JPH07200279A - Object management system and network management system - Google Patents

Abstract

PURPOSE:To reduce places being the objects of a switching-over processing accompanying an update processing and to reduce processing quantity by providing a repeating pointer between a reference source and a reference destination and indirectly referring to the reference destination at the reference source. CONSTITUTION:A regular processing process 201 indirectly refers to instance data 401 through an instance repeating pointer 301 and indirectly refers to a former generation method code 602 or a new generation method code 612 through a generation method repeating pointer 341. Instance data 401 indirectly refers to a former class method table 501 or a new class method table 511 through a method repeating pointer 321. A selection setting means 111 switches over the reference destination between the former class method table 501 and the new class method table 511 by rewriting the content of the method repeating pointer 321, and switches over the reference destination between the former generation method code 602 and the new generation method code 612 by rewriting the content of the generation method repeating pointer 341.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object management system and a network management system.

[0002]

2. Description of the Related Art In object orientation, an instance means a concrete object entity, data that holds the state of the instance, and a plurality of procedures (methods) for performing operations such as data reference and update. It consists of a procedure table (procedure table) that lists the storage areas for procedures. Instances that have the same data format and procedure belong to the same class. The data holding the instance state must exist for each instance, but procedures and procedure tables usually exist for each class. The property of having to use the procedure of the instance itself to refer to and update the data of the instance from outside the instance is called encapsulation. The modularity of the program is improved by using the encapsulated objects.

In a system that is used continuously for a long time such as a network control system, it is desirable that functions and services can be added or changed during operation.
For this reason, it is essential to update the control program during operation. By making the network control program an object-oriented program that consists of a collection of objects, you can utilize the characteristics of objects that encapsulate procedures (methods) and data to realize program updates in units of operating objects. There is a possibility. Here, updating the program in object units means old procedure tables, procedures, and / or
It also means replacing the data with a new procedure table, procedure code and / or data, including both class-wise updates and instance-wise updates.

As a conventional technique for updating the object-oriented software during operation, for example, "Switching Systems Research Group 91-78" Program Update in Object-Oriented Exchange Program "of the Institute of Electronics, Information and Communication Engineers," IS
S'92 Dynamic Software Conf
Iguration Management in Br
oadband Telecommunication
"Environments" and the like.

The former procedure will be described with reference to FIGS. 3 and 4.

In the first conventional example shown in FIG. 3, the new class method (procedure) table and the new method (procedure) code are loaded into the memory, and the procedure of the old instance (instance of the old class) is not executed. In the state, the procedure of switching the reference to the old method table to the new method table by rewriting the pointer to the class method table held by the instance data while exclusively controlling the execution of the normal processing process was performed.

In the second conventional example shown in FIG. 4, the new class method (procedure) table and the new method (procedure) code are loaded into the memory, and the procedure of the old instance (instance of the old class) is not executed stably. In the state, the procedure to inherit the state from the old instance to the new instance while exchanging the execution of the normal processing process and rewrite the pointer to the instance of the normal processing process to switch the reference to the old instance to the new instance Was going on.

Since the program update processing of the system which requires continuous operability is executed in parallel with the normally executed processing, it is desired to be executed without hindering the execution of the normal processing. Therefore, it is desirable that the load of update processing is small. Also, an object-oriented program
I aim to create a program by combining objects freely. Therefore, the update processing program does not depend on the knowledge of how the object to be updated and the object that refers to it are combined,
It is desirable to be able to create in a form that depends only on the structure of the object to be updated.

Further, in the above-mentioned conventional technique for updating object-oriented software, exclusive control of the current instance is performed in a stable state where the procedure of the current instance is not executed, and the state of the current instance is inherited to the state of the new instance. , The procedure for switching the reference to the current instance to the new instance is shown.

FIG. 19 shows an example of the configuration of these methods.
The processing execution means 901 calls the update processing means 910.
The called update processing unit 910 is the current instance 9
If 60 is stable, that is, if there is a normal processing means that is accessing the current instance at that time, it is judged by the flag 962, and if it is stable, that is, if the flag 962 is empty, By setting the flag 962, exclusive control of the current instance 960 is performed, the state 961 of the current instance 960 is inherited to the state 971 of the new instance 970, and the access destination of the normal processing means is switched from the current instance 960 to the new instance 970.

In a communication network composed of a plurality of terminals and communication devices, a network management station (NMS) 1572 shown in FIG. 49 has a manager function means 1581 shown in FIG. 50 and a management information storage means 1582 shown in FIG. The protocol processing unit 1583 of FIG. 50 is provided, and the terminal 1573 of FIG. 49 or the communication device 1574 of FIG. 49 is the agent function unit 1591 of FIG.
51, and the management information storage means 1592 of FIG. 51 and the protocol processing means 1593 of FIG. The manager function unit 1581 of FIG. 50 performs management operation on the agent function unit 1591 of FIG. 51 using the network management protocol to manage the management target in the management area. Further, the agent function means 1591 of FIG. 51 continuously manages individual communication devices or terminals, and FIG.
The result of the management operation from the manager function means 1581 is returned to the manager function means 1581 in FIG.

Here, the terminal 15 of FIG. 49 managed by the network management station 1572 of FIG. 49.
73 or the communication device 1574 of FIG. 49, the agent function means 1591 of FIG. 51 notifies the manager function means 1581 of FIG. 50 of the failure information, and the administrator notifies the network management station 1 of FIG. 49.
The failure is dealt with by obtaining the failure information through 572.

[0013]

However, in the above-mentioned conventional technique, firstly, by rewriting the pointer held by the reference source (instance data, normal processing process) that references the object to be updated, the old and new Since switching was performed, it is necessary to rewrite for the number of reference sources. Since one update unit (data, procedure table, and / or procedure code) is generally referred to from a plurality of places, the number of rewriting processes is larger than the number of update units. Further, the update processing program needs to know all the reference sources, and it is difficult to create an update processing program that depends only on the structure of the object to be updated and does not depend on the program structure of the reference source.
Since an object may be referred to by a plurality of types of normal processing programs, it has been necessary to create an update processing program that is compatible with all normal processing programs that can be the reference source.

Secondly, in the conventional real-time system requiring continuous operability, it is guaranteed that the time from the occurrence time of an asynchronously occurring event to the processing end time is smaller than a certain range. There must be. However, in the conventional program update method in operation, despite performing exclusive control for the current instance,
The method of guaranteeing the real-time property as described above has not been considered. Therefore, the activation of the update process breaks the real-time property of the normal process that requires real-time property, lowers the service quality of the system, and sometimes the service itself of the system becomes difficult to provide.

Thirdly, in the above-mentioned conventional technique, a network management station for managing terminals and communication devices to be managed is arranged, and an administrator manages the management target in the management area through the network management station. I managed it, but with the sophistication and complexity of the network, it is possible that the administrator will manage multiple networks at multiple network management stations,
In that case, the administrator does not always perform the management operation at the network management station that manages the management target in which the failure has occurred. It is also possible that the administrator cannot be stationed at the network management station due to various reasons, and in that case, the administrator cannot quickly perform fault management when a fault occurs in the management target.

On the other hand, conventionally, there is a network management station provided with means for notifying a failure by using an electronic mail, a pager, etc., but only a terminal / pager etc. registered in the management station. Only the failure notification was given, and the failure could not be dealt with, that is, the failure could not be managed.

The object management system of the present invention has been made by paying attention to such a problem. The object is to reduce the processing amount at the time of updating an object, and An object management system that facilitates the creation of an object update processing program that depends only on the update target and not on the reference source.
Another object of the present invention is to provide an object management system capable of providing a software update method that guarantees real-time performance and continuous operability.

A third object of the present invention is to provide a network management system capable of managing the network quickly and efficiently without being resident at the network management station in charge of the manager. .

[0019]

In order to achieve the above-mentioned first object, the present invention provides a plurality of data for holding an internal state of an object and a plurality of data for implementing processing based on these data. In the object management system for managing the object, which is made up of the procedure code and the plurality of procedure tables indicating the storage area contents of these procedure codes, the plurality of data identified by the first single identifier A first selection means for selecting one of the plurality of procedure tables identified by a second single identifier, and a third selection means for selecting one of the plurality of procedure tables identified by the second single identifier. At least one of third selecting means for selecting one of the plurality of procedure codes selected by one identifier is provided, and the first selecting means is provided. Wherein the plurality of data, the second of said plurality of procedures table selecting means should select to be-option,
The third selection means includes setting means for setting at least one of the plurality of procedure codes to be selected.

Further, in order to achieve the above second object, the present invention provides an object which is a collection of states and procedure contents constituting software existing in a computer system composed of one or more processors. In the managed object management system, one or more normal processing means for providing a predetermined service of the computer system, and a state matching between a current instance of the class of the object and a new instance of the class or another class. Of the series of processes for detecting and switching between the instances, the update processing means for processing at least a part and the end time defined for the execution of at least a part of the normal processing means are at least arbitrary. Execution or update processing of the normal processing means so as to be guaranteed by probability. And assurance means for adjusting at least one relative execution process of the execution means comprises an execution unit for executing a coordinated process.

In order to achieve the above third object, the present invention provides a network management system including a communication network including a plurality of communication devices and terminals, and a network management station that manages the communication network. , The first to manage a communication network
This network management station recognizes, at a certain point in time, when a terminal managed by another network management station that has not been accessed fails, and the recognition means that recognizes the terminal that is accessing at that time. In order for the specified terminal to respond to the failure,
And a setting means for setting this specific terminal as a second network management station.

[0022]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a first principle diagram according to the first embodiment of the present invention, and FIG. 2 is a second principle diagram according to the first embodiment of the present invention.
3 is a first conventional example, FIG. 4 is a second conventional example, FIG. 5 is a first embodiment of the present invention, FIG. 6 is a first data structure of instances and classes, and FIG. No. 2 data structure, FIG. 8 is a second data structure of an instance and a class being updated, FIG. 9 is a third data structure of an instance and a class, and FIG. 10 is a third data structure of an instance and a class being updated.
Data structure, FIG. 11 is an update procedure classification table, FIG. 12 is a flowchart of the entire update process, FIG. 13 is a flowchart of the update process body of the first procedure, and FIG. 14 is a flowchart of the update process body of the second procedure. 15 is a flowchart of the update processing body of the third procedure.

A first embodiment according to the first embodiment of the present invention shown in FIG.
The principle diagram of is the update processing process 101 that has the selection setting means 111 and executes the object update processing, the normal processing process 201 that executes the normal processing of the system, the instance relay pointer 301, and the method relay pointer 32.
1, a generated method relay pointer 341, instance data 401 holding the state of an instance, an old class method table 501 that is a list of storage areas of method codes of the old class, and a list of storage areas of method codes of the new class. A certain new class method table 511, old method code 601 that is an implementation of the method of the old class, old generated method code 602 that secures the storage area of the instance of the old class and initializes the area, and new implementation of the method of the new class A method code 611 and a new generation method code 612 that secures a storage area for an instance of a new class and initializes the area.

The normal processing process 201 indirectly refers to the instance data 401 via the instance relay pointer 301 and indirectly refers to the old generation method code 602 or the new generation method code 612 via the generation method relay pointer 341. Instance data 40
1 is indirectly the old class method table 501 or the new class method table 511 via the method relay pointer 321.
Refer to. The selection setting unit 111 switches the reference destination between the old class method table 501 and the new class method table 511 by rewriting the content of the method relay pointer 321 and also the generated method relay pointer 341.
The reference destination is switched between the old generation method code 602 and the new generation method code 612 by rewriting the contents of the above.

Second embodiment according to the first embodiment of the present invention shown in FIG.
The principle diagram of is the update processing process 101 having the selection setting means 111 for executing the object update processing, the normal processing process 201 for executing the normal processing of the system, the instance relay pointer 301, the old method relay pointer 3
21 (The old method relay pointer indicates the same entity as the method relay pointer of FIG. 1, and therefore the same number is assigned. Hereinafter, when there is no distinction between old and new and the old one is the same entity, The same number is assigned in the same way), new method relay pointer 331, generated method relay pointer 341, old instance data 401 holding the state of the old instance, new instance data 411 holding the state of the new instance, old An old class method table 501 that is a list of storage areas for method codes of a class, a new class method table 511 that is a list of storage areas for method codes of a new class, an old method code 601 that is an implementation of a method of an old class, Old generation method that secures the storage area of the old class instance and initializes the area Over de 602, the new method code 611 is the implementation of the method of the new class, and a new generation method code 612 to initialize the secure and space a storage area for an instance of the new class.

