JPH07295862A - System control method applying error tracing - Google Patents

Abstract

PURPOSE:To accurately and easily decide an error occurrence place based on the stored trace information, i.e., the attribute information on the state change of a program. CONSTITUTION:A trace processing hardware 22 functions to collect the attribute information on the state change of programs and stores these collected information in a tracing buffer 24 when the trace processing is started by a user program 21a or a system program 21b. In a system control method which applies the error tracing, the process attribute information is stored in a tracing buffer memory or a tracing buffer area of the memory when the state change of the program occurs. When an error is detected, the attribute information stored in the buffer memory or the buffer area is traced. Then an error deciding program decides the program of the error occurrence place.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system control management method by error tracing for facilitating the recognition of an error source and controlling and managing a system according to the recognized error source.

[0002]

2. Description of the Related Art A conventional computer system composed of a memory system operating at a low speed and a small amount of hardware operating at a high speed determines the source of an error by simultaneously detecting the error and detecting the error. It is assumed that the error detection is performed before the source causes the state change of the program. That is, the program that started the operation that causes the error and the program that was being executed when the error was detected are in the same program state (user program or system program), and the program that causes the error and the program that detected the error are Assuming that the processes are the same, the cause of the error and the source of the error were certified, and the error was processed.

FIG. 23 is an explanatory diagram schematically showing an error occurrence situation in a computer system with the passage of time, and in the figure, two processes A and B different from 1 and 2 are shown. Further, 3 indicates the execution state of the system program, and 4 indicates the execution state of the user program. Reference numerals 5 to 9 indicate timings at which an operation for generating an error is performed, and reference numerals 10 to 13 indicate the timings 5 to 5.
9 shows the timings at which the respective errors corresponding to the occurrence of the error due to the operation of causing the error of No. 9 are detected. Timing 5 and Timing 10, Timing 6 and Timing 11, Timing 7 and Timing 12, Timing 8 and Timing 13, Timing 9 and Timing 1
Reference numeral 4 is a set of error occurrence and error detection.

In the conventional error source qualification, which is performed on the assumption that the error source is detected before the error source causes a change in the state of the program, the conventional computer system has a memory system operating at a low speed and a small capacity. Since it is composed of hardware that operates at high speed (refers to cache memory, write buffer for speeding up writing, etc.), the timing of the operation that causes an error and the timing of detecting the error that occurs The interval is not so large as shown in FIG. 23, the error is detected and the error is detected at the same time, and the error is detected based on the premise that the error is detected before the error source causes the program state change. Was being done.

[0005]

While the conventional method of recognizing the error source of a computer system has been performed under the above-mentioned environment, the progress of hardware technology in recent years has made it possible to further improve the CPU and speed. A computer system that uses a large amount of cache memory can be configured,
For this reason, even if the CPU executes an operation that causes an error, the resulting error is not immediately detected, and on the other hand, the CPU can execute the program in the cache memory at high speed, so the program is not detected when the error is detected. There is a situation where it is running far from the error source,
The state of the program being executed and the state of the program where the error occurred may not match.

That is, timing 5 and timing 10, timing 6 and timing 11, timing 7 and timing 12, timing 8 and timing 1 shown in FIG.
3, among the pairs of error occurrence and error detection such as timing 9 and timing 14, among the pairs of error occurrence source and error detection of timing 6 and timing 11, timing 8 and timing 13, timing 9 and timing 14,
Since the program state at the time when an error is detected is different from the program state of the error source, the error source cannot be accurately identified. In the set of timing 6 and timing 11, the user program is the error source, but the system program is erroneously recognized as the error source, and the timing 8 and the timing 1 are set.
In the set of 3, the system program is the error source, but the user program is mistakenly recognized as the error source, and in the set of timing 9 and timing 14, the unrelated process B is recognized as the error source. It will be.

That is, in the conventional computer system, the precondition that the execution program that causes the error and the execution program that has detected the error are the same process is not satisfied, and it is difficult to accurately detect the error. The problem arises.

The present invention has been made to solve the above-mentioned problems, and the invention of claim 1 provides a system control management method by an error trace capable of accurately and easily recognizing an error occurrence source. The purpose is to get.

It is an object of the present invention to provide a system control management method by an error trace that facilitates recognition of an error occurrence source even when the program state at the time of error occurrence and the program state at the time of error detection are different. .

It is an object of the invention of claim 3 to provide a system control management method by error tracing which can accurately and easily identify a program using a resource that has caused an error.

According to a fourth aspect of the present invention, there is obtained a system control management method by an error trace capable of accurately and easily recognizing an error source even when a program state at the time of error occurrence and a program state at the time of error detection are different. The purpose is to

According to a fifth aspect of the present invention, there is provided a system control management method based on an error trace capable of accurately and easily recognizing an error source even when a program state at the time of error occurrence and a program state at the time of error detection are different. The purpose is to

It is an object of the present invention to provide a system control management method by error trace which enables effective use of a trace buffer and can accurately and easily identify an error occurrence source.

It is an object of the present invention to provide a system control management method by error tracing which can accurately and easily identify an error source.

It is an object of the present invention to provide a system control management method by error tracing, which can accurately and easily identify an error source.

It is an object of the present invention to provide a system control management method by error tracing, which can accurately and easily identify an error source.

It is an object of the invention of claim 10 to obtain a system control management method by an error trace capable of accurately and easily recognizing an error source.

It is an object of the invention of claim 11 to obtain a system control management method by an error trace capable of accurately and easily recognizing an error source.

It is an object of the invention of claim 12 to obtain a system control management method by an error trace capable of accurately and easily identifying an error occurrence source.

It is an object of the invention of claim 13 to obtain a system control management method by an error trace capable of accurately and easily recognizing an error source.

It is an object of the present invention to provide a system control management method by error tracing, which can accurately and easily identify an error source.

It is an object of the invention of claim 15 to obtain a system control management method by an error trace capable of accurately and easily recognizing an error occurrence source.

An object of the 16th aspect of the present invention is to obtain a system control management method by an error trace capable of accurately and easily recognizing an error occurrence source.

An object of the 17th aspect of the invention is to obtain a system control management method based on an error trace capable of accurately and easily recognizing an error occurrence source.

It is an object of the invention of claim 18 to obtain a system control management method by an error trace capable of accurately and easily recognizing an error occurrence source.

It is an object of the invention of claim 19 to obtain a system control management method by error tracing capable of accurately and easily recognizing an error occurrence source.

According to the twentieth aspect of the present invention, it is possible to obtain a system control management method by an error trace that can efficiently collect the data necessary for certifying the error source and can accurately and easily certify the error source. To aim.

It is an object of the invention of claim 21 to obtain a system control management method by an error trace capable of accurately and easily recognizing an error occurrence source.

It is an object of the invention of claim 22 to obtain a system control management method by error tracing capable of accurately and easily recognizing an error occurrence source.

An object of the 23rd aspect of the invention is to obtain a system control management method by error tracing, which can accurately and easily identify the error source.

It is an object of the invention of claim 24 to obtain a system control management method by error tracing capable of accurately and easily recognizing an error occurrence source.

It is an object of the invention of claim 25 to obtain a system control management method by an error trace for accurately and easily recognizing an error occurrence source and improving system reliability.

The object of the 26th aspect of the invention is to obtain a system control management method by an error trace for accurately and easily recognizing an error source, improving the utilization rate of effective resources, and improving system reliability. And

It is an object of the invention of claim 27 to obtain a system control management method based on an error trace for accurately and easily recognizing an error occurrence source and improving system reliability.

It is an object of the invention of claim 28 to obtain a system control management method by an error trace for accurately and easily recognizing an error occurrence source and improving system reliability.

It is an object of the invention of claim 29 to obtain a system control management method by error tracing that accurately and easily identifies an error source and improves system reliability.

It is an object of the invention of claim 30 to obtain a system control management method by an error trace for accurately and easily recognizing an error occurrence source and improving system reliability.

An object of the invention of claim 31 is to obtain a system control management method by error tracing which accurately and easily identifies an error source and improves system reliability.

[0039]

According to another aspect of the present invention, there is provided an error trace system control management method, wherein attribute information indicating a change in a program state is stored in a trace buffer, and an error is recorded based on the stored attribute information. It is equipped with a structure that certifies the source.

In the system control management method by error tracing according to the second aspect of the present invention, the attribute information indicating the program state change is stored in the trace buffer in the order of occurrence time, and the stored attribute information is advanced in the order of occurrence time. Alternatively, it is configured to certify the program related to the cause of the error by performing a return search.

According to the third aspect of the present invention, there is provided an error trace system control management method, which retrieves attribute information stored in a trace buffer, which allows a process to know which resource the process uses, and uses the attribute information. It is provided with a configuration for recognizing the program that uses the resource that has caused the error by knowing the existing resource.

