KR100315401B1 - Interrupt handling method in real-time OS - Google Patents

Abstract

It is an object of the present invention to separate interrupts into hardware interrupts and software interrupts, thereby minimizing the disable and mask times of interrupts, thereby ensuring the real-time nature of the entire system, and easily implementing interrupt service routines. The present invention provides an interrupt handling method for a real-time operating system. As described above, the present invention divides the interrupt processing method generated in the real-time operating system into a hardware interrupt processing routine and a software interrupt processing routine to shorten the interrupt disable and mask periods occurring during the normal interrupt processing period, thereby performing the interrupt. Minimizing the delay, thereby ensuring the real-time performance of the operating system and improving the system performance, it is also effective to facilitate the development of various interrupt service routines.

Description

Interrupt handling method in real-time operating system divided into hardware interrupt service routine and software interrupt service routine

The present invention relates to a method for processing an interrupt for a real-time operating system. In particular, the interrupt is divided into a hardware interrupt service routine and a software interrupt service routine, thereby minimizing the disabling and mask time of the interrupt and thereby real-time the entire system. Interrupt handling method for real-time operating system that guarantees personality.

In general, interrupt processing greatly affects the overall performance of the real-time operating system. In particular, in the case of a real-time operating system, the interrupt disable period and the interrupt mask period that occur during the interrupt processing period are shortened. Minimize the execution delay of interrupts. In addition, the interrupt processing routine should be configured in a structure that facilitates the development of various types of interrupt service routines.

The conventional method used at this time may use a method of minimizing an interrupt service routine to shorten an interrupt disable and a mask period, or use a method of associating an interrupt with a process scheduler to increase the overall scheduling effect. These existing methods have no special effect when dealing with complex structured interrupts.

In the conventional interrupt processing method, when an interrupt occurs, the interrupt is processed only by a hardware interrupt service routine. In other words, all the interrupt processes are processed in the interrupt mask state. Therefore, the period until the interrupt is generated and terminates becomes the interrupt mask period, and during this interrupt mask period, no response can be made to an interrupt having a level lower than the interrupt level currently being processed. That is, in the related art, the interrupt mask period generated during the interrupt processing is inevitably lengthened in proportion to the interrupt processing period, and the delay that occurs is a technical problem to be minimized in the real-time operating system. In addition, the conventional interrupt handling method has been difficult to deal with interrupts of complex structure that exist at the same level or more.

The object of the present invention is to separate interrupts into hardware interrupt processing routines and software interrupt processing routines, thereby minimizing the disabling and masking time of the interrupts, ensuring the real-time characteristics of the entire system, and easily implementing interrupt service routines. The purpose is to provide an interrupt handling method in a real-time operating system.

1 is a configuration diagram of a hardware system applied to an embodiment of the present invention;

2 is an overall flowchart illustrating a process of processing an interrupt generated in a real-time operating system according to the present invention;

3 is a flowchart illustrating a process of processing a software interrupt on a hardware system to which the present invention is applied.

<Description of the symbols for the main parts of the drawings>

1: Target System of Distributed Structure 2: Subscriber Call Processing Processor

3: operation maintenance processor 4: main processing processor board

5: main processing unit 6: main memory unit

7: ethernet 8: SPCA

9: Interrupt function control device 10: SCSI Hard Disk

11: switching network 12: communication network

13: host system

In the real-time operating system for achieving the object of the present invention, an interrupt processing method includes a first step of disabling an interrupt when an interrupt occurs, obtaining a level of the generated interrupt, and masking an interrupt below the generated interrupt level; ;

A second step of checking whether there is a hardware interrupt service routine to be processed in an interrupt mask state among the generated interrupt level service routines;

If the result of the check in the second step is a hardware interrupt service routine, enable the interrupt, perform the hardware interrupt service routine, disable the interrupt again, interrupt unmask, and the result of the check in the second step does not exist hardware interrupt service routine And a third step of immediately unmasking the interrupt;

A fourth step of calling a software interrupt processing routine to check whether there is a software interrupt service routine which can be processed in an interrupt unmask state among the generated interrupt level service routines;

If the check result of the software interrupt service routine of the fourth stage is present, enable the interrupt, perform the software interrupt service routine, disable the interrupt again, and terminate. The check result of the check of the fourth stage does not exist. And a fifth step of terminating.

