JP2008305238A - Measuring instrument - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To allow a plurality of index times to be measured by a simpler configuration in a measuring instrument for measuring index times showing execution modes of tasks executed by a microcomputer. <P>SOLUTION: One of a plurality of tasks is selected as a measurement object task (object task), and the object task is switched for every intervals of a prescribed time. An index time (wait time) from wake-up to actual execution of the object task, an index time (elapsed time) from actual execution to termination of the object task, an index time (slack time) from termination of execution to next wake-up of the object task, and an index time (execution time) when a CPU is actually assigned to execution of the object task are measured for the prescribed time by providing a timer A for measuring the first four index times and a timer B for measuring the last index time. Thus the plurality of index times can be measured with respect to the plurality of tasks. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a measuring device that measures an index representing an execution mode of a task.

  2. Description of the Related Art Conventionally, for example, a microcomputer mounted on an electronic control device for a vehicle executes a plurality of tasks for controlling a controlled object by using a multitask function by an OS (Operating System). In this type of microcomputer, the number of tasks to be executed increases as control specifications become more complex, and therefore the task processing load tends to increase.

Therefore, program design and debugging are performed to analyze the task execution mode so that the task is executed more efficiently.
In analyzing the task execution mode, the following indices (1) to (5) representing the task execution mode are used.
(1) Waiting time: The time from when a task is woken up until the task is actually started (until the work of the microcomputer CPU is assigned to the task).
(2) Elapsed time: The time from when the task is actually started (after the CPU work is assigned to the task) until the task is completed. For example, when another high-priority task is executed by an interrupt, the time during execution of the interrupt is also included.
(3) Execution time: Total time required only for execution of the target task among the elapsed time. For example, it does not include the execution time of other high priority tasks due to interrupts.
(4) Spare time: The time from when a task is completed until the task is woken up again.
(5) Number of misses: The number of times that a new task related to wake-up is not executed because the number of wake-ups allowed for each task has been exceeded in advance when the task wakes-up.

  For example, Patent Document 1 describes an execution time measuring device that measures the execution time of (3) above. In this execution time measuring device, the execution time can be measured for each of a plurality of tasks with one counter.

Specifically, for example, the execution time of task A and task B is measured as follows. It is assumed that task A is started first.
1. When the execution of task A is started, the counter starts counting.
2. When the execution target is switched from task A to task B, the count value of the counter at that time (count value for task A) is stored in the memory area for task A. For example, the count operation starts from 0.
3. Thereafter, when the execution target is switched from task B to task A, the count value of the counter at that time (count value for task B) is stored in the memory area for task B, and in the memory area for task A. The stored count value is set in the counter, and the counter continues counting from the set count value as the task A starts.
4). The count value of the counter when task A is completed is set as the task A execution time. The same applies to task B.

As described above, even when task execution is intermittent for each of task A and task B, the execution time can be measured with one counter.
For example, Patent Document 2 describes an apparatus for measuring the waiting time of (1), and Patent Document 3 describes an apparatus for measuring the margin time of (4).
JP-A-5-165655 JP 2000-242522 A JP 2006-90356 A

  However, in the execution time measuring device described in Patent Document 1, each time the task to be executed is switched, the execution time of the task whose execution is interrupted is acquired from the counter and stored in the memory. Many processes are required, such as reading the total execution time of the task whose execution is to be resumed from the memory and setting it in the counter. For this reason, the processing load of the execution time measuring device increases.

  Further, since it is necessary to provide a memory area for storing the execution time for each task, when the number of tasks to be measured increases, more memory resources are required.

  Furthermore, in the execution time measuring device of Patent Document 1, when measuring an index other than the execution time, it is necessary to provide a counter and a memory area for each index to be measured. For example, in Patent Document 1, if the waiting time or the margin time is measured as in the device of Patent Document 2 or the device of Patent Document 3, the counter and the memory area for waiting time measurement or margin time measurement are displayed. Further required.

  The present invention has been made in view of such problems, and an object of the present invention is to enable measurement of a plurality of indices with a simpler configuration in a measurement apparatus that measures an index representing an execution mode of a task executed by a microcomputer. To do.

  The measuring device according to claim 1, which has been made to achieve the above object, for each of a plurality of tasks executed by a microcomputer, for each task situation, a period during which the task is under that situation (hereinafter, Is called index time).

  Specifically, a counter that performs a counting operation at regular intervals is provided, and a task to be measured (hereinafter referred to as a target task) every time a predetermined time elapses among a plurality of tasks. Are sequentially selected, and based on the count value of the counter when the status of the target task changes, the index time for the status immediately before the status of the target task changes is calculated.

  In the measuring apparatus according to the first aspect, since the task to be measured (target task) is sequentially selected, the index time for a certain situation can be measured in order for each of the plurality of tasks. Note that there may be one situation type or a plurality of situation types.

  For example, in the measuring device according to claim 1, since the index time of the target task is measured until a predetermined switching condition is reached, the target task can be used until the predetermined switching condition is reached. A plurality of index times can be measured. In other words, until a predetermined switching condition is reached, if a certain situation appears and the situation is completed, the index time for the situation can be measured, and the appearance to the completed situation If there are multiple, the index time can be measured for the multiple situations.

