been started, the start-up time of each computer system differs according to the memory capacity and the number of processors mounted in the conventional technology. Therefore, the monitoring time from when the power is turned on until the start of work must be adjusted to a computer system with the largest memory capacity or a computer system with the least number of processors. Therefore, there is a problem that the monitoring time until the start of work must be readjusted every time the memory and the processor are increased or decreased.

In the prior art, the operating system can use all the memory areas immediately after startup, so that the processing capability of the computer system can be maximized. On the other hand, since all the memory areas are used, the hardware memory elements are used. Focusing on the above, all the memory elements are also used. If the failure occurrence probability of one memory element is F and the number of used elements is n, the normal operation probability of the entire memory is (1-F). It becomes the nth power. Therefore, there is a problem that the normal operation probability of the memory becomes minimum immediately after the start of the operating system.

[0008] Focusing on the operation mode of the computer system, the work does not concentrate immediately after the start of the operation.
It is common to concentrate on a fixed time every day. Therefore, the operating system is not required to have the maximum processing capability immediately after startup. Therefore, if the memory capacity is dynamically secured by paying attention to the operation mode of the computer system, the reliability of the memory can be secured for the above-described reason. In addition, the processing capability of the computer system according to the task can always be ensured.

The present invention solves the above-mentioned problem by diagnosing all the memories before starting the operation of the operating system and executing only a part of the memories and diagnosing the remaining undiagnosed areas again after the operation of the operating system starts. By doing so, it is intended to contribute to improvement of operability and reliability of the computer system.

[0010]

In order to solve the above-mentioned problems, a computer system according to the present invention has a means for storing a memory capacity to be diagnosed before starting an operating system in a nonvolatile memory such as an EEPROM. Means for reading the data, means for diagnosing the memory up to the specified capacity using the same, means for notifying the operating system of the memory address where the diagnosis has been completed and the memory capacity pre-installed in the computer system, and Means for scheduling the diagnostic processing of the memory area which has not been completed by the operating system; means for diagnosing the memory area which has not been diagnosed according to the instruction of the operating system;
Means for informing the operating system of this diagnosis result.

[0011]

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

FIG. 1A shows a procedure of an initial diagnosis process of a memory performed from a power-on or a reboot of a computer system to a start of an operation of an operating system. FIG. 1B shows a memory called by the operating system during a job. FIG. 1 shows the procedure of the additional diagnosis processing of FIG.
FIG. 1C shows a procedure of a process for updating an initial diagnosis range of a memory called by an operating system during a job, and FIG. 1D is a memory map of a computer system used in the embodiment of the present invention. FIG. 2 is an operation explanatory diagram used in the embodiment of the present invention.

It is assumed that the time at which the power is turned on to the computer system and the range in which the initial diagnosis of the memory is performed are in the non-volatile memory 200, the current time is in the hardware timer 201, the power is automatically turned on, the memory is diagnosed and the non-volatile The process of updating the memory 200 is performed by the controlware 20.
2, the operating system to be booted is in the boot device 203, and the interface area between the controlware 202 and the operating system is 204.
In addition, it is assumed that the nonvolatile memory 205 stores a time zone during which work is concentrated and an additional diagnostic time per unit memory capacity. It is also assumed that in the initial state of the computer system, the operating system has not started and the automatic power-on process linked to the hardware timer 201 is enabled. The operating system uses all the memory capacity of the computer system stored in the non-volatile memory 205 during the busy time of the business, and the memory capacity at the time of startup during the other time. It is assumed that a task is scheduled in advance to execute a task.

First, in the automatic power-on processing, the power-on time in the nonvolatile memory 200 is compared with the hardware timer 201, and when a scheduled time is reached, an initial diagnosis processing of the memory is started. In the initial diagnosis process, first, after the power of the computer system is turned on, all the memory capacities mounted on the system are calculated (100). Next, the nonvolatile memory 200
The initial diagnostic range is read from (101). Here, the memory mounting capacity of the system is Mmax, and the initial diagnostic range of the memory is Mi. Next, an initial diagnosis (102) of the memory is performed for the read diagnosis range. Next, the memory capacity mounted on the system and the memory capacity for which the initial diagnosis has been completed are compared with the interface area 20 of the operating system.
4 (103). The stored values in this embodiment are Mmax and Mi, respectively. Then, the operating system is stored from the boot device 203 in the memory area where the initial diagnosis has been completed and started up (104).