According to a fourth aspect of the present invention, there is provided a system control management method using error tracing, which has a configuration in which a program state before a program state change or a program state after a program state change is stored as attribute information in a trace buffer. It is a thing.

According to a fifth aspect of the present invention, there is provided a system control management method using error trace, wherein when a storage device such as a write buffer or a cache is used, when an error occurs, the data in the storage device is fixed and the program is stored. The attribute information stored in the trace buffer such as the program state before the state change or the program state after the state change is searched for, and the error source is identified.

An error tracing system control management method according to a sixth aspect of the present invention comprises a structure for cyclically using a trace buffer to store attribute information indicating a change in the state of a program.

According to a seventh aspect of the present invention, there is provided a system control management method using error tracing, which is a pointer to a management table of a program which is created and managed by the operating system according to the usage status of the program or is defined in advance, or the management. It is provided with a structure for storing a copy of a table or attribute information of a program identifier in a trace buffer.

In the system control management method by error tracing according to an eighth aspect of the present invention, the attribute information about the time when the program state change occurs or the elapsed time after the program state change is stored in the trace buffer. It is equipped with a configuration.

In the system control management method based on the error trace according to the ninth aspect of the present invention, the error occurrence time at which the error actually occurs is calculated from the error detection time and the error detection time required to detect the error. Further, by searching attribute programs such as the time when the program state change stored in the trace buffer or the elapsed time after the program state change, the program executed at the obtained error occurrence time is searched. It has a configuration for asking for and certifying the error source.

In the system control management method based on the error trace according to the tenth aspect of the invention, a list of error detection times required to detect an error corresponding to the type of error that has occurred is obtained in advance, and the type of error that has occurred. The error detection time according to
The actual error occurrence time according to the type of error is obtained from the actual error detection time and the error detection time obtained from the list, and the time or program in which the state change of the program stored in the trace buffer as attribute information occurs According to the information of the elapsed time from the change of the state of (1), the program executed at the obtained error occurrence time is searched for and obtained, and the error generation source is identified.

The system control management method by error tracing according to the eleventh aspect of the present invention has a configuration for storing a list of resources used by a program as attribute information in a trace buffer.

According to the twelfth aspect of the present invention, there is provided a system control management method using error tracing, which comprises a structure in which attribute information indicating a change in the state of a program is stored in a trace buffer based on the occurrence of a context switch between processes. It is a thing.

According to a thirteenth aspect of the present invention, there is provided a system control management method based on error tracing, wherein attribute information relating to a change in the state of a program is stored in a trace buffer by a CPU
The configuration is performed based on the transition of the execution state of.

In the system control management method by error tracing according to the fourteenth aspect of the present invention, the attribute information regarding the change of the state of the program in the trace buffer is stored in C
The configuration is such that the execution state of the PU is based on the transition from the user program to the system program or from the system program to the user program.

According to the fifteenth aspect of the present invention, there is provided a system control management method using error tracing, wherein the CPU execution state is changed to a trace buffer based on a transition from a non-privilege level to a privilege level or from a privilege level to a non-privilege level. It is provided with a configuration for storing attribute information regarding a change in the state of a program.

According to a sixteenth aspect of the present invention, there is provided a system control management method based on error tracing, which is used for tracing at a timing at which interrupt processing is started or ended, at a timing at which interrupt processing is started and ended, or at an arbitrary timing during interrupt processing. The buffer is provided with a configuration for storing attribute information relating to a change in the state of the program.

According to a seventeenth aspect of the present invention, there is provided a system control management method based on error tracing, which is an interrupt process which uses an error occurrence at the end of an input / output operation in an input / output device or during an input / output operation, a hardware error or an exception as an interrupt factor. It is provided with a configuration for storing attribute information regarding a change in the state of a program in a trace buffer at the timing of activation or termination, the timing of activation and termination of the interrupt processing, or an arbitrary time timing during the interrupt processing. is there.

According to the eighteenth aspect of the present invention, there is provided a system control management method using error tracing, in which a system call for transition from a user program to a system program is activated or a termination process is executed from the system call. Alternatively, the attribute information relating to the change of the program state is stored in the trace buffer based on the timing at which the system call is started and the timing at which the termination processing from the system call is executed, or the arbitrary timing during the processing of the system call. It is equipped with a configuration for performing.

According to a nineteenth aspect of the present invention, there is provided a system control management method using error tracing, which has a structure for storing attribute information regarding a change in the state of a program as trace information by using an interface provided to an application program. .

According to a twentieth aspect of the present invention, there is provided a system control management method using an error trace, wherein a program counter value,
CPU status value, time element, trace event number, contents of various registers, program name, process name, message,
A program using the interface provided to the application program, using any one of address space identifier, process identifier, program identifier, process attribute information, program attribute information, message queue identifier, message queue attribute information or a combination thereof as an interface parameter. The attribute information about the state change of is stored as trace information.

The system control management method by error tracing according to the twenty-first aspect of the present invention is provided with a structure for storing the attribute information relating to the change of the program state in the trace buffer only when a prespecified factor occurs. It is a thing.

According to a twenty-second aspect of the present invention, there is provided an error trace system control management method, including an interrupt type, a system call type, a program or process identifier,
A configuration that stores the attribute information regarding the program state change in the trace buffer only when a preset factor such as program name, process name, function name, writing to, reading from, or executing a specific address area occurs. It is equipped with.

The system control management method by error tracing according to the twenty-third aspect of the present invention is provided with a structure for storing the attribute information regarding the change of the state of the program in the trace buffer provided in the hardware control circuit. is there.

According to a twenty-fourth aspect of the present invention, there is provided a system control management method using error tracing, which is applied to hardware such as an interface circuit between a CPU and a main memory, an external bus control circuit, a local bus control circuit, an input / output control circuit and an output memory. The trace buffer provided is provided with a structure for storing attribute information regarding a change in the state of the program.

According to a twenty-fifth aspect of the present invention, there is provided a system control management method using error tracing, which stops the execution of the certified program or stops the system, depending on the type of the program certified as the error source. The system is configured to control and manage the system by suspending the execution of the program and resuming and continuing the process when the error cause is recovered.

According to a twenty-sixth aspect of the system control management method by error tracing, when the program identified as the error source is a user program or a user process, the resources used by the user program or the user process are The system is controlled and managed by returning it to the system management means executing the management of the system.

In the system control management method by error tracing according to the twenty-seventh aspect of the present invention, resources such as a memory area or various management tables or input / output devices used by a program recognized as an error generation source are allocated. It has a configuration that controls and manages the system by returning it to the free resource management pool.

According to the 28th aspect of the present invention, there is provided a system control management method using error tracing, wherein a memory page in which an error has occurred in a program recognized as an error source or other resource in which an error has occurred is set as an empty resource. It has a configuration that controls and manages the system by not returning it to the management pool.

According to a twenty-ninth aspect of the present invention, there is provided a system control management method by error tracing, wherein when the type of the program recognized as the error source is a system program such as an operating system, the system is stopped to perform the system stop processing. It has a configuration for controlling and managing the.

According to a thirtieth aspect of the present invention, there is provided a system control management method using error tracing, wherein when the program recognized as an error source is a user program or a user process, the user program or the user process and related programs related thereto. Alternatively, the system is controlled and managed by stopping the processing of the process group and continuously executing other processing.

In the system control management method by error tracing according to the thirty-first aspect of the present invention, the resource which has an error in the program recognized as the error source is replaced with the alternative resource to recover from the error, and the cause of the error is The system is configured to control and manage the system by restarting and continuing the process at the time of recovery.

[0070]

According to another aspect of the present invention, there is provided a system control management method based on error tracing, wherein attribute information indicating a program state change is stored in a trace buffer as trace information, and the stored attribute information indicating a program state change is used. Accurate and easy certification of error sources.

In the system control management method based on the error trace according to the second aspect of the present invention, the attribute information indicating the state change of the program is stored in the trace buffer in the order of occurrence time, and the stored attribute information is advanced or returned in the order of occurrence time. By doing so, it is possible to accurately and easily identify the error source.

In the system control management method based on the error trace according to the third aspect of the present invention, attribute information capable of knowing which resource a process is using is stored in the trace buffer, and this attribute information is searched. By knowing which resource the process uses based on the retrieved attribute information, if the source of the error is the resource owned by the process, an error occurs in the program using the resource. Certified as a source.

In the system control management method by error tracing according to the invention of claim 4, the program state before the change of the program state or the program state after the change is stored in the trace buffer as attribute information,
Accurate and easy recognition of error source by stored attribute information.

