KR20000046239A - Realtime multiprocessing system in data processing order according to time limit of processing - Google Patents

Abstract

PURPOSE: A realtime multiprocessing system in data processing order according to the time limit of processing is provided to distribute the process sources more effectively by reflecting the requirement about the time limit of data processing of the real life to the design of task. CONSTITUTION: A realtime multiprocessing system in data processing order according to the time limit of processing consists of three parts. A system core is used in actions between tasks and transmissions of messages by putting GS(Global Sequence) according to the time limit of tasks, generations and terminations of the tasks, the early processing. A task part handles data through executing the task by GS given to each task by the request of the system core. A GS list provides TID(Task ID), the time limit, GS information to the task part.

Description

Real-time multiprocessing system based on data processing sequence according to processing time

The present invention relates to a real-time multi-processing system based on the data processing order according to the processing time, in detail, a complex multi-processing (Multitask) that must process data in real time and one task (Multitask) must process multiple data types The present invention relates to a real-time multiprocessing system based on a data processing sequence to be applied to a system having a tasking structure.

An example of a system that can be applied is a mobile terminal having a limited number of processors while processing various control signals and pay information processing and control signals for various accessories.

1 shows a typical schematic diagram of a conventional real-time multiprocessing system.

As shown in the drawing, a conventional real-time multiprocessing system is largely composed of a system core 100 for managing tasks, which are programs for processing actual data, and corresponding to system requirements of the task, and a task unit 200 for processing actual data.

The system core 100 includes a process rescheduling service (PRS) 110 to select and execute a task to be executed, another system service 120, and a task control block (TCB) list 130 which stores information about a task. )

The PRS 110 stores an ETID (Executing Task ID) 111 for storing a TID of a running task, and a TQ1 (1st Priority Task Queue) 112 for storing TIDs and statuses of tasks according to priorities. 2nd Priority Task Queue) 113, and TQ3 (3rd Priority Task Queue) 114.

The task unit 200 includes a plurality of tasks 210 and MQ (Message Queue) 220 provided in each.

The system core 100 manages tasks, which are programs that process data, and corresponds to system requirements of tasks.

The PRS 110 is a service (system call) that selects and executes a task to execute.

The ETID 111 is a storage location for storing the TID of the currently running task. TID is the only number assigned to each task to identify the task.

The TQ1 112 is a queue type storage location that stores TIDs and statuses of tasks having the highest priority among tasks. Status indicates whether the task is Ready or Suspended. Suspended state is that task is waiting for a specific message. Ready state is not waiting for any message and can be executed immediately.

The TID and Status of the most recently executed task are stored in the last node of the queue.

The TQ2 113 is a queue that stores TIDs and statuses of tasks having the second highest priority among tasks.

The TQ3 114 is a third priority queue, and may have multiple queues such as TQ4, TQ5 ...

The other system services 120 are various services necessary for managing tasks according to the system, such as transfer of messages between tasks in addition to the PRS 110, initial processing of interrupts, creation and destruction of tasks, and priority change of tasks in the system core. Is present.

The TCB List 130 is a list of TCBs that hold information about a task. TCBs for all tasks are stored. As shown in FIG. 2, the TCB generally includes a TID, an area for storing the status of registers of a processor, and the like, and a priority area for storing the queue number to which it originally belongs.

The task 210 is a program execution unit that processes actual data.

Each task 210 has one MQ 220, which is a queue that holds messages coming to the task.

The external environment 300 refers to external hardware or software that sends and receives interrupts or data to this system. Interrupts include interrupts that signal external inputs, including Clock Tick interrupts.

The conventional real time multi-processing system configured as described above operates as follows.

When the PRS 110 is activated, the PRS 110 searches for a task whose Status is Ready, starting with the node in front of the TQ1 112. If there is no Task having a status of Ready in TQ1 112, TQ2 113 is found. If TQ2 113 is not present, TQ3 114 is found last. If not in the TQ3 114, the PRS 110 starts a random idle task, that is, a task that does nothing. As a result, the priority of the task is determined according to whether the task belongs to the TQ1 112, the task belonging to the TQ2 113, or the task belonging to the TQ3 114. The task to be started is moved to the end of the Task Queue indicated by the TCB Priority, which changes the node position on the Task Queue. Moving between task queues checks the time when no task is running and moves from the lower priority queue to the higher priority queue if it is not processed for a certain period of time. This is an example of a number of real-time multiprocessing systems, and there are many ways to prioritize tasks.