The normal processing process 201 indirectly receives the old instance data 401 or the new instance data 411 via the instance relay pointer 301 and indirectly the old generation method code 602 or the new generation method via the generation method relay pointer 341. Code 612
Refer to. The old instance data 401 indirectly references the old class method table 501 via the old method relay pointer 321, and the new instance data 411 indirectly references the new class method table 511 via the new method relay pointer 331. The selection setting unit 111 switches the reference destination between the old instance data 401 and the new instance 411 by rewriting the content of the instance relay pointer 301, and rewrites the content of the generation method relay pointer 341 to change the reference destination. Switch between old generated method code 602 and new generated method code 612.

The first conventional example shown in FIG. 3 has an update processing process 101 having selection setting means 111 for executing object update processing, a normal processing process 201 for executing normal processing of the system, an instance state, etc. Instance data 401 to be held, old class method table 501, which is a list of storage areas for method codes of old classes,
It is composed of a new class method table 511 which is a list of storage areas of method codes of the new class, an old method code 601 which is implementation of the method of the old class, and a new method code 611 which is implementation of the method of the new class. The normal process 201 directly sends the instance data 40
Refer to 1. The instance data 401 directly corresponds to the old class method table 501 or the new class method table 51.
Refer to 1. The selection setting unit 111 switches the reference destination between the old class method table 501 and the new class method table 511 by rewriting the content of the instance data 401.

The second conventional example shown in FIG. 4 has an update processing process 101 having selection setting means 111 for executing object update processing, a normal processing process 201 for executing normal processing of the system, a state of an old instance, etc. The old instance data 401 that holds the new instance data, the new instance data 411 that holds the state of the new instance, the old class method table 501 that is a list of the storage areas of the method code of the old class, and the list of the storage area of the method code of the new class A new class method table 511 which is a table, an old method code 601 which is an implementation of a method of an old class, and a new method code 611 which is an implementation of a method of a new class. The normal process 201 directly sends the old instance data 401 or the new instance data 411.
Refer to. The old instance data 401 directly refers to the old class method table 501, and the new instance data 411 directly refers to the new class method table 511. The selection setting unit 111 switches the reference destination between the old instance data 401 and the new instance 411 by rewriting the contents of the normal processing process 201.

The first embodiment of the invention shown in FIG. 5 is a detailed version of FIG. Specifically, in the update process,
Area acquisition / release means 112, state inheritance means 113 for referring to old instance data and setting new instance data accordingly, new class method table 511, new method code 611, new generation method code 612 from a disk or the like not shown. New class loading means 114 for loading a new class into the storage area, old and new method relay pointers 321 and 331 of the specified class, and old and new method table 501.5.
11, acquisition of class data for acquiring addresses of storage areas such as generation method relay pointer 321, old and new generation method codes 612 and 622, instance relay pointer 301 and the like, size of area required for storage, and number of instances belonging to a class Means 115, updating means deciding means 116 for deciding an object updating procedure according to the object updating condition described below, given update scheduled end time, schedule of the normal processing currently being executed, etc. , Schedule information such as priority, guarantee schedule 7
A schedule determining means 117 to be registered in 02 is added.

Further, each relay pointer 301, 321, 3
Reference counters 302 and 322 for 31 and 341, respectively
・ 332 ・ 342 and update counter 303 ・ 323 ・ 33
3.343 was added. The reference counter is a counter that is incremented when the relay pointer or its reference destination is referenced by the normal processing process or the update processing process, and is decremented when the reference is completed. The update counter is a counter that is incremented when the relay pointer or its reference destination is updated by the normal processing process or the update processing process, and is decremented when the update is completed. Further, according to the guarantee schedule 702, an operating system 701 for controlling execution of the update processing process 101 and the normal processing process 201 is added.

FIG. 6 shows a first data structure of instances and classes. It is possible to apply the update whose principle is shown in FIG. 1 or FIG. 2, and at the time of update, the data structure shown in FIG. 1 or FIG.

FIG. 7 shows a second data structure of instances and classes. Instance relay pointer 3 compared to FIG.
Since 01 is omitted, only the update whose principle is shown in FIG. 1 can be applied. At the time of updating, it becomes the second data structure of the instance and data being updated shown in FIG.

FIG. 9 shows a third data structure of instances and classes. Method relay pointer 321 compared to FIG.
Is omitted, only the update whose principle is shown in FIG. 2 is applicable. At the time of updating, it becomes the third data structure of the instance and data being updated shown in FIG.

FIG. 11 is a classification table of update procedures according to the object update conditions. The classification table will be described below.

The conditions for updating the object-oriented program in object units are listed below. Object update procedures are classified according to the combination of these conditions. (A: Correspondence between old and new classes) (A1) Addition: The old class does not exist, but the new class is added.

(A2) Delete: Delete the old class. There is no corresponding new class.

(A3) One-to-one update: old class and new class are one
Corresponding to 1 and updating the old class with the new class.

(A4) n-to-m update: n kinds of old classes correspond to m kinds of new classes, and the old classes are updated with new classes. (B: Method interface) (B1) No change: The method type and parameters are not changed, and the operation of the method seen from the caller side is not changed.

(B2) Addition: The type of method and / or parameter is added. However, the behavior of the method seen from the caller does not change when only the methods and parameters common to the old class and new class are used.

(B3) Delete: The method type and / or parameter is deleted. However, the behavior of the method seen from the caller does not change when only the methods and parameters common to the old class and new class are used.

(B4) Other changes: Changes in method interfaces other than the above. (C: Internal state of old instance) (C1) None: The instance of the old class does not have internal state data other than the method storage location related data.

(C2) Yes: The instance of the old class has internal state data other than the method storage position related data. (D: Life of old instance) (D1) Shorter than update period: The instance of the old class that existed at the start of object update disappears in normal processing by the end of object update.

(D2) Longer than the update period: The instance of the old class that existed at the start of the object update cannot be expected to disappear in the normal processing by the end of the object update. (E: Old and new instances coexist) (E1) None: It is guaranteed that during the object update period, there is no period during which normal processing is accessing the old class instance and the new class instance.

(E2) Yes: During the object update period, there may be a period during which normal processing is accessing an instance of the old class and an instance of the new class. (F: Change of internal state data format) (F1) None: The data structure of the internal state held by the old instance and the data structure of the internal state held by the new instance are equivalent.

(F2) Addition: The data structure of the internal state held by the new instance includes the data structure of the internal state held by the old instance.

(F3) Other changes: The data structure of the internal state held by the new instance does not include the data structure of the internal state held by the old instance.

The combination of these conditions has the following restrictions. In the case of adding a class (A1): Since the interface of the method is always added, only the condition (B2) is allowed.
Since the instance of the old class does not exist, condition D,
E and F have no meaning. In the case of class deletion (A2): Since the interface of the method is always deleted, condition B only (B3) is allowed.
Conditions E and F have no meaning because there is no new instance. For one-to-one class update (A3): Limit the method interface to no change (B1) and add (B2).

However, even with deletion (B3) and other changes (B4),
If the normal process existing at the time of update uses only the method interface of the new class, (B
Interpret it as included in 1) or (B2) and proceed with the subsequent discussion. When the instance has no internal state (C2): Since the internal state cannot be changed, the condition F is limited to (F1) or (F2).

The object update processing procedure is divided into the following stages. (p1) Allocating and loading area for new class (p2) Switching the reference destination of method from old class to new class. (p3) Switch instance generation method from old class to new class. (p4) Create new class instance. (p5) Handing over the old instance state to the new instance. (p6) Switch the reference destination from the old instance to the new instance. (p7) Release old instance area. (p8) Open old class area.

The restrictions on the combination of the object update condition and the object update procedure are listed below. In the case of adding a class (A1): (p3) does not change the instantiation method that the normal process refers to from the old class to the new class, but simply makes the normal process refer to the instantiation method of the new class. Interpreted as setting. The old class and its instances do not exist, so
(p2), (p5), (p6), (p7), and (p8) are unnecessary. In the case of class deletion (A2): Since the new class and its instance do not exist, (p1) to (p6) are unnecessary. (p8) is mandatory. In case of one-to-one class update (A3): (p8) is mandatory. If the life of the old instance is shorter than the renewal period (D1):
If the life of the old instance is shorter than the update period, the old instance is naturally deleted by normal processing and a new instance is generated, so the old instance is updated to the new instance, so (p2), ( p5),
(p6) is unnecessary. (p4) and (p7) are performed not by the update process but by the normal process. If the life of the old instance is shorter than the renewal period (D1):
When the life of the old instance is shorter than the update period, it cannot be expected that the old instance will be updated to the new instance by normal processing. Therefore, it is necessary to forcibly switch from the old class to the new class, (p3) or
The processing of (p6) is essential. In the case of (E2) where old and new instances coexist: In the process of (p2), the reference destinations of the methods of all the instances of the update target class are switched from the old class to the new class at the same time, so that the old and new instances do not coexist. When the internal state data format is changed (F3): (p5) is required because the state of the old instance needs to be inherited by the new instance. Accordingly, (p6) and (p7) are essential.

The relationships between the stages of the update procedure are shown below. (p1) accompanies (p3): After loading the new class, it is necessary to make it possible to refer to the instantiation method of the new class from normal processing. (p2) and (p5), (p6), and (p7) are exclusive: (p2) is used when switching only the old and new method parts while leaving the instance part as it is. Therefore, the inheritance of instance data (p5) is meaningless, and the instance switching (p6) is unnecessary. On the contrary, if the instance is switched from new to old by (p6), the instance will refer to the new method part, so there is no need to perform (p2) separately. Since (p6) is accompanied by (p7) as shown below,
(p2) and (p5), (p6), and (p7) are exclusive. (p6) is accompanied by (p7): When switching instances (p6),
Since the old instance area will not be released by normal processing, it is necessary to release the old instance area by update processing (p7). (p5) is accompanied by (p6) and (p7): Inheritance of instance data (p5) necessarily involves switching of instances (p6).

The update procedure classification table shown in FIG. 11 lists the object update conditions and required procedures according to the above-mentioned restrictions and relationships. In FIG. 11, ∘ indicates a procedure for update processing, Δ indicates a procedure for normal processing, and × indicates an unnecessary procedure.

The relationship between the data structure of the object shown in FIG. 6, FIG. 7 or FIG. 9 and the above update procedure will be described. FIG. 6 is a general-purpose configuration applicable to any of the update processing procedures shown in FIG. FIG. 7 shows a degenerate configuration applicable when the update process does not include (p6) and FIG. 9 is applicable when the update process does not include (p2).

The flow of update processing will be described below.

FIG. 12 shows the overall flow of this embodiment. In step 101, an update target, an update condition, etc. are designated. The update target is specified in the form of a class or a specific instance of a class. The latter is effective when it is desired to replace only a part with a new instance for program testing or the like. The update condition is used to determine the update procedure, and the update condition (for example, the above-mentioned conditions A to F), the update procedure classification number (for example, the classification number shown in FIG. 11), or the content of the update procedure to be executed. The item itself (for example, the necessary items (p1) to (p8) listed above are listed in the order of execution) is designated. Furthermore, a scheduled update end time that can be used to determine the activation cycle of execution of the update processing process 101, priority, etc. may be specified.

In step 102, the update procedure determining means 11
6 is applied to the update condition, and the update procedure to be executed in the update processing body of step 103 is determined.

In step 103, an update processing body, which will be described in detail below, is executed according to the determined update procedure. Details are shown in FIG. 13, FIG. 14 or FIG. 15 for each update procedure.