  Therefore, in the measuring apparatus according to the first aspect, it is possible to measure a plurality of index times for each of a plurality of tasks. In addition, it is advantageous because the index time can be calculated with a simple configuration using a counter. Specifically, it is possible to prevent the measuring device from becoming large and the cost from increasing, and for example, the index time can be calculated by a simple process of reading the count value. Can reduce the load.

Here, the measuring device according to claim 1 may be specifically configured as in claim 2.
According to a second aspect of the present invention, there is provided the measuring apparatus according to the first aspect, comprising at least two counters. Then, among the period of time from the start to the end of execution of the target task (hereinafter referred to as elapsed time) and the period of time from the start to the end of the situation, the The period of the situation (hereinafter referred to as execution time) that the work of the CPU of the computer is allocated is measured.

  More specifically, the elapsed time is the time from when the target task is actually started (after the CPU work is assigned to the task) to the end of the target task. For example, other high priority When a task is executed by an interrupt, it is intended to include the time during execution of the interrupt. The execution time is the total time required only for the execution of the target task in the elapsed time, and does not include the execution time of other high priority tasks due to interruption, for example.

Since the elapsed time and the execution time overlap with each other, they cannot be measured simultaneously by one counter.
On the other hand, since the measuring apparatus according to claim 2 includes at least two counters, different counters can be assigned for elapsed time measurement and execution time measurement. It becomes possible to measure in parallel (simultaneously). For example, a complicated configuration and processing are unnecessary, and the elapsed time and execution time can be measured with a simple configuration.

  Next, the measuring device according to claim 3 is the measuring device according to claims 1 and 2, in which the period from the time when the target task is woken up until the execution of the target task is started (hereinafter referred to as waiting time). It is characterized by measuring.

According to the third aspect of the present invention, the waiting time can be measured, and according to this, it is possible to contribute to the analysis of the task execution mode.
Next, in the measuring device according to claim 4, in the measuring device according to claims 1 to 3, after the execution of the target task is completed, the period of time until the target task is woken up (hereinafter referred to as a margin time). It is characterized by measuring.

In the measuring apparatus according to the fourth aspect, the surplus time can be measured, and according to this, it is possible to contribute to the analysis of the task execution mode.
The elapsed time, the waiting time, and the surplus time can be measured by one counter because the periods do not overlap.

  Next, the measuring device according to claim 5 is the measuring device according to claims 1 to 4, wherein each of the plurality of tasks has a predetermined wake-up timing, and the target task is executed at the wake-up timing of the target task. It is characterized by comprising a detecting means for measuring the number of phenomena that are not performed.

  For example, if the number of allowed multiple wakeups registered in advance for the target task has been exceeded at the wakeup timing of the target task, the target task is not executed (so-called “missing” occurs).

  In the measuring apparatus according to claim 5, the number of phenomena that the target task is not executed, that is, a phenomenon that the target task is lost can be measured at the wake-up timing of the target task. Can contribute.

  Next, in the measuring device according to claim 6, in the measuring device according to claims 1 to 5, the index time for the same situation is measured a plurality of times for any one of the plurality of tasks, and the measurement can be performed. The information on the index time is configured so that information that can be measured is stored in an identifiable manner.

  According to this, for example, a change in the length of the index time can be understood in time series. Therefore, statistical verification is possible, which is advantageous for analysis of task execution modes. For example, it is possible to perform more precise verification by processing index time data and organizing the processed data in time series.

  Next, in the measuring device according to claim 7, in the measuring device according to claims 1 to 6, the index time for the same situation is measured a plurality of times for any one of the plurality of tasks, and the measurement can be performed. Of the length of the index time, the maximum value is calculated for each type of task status.

  If the maximum value of the index time is known, it will be possible to determine whether the maximum value is acceptable, adjust the priority for each task so that the maximum value is within the allowable range, It is possible to take measures such as dividing.

  Next, the measurement device according to claim 8 is the measurement device according to claims 1 to 7, and the index time for the same situation is measured a plurality of times for any one of the plurality of tasks, and the measurement can be performed. Of the length of the index time, the minimum value is calculated for each type of task status.

  If the minimum value of the index time is known, it is possible to determine whether or not the minimum value is acceptable as in the case of claim 7, and the priority for each task so that the minimum value is within the allowable range. You can take measures such as adjusting tasks and dividing tasks.

  Next, in the measuring device according to claim 9, in the measuring device according to claims 1 to 8, the index time for the same situation is measured a plurality of times for any one of the plurality of tasks, and the measurement can be performed. An average value of the length of the index time is calculated for each task status type.

The load on the microcomputer changes every moment, and the length of the index time varies greatly. For this reason, it is difficult to perform an effective analysis only with individual index times.
In this regard, in the measurement device according to the ninth aspect, since the average value of the index time is calculated, the task execution mode can be analyzed more effectively. For example, it is possible to determine whether or not the average value is acceptable, and to adjust the priority for each task so that the average value is within the allowable range or to divide the task. Can be dealt with.

  For example, if the average value for the above execution time is calculated for each of a plurality of tasks, the CPU occupancy of the microcomputer can be calculated for each of the tasks. Can be estimated. Further, as described above, it is possible to take measures such as adjusting the priority for each task or dividing the task so that the degree of load on the CPU becomes appropriate.