Next, the started operating system reads from the interface area 204 of the operating system the capacity of all the memories mounted on the computer system and the capacity of the memory which has been subjected to the initial diagnosis (110). Then, the operation is started (111) within the range of the memory for which the initial diagnosis has been completed. In the case of the present embodiment, the business is started from address 0 to Mi of the memory. Next, the operating system compares the amount of memory that can be additionally diagnosed, the current time in the hardware timer 201, the time of business congestion stored in the non-volatile memory 205, and the additional diagnostic time per unit memory capacity. The time to start the additional diagnosis, the additional diagnosis capacity per one time, and the interval are determined, and the time until the time is reached (11).
2). Next, when it is time to start the additional diagnosis, the operating system stores the capacity for the additional diagnosis in the interface area 204 of the operating system (114). Then, an additional diagnostic process in the controlware 202 is called (115). Here, the diagnosis start address of the additional diagnosis process is a diagnosed address in the interface with the operating system. Next, in the called additional diagnostic process, the memory is diagnosed according to the instruction of the operating system (116), and when the diagnosis is completed, the diagnosed address in the interface area 204 for the operating system is updated (117).
Next, the operating system incorporates and uses the area for which additional diagnosis has been completed in the business (118). Then, scheduling (112) is performed so that all the memory diagnoses are completed by the time when the business stored in the nonvolatile memory 205 is busy.

Next, when the operating system determines that the time period during which the business is congested has passed, the memory area used for the business is degenerated until immediately after startup, and its address is diagnosed in the interface area 204 for the operating system. Write back to address. In this embodiment, the value to be written is Mi. By the way, FIGS. 1 and 2
Does not show the degeneracy process.

If the operating system analyzes the business (120) and determines that a larger amount of memory is required even in the normal business (121), the operating system determines the required memory capacity in the interface area 204 for the operating system. Is stored in the update address of the initial diagnostic range (122), and the update process is called (123). The called update process is performed according to the instruction of the operating system, and the nonvolatile memory 200 in which the initial diagnosis range is stored.
Is updated (124).

[0018]

According to the present invention, before starting the operation of the operating system, all the memories are not diagnosed but only partially executed, and after the operation of the operating system is started, the remaining undiagnosed areas are diagnosed again. To improve the problem.

Since the rise time of the operating system is related to the initial diagnostic capacity, it is always constant without being affected by the memory capacity mounted on the computer system, and is not affected by the number of processors mounted. Therefore, even during an emergency reboot due to a processor failure or the like, the recovery time of the operating system does not change as compared to before the failure occurred. Further, since the rise time of the operating system is related to the initial diagnostic capacity, even if a plurality of computer systems equipped with different memory capacities are mixed, if the initial diagnostic capacity is unified, the rise time of the entire computer system is always constant. Therefore, it is not necessary to readjust the monitoring time until the entire computer system starts up every time the memory and the processor are increased or decreased.

Further, since the operating system can dynamically change the memory capacity to be used as needed, the number of memory elements to be used also changes, and the reliability of the memory can be improved as compared with the conventional technology. it can. further,
The operating system can use the memory area for which the diagnosis has been completed immediately before the business use of the memory, so that business stoppage due to a memory failure can be avoided.

Further, the operating system can dynamically change the memory capacity that can be used in a business operation mode in which the business is concentrated only during a certain period of time, for example. Nature can be secured. In addition, the processing capability of the computer system according to the task can always be ensured.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the operation principle of the present invention.

FIG. 2 is a diagram illustrating the operation of the present invention.

FIG. 3 is a diagram illustrating the operation principle of the conventional technique.

[Explanation of symbols]

130: memory map of a computer system used in the description of the operation of the present invention; 200: nonvolatile memory;
201: hardware timer, 202: control wear, 203: boot device, 204: interface with operating system, 205: non-volatile memory, 310: memory map of a computer system used in the description of the operation of the conventional technology.

Claims (2)

[Claims] In a method for diagnosing the normal operation of a computer system, a diagnosis of a memory mounted on the computer system is performed not in the entire area but in a part of the memory from the power-on or reboot of the computer system to the operation of the operating system. Only after
The operation of the operating system is started in the diagnosed area, and the remaining undiagnosed area is re-executed during the operation of the operating system, and released to the operating system, so that the rise time of the operating system can be reduced to the actual unit memory capacity. A memory diagnosis method for a computer system, wherein the memory diagnosis method is apparently shortened without being proportional to a diagnosis time per hit. 2. After a certain period of time has elapsed since the start of the operating system, a memory area that has not yet been diagnosed is diagnosed and released to the operating system, or a certain period of time has passed since the operating system was started. 2. The computer system according to claim 1, wherein the processing capacity of the computer system is dynamically adapted to, for example, a concentration of business by reducing the memory area used by the operating system and re-diagnosing and securing the memory area as needed. Memory diagnostic method.