In the following, the flow will be described by roughly classifying the update procedure shown in FIG. 11 into three. The first update procedure shown in FIG. 13 represents the procedure of g3-1 in FIG. 11, and by omitting a part of this procedure as shown in FIG. 16, g1-1,
Express g2-1. The second updating procedure shown in FIG. 14 represents the procedure of g4-1. The third updating procedure shown in FIG. 15 is g4-3.
The above procedure is represented, and g4-2 is expressed by omitting a part of this procedure as shown in FIG.

The procedure of FIG. 13 will be described below. This procedure is applicable to the data structures shown in FIGS. 6, 7, and 9. (Step 201) The class data acquisition means 115 is used to acquire data relating to the class designated in Step 101. The data required here includes the address of the storage area of the generated method relay pointer of the class, the new method relay pointer 331, the new class method table 511, the new method code 611, and the area size required to store the newly generated method code 612. Is. (Step 202) The execution schedule of the update processing process 101 is decided using the schedule decision means 117. In the first update procedure, step 204 is performed by blocking execution of the normal processing process 201.
This does not prevent the real-time processing of the normal processing process 201, the update is completed at the scheduled update end time given in step 101, and the schedule of the currently executed normal processing process 201, etc. The operating system 701 is requested to determine schedule information such as the execution cycle and priority of the update processing process 101 and execute the schedule of the update processing process 101 based on the schedule information. That is, the determined schedule is set as the guarantee schedule 702. (Step 203) Using the area acquisition / release means 112,
The new method relay pointer 331 obtained in step 201,
New class method table 511, new method code 611,
A memory area for loading a new class is secured according to the size of the new generation method code 612, and the new class loading means 114 is used to create a new class method table 511 on the disk, network, etc., new method code 6
11. Load the newly generated method code 612 into the secured memory area. (Step 204) The selection setting means 111 is used to rewrite the contents of the generation method relay pointer 341 to the address of the new generation method code 612. This rewriting needs to be performed by blocking the execution of the normal processing process 201 that uses the update target class. Further, the blocking must not prevent the normal processing process 201 from being impaired in real time. As a consideration for this, at the start of step 204, the priority is set high so that the update processing process 101 is not interrupted by the normal processing process 201.
The block time is minimized by allowing the switching process using the selection setting means by the update process 101 to be continuously executed. Generation method relay pointer 3
The rewriting of 41 is performed after confirming that the reference counter 342 and the update counter 343 of the generation method relay pointer 341 are 0 and are not being used by the normal processing process 101 at that time. If it cannot be rewritten, the priority of the update processing process 101 is lowered and the execution right is passed to the normal processing process 201 using the generation method relay pointer 341, and then the update processing process 1 is executed.
When the execution right is assigned to 01, step 204 is executed. (Step 205) Confirm that all old instances have been deleted. The implementation method may be to periodically call the class data acquisition unit 115 and refer to the number of instances of the old class. Alternatively, as a function of the old class, it may be possible to notify when the number of instances becomes 0. Note that the deletion of the old instance itself is naturally performed as the processing of the normal processing process 101 progresses. Step 206) After confirming that all instances of the old class have been deleted, the old method relay pointer 32
1, old class method table 501, old method code 60
1. The memory area storing the old generated method code 602 is released.

The procedure of FIG. 14 will be described below. This procedure can be applied to the data structures shown in FIGS. (Step 301) The class data acquisition means 115 is used to acquire data relating to the class designated in Step 101. The data required here are the method relay pointer of the class, the address of the storage area of the generated method relay pointer, the new class method table 511, the new method code 611, and the area size required to store the new generated method code 612. . (Step 302) The schedule determining means 117 is used to determine the execution schedule of the update processing process 101. In the first update procedure, step 304 and step 305 are performed by blocking the execution of the normal processing process, so that the real time property of the normal processing process 201 is not hindered, and the update end schedule given in step 101 is scheduled. Schedule information such as the execution cycle and priority of the update processing process 101 is determined in consideration of the completion of the update at time and the schedule of the currently executed normal processing process 201, and the update processing is performed based on this. Request the operating system 701 to execute the schedule of process 101. That is, the determined schedule is set as the guarantee schedule 702. (Step 303) Using the area acquisition / release means 112,
A memory area for loading the new class is secured according to the sizes of the method table 511 of the new class obtained in step 301, the new method code 611, and the newly generated method code 612, and the new class loading means 114 is used to create a disk. Method of new class on top, network etc. Table 5
11, the new method code 611 and the new generated method code 612 are loaded into the secured memory area. (Step 304) The switching means 111 is used to rewrite the content of the method relay pointer 321 to the address of the new class method table 511. This rewriting needs to be performed by blocking the execution of the normal processing process 201 that uses the update target class. The method of not blocking the real-time property of the normal processing process 201 by blocking is the same as that of step 204. (Step 305) The contents of the generation method relay pointer 341 are rewritten to the address of the new generation method code 612 by using the selection setting means 111. This rewriting needs to be performed by blocking the execution of the normal processing process 201 that uses the update target class. The method of not blocking the real-time property of the normal processing process 201 by blocking is the same as that of step 204.

Step 306) Old class method table 50
1, the memory area storing the old method code 601 and the old generated method code 602 is released.

The procedure of FIG. 15 will be described below. This procedure can be applied to the data structures shown in FIGS. (Step 401) The class data acquisition means 115 is used to acquire data relating to the class designated in Step 101. The data required here is the storage address of the instance relay pointer 301 of the class, the storage area of the generated method relay pointer 341, the new method relay pointer 331, the new class method table 511, the new method code 611, and the new generated method code 612. Area size, number of instances of the class, etc. (Step 402) The execution schedule of the update processing process 101 is decided by using the schedule decision means 117. In the first update procedure, step 204 is performed by blocking execution of the normal processing process 201.
This does not hinder the real-time property of the normal processing process 201, that the update ends at the scheduled update end time given in step 101, the schedule of the currently executed normal processing process, and the number of instances that must be updated. In consideration of the above, the operating system 701 is requested to determine the schedule information such as the execution cycle and priority of the update processing process 101 and execute the schedule of the update processing process based on the schedule information. That is, the determined schedule is the guaranteed schedule 70
Set to 2. (Step 403) Using the area acquisition / release means 112,
New method relay pointer 331 of new class obtained in step 301, new method table 511, new method code 61
1. A memory area for loading a new class is secured according to the size of the newly generated method code 612, and a new class loading method is used to create a new class method table 511 on the disk or on the network. 61
1. Load the newly generated method code 612 in the secured memory area. (Step 404) The contents of the generation method relay pointer 341 are rewritten to the address of the new generation method code 612 by using the selection setting means 111. This rewriting needs to be performed by blocking the execution of the normal processing process 201 that uses the update target class. The method of not blocking the real-time property of the normal processing process 201 by blocking is the same as that of step 204.

Step 405) Class data acquisition means 1
15, the instance of the old class existing at that time is selected as the processing target of the following steps. Specifically, the address of the storage area of the instance relay pointer 301 is obtained. A plurality of old instances may be selected at a time so that processing efficiency can be improved by collectively processing a plurality of old instances in each step. (Step 406) If an instance of the old class can be selected, Step 407 is executed, and if not (that is, if all the instances have been deleted), Step 411 is branched to. (Step 407) The new generation method is called to generate a new class instance corresponding to the old class instance selected in step 405. As a result, the area of the new instance data 411 is allocated, and necessary initialization (setting of the address value of the new method relay pointer 331, etc.) is performed. (Step 408) The state inheritance means 113 is used to refer to the current state held in the old instance data 401 and inherit it as the state of the new instance data 411. The takeover process may be a simple copy or may involve data conversion. (Step 409) The contents of the instance relay pointer 301 are rewritten to the address of the new instance data 411 by using the selection setting means 111. This rewriting needs to be performed by blocking the execution of the normal processing process 101 that uses the update target instance. The method of not blocking the real-time property of the normal processing process 201 by blocking is the same as that of step 204. (Step 410) The area of the old instance data 401 that is no longer referenced after switching in step 409
The area is acquired and released by using the area acquisition / release means 112. (Step 411) After confirming that all the instances of the old class have been deleted in step 406, the memory area storing the old method relay pointer 321, the old class method table 501, the old method code 601, and the old generated method code 602. Open up.

The second embodiment of the present invention will be described in detail below with reference to the drawings.

The block diagram of the second embodiment of the object management system shown in FIG.
1, guarantee schedule 802, update processing means 810, normal processing means 820, one or more states 861 and flags 86
2 current instances 860, one or more states 8
It is composed of a new instance 870 composed of 72 and a flag 872.

A block diagram of a conventional example of the object management system shown in FIG. 19 is a process execution means 901, an update processing means 910, a normal processing means 920, and one or more states 96.
1 and current instance 960 consisting of flag 962, 1
It consists of a new instance 970 consisting of one or more states 972 and flags 972.

The flow chart of the schedule executing means of the second embodiment shown in FIG. 20 is from step 111 to step 120.
Composed by.

The flow chart of the acceptance processing for execution of the normal processing means of the second embodiment shown in FIG. 21 comprises steps 211 to 216.

The flow chart of the acceptance processing for execution of the update processing means of the second embodiment shown in FIG. 22 is step 311 and step 3
It is composed of 12.

The flowchart of the flag processing of the second embodiment shown in FIG. 23 is composed of steps 421 to 430.

The flow chart of the update processing means of the second embodiment shown in FIG. 24 comprises steps 501 to 507.

The flow chart of the normal processing means of the second embodiment shown in FIG. 25 comprises steps 601 to 604.

FIG. 26 is a block diagram of a first modified example of the object management system according to the second embodiment. Schedule execution means 1301, guarantee schedule 1302,
Schedule determination means 1303, switching processing means 1311
And update processing means 13 including state matching processing means 1312
10, normal processing means 1320, switching processing amount estimation means 13
31 and the state coincidence processing amount estimation means 1332, the update processing amount estimation means 1330 and the shortest activation period acquisition means 134.
0, processor load estimation means 1350, threshold value determination means 1355, state change counter 1361, state reference counter 1362, state coincidence flag 1363, state change flag 1364, state reference flag 1365 and state 1366.
A current instance 1360 consisting of a state change flag 1374, a state reference flag 1375, and a state 137.
It is composed of a new instance 1370 of 6.

The flow chart of the schedule executing means of the first modified example shown in FIG. 27 is from step 701 to step 714.
Composed by.

The flow chart of the acceptance processing for executing the normal processing means of the first modification shown in FIG. 28 is composed of steps 801 to 805.

FIG. 29 shows a flow chart of the scheduling decision process of the execution of the update process means of the first modified example.
01 to step 908.

The flow chart of the state change flag processing of the first modification shown in FIG. 30 is from step 1001 to step 101.
It is composed of three.

The flow chart of the state reference flag process of the first modification shown in FIG. 31 is from step 1101 to step 111.
It is composed of 0s.

The flow chart of the state match flag processing of the first modification shown in FIG. 32 is from step 1201 to step 120.
It is composed of 5.

The flow chart of the update processing means of the first modification shown in FIG. 33 comprises steps 1301 to 1308.

The flow chart of the state matching processing means of the first modification shown in FIG. 34 is from step 1401 to step 1405.
Composed by.

The flow chart of the switching processing means of the first modification shown in FIG. 35 comprises steps 1501 to 1509.

The flow chart of the normal processing means of the first modification shown in FIG. 36 comprises steps 1601 to 1604.

The flow chart of the update processing by the normal processing means of the first modification shown in FIG. 37 is composed of steps 1701 to 1704.

FIG. 38 is a block diagram of a second modified example of the object management system according to the second embodiment. The schedule execution means 1401, the guarantee schedule 1402, the update processing means 1410, the normal processing means 1420, and one or more states. 1461, counter 1462, and update flag 14
63 current instances 1460, one or more states 1472, counters 1472 and update flags 1473.
It is composed of a new instance 1470.

The flow chart of the schedule executing means of the second modification shown in FIG. 39 comprises steps 1801 to 1808.

The flow chart of the update processing means of the second modification shown in FIG. 40 comprises steps 1901 to 1908.

