JP2001312413A - Method for retrial control in access collision and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase the success rate of access in retrial processing through simple processing if plural programs collide when accessing a common resource. SOLUTION: A program periodically accesses the common resource 2 to update it. A program 1B optionally accesses the common resource 2 to update it. When access from the program 1B collides against access from the program 1A, the program retries access after a retrial time set in a retrial timer 4 elapses. When the retrial is successful, a counter which counts a retrial frequency 3 is cleared. If the retrial fails, the set value of the retrial timer 4 is varied through simple arithmetic processing for dividing the set value by a constant or subtracting a constant and access is retried while the retrial time is varied until the retrial frequency is exceeded, so the success rate of the retrial can be increased.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a retry control method at the time of access contention and a recording medium for enabling a plurality of programs to operate in parallel while sharing resources in multitasking or the like.

[0002]

2. Description of the Related Art In a computer system capable of multitasking, one processor executes a plurality of programs in parallel. Multiple programs running in parallel
Resources such as the processor itself that executes the program, various input / output devices, and memory are shared. Depending on the operating system, the processor operates by switching a plurality of programs little by little in time, or switches to the operation of another program when one program is waiting for an event and input, etc. Perform parallel operation of the program.

[0003] When a shared resource such as an input / output device is accessed, the purpose of the access is not always achieved immediately. For example, CD-ROM and DVD-RO
When reading information recorded on a recording medium such as an M, the preparation for reading the recording medium is started after an access request is made, and until the information can be actually read from the recording medium. It may take some time. A program that accesses such a resource needs to wait until it becomes readable after once making an access request, during which time another program can operate. However, when another program accesses a resource for which an access request has already been issued, access conflict occurs, and some adjustment is required for smooth execution of the program. In general, until the previously performed access is completed, the earlier access takes precedence, and the later accessed program needs to change the time and retry to access again. Further, when a plurality of programs simultaneously access the shared resource, none of the programs can complete the access, and it is necessary to perform a retry so that the access does not overlap with the access of another program. In order to simplify the retry process, a method of retrying at regular time intervals can be considered.

As an example in which a plurality of accesses compete for a common resource, a CSMA / CD (Carrier Sense Multiple) for smoothly using a bus among a plurality of nodes sharing one communication line as a bus. Access With
Collision Detection) method is LAN (Local Area Net)
work) is widely known. In the CSMA / CD system, when one node uses the bus, it is first determined whether or not communication has already been performed on the bus. When it is determined that the bus is free, communication is started, but it is also monitored whether the communication collides with communication from another node. When the bus is free, multiple nodes may start communication at the same time, in which case the communication will collide. Upon detecting a communication collision, each node temporarily stops communication. Therefore, although the communication line becomes empty again, if the same communication is started again from each node, there is a high possibility that collision will occur again, and it will not be possible to communicate even after repeating collision and standby. there is a possibility. For this reason, in order to change the time interval at which each node starts the next communication, for example, by using a random number, the standby time is changed at random, so that the possibility of collision when starting the next communication is reduced. ing. Such a concept of randomly changing the retry waiting time can be generally applied at the time of contention for access to a shared resource.

Japanese Patent Application Laid-Open No. Hei 9-114750 discloses that a retry interval when an abnormality is detected on a system bus is defined as:
The concept of switching between values specified in advance in a plurality of registers or switching to a value specified by an output of a bus monitoring control circuit is shown.

[0006]

In a computer system in which a plurality of programs operate in parallel in a multitask, a program that periodically accesses a shared resource at predetermined time intervals may be operating. In such a case, if an access conflict occurs when another program tries to access the same shared resource, when retrying at a fixed time interval, the time interval and the periodic access cycle match, and Even if retry is repeated, it may not be possible to access the shared resource. If the time interval of the retry is randomly changed as in the CSMA / CD system, the possibility of successful access in the retry can be increased. However, if the retry time is set at random, it is difficult to make the number of retries correspond to the sum of the retry times. Even if the number of retries is limited, the time is undefined, and if the retry time is specified, the number of retries is undefined. . Also, processing for generating random numbers is required.

Further, when the retry time is switched as in the prior art of Japanese Patent Application Laid-Open No. Hei 9-114750, it is necessary that the retry time to be switched does not coincide with the access cycle of a program that accesses periodically. For this purpose, it is necessary to acquire information about the cycle in advance and create a program, and there are great restrictions and burdens when creating the program.

An object of the present invention is to provide a retry control method and a recording medium at the time of access conflict in which the success rate of retry can be easily improved and the relationship between the number of times of retry and time can be accurately defined. .