In the system control management method based on the error trace of the fifth aspect of the invention, when an error occurs, the data in the storage device such as the write buffer and the cache is fixed, and the program before execution changes the state of the program. The attribute information stored in the trace buffer, such as the state or the program state after the change, is searched to accurately and easily identify the error source.

According to a sixth aspect of the present invention, there is provided a system control management method based on an error trace, in which trace attribute information indicating a change in the state of a program when an error occurs is stored by cyclically using the trace buffer. Accurately and easily certify the error source while achieving effective utilization.

According to a seventh aspect of the present invention, there is provided a system control management method based on error tracing, which is a pointer to a management table of a program which is created and managed by the operating system according to the usage status of the program or which is defined in advance, or the management table. A copy or program identifier is stored as attribute information in the trace buffer, and the stored attribute information is searched to accurately and easily identify the error source.

In the system control management method based on the error trace according to the invention of claim 8, the time when the program state change occurs or the elapsed time after the program state change is stored in the trace buffer as attribute information, The stored attribute information is searched to accurately and easily identify the error source.

According to the ninth aspect of the present invention, there is provided a system control management method based on error tracing, wherein an error detection time, an error occurrence time at which an error actually occurs, which is obtained from the error detection time, and a state of the program stored in the trace buffer. Using the attribute information such as the time when the change occurred or the time elapsed since the state of the program changed, the program that was being executed at the time when the error occurred is searched for and obtained, and the error source is accurately and easily identified. To enable that.

According to a tenth aspect of the present invention, there is provided a system control management method based on error tracing, wherein an error is detected from an error detection time required to detect an error obtained from a list according to the type of error that has occurred and an error detection time. The actual error occurrence time according to the type of the error is obtained, and the error obtained by the attribute information such as the time when the program state change stored in the trace buffer occurs or the elapsed time after the program state changes The program that was being executed at the time of occurrence can be searched and obtained, and the program that is the source of the error can be accurately and easily certified.

In the system control management method by error tracing according to the invention of claim 11, the list of resources used by the program is stored in the trace buffer as attribute information, and the attribute information makes it possible to accurately and easily identify the error source. Allows you to certify.

According to the twelfth aspect of the present invention, there is provided a system control management method based on error tracing, in which attribute information indicating a program state change is stored in a trace buffer based on the occurrence of a context switch between processes, and based on this attribute information. Enables accurate and easy identification of error sources.

According to a thirteenth aspect of the present invention, there is provided a system control management method based on an error trace, in which attribute information indicating a change in the state of a program is stored in a trace buffer on the basis of a transition of an execution state of a CPU. Allows for accurate and easy certification.

According to the system control management method by error tracing in the fourteenth aspect of the present invention, the execution state of the CPU is traced by changing the state of the program from the user program to the system program or from the system program to the user program. By storing the attribute information indicating the program state change in the buffer, it is possible to identify the error source accurately and easily.

In the system control management method by error tracing according to the fifteenth aspect of the present invention, for the tracing, the CPU execution state transitions from the non-privileged level to the privileged level or from the privileged level to the non-privileged level as a change in the state of the program. By storing the attribute information indicating the program state change in the buffer, it is possible to identify the error source accurately and easily.

According to the sixteenth aspect of the present invention, there is provided a system control management method based on error tracing, in which the state of the program is changed to the trace buffer at the timing at which the interrupt processing is activated or terminated, or both, or at an arbitrary time timing during the interrupt processing. By storing the attribute information indicating the, it is possible to identify the error source accurately and easily.

According to a seventeenth aspect of the present invention, there is provided a system control management method based on error tracing, wherein an interrupt process is started when an input / output operation of an input / output device ends or an error occurs during the input / output operation or a hardware error or an exception is an interrupt factor. Alternatively, the attribute information indicating the program state change is stored in the trace buffer at the timing of completion, at the timing of starting and ending the interrupt processing, or at any timing during the interrupt processing, and the error source is accurately identified. And allow easy certification.

According to the eighteenth aspect of the present invention, there is provided a system control management method using error tracing, in which a system call for transition from a user program to a system program is activated or a termination process is executed from the system call, or The attribute information indicating the state change of the program is stored at the timing when the system call is started and the timing when the termination processing from the system call is executed, or at an arbitrary time during the processing of the system call, and the error source is stored. Allows for accurate and easy certification.

In the system control management method by error tracing according to the nineteenth aspect of the present invention, the attribute information of the program is stored in the trace buffer by using the interface provided to the application program, and the error source is accurately and easily identified. To enable that.

A system control management method by error tracing according to the invention of claim 20 is a program counter value, CPU state value, time element, trace event number, contents of various registers, program name, process name, message, address space identifier, Any one or a combination of a process identifier, a program identifier, a process attribute information, a program attribute information, a message queue identifier, a message queue attribute information is used as an interface parameter, and the attribute information of the program is provided by using the interface provided to the application program. Is stored in the trace buffer, and the data necessary for certifying the error source can be collected efficiently to certify the error source accurately and easily.

In the system control management method by error tracing according to the twenty-first aspect of the present invention, the attribute information indicating the program state change is stored in the trace buffer only when a prespecified factor occurs, and the error source is set. Allows for accurate and easy certification.

According to the 22nd aspect of the present invention, there is provided a system control management method based on an error trace, which includes an interrupt type, a system call type, a program or process identifier, a program name or a process name or a function name, and a write or read to a specific address area. , Attribute information indicating program state change is stored in the trace buffer based on occurrence of context switch between processes or transition of CPU execution state when a predetermined factor such as execution occurs, and error occurs Allows the source to be accurately and easily certified.

In the system control management method by error trace according to the twenty-third aspect of the present invention, attribute information indicating a change in the state of the program is stored in the trace buffer provided in the hardware control circuit, and the stored attribute information causes an error. Accurate and easy source identification.

A system control management method by error tracing according to a twenty-fourth aspect of the present invention is provided in hardware such as an interface circuit between a CPU and a main memory, an external bus control circuit, a local bus control circuit, an input / output control circuit, and an output memory. The trace information, which is the attribute information indicating the state change of the program, is stored in the trace buffer,
Accurate and easy identification of error source by the stored trace information.

According to the system control management method by error tracing in the twenty-fifth aspect of the invention, the execution of the certified program is stopped or the system is stopped according to the type of the program certified as the error source.
Alternatively, the execution of the program is postponed, and when the cause of the error is recovered, the execution of the program or the operation of the system is resumed and continued, thereby improving the reliability of the system with respect to the generated error.

According to a twenty-sixth aspect of the present invention, in the system control management method by error tracing, when the program recognized as the error source is a user program or user process, the resources used by the user program or user process are The system is returned to the system management means executing the system management, the utilization rate of effective resources is improved, and the reliability of the system with respect to the generated error is improved.

According to a 27th aspect of the present invention, there is provided a system control management method based on an error trace, wherein when a program identified as an error source is a user program or a user process, a memory area used by the user program or the user process or Resources such as various management tables or input / output devices are returned to the respective free resource management pools, the utilization rate of effective resources is improved, and the reliability of the system against an error that occurs is improved.

In the system control management method based on the error trace according to the twenty-eighth aspect of the present invention, a program identified as an error generation source manages a memory page in which an error has occurred, or other resources in which an error has occurred, by managing each free resource. By not returning the error to the pool, it is possible to prevent the error from occurring due to the reuse of the resource that caused the error and improve the reliability of the system against the error.

According to the 29th aspect of the present invention, in the system control management method based on the error trace, when the type of the program recognized as the error generation source is a system program such as an operating system, the entire system is stopped and an error occurs. Improve system reliability.

According to a thirtieth aspect of the present invention, there is provided a system control management method by error tracing, wherein when the user program or user process is recognized as an error source, the user program or user process and the related program or process related thereto. The processing of the group is stopped and the other processing is continuously executed to improve the reliability of the system with respect to the generated error.

According to a thirty-first aspect of the present invention, there is provided a system control management method based on error tracing, which stops the execution of the certified program according to the program certified as the error generation source, and executes the program certified as the error generation source. By replacing the resource in which the error has occurred with the alternative resource, recovery from the error is performed, and when the cause of the error is recovered, the process is restarted and continued to improve the reliability of the system with respect to the error.

[0101]

【Example】

Example 1. An embodiment of the invention of claim 1 will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a system control management method by error tracing according to this embodiment.
FIG. 9A is a block diagram when the system control management method by the error trace is realized by the system program, and FIG. 11B shows the configuration of the system control management method by the error trace using the trace processing hardware. In FIG. 1A, reference numeral 21 indicates an execution process of a system program or a user program, 22 is a trace buffer in which trace information that is attribute information of the program is written, and 23 is an error determination program for determining an error occurrence source. Shows. Further, in FIG. 9B, 22 executes a process of collecting the attribute information regarding the state change of the program and storing it in the trace buffer 24 by starting the trace process by the user program 21a and the system program 21b. This is the hardware for trace processing.