The flow chart of the reference processing by the normal processing means of the second modification shown in FIG. 41 is composed of steps 2001 to 2004.

The flow chart of the update processing of the normal processing means of the second modification shown in FIG. 42 is from step 2101 to step 2
It is constituted by 106.

Regarding the object management systems of the second embodiment and the first and second modified examples configured as described above,
The operation will be described below with reference to the drawings.

In the second embodiment and the first and second modifications, the objects currently used are provided during the operation of the real-time computer system managed by the object management system of the present invention. The old class is replaced by a new class that includes the state definition and procedure definition of the old class and has an upward compatibility relationship with the old class, and the current instance created as an instance of the old class is replaced by the new class. Assume the process of updating to a new instance which is an instance of.

The second embodiment will be described below.

FIG. 20 shows the flow of the schedule executing means 801.

At step 111, it is judged whether or not it is an execution request for the new normal processing means 820. In that case, step 112 is executed, otherwise step 113 is executed.

At step 112, the new normal processing means 820 receives an execution request. See the description of FIG. 21 for details. After the end, step 119 is executed.

At step 113, it is judged whether or not it is an execution request for the new normal processing means 810. In that case, step 114 is executed, otherwise step 115 is executed.

At step 114, the process of accepting the execution request of the new normal processing means 810 is performed. For details, see the description of FIG. After the end, step 119 is executed.

At step 115, it is determined whether or not the flag processing is being performed. If it is flag processing, step 116 is executed, otherwise step 117 is executed.

At step 116, flag processing is performed.
The normal processing unit 820 or the update processing unit 810 acquires or releases the flag 862. See Figure 23 for details.
checking ... After the end, step 119 is executed.

At step 117, it is judged whether or not the timer processing is being performed. If it is a timer process, step 118 is executed, otherwise step 119 is executed.

At step 118, timer processing is performed.
This process is a process executed at the time of a timer interrupt of the hardware, and by referring to the current time, the normal processing means 820 or the update processing means 810 to be activated at that time is selected from the registered contents of the guarantee schedule 802. , Put the vehicle in a waiting state. After the end, step 119 is executed.

At step 119, the guarantee schedule 8
With reference to the registered contents of 02, the one having the highest priority is selected as the target to be executed next from all the normal processing means 820 or the update processing means 810 in the traveling state or the traveling waiting state. After the end, the selected normal processing means 820 or update processing means 810 is executed.

FIG. 21 shows the contents of the acceptance processing of the normal processing means 820 in step 112.

At step 211, the activation cycle T _i of the normal processing means 820 and the processing amount L _i per activation are designated, and the acceptance processing of the execution request of the new normal processing means 820 is started. The processing amount is expressed, for example, by dividing the number of instructions of the processor executed by one activation of the normal processing means 820 by the number of processing instructions per unit time of the processor.

At step 212, the execution loads U _p of all currently executed normal processing means 820 are acquired. This is obtained by summing the usage rates L _i / T _i of the declared processors of the normal processing means 820. In this case, it is theoretically possible to completely guarantee real-time property as described later.

[0107]

[Equation 1] It can also be obtained by actually measuring the processor load. In this case, real-time property is guaranteed stochastically.

At step 213, the threshold value U _T is acquired. The background for determining the threshold value U _T will be described.

Late monotonic scheduling, deadline driven scheduling, and the like are known as algorithms for deciding a schedule that guarantees real-time performance (CLLiu, JW Layland, "Scheduling Algorithm).
ms for Multiprogramming ina Hard-Real-Time Environ
ment ", J. ACM Vol.20, No.1, pp.46-61). In each case, the process has a priority, and when a request to execute a process with a high priority occurs, In the late monotonic scheduling, the higher the priority, the shorter the activation cycle of the process (task) is given to the processor that is assigned to execute the process with the lower priority. .. The priority is basically fixed. With deadline driven scheduling, the time when processing is required to end is
Give the highest priority to the one closest to the present. Therefore, the priority is dynamically changed. In these algorithms, when a process activation period T _i and a process amount L _i per period are given and the processes are numbered consecutively from 1 to n,

[0110]

[Equation 2] It has been proved that the real-time property can be guaranteed if is satisfied. U _M is a late monotonic scheduling

[0111]

[Equation 3] With deadline driven scheduling,

[0112]

[Equation 4] Is. Actually, it is possible to consider the process switching time, but for simplicity, it will be ignored here.

Based on the above, the threshold value U _T is set as follows, for example. The execution of the update processing means 810 has a portion (critical portion) that must be performed by continuously blocking the execution of the i-th normal processing means 820. Let the processing amount of this part be L _c . When the i-th normal processing means 820 is blocked, the i-th normal processing means 820
Is executed after waiting for the completion of the processing of the critical portion, so it can be considered that this processing amount has become L _i + L _c in practice. For example, L _c is considered to be a value obtained by adding a fixed processing amount to the product of the processing amount of the critical part per instance set and the product of the number N of instances of the update processing means 810 at the same time. Also, the normal processing means 820 that may be blocked
There is even several there, the shortest period of these T _c
And The processor usage rate of the normal processing means 820 having the shortest cycle is (L _i + L _c ) / T _c in the worst case.

In some cases, the update processing means 810 has a part that can execute the normal processing means 820 without blocking it. The processing amount per cycle of the update processing means 810 including this is L _N, and the activation cycle of the update processing means 810 is T _N. However, all target instances may be processed in one cycle. If the load for executing the update processing means is shifted to the right side of (Expression 2), the following expression is obtained.

[0115]

[Equation 5] As long as the above conditions are met, even if the update processing means 810 is executed, it is possible to perform scheduling that guarantees real-time execution of the normal processing means 820. This condition can be used to determine whether or not the normal processing unit 820 can accept the execution request, and can also be used to determine whether or not the update processing unit 810 can accept the request and the scheduling content. Update processing means 8
If it is not necessary to guarantee the end time of the execution of 10, then T
_{Since N} → ∞, the last term of the second equation of (Equation 5) may be ignored. However, even in this case, the normal processing means 8
If the probability of blocking 20 is known, the update processing means 8
The end time of execution of 10 can be guaranteed stochastically.

Further, when the processing is blocked by the late monotonic scheduling, the activation cycle of the _mth normal processing means 820 to be blocked is T _m , and the mth normal processing means 8 is activated.
When the time when 20 is closed is B _m ,

[0117]

[Equation 6] It is known that scheduling is possible if If B _m is a critical part of the update processing means 810, the processor utilization efficiency can be improved more than in (Equation 5). However, it is necessary to apply a plurality of judgment conditions (that is, threshold values). Hereinafter, for simplicity, it is assumed that the determination is made by (Equation 5).

In step 214, the new normal processing means 82
Assuming that the processing amount of 0 is L _i and the activation period is T _i , (Equation 5)
Whether to accept or not is determined by. In this case, since U _T is fixed, the load for update processing is always taken into consideration regardless of whether or not the update processing unit 810 is executed. If possible, execute step 216, otherwise execute step 215.

At step 215, the new normal processing means 82.
The request source was informed that the execution request of 0 was rejected,
The acceptance processing of the normal processing means in step 112 is ended.

At step 216, the new normal processing means 82.
0 is registered in the guarantee schedule 802. The registration contents are
The activation period T _i , the processing amount L _i , the identifier given by the system side, and the like. As a result, the acceptance processing of the normal processing means 820 in step 112 ends.

FIG. 22 shows the contents of the acceptance processing of the update processing means 810 at step 114.

At step 311, execution of the update processing means 810 is requested by designating an update target class.

At step 312, the update processing means 810 is registered in the guarantee schedule 802. The registered contents are the activation cycle T _N , the processing amount L _N , and the identifier given by the system side. Here, at the time of request, since it is assumed that T _N and L _N are not given, appropriate values must be determined. Since L _N / T _N is given in determining the U _T, it is possible to determine the T _N be determined the L _N. For example, L _N may be obtained by multiplying the number of instances of the update target class by the processing amount of each instance and further adding the fixed processing amount. However, in the case of late monotonic scheduling, T _N is a normal process that requires real-time processing so that the execution of the update processing unit 810 has a lower priority than the execution of the normal processing unit 820 that requires real-time processing. Means 820
Is larger than the starting cycle T _{i of}

In the case of deadline scheduling,
When the execution of the update processing means 810 of a certain start cycle is completed, T _N is set so that the deadline becomes farther from the deadline of the normal processing means 820 that requires real-time processing. It is assumed here that the number of update processing means 810 activated at the same time is one, but it is easy to expand to a plurality of cases.

FIG. 23 shows the contents of the flag processing in step 115. The flag 862 is defined for each instance, for example, and gives the owner of the flag 862 access to the instance. reading/
Although it is possible to define each flag 862 for the write access right, it is assumed here that a single flag 862 is given. Further, this flag 862 temporarily lowers the priority of the owner to the priority of the requester (inheriting the priority) when the priority of the flag owner is lower than the priority of the flag requester. , A flag including a priority inheritance function that operates so as to end the processing of the owner early and prompt the opening of the flag 862. Assuming here, the normal processing means 820 has a higher priority than the update processing means 810. Therefore, when the update processing unit 810 owns the flag 862 for processing the critical portion, the processing amount of the normal processing unit 820 is at most L _c.
It increases, but this is the normal processing 820 based on (Equation 5).
Since it is considered in the reception processing of the normal processing means 820
The real-time property of is guaranteed.

At step 421, it is judged whether the request is a flag acquisition request by the normal processing means 820 or the update processing means 810. If so, branch to step 422, else branch to step 427.

At step 422, the requested flag 86
It is determined whether or not 2 is empty. If it is empty, execute step 423, otherwise step 4
Execute 24.

At step 423, the requester is flagged as 862.
Register as the owner of. Specifically, the identifier given to the normal processing means 820 or the update processing means 810 which is the requester is written in the owner field.

In step 424, it is determined whether the requester of the flag 862 has a higher priority than the owner.
For example, in the case of late monotonic scheduling, the shorter the activation period T _i is, the higher the priority is, and in the case of deadline scheduling, the end time required for the current process (corresponding to the next activation time if it is periodic) is closer. Has a higher priority. If so, step 425 is performed, otherwise step 427 is performed.

At step 425, the priority of the owner of the current flag 862 is temporarily set to the same priority as the requester until the flag is opened at step 427.

At step 426, the requester with flag 862 is added to the waiting list.

At step 427, it is judged whether or not the request is a flag release request by the normal processing means 820 or the update processing means 810. If so, the process branches to step 428, otherwise the flag processing of step 115 ends.

At step 428, the requester of the current flag 862 is deleted.

In step 429, it is determined whether or not the waiting is registered in the waiting person list of the flag 862. If it is registered, step 430 is executed, otherwise, the flag processing of step 115 is ended.

At step 430, the normal processing means 820 or the update processing means 810 registered in the waiting person list.
The one with the highest priority is selected from among the above as the owner, and the traveling waiting state is further set, and the flag processing of step 115 is ended.

FIG. 24 shows the flow of the update processing means 810. Before the processing of the update processing means 810 starts, acceptance of execution of the new update processing means is performed in step 114.

In step 501, preprocessing for update processing is performed. For example, a file storing the state definition and procedure definition of the new class corresponding to the old class to be updated is loaded, and the program of the procedure is placed in the executable memory.

At step 502, the current instance 860 of the target class is selected. The number of instances selected at the same time is such that the update processing amount does not exceed the processing amount L _c of the critical part of the update processing means 810 even if they are processed at the same time.

At step 503, the current instance 86
It is determined whether 0 is present. If not, it means that all the current instances 860 for the target class have been processed, so step 507 is executed. Otherwise, execute step 504.

At step 504, the current instance 86
The flag 862 of 0 is acquired. Therefore, the process of the schedule executing means 801 of FIG. 20 is entered, and specifically, the flag process of step 116 is executed. Here, it is assumed that the processing is stopped until the flag 862 is acquired. In an actual system, it may be possible to request a flag of another current instance 860 when the flag 862 cannot be acquired. Step 50 after acquiring the flag
Execute 5.