  Next, the measurement device according to claim 10 is the measurement device according to claims 1 to 9, wherein the index time for the same situation is measured a plurality of times for any one of the plurality of tasks, and the measurement can be performed. Frequency distribution information indicating the length of the index time and the occurrence frequency for each length is output as analysis information.

According to this, the frequency distribution of the length of the index time can be easily understood, and the task execution mode can be more easily analyzed.
Next, the measurement apparatus according to claim 11 is the measurement apparatus according to claims 1 to 10, wherein, for any one of the plurality of tasks, the measurement time for the same situation is measured a plurality of times, and the microcomputer is provided. The load on the CPU changes, and frequency distribution information indicating the length of the index time and the occurrence frequency for each length for each degree of the load on the CPU is output as analysis information. ing.

According to this, the frequency distribution of the length of the index time for each degree of load of the CPU can be understood, and the task execution mode can be easily analyzed.
Next, a measurement device according to a twelfth aspect is the measurement device according to the tenth or eleventh aspect, wherein the analysis information is a graph and the plot points of the graph are color-coded for each type of task. It is said.

According to this, the frequency distribution of the index time is easily visually recognized. In addition, it becomes easier to recognize information for each task.
Next, in the measurement device according to claim 13, in the measurement device according to claims 10 to 12, the analysis information is a graph, and the plot points of the graph are color-coded for each type of task status It is characterized by.

According to this, as in the twelfth aspect, the frequency distribution of the index time is easily visually recognized. Further, it becomes easier to recognize information for each type of task status.
Next, a measuring device according to a fourteenth aspect is characterized in that, in the measuring device according to the first to thirteenth aspects, the switching condition is that a predetermined time elapses.

  Note that the predetermined time may be any length. For example, the same time may be used for each task, or a different time may be used for each task. Furthermore, for example, the length of time may be stored in advance in the memory, or a certain rule or calculation formula is determined, and the time length is calculated each time according to the rule or calculation formula. Also good.

  If the switching condition is that a predetermined time elapses, the setting of the switching condition can be simplified, and it is easy to determine whether or not the switching condition has been reached, thereby simplifying the configuration of the measuring device.

Next, a measuring device according to a fifteenth aspect is characterized in that, in the measuring device according to the first to thirteenth aspects, the switching condition is that the index times are calculated a predetermined number of times.
As for the number of times of calculation, the term “one time” means that each of a plurality of types of index time is counted once when each time is calculated. According to this, regardless of the length of the index time, a plurality of types of index times can be reliably measured. Of course, every time one of a plurality of types of index times is calculated, the number of calculations may be counted as one.

Embodiments of the present invention will be described below with reference to the drawings.
In FIG. 1, the microcomputer 10 and the debugging device 3 correspond to the measuring device of the present invention. In particular, the functions of the software processing unit 12, the hard counter unit 14, the task ID storage unit 16, the task ID rewriting unit 18, and the measurement result storage unit 20 described later of the microcomputer 10 correspond to the measurement device.

  And about the task which CPU which the microcomputer 10 does not illustrate performs, waiting time, elapsed time, execution time, and margin time are measured as index time. The contents of these index times are as described in the description section of the prior art, and the description is omitted here.

  The microcomputer 10 is mounted on an electronic control device (hereinafter referred to as ECU) 2 that controls the vehicle. The debugging device 3 is provided outside the ECU 2 and connected to the ECU 2 so as to be communicable.

Next, the configuration of the microcomputer 10 will be described.
As described above, the microcomputer 10 includes the software processing unit 12, the hard counter unit 14, the task ID storage unit 16, the task ID rewriting unit 18, and the measurement result storage unit 20.

  The software processing unit 12 executes a task under the management of an operating system (hereinafter referred to as OS). The OS installed in the microcomputer 10 is a multitasking OS that performs a plurality of tasks in parallel.

The hard counter unit 14 includes two hard counters 14A and 14B. The hard counters 14A and 14B are configured by hardware as the name suggests.
FIG. 2 shows details of the hard counters 14A and 14B.

  As shown in FIG. 2, each of the hard counters 14A and 14B includes a timer that counts up every predetermined time, a task measurement timer register, and a task measurement timer control register.

  In this example, the task measurement timer register is a 32-bit read-only register, and the count value of the timer is stored in the upper 24 bits as shown in FIG. The lower 8 bits are not used. In this example, 0 is stored as dummy data in the lower 8 bits.

The task measurement timer control register is a register in which information for controlling the operation of the timer is written, and has four areas according to the operation of the timer.
For example, there are four timer operations: stop, start, restart, and reset. Stopping is to stop the count operation while keeping the count value as it is (starting), starting is to continue counting up from the current count value, and restarting returns the count value to 0 The count-up is started from 0, and the reset is to return the count value to 0 and stop the count-up.

  The task measurement timer control register has four areas to which items of stop, start, restart, and reset are respectively assigned. For example, the timer is stopped by writing 0 in the area to which the stop item is assigned. Similarly, the timer starts when 0 is written in the area to which the start item is assigned, and the timer restarts when 0 is written in the area to which the restart item is assigned. The timer is reset by writing 0 in the area to which is assigned.

