JP2020112918A - Method of measuring execution load factor of software in control device - Google Patents

Abstract

To provide a method of easily measuring an execution load factor of software with a small error while suppressing a required time to 100 ns or less by limiting processing for measuring the execution load factor into necessary minimum.SOLUTION: In executing software comprising a plurality of pieces of interruption processing 401 to 404 and idling processing 405, a high-level voltage is output to a digital terminal when interruption processing being a measurement object is started, and a low-level voltage is output when the interruption processing is terminated. The low-level voltage is output when interruption processing not being the measurement object is started, and the high-level voltage is output when the interruption processing being the measurement object occurs and interruption processing having a priority level higher than that of the interruption processing being the measurement object does not occur when the interruption processing not being the measurement object is terminated. While the idling processing is executed, the low-level voltage is output (406 and 407), and an execution load factor of the software is calculated from an average voltage at the digital terminal in a prescribed period.SELECTED DRAWING: Figure 5

Description

The present invention relates to a technique for measuring an execution load factor of software in a control device that operates software at a high-speed fixed cycle such as a motor drive device of an electric vehicle.

As a method to measure the execution time rate of software execution time or execution time per unit time, the counter value of the CPU built-in timer is recorded and started at the start and end of the execution of tasks and interrupt processing belonging to the target software. There is known a technique of calculating an execution time and an execution load factor from a difference between time and end time.

In this technique, during execution of a task or interrupt process belonging to target software, if a task or interrupt process not belonging to the software is preferentially executed, it is necessary to subtract the execution time thereof.

In the execution time measuring method for tasks and the like in the control device described in Patent Document 1, when tasks or interrupts having a higher priority than the measurement target are executed in multiple, the execution time is subtracted by operating the stack. ing.

Specifically, the control device uses the free-run counter provided for the operating system, reads the time value of the free-run counter at the start of the task or interrupt processing, stores it as stack data by a push operation, and ends. The execution time of each task and interrupt process is calculated from the time value of the free-run counter at that time and the stack data recalled by the pull operation.

JP, 2003-288237, A

However, the method of Patent Document 1 has the following problems.

First, it takes time for the process of calculating the execution time of each task and interrupt process, and the process itself for subtracting the execution time of the task and interrupt process executed with a higher priority than the measurement target. Further, stack operations such as interrupt masks and signed operations accompanying execution of interrupt processing and calculation of execution time require more time. Therefore, in a CPU with low performance, the time required for executing the software and measuring the execution time may exceed the allowable time determined by the control target and the control content.

Then, in this method, the execution time includes the measurement time. This is a non-negligible error in a control device that operates software at a high-speed fixed cycle of several hundreds of μs or less, such as a motor drive device of an electric vehicle, and a large error occurs in the execution load factor calculated from the execution time.

The present invention is for solving the above-mentioned problems, and it is possible to measure the execution load factor of software easily and with a small error by minimizing the processing for measuring the execution load factor to a required time of about 100 ns or less. Provide a way to do.

In a control device having a digital terminal that outputs a high level voltage or a low level voltage, software including a plurality of interrupt processes with different priorities and an idling process executed when none of the interrupt processes is executed When executing, the high level voltage is output to the digital terminal during execution of each interrupt process, and the low level voltage is output during execution of the idling process.

The voltage of this digital terminal is measured with equipment such as an oscilloscope or memory high coder, and the ratio of the average voltage to the high level voltage in a predetermined period or the ratio of the period in which the high level voltage was output in the predetermined period is calculated. By doing so, the execution load factor of the software can be obtained.

When measuring the execution load factor in interrupt processing units, a high level voltage is output to the digital terminal at the start of interrupt processing of the measurement target, a low level voltage is output at the end, and a low level voltage is output at the start of interrupt processing other than the measurement target. At the end of the voltage measurement, when the interrupt process of the measurement target is occurring and the interrupt process of higher priority than the interrupt process of the measurement target is not occurring, the High level voltage is output and the voltage of the digital terminal is measured in the same manner as above. As a result, the execution load ratio of the target interrupt process can be obtained.

Further, a global variable is provided in the software, and the processing of replacing the digital terminal, the high level voltage and the low level voltage with the global variable, “1” and “0”, respectively, is executed, and the state of the global variable is monitored by a RAM monitor or the like. The execution load factor of software and interrupt processing can also be obtained by measuring with equipment.