At step 505, the current instance 86
The main body of the update process of 0 is executed. Specifically, a storage area for recording the state of the new instance 870 is secured, and the state 861 of the current instance 860 is referred to,
The state 871 of the new instance 870 is set accordingly. Furthermore, the access by the normal processing means 820
Switch from the current instance 860 to the new instance 870.

At step 506, the flag 861 of the current instance 860 of the old class is released. Therefore, the process of the schedule executing means 801 of FIG. 20 is entered, and specifically, the flag process of step 116 is executed. If the normal processing means 820 requesting to acquire the flag to access the current instance 860 is registered in the waiting list of the flag 862, the normal processing means 820 is of a new class because the switching processing is completed. Access the new instance 870. After this, step 50
Return to 2.

At step 507, post-processing is performed. For example, if the procedure program of the old class of the target class is unnecessary, the area storing it is released. Further, it notifies that the execution of the update processing means 810 for the target class is completed. As a result, the update processing means 810 ends.

FIG. 25 shows only the part for accessing a certain instance from the processing flow of the normal processing means 820.

In step 601, the current instance 860 to access is selected.

At step 602, the current instance 8
Acquire 60 flag 862. Therefore, the process of FIG. 20 is entered, and specifically, the flag process of step 116 is executed. Here, it is assumed that the processing is stopped until the flag 862 is acquired. After obtaining the flag, step 603 is executed.

At step 603, various processes for executing the normal process are performed.

At step 604, the flag 862 is opened. For this purpose, the schedule executing means 801 shown in FIG.
The process of (1) is entered, and specifically, the flag process of step 116 is executed.

The first modification of the second embodiment described above will be described below.

FIG. 27 shows the flow of the schedule executing means 1301 shown in FIG.

At step 701, it is judged whether the request is an execution request for the new normal processing means 1320. In that case, step 702 is executed, otherwise step 703 is executed.

At step 702, the new normal processing means 1320 receives an execution request. See the description of FIG. 28 for details. After the end, step 713 is executed.

At step 703, it is judged whether or not the request is an execution request of the new update processing means 1310. In that case, step 704 is executed, otherwise step 705 is executed.

At step 704, the process of accepting the execution request of the new update processing means 1310 is performed. For details, see the description of FIG. After the end, step 711 is executed.

At step 705, it is judged whether or not the state change flag process is being executed. If it is the state change flag process, step 706 is executed, otherwise step 707 is executed.

At step 706, a state change flag process is performed. Normal processing means 1320 or update processing means 13
The process 10 acquires or releases the state change flag 1364. See FIG. 30 for details. After the end, step 713 is executed.

At step 707, it is determined whether or not the state reference flag process is being performed. If it is the state reference flag process, step 708 is executed, otherwise step 709 is executed.

At step 708, the state reference flag process is performed. State reference flag 1 by the normal processing means 1320
The 365 is acquired or released. See FIG. 31 for details. After the end, step 713 is executed.

At step 709, it is determined whether or not the state coincidence flag process is being performed. If it is the state match flag process, step 710 is executed, otherwise step 711 is executed.

At step 710, a state coincidence flag process is performed. Status match flag 1 by update processing means 1310
363 is acquired or released. See FIG. 32 for details. After the end, step 713 is executed.

At step 711, it is determined whether or not the timer processing is being performed. If it is a timer process, step 712 is executed, otherwise step 713 is executed.

At step 712, timer processing is performed.
This process is a process executed at the time of a timer interrupt of hardware, and by referring to the current time, the normal processing means 1320 or the update processing means 1310 to be activated at that time is selected from the registered contents of the guarantee schedule 1302. , Put the vehicle in a waiting state. After the end, step 713 is executed.

At step 713, the guarantee schedule 1
With reference to the registered contents of 302, the one having the highest priority is selected from all the normal processing means 1320 or the update processing means 1310 in the traveling state or the traveling waiting state as the target to be executed next.

At step 714, the normal processing means 1320 or the update processing means 13 selected at step 711.
10 is executed.

FIG. 28 shows the contents of the acceptance processing for executing the normal processing means in step 702. This processing content is almost the same as the processing content of FIG. 21 of the second embodiment.

In step 801, the normal processing means 1320
The start-up cycle T _i and the processing amount L _i per start-up are designated, and a new normal process request acceptance process is started. The processing amount is expressed, for example, by dividing the number of instructions of the processor executed by one activation of the normal processing means 1320 by the number of processing instructions per unit time of the processor.

At step 802, the execution load U _p of the existing normal processing means 1320 and update processing means 1310 currently being executed is acquired. This is obtained by summing the usage rates L _i / T _i of the processors declared for the existing processing means according to (Equation 1).

At step 803, the threshold value determining means 13
55 determines the threshold U _T. When late monotonic scheduling is performed, U _M of (Equation 3) is set to U _T (however, the new normal processing means 1320 is included in the number), and when deadline scheduling is performed, (Equation 4) of Let U _M be U _T. When scheduling other than the above is performed, the threshold value U _T is determined by a different method.

In step 804, the new normal processing means 13
With 20 as the processing amount of L _i and T _i as the activation period, the acceptance / rejection is determined by (Equation 5). If possible, execute step 806, else execute step 805.

In step 805, the new normal processing means 13
The fact that the execution request of 20 is rejected is notified to the request source, and the acceptance processing of the normal processing means 1320 of step 702 is ended.

In step 806, the new normal processing means 132
Register 0 in the guarantee schedule. The registered contents are the activation cycle T _i , the processing amount L ₀ , the identifier given by the system side, the object used, and the like. This results in step 70
The new normal processing means 1320 of No. 2 performs acceptance processing.

FIG. 29 shows the content of the acceptance processing for executing the update processing means in step 704. This processing is performed by the schedule determining means 1303.

At step 901, switching processing amount estimation means 1
From 331, the switching processing amount L _c per instance of the update processing target class is estimated.

At step 902, the state matching processing amount estimation means 1332 estimates the state matching processing amount L _n per instance of the class of the update processing target.

At step 903, the shortest starting cycle acquisition means 1340 acquires the shortest starting cycle T _c of the normal processing means 1320 that accesses the current instance 1360 of the update processing target class. The called shortest startup period acquisition unit 1340 may read a value determined in advance at the time of system design from a certain area and return it as the shortest startup period, or the normal processing unit 1302 recorded in the guarantee schedule 1302. The shortest one of the starting cycles may be returned as the shortest starting cycle.

In step 904, the processor load estimation unit 1350 acquires the processor load U _p due to the execution of the existing normal processing unit 1320 and update processing unit 1310 that are currently being executed. This is the processor usage rate L declared for the existing processing means by (Formula 1).
It is obtained by summing _i / T _i .

At step 905, the threshold value determining means 13
55 determines the threshold U _T. If late monotonic scheduling is performed, U _M in (Equation 3) is set to U _T (however, the number of new normal processing means is included), and if deadline scheduling is performed, U in (Equation 4) is performed. _{Let M} be U _T. An appropriate margin may be included for convenience of management. For example, if you want to finish the update process as soon as possible, decide the threshold value without including any margin, and if there is no particular urgency, include the margin and leave the processor power for other processing. Can be kept. In the case of execution of the normal processing means, the amount of processing to be performed within a fixed time is generally determined, so adjustment of the processing amount is meaningless, but in the case of execution of the update processing means, it is performed within a fixed time. The amount of work to be done can be adjusted.

In step 906, the number of instances to be processed within a fixed time is determined by the threshold value. Therefore, it is a system management that the apparent threshold value is changed by such a margin and the processing amount is adjusted accordingly. It is very effective. When scheduling other than the above is performed, the threshold value U _T is determined by a different method.

At step 906, the maximum number of instance sets N ₁ to be the target of the switching processing means 1331 is determined at the same time. The switching processing unit 1331 is the update processing unit 133.
This is a part that closes the execution of the normal processing means 1320 with the processing of 0 and executes the normal processing means 1320 from the current instance 1360 to the new instance 137.
Including the process of switching to 0. Therefore, the switching processing means 13
The execution of 31 causes normal processing means 132 during L _c × N _1.
The processing of 0 is continuously blocked. Therefore, for example, N ₁ is determined so as to satisfy the following (Formula 7).

[0180]

[Equation 7] In step 907, the activation cycle T _N of the update processing means 1310 including the state matching processing means 1312 and the switching processing means 1311 is determined. Here, the state matching processing means 131
2 is the processing of the update processing means 1310 and the normal processing means 1
320 is a part to be executed without being blocked, and refers to the state of the current instance 1360 to create a new instance 137.
It includes the process of setting the state of 0. Therefore, it is assumed that (L _c + L _n ) × N _{l is} required to execute the update processing means 1310, where N _l is the maximum number of groups to be processed simultaneously. Therefore, in the case of late monotonic scheduling, T _N is determined so as to satisfy the following (Equation 8).

[0181]

[Equation 8] In step 908, the content determined as above for the update processing means 1310 is registered in the guarantee schedule 1302. The registered contents are the (virtual) activation cycle T _{c of} the switching processing means 1311 that closes the normal processing means 1320, the processing amount L _c , and the activation cycle T of the entire update processing means 1310.
_N , processing amount L _c + L _n , an identifier given by the system, and the like. In this embodiment, since it is assumed that the switching processing means 1311 is executed by dynamically raising the priority of a part of the updating processing means 1310, the switching processing means 13
No independent identifier is given to 11.

FIG. 30 shows the contents of the state change flag process of step 706. The state change flag 1364 is provided for each instance, for example, and gives the owner of the state change flag an access right to change the state of the instance.

At step 1001, the status change flag 1
It is determined whether the request is an acquisition request of 364. If so, perform step 1002, else step 1
010 is executed.

At step 1002, the state reference flag 1
It is determined whether or not 363 is empty. If the state reference flag 1363 is not empty, the state change flag 1364 cannot be acquired because the normal processing means 1320 refers to the state of the instance having the state reference flag 1363. In this case, step 1007 is executed.
Otherwise, execute step 1003.

At step 1003, the state change flag 1
It is determined whether 364 is empty. If it is empty, step 100 is executed to acquire the status change flag 1364.
4 is executed, otherwise step 1007 is executed.

At step 1004, the requester making the request to acquire the status change flag is registered as the owner of the status change flag 1364. Specifically, the identifier given to the normal processing means 1320 which is the requester is written in the storage area of the owner.

At step 1005, the state match flag 1
It is checked whether 363 is empty. If it is not available, the state matching processing means 1312 causes the current instance 13
Reference is made to 60. At this time, step 1006
To execute. Otherwise, the status change flag processing is ended.

At step 1006, the state match processing means 1312 increments the value of the state change counter 1361 by 1 to indicate that the state 1366 of the current instance 1360 is changed while referring to the current instance 1360. The flag processing ends.

At step 1007, the state change flag 1
364 acquisition requester has a current status change flag 1364.
Alternatively, it is determined whether or not the priority is higher than the owner of the status reference flag 1365. If the priority is high, step 1008. Otherwise, execute step 1009.

At step 1008, regarding the owner of the current state change flag 1364 or the state reference flag 1365, the state reference flag 1365 is opened until the state change flag 1364 is opened at step 1011 or at step 1108 of FIG. Until that time, the priority of the owner is temporarily set equal to the priority of the requester.

At step 1009, the normal processing means which is the requester making the request to acquire the status change flag is registered in the status change flag waiter list. Specifically, the identifier given to the normal processing means 1320 which is the requester is written in the storage area of the waiting party list.

At step 1010, the state change flag 1
It is determined whether or not the request is a release request of 364. If so, step 1011 is executed, and if not, the state change flag process ends.

At step 1011, after confirming that the release requester is the owner of the status change flag 1364, if the priority of the owner is temporarily raised, it is returned to the original state and the owner is registered. To the periphery.

At step 1012, the state reference flag 1
It is checked whether or not the waiting person is registered in the waiting person list of 365 or the state change flag 1364. If so, step 1013 is executed, and if not, the state change flag process ends.

At step 1013, the state reference flag 1
From the waiters registered in the waiter list of 365 or the state change flag 1364, the one with the highest priority is selected according to the determination criterion determined by the scheduling method, and this is referred to as the requester for the state reference flag 13
65 or the state change flag 1364 is acquired, and a traveling waiting state is set. This completes the processing of the status change flag 1364.

