KR100675136B1 - Method for operating system based duplication by using MMU - Google Patents

Abstract

The present invention provides a method for operating system-based redundancy using the MMU, comprising: a first step of generating a page error exception in the MMU when a process is running and storing it in a dirty page list by the duplication-only operating system; If there is a context switching between the processes after the first step, a second step of transmitting information on the previously stored dirty page and contexts for the currently running process to the other party through the general purpose communication device; If an error occurs in the process after the second step, the kernel generates an exception to transmit the contents of the currently stored dirty page and the process context to the other party, and the third party performs error processing with the received information. By including the configuration, it is possible to duplicate only the operating system without modifying hardware or application programs, and to recover all hardware errors and software errors that the existing duplexing system cannot cope with.

Description

Method for operating system based duplication by using MMU}

1 is a block diagram of a duplexing device using conventional hardware;

2 is a block diagram of a duplication apparatus using a conventional application program library,

3 is a block diagram showing the operation of FIG.

4 is a block diagram showing the operation of FIG.

5 is a block diagram of an OS-based redundancy apparatus using an MMU to which the present invention is applied.

6 is a flowchart illustrating an operating system based redundancy method using an MMU according to the present invention;

7 is a flowchart illustrating an operation example of a first step in FIG. 6;

8 is a block diagram illustrating an operation example of FIG. 7;

9 is a flowchart illustrating an operation example of a second step in FIG. 6,

FIG. 10 is a block diagram illustrating an operation example of FIG. 9.

FIG. 11 is a flowchart illustrating an operation example of a third step in FIG. 6;

12 is a block diagram illustrating an operation example of FIG. 11.

Explanation of symbols on the main parts of the drawings

20: system

21: general purpose applications

22: Redundant operating system

23: CPU

24: memory

25: general purpose hardware

The present invention relates to an operating system based redundancy method using a memory management unit (MMU) in a central processing unit (CPU), and in particular, a hardware (H / W) or an application program. Redundancy with only operating system (OS) without modification, and OS-based redundancy using MMU that is suitable for recovering both hardware faults and software (software / software) errors that existing duplication methods cannot respond to It is about a method.

In general, redundancy is one of the recovery techniques of a system, and the error is recovered by overlapping the system in two and updating it simultaneously.

There are many methods and techniques for redundancy, but they can be divided into two types: using dedicated hardware and using dedicated libraries.

1 is a block diagram of a duplexing device using conventional hardware.

Where reference numeral 10 is a system, 11 is a general purpose application, 12 is a redundancy only operating system, 13 is a CPU, 14 is a memory, 15 is a redundancy only hardware, and 16 is a general purpose hardware.

Therefore, the redundancy method using dedicated hardware includes two dedicated dedicated hardware 15 in the same system 10, a redundant dedicated operating system 12 for controlling the same, and a general application 11 using the same.

2 is a block diagram of a duplication apparatus using a conventional application program library.

Where reference numeral 10 is a system, 13 is a CPU, 14 is a memory, 16 is general purpose hardware, 17 is a dedicated application, 18 is a dedicated library and tool, and 19 is a general purpose operating system.

Thus, the redundancy method using a dedicated application program library is composed of a general purpose hardware 16 and a general purpose operating system 19, and a redundancy only application program 17 using a redundancy only library 18 in two identical systems 10.

The operation of the prior art will be described in detail with reference to the accompanying drawings.

1. Redundancy using hardware

Thus, the redundancy dedicated hardware synchronizes the CPU 13, the memory 14, and various devices of the two systems 10 with the help of the redundancy dedicated operating system.

The redundancy dedicated hardware 15 communicates with the other system 10 to perform redundancy processing.

Since the redundancy dedicated hardware 15 performs the redundancy processing, there is almost no separate CPU 13 load, and since the redundancy is handled by the lower layer (OS and hardware), the upper general purpose application 11 does not have to consider the redundancy. do.

