JP2008250419A - Competition arbitration apparatus, master-slave system, and method for competition arbitration - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a competition arbitration apparatus for arbitrating between requests from a plurality of master devices to a slave device and which, even if there is a dependency relationship in the order of executing tasks on the plurality of master devices, can provide corresponding arbitration. <P>SOLUTION: Basic priorities are derived based on a table in which the content of a competition process matching a combination of the operating conditions of a plurality of master devices 101 to 103 is preset. Corrections for taking into consideration of a dependency relationship involving tasks in a standby state on the master devices 101 to 103 are made to the basic priorities so as to determine final priorities. The final priorities determined are assigned to access requests from the master devices 101 to 103. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a contention arbitration device, a master / slave system, a contention mediation method, and a program thereof.

  In recent years, in addition to advances in semiconductor microfabrication technology, due to market demands such as cost reduction and performance improvement, it has become natural to mount the entire system on-chip. That is, a plurality of devices that execute different processes have been mounted on the same chip. Furthermore, as measures for improving performance and reducing power consumption, it has become possible to form multi-core processors on the same chip and simultaneously execute software applications that perform different operations on the same chip in parallel.

  When multiple processor cores operate on-chip, the required number of processor cores exist independently. However, slave devices such as memory are not only used for active communication between cores, but also cost-effective. It is often shared for the purpose of down. In other words, a configuration in which a plurality of master devices share a slave device via a bus as a connection means is common in a system-on-chip.

  By the way, as one of the factors that determine the quality of service provided by a multi-core based system, how is arbitration performed when access requests from individual cores in a system to a specific slave device occur at the same time? There is something to do. In other words, it is necessary to determine the relative priority between access requests generated simultaneously from different cores according to the task that issued the request.

  In addition, in such a determination, there may be a case where it is necessary to consider a task that is in a waiting state and does not occupy a processor core at that time. Specifically, a task having a high processing priority in one processor core is forced to enter a wait state because another task in another processor core has not been completed.

  Various mediation methods have been proposed. For example, in the conventional arbitration method described in FIG. 1 of Patent Document 1, a method is used in which the frequency of access by each master device is counted and an upper limit is set for the frequency. The effect is realized. The technique of Patent Document 1 is based on the premise that the master devices that generate access requests to the slave devices are equal to each other, and there is no need to consider the processing order. Such an arbitration method is effective in improving the processing throughput of many symmetrical transactions.

On the other hand, mediation methods aiming at individual optimization have also been proposed. For example, in the conventional arbitration method described in FIG. 2 of Patent Document 2, the priority level of each access request is defined in real time in the form of self-reporting by each master device. The technique of Patent Document 2 makes it possible to cope with individual special requests that cannot be handled by the above-described averaging method in an environment in which various processes of asymmetry are mixed.
Japanese Laid-Open Patent Application No. 08-083211 (Page 11, Fig. 1) Japanese Laid-Open Patent Publication No. 2004-199374 (page 9, FIG. 2)

  However, the competitive arbitration device disclosed in Patent Document 2 has the following problems.

  A first problem is that contention cannot be performed in consideration of the relative relationship between states of a plurality of master devices. The cause of this problem is that when the priority is generated in each master device, it is not possible to make a setting that takes into account the relative relationship with the state of other master devices. When the priority is given as an absolute value, the priority of specific processing is uniquely determined, and an operation that changes according to the situation, that is, the operating state of another master device, cannot be realized.

  For example, for a request issued by one master device, the request must be immediately executed while another master device is executing a predetermined process, but when the other master device is not executing a predetermined process. The request shall be allowed to be delayed. In this case, with respect to a request issued by the one master device, the priority assigned to the one master device should not be uniquely determined by a determination based only on the operating state of the one master device. .

  A second problem is that in a system in which a plurality of tasks are operating on each master device while being switched by the operating system, the priority is that the processing is delayed even though the processing is high in priority. It is that the state called inversion cannot be resolved. The reason why this problem occurs is that consideration is not given to the transition to a waiting state for specific processing performed in many operating systems.

