JPS63211031A - Task state monitoring system - Google Patents

Abstract

PURPOSE:To increase the flexibility of debugging and to reduce an increase in overhead at the time of stationary operation by loading an optional task monitor program corresponding to the purpose of debugging in a system when the debugging is performed. CONSTITUTION:The task monitor program 41 consists of an initialization program 411 and a task monitor data gathering program 412, and the task monitor program 41 is loaded in a storage memory 2 from a disk drive 6 as a task at the time of debugging operation to execute the initialization program 411. Then the initialization program 411 reads out the head address of an address table 34 to know the gathering position of task monitor data and replaces branch instructions set at the positions 311 and 321 with an INT instruction for calling the task monitor data gathering program 412, thereby merging system programs 31 and 32 with the task monitor data gathering program 412.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for monitoring the operating status of each task and the operating status of a system program in a computer system that processes a plurality of tasks in parallel.

[Conventional technology]

Conventionally, in a computer system that processes a plurality of tasks in parallel, an in-circuit crystallizer has been used to debug a system program or a task processing program. With this method, when the operations of multiple tasks are intertwined and the state of a task changes due to an interrupt, it is easy to clearly trace which tasks are being processed and in what order. However, debugging was difficult.

Further, as shown in Japanese Patent Laid-Open No. 60'-11948, a task identification code is included in the system program.

Function to output task status, cause of task status transition 9 time, and VC for each task with the above output information! There is also a known method for monitoring task state transitions, which has a function of displaying the filled-in data on a screen.

[Problem that the invention seeks to solve]

In the above conventional technology, the task monitoring processing program is always waiting in the system program.

The system program capacity is large, it is not easy to change the collection items Q of task monitoring data or select the collection location depending on the debugging purpose, and the same overhead is incurred during normal operation as during debugging. There was no consideration given to this, and there were problems in terms of usability.

An object of the present invention is to solve the above-mentioned problems in the prior art, and to make it possible to install any task monitoring program suitable for the purpose of debugging into the system during debugging, thereby reducing the increase in overhead during steady operation. less,
The objective is to provide a task monitoring method that allows efficient debugging.

[Means for resolving the problem] The above purpose is to create a system program in which the reception section for soft interrupts (interrupts used to access functions that are standardly supported by the system) and the The reception section, task switching processing section, and other task monitoring data collection points are determined in advance, and the collection points are collectively managed in one table, and any task monitoring program is loaded from the disk device to the memory during debugging. This is achieved by placing a call command for the task monitoring program at the collection location.

[For production]

An area for storing the identification code of the collection point and 9 branch instructions are placed at the scheduled collection point of task monitoring data in the system program, and the branch instruction is executed during steady operation.
Skip task monitoring processing. Thereby, the increase in overhead during steady operation can be minimized.

On the other hand, during debugging, the branch instruction placed at the scheduled collection location of task monitoring data is replaced with a call instruction for the task astigmatism program. The collection location already has a collection location identification code area, and this is used as the task monitoring program call command (7) Harameter. Thereby, the reception port for the task monitoring program can be placed in one place, which simplifies the structure of the task monitoring program and the process of replacing the program call command.

〔Example〕

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

FIG. 2 is an overall configuration diagram of a computer system according to an embodiment of the present invention, in which 1 is a central processing unit (CPU);
2 is a storage memory within the central processing unit, 5 is a console input/output device, 6 is a disk device, and 7 is a printer device.

FIG. 5 shows software stored in the storage memory 2, in which 5 is a system program, 4! '! .

A group of task processing programs.

System program 5 that controls parallel processing of multiple tasks
The interrupt receiving unit 511 and the interrupt processing unit 55. Task scheduling 52. Address tape A154 for storing collection points of task monitoring data. It consists of a functional process @55 that reads the first address of the address table 54.

The operation of the computer system constructed using VCL as described above will be explained below.

The operation of the system program 5 begins when a soft or hard interrupt occurs.

When an interrupt occurs while task 41 is being executed, interrupt reception s5
1 examines the causes of the interrupt occurrence and branches to the corresponding interrupt processing section 55.

The interrupt processing unit 55 performs tasks such as activating a task, terminating a task, waiting for an event, notifying the occurrence of an event, and accessing a file, in accordance with a request from a task. In these interrupt processes, the state of the task changes.