In addition, since redundancy dedicated hardware 15 handles redundancy, switching over from an active and running system to a standby stand-by system can be accomplished quickly, resulting in complex and high reliability. It is suitable for use in large systems where this is required.

3 is a block diagram illustrating the operation of FIG. 1.

1) The CPU 13 updates data to the memory 14.

2) The BUS between the CPU 13 and the memory 14 is monitored, and the redundancy dedicated hardware 15 passes the contents to the redundancy dedicated hardware 15 of the other system (standby).

3) The redundancy dedicated hardware 15 on the standby side updates the updated contents in the BUS between the CPU 13 and the memory 14.

4) The device performs direct DMA (Direct Memory Access) operation without going through the CPU 13.

5) The redundancy dedicated hardware 15 on the active side monitors the contents of the DMA operation.

6) The redundancy dedicated hardware 15 on the active side passes the contents of the DMA operation to the redundancy dedicated hardware 15 on the standby side.

7) The redundancy dedicated hardware 15 on the standby side updates the contents of the memory 14.

2. Redundancy Using Application Program Library

This is a method of application redundancy using libraries that can implement redundancy on general-purpose hardware and general-purpose OS. The redundancy-only library synchronizes two systems using general-purpose communication hardware.

The application program assigns the contents to be synchronized to the redundancy dedicated library, and the redundancy dedicated library synchronizes them.

Therefore, since it uses general-purpose hardware and OS, it has the advantage of high hardware portability, and is suitable for use in small and medium systems where high reliability is not required.

4 is a block diagram illustrating the operation of FIG. 2.

1) When there is a content to be synchronized with the other party (standby), the dedicated application program 17 using the redundant library makes an update request using the redundant dedicated library 18.

2) The redundant library 18 is transmitted to the other side through the general purpose communication device.

3) The universal communication device on the active side transmits data to the universal communication device on the standby side.

4) The universal communication device on the standby side uploads the data to the redundant library.

5) The redundant library sends the updated contents to the application program.

However, this conventional technology has the following problems.

1) Problems of Redundancy Using Hardware

This requires a lot of redundancy dedicated hardware and a dedicated redundancy OS, which takes a lot of cost and development time, and has a problem of inferior hardware compatibility.

In addition, in the case of a software logical error, there was a problem that recovery is impossible.

2) Problem of Redundancy Using Application Program Library

This requires the use of a library dedicated to redundancy, so there was a problem of redesigning the application.

In addition, when switching over, the context of a running process cannot be recovered, or when recovering, a very high CPU load is required, and software compatibility is inferior.

Accordingly, the present invention has been proposed to solve the above-mentioned conventional problems, and the object of the present invention is to duplicate only the operating system without modifying hardware or application programs, and to solve both hardware error and software error that the existing duplication method does not correspond to. An operating system-based redundancy method using a recoverable MMU is provided.

Operating system based redundancy method using the MMU according to an embodiment of the present invention to achieve the above object,

10. A redundancy device comprising active and standby, comprising: a first step of generating a page fault exception in an MMU when a process on the active side is executing, and storing the page error exception in a dirty page list by a duplication-only operating system; If there is a context switching between the processes after the first step, a second step of transmitting information on a dirty page previously stored and contexts for a currently running process to a standby side via a universal communication device; If an error occurs in the process on the active side after the second step, an exception is generated in the kernel to transmit the contents and the process context of the dirty page currently stored to the standby side , and the standby side performs the error processing with the received information. Including the third step to perform is characterized by its technical configuration.

Hereinafter, an embodiment of the present invention as described above, according to the technical spirit of the OS-based duplication method using the MMU will be described with reference to the drawings.

5 is a block diagram of an OS-based redundancy apparatus using an MMU to which the present invention is applied.

Where reference numeral 20 is a system, 21 is a general purpose application, 22 is a redundant dedicated operating system, 23 is a CPU, 24 is a memory, and 25 is general hardware.

6 is a flowchart illustrating an operating system based redundancy method using an MMU according to the present invention.