In the system control management method based on this error trace, the attribute information of the process is stored in the trace buffer memory or the trace buffer area in the memory when the program state change occurs.
When an error occurs, the error determination program recognizes the program that is the source of the error by tracing the attribute information stored in the trace buffer memory or the trace buffer area.

FIG. 2 is a flow chart showing a series of these operations. According to this flowchart, when a program state change occurs (step ST1), the attribute information of the program when the program shows the state change is stored in the trace buffer 24 (step ST).
2). Then, when an error occurrence is detected by the activated error occurrence detection processing function and the error processing function for the detected error occurrence (step ST
3) Next, the cause of the error is investigated (step ST
4). In investigating the cause of this error, the cause of the error is known from the error information, the attribute information of the program is read from the trace buffer 24 (step ST5), and the error source is investigated and certified from this attribute information (step ST).
6). Then, error processing is performed on the recognized error source according to the cause of the error (step ST
7).

As a result, according to the present embodiment, when there is no trace buffer or memory area used as a trace buffer area, a table of all processes or other explicit or implicit system management information is covered. Whereas the error source is identified by performing a manual search, the attribute information related to the program status change is stored in the trace buffer or the memory area used as the trace buffer area. It is possible to efficiently certify the program of the error source.

Example 2. An embodiment of the invention of claim 2 will be described below with reference to the drawings. FIG. 3 shows the attribute information of the program stored in the trace buffer 24,
The program attribute information 1 to n are stored in the trace buffer 24 in the order of the time when the program state change occurs. The attribute information of the programs stored in the trace buffer 24 of the present embodiment is in ascending order (following the sequence of time, the attribute information of the past programs is traced in reverse order) or in descending order (sequential time of the program). Follow the attribute information of the program in the same direction as the flow).

Therefore, by storing the attribute information in the order of time of occurrence, it becomes easy to retrieve new ones in time order or newest to oldest ones according to changes over time, and the error source can be efficiently identified. , It is possible to efficiently certify the program of the error source based on the attribute information of the program.

Example 3. An embodiment of the invention of claim 3 will be described below with reference to the drawings. FIG. 4 is a detailed flowchart of step ST6 of the flowchart of FIG. 2 described in the first embodiment. In the present embodiment, the resource in which an error has occurred is detected, and the program that is the error source is identified from the resource that is the error source.

According to these flowcharts, the resource used by the process is confirmed from the attribute information of the program read from the trace buffer 24 in step ST5 (step ST6a). That is, the attribute information of the program includes information such as resources used by the process. Further, the resource in this case is a memory area or an input / output device used by the program. Next, the resource information confirmed in step ST6a is detected (step ST6b),
It is determined whether or not the error cause that has occurred is caused by the detected resource (step ST6c). When it is determined that the cause of the error that has occurred is not caused by the detected resource, the process returns to step ST6a,
The processing from step ST6b to step ST6c
The process is repeated until it is determined that the cause of the error that has occurred is from the detected resource. Step ST6
When it is determined that the cause of the error that occurred in c is caused by the detected resource, the process is identified as the source of the error (step ST6).
d).

Therefore, also in this embodiment, it is possible to efficiently certify the program of the error source based on the attribute information of the program.

Example 4. An embodiment of the invention of claim 4 will be described below with reference to the drawings. FIG. 5 shows the attribute information of the program stored in the trace buffer 24 of this embodiment. The attribute information 1 to n of the program are stored in the trace buffer in the order of, for example, the time of occurrence of a change in the state of the program. The stored attribute information is the attribute information of the immediately preceding program when the state change occurs. In this case, the program attribute information stored in the trace buffer 24 may be searched in ascending or descending order according to the flow of the program generation time.

FIG. 6 also shows the attribute information of the program stored in the trace buffer 24. The attribute information 1 to n of the program is stored in the trace buffer 24 in the order of, for example, the time of occurrence of a change in the state of the program. The stored attribute information is the attribute information of the program after the state change when the state change occurs. in this case,
The program attribute information stored in the trace buffer may be searched in ascending or descending order according to the flow of program generation time.

Therefore, in this embodiment, the program of the error occurrence source is efficiently identified based on the attribute information before or after the state change of the program stored in the trace buffer 24 due to the state change of the program. It becomes possible to do.

Example 5. An embodiment of the invention of claim 5 will be described below. In the system control management method by error trace of the present embodiment, when the hardware configuration has a write buffer or cache,
When an error occurs, the write buffer is frozen and the trace buffer is searched. However, here, the cache or the write buffer is assumed to operate in the FIFO method. In other words, if the state of a new program after the state change is stored in the trace buffer, the state of the program changes immediately after the operation that causes the error occurs, and the CPU changes to the new state. When an error is detected, the CPU writes the state change to the trace buffer. At this time, the write buffer still has the operation to write the state change to the trace buffer. Therefore, as shown in the flowchart of FIG. At this point, the write buffer is temporarily frozen to stop the operation of writing the state change to the trace buffer. Then, the attribute information stored in the trace buffer is searched and checked to facilitate the recognition of the error occurrence source. In this case, the attribute information stored in the trace buffer may be searched in ascending order or descending order described in the second embodiment when the attribute information is stored in the order of occurrence time.

Example 6. An embodiment of the invention of claim 6 will be described below with reference to the drawings. FIG. 8 is an explanatory diagram showing the configuration of the trace buffer in the system control management method by error tracing according to this embodiment. In the figure, 31 is the attribute information stored most recently in the ring-shaped trace buffer, and 32 to 33 are attribute information stored in the ring-shaped trace buffers in the order of oldness.

In the present embodiment, the trace buffer or trace buffer area is formed into a ring ring structure, and the trace buffer or trace buffer area is used cyclically. As a result, it is possible to improve the usage efficiency of the trace buffer by repeatedly using the buffer that stores the oldest information when the ring buffer is full without setting the trace buffer or trace buffer area to a large value. Becomes

Further, instead of configuring the ring buffer by the connection information by the pointer, it is also possible to configure the ring buffer by using the index in the array to make the array configuration into a ring shape. It is also possible to store a flag indicating whether or not the information stored in each buffer constituting the ring buffer is valid and information indicating the storage time.

Example 7. As the attribute information described in the first to sixth embodiments, as shown in FIG. 9, a pointer to the management table of the program, a part or all of the copy of the management table, or a process for identifying the process The program identifier such as the ID or the task ID for identifying the task may be stored in the trace buffer together with other information.

Further, it is possible to manage this information for each resource at the bit position of the management variable area.

Example 8. Moreover, the said Example 1-Example 6
As the attribute information described above, the time at which the program state change occurs or the elapsed time after the program state change may be stored in the trace buffer. It is explanatory drawing for demonstrating the characteristic of one Example of the system control management method by the described error trace. In the figure (a), the error occurrence time T1 and the error that has occurred since the error occurred are C
An explanatory diagram showing a time T2 and the like until it is recognized by the PU on a time axis, and FIG. 8B is an explanatory diagram showing transitions of process executions that are simultaneously proceeding.

The operation of this embodiment will be described with reference to FIG. The trace buffer stores, as attribute information, the time at which a program state change occurs or the elapsed time after the program state change. As described above, when the error occurrence time is T1 and the time required from the occurrence of the error until the error is recognized by the CPU is T2, as shown in (a) of FIG. It can be recognized that the program executed at the time indicated by (T1-T2) has caused an error. Therefore, from the explanatory diagram showing the transition of the process execution shown in (B) of FIG. 10, since the user program is being executed at time Te, it can be determined that the error source is the user program.

Example 9. In addition, when the time T2 required from the occurrence of the error until the CPU recognizes the generated error depends on the type of the generated error, until the CPU recognizes the error for each type of error in advance. It is also possible to determine the time T2 required for the calculation and make a table as shown in FIG. 11, and read an appropriate occurrence error recognition time T2 according to the type of error that has occurred to accurately identify the error occurrence source.

Example 10. Further, as shown in FIG. 12, by storing a list of programs and resources used by the programs in the trace buffer, it is possible to search the trace buffer for a program using the resource that has caused the error. Then, it is possible to accurately and easily set the error source program.

Example 11. It is also possible to employ a process context switch as the timing for storing the program attribute information in the trace buffer. In this embodiment, the timing at which one process is switched to another process in step ST2 of FIG. To store the attribute information in the trace buffer. As a result, information about the execution order of processes will be stored as attribute information in the trace buffer, and when an error occurs, the attribute information stored in the trace buffer will be searched before the error occurs. It is possible to easily know the execution order of the processes that have been executed, and it is possible to easily identify the process that was being executed when the error occurred.