The contents of the state reference flag process of step 708 are shown in FIG. The state reference flag 1365 is provided, for example, for each instance, and the state reference flag 1
It is like giving the owner of 365 access to see the state of the instance.

At step 1101, the status reference flag 1
It is determined whether the request is a 365 acquisition request. If so, perform step 1102, else step 1
Execute 107.

At step 1102, the state change flag 1
It is determined whether 364 is empty. If the status change flag 1364 is not empty, the status of the instance having the status change flag 1374 has been changed by the normal processing means 1320, and the status reference flag 1365 cannot be acquired. In this case, step 1104 is executed.
Otherwise, step 1103 is executed.

At step 1103, the requester making the status reference flag acquisition request is registered in the owner list of the status reference flag 1365. Specifically, the identifier given to the normal processing means 1320 which is the requester is written in the storage area of the owner list. Furthermore, the state reference counter 13
62 is incremented by 1, and the state reference flag process is ended.

At step 1104, the state reference flag 1
The 365 requester of the acquisition requests the current status change flag 1364.
It is determined whether it is a higher priority than the owner of the. If the priority is high, step 1105 is executed. Otherwise, step 1106 is executed.

In step 1105, regarding the owner of the current state change flag 1364, step 101 in FIG.
Until the state change flag 1364 is opened at 1, the priority of the owner is temporarily set equal to the priority of the requester.

At step 1106, the normal processing means 132 which is the requester making the status reference flag acquisition request.
0 is registered in the status reference flag waiter list. Specifically, the identifier given to the normal processing means 1320 which is the requester is written in the storage area of the waiting party list. Then, the state reference flag process ends.

At step 1107, the state reference flag 1
It is determined whether or not the request is a 365 release request. If so, step 1108 is executed, otherwise, the state reference flag process ends.

At step 1108, after confirming that the release requester is registered in the owner list of the status reference flag 1365, if the priority of the owner is temporarily raised, it is returned to the original state and the owner Register from the list of persons.

At step 1109, the state reference flag 1
It is checked whether or not the waiting person is registered in the waiting person list of 365 or the state change flag 1364. If so, step 1110 is executed, otherwise, the processing of the status reference flag 1365 is ended.

At step 1110, the state reference flag 1
From the waiters registered in the waiter list of 365 or the state change flag 1364, the one with the highest priority is selected according to the determination criterion determined by the scheduling method, and this is referred to as the requester for the state reference flag 13
65 or the state change flag 1364 is acquired, and a traveling waiting state is set. This completes the processing of the status reference flag 1365.

FIG. 32 shows the contents of the status match flag processing in step 710.

At step 1201, the status match flag 1
It is determined whether the acquisition request is for 363. If so, step 1202 is executed, otherwise the process ends. The state coincidence flag 1363 is defined for each instance, for example, and the state coincidence processing means 1312 refers to the state of the instance, or indicates that the state coincidence processing unit 1312 has finished referencing but has not yet finished the switching process to the new instance 1370. Show.

At step 1202, the state change flag 1
It is determined whether 364 is empty. If the status change flag 1364 is not empty, the status 1366 of the current instance 1360 having the status change flag 1364 may be changed by the normal processing means 1320, so the status match flag 1363 cannot be acquired. In this case, step 1205 is executed. Otherwise, execute step 1203.

At step 1203, the state coincidence flag 1
Check if 363 is free. If it is free, step 1204 is executed. If not free, step 1205 is executed.

At step 1204, the requester requesting acquisition of the status match flag is registered in the owner list of the status match flag 1363. Specifically, the identifier given to the update processing means 1310 which is the requester is written in the storage area of the owner list. This completes the state match flag processing.

At step 1205, the state coincidence flag 13
The update processing means 1310 making the acquisition request is notified that the acquisition of 63 has not been completed, and the status match flag 136 is sent.
The process of 3 is ended.

FIG. 33 shows the flow of the update processing means 1310. Before the processing of the update processing means 1310 starts,
In step 704, acceptance of execution of new update processing means is being performed. Further, the part other than step 1305 is executed with the priority of the state matching processing means 1312.

At step 1301, preprocessing for update processing is performed. For example, a file storing the state definition and procedure definition of the new class corresponding to the old class to be updated is loaded, and the program of the procedure is placed in the executable memory.

At step 1302, the number of current instances n _l to be processed by the state matching processing means 1312 is determined by repeating this time. As the state matching process target current instance, n _l, which is determined by a rule described later, is selected from the set of the current instances 1360 for which neither the state matching process nor the switching process has been performed. However, if the number of current instances in the set is less than n ₁ , all the current instances 1360 in the set are targeted. on the other hand,
The switching target current instance is composed of n _n current instances that have undergone the state matching process but have not been switched, and n _l current instances that are the current state matching process target. _n n it is determined at the time when the switching process of the previous iteration has been completed. However, n _n = 0 at the first iteration. Since the switching process processes n _l current instances at one time, when N _l ≧ n _n ,

[0216]

[Equation 9] The number n ₁ of current instances for state matching processing is determined by. However, if N ₁ <n _n , then n _l = 0.

In step 1303, if there is neither a state matching target current instance to be processed in this iteration, nor a switching process target current instance (n = 0 and n _n = 0), that is, all the target classes are processed. If the update process for the current instance 1360 of 1 is completed, the process branches to step 1308 to perform post-processing for ending the update process. Otherwise, step 13
The following processing is performed.

At Step 1304, Step 1302
Status matching processing target current instance 1360 determined in
Then, the state matching process is performed for the new instance 1370 that is paired with it. See the description of FIG. 34 for details.

At step 1306, the necessity of reschedule is judged, and if reschedule is necessary,
In order to execute the schedule determination means 1303, the process branches to step 1307. Else step 130
Return to 2. When the reschedule is necessary, for example, the processor load becomes high, and the new normal processing means 132
There is a high possibility that the acceptance of 0 will be rejected, a case where it is actually rejected, or a case where the processor load becomes low and more resources are allocated to the update process. For this determination, for example, a task that regularly monitors the processor load or the like refers to the memory area in which the monitoring result is written in this step, or the schedule execution means 1301 leaves the new normal processing means 1320. The history of acceptance / rejection may be referred to.

At step 1307, in order to change the schedule, the schedule deciding means 1303 is called and the processing of FIG. 29 is executed.

At step 1308, post-processing for ending the update processing is executed. Specifically, the old class procedure program is deleted from the executable area, the class state viewed from the object management system is changed from the transitional state of change to a stable state, and the update processing request source ends the update processing. Is notified, and the registration of the update processing execution means 1310 is deleted from the guarantee schedule. As a result, the update processing executing means is ended.

When step 1304 is activated, FIG.
The state matching process is performed.

In step 1401, one is selected from the current instances 1360 for which the state matching process has not been performed yet as the state matching process target. However, the current instance number selected by this repetition, so as not to exceed the n _l determined in step 1302.

In step 1402, a branch is made depending on whether or not the state matching process target current instance in step 1401 has been selected. If so, step 1403
Otherwise, that is, if the number of current instances in this iteration exceeds n _l , the state matching process in this iteration ends.

At step 1403, the procedure for requesting the acquisition of the status match flag 1363 of the selected current instance 1360 is called. Further, the schedule executing means 1301 is called in the status match flag acquisition request procedure,
Step 1710 is executed.

At step 1404, the status match flag 1
A branch is made depending on whether or not 363 can be acquired. If it can be acquired, the processing proceeds to step 1405, and if it cannot be acquired, the processing returns to step 1401 to select a new current instance 1360.

At step 1405, the state 1366, the state change flag 1364, and the state reference flag 1365 of the current instance 1360 are referred to, and the state 1376, the state change flag 1374, and the state of the new instance corresponding to the result are referenced. The reference flag 1375 is set. Basically the current instance state 1366,
State change flag 1364 and state reference flag 136
State 1 of new instance so that 5 becomes the same state as
376, the state change flag 1374, and the state reference flag 1375 are set.

When step 1305 is activated, FIG.
The switching processing means 1311 is started.

In step 1501, the normal processing means 132
Block execution of 0. For example, when the switching processing unit 1311 and the state matching processing unit 1312 are realized as one task, the task is executed with a lower priority than the other normal processing unit 1320 when performing the state matching process. However, at this stage, the priority of the task executing the update processing means 1310 is made higher than the priority of the task executing the other normal processing means 1320.

[0230] In step 1502, one of the current instances 1360 for which the state matching process has been completed but the switching process has not been performed is selected as a switching process target. However, the value of the counter that holds the current number of instances selected by this repetition is N determined in step 906.
Do not exceed _l

In step 1504, the value of the state change counter 1361 of the instance is referred to. If it is 1 or more, the process branches to step 1505. Otherwise, step 1506 is executed. The state change counter 136
If 1 is 1 or more, it means that the state of the current instance 1360 is changed by the normal processing means 1320 during the state matching process. Therefore, it is not guaranteed that the states match between the current instance 1360 and the new instance 1370 paired with it.

At step 1505, the state change counter 1361 is reset to 0 and the current instance 136 is reset.
0 is returned to the state where the state matching process is not performed, and the process returns to step 1502.

At step 1506, the value of the state reference counter 1362 of the current instance 1360 is referred to, and 1
When it is above, it branches to step 1507, otherwise, step 1508 is executed. Note that the state reference counter 1362 being 1 or more means that there is a normal processing unit that refers to the state 1366 of the current instance 1360 at that time. Therefore,
It is not possible to immediately switch from the current instance 1360 to the new paired instance 1370.

At step 1507, the current instance 1360 is brought into a state in which the state matching process is completed but the switching process is not completed, and the process returns to step 1502.

At step 1508, the new instance 1370 paired with the current instance 1360 is processed as a new current instance. That is, when the normal processing means 1320 tries to operate the instance, the operation target is switched from the current instance 1360 to the new instance 1370, and the normal processing means 13 is further changed.
The procedure program itself, which is called by 20 to operate the instance, is also switched from the procedure program of the old class to the procedure program of the new class. However, this switching is actually effective only after the blockade of the normal processing means 1320 is released in step 1509.

In step 1509, the normal processing means 132
The block of execution of 0 is released, and the switching process by this repetition is ended. Specifically, when the switching processing unit 1311 and the state matching processing unit 1312 are realized as one task, the priority of the task that executes the updating processing unit 1310 can be returned to the priority of the state matching processing unit 1312. Good.

In order to realize the above processing, the normal processing means 1320 performs the processing shown in FIG. 36 or FIG. 37, respectively, when referring to or changing the state 1366 of the current instance 1360.

Normal processing means 13 referring to the state 1366
The process flow of 20 is shown in FIG. Step 1601
, Select the current instance 1360 to which to reference state 1366.

At step 1602, a procedure for requesting acquisition of the status reference flag 1365 of the current instance 1360 is called. The schedule executing means 1301 is further called in the status reference flag requesting procedure, and step 7
08 is executed.

At step 1603, normal processing for referring to the state 1366 of the current instance 1360 is performed.

At step 1604, the status reference flag 1
Call the procedure to release 365. The schedule executing means 1301 is further called in the state reference flag release request procedure, and step 708 is executed.

Normal processing means 13 for changing the state 1366
The processing flow of 20 is shown in FIG.

At step 1701, the current instance 1360 whose state 1366 is to be changed is selected.

At step 1702, a procedure for requesting acquisition of the status change flag 1364 of the current instance 1360 is called. The schedule executing means 1301 is further called in the state change flag acquisition request procedure, and step 706 is executed.

At step 1703, normal processing for changing the state 1366 of the current instance 1360 is performed.

At step 1704, the status change flag 1
Call the procedure to release 364. The schedule executing means 1301 is further called in the state change flag release request procedure, and step 706 is executed.

A second modification of the second embodiment will be described below.

