JPH0421893B2 - - Google Patents

Description

Detailed Description of the Invention [Field of Industrial Application] The present invention provides a system in which a single processing device sequentially processes multiple priority processing requests placed in memory. This relates to a method for serializing accesses from multiple processing requests to (regardless of the content).

In the system shown in FIG. 3, there are processing requests 2a, 2b, and 2c in the memory 1, and each of these processing requests has a priority, and the processing device 4 uses the common resource 3. It is assumed that processing is performed sequentially by interrupts from external devices. In this case, the processing device 4 processes the one with the highest priority among the executable processing requests according to its processing procedure. All of these processes are performed under program control, and all processing requests are in an executable state or in an event waiting state. The transition from the event waiting state to the executable state is performed in response to a signal from an external device or an instruction from another processing request. In particular, a processing request (interrupt request) that can be executed by a signal from an external device cannot wait for an event until all its processing procedures are completed.

[Conventional technology]

When a processing request with such a priority attempts to access a common resource, if no other processing request is currently accessing, it displays "Accessing" and then accesses. If another processing request is already accessing, an access request is displayed and the current processing request is placed in the event waiting state. When the access ends, if there is an access request, the waiting processing request can be executed. This conventional method will be explained in detail with reference to FIG.

In FIG. 4A, processing request A (priority level 3) is using resources, and processing request B (priority level 4) is waiting for an event. In Figure 4 A, ECB
(Event Control Bloek) indicates control information for managing resources. In FIG. 4B, processing request B has a higher priority than A, so it can be executed and should be processed first. However, as shown in Figure 4C, processing request A is currently using the resource, so even if processing request B tries to use the resource, it will wait for an event because it is currently using the resource, and processing request A will be processed. The completed post-processing request B becomes executable and is processed first. This state is shown in FIG. 4D.

[Problem that the invention seeks to solve]

There are supervisor functions such as ENQ/DEQ macros for the purpose of exclusive control/sequential use of resources between execution units (tasks). In this method, when there is an execution unit that is using resources, the execution unit that requests it later is placed in a waiting state until the resources are returned, but it has the following problems.

(b) It cannot be used for exclusive control with execution units that cannot enter the wait state (such as interrupt processing). Therefore, it is necessary to operate with interrupts masked, resulting in poor responsiveness. That is, since it cannot be used for exclusive control with processing requests for interrupts from external devices, conventionally exclusive control has been achieved by masking interrupts using hardware. If there are multiple interrupt sources, there is no problem if they can be masked individually, but if they cannot be masked all at once,
(This kind of hardware is more common)
Responsiveness to interrupts not related to exclusion deteriorates.

(b) There are processing requests A, B, and C, and their priorities are P _A , P _B , and P _C , with P _A > P _B > P _C. Now C
A enters the running state while A is using resources, and when it requests the same resource, A enters the waiting state. Here, when B enters the execution state, it gains control because it has a higher priority than C, and as a result, the waiting release is delayed to A, which has a higher priority, because C does not move. This state is shown in FIG. In FIG. 5A, processing request C (priority 2) is in the execution state,
Processing request A (priority level 4) and processing request B (priority level 3) are in an event waiting state. Since processing request A has a priority of 4 and has a higher priority than processing request C, it is allowed to be executed and is processed first. However, when processing request A accesses the resource, processing request C is in use, so it is returned to the event waiting state as shown in FIG. 5C. In this state, when processing request B with priority 3 becomes executable, it is processed first because it has a higher priority than processing request C.
Even though A has a higher priority than B, it cannot be processed because C is not processed.

[Means for solving problems]

The present invention provides an exclusive control method with good responsiveness that solves the above-mentioned problems, and the means for doing so is to serialize access to a common resource by serializing multiple priority processing requests stored in memory. In a system in which processing devices sequentially perform processing, when the processing request accesses the resource, when the resource is being used by a processing request whose priority is lower than that of the accessed processing request, the usage The priority of the processing request in use is raised to the same level as the processing request accessed later and processed first, and when the processing request in use is finished, the priority of the processing request is returned to the original priority. Next, a processing request that accesses the resource later is processed using an exclusive control method.

[Effect]

The above exclusive control method can be used even by processing requests that cannot be placed in a waiting state, does not impair interrupt responsiveness, and does not cause an undue delay in securing resources for processing requests with lower priority than itself.

〔Example〕

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

FIG. 1 shows one embodiment of the present invention. In FIG. 1, at A, processing request C (priority level 2) is in an executable state, and processing requests A (priority level 4) and processing requests B (priority level 3) are in an event waiting state. Figure 1A
Since processing request A has a priority of 4 and has a higher priority than processing request C, it is allowed to be executed and is processed first. This is shown in FIG. 1B. However, when processing request A accesses the resource, processing request C is in use, so the priority of processing request C is increased to 4, which is the same as processing request A, and processing request C is processed first. This state is shown in FIG. 1C. In Figure 1D, processing request B with priority 3 has become executable, but processing request C with priority 4 has become executable.
Indicates a state in which the item is processed first. When the resource usage of processing request C is completed, the priority is returned to the original 2, and processing request A with priority 4 is processed next. This is shown in Figure 1E.

Next, an embodiment of the present invention will be described with reference to the flowcharts of FIGS. 2A and 2B.

When a resource request is made (block 21), its current priority is memorized (block 22). A check is then made to see if there is a processing request in use (block 23). If there is no processing request in use, mark it as a processing request in use (block 24).
and returns (block 25). If there is a processing request in use in block 23, set it to the same priority level as yourself (block 23).
6) First processed by dispatcher (block 27).

In the case of resource return (block 31), processing requests in use are cleared (block 32). Then, compare the current priority with the memo (block 33), and if the priority is the same, return (block 3).
4). If it is different, the priority is returned to the memo (block 35) and the dispatcher returns (block 36).

〔Effect of the invention〕

As explained in detail above, in the present invention, when a resource is accessed, if there is a processing request that is currently using that resource and has a lower priority, its priority is raised to the same level as itself, and the dispatch request is You put it in front of you and process it first. At this time, I remain in the running state,
Problem A mentioned above is eliminated, and the problem of poor responsiveness is solved. Furthermore, since the priority of the processing request in use is raised to the same level, the aforementioned problem (2), in which resource reservation is delayed due to a processing request with a lower priority than the processing request, is also solved.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the procedure of the exclusive processing control method according to the present invention, FIG. 2 is a flowchart for explaining the method according to the present invention, and FIG. 3 is a block diagram for explaining the field of application of the present invention. FIG. 4 and FIG. 5 are diagrams showing the procedure of the conventional method. In the drawing, 26 indicates a block that raises the priority of a processing request in use to the same level as that of a processing request accessed later, and 35 indicates a block that returns a processing request whose priority has been raised to its original priority.

Claims (1)

[Claims] 1. In a system in which a single processing device sequentially processes multiple priority processing requests placed on memory by serializing access to a common resource, when the processing request accesses the resource, the When a resource is being used by a processing request whose priority is lower than that of the processing request that was accessed, the priority of the processing request that is in use is raised to the same level as the processing request that was accessed later. An exclusive control method characterized in that when processing of the processing request in use is completed, the priority of the processing request is returned to the original priority, and then a processing request that accesses the resource later is processed.