The above description will be more specifically described with reference to FIG. 10. Every time a process execution transition (process A.fwdarw.process B.fwdarw.process C) is performed, the process attribute information is changed to time information or time information. When it is determined that the actual error occurrence time is Te, the process executed at time Te based on the attribute information stored together with the time information is process B when the actual error occurrence time is found to be Te. Is easily certified.

Example 12. Hereinafter, claim 13 to claim 1
An embodiment of the invention of No. 8 will be described. In this embodiment,
In step ST2 of FIG. 2, each time the execution level of the CPU changes, the change in the state of the program is stored in the trace buffer. That is, by storing the program attribute information in the trace buffer each time a system call, privilege level change, or interrupt occurs, it is possible to identify the error source as to whether the error source is a user program, system program, or interrupt program. It will be easier.

In FIG. 10, (S) during the execution transition of the process is the operating state of the CPU by the system program,
(U) shows the operating state of the CPU according to the user program. By storing the execution operation state of the CPU in the trace buffer together with the time information, it can be easily determined that the CPU is in the execution state by the user program at time Te and the source of the error at time Te is in the user program. .

Furthermore, by storing the change in the state of the program at two timings in the trace buffer, the execution history of the program on the time axis (that is, process A "user state"-> process A "system state" → process B “system state” → process B “user state” execution history),
It is possible to certify the error source more reliably.

When the program attribute information is stored in the trace buffer each time an interrupt or the like occurs, the program attribute information is displayed each time an interrupt occurs, which causes an input / output interrupt, an exception, a hardware error, or the like. The attribute information of the program may be stored in the trace buffer, and further, the attribute information of the program may be stored in the trace buffer at the timing at which these interrupt processes are activated or ended, or both, or at any timing during the interrupt process.

Example 13 An embodiment of the inventions of claims 19 and 20 will be described below with reference to the drawings. FIG.
FIG. 3 is a conceptual diagram showing a configuration of a system control management method by error tracing according to this embodiment. In the present embodiment, a software interface by an application program for explicitly storing attribute information in a trace buffer from a user program is provided. In the figure, 41 is a user program, 42 is a system program, 43 is an error judgment program for recognizing the error source based on the attribute information stored in the trace buffer, and 44 is a trace buffer for explicitly displaying attribute information from the user program. Is a software interface with an application program for storing in.

In this embodiment, a software interface for explicitly storing the attribute information in the trace buffer is provided from the user program, and this software interface allows the user program 41 to explicitly store the attribute information in the trace buffer 24. Is stored. In this case, the user program becomes clearly aware of storing the attribute information in the trace buffer 24 via the software interface 44.

Further, as the parameters of the software interface, the value of the program counter, the state value of the CPU, the time element, the trace event number, the contents of various registers, the program name or the process name, the message, the address space identifier, the process identifier, the program There are identifiers, process attribute information, program attribute information, message queue identifiers, message queue attribute information, and the like, and these pieces of attribute information are stored in the trace buffer 24 via the software interface 44.

As a result, the value of the program counter,
CPU status value, time element, trace event number, contents of various registers, program name or process name, message, address space identifier, process identifier, program identifier, process attribute information, program attribute information, message queue identifier, message queue attribute The storage of the attribute information such as information in the trace buffer 24 can be controlled by software, and the data effective for certifying the error occurrence source can be efficiently stored in the trace buffer 2.
It will be possible to collect in 4.

Example 14 An embodiment of the inventions of claims 21 and 22 will be described below. Although the eleventh and twelfth embodiments have been described as the timing of storing the attribute information of the program in the trace buffer each time the context switch of the process or the execution level of the CPU changes, a prespecified factor occurs. The attribute information may be stored in the trace buffer only when this is done. In this case, the factors to be specified in advance are interrupt type, system call type, process or program identifier, process name or program name or function name. The read / write and execution of a specific address area can be the factors.

Example 15. An embodiment of the inventions of claims 23 and 24 will be described below. In this embodiment, the trace buffer is set in the hardware control circuit, and the control circuit is an interface circuit between the CPU and the main memory, an external bus control circuit, a local bus control circuit, or an input / output control circuit. Also, a trace buffer area is set in a predetermined area of the main memory.

FIG. 14 is a conceptual diagram showing the structure of a system control management method by error tracing according to this embodiment. In the figure, 51 is a CPU, 52 is a CPU 51 and a main memory 5.
3, 54 is an external bus control circuit, 55 is a local bus control circuit, and 56 is an input / output control circuit.

FIG. 15 shows an interface 52 (hereinafter, PMI, Process) between the main memory 53 and the CPU 51.
or Memory Interface) is a conceptual diagram when a trace buffer (region) is provided. In FIG. 15, the same or corresponding portions as those in FIG. 14 are designated by the same reference numerals and the description thereof is omitted, but 60 is a cache memory, 61 is a write buffer, and 62 is a trace buffer (area).

When the CPU 51 writes data to the outside via the PMI 52, the CPU 51 performs writing by using the cache memory 60 and the write buffer 61, thereby realizing high-speed writing processing. That is, the CPU 51
In the external write operation, if the write target is an area to be cached, it is once stored in the cache memory 60, and then queued in the write buffer 61. The CPU 51 shifts the control to another operation at the stage when the writing to the cache memory 60 is completed, and then the write operation is executed. Therefore, the CP
U51 is not affected by the processing delay due to writing. The contents of the cache memory 60 are written to the main memory 53 or the like via the write buffer 61.
Assuming that the write buffer 61 and the cache memory 60 of the CPU 51 are executing the write processing by the FIFO method, the write operation of the actual CPU 51 and the actual physical write are performed because of the write buffer 61 and the cache memory 60 of the CPU 51. Are not performed synchronously, and while speeding up the processing is realized, there is a time lag between the occurrence of an error and the detection of the error that has occurred, making it difficult to detect the error. Therefore, the problem caused by the asynchronous nature of the error trace processing of the computer system during the write processing is solved by setting the trace buffer in the control circuit of the specific hardware.

In the present embodiment, when the program state changes, the new program state after the state change is stored in the trace buffer 62. For example, there is a change in the state of the program immediately after the error occurrence operation is performed, and the CPU 5
Assume that 1 has detected an error after transitioning to a new state.

That is, (1) the CPU 51 performs the operation of causing an error, but the actual write operation of the state change trace buffer 62 is stored in the write buffer 61. (2) The state of the program further transits. Although the operation of writing the state transition to the trace buffer 62 is performed, the actual write operation of the state change to the trace buffer 62 is accumulated in the write buffer 61. (3) Since the write buffer 61 is a FIFO, the operation that causes an error is executed first and the error is detected. The operation of writing the state change to the trace buffer 62 when the above processes (1), (2), and (3) proceed in order must be left in the write buffer 62 when an error is detected by (3). become. So in this case,
An error detected by temporarily freezing the write buffer 61, fixing the behavior of the program state change in the write buffer 61, and checking the attribute information about the program state change stored in the trace buffer 62 The source can be easily certified.

FIG. 16 is an explanatory diagram showing a state of the write buffer 61 before an error occurs, and 25 is error write operation information of a write operation that causes an error.
Reference numeral 26 is state change write operation information for performing a write operation in the trace buffer 62 regarding a change in the state of the program. In this case, a write operation that causes an error is performed according to the error write operation information 25, and when an error occurs, the write buffer 61 is temporarily frozen,
The write operation to the trace buffer 62 is prevented from being executed with respect to the subsequent change in the state of the program. Therefore, the state before the error remains in the trace buffer 62 when the error occurs, and it is possible to easily identify the error source by the attribute information remaining in the trace buffer 62 when the error occurs.

Next, the method of recognizing the error occurrence source in this case will be specifically described with reference to FIGS. 17, 18 and 23. In FIG. 23, S indicates that the system program is being executed, and U indicates that the user program is being executed. As described in the related art, it is not possible to correctly certify the error source for the pairs (6, 11), (8, 13) and (9, 14) of error occurrence and error detection. On the other hand, in the present embodiment, since the process at the time of error detection is different for the set of error occurrence and error detection (9, 14), an error occurs by writing the process identifier in the attribute information of the trace buffer 62. The original certification is possible.

FIG. 17 is an explanatory diagram showing a correspondence table for recognizing the error source in other cases. That is, in FIG. 23, at the time of error detection 11 and at the time of error detection 13, the request to update the attribute information of the program regarding the occurrence of errors 6 and 8 remains in the write buffer 61, so check the trace buffer 62. It is possible to know that the program state of the error detected at the time of error detection 11 is the user state and the program state of the error detected at the time of error detection 13 is the system state.

