JP2788353B2 - Task trace method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a task tracing method in a data processing system.

[0002] The task trace function is a function provided as a debug function and an RAS function when developing software, and is generally a function of storing input / output information of a task in a trace area in time series. The purpose of using the task trace function is to analyze the stored information at the time of software testing and at the time of failure, and to use it for investigating the cause of the failure.

[0003]

2. Description of the Related Art There are various methods for performing task tracing, and two typical methods will be described below.

[0004] One is a method of realizing each application program task (hereinafter, referred to as an AP task) created by a user. In this method, as shown in FIG. 3, a program is created so as to equip each AP task with a task trace function, and each AP task uses the task trace function when the AP task is started to execute a trace area of a memory. This is a method of directly accessing the server and storing the input / output information of the inter-task interface.

In this method, since the task trace stores the start / stop of all tasks in time series, a plurality of A
P task will access the trace area,
In this case, if a high-priority task is started while a low-priority task is accessing the trace area, the priority processing interrupts the former processing and the input / output information may be lost. It is necessary to perform exclusive control to wait for access from other tasks while accessing the trace area. Therefore, each task acquires / releases a buffer for transferring input / output information to the operating system. It is necessary to do.

Another method is to create and implement a task trace task on the operating system side. As shown in FIG. 4, each AP task starts a task when it starts up. In this method, the input / output information of the own task is passed to the trace task, and the input / output information received from each AP task by the task trace task is stored in the trace area.

[0007]

The task tracing method realized by each of the AP tasks has the following problems.・ Obtain buffer for operating system /
It is necessary to perform release exclusive control processing, and the overhead of the operating system increases by the exclusive control processing. In each AP task, the number of program steps increases by the amount of performing the exclusive control processing. In some cases, a task with a higher priority waits until the processing of a task with a lower priority ends for exclusive control processing. In other words, as indicated by * 1 in FIG. 5, the exclusive control is applied to the AP task 1, so that the high priority A
When the exclusive control request is issued by the P2 task, the AP task 2 waits until the exclusive control release request without returning from the request ends. Management of execution / stop of task trace is required for each AP task.

The latter task tracing method for creating and realizing a task tracing task has the following problems. Since the task trace task prepared on the operating system side is started, the overhead of the operating system increases. -More interfaces between tasks. In each AP task, the number of program steps increases by the amount of processing described above.

The present invention has been made in view of the above problems, and has as its object the overhead of an operating system, the interface between tasks,
An object of the present invention is to provide a task trace method capable of reducing the number of task programs and the like.

[0010]

FIG. 1 is an explanatory view of the principle according to the present invention. In the task trace method according to the present invention, a task trace processing routine is prepared on the operating system side, a trace processing return address area is added to the task management table, and an entry address of the task is set here, and the task When tracing is performed, the entry address of the task trace processing routine is set in the task entry address area of the task management table of the task, so that when the operating system processes the task, the task entry address of the task management table is set. By accessing the area, the task trace processing routine is started in place of the task, and the task trace processing routine restores the trace processing of the task management table when ending the trace processing for the task. It accesses the address area is obtained so as to pass control to the task.

[0011] The task tracing method according to the present invention further comprises:
In the above-described method, by rewriting the contents of the task entry address area of the task management table to the entry address of the task trace processing routine on a task basis by a task trace start command during system operation,
The operation mode can be changed from stop to start on a per-task basis, and the contents of the task entry address area of the task management table can be rewritten on a per-task basis by the task trace stop command to the contents set in the trace processing return address area. The mode can be changed from start to stop for each task.

[0012]

When performing a task trace for a certain task, the entry address of the task trace processing routine is set in the task entry address area of the task management table of the task. As a result, when the operating system accesses the task entry address area of the task management table when processing the task, the entry address of the task trace processing routine is read instead of the entry address of the task. Instead, a task trace processing routine is started. The task trace processing routine executes the trace processing for the task, and when the task ends, accesses the trace processing return address area of the task management table. Since the entry address of the task is stored in the trace processing return address area, control is passed to the task based on the entry address.

