KR100294876B1 - Operation system capable of dynamic reconfiguration and method for the same - Google Patents

Abstract

PURPOSE: An operation system capable of dynamic reconfiguration is provided to selectively execute a necessary algorithm without compiling, system shutdown and booting according to an application program by performing the algorithm desired for each kernel component in a kernel. CONSTITUTION: The method comprises a module configuration part(200) statically configuring the configurable modules, a module discriminator discriminating the module to be reconfigured and a module executor executing the reconfiguration by branching to the module to be reconfigured. The module configuration part comprises a compiler/loader/linker(202) making the operation system to an executable format by compiling/loading/linking the source code and a mapping table manager(204) making a mapping table by extracting the information for the various algorithm included in one data structure of source code and the algorithm module being executed. The module discriminator comprises a reconfiguration deciding part and a combination deciding part. The algorithm selected by the module configuration part and module discriminator is called and executed by the module executor.

Description

Dynamic Reconfiguration and Operating System Dynamic Reconfiguration Method

1 illustrates the environment of an operating system on which the present invention is performed.

2 shows the functional configuration of the present invention.

3 shows a detailed view of the module configuration of FIG.

4 is a detailed view of the module identification unit of FIG.

5 is a flowchart showing the operation of the module unit over time.

6 is a flowchart illustrating the operation of the module identification unit over time.

7 shows a data structure in source code for explaining an embodiment of the present invention.

Figure 8 shows an example of a combination table for explaining an embodiment of the present invention.

9 shows an example of a scheduling algorithm for explaining an embodiment of the present invention.

The present invention relates to an operating system of a single processor system, and more particularly, to an operating system capable of dynamic reconfiguration of a single processor system and a method of dynamically reconfiguring the operating system.

It is composed of a memory device that can store various data, I / 0 devices that can process external signals, and a central processing unit that can process and manage these signals and data in a proper form. In a single processor system, an operating system implemented in the single processor system has been recently developed in a different concept from a conventional design concept from various angles.

Up to now, since the operating system maintains a fixed structure, it has been difficult to flexibly deal with various applications in which the amount of information to be processed or the relationship between the information is complicated. This meant redesigning a separate operating system for each of the various applications. Under these circumstances, a new concept of operating system design has emerged that provides an operating system by designing an operating system suitable for various application ranges and easily reconfiguring the operating system for each situation in consideration of various applications that are increasing day by day. The category of the model applying the operating system may be a supercomputer, a mainframe, a minicomputer, a workstation, a personal computer, and the like.

Therefore, in order to design an operating system suitable for various types of applications as described above, it is necessary to introduce a common concept, which is a microkernel. As the term suggests, the microkernel means that there are fewer quantitative aspects, that is, the functionality provided, than the existing operating system.

In other words, when the operating system 100 is divided into a kernel unit 120 and a server unit 110 as illustrated in FIG. 1, the kernel unit 120 processes multiple processes managed by the operating system. Process management unit 122 that is responsible for creation and removal of interprocess communication, interprocess communication management unit managing data structures and algorithms for interprocess communication techniques (Interprocess Communication Management, 126), and because the process is performed in memory There is a main function such as the memory management unit 124 for management, and there may be another portion in charge of interrupt processing, and a device management portion in charge of work related to the device.

Meanwhile, the server unit 110 supports various services not provided by the kernel unit, for example, a file management unit 112 used to support a file system, a network management unit for managing various network services. (114), the virtual memory management unit 116 to further extend the memory management unit supported by the kernel to support the virtual memory.

Such functionally necessary parts are placed in the kernel part, and other parts with strong aspects of extending or strengthening the functions of the operating system are separated in the server part, thereby having many advantages in the design of the operating system. If only the interface to the server part of the kernel part is well defined, the operating system can be easily scaled up and down (Scalability), and since the kernel part mainly contains hardware dependencies, the developed operating system is ported to other hardware systems. In this case, only localized parts can be modified to improve maintenance and maintainability, such as reducing development time.

