JP2002318713A - Cpu occupancy time measuring method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a CPU occupancy time measuring method, with which CPU occupancy time in interrupt processing can be measured from the measured result of the CPU occupancy time in non-interrupt processing. SOLUTION: In the CPU occupancy time measuring method for an integrated system configured by combining one or a plurality of interrupt routines to be started in a fixed cycle and non-interrupt processing to be run while no interrupt routine is not run, by measuring time 4 for non-interrupt processing to be run, the CPU occupancy time in the interrupt routine is measured from a period 2 of interruption 2 and the time 4.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method for measuring the CPU occupation time of an interrupt routine in an industrial embedded system.

[0002]

2. Description of the Related Art In a recent embedded system of a plurality of tasks, as seen in a real-time OS (RTOS), a module (source code) is required to be reused while maintaining independence between modules. Because we try to reuse it as a software part,
As the program size increases, the processing becomes more complicated and the quality of bugs tends to become more complicated. In such embedded systems, real-time performance that obeys time constraints and high reliability that it must operate stably are always required, and in terms of time constraints, processing must be completed within a predetermined time. This is used in an application environment in which resetting or turning on the power in the middle cannot be performed in terms of reliability, and continuous operation may be required.

[0003]

However, in the conventional embedded system, it is mainly a method of confirming whether a task such as an interrupt routine started at a constant cycle is completed within a cycle by actually moving the task. Therefore, there is a problem that the normal processing of the system itself cannot be realized unless the interrupt routine started at a constant cycle completes the processing accurately within the cycle. In addition, even if the system has already started to operate normally, if the processing load changes due to specification changes, etc., there is a problem that it is necessary to actually operate the system again to confirm normal processing. . Therefore, the first invention of the present invention is an embedded system in which a plurality of interrupt routines and non-interrupt processing are combined, and the running time is measured by the non-interrupt processing to know the CPU occupation time of the interrupt routine group. CPU that can grasp the limit of system processing load
It aims to provide an occupancy time measurement method. Further, a second invention of the present invention is an embedded system in which a plurality of interrupt routines and non-interrupt processing are combined, the execution time of non-interrupt processing in a state where all interrupts are disabled, and
It is an object of the present invention to provide a CPU occupation time measuring method capable of grasping a limit of a system processing load by knowing a CPU occupation time of an interrupt routine group from an execution time in a state where an interrupt is permitted.

[0004]

In order to achieve the above object, according to the first aspect of the present invention, there is provided one or more interrupt routines that are started at a fixed period, and runs when no interrupt routine is running. In a CPU occupation time measuring method for an embedded system configured by combining a non-interrupt process, a CPU of the interrupt routine is measured by measuring a running time of the non-interrupt process.
The occupancy time is measured. The invention according to claim 2 is an embedded system configured by combining one or more interrupt routines started at a fixed period and non-interrupt processing that runs when no interrupt routine is running. In the CPU occupation time measuring method, the CPU occupation time of an interrupt routine is measured from the execution time of the non-interrupt processing in a state where all interrupts are prohibited and the execution time in a state where interrupts are permitted. Features. According to the CPU occupation time measuring method, the running time is measured by the CPU occupation time measuring unit provided in the non-interrupt processing unit of the continuous loop.
This measures the CPU occupation time of non-interrupt processing directly, and consequently calculates the CPU occupation time of the interrupt routine group by ratio calculation, grasps the limit of the system processing load, and connects the whole system. Time management becomes possible. Alternatively, the CPU occupation time ratio is calculated as (T1−T0) / T1 from the measurement execution time T0 in the non-interrupt processing in a state in which all interrupts are prohibited and the measurement execution time T1 in a state in which interrupts are permitted. Then, by knowing the CPU occupation time of the interrupt routine group and the limit of the system processing load, it becomes possible to manage the connection time of the entire system.

[0005]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a timing chart of the CPU occupation time measuring method according to the first embodiment of the present invention. FIG. 2 is a flowchart of the processing of the CPU occupation time measuring method shown in FIG. FIG. 1 shows the running state of each part of the embedded system in which interrupts 1 and 2 of two fixed-cycle interrupt routines and one non-interrupt process are performed. 1 is a cycle of the interrupt 1 of a fixed cycle. Reference numeral 2 denotes a cycle of the interrupt 2, and the interrupt priority of the interrupt 1 is higher than that of the interrupt 2. Reference numeral 3 denotes a measurement period 1 (representing waiting until the start of measurement), 4 denotes a measurement period 2 (represents that measurement is being performed), and 5 denotes a measurement period 3 (represents post-measurement processing).