  In this embodiment, the hardware counter 14A measures the waiting time, elapsed time, and margin time. Then, the execution time is measured by the hard counter 14B. Hereinafter, the timer provided in the hard counter 14A is referred to as timer A, and the timer provided in the hard counter 14B is referred to as timer B.

  The task ID storage unit 16 has a function of storing a task ID for specifying a task as a measurement target for measuring the index time. Here, in this embodiment, a task ID unique to each task is assigned to each task executed by the CPU of the microcomputer 10. Specifically, a serial number is assigned as a task ID for each task. For example, if the number of tasks is 10, numbers 0 to 9 are assigned to the tasks. The task ID storage unit 16 is specifically composed of a RAM.

FIG. 3 is a configuration diagram of the task ID storage unit 16. As illustrated in FIG. 3, the task ID storage unit 16 has an area for storing a task ID to be measured.
The task ID rewriting unit 18 has a function of rewriting the task ID stored in the task ID storage unit 16.

The measurement result storage unit 20 has a function of storing measurement results, that is, information on the measured index time. Specifically, it is composed of a RAM.
FIG. 4 is a configuration diagram of the measurement result storage unit 20.

  As shown in FIG. 4, the measurement result storage unit 20 has areas for storing measurement results for each of the waiting time, elapsed time, execution time, and margin time. This area exists for each task type. In other words, in the measurement result storage unit 20, the index time is stored so as to be identifiable as to which task. In each area, the latest measurement result is stored. Specifically, the measurement results are sequentially overwritten in the measurement result storage unit 20. Further, the debugging device 3 periodically acquires the measurement results stored in the measurement result storage unit 20 and stores the measurement results in an identifiable manner.

  Next, FIG. 5 is a flowchart showing a task ID rewriting process executed by the task ID rewriting unit 18. The task ID rewriting unit 18 rewrites the task ID in the task ID storage unit 16 at predetermined switching time intervals by periodically executing the process of FIG. The length of the switching time is determined in advance. The length of the switching time may be stored in advance in a ROM (not shown) included in the microcomputer 10 or may be stored in the RAM.

  In the process of FIG. 5, first, in S110, it is determined whether or not the length of time measured by a counter (not shown) for time measurement (hereinafter referred to as counter measurement time) is greater than the length of the switching time. To do.

If it is determined in S110 that the counter measurement time is equal to or shorter than the switching time (S110: NO), the process is terminated as it is.
On the other hand, if it is determined in S110 that the counter measurement time is longer than the switching time (S110: YES), the process proceeds to S120.

In S120, the value of a counter (not shown) for time measurement is set to zero. That is, the counter measurement time is reset to zero.
Next, the process proceeds to S130, and interrupts are prohibited. This is to ensure that the next step S140 is executed.

  In subsequent S140, the task ID stored in the task ID storage unit 16 is incremented. That is, the number as the task ID stored in the task ID storage unit 16 is incremented.

  In step S150, it is determined whether the task ID number stored in the task ID storage unit 16 is smaller than the upper limit value of the number of tasks. In other words, it is determined whether each of the plurality of tasks has been designated as a measurement target.

  If it is determined in S150 that the task ID is equal to or greater than the upper limit of the number of tasks (S150 :: NO), it is determined that all of the plurality of tasks are all designated as measurement targets, the process proceeds to S160, and the task ID storage unit 16 Write 0 to Then, the process proceeds to S170 and interrupts are permitted. Thereafter, the process is terminated.

  On the other hand, if it is determined in S150 that the number as the task ID stored in the task ID storage unit 16 is smaller than the upper limit value of the number of tasks (S150 :: YES), the process proceeds to S170 and thereafter the process is terminated. .

Next, FIG. 6 is a flowchart of the wake-up process executed by the software processing unit 12. This wake-up process is started when a task wakes up.
In this process, first, in S210, it is determined whether or not the task ID related to getting up is the same as the task ID stored in the task ID storage unit 16. That is, it is determined whether or not the task related to getting up is a task to be measured.

  If it is determined in S210 that the task ID relating to the wake-up is not the same as the task ID stored in the task ID storage unit 16 (S210: NO), it is determined that the task woken up this time is not a measurement target, and the process ends. .

On the other hand, if it is determined in S210 that the task ID related to getting up is the same as the task ID stored in the task ID storage unit 16 (S210: YES), the process proceeds to S220.
In S220, it is determined whether or not the task is successfully woken up. If it is determined that the task is successful (S220 :: YES), the process proceeds to S230.

  In S230, the count value of the timer A is stored in the measurement result storage unit 20 as a raw value of the margin time. As will be described later, the timer A starts counting up from 0 when the execution of the task is completed (restart). Therefore, the count value of the timer A when the task is woken up is not enough. Represents time.

  Next, in S240, the timer A is restarted. In step S250, the timer B is reset. Thereafter, the process is terminated. Note that the processes of S230 to S250 are executed simultaneously.

On the other hand, if it is determined in S220 that the task wakeup has not succeeded, that is, the task wakeup has failed (S220: NO), the process proceeds to S260.
In S260, it is determined whether or not the task wake-up failure is caused by trying to wake up the task more than a predetermined number of times. In the multitask OS, for example, the number of tasks that can be woken up simultaneously is preset. Further, it is impossible to wake up tasks exceeding the number of tasks that can be woken up at the same time.