In the second embodiment, the flag 862 having the priority inheritance function is used, and the update processing means 810 having a low priority is used.
While holding the flag 862, the normal processing means 82
When a processing request of 0 is generated, the priority of the update processing unit 810 is temporarily raised to accelerate the opening of the flag 862. The continuous processing time upper limit of the update processing means 810 is L
The real-time property of the normal processing means 820 was guaranteed because the normal processing means 820 was given by the above and is considered at the time of reception of the normal processing. However, since the flag processing is complicated, the overhead is large. This modification is intended to simplify the process corresponding to the flag process.

In this modification, the normal processing means 1 of FIG. 38 is used.
Counter 14 representing the number of instances used by 420
62 and an update flag 1463 indicating that the normal processing means 1420 is updating. These do not involve complicated processing such as priority inheritance. Instead, the update processing means 1
The priority of the critical portion having the processing time upper limit of 410 is explicitly raised to the extent that the normal processing means 1420 is not allowed to interrupt, and the counter 1462 and the update flag 146 are set.
Current instance 14 determined to be updatable from the value of 3
60 is selected to perform the updating process. If it was not updatable, try again later. Hereinafter, differences from the second embodiment will be described.

The flow chart of the schedule executing means of FIG. 39 is the same as the processing of FIG. 20 except that the flag processing is not provided.

In the flow chart of the update processing means of FIG. 40, steps 1901-1903 are steps 501-50 of FIG.
Equivalent to 3.

At step 1904, the update processing means 141
The normal processing means 1420 is closed by a method such as setting the priority of 0 higher than that of any normal processing means 1420 using the current instance 1460 to be updated.

At step 1905, among the plurality of current instances 1460 selected at step 1902,
Counter = 0, that is, the current instance 1460 not used by the normal processing means 1420 is set to step 190
6 is selected as the update target, and the others are again step 1902
Return to the selection target candidate of.

The updating process of step 1906 is the same as step 505 of FIG.

In step 1907, the update processing means 141
The priority of 0 is returned to a value lower than that of the normal processing means 1420.

FIG. 41 shows only the reference processing part for a certain instance from the processing flow of the normal processing means 1420.

At step 2001, the current instance 1460 to be referred to is selected.

At step 2002, the value of the counter 1462 of the current instance 1460 is incremented by 1.

At step 2003, processing based on the reference of the state of the current instance is performed.

At step 2004, the counter is decremented by 1.

FIG. 42 shows only the update processing part for a certain instance from the processing flow of the normal processing means 1420.

In step 2101, the current instance 1460 to be referred to is selected.

At step 2102, the value of the counter 1462 of the current instance 1460 is incremented by 1.

In step 2103, it is confirmed that the update flag 1463 of the current instance 1460 is empty and the value of the counter 1462 is 1, and the update flag 1463 is acquired. The processing when it cannot be acquired is omitted.

At step 2104, processing based on the update of the state of the current instance is performed.

In step 2105, the update flag 1463
Open up.

At step 2106, the counter is decremented by 1.

The basic method of the second modification described above should be applied when the update processing is divided into non-critical state matching processing and critical switching processing as in the first modification already described. You can also

A network management system according to the third embodiment of the present invention will be described in detail below with reference to the drawings.

FIG. 43 is a block diagram showing the detailed arrangement of the primary management station in the network management system of the third embodiment. In FIG. 43, reference numeral 1511 denotes a network management execution means having a means for executing this embodiment, 1512 a management information storage means for storing management information at the time of performing network management, and 1513 when a failure occurs in the management area. An automatic terminal recognizing unit having a unit for automatically finding and recognizing a terminal being accessed by an administrator, 1514 is an access terminal setting unit having a unit for setting a terminal on which an administrator executes a management operation, and 1515 is an embodiment. 1516 is a protocol processing means for communicating with a secondary management station or the like through a communication network, and 1527 is a communication network.

FIG. 44 is a block diagram showing the detailed arrangement of the secondary management station in the network management system of this embodiment. Here, the secondary management station is a terminal within the management area of the administrator (within the management areas of all network management stations that the administrator is in charge of). In FIG. 44, reference numeral 1521 denotes a network management execution unit having a unit for executing this embodiment, and 1522.
Is a management information storage unit that stores management information when performing network management, 1523 is an agent function unit that directly manages its own terminal or a communication device, and 1524 is for communicating with a primary management station through a communication network. 1525 is a communication network.

FIG. 45 is a flow chart when the primary management station performs the processing of this embodiment, and FIG. 46 is a flow chart when the secondary management station performs the processing of this embodiment.

The processing of this embodiment will be described below with reference to FIGS. 45 and 46.

FIG. 45 is a flowchart of the processing performed by the primary network management station of FIG. In this embodiment, steps A1 to A11 are shown.

Step A1 The primary management station normally determines the configuration and failure of terminals and communication devices within the management area.
Performance, confidentiality, etc. are managed, and the state is step A1 in FIG. The network management is executed by the network management executing means 1511 shown in FIG. 43, and the protocol processing at the time of network management is executed by the protocol processing means 1516 shown in FIG.
Management information (configuration information, failure information, etc.) executed by the management information storage means 151 of FIG.
Accumulated in 2.

Step A2: When a failure occurs in the terminal or the communication device, the agent of the managed device issues a failure notification to the primary management station, or the network management execution means 1511 of FIG. 43 in the primary management station is executed. It can be used and recognized by regular management operations.

Step A3 After the network management execution means 1511 of FIG. 43 in the primary management station receives and recognizes the failure notification through the protocol processing means 1516 of FIG. 43, the failure information is stored in the management information storage means 1512 of FIG. accumulate.

Step A4 In order for the administrator to promptly deal with the failure, when the failure information of the terminal or the communication device is received, the terminal automatic recognition means 151 of FIG.
3 is a terminal that the administrator is accessing or logging in (including network management stations outside the management area)
Recognize.

Step A5 The access terminal setting means 1513 of FIG. 43 sets one of the terminals accessed or logged in by the administrator as the secondary management station.

Step A6 The agent function execution means 1515 of FIG. 43 sends a failure notification through the protocol processing means 1516 of FIG. 43 to notify the terminal set as the secondary management station of the failure of the managed device.

Step A7 The agent function executing means 1515 of FIG. 43 receives the management operation executed by the network management executing means 1521 of FIG. 44 and transmitted through the protocol processing means 1524 of FIG.

Step A8 The agent function execution means 1515 in FIG. 43 interprets the management operation received in step A7, and the network management execution means 151 in FIG.
1 executes management operation. At this time, when referring to past failure information, configuration information, or the like, the information is retrieved from the management information storage means 1512 in FIG.

Step A9 The agent function executing means 1515 in FIG. 43 sends the result of the management operation executed in step A8 to the secondary management station through the protocol processing means 1516 in FIG.

Step A10: It is determined whether the secondary management station has completed the management operation. If the agent function execution means 1515 of FIG. 43 receives the management operation end report through the protocol processing means 1516 of FIG. 43, the process proceeds to the next step A11, and if the management operation end report is not received, the agent function execution means 151 of FIG.
Step 5 waits for the reception of the management operation from the secondary management station, and the process returns to step A7.

Step A11 The access terminal setting means 1514 of FIG. 43 cancels the terminal setting of the secondary management station when the management operation of the secondary management station is completed in step A10.

FIG. 46 is a flow chart when a terminal (including a network management station outside the management area) accessed by the administrator by the network management station is executed as a virtual secondary management station. In this embodiment, steps B1 to B7 are shown.

Step B1 The agent function means 1523 shown in FIG. 44 manages the managed device in response to the management operation from the primary management station and normally fails the managed device in response to the management operation from the primary management station. Executes the function that issues a fault when occurs.

Step B2 The network management executing means 1521 of FIG. 44 is the protocol processing means 15 of FIG.
Through 24, it is recognized whether the terminal setting information is received from the primary management station. When you receive the device setting information,
The primary management station proceeds to the next step B3 in order to set the terminal as the secondary management station, and when the terminal setting information is not received, executes step B1.

Step B3 The network management executing means 1521 of FIG. 44 is the protocol processing means 15 of FIG.
Fault notification is received through 24.

Step B4 The administrator judges the failure and inputs the management operation from the console of the terminal which is the secondary management station, and the network management executing means 15 of FIG.
Reference numeral 21 denotes a management operation of the protocol processing means 152 of FIG.
4 to the primary management station.

Step B5 The network management execution means 1521 of FIG. 44 receives the result of the management operation transmitted by the primary network management station through the protocol processing means 1524 of FIG.

Step B6: As a result of the management operation, it is judged whether the administrator does not perform any further management operation. When the management operation is completed, the management operation completion report is transmitted to the primary network management station through the protocol processing means 1524 of FIG.

Step B7 When the setting of the terminal to the management station is canceled by the access terminal setting means 1514 of FIG. 43 by step A11, FIG.
Execution of the network management execution means 1521 of
The agent function unit 1523 in FIG. 44 continues to manage the terminal itself and the communication device.

The first modification of the third embodiment described above will be described below.

In the third embodiment, the second station is a terminal within the management area of the administrator (within the management areas of all network management stations that the administrator is in charge of), but in the present embodiment, the second station A case where the management station is a network management station (excluding the primary management station) within the management area of the administrator will be described.

Since the secondary management station in this embodiment is the primary management station in the management area, it includes both the functions of the primary management station and the functions of the secondary management station. Therefore, the block configuration diagram showing the detailed configurations of the primary management station and the secondary management station of this embodiment is the same as the block configuration diagram (FIG. 43) showing the detailed configuration of the primary management station of the third embodiment. Will be the same. Further, the flowchart of the processing of the primary management station and the secondary management station in this embodiment is different from that of the secondary management station managing the managed devices in the management area in step B1 of FIG. 45 and 46 are almost the same. As a result, it becomes possible to manage not only the managed devices within the management area but also the managed devices outside the management area while performing communication between the managers.

A second modification of the third embodiment will be described below.

Step A4 of the flowchart of the process in the third embodiment A first method for automatically recognizing the access terminal of the administrator will be described with reference to the flowchart of FIG.

One of the methods for the primary management station to automatically recognize the administrator login terminal is in the management area of the administrator (in the management areas of all network management stations that the administrator is in charge of). It is realized by polling the terminal. When the administrator logs in to the terminal, information about the login is issued, and the first
The next management station is recognized by collecting that information. As the polling method, as shown in the flowchart of FIG. 47, the login information is searched in the order of the network management station and the terminal in the management area of the administrator. When the login information is collected by this search, the primary management station recognizes those network management stations / terminals, and if it is not collected, it recognizes that the administrator is not logged in to the network management station / terminal. . With that, the polling ends.

In order to execute the method for automatically recognizing the access terminal of the administrator according to this modification, all the configuration information in the administrator's management area is set in FIG. 43 in each network management station in the administrator's management area. It is preferable that the management information is stored in the management information storage means 1512. This can be realized by providing the network management execution means 1521 of FIG. 44 with means for transferring the configuration information to another network management station when the network configuration in the management area of the network management station is changed.

The third modification of the third embodiment will be described below.

Step A4 of the flowchart of the process in the third embodiment A second method for automatically recognizing the access terminal of the administrator will be described with reference to FIG.

One of the methods for the primary management station to automatically recognize the login terminal of the administrator is to put it in the management area of the administrator (in the management area of all network management stations in charge of the administrator). When an administrator logs in to a certain network management station / terminal, the login information is transferred to each network management station within the management area of the administrator at appropriate intervals. When the administrator finishes logging in, the transfer of login information also ends. The primary management station recognizes the network management station / terminal to which the administrator is logged in based on the information. As a method of transferring the login information, a method of broadcasting to the network management station / terminal in the management area of the administrator or a method of multicasting only to each network management station in the management area of the administrator can be considered. The former method has a higher load in the network than the latter method, but is realized by a simple function, and the latter method is realized by registering the location of the network management station in the management area of the manager.

[0304]

As described in detail above, according to the present invention,
First, when updating an object-oriented program in object units, a reference source (normal processing process, instance data) and a reference destination (normal processing process are instance data or generated method code,
For instance data, a relay pointer, which is a selection means, is provided between the class method table) so that the reference source indirectly refers to the reference destination. Therefore, (a) the target portion of the switching processing accompanying the update processing is The amount of processing can be reduced. This can reduce the influence on normal processing. (B) Since the switching process associated with the update process does not need to know the reference source (especially the normal process) to be updated,
It becomes easy to create an update processing program that does not depend on the reference source program structure.