With this in mind, as described above, an operating system having a structure suitable for various application systems does not require redesigning efforts, and the redesign problem is solved only by changing the configuration of the operating system. This is the micronization of the kernel, which maximizes the benefits of separating the kernel and server.

The microkernel was created by simplifying the role of the kernel, making services that can be provided with low-level system calls, and moving the parts to the server. Thus, from the operating system's point of view, the strategy and processing parts are separated.

As a result, various system calls in the microkernel lead to an abstraction of the target hardware, so that hardware independent parts and hardware dependent parts are thoroughly divided, and functions are divided into small units in addition to the advantages described above, making it easy to implement modularity. This increases the productivity of the software by increasing modularity. In this example, since the scheduling algorithm of processes in charge of the process management unit corresponds to the strategy unit, there may be various strategies. Therefore, the scheduling algorithm varies depending on the application system, so it is to be implemented in the server part.

The implementation of the operating system divided into two main parts can be easily applied to various systems by relieving the heavy burden of redesign by overcoming the inability to adapt to various application systems of the existing operating system. It is difficult to dynamically reconfigure various services with different structures. In other words, it is easier to apply a new algorithm to the kernel in recent years than the existing operating systems, but it is basically done by separate compilation, and after executing the new kernel, the entire system is shut down and initialized. From now on, the newly applied algorithm can actually be executed in the kernel only after a series of processes are executed. This process is time-consuming and complicated as described above, which causes unnecessary overhead in operating the system. Therefore, in designing an operating system that can be dynamically reconfigured, it is necessary to study what kind of architecture is suitable and to design the kernel unit accordingly.

Therefore, the present invention was created to solve the above-mentioned problems, and when one algorithm is selected and a kernel is used accordingly, unlike the existing structure that undergoes a process of compiling, shutting down, restarting, etc. when another algorithm is needed. In addition, the necessary algorithms for each element of the kernel are implemented in the kernel, and they are registered in the table so that the user can select and execute the algorithms needed at that time without any separate compilation, shutdown, or system restart depending on the application. Its purpose is to provide an operating system capable of dynamic reconfiguration and a method for dynamically reconfiguring the operating system.

In order to achieve the above object, a dynamic reconfigurable operating system for reconfiguring an operating system performed on a single processor system according to the present invention into an operating system having different characteristics makes the module reconfigurable by writing the source code of the operating system, thereby A module configuration unit for generating a table (combination table) having information on a module which can be reconfigured and a module currently being used; If the operating system needs to dynamically reconfigure the operating system, the module identification unit for reconfiguring the operating system by selecting a desired module using a combination table of the module configuration unit may select the desired module. And a module execution unit that executes the selected module.

And a method for reconfiguring the operating system running on a uniprocessor system according to the present invention to an operating system having different characteristics to achieve the above object is composed of a module of the executable form of the operating system, reconfigurable module A first step of generating a table (combination table) having information on the second step: Dynamically reconfiguring the current operating system by dynamically changing the content of the information about the reconfigurable module generated in the step while the operating system is running And the next call to the modified module includes a third step of executing the dynamically reconfigured module.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. The system architecture in which the present invention operates has a functional configuration as shown in FIG. 2, assuming that it is performed in the operating system layer of FIG.

2 is a functional diagram of the present invention is composed of three functional modules, all of which consists of a module configuration unit 200, statically reconfigurable module, the modules to be reconfigured (reconfigurable) It consists of a module identification unit 210 for identifying whether the module execution unit 220 responsible for the actual execution by branching to the module to be reconfigured.

The module configuration unit 200 is shown in more detail as shown in FIG. 3. Compiler / Linker / Loader (202) and Compiler / Linker / Loader (202) to compile, link, and load the source code that design the operating system to make the operating system executable. It is composed of a combination table management unit 204 for extracting the information of the various algorithms and the algorithm module currently being executed to create a combination table. The module configuration unit 200 outputs a mapping table (208) of codes and modules of the operating system that can be executed.