Next, the operation will be described. The routine of the interrupt 1 is started in one cycle, and ends when the processing is completed. The routine of the interrupt 2 is started in a cycle of 2, but the execution of the routine of the interrupt 1 having a higher priority is delayed or interrupted during the running. In FIG. 1, the execution time of the interrupt 2 routine is delayed, and the execution of the interrupt 1 routine is interrupted by the execution of the interrupt 1 routine until the processing is completed even after the start, as indicated by the period 2. The execution of the non-interrupt processing is executed only when no interrupt routine is running. This time the CPU for non-interrupt processing
This is the occupation time, and the other time is the CPU occupation time of the interrupt routine group. The interrupt routine is 1
Although the process is completed once, the non-interrupt process is a loop of the program itself, there is no concept of completion, the execution is frequently interrupted, but the process is continuous and it is a continuous operation application . According to the present invention, the CPU occupation time measuring unit is provided in the non-interrupt processing unit, and when entering the “measurement mode”, this portion runs and measures time. FIG. 2 is a flowchart of this processing.
The routine has an execution counter, which is incremented each time it is executed. This will be described with reference to FIG. First, when the non-interrupt processing unit enters the measurement mode, the count value n of the interrupt 2 is read (S100). Wait for this to be updated while looping (S10
1). This is a process for adjusting the timing of the measurement start because the timing of entering the measurement mode is not known. In FIG. 1, this period is three measurement periods 1 (waiting until the start of measurement), and is the measurement period 1 shown on the right side of FIG. When measurement is started, the count value m of the next interrupt 2
(= N + 1) is read (S102). Loop until the count value m is updated (S10
3) While waiting for the update (S104), a constant wait process (addition of the number of loops such as i ++) is performed (S10).
5). When the count value of the interrupt 2 is updated, the process exits the loop. This period is 4 measurement periods 2 in FIG. The subsequent processing corresponds to the measurement period 3 of 5 in FIG.
106). The measurement time of 4 in FIG.
It can be measured by counting the number of times 105) and the waiting time per loop. In this case, the interrupt counter having the lowest priority is used. This is because all interrupts are included. Thus, the running time (4 in FIG. 1) of the non-interrupt processing is obtained, and the ratio of the interrupt 2 to the cycle (2 in FIG. 1) is the CPU occupation time ratio of the non-interrupt processing, and the rest is the CPU occupancy of the interrupt processing group. Rate.

Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 3 is a flowchart of the non-interrupt processing of the CPU occupation time measuring method according to the second embodiment of the present invention. FIG. 4 is a diagram showing the time required for one round when the interrupt of the non-interrupt processing shown in FIG. 3 is prohibited. FIG.
FIG. 4 is a diagram showing a time required for one round when an interrupt of non-interrupt processing shown in FIG. 3 is permitted. In FIG. 3, 10 is I /
This is processing such as O output or memory writing. Next, the operation will be described. The non-interrupt processing as shown in FIG. 3 usually has an infinite loop configuration. By inserting a process such as 10 in one place of the loop, the timing at which the program has passed this place can be known externally, and the time required for one round of processing can be measured. First, interrupts are prohibited, and the time required for one round of non-interrupt processing is measured. This is shown in FIG.
The execution of the non-interrupt processing is not interrupted by the interrupt processing, and the measured value of one round time at this time is 11: T0. Next, an interrupt is permitted, and the time required for one round of this processing is measured. As shown in FIG. 5, the execution of the non-interrupt processing is interrupted by the interrupt processing 13 and the interrupt processing 14. Here, the suspended time is, in other words, the time required for interrupt processing. Non-interrupt processing 1 in FIG.
The measurement result of the time required for the circumference is 12: T1. This one
2: T1 includes time required for interrupt processing in addition to non-interrupt processing. Therefore, the time obtained by subtracting the time 11: T0 from the time 12: T1 is the time required for the interrupt processing. The CPU occupancy of the interrupt processing is given by the following equation. (T1-T0) / T1 Thus, the CPU not occupied by the interrupt routine
By measuring time, i.e. CPU idle time,
The occupancy time of the interrupt routine can be measured.

[0008]

As described above, according to the present invention,
By measuring the CPU time that is not occupied by the interrupt routine, that is, the CPU idle time, the occupancy time of the interrupt routine is measured, and the CPU occupation ratio required for interrupt processing is obtained to achieve system optimization and efficient specification change. There is an effect that can be done.

[Brief description of the drawings]

FIG. 1 is a timing chart of a CPU occupation time measuring method according to a first embodiment of the present invention.

FIG. 2 is a flowchart of a process of a CPU occupation time measuring method shown in FIG. 1;

FIG. 3 is a flowchart of non-interrupt processing of a CPU occupation time measuring method according to a second embodiment of the present invention.

FIG. 4 is a diagram showing a time taken for one round when interrupts in the non-interrupt processing shown in FIG. 3 are prohibited.

FIG. 5 is a diagram showing a time taken for one round when an interrupt of non-interrupt processing shown in FIG. 3 is permitted.

[Explanation of symbols]

 1 Period of interrupt 1 2 Period of interrupt 2 3 Measurement period 1 4 Measurement period 2 5 Measurement period 3 10 I / O output or memory write 11 T0 (one round when interrupt is disabled) 12 T1 (one round when interrupt is enabled) time)

Continued on the front page (72) Inventor Kazuhiko Hiramatsu 2-1 Kurosaki Castle Stone, Yawatanishi-ku, Kitakyushu-shi, Fukuoka F-term (reference) 5B042 GA23 HH20 MC21 MC29 MC33 5B098 GA02 GA04 GC05 GC10 JJ05

Claims (2)

[Claims] 1. A CPU occupation time measuring method for an embedded system configured by combining one or a plurality of interrupt routines started at a fixed period and a non-interrupt process running when no interrupt routine is running. 5. The method according to claim 1, wherein the CPU occupation time of the interrupt routine is measured by measuring a time during which the non-interrupt process runs. 2. A method for measuring a CPU occupation time of an embedded system configured by combining one or more interrupt routines started at a fixed period and a non-interrupt process running when no interrupt routine is running. Wherein the CPU occupation time of an interrupt routine is measured from the execution time of the non-interrupt processing in a state where all interrupts are prohibited and the execution time in a state where interrupts are permitted. Measuring method.