As shown in FIG. 1, the MMU generates a page fault exception when the process is running, and stores it in a dirty page list by a redundant dedicated operating system (ST1). Wow; If there is a context switching between the processes after the first step (Transition from Process # 1 to Process # 2), the information about the existing dirty page (Dirty Page) and the contexts for the currently running process A second step ST2 for transmitting to the opposite party (standby side) via the communication device; If an error occurs in the process after the second step, an exception is generated by the kernel to transmit the contents of the dirty page and the process context that are currently being stored to the other side (standby side), and the other side ( The standby side) includes a third step ST3 for performing error processing with the received information.

7 is a flowchart illustrating an operation example of the first step in FIG. 6, and FIG. 8 is a block diagram illustrating an operation example of FIG. 7.

As shown therein, the first step includes: an eleventh step (ST11) (ST12) for generating a page fault exception when the MMU attempts to modify the contents of the memory in process # 1; After the eleventh step, the duplication-only operating system 22 includes a twelfth step ST13 of storing a location of a corresponding page in a dirty page list.

FIG. 9 is a flowchart illustrating an operation example of the second step in FIG. 6, and FIG. 10 is a block diagram illustrating an operation example of FIG. 9.

As shown in the drawing, if the second step includes context switching from the process # 1 to the process # 2 after the first step, the redundancy-only operating system 22 is configured for the dirty page for the process # 1. A twenty-first step ST21 (ST22) for collecting location and process context information; After the twenty-first step, the redundancy dedicated operating system 22 transmits the contents of the dirty page and the process context information to the general-purpose communication device (ST23); A twenty-third step (ST24) of transmitting data to the general-purpose hardware 25 of the other side (standby side) to the general-purpose communication device after the twenty-second step; After the twenty-third step, the standby dedicated redundancy operating system 22 reads and updates dirty pages and process context information from the general purpose communication device (ST25); After the twenty-fourth step, the redundant dedicated operating system 22 on the standby side includes a twenty-fifth step ST26 that performs context switching to process # 2 in the same manner as the active side.

FIG. 11 is a flowchart illustrating an operation example of the third step in FIG. 6, and FIG. 12 is a block diagram illustrating an operation example of FIG. 11.

As shown in FIG. 3, in the third step, if an error occurs in process # 2 after the second step, the redundancy operating system 22 detects an error state through a page error exception (ST31) ( ST32); A thirty-second step (ST33) of transmitting a dirty page and a process context to the opposite party (standby side) by the redundant dedicated operating system 22 after the thirty-first step; A thirty-third step (ST34) of transmitting data to the other party (standby side) to the general-purpose communication device after the thirty-second step; After the thirty-third operation, the standby dedicated redundancy operating system 22 includes a thirty-fourth step ST35 for reading a dirty page and a process context from a general-purpose communication device; After the 34 th step, the redundant dedicated OS 22 performs the same operation as the active through the dirty page and the process context, and includes the 35 th step (ST36) (ST37) for performing error processing. do.

In the above-described step 35, when the cause of the error on the active side is hardware, the standby side operates normally when there is no abnormality in the hardware on the standby side.

In the above-mentioned step 35, when the cause of the error on the active side is a software error, the error occurs on the standby side as in the active side.

In the above step 35, when an error occurs on the standby side as in the active side, the process returns to the previous state by rolling back through the process context information of the previous process # 1 previously stored. do.

The operation of the OS-based duplication method using the MMU according to the present invention configured as described above will be described in detail with reference to the accompanying drawings.

First, the present invention intends to duplicate only the operating system without modifying hardware or application programs, and to recover both hardware error and software error that the existing duplication method does not support.

Thus, in the present invention, the redundant dedicated operating system 22 finds a dirty page using the MMU of the CPU 23 and transmits it to another system (standby side) using a general-purpose communication device in the general-purpose hardware 25. To synchronize.

Each time the general purpose application 21 changes the contents of the memory 24 using the MMU of the CPU 23, a page fault exception can be generated.