Preferably, the fourth step examines the software interrupt service flag of the generated interrupt level to check whether a software interrupt service routine is being executed and immediately terminates if it is being executed, and if not, to execute the software of the generated interrupt level. Setting an interrupt service flag to a running state and then checking whether a software interrupt service routine of the generated interrupt level is registered;

The fifth step is characterized by resetting the software interrupt service flag of the generated interrupt level to the non-executed state immediately before termination.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram illustrating a hardware system according to the present invention.

In the target system 1, a process 2 for call processing of a subscriber and a process 3 for operation maintenance are connected to the switching network 11. Here, the process (2) for the subscriber's call processing and the process (3) for the preservation of operation have a common AMPA (4), which is the main processing process board. It operates on the main memory device 6, and performs a function of controlling the Ethernet 7, which is a connection device with the communication network 12, and the SPCA 8, which is a connection device with the switching network 11. do. In addition, each processor has an interrupt function control device 9 through which an interrupt can be enabled, disabled, masked, unmasked, and the level of the interrupt generated can be obtained. Consists of

In particular, in the operation maintenance process (3) there is an additional SCSI hard disk (10) for connecting to the secondary storage device. The result processed when the interrupt occurs on the target system 1 is transmitted to the host system 13 through the communication network 12 in the operating system of the target system 1, and finally the contents thereof are transmitted to the system operator.

In the present invention, when an interrupt occurs, the hardware interrupt processing routine and the software interrupt processing routine are separated and processed. In the hardware interrupt processing routine, only the functions that can be processed in the masked state are collected and serviced, and the software interrupt processing routine does not need to mask the processing. Collect the possible functions to service. For example, in the case of Ethernet interrupt, in the conventional method, all of these processes are handled by a hardware interrupt service routine. However, in the present invention, only an important part such as a hardware register value change or a global variable value is shortened to a hardware interrupt service routine. It takes a long time, such as copying the rest of the actual message, but separates the functions that can be handled while servicing external interrupts and processes them with software interrupt service routines. Therefore, the interrupt mask period in the present invention can be shortened to the hardware interrupt processing period rather than the entire interrupt processing period.

2 is a flowchart illustrating an operation of a hardware interrupt processing routine according to the present invention.

First, when an interrupt occurs, it is disabled 14 to prevent further interrupts. After that, the level of the interrupt generated after that is obtained (15), and then the interrupt below the level is masked (16).

After masking an interrupt in this way, an interrupt with a level below the level cannot be generated. For example, assuming that the interrupt level of an Ethernet (7) device is 4 and the interrupt level of an SPCA (8) device is 5, the interrupt if the currently occurring hardware interrupt is from an SPCA (8) device of level 5 During the mask period, the Ethernet 7 device does not generate an interrupt. Thus, if this interrupt mask period is prolonged, real-time operation using Ethernet is a major obstacle.

After the interrupt mask, the value of the registered hardware interrupt service routine table corresponding to the interrupt level currently generated is obtained (17). It then checks whether there is a registered hardware interrupt service routine according to this value (18). The hardware interrupt service routine according to the interrupt level collects and registers only the functions that need to be handled only in the interrupt mask state in order to handle the interrupt of the level.

If there is no registered hardware interrupt handling routine, go directly to step 22, if there is a registered hardware handling routine, enable the corresponding interrupt (19) and perform all hardware interrupt service routines registered at the corresponding level ( 20). After all registered hardware interrupt service routines have been performed, disable interrupts (21), unmask interrupts below that level (22), call software interrupt handling routines (23), and before interrupts are generated. Return to the level of (24), and terminate the hardware interrupt processing routine.