The other system service 120 processes the work to be performed, and generally starts the PRS 110 to reselect the priority of the task and the next task to be executed. In a real-time multiprocessing system, interrupts occur at regular intervals, and the PRS is called at this time to prevent a task from taking over the processor continuously.

Each task 200 calls a system service to change its priority in order to create or destroy a task when delivering or waiting for messages.

As described above, various techniques have been proposed for real-time data processing, such as priorities of tasks by themselves, or by priorities of tasks. However, in the real world, the real time system has a processing deadline for each data, but does not provide a mechanism to manage the distribution of resources to ensure that these processing deadlines are met. There is no choice but to keep it. For example, if the data of A should be processed within 1 second after input and the data of B should be processed within 10 seconds after input, the conventional real-time multi-processing system is that if the data of B is input and A data is input within 9.9 seconds, A data. There is a problem that it is difficult to design a task to give a higher priority to a task that processes B data than a task that processes a.

The present invention has been made to solve the above problems, the data processing according to the processing period to enable more efficient distribution of processor resources than the previous system by reflecting the real-time system request for the data processing deadline of the real world in the design of the task as it is It is an object of the present invention to provide a real-time multiprocessing system based on a sequence.

Real-time multi-processing system according to the data processing order according to the processing time limit according to the present invention for achieving the above object, in the real-time multi-processing system according to the data processing order, priority according to the processing time to the task (GS: Global Sequence) Task to process data by performing task by GS assigned to each task at the request of the system core and system core for transition and message transfer between task, creation and destruction of task, and initial processing of interrupt And a GS List provided to the Task unit by including a TID (Task ID), a processing period, and GS information of each task.

1 is a typical block diagram of a conventional real-time multi-processing system,

2 is a structural diagram of a conventional TCB and a task queue;

3 is a configuration diagram of a real-time multiprocessing system based on a data processing sequence according to an embodiment of the present invention;

4 is a structural diagram of a TCB and a task queue of the present invention.

Explanation of symbols on the main parts of the drawings

400: system core 410: PRS

411: ETID and ETGS 412: RTQ

413: STQ 420: IHS

430: GSS 440: Other System Services

450: TCB List 500: Task

510: GSM Task 520: IS Task

530: Idle Task 540: App Task

550: MQ 600: GS List

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

3 is a block diagram of a real-time multiprocessing system based on a data processing sequence according to an embodiment of the present invention.

As shown, the real-time multi-processing system according to an embodiment of the present invention manages tasks that are large programs for processing actual data, and processes a system core 400 corresponding to system requirements of the task and a task 500 for processing actual data. And a GS (Global Sequence) List 600.

The system core 400 includes a PRS 410 for selecting and executing a task to be executed, an Interrupt Handing Sevice (IHS) 420 for handling interrupts of an external environment, and a GSS (for changing and rearranging GS values of a specific task). Global Sequence Service (430), other system services (440), and TCB List (450) that stores information about the task.

The PRS 410 includes an ETID / ETG (Executing Task Global Sequence) 411 for storing TIDs and GSs of a running task, an RTQ (Ready Task Queue) 412 for storing TIDs and GSs of various tasks; It is composed of a suspended task queue (STQ) 413.

The task unit 500 is a GSM Task (Global Sequence Manager Task) (510) to grant and change the GS, the IS Task (Interrupt Service Task) 520 to send a message to the specified Task, and do nothing It consists of an Idle Task 530, an App Task 540 that actually handles data, and an MQ 550 that stores messages that have been delivered to the Task but have not yet been processed.

The system core 400 is a collection of programs or services and data structures necessary for performing transitions between tasks and message transmission, creation and destruction of tasks, and initial processing of interrupts.

The PRS 410 is a service that executes a task to be processed next by rearranging a task execution order.

The ETID / ETGS 411 is a storage location for storing the TID and GS of the currently running task. TID is a unique number assigned to each task to identify the tasks.

The RTQ 412 is a queue that stores TIDs and GSs of various tasks. There are no messages waiting in this queue, so if only tasks that can be executed immediately are stored, the order is determined by the value of GS. GS is assigned to each task in the GSM task and has a value between a specific GSmin and GSmax. GSwnd is a value of (GSmax_GSmin) / 2, and the size comparison between GS is said to be higher priority when GS is small when │GS_a-GS_b│ <GSwnd. do. In the RTQ 412, GS is sorted from high priority items to low items. GS implies a data processing sequence. Special task Tasks IS Task 520, GSM Task 510 and Idle Task 530 are not given GS separately, IS Task 520 has highest priority and GSM Task 510 has next priority It is always assumed to have a higher priority than a Task that has GS. The Idle Task 530 always sees a lower priority than the General Task.