As described above, a task for which the operation mode is set to perform the task trace is automatically executed even if the operating system does not care about the task trace processing routine when processing the task. The routine is started to execute the task trace, and then the processing of the target task is executed.

By rewriting the contents of the task entry address area of the task management table with the entry address of the task trace processing routine by using the task trace start command, the operation mode can be changed from stop to start for each task. By rewriting the contents of the task entry address area of the task management table with the contents set in the trace processing return address area for each task, the operation mode can be changed from start to stop for each task.

[0015]

Embodiments of the present invention will be described below with reference to the drawings. In this embodiment, the task trace function is realized by an operating system. First, the outline of the method of this embodiment will be described. A test trace processing routine for performing the trace processing of the AP task is prepared on the operating system side, and a trace processing return address is added to the task management table (TCB). When performing a trace process for a certain AP task, the contents of the task entry address area of the task management table of the AP task (that is, the entry address of the AP task) is rewritten to the entry address of the task trace processing routine. When the operating system activates the AP task, the contents of the task entry address area are rewritten to the entry address of the task trace processing routine.
A task trace processing routine is automatically started instead of a task. The task trace processing routine is the AP
The task trace processing is executed, and when the processing is completed, the control is passed to the original AP task by referring to the task processing return address in the task management table, and the AP task is executed. The task trace is automatically performed without being aware of the activation of the trace processing routine.

The task management table is provided for each AP task. Usually, the entry address of the corresponding AP task is stored in the task entry address area, and the operating system stores the entry address in the task entry address area. By accessing the AP
The task is running. In this embodiment, the following information is added to the task management table in order to realize the task trace function by the operating system. That is, the trace processing return address is the address of the trace memory. This trace processing return address stores the entry address of the AP task corresponding to the task management table.

Hereinafter, a method of realizing the task trace function by the operating system will be described. (1) System editing (system generation) method The user sets the following information to use the task trace function. During system generation ・ Start address of trace area ・ Trace area size During task generation ・ Trace operation mode (Stop / Start: Set the operation mode of a task to start when performing a task trace, and to stop when not performing the task. -Memory address to be traced (set the memory address for storing trace information)

(2) Initial Operating System
The operating system initializes the task management table using the start address and size of the trace area during system generation. At the time of task generation, the operation mode of the trace (that is, the stop mode or the start mode) is determined. Then, in the stop mode, the entry address of the task is set in the task entry address area of the task management table. on the other hand,
In the start mode, the entry address of the task trace processing routine is set in this task entry address area. Further, the memory address to be traced set by the user in (1)-is set in the address area of the trace memory of the task management table, and the entry address of the task is set in the trace processing return address area.

(3) Trace processing start processing (A in FIG. 2)
When starting / stopping a task, the operating system accesses the task entry address area of the task management table of the task, obtains an entry address, and starts / stops the processing.

At this time, if the operation mode of the task is "stop", the entry address of the task is set in the task entry address area at the time of the initialization processing of the operating system. Is started / stopped.
Therefore, no task trace is performed in this case.

On the other hand, when the operation mode of the task is “start”, the entry address of the task trace processing routine is set in the task entry address area at the time of the initialization processing of the operating system. Control is passed to the routine,
The trace processing for the task is executed by this task trace processing routine. Therefore, the operating system performs the conventional dispatch processing, that is, the trace processing is started if the operation mode of the trace is "start", even though the task trace processing routine is completely unconscious.

(4) Trace Processing When the trace processing is started by the task trace processing routine, the task trace processing routine uses the management table of the trace area to execute the trace information (start task number, start task number, start factor, buffer) address,
(For example, memory contents) in the trace area. When this is completed, the task trace processing routine refers to the trace processing return address area of the test management table, and
An unconditional branch is made to the address stored therein, and the trace processing is terminated. In this trace processing return address area, the entry address of the task is set at the time of the initialization processing of the operating system, so that the control is passed to the task which the operating system tried to start / stop, and thereafter, the Task processing will be executed.