3 is an operational flow diagram illustrating a software interrupt processing routine in accordance with the present invention.

The software interrupt service flag SIS [HIL] value corresponding to the hardware interrupt level HIL generated in FIG. 2 is checked (24). Here, Software Interrupt in Service (SIS) [HIL] is a flag value indicating whether there is a software interrupt processing routine currently in operation. If SIS [HIL] is not 0, it indicates that the software interrupt processing routine is already running. ] = 0 indicates that no software interrupt processing routine is currently running. This SIS [HIL] value is used to prevent software processing that is performed with duplicated interrupt levels.

That is, if the SIS [HIL] value is not 0, the software processing routine is terminated immediately because the software processing routine is already running.

However, if the SIS [HIL] value is 0, the value is first set to 1 (25), and the value of the software interrupt service routine table registered at the corresponding interrupt level is obtained (26). According to the value, it is checked whether there is a registered software interrupt service routine (27). If there is no registered software interrupt service routine, go directly to step 31. If there is a registered software interrupt service routine, enable the interrupt (28) and perform all software interrupt service routines registered at the level. (29).

After the software interrupt service routine is performed in this manner, after disabling the interrupt (30), the SIS [HIL] value is set to 0 (31) and the software interrupt service routine is terminated.

In the conventional interrupt processing method, since all interrupt processing is performed in the hardware interrupt service routine, that is, all interrupt processing is performed in the interrupt disable and mask states, until interrupt generation occurs and the interrupt is disabled. , The mask period is increased, which increases the total disable period and the mask period. In other words, there are many functions that can be handled without interrupt disabling and masking during all interrupt processing, but these functions are handled under interrupt disabling and masking status. Will adversely affect.

However, in the present invention, the functions to be processed in the mask state in the interrupt processing are collected and registered in the hardware interrupt service routine, and the functions that can be processed even in the unmasked state are collected and registered in the software interrupt service routine. By executing the hardware interrupt service routine in the mask state and the software interrupt service routine in the interrupt unmask state, the mask period can be greatly reduced. This also improves interrupt response times in real-time systems. In addition, by performing interrupt enable and disable operations appropriately during service routines, the overall interrupt disable period can be shortened.

As described above, the present invention divides the interrupt processing method generated in the real-time operating system into a hardware interrupt processing routine and a software interrupt processing routine to reduce interrupt disable and mask periods occurring during a general interrupt processing period, thereby delaying interrupt execution. Minimize the number, and guarantee the real-time performance of the operating system and improve the system performance, and also make it easy to develop the interrupt service routine of various structures.

Claims (2)

Disabling an interrupt when an interrupt occurs, obtaining a level of the generated interrupt, and masking an interrupt below the generated interrupt level; A second step of checking whether there is a hardware interrupt service routine to be processed in an interrupt mask state among the generated interrupt level service routines; If the result of the check in the second step is a hardware interrupt service routine, enable the interrupt, perform the hardware interrupt service routine, disable the interrupt again, interrupt unmask, and the result of the check in the second step does not exist hardware interrupt service routine And a third step of immediately unmasking the interrupt; A fourth step of calling a software interrupt processing routine to check whether there is a software interrupt service routine that can be processed in an interrupt unmask state among the generated interrupt level service routines; If the check result software interrupt service routine exists in the fourth step, enable the interrupt, perform the software interrupt service routine, disable the interrupt again, and exit. If the check result software interrupt service routine does not exist in the fourth step, immediately Interrupt processing method in a real-time operating system, characterized in that it comprises a fifth step of terminating. The method of claim 1, In the fourth step, the software interrupt service flag of the generated interrupt level is examined to check whether a software interrupt service routine is being executed, and if it is being performed, the process is terminated immediately. Checking whether a service routine of a software interrupt of the generated interrupt level has been registered after setting to a running state; The fifth step may include resetting the software interrupt service flag of the generated interrupt level to a non-executed state immediately before termination.