The IHS 420 is a service that handles interrupts from an external environment.

The GSS 430 is a service for changing the GS value of a specific task in the RTQ 412 in response to a request from the GSM Task 510 and relocating the RTQ 412 accordingly.

Other system service 440 may have a number of system services, such as message delivery, task generation, like the conventional system.

The TCB List 450 is a database that stores information about all tasks. As shown in FIG. 4, the TCB List 450 stores various information such as PS, which is an area for storing values such as a TID and a processor register.

The task 500 is a program and data structure unit managed by the system core 400.

The GSM task 510 is a task that receives the message of the App Task 540 and grants and changes the GS of the App Task 540.

The GS List 600 managed by the GSM Task 510 is an annular Linked List composed of items having GS, processing time, and TID information.

The IS task 520 is executed at the request of the IHS 420 and sends a message to a predetermined task according to the interrupt signal. It is assumed that the GS value is virtually assigned to have the highest priority.

The Idle Task 530 is a task that does nothing and assumes that the GS value of the lowest priority is given.

The App Task 540 is actually a task in charge of data processing.

The MQ 550 has one task except for the Idle Task 530 and the IS Task 520, and is a FIFO queue that stores messages that have been delivered to the Task but have not yet been processed.

The real-time multiprocessing system according to the data processing sequence according to the embodiment of the present invention configured as described above operates as follows.

The PRS 410 is started at the request of another system core service or task and, like the services of all system cores, does not reenter until the end of the task. The PRS 410 records the state of the processor (ie, values of all registers, etc.) before the system core call in the PS area of the TCB indicated by the ETID.

Next, GS and ETGS of the foremost node of the RTQ 412 are compared and stored in the ETID and ETGS 411 if the priority of the ETGS is low. Then, the front node of the RTQ 412 is deleted. If the value of the first ETGS is large, the aforementioned operation is not performed. The next task is to restore the values stored in the PS area of the TCB specified by the ETID to the processor state. This action causes the processor to resume the task specified by the ETID.

Forwarding of messages among other system services The items of RTQ 412 and STQ 413 can be interchanged by related system services. At this point, the items added to the queue are always found to be sorted by GS priority. It must be inserted.

IHS 420 is started by the occurrence of an interrupt from the outside. The IHS 420 changes the interrupt handling information of the TCB of the IS Task 520 and moves the node to the first node of the RTQ 412 if the IS Task 520 is in the first node of the STQ 413. Since the IS Task 520 is considered to have the highest priority, it is located in the first item of the RTQ 412 or the STQ 413.

GSS 430 is initiated by a GS change service request from GSM Task 510. The GSS 430 searches the given TID in the RTQ 412 and finds the corresponding node. The GSS 430 rearranges the node on the RTQ 412 by changing the value of GS to the given GS value. The reason for not having to search for the STQ 413 is that when the task sends a GS grant request message to the GSM Task 510, the GSM Task 510 finishes all tasks and waits because the GSM Task 510 has a high priority. This is because the task will be inserted into the RTQ 412 until the state is entered.

The GSS then starts the PRS.

Regardless of the general task form, the GSM task 510 is also implemented in a loop that waits for a message (when a message is received), processes the message, and waits for the message again. The messages waiting for the GSM task 510 are GS grant request message and GS abandoned request message. The GS grant request message contains the absolute time information of the processing deadline. When the GSM task 510 receives the GS grant request message, the GSM task 510 searches the GS List 600, which is an annular list sorted by the due date, to find a position where the given due date is inserted. When the location is found, the GS value of the previous item and the next item is used to calculate a new GS value.

First, the difference between the value of the preceding item and the next item GS is calculated. If the value (hereinafter referred to as GSdif) is greater than GSwnd, GSwnd * 2-GSdif is regarded as a difference (replacement with GSdif). If the value of GSdif is greater than the value of a specific GSstep, the new GS value is equal to the GS value of the previous item plus GSstep (or, if this value is greater than GSmax, this value minus GSwnd * 2). If the value of GSdif is smaller than the value of a particular GSstep, the new GS value is the GS value of the previous item plus GSdif / 2 (or, if this value is greater than GSmax, this value minus GSwnd * 2). do. If there is no preceding item (none includes the case where the item's deadline is later than the given new deadline, even if there is a preceding item in the annular List), the GS value to be added is the GSstep value from the GS value of the latter item. The sum is obtained (or if GSmin is less than GSmin, GSwnd * 2 is added to this value). If there is no trailing item (there is no trailing item in the annular List, even if the deadline is earlier than the given deadline to add), then the GS value to be assigned is the GSstep value of the trailing GS value. The sum is obtained (or this value is less than GSmax minus GSwnd * 2). If there is no front item and no back item, the newly assigned GS is an arbitrary value between GSmin and GSmax. The values of GSwnd and GSwtep are appropriately designed according to the number of tasks and data processing time of the system.