The combination table is a main part of the present invention, which modules currently use which algorithms and which algorithms are selected and which algorithms can be selected and used, with various operating algorithms applicable to each kernel element stored in the kernel. Information management table. And the combination table is composed of a modular system call or a list of various algorithms that can be used in the operating system, the maximum number of algorithms that can be reconstructed in one item, and the number of algorithms currently adopted and used as shown in FIG. .

4 illustrates the module identification unit 210 of FIG. 2 in more detail. The module identification unit 210 determines whether a reconfigurable module is called in a dynamic situation in which an operating system is being performed. It is composed of a reconstruction determination unit 212 and a combination determination unit 214 for branching to an appropriate algorithm by using a combination table generated in the module configuration unit 200 to determine which algorithm is combined with the reconstruction possible. . The output of the module identifier 210 is an appropriately selected algorithm.

The algorithm selected by the module configuration unit 200 and the module identification unit 210 is finally called in the module execution unit 220 shown in FIG. 2 and transferred to execution to complete the construction of an operating system of desired characteristics. .

In the above, the module configuration unit 200 is included in the source code of the operating system when the initial generation of the operating system is generated in a 'static' up to the combination table, the module identification unit 210 is generated by the operating system is performed in a specific system It is the 'dynamic' part that, when present, allows for proper identification by join tables to perform the desired algorithm, that is, the desired operating system.

On the other hand, FIG. 5 and FIG. 6 show the above-described module configuration in a flow chart in terms of time flow. FIG. 5 is a flowchart illustrating the module configuration unit 200. In FIG. 5, a plurality of modules having the same function are designed and edited so as to be reconfigured in a state before an operating system is generated. (Step 500) Then, the source code of the designed and edited operating system is compiled and linked to load the image to be performed into the memory of the target system (step 510), and then generate a binding table (step 520). A possible operating system is created.

FIG. 6 illustrates a flowchart of the module identification unit 210. When the operating system is running (step 600), when a module consisting of several algorithms is called in the operating system (step 610), the called module is reconfigured. Check if this is possible. If it is impossible to reconstruct (step 620), it means that there is no other choice, so the module is performed as it is (step 630). Obtain and branch to the desired module to match it. (Step 650)

Meanwhile, the operating system may be dynamically reconfigured in the dynamically reconfigurable operating system according to the present invention as follows.

As shown in FIG. 5, the static table is a binding table. The association table contains information about which specific algorithm is used and can be used by a module in the dynamic state as described in Figure 3. Therefore, the contents of the binding table can be said to be the core of the present invention. In order to explain in detail, the scheduling of a process that can be referred to as the core of the operating system will be described as an example.

Hereinafter, other functions go through the same process, and thus description thereof will be omitted.

Figure 7 is source code that shows part of the internal data structure for policy in scheduling the operating system to enable dynamic reconfiguration. This corresponds to the first process (step 500) of FIG. 5, which determines which process scheduling algorithm the operating system to design supports and designs / implements codes for. Priority-based, round-robin, and late-monotonic algorithms are defined in a structure called Schedule_PTR. Then, when the second step of step 5 (step 510), that is, other modules and the like, is compiled, a coupling table as shown in Fig. 8 is generated in the kernel. In the combination table of FIG. 8, the Schedule module has five available algorithms in addition to the three mentioned above, and the second algorithm (Round-Robin) is currently used in the Schedule module. In addition, three algorithms can be applied to the replacepage module, indicating that the first one is currently in use.

On the other hand, it is called when the operating system is running an application made up of processes (system call: tcreate, the third item in the combination table) or operating on a module such as Schedule () called by internal operating system requests. When shown in a flow chart, it can be represented as shown in FIG.