  If it is determined in S260 that it is not because the task has been attempted to wake up more than a predetermined number of times (S260: NO), it is determined that some abnormality has occurred, and the process is terminated as it is.

  On the other hand, if it is determined in S260 that the task has been attempted to wake up over a predetermined number of times (S260: YES), it is determined that “missing” has occurred, and the process proceeds to S270. In the present embodiment, “missing” refers to a phenomenon in which a task that has attempted to wake up cannot be woken up because the task has been woken up more times than can be woken up.

In S270, a counter (not shown) for counting the number of omissions is incremented. Thereafter, the process is terminated.
Next, FIG. 7 is a flowchart of a start process executed by the software processing unit 12. This start time process is started when the task is actually started.

  In this process, first, in S310, it is determined whether or not the start task ID is the same as the task ID stored in the task ID storage unit 16. That is, it is determined whether the task to be started this time is a measurement target task.

  If it is determined in S310 that the task ID to be started is not the same as the task ID stored in the task ID storage unit 16 (S310: NO), it is determined that the task to be started this time is not a measurement target, and the process Exit.

On the other hand, if it is determined in S310 that the task ID to be started is the same as the task ID stored in the task ID storage unit 16 (S310: YES), the process proceeds to S320.
In S320, the count value of the timer A is stored in the measurement result storage unit 20 as a raw value of the waiting time. In S240 described above, since the timer A is restarted when the task wakes up, the count value of the timer A when the execution of the task is started represents a waiting time.

  Next, in S330, the timer A is restarted. In step S340, the timer B is restarted. Thereafter, the process is terminated. Note that the processes of S320 to S340 are executed simultaneously.

  Next, FIG. 8 is a flowchart of the ready process executed by the software processing unit 12. In this ready process, when the state of the task to be measured becomes Ready, specifically, the CPU work is assigned to the processing of the task other than the task to be measured, and the task to be measured is executed. It is a flowchart of the process started when it is suspended.

  In this process, first, in S410, it is determined whether or not the task ID related to the suspension of execution is the same as the task ID stored in the task ID storage unit 16. That is, it is determined whether or not the task whose execution is suspended is a task to be measured.

  If it is determined in S410 that the task ID related to the suspension of execution is not the same as the task ID stored in the task ID storage unit 16 (S410: NO), it is determined that the task whose execution is suspended is not a measurement target, The process ends.

On the other hand, if it is determined in S410 that the task ID related to the suspension of execution is the same as the task ID stored in the task ID storage unit 16 (S410: YES), the process proceeds to S420.
In S420, the timer B is stopped. Thereafter, the process is terminated.

Next, FIG. 9 is a flowchart of the restart process executed by the software processing unit 12. This restart processing is started when the processing of a task whose execution has been suspended is restarted.
In this process, first, in S510, it is determined whether or not the task ID related to resumption is the same as the task ID stored in the task ID storage unit 16. That is, it is determined whether or not the task whose processing is restarted this time is a task to be measured.

  If it is determined in S310 that the task ID related to the restart is not the same as the task ID stored in the task ID storage unit 16 (S310: NO), it is determined that the task whose process is to be restarted is not a measurement target, and the process Exit.

On the other hand, if it is determined in S510 that the task ID related to resumption is the same as the task ID stored in the task ID storage unit 16 (S510: YES), the process proceeds to S520.
In S520, timer B is started. As described above, the timer B is stopped when the CPU work is assigned to the processing of the task other than the task to be measured (S420), and the timer B is started when the processing of the task to be measured is resumed (S520). ). In this way, the execution time of the task to be measured is measured.

Next, FIG. 10 is a flowchart of end processing executed by the software processing unit 12. This end-time process is started at the end of task execution.
In this process, first, in S610, it is determined whether or not the task ID related to the end is the same as the task ID stored in the task ID storage unit 16. That is, it is determined whether or not the task to be finished this time is a measurement target task.

  If it is determined in S610 that the task ID related to the end is not the same as the task ID stored in the task ID storage unit 16 (S310: NO), it is determined that the task to be ended this time is not a task to be measured, and the process is performed. finish.

On the other hand, if it is determined in S610 that the ID of the task related to the end is the same as the task ID stored in the task ID storage unit 16 (S610: YES), the process proceeds to S620.
In S620, the count value of the timer A is stored in the measurement result storage unit 20 as a raw value of elapsed time. As described above, when the execution of the task to be measured is actually started, the timer A is restarted (S330). For this reason, the count value of timer A when the execution of the task to be measured ends indicates the elapsed time.

  Next, it progresses to S630 and the count value of the timer B is memorize | stored in the measurement result storage part 20 as a raw value of execution time. As described above, the timer B counts up only during the execution of the measurement target task (S340, S520), and stops during the execution of tasks other than the measurement target task (S420). For this reason, the count value of the timer B represents the execution time.

Next, it progresses to S640 and the timer A is restarted.
Next, in S650, the timer B is reset. Thereafter, the process is terminated. Note that the processes of S620 to S650 are executed simultaneously.