According to the present invention, in a control device that operates software consisting of a plurality of interrupt processes having different priorities at a high-speed fixed cycle, it is possible to minimize the execution load factor measurement process of the software and the interrupt processes constituting the software. Therefore, the time required for the measurement is short, and the execution load factor can be measured with a small error.

The block diagram of the system which measures the execution load factor of software by the method of this invention. The block diagram of the software used as the object of measurement by the present invention. The timing chart showing 1st Embodiment of this invention. Numerical example of the digital terminal output signal in FIG. The timing chart showing 2nd Embodiment of this invention. Transition of interrupt process execution status variables and digital terminal output status corresponding to Fig. 5.

Embodiments of the invention will be described below with reference to the drawings.
[System and software configuration]
FIG. 1 shows an example of the configuration of a system for measuring the execution time and execution load factor of software by the method of the present invention. It is composed of a test device 10, a drive device 20, a motor 30, and a measuring device 40.

The test device 10 is connected to the drive device 20 with a cable for communication such as twisted pair, and gives a command to the drive device 20. The drive device 20 is a control device that controls the motor 30 in response to a command from the test device 10. The drive device 20 gives a command to the motor 30 and responds to the test device 10 with the execution result of the command and the state of itself and the motor 30. To return as. A power source for supplying a driving force to the motor 30, a power converter, and a device for detecting the current and the rotation state of the motor 30 are not shown. The measuring device 40 measures the state of the digital terminals and the memory of the driving device 20 through the I/O port of the driving device 20, and specifically uses an oscilloscope, a memory hi-coder, a RAM monitor or the like.

FIG. 2 is a configuration example of software (hereinafter, referred to as a control program) that operates in the drive device 20. The control program includes an interrupt process A (201), an interrupt process B (202), an interrupt process C (203), and an idling process 204.

The priority order of the interrupt processing is interrupt processing A (201)>interrupt processing B (202)>interrupt processing C (203). When no interrupt processing is executed, idling processing 204 is executed. Since the function of each process is not related to the execution load factor measurement of the present invention, the description thereof will be omitted.

In the system configuration of FIG. 1 and the control program configuration of FIG. 2, as a process for measuring the execution load factor, the drive unit 20 uses the interrupt process A (201), the interrupt process B (202), the interrupt process C (203), and the idling process 204. The high level voltage or the low level voltage is output to the digital terminal according to the execution state of.

The output voltage is held until a reverse level voltage (Low level voltage for High level voltage, High level voltage for Low level voltage) is output. Such a mechanism can be easily and inexpensively realized by a flip-flop circuit or a logic IC incorporating the flip-flop circuit.

In the following, the High level voltage and the Low level voltage are simply written as High and Low, respectively, and the digital terminal output voltage is also called a digital terminal output signal or a digital terminal output state.
[First Embodiment]
In the first embodiment, the entire control program shown in FIG. 2 is the measurement target.

As the processing for measuring the execution load factor, a high level voltage is applied to the digital terminal at the start of each of the interrupt process A (201), the interrupt process B (202), and the interrupt process C (203), and the low level voltage is started at the start of the idling process 204. Is output.

FIG. 3 is an operation example of the control program and a timing chart of the digital terminal output signal when the first embodiment is applied thereto. In the following, the circled numbers in the figures are represented by numbers in parentheses.

At the time of (1) when the interrupt processing A (301) is started, the digital terminal output signal 305 is changed from Low to High. Subsequently, the interrupt process B (302), the interrupt process C (303), the interrupt process A (301), and the interrupt process C (303) are sequentially executed.

At the time of (2) when the interrupt process C (303) is completed, no interrupt process is executed and the idling process 304 is started, so that the digital terminal output signal 305 is changed from High to Low. The period of (1)-(2) is T1.

Similarly, the periods of (3)-(4), (5)-(6), and (7)-(8) in which any of the interrupt processes are executed are T2, T3, and T4, respectively. The period of (1)-(9) is set as the measurement period, and its length is set as the total time Tall.

The execution load factor of the control program in the measurement period (1)-(9) is Tsum÷Tall×100% when the total interrupt processing time is Tsum=T1+T2+T3+T4.