(5) Operation mode change processing during operation (FIG. 2)
The operation mode can be changed for each task by inputting a command from the online debugger.

The task trace start command is shown in FIG.
As shown in (2), the content of the task entry address area set in the task management table is rewritten to the entry address of the task trace processing routine, and this processing is performed for all the tasks specified to be changed.

On the other hand, the task trace stop command sets the content (task entry address) set in the trace processing return address area of the task management table in the task entry address area, as shown in FIG. 2B. This process is performed for all tasks for which the change is specified.

(7) Display processing of trace information during operation By connecting a terminal by inputting a command from the online debugger , trace information is displayed on the screen of the terminal in the indicated mode. Further, by inputting another command, it is possible to initialize the trace area.

As described above, the operating system described above
Since the task trace function by the system starts the task via the trace processing, the overhead of the operating system is added and the time until the task starts is slower than when the task trace function is not activated. Since the task implements a task trace function as internal processing of the task and requires a processing time there, this delay can be substantially ignored. On the other hand, a high-performance task having a short trace processing time may be affected. In such a case, the effect can be prevented by stopping and starting the trace function of the task.

If it is desired to use the task trace function only during debugging or when investigating a fault, the operation mode of the task trace is set to "stop" in the generation of the system, the operation is performed, and the task trace function is used. In this case, the task trace may be set to “start” by inputting a command from the online debugger to activate the task trace function, and then “stop” after use. Even in such a case, it is possible to trace only the task to be investigated by designating only the task to be investigated.

[0029]

As described above, according to the present invention, by supporting the task trace function in the operating system, the development burden on the application program developer can be reduced. Also, the overhead of the operating system can be reduced. In addition, it is possible to reduce the inter-task interface required when the test trace task is used.

By inputting the command of the online debugger and setting the trace mode for system editing, it is possible to activate / stop the task trace function and activate the trace function only for a specific task, so that an operation mode suitable for use can be specified. Becomes Therefore, it can be applied to any system.

Further, by connecting a terminal by inputting a command of the online debugger, trace information can be analyzed even during operation, and efficiency during debugging and failure analysis can be improved.

Further, since it is possible to cope with various tests and to easily analyze the cause of a bug, it is possible to contribute to improvement of the quality of development application software.

By registering the entry address of the trace processing routine in the task control area (TCB), it is possible to prevent the processing performance of the operating system from decreasing when the trace function is stopped.

[Brief description of the drawings]

FIG. 1 is an explanatory view of the principle according to the present invention.

FIG. 2 is a flowchart illustrating an embodiment of the present invention.

FIG. 3 is a diagram illustrating a conventional method for realizing a task trace function in each AP task.

FIG. 4 is a diagram illustrating a conventional method for implementing a task trace function by creating a task trace task.

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Makoto Ueno 3-9-1, Shin-Yokohama, Kohoku-ku, Yokohama-shi, Kanagawa Prefecture Inside Fujitsu Communication Systems Limited (56) References JP-A-63-174142 (JP, A (58) Fields surveyed (Int.Cl. ⁶ , DB name) G06F 11/28-11/34

Claims (2)

(57) [Claims] 1. A task trace processing routine is prepared in the operating system, a trace processing return address area is added to a task management table (TCB), and an entry address of the task is set therein. When performing the task, the entry address of the task trace processing routine is set in the task entry address area of the task management table of the task, so that when the operating system performs the task, the task entry address of the task management table is set. By accessing the area, the task trace processing routine is started in place of the task, and the task trace processing routine returns to the task management table when ending the trace processing for the task. Task Trace method to pass control to the task by accessing the-less region. 2. The operation mode is changed from stop to start for each task by rewriting the contents of a task entry address area of a task management table with an entry address of a task trace processing routine by a task trace start command during system operation. 2. The operation mode can be changed from start to stop for each task by rewriting the contents of a task entry address area of a task management table to contents set in a trace processing return address area by a task trace stop command. The described task trace method.