Next, FIGS. 11 to 13 are flowcharts of processing executed in the debugging device 3.
FIG. 11 is a flowchart of a maximum value calculation process for calculating the maximum value of waiting time, elapsed time, execution time, or margin time (hereinafter also referred to as index time together), and FIG. FIG. 13 is a flowchart of an integrated value calculation process for calculating the integrated value of the index time. In the debugging device 3, the maximum value, the minimum value, and the integrated value of the index time are stored. The integrated value calculation process is executed to calculate an average value from the integrated value and the number of integrations. In other words, the integrated value calculation process is a process for calculating an average value.

  In the maximum value calculation process of FIG. 11, first, in S710, it is determined whether or not the current value (latest value) of the raw value of the index time is larger than the stored maximum value of the index time. The current value is acquired from the measurement result storage unit 20. If it is determined that the current value is larger than the stored maximum value (S710: YES), the process proceeds to S720, and the current value is newly stored as the maximum value of the index time. Thereafter, the process is terminated.

On the other hand, if it is determined in S710 that the current value is less than or equal to the stored maximum value (S710: NO), the process is terminated as it is.
Next, in the minimum value calculation process of FIG. 12, first, in S810, it is determined whether or not the current value of the raw value of the index time is smaller than the stored minimum value of the index time. If it is determined that the current value is smaller than the stored minimum value (S810: YES), the process proceeds to S820, and the current value is newly stored as the minimum value of the index time. Thereafter, the process is terminated.

On the other hand, if it is determined in S810 that the current value is greater than or equal to the stored minimum value (S810: NO), the process is terminated as it is.
Next, in the integrated value calculation process of FIG. 13, first, in S910, a counter (not shown) for counting the number of integrations is incremented.

Next, the process proceeds to S920, and the value obtained by adding the current value of the raw value of the index time to the integrated value of the index time up to the previous time (index value integrated value) is set as a new index time integrated value.
Next, proceeding to S930, it is determined whether or not the value of the counter that counts the number of integrations is equal to or greater than a predetermined number of integrations. That is, it is determined whether or not integration has been performed for a specified number of integrations.

  If it is determined that the value of the counter for counting the number of integrations is equal to or greater than the specified number of integrations (S930: YES), the process proceeds to S940, and the index time integration intermediate value calculated in 920 is set as the index time integration value.

Next, in S950, the counter value for counting the number of integrations is reset to zero.
Next, the process proceeds to S960, and the index time integration intermediate value is reset to zero. Thereafter, the process is terminated.

In this case, the average value of the index time can be calculated by dividing the index time integrated value by the specified number of integrations.
On the other hand, if it is determined in S930 that the value of the index time integration counter is smaller than the specified number of integrations (S930: NO), the processing is terminated as it is. In this case, the average value of the index time can be calculated by dividing the intermediate value of the index time by the count value of the counter.

  Next, FIG. 14 is a time chart showing the operation of the present embodiment, and shows the relationship between the index time, the event that occurs in the OS, the state of the task, and the processing executed by the software processing unit 12. ing. In the following description, it is assumed that the task is a measurement target task (hereinafter referred to as a target task). Regarding the task state, “Ready” means the state from wake-up to dispatch (actually assigning CPU work to the task execution), or execution by assigning CPU load work to other tasks. "Running" means that the task is actually being executed, and "Suspended" means that it will be woken up after the task has been executed. It is the state until.

  First, as the target task wakes up, the state of the target task becomes “Ready”, the wake-up process (FIG. 6) is executed, and timer A starts counting from 0 (S210: YES → S220: YES → (S230 ) → S240).

  When the target task is dispatched, the state of the target task becomes “Running”, the start process (FIG. 7) is executed, and the value of timer A at that time is stored in the measurement result storage unit 20 as a raw value of the waiting time. Store (S310: YES → S320).

Further, the timer A and the timer B are restarted (S330, S340).
When a preemption (assignment of CPU work to the execution of a task different from the target task) occurs when the target task status is “Running”, the status of the target task becomes “Ready”, and the ready process (FIG. 8). ) Is executed and the timer B is stopped (S410: YES → S420). At this time, the timer A continues to count up.

  Thereafter, when the target task is dispatched again, the state of the target task becomes “Running”, the restart process (FIG. 9) is executed, and the timer B is started (S510: YES → S520).

  When the execution of the target task ends, the state of the target task becomes “Suspended”, the process at the end (FIG. 10) is executed, and the count value of the timer A is stored in the measurement result storage unit 20 as the raw value of the elapsed time. At the same time (S620), the count value of the timer B is stored in the measurement result storage unit 20 as a raw value of the execution time (S630). Further, the timer A starts counting up from 0 (S640), and the count value of the timer B returns to 0 (S650).

  After that, when the target task is woken up again, the state of the target task becomes “Ready”, the wake-up process (FIG. 7) is executed, and the count value of the timer A is stored in the measurement result storage unit 20 as a raw value of the margin time. It is stored (S210: YES → S220: YES → S230). The subsequent steps are as described above.

  In this way, the waiting time, elapsed time, execution time, and margin time are measured for the target task. Further, the number of omissions is also measured (S270). Further, since the target task is switched every predetermined switching time by the task ID rewriting unit 10, in this embodiment, a plurality of index times are measured for a plurality of tasks by the two timers (hard counters 14A and 14B). Can be done.