In the case of error detection 10 and error detection 12, since the attribute information of the trace buffer 62 and the state of the CPU at the time of error detection match, the error occurrence source is determined by the attribute information of the trace buffer 62. It is possible to know the state of the CPU. Therefore, as shown in FIG. 17, it is clear that the error source at error detection 10 and error detection 11 is in the user program, and the error source at error detection 12 and error detection 13 is in the system program. Become. In other words, the combination of error occurrence and error detection that could not be correctly recognized in the past (6, 11)
With respect to (8, 13), the error source is correctly identified.

The same effect can be obtained by storing the program state before the program state is changed in the trace buffer 62 when the program state is changed. FIG. 18 is a correspondence table when the program state before the change of the program state is stored in the trace buffer 62 when the state of the program changes, and an error similar to the case shown in the correspondence table of FIG. The error occurrence source at the time of detection 10 and the error detection 11 is in the user program state after the change of the system program state, and the error source at the time of error detection 12 and the error detection 13 is the system program after the change of the user program state It becomes clear that it is in a state. That is, with respect to the error occurrence and error detection pairs (6, 11) and (8, 13), which could not be correctly recognized in the past, the error source is correctly recognized.

Example 16. Hereinafter, claim 25 to claim 3
An embodiment of the invention described in 1 will be described. The present embodiment describes a process executed when the error source is identified by the first to fifteenth embodiments described above. Regarding the processing based on the recognition of the error occurrence source, it is possible to erase only the program in which the error has occurred from the system and continue execution of other processing, or it is possible to stop the entire system. For example, if a user process is the source of an error, all resources used by that program are released to the free resource pool, and if a system program is the source, the entire system is stopped. Alternatively, it is possible to move all or some of the currently running programs to another system.

Further, there is a method such as stopping the execution of the program which caused the error and continuing the execution of the other programs, or postponing the execution process of the error and continuing the execution when the cause of the error is recovered. It is possible. That is, the cause of the error is an I / O error,
If the program cannot continue running without the I / O device, stop the program, and if an alternate I / O device is available later, wait until the alternate I / O device becomes available. Suspend and suspend the execution of the program that caused the error.

Further, it is possible to improve the utilization rate of effective resources by restoring the resources used by the program that has generated the error.

Furthermore, it is possible to prevent the recurrence of the error by not returning the resource that caused the error to the free resource management pool. In other words, the cause of the error is a memory failure, and the main memory used by the program that has not failed is returned to the system as free memory. Do not return to, and prohibit use until the cause of the error is recovered. If the resource that caused the error can be replaced with a substitute resource, do not return it to the free resource management pool as an unusable resource, and improve system reliability by prohibiting use until the cause of the error is recovered. Is possible. In other words, if an error occurs due to a memory device failure and the operation at the time of error occurrence is a write process to the memory device, another memory is prepared as an alternative memory and the write process is performed using that memory. By doing this, error recovery becomes possible.

FIG. 19 is a flow chart showing the operation after recognizing the error source in the embodiment of the invention of claim 25, claim 29 or claim 30, and FIG.
FIG. 21 is a flowchart showing the operation after recognizing the error source in the embodiment of the invention described in claim 26 or claim 27. FIG. 21 is the operation after recognizing the error source in the embodiment of the invention described in claim 28. A flow chart showing
FIG. 22 is a flow chart showing the operation after recognizing the error source in the embodiment of the invention of claim 31.

That is, according to the operation shown in the flowchart of FIG. 19, a program of the same type as the type of program (type of system program or user program) recognized as the error source is identified (step ST11). Then, it is determined whether or not the identified program type is a system program (step ST12), and when it is determined to be a system program, the entire system is stopped (step ST1).
3). On the other hand, when it is determined in step ST12 that the program is not the system program, that is, the user program, the execution of the user program corresponding to the program specified in step ST11 is stopped (step ST14). In this state, the user program corresponding to the specified program is in the stopped state, but the other programs that are not the source of the error continue to be executed without stopping, and the operation for recovering the cause of the error is performed.

In the state where the user program is stopped, it is next determined whether or not the cause of the error is recovered (step ST15). When it is determined that the cause of the error is recovered, the stopped user program The execution is resumed and continued (step ST16).

Further, according to the operation shown in the flow chart of FIG. 20, a program of the same type as the type of the program (system program or user program) recognized as the error source is identified (step In ST21), the resources used by the further specified program are confirmed (step ST22). The resources include a memory area used by a program that is identified as the source, various management tables, an input / output device, and the like. Then, all or part of the resources confirmed in step ST22 is returned to the resource management system managing the resources (step ST23).

Further, in the flow chart shown in FIG.
The operation of returning the resource to the resource management system managing the resource in step ST23 shown in the flowchart of FIG. 20 is not performed (step S
T33), the resource causing the error is prevented from being returned to the resource management system and used again to cause the error again.

Further, according to the operation shown in the flow chart of FIG. 22, when the program which is the source of the error is certified, the error processing is started and the execution of the certified program is stopped (step ST41). Then, the resource used by the program recognized as the source of the error is confirmed (step ST42), and it is determined whether or not there is an alternative resource replacing this resource (step ST4).
3). When it is determined that there is no alternative resource, and it is determined that the subsequent processing cannot be continued unless the alternative resource is used,
An interruption process for interrupting the process is performed (step ST44).
On the other hand, when it is determined in step ST43 that there is an alternative resource, the resource that causes the error is replaced with the alternative resource (step ST45). Then, necessary setting data and the like are reset so that the state of the newly replaced alternative resource becomes the same as the state of the resource before the replacement (step ST46), and the execution of the program interrupted in step ST41 is started. Continue (step ST47).

[0155]

As described above, according to the first aspect of the invention, the trace buffer stores the trace information which is the attribute information relating to the change in the state of the program, and the error source is identified based on the stored attribute information. I configured it to
Accurate and easy recognition of the error source is effective.

According to the second aspect of the present invention, the attribute information relating to the state change of the program being executed in the trace buffer is stored as trace information in the order of occurrence time, and the stored attribute information is advanced or returned in the order of occurrence time. Since the program related to the cause of the error occurrence is certified by searching, it is possible to accurately and easily identify the error source.

According to the third aspect of the present invention, the attribute information of the process, which is stored in the trace buffer and which can be used to know which resource the process is using, is searched, and based on this attribute information, the process is searched. Since the program that uses the resource that has caused the error has been configured to be recognized by knowing the resource that is being used, the error source can be accurately and easily identified.

According to the fourth aspect of the present invention, the trace buffer stores the program state before the state of the program being executed or the program state after the change as attribute information, and the stored attribute information. Since the error source is identified based on the above, there is an effect that the error source is identified accurately and easily.

According to the fifth aspect of the invention, when an error occurs while using a storage device such as a write buffer or a cache, the state of the program being executed while fixing the data in the storage device changes. The attribute information stored in the trace buffer, such as the program state before the change or the program state after the change, is searched and the source of the error is configured to be certified, so that the source of the error is accurately and easily certified. effective.

According to the sixth aspect of the present invention, the attribute information regarding the state of the program at the time of error is stored by using the trace buffer cyclically, and the error source is identified based on the stored attribute information. With this configuration, the trace buffer can be effectively used, and the error source can be accurately and easily identified.

According to the invention of claim 7, a pointer to a management table of a program which is created and managed by the operating system according to the usage status of the program or is defined in advance, or a copy of the management table or an identifier of the program, etc. Is stored as the attribute information in the trace buffer and the error source is identified based on the stored attribute information. Therefore, the error source can be identified accurately and easily.

According to the eighth aspect of the present invention, the time when the program state change occurs or the elapsed time after the program state change is stored in the trace buffer as attribute information, and the stored attribute information is used as a basis. Since it is configured to certify the error source, the error source can be accurately and easily certified.

According to the ninth aspect of the invention, the error occurrence time at which an error actually occurs is obtained from the error detection time and the error detection time required to detect the error that has occurred, and is further stored in the trace buffer. Based on the attribute information such as the time when the status change of the program occurred or the elapsed time since the status of the program changed, the program that was executed at the time when the error occurred was searched for and found, and the error source was identified. Since it is configured to do so, it is effective in certifying the error source accurately and easily.

According to the tenth aspect of the invention, a list of error detection times required to detect an error corresponding to the type of error that has occurred is obtained in advance, and the error detection time corresponding to the type of error that has occurred is obtained. From the list, the actual error occurrence time corresponding to the type of error is found from the error detection time and the error detection time found from the list, and the state of the program stored in the trace buffer as attribute information Since it is configured to search for the program that was being executed at the obtained error occurrence time by the information of the time when the change occurred or the time elapsed since the state of the program changed, and to determine the error source. Accurate certification of the error source even when the program state when the error occurs and the program state when the error occurs are different One is easy to become effective.