[0009]

SUMMARY OF THE INVENTION According to the present invention, in an environment in which a plurality of programs can operate in parallel, at least one program periodically accesses a shared resource.
This is a retry control method at the time of access conflict which is performed when another program arbitrarily accesses and a conflict occurs. When a conflict occurs at the first access, a retry is performed after a predetermined retry time elapses, and even at the time of the retry, the access conflict occurs. Is a retry control method at the time of access contention in which the retry time is changed by a predetermined constant arithmetic processing, and the next retry is controlled after the changed retry time elapses.

According to the present invention, when at least one of a plurality of programs operating in parallel is accessing the shared resource periodically, another program accesses the shared resource. However, at the time of access conflict, retry is performed after a predetermined retry time has elapsed. Even if the access cannot be performed in conflict with the periodic access at the time of the first access, the timing of the periodic access and the access is shifted at the time of the retry, thereby increasing the possibility that the access can be performed. . Further, when an access conflict also occurs during a retry, there is a possibility that the initially determined retry time coincides with the period of the periodic access to the shared resource, and the retry time is changed by a predetermined arithmetic processing. By doing so, access conflict does not occur at the time of the next retry, and it is possible to increase the probability of being able to access reliably.

Further, in the present invention, when the access conflict occurs even in the second and subsequent retries, control is performed such that the predetermined arithmetic processing is performed again for the retry time.

According to the present invention, when an access conflict occurs at the time of the first access and an access conflict occurs at the time of the retry, and the retry is performed a plurality of times, the retry time, which is a waiting time between each retry, is predetermined. , The retry time can be changed for each retry, and the success rate of the retry can be improved. In addition, since the change in the retry time is performed according to a predetermined calculation process, the retry time changes within a predictable range, and the total retry time can be easily calculated regardless of how many times the retry is repeated. it can. Since the retry time is changed at each retry, the possibility of successful access at the time of retry can be increased.

In the present invention, the predetermined arithmetic processing is division by a constant.

According to the present invention, since the retry time is changed by division by a constant, the retry time is reduced by the constant, so that it is difficult to compete with periodic access.
In addition, it is possible to easily grasp the retry time.

Further, in the present invention, the predetermined arithmetic processing is a subtraction of a constant.

According to the present invention, when the retry is repeated, the retry time is decremented by a constant. Therefore, even if contention with periodic access occurs, the possibility of success in the next retry can be changed by changing the retry time. Can be enhanced. Since the retry time is changed by subtracting a constant, the correspondence between the number of retries and the retry time can be easily grasped.

Further, the present invention is a computer-readable recording medium in which the above-described retry control method at the time of access conflict is recorded as a program for causing a computer to execute the method.

According to the present invention, a plurality of programs can be operated in parallel by a computer, and access competition for a shared resource can be efficiently adjusted.

[0019]

FIG. 1 shows a schematic configuration of a system for executing a retry control method at the time of access contention as one embodiment of the present invention. A plurality of programs 1A and 1B operate in parallel and can access the shared resource 2 together. One program 1A
Periodically accesses the shared resource 2 and updates its content. The other program 1B arbitrarily accesses the shared resource 2 and updates the content. When the program 1B accesses the shared resource 2 and the access conflicts with the program 1A, the program 1B counts the number of retries 3 and sets a retry time, which is a retry waiting time, in the retry timer 4. The present invention can be similarly applied to a case where three or more programs operate in parallel and access the shared resource 2.

FIG. 2 shows a periodic update operation for the shared resource 2 by the program 1A shown in FIG. The operation is started from step a1, and in step a2, the shared resource 2 is accessed to acquire a semaphore. The shared resource 2 is assumed to be accessed from a plurality of programs, and employs a semaphore method to adjust access competition. The semaphore method is a method in which the program that has access for the first time acquires the semaphore from the state where there is no access from any program, and thereafter, the access is permitted only from the program that has acquired the semaphore, and the access ends Then, the semaphore is released, and the system returns to the state of waiting for access again. That is, if one program has acquired a semaphore, that program can exclusively access the shared resource.

In step a3, it is determined whether or not the semaphore has been successfully acquired. If succeeded, the shared resource is accessed in step a4, and the content is updated. When the update processing of the shared resource by the access ends, the semaphore is released in step a5. Step a5
After the semaphore release processing is completed, or
If it is determined in step 3 that the semaphore acquisition has not succeeded, the process shifts to waiting for a certain time in step a6. When the waiting for the predetermined time in step a6 ends, the process returns to step a2. That is,
The program 1A accesses the shared resource at regular intervals, acquires the semaphore, updates the shared resource if the access is successful, and attempts to access again after the next fixed time if the acquisition of the semaphore is not successful. Access to the shared resource 2 at regular time intervals.