  Next, in the present embodiment, the debug device 3 displays a frequency distribution graph representing the relationship between the length of the index time and the frequency of occurrence thereof. 15 and 16 are examples of graphs.

  FIG. 15 is a two-dimensional graph of frequency distribution. In the two-dimensional graph of FIG. 15, the frequency distribution of the waiting time is represented for each of the three types of tasks S, T, and U. The X axis is the waiting time (μs), and the Y axis is the number of occurrences per second (times / 1 s).

  In FIG. 15, the plot points are color-coded for each task type. Also, the shape of the plot points is made different for each task. Specifically, the shape of the plot point of task S is represented by □, the shape of the plot point of task T is represented by ◯, and the shape of the plot point of task U is represented by △.

This makes it easy to visually recognize.
Next, FIG. 16 is a three-dimensional graph of frequency distribution. In FIG. 16, the frequency distribution of the waiting time of task S is shown. The X axis is waiting time (μs), the Z axis is the number of occurrences per second (times / 1 s), and the Y axis is time. In this example, the relationship between the waiting time and the number of occurrences at a predetermined time from time 1 to time 15 is shown. Here, the load on the CPU changes every moment, and the load on the CPU differs from time 1 to time 15. Specifically, the load on the CPU is the smallest at time 1, the load on the CPU is sequentially increasing toward time 15, and the load on the CPU at time 15 is the largest. That is, the three-dimensional graph of FIG. 16 represents the relationship between the length of the waiting time and the occurrence frequency for each length for each degree of CPU load.

  As described above, in this embodiment, by sequentially switching the measurement target task (target task), the index time (specifically, waiting time, elapsed time, execution) is sequentially performed for each of the plurality of tasks. Time and margin time). Therefore, a plurality of index times can be measured for each of a plurality of tasks. In particular, two timers for measuring the index time are provided, and the elapsed time and execution time can be measured simultaneously. In addition, the index time can be measured by a simple process of configuring the timer by hardware and reading the value of the timer, which is advantageous because a complicated configuration and process are unnecessary.

  Moreover, in the said embodiment, since it tried to wake up a task exceeding predetermined number of times, the frequency | count that failed to wake up can also be counted and measured. According to such a measuring device, it can contribute to the analysis of the task execution mode.

  In the above-described embodiment, the index time is stored in the debug device 3 so as to be identifiable when the index time can be measured. Further, the maximum value, minimum value, and integrated value (average value) of the index time are calculated. According to this, the user of the measuring device can statistically verify the index time, and can analyze the task execution mode in more detail. This makes it possible to design a more optimal program. For example, it becomes easy to take measures such as adjusting the priority for each task or dividing the task.

  Moreover, in the said embodiment, since the frequency distribution graph about the length of index | exponent time is displayed, it becomes easy for the user of a measuring device to recognize the execution aspect of a task visually.

In the above embodiment, the process of S270 corresponds to a detection unit.
As mentioned above, although one Embodiment of this invention was described, this invention is not limited to the said embodiment, A various form can be taken within the technical scope of this invention.

  For example, in the above embodiment, FIGS. 15 and 16 show the frequency distribution for the waiting time, but may show the frequency distribution for other index times. When the frequency distribution of a plurality of index times is shown on one graph, the plot points are preferably color-coded for each index time. Then, it becomes easier to recognize visually.

  In the above embodiment, the timer B is reset when the task is woken up (S250). However, if the time is from when the task is woken up until the execution of the task is actually started, the timer B is reset. The reset timing of B may be anytime. Further, the timer B may be reset and the count-up may be started at the task execution start timing.

  In the above embodiment, the timer B is reset at the timing when the execution of the task is completed (S650), but from the end of the execution of the task until the actual execution of the task is started. If so, the reset timing of the timer B may be anytime. Further, the timer B may be reset and the count-up may be started at the task execution start timing.

  In the above embodiment, the maximum value calculation process (FIG. 11), the minimum value calculation process (FIG. 12), and the integrated value calculation process (FIG. 13) may be configured to be executed by the software processing unit 12, respectively. good.

  Moreover, in the said embodiment, although only the newest measured value of index time is memorize | stored in the measurement result storage part 20, you may make it memorize | store the measured value for multiple times, for example. In this case, the old data may be replaced with new data in order.

In the above embodiment, the task ID is not limited to a serial number, and may be data in any format.
Moreover, in the said embodiment, the structure which does not perform the process of S130 and S170 may be sufficient.

  In the above embodiment, a task having a higher priority than the task related to wake-up is being executed, and the task with the higher priority does not end until the next wake-up timing of the task related to wake-up comes. The number of times that the execution of the previous wake-up is canceled may be counted.

In the above embodiment, the function of the hard counter unit 14 may be configured by software.
In the above embodiment, the maximum value calculation process of FIG. 11 only needs to calculate at least the maximum values of the waiting time, the elapsed time, and the execution time, and the maximum value of the margin time may not be calculated. This is because no particular inconvenience occurs even if the margin time increases on the system. Rather, it can be said that it is preferable that the margin time is large.

  In the above embodiment, in the minimum value calculation process of FIG. 12, it is sufficient that at least the minimum value of elapsed time, execution time, and margin time can be calculated, and the minimum value of waiting time may not be calculated. This is because there is no particular inconvenience even if the minimum waiting time is reduced on the system. Rather, it can be said that a smaller waiting time is preferable.