At this time, the average voltage VA of the digital terminal is VA=(VH−VL)×Tsum÷Tall, where VH is the high level voltage and VL is the low level voltage. The execution load factor can be calculated by ÷(VH−VL)×100%.

For example, when the high level voltage is 5.0 V and the low level voltage is 0.0 V in the digital terminal output signal as shown in FIG. 4, the average voltage is 3.75 V when measured. When the execution load factor of the control program is calculated from this average voltage, it becomes 3.75V÷(5.0V-0.0V)×100%=75%.
[Second Embodiment]
In the second embodiment, a specific interrupt process among the processes forming the control program is a measurement target.

As a process for measuring the execution load factor, a high-level voltage is output to the digital terminal at the start of interrupt processing of the measurement target, and a low-level voltage is output at the end, and the low-level voltage is measured at the start of interrupt processing outside the measurement target and measured at the end. When the target interrupt process is occurring (after starting and before ending) and the interrupt process having a higher priority than the measurement target interrupt process is not occurring, the high level voltage is output. At the start of the idling process, a low level voltage is output to the digital terminal.

FIG. 5 is an operation example of a control program in which an interrupt process D (404) is added to the control program shown in FIG. 2 and a timing chart of a digital terminal output signal when the second embodiment is applied thereto.

The priority of the interrupt processing is interrupt processing A (401)>interrupt processing B (402)>interrupt processing C (403)>interrupt processing D (404). When no interrupt processing is executed, the idling processing 204 is To be executed.
[Second Embodiment: When Interrupt Process B is Measured]
When the interrupt process B (402) is set as the measurement target, the digital terminal output signal 406 is set to High and ended (4) at the time points (3), (10), and (15) when the interrupt process B (402) is started. ), (11) and (18), the interrupt process A (401), C (403) and D (404) which is not the measurement target is started while it is set to Low (1), (2), ( At the time of 4), (5), (9), (11), (12), and (16), the digital terminal output signal 406 is set to Low and ended (3), (6), (7), (8 ), (10), (13), (14), and (17), when the interrupt process B (402) is occurring, the digital terminal output signal 406 is set to High.

During the period of (3)-(4) and (10)-(11), the interrupt process having a higher priority than the interrupt process B (402) is not executed during execution. Therefore, the period is directly the period when the digital terminal output signal 406 is High.

During the period (15)-(18), the interrupt process A (401) having a higher priority than the interrupt process B (402) is generated. Therefore, the digital terminal output signal 406 is set to Low at the time (16) when the interrupt processing A (401) is started, and the digital terminal output signal 406 is set to High at the time (17) when the interrupt processing B is being generated. To Then, the digital terminal output signal 406 becomes High in the periods (15)-(16) and (17)-(18) of the period (15)-(18).

Here, the periods (3)-(4), (10)-(11), (15)-(16), and (17)-(18) are referred to as Tb1, Tb2, Tb3, and Tb4, respectively. The period (1)-(19) is set as the measurement period, and its length is set as the total time Tall.

The execution load ratio of the interrupt processing B (402) in the measurement period (1)-(19) is Tsum÷Tall×100% when the total interrupt processing time is Tsum=Tb1+Tb2+Tb3+Tb4, which is the same as in the first embodiment. Can be calculated from the measured value of the average voltage of the digital terminal.
[Second Embodiment: When the interrupt processing D is the measurement target]
When the interrupt process D (404) is set as a measurement target, the digital terminal output signal 407 is set to High at the time point (1) when the interrupt process D (404) is started, and at the time point when the interrupt process D (404) is ended (8). , Interrupt processing A(401), B(402), C(403) outside the measurement target was started (2), (3), (4), (5), (9), (10), ( At the time of 11), (12), (15), and (16), the digital terminal output signal 407 is set to Low and ended (3), (4), (6), (7), (10), (11). ), (13), (14), (17), and (18), when the interrupt process D (404) is occurring and the interrupt process with a higher priority is not occurring, the digital terminal output signal 407 is set to High. To

During the period (1)-(8), the interrupt processes A (401), B (402), and C (403) having a higher priority than the interrupt process D (404) are generated.