According to the invention of claim 11, the list of resources used by the program is stored in the trace buffer as attribute information, and the error source is identified based on this attribute information. Therefore, it is effective to accurately and easily identify the error source.

According to the twelfth aspect of the present invention, the attribute information is stored in the trace buffer based on the occurrence of the context switch between processes, and the error source is identified by the attribute information stored in the trace buffer. Since it is configured as described above, it is possible to easily know the execution order of the processes, and it is possible to accurately and easily identify the error source.

According to the thirteenth aspect of the present invention, the attribute information is stored in the trace buffer based on the transition of the execution state of the CPU, and as a result, the error occurrence source is determined based on the attribute information stored in the trace buffer. Since it is configured to certify, the effect of certifying the error source is accurate and easy.

According to the fourteenth aspect of the present invention, the attribute information is stored in the trace buffer based on the transition of the CPU execution state from the user program to the system program or from the system program to the user program. Since the error source is configured to be certified based on the attribute information related to the program state change stored in, the error source can be certified even when the program state when an error occurs and the program state when an error is detected are different. It has the effect of being accurate and easy.

According to the fifteenth aspect of the present invention, the attribute information is stored in the trace buffer when the execution state of the CPU transits from the non-privileged level to the privileged level or from the privileged level to the non-privileged level as the state change of the program. Then
Since the error source is identified based on the attribute information stored in the trace buffer, the error source can be identified accurately and easily.

According to the sixteenth aspect of the present invention, the attribute information is stored in the trace buffer at the timing when the interrupt processing is started or ended, the timing when the interrupt processing is started and ended, or an arbitrary time during the interrupt processing. Since the error source is identified based on the attribute information stored in the trace buffer, the error source can be identified accurately and easily.

According to the seventeenth aspect of the present invention, when an input / output operation in the input / output device is ended or during an input / output operation, an error occurrence, a hardware error or an exception is used as an interrupt factor, and when these interrupt factors occur, tracing is performed. Since the attribute information is stored in the buffer and the error source is identified based on the attribute information stored in the trace buffer, even if the program state when an error occurs and the program state when an error is detected are different. Accurate and easy recognition of the error source is effective.

According to the eighteenth aspect of the present invention, the timing at which the system call for transitioning from the user program to the system program is activated, the timing for performing the termination processing from the system call, or the system call is activated. The attribute information is stored in the trace buffer at the timing when the system call is terminated and when the system call is terminated, or at any time during the processing of the system call, and based on the attribute information stored in the trace buffer. Since the error source is configured to be certified, even if the program state at the time of error occurrence and the program state at the time of error detection are different, the error source can be certified accurately and easily.

According to the nineteenth aspect of the invention, since the attribute information is stored using the interface provided to the application program and the error occurrence source is identified based on the stored attribute information, an error occurs. This has the effect of making the source identification accurate and easy.

According to the twentieth aspect of the invention, the program counter value, the CPU state value, the time element, the trace event number, the contents of various registers, the program name, the process name,
The interface provided to the application program is used with the message, the address space identifier, the process identifier, the program identifier, the process attribute information, the program attribute information, the message queue identifier, the message queue attribute information, or a combination thereof as the interface parameter. The attribute information is stored by using the attribute information, and the error source is certified based on the stored trace information. Effective data for certifying the error source can be collected efficiently, and the error source can be certified accurately and easily. Is effective.

According to the twenty-first aspect of the present invention, the attribute information is stored in the trace buffer only when a prespecified factor occurs, and the error occurrence source is identified based on the stored attribute information. Since it is configured, it has the effect of certifying the error source accurately and easily.

According to the twenty-second aspect of the invention, the type of interrupt, the type of system call, the identifier of the program or process, the program name or the process name or the function name, and the writing, reading, and execution in a specific address area are designated in advance. The attribute information is stored in the trace buffer only when the specified factor occurs, and the error source is identified based on the stored attribute information, so the error source can be identified accurately and easily. Is effective.

According to the twenty-third aspect of the present invention, the attribute information indicating the state change of the program is stored in the trace buffer provided in the hardware control circuit, and the error source is identified based on the stored attribute information. With this configuration, even if the program state at the time of error occurrence and the program state at the time of error detection are different, it is possible to accurately and easily identify the error source.

According to the twenty-fourth aspect of the present invention, the interface circuit between the CPU and the main memory, the external bus control circuit, the local bus control circuit, the input / output control circuit, and the output memory are used as hardware control circuits, and these hardware are used. Since the attribute information is stored in the trace buffer provided and the error source is identified based on the stored attribute information, the error source can be identified accurately and easily.

According to the twenty-fifth aspect of the invention, the execution of the certified program is stopped, the system is stopped, or the execution of the program is postponed according to the type of the program recognized as the error source. When the cause of the error is recovered, the program is restarted and the operation of the system is resumed at the time of recovery, so that
There is an effect that a system control management method by error tracing that improves system reliability can be obtained.

According to the twenty-sixth aspect of the present invention, when the program recognized as the error source is a user program or a user process, the resources used by the user program or the user process are managed by the system. Since it is configured to control and manage the system by returning it to the existing system management means, there is an effect that a system control management method by error tracing that improves the utilization rate of effective resources and improves system reliability can be obtained. .

According to the twenty-seventh aspect of the present invention, the resources such as the memory area used by the program identified as the error source, the various management tables, and the input / output devices are returned to the respective free resource management pools. Since the system is configured to be controlled and managed by the system, there is an effect that a system control management method by error tracing that improves the utilization rate of effective resources and improves system reliability can be obtained.

According to the twenty-eighth aspect of the invention, the memory page in which an error has occurred in the program identified as the error source or the resource in which another error has occurred are not returned to the respective free resource management pools. Since the system is configured to be controlled and managed, it is possible to obtain a system control management method by error tracing that improves the reliability of the system by preventing the occurrence of errors repeatedly.

According to the twenty-ninth aspect of the invention, when the type of the program recognized as the error generation source is a system program such as an operating system, the system is configured to execute the stop processing of the entire system. There is an effect that a system control management method by error tracing that controls the system according to the source and improves the reliability of the system can be obtained.

According to the thirtieth aspect of the invention, when the program recognized as the error source is the user program or the user process, the processing of the user program or the user process and the related program or process group related thereto is stopped. Since the other processing is configured to be continuously executed, the system control management method by the error trace that controls the system according to the recognized error source and improves the system reliability can be obtained.

According to the thirty-first aspect of the invention, the error is recovered by replacing the resource in which an error has occurred in the program recognized as the error source with the alternative resource, and the process is restarted when the cause of the error is recovered. Since it is configured to continue the operation, there is an effect that a system control management method by error tracing that improves system reliability can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a system control management method by error tracing according to an embodiment of the invention described in claim 1.

FIG. 2 is a flowchart showing an operation of a system control management method by error tracing according to an embodiment of the invention described in claim 1.

FIG. 3 is an explanatory diagram showing attribute information of programs stored in a trace buffer in order of occurrence time in a system control management method by error tracing according to an embodiment of the invention as set forth in claim 2;

FIG. 4 is a flow chart showing an operation of error source investigation / authorization processing in a system control management method by error tracing according to an embodiment of the invention as set forth in claim 3;

FIG. 5 is an explanatory diagram showing attribute information of a program stored in a trace buffer in a system control management method by error tracing according to an embodiment of the invention described in claim 4;

FIG. 6 is an explanatory diagram showing attribute information of a program stored in a trace buffer in a system control management method by error tracing according to an embodiment of the invention described in claim 4;

FIG. 7 is a flow chart showing an operation of a system control management method by error tracing according to an embodiment of the invention described in claim 5;

FIG. 8 is an explanatory diagram showing a configuration of a ring-shaped trace buffer in a system control management method by error tracing according to an embodiment of the present invention.

FIG. 9 is an explanatory diagram showing attribute information stored in a trace buffer of a system control management method by error tracing according to an embodiment of the invention described in claim 6;

FIG. 10 is an explanatory diagram for explaining the features of the system control management method based on error tracing according to an embodiment of the invention described in claims 8 and 9;

FIG. 11 is an explanatory diagram showing a list of error detection times in a system control management method by error tracing according to an embodiment of the invention as set forth in claim 10;

FIG. 12 is an explanatory diagram showing a list of resources used in a system control management method by error tracing according to an embodiment of the invention described in claim 11;

FIG. 13 is a conceptual diagram showing a configuration of a system control management method using error tracing according to an embodiment of the invention described in claims 19 and 20.