FIG. 3 shows an operation of accessing the shared resource 2 by the program 1B of FIG. The operation is started from step b1, and an event is waited in step b2.
In the event waiting state, the operation of another program, for example, the program 1A, is actually performed, and the program 1B
Is a state of waiting for a request interrupt or an interrupt from the retry timer 4. The request interrupt is performed by the operating system, for example, according to a request from a user of the computer system. When an event such as a request interrupt or an interrupt from the retry timer 4 occurs, access to the shared resource 2 is required as part of the program processing accompanying the event, and a semaphore is obtained in step b3.

In step b4, it is determined whether or not the semaphore acquisition has been successful. When it is determined that the acquisition of the semaphore is successful, the retry count 3 is cleared in step b5, the shared resource 2 is accessed and updated in step b6, and the semaphore is released in step b7. If it is determined in step b4 that acquisition of the semaphore has not been successful, a counter for counting the number of retries 3 is incremented in step b8.

At step b9, it is determined whether or not the number of retries 3 exceeds the maximum value set in advance. When it is determined that the number of times has not been exceeded, the value of the retry timer 4 is calculated in step b10, and the value of the retry timer 4 is calculated in step b11.
And the retry timer 4 is started. When the retry timer 4 starts, an event is issued after a specified time according to the calculation result of step b10. When it is determined in step b9 that the number of times is excessive, the number of retries 3 is cleared in step b12, and step b1
In step 3, retry count excess processing is performed. Upon completion of the processing in the step b7, the step b11, or the step b13, the process returns to the step b2.

When returning to the event wait of step b2 after the release of the semaphore in step b7, it waits for the next request interrupt. When waiting for an event in step b2 after the start of the retry timer in step b11, the system waits for an interrupt from the retry timer 4 as an event. In the retry count excess process of step b13, it is determined that the shared resource 2 cannot be accessed.
It performs processing for notifying the user.

As shown in FIG. 2, the program 1A periodically accesses the shared resource 2 according to the following procedure. 1) Acquire a semaphore before accessing. 2) Do not access when semaphore acquisition fails. 3) If semaphore acquisition is successful, access shared resource 2 and release the semaphore after termination.

When an event including an access to the shared resource 2 occurs, the program 1B accesses according to the following procedure. 1) Acquire a semaphore before accessing. 2) If the semaphore acquisition is successful, access the shared resource 2 and release the semaphore after the end. 3) When semaphore acquisition fails, first retry timer 4
To wait for the time. The first retry time is n1 seconds. 4) Acquire a semaphore. 5) If semaphore acquisition is successful, access shared resource 2 and release the semaphore after termination. 6) When semaphore acquisition fails, the second retry timer 4
To wait for the time. The second retry time is n2 seconds. 7) Acquire a semaphore. 8) If semaphore acquisition is successful, access shared resource 2 and release the semaphore after termination. 9) When semaphore acquisition fails, the third retry timer 4
Wait for the set value of. The set value of the retry timer 4 is n3 seconds. 10) Acquire a semaphore. 11) When semaphore acquisition is successful, access shared resource 2 and release the semaphore after completion. 12) If the semaphore acquisition fails, perform the error processing with the retry count exceeded.

In the above procedure, when the shared resource 2 cannot be accessed even after performing the retry up to three times, it is determined that the retry number has been exceeded. In this case, the time until the retry count is exceeded can be defined as x by the following equation. x = n1 + n2 + n3

The setting of the retry timer 4 can be changed as follows. 1) Method of dividing by a constant Example: If the constant is 2, n1 = 10, n2 = 5, n3 =
2.5, and x = 17.5. 2) Method of dividing a constant Example: Assuming that the constant is 2, n1 = 10, n2 = 8, n3 =
6 and x = 24.

If the set value of the retry timer 4 is changed and the retry time is changed by these simple arithmetic operations, the repetition of the number of retries can be performed without knowing the period for the periodic access from the program 1A. Can increase the retry success rate.

The concept of the computer device for accessing the shared resource 2 by the multitask method as described above can be applied to, for example, an in-vehicle navigation device. In the navigation device, a program for periodically detecting the current position of a vehicle at regular time intervals accesses a recording medium that stores road map data, and uses a road map to detect the current position based on dead reckoning navigation or map matching. Use data. Further, a road map centering on the current position is displayed on the display screen based on the read road map data. The user of the navigation device
For example, when requesting to view a map of a specific point, a program for reading map data in response to a user's request is activated. This program accesses a recording medium of road map data, which is a shared resource, at an arbitrary timing according to a user's request, and displays the read map data on a display screen if the access is successful.