Therefore, first, at the time point (2) when the interrupt processing A (401) is started, the digital terminal output signal 407 is changed from High to Low. At the time of (3) when the interrupt process A (401) is completed, the interrupt process D (404) that is the measurement target is being executed, but the interrupt process B that has been waiting for a turn while the interrupt process A (401) is being executed. Since (402) is continuously started, the digital terminal output signal 407 remains Low. The same applies at the time point (4) when the interrupt process B (402) is ended and the interrupt process C (403) is started, and the digital terminal output signal 407 is kept low during the period (2)-(4). It will be.

Then, the digital terminal output signal 407 remains Low even at the time of (5) when the interrupt processing A (401) is started again. At the time of (6) when the interrupt process A (401) is completed, the interrupt process D (404) to be measured is still being executed, but since the interrupt process C (403) having a higher priority is being executed, it is digital. The terminal output signal 407 remains Low. At the time of (7) when the interrupt process C (403) is completed, the interrupt process D (404) to be measured is being executed, and the interrupt process having a higher priority than this is not being executed. Therefore, the digital terminal output signal 407 is changed from Low to High. To

After that, the interrupt processing D (404) is restarted, and the digital terminal output signal is changed from High to Low at the time of the completion (8). Since the interrupt process D (404) is not executed thereafter, the digital terminal output signal is kept low.

Therefore, the digital terminal output signal 407 becomes High in the periods (1)-(2) and (7)-(8) of the period (1)-(19).

Here, the periods (1)-(2) and (7)-(8) are referred to as Td1 and Td2, respectively. The period (1)-(19) is set as the measurement period, and its length is set as the total time Tall.

The execution load ratio of the interrupt processing D (404) in the measurement period (1)-(19) is Tsum÷Tall×100% when the total interrupt processing time is Tsum=Td1+Td2, which is also the same as in the first embodiment. Can be calculated from the measured value of the average voltage of the digital terminal.

The process of changing the output to the digital terminal according to the execution state of each interrupt process as described above can be realized by providing an interrupt process execution state variable in the control program.

An interrupt process execution state variable indicating the execution state is provided for each interrupt process, and the interrupt process execution state variable is turned ON (running) at the start of the interrupt process and turned OFF (not being executed) at the end. The interrupt processing execution state=ON includes the execution restart waiting state, and is the period from the start of the interrupt processing to the end thereof.

FIG. 6 shows transitions of interrupt processing execution state variables and digital terminal output states of each interrupt processing in FIG.

For example, in the period from (4) to (7) when the interrupt process D (404) is the measurement target in the second embodiment, another interrupt process is interrupted at the time (4) when the interrupt process B (402) is completed. When the execution state variable is referred to, the execution state variable of the interrupt process D (404) that is the measurement target is ON, but the execution state variable of the interrupt process C (403) having a higher priority is ON, so the output to the digital terminal is made. Becomes Low, and the execution state variable of the interrupt process D (404) to be measured is still ON at the time of (6) when the interrupt process A (401) is completed, but the interrupt process C (403) having a higher priority than this. Since the execution state variable of is ON, the output to the digital terminal is also Low. At the time of (7) when the interrupt processing C (403) is completed, the execution state variable of the interrupt processing D (404) to be measured is ON and the interrupt processing A (401), B (402), C having a higher priority than this. Since none of the execution state variables of (403) are ON, the output to the digital terminal becomes High.

Since such a process consists only of data comparison and output and does not require the complicated calculation as mentioned in the problem, the required time can be suppressed to about 100 ns or less.
[Application of the first and second embodiments]
In the first and second embodiments, by providing the control program with a variable for storing the measurement mode, a mode for measuring the execution load factor of the entire control program as in the first embodiment and a mode as in the second embodiment. In addition, the test apparatus 10 may be allowed to specify the mode for measuring the execution load rate in units of interrupt processing.

Further, in the second embodiment, an identifier for distinguishing each interrupt process is defined, and a variable for storing the identifier of the interrupt process to be measured is provided in the control program, so that the interrupt process to be measured is tested by the test apparatus 10. It may be possible to specify from.

In addition, in the first embodiment and/or the second embodiment, a plurality of digital terminals are provided, and a plurality of processes out of the processes targeted for measurement of either the entire control program or the interrupt process are performed. The output of each process may be assigned to a plurality of digital terminals by performing the process at the start and end of each interrupt process.
[Third Embodiment]
In the third embodiment, a global variable provided in the control program is used instead of the digital terminal.