FIG. 14 is a conceptual diagram showing the configuration of a system control management method by error tracing according to an embodiment of the invention described in claims 23 and 24.

FIG. 15 is a conceptual diagram in the case where a trace buffer is provided in the PMI 52 of the configuration of the system control management method by error trace according to the embodiment of the invention described in claims 23 and 24.

FIG. 16 is an explanatory diagram showing a state of a write buffer before an error occurs in the configuration of the system control management method by error tracing according to the embodiment of the invention described in claims 23 and 24;

FIG. 17 shows a correspondence table for knowing an error occurrence source from the state of the program stored in the trace buffer in the configuration of the system control management method by error trace according to the embodiment of the invention described in claims 23 and 24. FIG.

FIG. 18 shows a correspondence table for knowing an error occurrence source from the state of the program stored in the trace buffer in the configuration of the system control management method by error tracing according to the embodiment of the invention described in claims 23 and 24. FIG.

FIG. 19 is a flow chart showing an operation after recognizing an error source in one embodiment of the invention of claim 25, claim 29 or claim 30.

FIG. 20 is a flowchart showing an operation after recognizing an error source in one embodiment of the invention of claim 26 or claim 27.

FIG. 21 is a flowchart showing an operation after recognizing an error occurrence source in one embodiment of the invention of claim 28.

FIG. 22 is a flow chart showing an operation after recognition of an error source in one embodiment of the invention of claim 31;

FIG. 23 is an explanatory diagram schematically showing an error occurrence situation in a conventional computer system over time.

[Explanation of symbols]

24, 62 trace buffer, 44 software interface, 51 CPU, 53 main memory, 5
4 external bus control circuit, 55 local bus control circuit,
56 input / output control circuit, 60 cache memory, 61
Write buffer.

Claims (31)

[Claims] 1. A system control management method using an error trace, wherein trace information at the time of error occurrence is stored in a trace buffer, an error generation source is identified by the stored trace information, and a system is controlled and managed. A system control management method by error tracing, wherein attribute information relating to a change in the state of a program is stored as trace information in a buffer for use, and an error generation source is identified based on the stored attribute information. 2. The trace buffer stores attribute information relating to a change in the state of a program being executed as trace information in the order of occurrence time, and the stored trace information is searched for forward or backward in the order of occurrence time, resulting in an error. 2. The system control management method by error trace according to claim 1, wherein a program related to the cause of occurrence is certified. 3. Retrieval of attribute information that allows the process stored in the trace buffer to know which resource the process is using, and based on this attribute information, retrieves the resource used by the process. The system control management method by error trace according to claim 1, wherein the program that uses the resource that caused the error is recognized by knowing. 4. The error trace according to claim 1, wherein the trace buffer stores the program state before the state of the program being executed changes or the program state after the change as attribute information. System control management method. 5. When a storage device such as a write buffer or a cache is used, when an error occurs, the data in the storage device is fixed and the state of the program being executed or change before the state of the program being executed changes. 2. The system control management method by error trace according to claim 1, wherein the attribute information stored in the trace buffer such as the program state after the search is searched to identify the error source. 6. The system control management method by error trace according to claim 1, wherein the trace buffer is cyclically used to store the attribute information when an error occurs. 7. A trace buffer as a pointer to a management table of a program which is created and managed by the operating system according to the usage status of the program or is defined in advance or a copy of the management table or an identifier of the program as attribute information. The system control management method by error trace according to claim 1, wherein 8. The error tracing system according to claim 1, wherein the trace buffer stores attribute information relating to a time when the program state is changed or an elapsed time after the state of the program is changed. Control management method. 9. The error occurrence time at which an error actually occurs is determined from the error detection time and the error detection time required to detect the error, and the status of the program stored in the trace buffer changes. The error source is identified by searching for the program that was being executed at the obtained error occurrence time, based on the attribute information such as the time when the error occurred or the time elapsed since the state of the program changed. A system control management method based on the error trace according to item 1. 10. A list of error detection times required to detect an error corresponding to the type of error that has occurred is obtained in advance, and the error detection time corresponding to the type of error that has occurred is searched from the list. The error detection time and the error detection time obtained from the above list are used to obtain the actual error occurrence time according to the type of error, and the time when the state change of the program stored in the trace buffer occurs or the state of the program is 2. The error trace source according to claim 1, wherein the program that was executed at the obtained error occurrence time is searched and obtained by attribute information such as the elapsed time since the change, and the error occurrence source is identified. System control management method. 11. The system control management method according to claim 1, wherein a list of resources used by the program is stored as attribute information in the trace buffer. 12. The system control management method according to claim 1, wherein the attribute information is stored in the trace buffer based on the occurrence of a context switch between processes. 13. The system control management method according to claim 1, wherein the attribute information is stored in the trace buffer based on the transition of the execution state of the CPU. 14. The transition of the execution state of the CPU is a transition of the execution state of the CPU from a user program to a system program or from a system program to a user program as a state change of the program. System control management method by error trace of the. 15. The transition of the execution state of the CPU is a transition of the execution state of the CPU from a non-privileged level to a privileged level or from a privileged level to a non-privileged level as a state change of a program. 13. A system control management method based on the error trace described in 13. 16. The transition of the execution state of the CPU is a timing at which the interrupt processing is activated or terminated, a timing at which the interrupt processing is activated and terminated, or an arbitrary time timing during the interrupt processing. 13. A system control management method based on the error trace described in 13. 17. The error trace according to claim 16, wherein the interrupt processing uses an error occurrence at the end of the input / output operation in the input / output device or a hardware error or an exception during the input / output operation as an interrupt factor. System control management method. 18. The transition of the execution state of the CPU is performed at the timing of activating a system call for transition from the user program to the system program or at the timing of executing termination processing from the system call, or at the activation of the system call. 14. The method for managing system control by error trace according to claim 13, wherein the timing is a timing at which the system call is terminated and a timing at which a termination process from the system call is executed, or an arbitrary time timing during the processing of the system call. 19. The system control management method according to claim 1, wherein the attribute information is stored as trace information by using an interface provided to the application program. 20. When storing attribute information as trace information by using an interface provided to an application program, a program counter value, a CPU state value, a time element, a trace event number, contents of various registers, a program name, a process A name, a message, an address space identifier, a process identifier, a program identifier, a process attribute information, a program attribute information, a message queue identifier, a message queue attribute information, or a combination thereof is used as an interface parameter. 21. A system control management method based on the error trace described in 19. 21. The system control management method by error tracing according to claim 1, wherein the attribute information regarding the change in the state of the program is stored in the trace buffer only when a predetermined factor occurs. 22. When a predetermined factor such as an interrupt type, a system call type, a program or process identifier, a program name or a process name or a function name, writing to, reading from, and executing a specific address area occurs. 22. The system control management method by error tracing according to claim 21, wherein the attribute information is stored in the trace buffer only in the case. 23. The error trace system control management method according to claim 1, wherein attribute information indicating a program state change is stored in a trace buffer provided in a hardware control circuit. 24. The error according to claim 23, wherein the interface circuit between the CPU and the main memory, the external bus control circuit, the local bus control circuit, the input / output control circuit, and the main memory are hardware control circuits. System control management method by trace. 25. When the execution of the certified program is stopped, the system is stopped, or the execution of the program is postponed and the cause of the error is recovered according to the type of the program recognized as the error source. 2. The system control management method by error trace according to claim 1, wherein the system is controlled and managed by executing the program and restarting and continuing the operation of the system. 26. When the program recognized as the error source is a user program or a user process, the resource is returned to the system management means that manages the resource used by the user program or the user process. The system is controlled and managed by performing the following.
A system control management method based on the error trace described in 5. 27. The system is controlled and managed by returning resources such as a memory area used by a program recognized as an error source or various management tables or input / output devices to each free resource management pool. Claim 2
6. A system control management method based on the error trace described in 6. 28. A system for controlling and managing the system by not returning the memory page in which an error has occurred or other resource in which an error has occurred in a program identified as an error source to each free resource management pool. Item 27. A system control management method by error tracing according to Item 26. 29. A system based on an error trace according to claim 25, wherein when the type of the program recognized as the error source is a system program such as an operating system, the system is controlled and managed by performing a stop processing of the entire system. Control management method. 30. When the program recognized as the error source is a user program or a user process, the processing of the user program or the user process and related programs or processes related to them is stopped and other processing is performed. 26. The system control management method by error trace according to claim 25, wherein the system is controlled and managed by continuously executing it. 31. Recovery from an error is performed by replacing a resource in which an error has occurred with a program identified as an error source with an alternative resource, and the system is restarted by continuing the process when the cause of the error is recovered. 26. The system control management method according to claim 25, wherein the system control is performed by controlling.