  For example, there is a condition that even if the priority of the task of the second master device is high, the task of the second master device can only be executed after waiting for the completion of processing by the task of the first master device. Suppose. In this case, while the task is being processed by the first master device, the task of the second master device is in a waiting state, and an access request from the second master device does not appear in bus access. However, in order to quickly process the task of the second master device, which is a task with a high priority, it is necessary to prioritize the task of the first master device that is not inherently high in priority. That is, it is not possible to simply perform appropriate arbitration processing by simply comparing priorities among access requests appearing in bus access.

  An object of the present invention is to make an erroneous determination in the arbitration of the entire system in a race condition in which an access request is generated from a plurality of master devices to slave devices shared by the plurality of master devices at the same time. It is an object of the present invention to provide a contention arbitration device, a master / slave system, a contention arbitration method, and a program therefor that can avoid this.

  In order to solve the above-described problem, the contention arbitration device of the present invention includes a plurality of master devices that issue access requests, and a slave device that executes processing according to the access requests issued from the plurality of master devices. In the contention arbitration apparatus provided in the system, basic priority assignment means for assigning basic priority to access requests by the plurality of master devices, and priority change amount of access request by one master device of the plurality of master devices Priority change amount deriving means for deriving according to the operating state of the other master device, and correcting the basic priority by the priority change amount, thereby giving the final priority given to the access request by the one master device And a final priority determining means for determining the degree.

  In the contention arbitration device of the present invention, the priority change amount deriving means derives a priority change amount such that the final priority of the access request by the one master device does not change from the basic priority, or It is preferable to derive a priority change amount such that the final priority of the access request by the one master device is lower than the priority given to the access request by the other master device.

  In the contention arbitration device of the present invention, the priority change amount deriving means may derive a priority change amount of an access request by the one master device according to a task identifier or an interrupt number of the other master device. preferable.

  In the contention arbitration device of the present invention, the priority change amount derivation means derives a priority change amount of an access request by the one master device according to a task identifier of the other master device, The priority change amount deriving unit holds a task identifier of the task of the other master device that is preset as a task to be prioritized by the one master device over the task of the one master device, and the held task When the identifier matches the current task identifier of the other master device, the final priority of the access request by the one master device is lower than the priority given to the access request by the other master device; It is preferable to derive such a priority change amount.

  In the contention arbitration device according to the present invention, it is preferable that the basic priority assigning means assigns a basic priority to an access request from each master device based on a combination of operating states of the plurality of master devices.

  In the contention arbitration device according to the present invention, it is preferable that the basic priority assigning means recognizes the operation state of each master device by a task identifier or an interrupt number of each master device.

  A master-slave system of the present invention includes a plurality of master devices that issue access requests, a slave device that executes processing according to access requests issued from the plurality of master devices, and a contention arbitration device of the present invention. It is characterized by providing.

  A contention arbitration method according to the present invention provides for contention of access requests in a system comprising a plurality of master devices that issue access requests and a slave device that executes processing according to the access requests issued from the plurality of master devices. In the arbitration method, a first step of assigning a basic priority to access requests by the plurality of master devices, and a priority change amount of an access request by one master device of the plurality of master devices are set to another master device. A second step of deriving according to the operation state of the device; and a third priority for determining a final priority given to an access request by the one master device by correcting the basic priority by the priority change amount And a process.

  The program of the present invention is a computer-executable program that causes the computer to execute the contention arbitration method of the present invention.

  According to the present invention, the final priority given to an access request by one master device can be changed according to the operating state of the other master device. Competitive mediation is possible.

  Embodiments according to the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram showing the overall configuration of a master / slave system 100 according to the present embodiment.

  First, the configuration will be described.

  As shown in FIG. 1, a master / slave system 100 according to the present embodiment is shared by a plurality (three in the example of FIG. 1) of master devices 101, 102, and 103 and the plurality of master devices 101 to 103. A slave device 104 that executes processing according to an access request issued from the master devices 101 to 103, and a bus control unit 105 that performs connection between the master devices 101 to 103 and the slave device 104. Yes.

  The master devices 101 to 103 are connected to the bus control unit 105 via data and control signal transmission paths.

  The slave device 104 is connected to the bus control unit 105 via a data and control signal transmission path.

  Therefore, each of the master devices 101 to 103 can access the slave device 104 via the bus control unit 105.