A global variable is provided in the control program, and the processing in which the digital terminals, the high level voltage, and the low level voltage of the first and second embodiments are replaced with the global variables, "1" and "0", respectively, is executed.

If the state of this global variable is sampled at regular intervals using a device such as a RAM monitor, the interrupt processing time of the measurement target can be known from the number of times the global variable was "1", and the execution load factor can be calculated from this. it can.

For example, when the control program in which the digital terminal output signal is as shown in FIG. 4 is executed in the first embodiment, Tsum=T1+T2+T3+T4+T5=15 ms and Tall=20 ms are found in this embodiment, and the execution load factor is 15 ms÷20 ms× It is calculated as 100%=75%.

The accuracy of the execution load factor measurement can be improved by using this method in combination with the first embodiment and the second embodiment and matching the calculation results. Further, if this method is used alone, the execution load factor can be measured without a digital terminal or a voltage output means to the digital terminal.

10 Test device 20 Drive device 30 Motor 40 Measuring device 201, 301 Interrupt processing A
202, 302 Interrupt processing B
203, 303 Interrupt processing C
204, 304 Idling processing 401 Interrupt processing A
402 Interrupt processing B
403 Interrupt processing C
404 Interrupt processing D
405 Idling process 305, 406, 407 Digital terminal output signal

Claims (6)

In a control device having a digital terminal for outputting a high level voltage or a low level voltage,
When a software consisting of a plurality of interrupt processes with different priorities and an idling process executed when none of the interrupt processes is executed is executed,
A High level voltage or a Low level voltage is output to the digital terminal according to the execution state of each interrupt process, and
A High level voltage or a Low level voltage is output to the digital terminal according to the execution state of the idling process,
Calculating the execution load factor of the software or the interrupt processing from the voltage measurement value of the digital terminal,
Software execution load factor measurement method. A method for measuring an execution load factor of software according to claim 1, comprising:
A high level voltage is output to the digital terminal during execution of each interrupt process,
During execution of the idling process, a Low level voltage is output to the digital terminal,
An execution load factor of the software is calculated from an average voltage of the digital terminal in a predetermined period, or a length of a period in which a high level voltage is output to the digital terminal in the predetermined period.
Software execution load factor measurement method. A method for measuring an execution load factor of software according to claim 1, comprising:
One of the plurality of interrupt processes is the measurement target,
A high level voltage is output to the digital terminal at the start of the interrupt processing of the measurement target and held, and a low level voltage is output to the digital terminal at the end.
When a non-measurement target interrupt process is started, a low-level voltage is output to the digital terminal and held, and at the end of the measurement target interrupt process, an interrupt process having a higher priority than the measurement target interrupt process is in progress. When not being generated, output a high level voltage to the digital terminal,
An execution load factor of the interrupt process of the measurement target is calculated from an average voltage of the digital output terminal in a predetermined period or a length of a period in which a high level voltage is output to the digital terminal in a predetermined period,
Software execution load factor measurement method. A method for measuring an execution load factor of software according to claim 1, comprising:
We have global variables in the software,
In addition to outputting "1" or "0" to the global variable according to the execution state of each interrupt process,
“1” or “0” is output to the global variable according to the execution state of the idling process,
The execution load factor of the software or each interrupt process is calculated by sampling the global variable at regular intervals,
Software execution load factor measurement method. A method for measuring an execution load factor of software according to claim 4,
"1" is output to the global variable during execution of each interrupt process,
"0" is output to the global variable during execution of the idling process,
The execution load factor of the software is calculated from the length of the period in which "1" is output to the global variable in a predetermined period,
Software execution load factor measurement method. A method for measuring an execution load factor of software according to claim 4,
One of the plurality of interrupt processes is the measurement target,
At the start of the interrupt process of the measurement target, "1" is output to the global variable and held, and at the end, "0" is output to the global variable.
When the non-measurement target interrupt process starts, "0" is output to the global variable and is held, and at the end, the measurement target interrupt process is occurring and an interrupt process with a higher priority than the measurement target interrupt process is executed. If not, output "1" to the global variable,
Calculating the execution load factor of the interrupt process of the measurement object from the length of the period in which "1" was output to the global variable in a predetermined period;
Software execution load factor measurement method.