However, in general, the size of a page used by the CPU 23 is usually 4096 bytes, and therefore, transferring and synchronizing data each time generates too much CPU load.

In order to prevent this, the present invention stores only the page position without synchronizing every page error. When context switching occurs or an error occurs, the pages are sent to the other side (standby side).

The MMU (Memory Management Unit) is one of the main devices included in most CPUs 23, and is a device that performs memory mapping and memory protection functions.

In addition, redundancy refers to a method in which a system is switched over in case of an error so that there is no disconnection of service in case one system fails to perform normally due to a hardware or software error.

In addition, direct memory access (DMA) refers to an operation in which a device writes a value directly to memory.

In addition, a page fault exception is an exception that the MMU generates when there is no corresponding page in the MMU's page table entry or when there is no access permission.

Also, a dirty page list is a list that contains the address of the memory page where the application modified the value.

In addition, context refers to resource information (CPU register information, I / O (Input / Output, input / output) resource information, etc.) for the process to perform.

In addition, context switching refers to a process in which one process transitions to another process.

Referring to the operation of the present invention in more detail as follows.

1. The first step: run the process (see Figs. 7 and 8)

This is the process when the process is running.

Therefore, when the process is running, the MMU generates a Page Fault Exception, which causes the redundant operating system to store it in a dirty page list (ST1).

1) If process # 1 attempts to modify the contents of memory (ST11), the MMU generates a page fault exception (ST12).

2) The redundant dedicated operating system 22 stores the location of the page in a dirty page list (ST13).

2. Second step: context switching (see FIGS. 9 and 10)

This is the process during context switching.

So if there is a context switching between processes (process # 1 to process # 2), the information about the dirty page stored previously and the contexts for the currently running process can be stored through the general purpose communication device. Transmit to the other side (standby side) (ST2)

1) If there is a context switching from the process # 1 to the process # 2 (ST21), the redundant dedicated operating system 22 collects the position and process context information for the dirty page for the process # 1 (ST22).

2) The data is transmitted to the general purpose hardware 25 of the other side (standby side) to the general purpose communication device (ST24).

3) The standby dedicated redundancy operating system 22 reads and updates the dirty page and process context information from the general purpose communication device (ST25).

4) The redundant dedicated operating system 22 on the standby side performs context switching to process # 2 in the same manner as the active side (ST26).

3. Third step: error handling (see Figures 11 and 12)

This is the process when an error occurs.

Therefore, if an error occurs in the process, the kernel raises a page error exception, transferring the contents of the currently stored dirty page and the process context to the other side (standby side), and then to the other side (standby side). Side) performs the error processing with the received information (ST3).

1) If an error occurs in process # 2 (ST31), the duplication-only operating system 22 detects an error state through a page error exception (ST32).

2) The redundant dedicated operating system 22 transmits the dirty page and the process context to the other party (standby side) via the general purpose communication device (ST33).

3) The data is transmitted to the other party (standby side) to the general-purpose communication device (ST34).

4) The redundant dedicated operating system 22 on the standby side reads the dirty page and the process context from the general purpose communication device (ST35).

5) The redundant dedicated operating system 22 on the standby side performs the operation and performs error processing in the same manner as the active through the dirty page and the process context.

At this time, if the cause of the error on the active side is hardware, the standby side operates normally when there is no abnormal hardware on the standby side.

If the cause of the error on the active side is a software error, the error occurs on the standby side as in the active side (ST36).

6) When an error occurs on the standby side like the active side, the user rolls back to the previous state through the process context information of the previously stored process # 1 (ST37).

As described above, the present invention is to duplicate only the operating system without modifying hardware or applications, and to recover both hardware error and software error that the existing duplexing method does not support.

The present invention can be applied to the main communication system product line (Gateway equipment, IMT2000, switch, etc.), the network product line (switch, router, security equipment, etc.), the server product line (web server, game server, file server, etc.).