[0032] Since the display screen of the in-vehicle navigation device is limited, the display for displaying images is also a shared resource. When periodically displaying the current position on the display screen, the entire screen is used for displaying the current position. When the user is required to display information about a specific place, the display of the current position and the display screen are performed in parallel, or the display of the map data requested by the user is prioritized and the entire display screen is displayed. Or use it. However, the processing for displaying on the display screen is
If you access, you can respond immediately, so in the case of multitasking that the processor is also a shared resource,
In particular, there is no need for semaphore management.

For access to a recording medium, since it takes time from the access request to the actual reading of the necessary map data, it is necessary to perform management to avoid access conflicts by using a semaphore or the like. Without exclusive management, for example, in an access request to detect the current location,
During the preparation for reading the map data, an access is made from the user for a request for road information or the like at a specific point, road data of the current position is given to the access, and then the reading of the current position is performed. When there is a request, if the road map data of another point requested by the user is given,
A malfunction will occur.

The environment in which a plurality of programs operate in parallel is not limited to a computer device that performs multitasking with a single processor, but may be a multiprocessor computer system in which a plurality of processors operate simultaneously,
This may occur between a plurality of computer systems interconnected by a LAN or the like. For example, many electronic control devices are mounted on vehicles today in a distributed manner, and each device is connected by a communication line. By connecting a plurality of electronic control devices to be mounted with a common communication line, the number of cables required as a wire harness or the like can be reduced, and the weight can be reduced. For data transmission via such a communication cable, for example, various types of data necessary for vehicle operation are transmitted at a constant cycle, an operation unit and a display unit are arranged near a driver's seat, and many recording media are stored. Control in an audio-visual system in which a replaceable changer, a high-output amplifier, and the like are disposed in a trunk room or the like can also be applied to a case where the control is performed via a communication cable. That is,
The use of a communication cable as an audiovisual device may be performed in a time interval between communications for periodically operating a vehicle, and in the event of contention for access to a communication line as a shared resource, the following as an audiovisual device. If the retry waiting time is changed by applying the present invention, the success rate of the retry can be increased and the communication line can be used effectively.

[0035]

As described above, according to the present invention, a program for arbitrarily accessing a shared resource causes a conflict with an access from a program that periodically accesses the shared resource when the shared resource is accessed, and a predetermined period of time. When a conflict occurs even after retry after the elapse, the next retry is performed after the elapse of the retry time after performing a predetermined fixed arithmetic processing, so the possibility of contention with periodic access is reduced and the retry success rate is improved. Can be done. The change in the retry time is performed not by random numbers but by a predetermined arithmetic processing, so that processing for generating random numbers is unnecessary, and the waiting time required for retry can be clarified by simple processing.

Further, according to the present invention, since retry is repeatedly performed while changing the retry time by a predetermined constant arithmetic processing, it is not necessary to use a random number, and it is easy to grasp the number of times of retry and the retry time. Can be.

Further, according to the present invention, it is possible to increase the success rate of retry and easily grasp the correspondence between the number of retries and the retry time by a simple operation of dividing by a constant.

Further, according to the present invention, the success rate of retry can be increased and the correspondence between the number of retries and the retry time can be easily denied by a simple operation of subtracting a constant.

Further, according to the present invention, a plurality of programs can be operated in parallel by the computer, and the adjustment at the time of contention for access to the shared resource can be efficiently performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic system configuration as one embodiment of the present invention.

FIG. 2 is a flowchart showing an operation procedure of a program 1A in FIG.

FIG. 3 is a flowchart showing an operation procedure of a program 1B of FIG. 1;

[Explanation of symbols]

 1A, 1B program 2 shared resource 3 retry count 4 retry timer

Claims (5)

[Claims] 1. An access conflict that occurs when another program arbitrarily accesses a shared resource that is periodically accessed by at least one program in an environment where a plurality of programs can operate in parallel, and a conflict occurs. When contention occurs in the first access, retry is performed after a predetermined retry time elapses. A retry control method at the time of access contention in which the next retry is controlled after the changed retry time has elapsed. 2. The system according to claim 1, wherein when the access conflict occurs even in the second and subsequent retries, control is performed such that the predetermined arithmetic processing is performed again on the retry time. Retry control method at the time of access conflict. 3. The method according to claim 1, wherein the predetermined arithmetic processing is division by a constant. 4. The retry control method at the time of access contention according to claim 1, wherein said predetermined arithmetic processing is a subtraction of a constant. 5. A computer-readable recording medium in which a retry control method at the time of access conflict according to claim 1 is recorded as a program for causing a computer to execute the method.