In the task activation process, the task to be processed changes from a dormant state to an executable state. In task termination processing, the self-task changes from the running state to the dormant state. In the event wait process, the invoking task changes from the executing state to the waiting state. In the event occurrence notification process, the task to be processed changes from a waiting state to an executable state.

When the interrupt processing 55 is completed, the task monitoring data 52
Branch to.

The task scheduler 1-ra 52 selects any one of the running and executable tasks according to the scheduler algorithm, and places the selected task in the running state. When the previous running state task and the new running state task are different, task switching is also performed. When the task scheduling process is completed, execution of the selected new currently executing task begins.

If a soft interrupt or a hard interrupt occurs while the processing of the task group 4 is being executed, the operation of the system program 5 is repeated again.

During debugging, the interrupt reception unit 51 collects data such as the type of punishment and factor 1 interrupt occurrence time of task number 9 in the running state, and the task scheduler 521C returns the task to the running state. By collecting data such as the task number and time of the selected task, organizing this information by task, and displaying it in the order of collected time, you can check the operating status of the task and the operating status of the system program. can be made clear.

Figure 4 shows the processing method during steady operation.

Interrupt reception unit 61. The collection points 511 and 521 of the cutting monitoring data are determined in the task squeegee 252, and the addresses thereof are stored in the address table 54. 5
11, 521 is a branch instruction (JMP instruction) that skips the task monitoring process, and 512, 522 is an identification code area of the collection location.

Figure 1 shows the processing method during debugging.

The task monitoring program 41 includes an initialization program 411 and a task monitoring data collection program 412. During debugging, task supervisory program 41
is loaded from the disk device 6 to the storage memory 2 as a task, and its initialization program 411 is executed.

FIG. 6 shows details of the initialization program 411, and a functional process 55 for reading out the first address of the 5th@<7 address table 54.

The initialization program 411 calls the function processing 55, reads the first address of the address table 54, learns the collection point of task monitoring data from the address table 54, and
The branch instruction (JMP instruction) placed in 1iWr511, 4521 is transferred to the task monitoring data collection program
Replace it with an INT instruction to call 12. With the above, the system programs 51 and 52 and the task monitoring data collection program 412 have been combined, and task monitoring processing will be performed from now on.

The task monitoring data collection program 412 reads the occupation code of the collection location from the identification code area 512, 522 of the task monitoring data collection location passed as a parameter, and collects task monitoring data suitable for the collection location. Then, the return address is updated by the number of identification areas and the process returns.

41 task monitoring program is System File.

Since it is a separate file from the deno file, you can run any task monitoring program depending on the purpose.

For example, ■ Debugging that immediately outputs the collected task monitoring data to a printer, ■ Debugging that sends the collected task monitoring data to another computer using a communication line, ■ Collecting specific data at a specific collection point. Debugging etc. can be done easily.

〔Effect of the invention〕

According to the present invention, any task monitoring program suitable for the purpose of debugging can be installed in the system during debugging operation, so debugging can be performed with great flexibility, and the number of debugging steps can be greatly reduced. Furthermore, during steady operation, the task monitoring progeran does not operate and is non-resident in the memory 4TJ, so there is an effect that there is little increase in overhead during steady operation, and there is no increase in system program capacity.

[Brief explanation of the drawing]

l! Figure 1 shows a processing method during debugging of a system according to an embodiment of the present invention, Figure 2 shows the overall configuration of the computer system, and Figure 5 shows the configuration of software stored in the storage memory. The figure shows a processing method during steady operation of the system, FIG. 5 is a functional processing diagram for checking the first address of the address table, and FIG. 6 is a diagram showing an initialization processing method for a task monitoring processing program. 1...Central processing unit 2...Storage memory 5...System program 4...Task 51...Interrupt receiving unit 52...Task scheduler 54...Address table 55...Address table Functional processing 41 to read the start address of 54...Task monitoring program $7 @ No. 22 No. 3 Fig. 4 Fig. 85 Fig. 6

Claims (1)

[Claims] 1. In a computer system that processes multiple tasks in parallel, configure the system program that controls the parallel processing of multiple tasks and the task monitoring program as separate files,
Place the collection point identification code area and branch instruction in the place where task monitoring data is scheduled to be collected in the system program.
and has an address table for storing the address,
During system debugging, the task status monitoring program is loaded from the disk device to the storage memory, and by replacing the branch instruction at the address indicated in the address table with an instruction that calls the task monitoring program, any task status monitoring program can be executed during debugging. A task status monitoring method that is characterized by its ability to be combined with system programs.