Although the preferred embodiment of the present invention has been described above, the present invention may use various changes, modifications, and equivalents. It is clear that the present invention can be applied in the same manner by appropriately modifying the above embodiments. Accordingly, the above description does not limit the scope of the invention as defined by the limitations of the following claims.

As described above, the OS-based redundancy method using the MMU according to the present invention has the effect of dualizing only the operating system without modifying hardware or applications, and recovering all hardware and software errors that the existing duplexing method does not support. Will be.

Therefore, the present invention can cope with a software error that the duplication method using the duplication hardware does not cope with, and can cope with a hardware error that the duplication method using the duplication library does not cope with.

In addition, the present invention also has the advantage that it is possible to reuse the existing general-purpose application program in a system duplexed to the binary level (Binary Level).

In addition, since the present invention supports the redundancy in the operating system, the development time is reduced, the complexity of the redundancy is lowered, and the compatibility is increased because the duplication does not have to be considered when developing new applications.

In addition, since the present invention uses the MMU inside the CPU and synchronizes only necessary portions, the CPU load required for the duplication is lower than that of the duplication using the duplication library.

In addition, since the present invention uses general-purpose hardware, it is possible to reduce development costs and time.

Claims (7)

In the redundancy device configured as active and standby separately , A first step of generating a page error exception in the MMU when the active side of the process is executing, and storing the page error exception in a dirty page list by a duplication-only operating system; If there is a context switching between the processes after the first step, a second step of transmitting information on a dirty page previously stored and contexts for a currently running process to a standby side via a universal communication device; If an error occurs in the process on the active side after the second step, an exception is generated in the kernel to transmit the contents and the process context of the dirty page currently stored to the standby side , and the standby side performs the error processing with the received information. Operating system-based redundancy method using the MMU, characterized in that it comprises a third step to perform. The method of claim 1, wherein the first step, If it is desired to modify the contents of the memory in process # 1, the MMU includes an eleventh step of raising a page error exception; And the twelfth dedicated operating system after the eleventh step includes a twelfth step of storing a location of a corresponding page in a dirty page list. The method of claim 1, wherein the second step, If there is a context switching from the process # 1 to the process # 2 after the first step, the active dedicated redundancy operating system collects the location and process context information for the dirty page for the process # 1; A twenty-second step of transmitting the contents of a dirty page and process context information to a general purpose communication device after the twenty-first step; A twenty-third step of transmitting data to the general purpose hardware on the standby side to a general purpose communication device after the twenty-second step; After the twenty-third step, the standby dedicated redundancy operating system includes: a twenty-fourth step of reading and updating dirty pages and process context information from a general-purpose communication device; And a twenty-fifth step of performing context switching to process # 2 in the same manner as the active side after the twenty-fourth step. The method according to any one of claims 1 to 3, wherein the second step is A 31st step of detecting an error state through a page error exception by a duplication-only operating system on the active side when an error occurs in the process # 2 after the second step; A thirty-second step of transmitting, by the active dedicated redundancy operating system, a dirty page and a process context to a general purpose communication device after the thirty-first step; A thirty-third step of transmitting data to the standby side via the universal communication device after the thirty-second step; After the thirty-third operation, the standby dedicated redundancy operating system includes: a thirty-fourth step of reading a dirty page and a process context from a general purpose communication device; After the thirty-fourth step, the redundant dedicated OS on the standby side performs an operation in the same manner as the active through the dirty page and the process context, and performs an error process, including the thirty-five step of performing error processing. Based redundancy method. The method of claim 4, wherein the 35th step is Operating system-based redundancy method using the MMU, characterized in that the standby side is operating normally when the hardware of the standby side if the hardware of the active side is not abnormal. The method of claim 4, wherein the 35th step is Operating system-based redundancy method using MMU, if the error of the active side is caused by a software error, the error occurs on the standby side as well as the active side. The method of claim 4, wherein the 35th step is If an error occurs on the standby side like the active side, the OS-based redundancy method using the MMU to roll back through the process context information of the previously stored process # 1 to return to the previous state.

Images