  In the above embodiment, the target task is switched every predetermined switching time. However, for the target task, the target task is switched before the calculation time of the index time reaches the predetermined calculation number. May be. Specifically, when the waiting time, elapsed time, execution time, margin time, and number of omissions are measured, the number of calculations is counted as one, and the target task is switched when the number of calculations reaches a predetermined number. Anyway. Of course, the number of times of calculation may be counted as one when at least one of the waiting time, elapsed time, execution time, margin time, and number of omissions is measured.

It is a figure showing the structure of the measuring device of this embodiment. It is explanatory drawing of the hard counter of the hard counter part. 4 is an explanatory diagram of a task ID storage unit 16. FIG. It is explanatory drawing of the measurement result storage part. It is a flowchart of the task ID rewriting process which the task ID rewriting part 18 performs. It is a flowchart of the process at the time of wakeup which the software process part 12 performs. It is a flowchart of the start time process which the software process part 12 performs. It is a flowchart of the process at the time of ready which the software process part 12 performs. It is a flowchart of the process at the time of restart which the software process part 12 performs. It is a flowchart of the process at the time of completion | finish which the software process part 12 performs. 6 is a flowchart of a maximum value calculation process executed in the debugging device 3. 6 is a flowchart of a minimum value calculation process executed in the debugging device 3. 5 is a flowchart of integrated value calculation processing executed in the debugging device 3; It is a time chart showing the effect | action of this embodiment. 3 is a two-dimensional graph displayed on the debugging device 3; 3 is a three-dimensional graph displayed on the debugging device 3;

Explanation of symbols

DESCRIPTION OF SYMBOLS 2 ... ECU, 3 ... Debugging device, 10 ... Microcomputer, 12 ... Software processing part, 14 ... Hard counter part, 14A, 14B ... Hard counter, 16 ... Task ID memory | storage part, 18 ... Task ID rewriting part, 20 ... Measurement result Storage.

Claims (15)

For each of a plurality of tasks executed by the microcomputer, a measuring device that measures a period (hereinafter referred to as an index time) during which the task is under the status for each status of the task,
It has a counter that counts every certain time,
Each time one of the plurality of tasks reaches a predetermined switching condition, the task is sequentially selected as a measurement target task (hereinafter referred to as a target task), and the status of the target task changes. The measurement apparatus is characterized in that, based on the count value of the counter at the time, the index time for the situation immediately before the situation of the target task changes is calculated. The measuring device according to claim 1,
Comprising at least two counters;
Measure the period from the start to the end of execution of the target task and the period from the start to the end of the period to which the work of the microcomputer CPU is allocated for the execution of the target task A measuring device characterized by: In the measuring device according to claim 1 or 2,
A measuring apparatus that measures a period from when the target task is woken up to when execution of the target task is started. In the measuring device according to any one of claims 1 to 3,
A measuring apparatus for measuring a period from when execution of the target task is completed until the target task is woken up next time. In the measuring device according to any one of claims 1 to 4,
Each of the plurality of tasks has a predetermined wake-up timing,
A measuring apparatus comprising: a detecting unit that measures the number of times that the target task is not executed at the wake-up timing of the target task. In the measuring device according to any one of claims 1 to 5,
For any one of the plurality of tasks, the index time for the same situation is measured a plurality of times, and the index time information that can be measured is stored so that it can be identified when the information can be measured. It is comprised so that the measuring device characterized by the above-mentioned. In the measuring device according to any one of claims 1 to 6,
Among the plurality of tasks, for any one of the tasks, the index time for the same situation is measured a plurality of times, and the maximum value of the measured index time length is determined for each task status type. A measuring device characterized by being calculated. In the measuring device according to any one of claims 1 to 7,
Among the plurality of tasks, for any one of the tasks, the index time for the same situation is measured a plurality of times, and the minimum value of the measured index time length is determined for each task status type. A measuring device characterized by being calculated. In the measuring device according to any one of claims 1 to 8,
For any one of the plurality of tasks, the index time for the same situation is measured a plurality of times, and an average value of the measured index times is calculated for each type of task situation A measuring device characterized by that. In the measuring device according to any one of claims 1 to 9,
Among the plurality of tasks, the frequency distribution information that represents the length of the index time that can be measured and the occurrence frequency for each length, while measuring the index time for the same situation multiple times for any one of the tasks Is output as information for analysis. In the measuring device according to any one of claims 1 to 10,
For any one of the plurality of tasks, the index time for the same situation is measured a plurality of times, and the CPU load of the microcomputer is changed, and the degree of the CPU load A frequency measuring apparatus that outputs frequency distribution information representing the length of each index time and the occurrence frequency for each length as analysis information. In the measuring device according to claim 10 or 11,
The analysis information is a graph, and plot points of the graph are color-coded for each type of task. The measuring device according to any one of claims 10 to 12,
The analysis information is a graph, and plot points of the graph are color-coded for each type of task status. In the measuring device according to any one of claims 1 to 13,
The switching device is characterized in that a predetermined time elapses. In the measuring device according to any one of claims 1 to 13,
The switching device is characterized in that the calculation time of the index time is a predetermined number.