When the GS value to be assigned is determined, the GSM Task 510 makes a GSS service request to change the GS of the corresponding Task on the RTQ 412, and then saves the newly assigned GS, the deadline, and the TID information. Insert it at the insertion position of (600). It searches for other items in the GS List, finds an item with a matching TID, and if there is such an item, deletes it from the GS List (600).

Again the GSM Task 510 repeats from the message waiting state.

When the GSM Task 510 receives the GS Abandon Request message, the GSM Task 510 retrieves the corresponding TID from the GS List 600.

When the node is found, the GS list 600 deletes the node and makes a GS change service request without assigning a GS value. In this case, the GSS 430 finds the corresponding TID in the RTQ 412, deletes the GS value, and moves the message immediately before the Idle Task 530 at the end of the RTQ 412.

Therefore, tasks that do not have GS assigned will be in the slowest order of processing. Although not mentioned here, when creating and deleting a new task among the system services, a task of automatically sending a GS grant request message to the GSM task 510 is processed. Also, in the message delivery service, the GS of the message receiver may be changed according to the processing deadline of the message.

The IS task 520 makes a message waiting system request without specifying the message to wait, and when the IS task 520 resumes, checks the types of interrupts to be processed and interrupts them with a predetermined task according to the interrupt type to be processed. Write a message to let you know that there is a message delivery service request.

When the IS task is resumed, the IS task 520 checks whether there is an interrupt to be processed, and if so, repeats from the interrupt type check, and if not, repeats the message waiting system request.

Idle Task 530 is composed of an endless loop for repeating the processor resume instruction and the processor resume instruction.

The Idle Task 530 has the lowest GS and will only resume when all other tests are in standby.

App Task 540 is a task that processes the actual data are the parties that benefit from the real-time multi-processing system with variable priority described so far. In order to process the priority according to the data to be processed by the App Task 540, App Task 540 should be designed with information on the data processed by each opponent of the App Task 540 and the processing time. .

If a task deadline for a specific data processing is required in the App Task 540, the App Task 540 transmits a GS grant request message to the GSM Task 510. The processing period is the final processing completion time plus the processing period of the data from the current time. If a task requires multiple steps to process data and it takes a lot of time, each request for GS grant in each step increases the efficiency of real-time processing.

If this system is applied and merged with the prioritization method by priority queue, it is possible to add several special-purpose tasks not limited to GS.

Such a system can be applied to any complex and low-cost real-time system, such as a mobile phone, a handheld game console, or a home PC.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and various substitutions, conversions, and modifications are possible within the scope without departing from the technical spirit of the present invention. It will be apparent to those who have knowledge.

As described above, if the real-time multi-processing system according to the data processing sequence according to the present invention is applied, it is possible to distribute the processor resources more efficiently than the conventional system by reflecting the demand for the real-time data processing deadline in the design of the task. It can be done.

Claims (4)

In a real-time multiprocessing system based on a data processing sequence, A system core for assigning a priority (GS: Global Sequence) according to a processing time to a task, for transitioning between tasks, delivering a message, creating and destroying a task, and initial processing of an interrupt; A Task unit for processing data by performing a task by a GS granted to each task by a request of the system core; Real-time multi-processing system according to the data processing sequence according to the processing period, characterized in that it comprises a TID (Task ID) of each task, the processing period, GS information provided to the Task unit. The method of claim 1, wherein the system core, A PRS for rearranging the execution order of tasks according to the TID of the task and the GS given to the task to execute the next task; An IHS for handling interrupts from an external environment; Real-time multi-processing system according to the data processing sequence according to the processing time period characterized in that it comprises a GSS that gives the GS value of the task. The method of claim 2, wherein the PRS is ETID / ETGS for storing the TID and GS of the currently running task; RTQs that store TIDs and GSs of several tasks that can be executed immediately; Real-time multi-processing system based on the data processing sequence according to the processing time period characterized in that it comprises a STQ storing the TID and GS of the various tasks waiting for a specific message. The method of claim 1, wherein the Task unit, A GSM task managing the GS of each task; An IS task for sending a message to a predetermined task according to an interrupt signal from the system core; An App Task that actually handles data; Real-time multi-processing system based on the data processing sequence according to the processing time period, characterized in that it comprises a MQ (Manage Queue) to store the unprocessed message delivered to the task.