For example, assume that for some reason the operating system takes over control and schedules it properly. Then, if the module has been scheduled to the same state as in FIG. 9 (step 610 in FIG. 6), as shown in FIG. 6, the module must be reconfigured with the new algorithm before executing the module of the algorithm to be performed. If it is not a reconstruction module (step 620), the next step is performed immediately (step 630). However, during reconfiguration, the binding table of FIG. 8 is searched to find a current execution algorithm connected to Schedule (). (Step 640) Since the current algorithm of Schedule () is the second, it receives the number of the new algorithm module as a parameter and records it in the current execution item of the join table so that the next algorithm can be executed.

That is, Round Robin, the second item in the data structure of FIG. 7, becomes a true scheduler and branches to be executed as a function pointer.

However, if a situation arises and the scheduling algorithm selected above needs to be changed to Rate Monotonic, which is scheduled periodically, change the number of the current algorithm, 3rd in the combination table of FIG. 8, to 3 . Since various algorithms can be dynamically changed easily, as described above, it can serve as an operating system suitable for various characteristics.

As described above, according to the present invention, since the operating system up to now maintains a fixed structure, it does not flexibly cope with various applications in which not only the information to be processed is increased but also the relationship between the information is complicated, and it is suitable for the field. There was a cost to redesign the operating system or to port a new operating system, but by creating a dynamically reconfigurable operating system architecture, the operating system could be seen as two completely different operating systems from the standpoint of end-user and application programmers. As a result, it is possible to dynamically reconfigure a specialized operating system for various applications and to reduce the above costs. As an example, when emphasizing periodic tasks, ie processes for managing I / O data generated from external devices, the scheduling algorithm for them is rate-monotonic for periodic scheduling. A round robin algorithm capable of time sharing would be suitable for the purpose of allowing the algorithm to be performed and for the fair distribution of resources over time.

In addition, since the functions of the operating system are divided into independent small units, it is easy to implement modularity, so the modularity is increased, thereby increasing the productivity of the software. For example, the scheduling algorithm of processes in charge of the process management unit may have various strategies according to various applications.

Claims (4)

CLAIMS 1. A dynamic reconfigurable operating system for reconfiguring an operating system running on a single processor system into an operating system having different characteristics, the operating system comprising: a compiler / linker / loader for creating an executable form by writing source code of the operating system; A combined table manager for generating a combined table by extracting information of various algorithms included in one data structure and an algorithm module currently being executed from the source code; A reconfiguration determining unit that determines whether a reconfigurable module is called in a dynamic situation in which an operating system is being executed; If the reconstruction determining unit determines that reconstruction is possible, to determine which algorithm is combined, the branching to the appropriate algorithm using the combined table generated by the combined table management unit, if the reconstruction determination unit is not possible If not, the combination determination unit for selecting to perform the module as it is; And a module execution unit configured to execute the module selected by the combination determination unit. The method of claim 1, wherein the association table comprises: a name of a module designed to be reconfigurable; The maximum number of algorithms supported by the module; And information about an algorithm currently being used. CLAIMS 1. A method for reconfiguring an operating system running on a single processor system into an operating system having different characteristics, the method comprising: (a) writing source code of the operating system; (b) compiling, linking and loading the source code of the operating system created in step (a) into an executable module; (c) extracting information of the various algorithms included in the data structure and the algorithm module currently being executed by using the executable module, and generating a table (combination table); (d) calling a module in the operating system while the operating system is running; (e) determining whether the called module is reconfigurable; (f) if the reconstruction is not possible as a result of the determination of step (e), performing the current module as it is, and if reconfigurable, searching the binding table and changing the contents of the binding table to a desired module; And (g) the next call to the module modified in step (f) includes executing the dynamically reconfigured module. 4. The method of claim 3, wherein retrieving the association table in step (g) and changing the contents of the association table to information about a desired module comprises: passing an address of a called module; Finding the matched address among the entries of the passed address and the join table and passing the index of the currently combined algorithm; Identifying the number of times the called module is in a data structure on a source code of an operating system; And writing down at the same position as the identified position among the items of the combination table.