Secondly, according to the present invention, secondly, when the object management system is started by designating the update processing target class, the end time of the normal processing means of which the end time is set can be guaranteed in advance. Since at least one of the normal processing means and the update processing means is executed in accordance with the guaranteed schedule that is dynamically or dynamically determined, the software update configured by the object can be performed during the operation of the system while guaranteeing the real-time property. . That is, software update during operation can be performed without degrading the service quality of the system or making the service provision itself difficult.

Further, when the object management system is started by designating the update processing target class, the schedule determining means estimates the switching processing amount and the state matching processing amount, which are the update processing amounts, by the update processing amount estimating means, and starts the shortest time. The period acquisition means acquires the shortest startup cycle, the processor load estimation means estimates the processor load, and the update processing amount, the shortest startup cycle, and the processor load are used to determine the schedule of the update processing means including the state matching processing means and the switching processing means. A schedule in which the end time defined for the execution of at least a part of the normal processing means is dynamically determined as a guarantee schedule of the update processing means that is guaranteed with at least a certain probability, and the schedule execution means is determined. Perform the update processing by executing the update processing means according to The software update that is configured by an object, while guaranteeing real time property, and in a manner corresponding to the system load at that time, can be performed during operation of the system.

Further, the update processing means is separated into the state matching processing means and the switching processing means, the interruption of the operation of the normal processing means for the current instance is allowed during the execution of the state matching processing means, and the normal processing is performed only during the execution of the switching processing means. Since the interruption of the operation on the current instance by the means is prohibited, the time during which the execution of the update processing means prevents the execution of the normal processing means can be minimized, and the execution of the normal processing means is prohibited by the above separation. Since the number of current instances that can be continuously processed as it is can be increased, it is possible to suppress the number of times of switching between execution of update processing means and execution of normal processing means by scheduling.

Furthermore, since the schedule deciding means decides the guarantee schedule in accordance with a certain management policy, it becomes possible to allocate an appropriate processor capacity to the software update process within the range of guaranteeing the real-time property. If you want to complete the update process promptly on demand, allocate a large amount of processor power, and if you are not in a hurry, allocate a small amount of processor power to leave room for other processes. It enables flexible operations such as.

Further, according to the present invention, thirdly, the administrator does not always reside in one network management station, but when a failure occurs in a managed device (terminal and communication device), the administrator himself / herself. The terminal that is being accessed by a virtual secondary management station can receive fault information and perform management operations for the fault,
Allows administrators to quickly address failures without having to go to a network management station. Further, the administrator has an advantage that it can efficiently manage a plurality of network management stations, that is, a plurality of networks, and can perform flexible network management.

[Brief description of drawings]

FIG. 1 is a first principle diagram of an object management system according to a first embodiment of the present invention.

FIG. 2 is a second principle diagram of the object management system according to the first embodiment of the present invention.

FIG. 3 is a first conventional example.

FIG. 4 is a second conventional example.

5 is a diagram showing a detailed structure of FIG. 2. FIG.

FIG. 6 is a diagram showing a first data structure of instances and classes.

FIG. 7 is a diagram showing a second data structure of an instance and a class.

FIG. 8 is a diagram showing a second data structure of an instance and a class being updated.

FIG. 9 is a diagram showing a third data structure of instances and classes.

FIG. 10 is a diagram showing a third data structure of an instance and a class being updated.

FIG. 11 is a classification table showing an update procedure.

FIG. 12 is a flowchart of the entire update process.

FIG. 13 is a flowchart of the update process body of the first procedure.

FIG. 14 is a flowchart of the update processing body of the second procedure.

FIG. 15 is a flowchart of the update process body of the third procedure.

FIG. 16 is a procedure selection table of the update processing body of the first procedure.

FIG. 17 is a procedure selection table of the update processing body of the third procedure.

FIG. 18 is a block diagram of an object management system according to a second embodiment of the present invention.

FIG. 19 is a block diagram of a conventional example of an object management system.

FIG. 20 is a flowchart of the schedule executing means of the second embodiment.

FIG. 21 is a flowchart of acceptance processing for execution of normal processing means according to the second embodiment.

FIG. 22 is a flowchart of acceptance processing for execution of update processing means according to the second embodiment.

FIG. 23 is a flowchart of flag processing of the second embodiment.

FIG. 24 is a flowchart of update processing means of the second embodiment.

FIG. 25 is a flowchart of the normal processing means of the second embodiment.

FIG. 26 is a block diagram of a first modification of the second embodiment.

FIG. 27 is a flowchart of the schedule executing means of the first modified example.

FIG. 28 is a flowchart of acceptance processing for execution of normal processing means of the first modified example.

FIG. 29 is a flowchart of a scheduling determination process for executing the update processing means of the first modified example.

FIG. 30 is a flowchart of a state change flag process of the first modified example.

FIG. 31 is a flowchart of a state reference flag process of the first modified example.

FIG. 32 is a flowchart of a state matching flag process of the first modified example.

FIG. 33 is a flowchart of update processing means of the first modified example.

FIG. 34 is a flowchart of the state matching processing means of the first modified example.

FIG. 35 is a flowchart of the switching processing means of the first modified example.

FIG. 36 is a flowchart of reference processing by the normal processing means of the first modified example.

FIG. 37 is a flowchart of update processing by the normal processing means of the first modification.

FIG. 38 is a block diagram of a second modification of the second embodiment.

FIG. 39 is a flowchart of the schedule executing means of the second modified example.

FIG. 40 is a flowchart of update processing means of the second modified example.

FIG. 41 is a flowchart of reference processing by the normal processing means of the second modified example.

FIG. 42 is a flowchart of update processing by the normal processing means of the second modification.

FIG. 43 is a block configuration diagram showing a detailed configuration of a primary network management system according to a third exemplary embodiment of the present invention.

FIG. 44 is a block configuration diagram showing a detailed configuration of a secondary network management system.

FIG. 45 is a flowchart when the primary management station performs the process of the present embodiment.

FIG. 46 is a flowchart when the secondary network management station performs the processing of this embodiment.

FIG. 47 is a flowchart of a method for automatically recognizing an access terminal of an administrator.

FIG. 48 is a diagram showing another method of automatically recognizing the access terminal of the administrator.

FIG. 49 is a configuration diagram of a conventional network management system.

FIG. 50 is a block configuration diagram showing a detailed configuration of a conventional network management station.

FIG. 51 is a block diagram showing a detailed configuration of an agent in a conventional terminal or communication device.

[Explanation of symbols]

101 ... Update processing process, 111 ... Selection setting means, 1
12 ... Area acquisition / release means, 113 ... State inheritance means, 1
14 ... New class loading means, 115 ... Class data acquisition means, 116 ... Update procedure determining means, 117 ... Schedule determining means, 201 ... Normal processing process, 301 ... Instance relay pointer, 302 ... Reference counter, 303
... update counter, 321 ... (old) method relay pointer, 322 ... reference counter, 323 ... update counter, 3
31 ... New method relay pointer, 332 ... Reference counter, 333 ... Update counter, 341 ... Generation method relay pointer, 342 ... Reference counter, 343 ... Update counter, 401 ... (Old) instance data, 411 ... New instance data, 501 ... (Old) class method table,
511 ... New class method table, 601 ... (old) method code, 602 ... (old) generated method code, 611 ...
New method code, 612 ... Newly generated method code, 7
01 ... Operating system, 702 ... Guaranteed schedule, 801 ... Schedule execution means, 802 ... Guaranteed schedule, 810 ... Update processing means, 820 ... Normal processing means, 860 ... Current instance, 861 ... Status, 8
62 ... Flag, 870 ... New instance, 871 ... Status, 872 ... Flag, 901 ... Process execution means, 910 ...
Update processing means, 920 ... Normal processing means, 960 ... Current instance, 961 ... Status, 962 ... Flag, 970 ...
New instance, 971 ... Status, 972 ... Flag, 1
301 ... Schedule executing means, 1302 ... Warranty schedule, 1303 ... Schedule determining means, 1310 ...
Update processing means, 1311 ... Switching processing means, 1312 ... State matching processing means, 1320 ... Normal processing means, 1330 ...
Update processing amount estimating means, 1331 ... switching processing amount estimating means,
1332 ... State coincidence amount estimation processing means, 1320 ... Normal processing means, 1340 ... Shortest start cycle acquisition means, 1350 ...
Processor load estimation means, 1355 ... Threshold value determination means, 1360 ... Current instance, 1361 ... Status change counter, 1362 ... Status reference counter, 1363 ... Status match counter, 1364 ... Status change flag, 1365
... Status reference flag, 1366 ... Status, 1370 ... New instance, 1374 ... Status change flag, 1375 ... Status reference flag, 1376 ... Status, 1401 ... Schedule execution means, 1402 ... Assurance schedule, 1410 ...
Update processing means, 1420 ... Normal processing means, 1460 ... Current instance, 1461 ... Status, 1462 ... Counter, 1463 ... Update flag, 1470 ... New instance, 1471 ... Status, 1472 ... Counter, 1473 ...
Update flag, 1511 ... Network management execution means, 1
Reference numeral 512 ... Management information storage means, 1513 ... Automatic terminal recognition means, 1514 ... Access terminal setting means, 1515 ... Agent function execution means, 1516 ... Protocol processing means, 1517 ... Communication network, 1521 ... Network management execution means, 1522 ... Management information Storage means, 15
23 ... Agent function means, 1524 ... Protocol processing means, 1525 ... Communication network, 1561 ... Communication network, 1562 ... Network management station, 1563 ... Terminal, 1571 ... Communication network, 1
572 ... Network management station, 1573 ... Terminal, 1574 ... Communication equipment, 1581 ... Manager function means, 1582 ... Management information storage means, 1583 ... Protocol processing means, 1584 ... Communication network, 1591 ...
Agent function means, 1592 ... Management information storage means,
1583 ... Protocol processing means, 1594 ... Communication network.

Claims (5)

[Claims] 1. A plurality of data holding an internal state of an object, a plurality of procedure codes for realizing processing based on these data, and a plurality of procedure tables showing contents of storage areas of these procedure codes. In an object management system for managing the configured object, a first selection means for selecting one of the plurality of data identified by a first single identifier, and a second single identifier. Second selecting means for selecting one of the plurality of procedure tables identified by the third and third selecting for selecting one of the plurality of procedure codes selected by the third single identifier. At least one of the plurality of data to be selected by the first selection means and the plurality of data to be selected by the second selection means. Procedure table, said third object management system characterized by comprising setting means for setting at least one of selection means of the plurality of procedural code to be selected. 2. An object management system for managing an object, which is a collection of statuses and procedure contents constituting software existing in a computer system composed of one or more processors, provides a predetermined service of the computer system. One or more normal processing means for doing so, and a series of processing for detecting a match of the state between the current instance of the class of the object and a new instance of the class or another class, and switching between the instances, Update processing means for processing at least a part, and execution or update of the normal processing means such that the end time set for execution of at least a part of the normal processing means is guaranteed with at least an arbitrary probability. Adjust the execution process for at least one of the executions of the processing means. An object management system comprising: a guaranteeing means for executing the adjusted processing steps; and an executing means for executing the adjusted processing steps. 3. A network management system comprising a communication network composed of a plurality of communication devices and terminals, and a network management station managing the communication network, wherein a first network management station managing a certain communication network is provided. At a specific point in time, when a terminal managed by another network management station that has not been accessed fails, a recognition unit that recognizes the terminal that is accessing at that time, and the recognized specific terminal are A network management system comprising: setting means for setting this specific terminal as a second network management station so as to be able to cope with a failure. 4. The second network management station has network management execution means for receiving a setting request from the first network management station and managing the terminal in which the failure has occurred. The network management system according to claim 3, wherein 5. The network management system according to claim 4, wherein the first network management station has a receiving unit that receives a result of a management operation by the second network management station.