  Each of the master devices 101 to 103 is composed of, for example, a processor core, and the slave device 104 is composed of, for example, a memory.

  The bus control unit 105 includes a contention arbitration unit (a contention arbitration device) 106 that performs mediation processing when access requests from the master devices 101 to 103 to the slave device 104 compete, and the master devices 101 to 103 and the slave device 104. And a master-slave data transfer unit 107 for transferring control signals and data between them.

  FIG. 2 is a block diagram showing the overall configuration of the contention arbitration unit 106 as the contention arbitration device according to the present embodiment.

  As shown in FIG. 2, the contention arbitration unit 106 includes a priority processing table register unit 201 that assigns a basic priority to access requests from the master devices 101 to 103, and any one of the master devices 101 to 103. The relative priority correction determination unit 202 for deriving the priority change amount of the access request according to the operation state of the other master device, and the access request by one master device by correcting the basic priority by the priority change amount A priority request determination unit 203 that determines the final priority given to the request, and a request selection unit 204 that permits an access request according to the determined final priority.

  Among these, the priority processing table register unit 201 and the relative priority correction determination unit 202 are connected to each of the master devices 101 to 103 via task identifier signal lines corresponding to the master devices 101 to 103, for example. The task identifier indicating the current operation state is input.

  Alternatively, the task identifier is output to the priority table register unit 201 and the relative priority correction determination unit 202 by the operating system when the task is switched by the operating system, and the priority table register unit 201 and the relative priority level are output. It may be held by the correction determination unit 202.

  Here, a task is a basic unit of a program operating on the master device, and is handled as a unit of processing by the operating system. The operating system executes a plurality of processes divided into task units while exchanging them on the time axis triggered by various events. The task identifier is assigned with a different value for each task in order to identify the task, and is managed on the operating system. One master device can execute one task at a time.

  The priority processing table register unit 201 has a priority processing table for deriving a basic priority to be given to an access request from each master device 101 to 103 in accordance with a combination of operation states of each master device 101 to 103. keeping.

  Here, the priority (including the basic priority) is a value used to determine which task processing is to be prioritized (first to be performed) between tasks having conflicting access requests.

  The priority is expressed by an integer such as 1 to 32 or 1 to 256, for example. The higher the number, the higher the priority (the higher the priority), and vice versa. The degree is low (priority is slow).

  The priority processing table register unit 201 uses the priority processing table to derive the basic priority of the access request from each master device 101 to 103 in response to the input of the task identifier from each master device 101 to 103. The basic priority is output to the priority request determination unit 203 and the relative priority correction determination unit 202.

  The task identifier is input from each master device 101 to 103 to the priority processing table register unit 201 and the relative priority correction determination unit 202 at the input timing of the access request from each master device 101 to 103 to the bus control unit 105. Alternatively, the task identifiers after the change are changed from the master devices 101 to 103 to the priority processing table when the operation state of each of the master devices 101 to 103 is changed asynchronously with the access request. The information may be input to the register unit 201 and the relative priority correction determination unit 202. In the latter case, the priority processing table register unit 201 and the relative priority correction determination unit 202 store and hold the input task identifier (stored until a new task identifier is input from the corresponding master device. Hold) function.

  The relative priority correction determination unit 202 includes, for example, an inter-task linkage described below from each of the master devices 101 to 103 via an inter-task linkage information signal line corresponding to each of the master devices 101 to 103. Information is entered as needed.

  Inter-task linkage information is information that is input to the relative priority correction determination unit 202 from, for example, a master device that needs to lower its priority when there is a task processing order among the master devices 101 to 103. It is.

  That is, “when one master device executes a task, if another master device is executing a predetermined task, the one master device needs to wait for the end of the predetermined task.” Suppose that there is a task processing order. In this case, the one master device has the predetermined task in order to lower the priority of the access request from the one master device than the access request from the predetermined task (of the other master device). Inter-task linkage information specifying that the priority of its own access request is lower than that is input to the relative priority correction determination unit 202.

  The relative priority correction determination unit 202 that has received such inter-task linkage information determines whether a predetermined task identifier is input to the relative priority correction determination unit 202 from another master device, or from another master device. It is determined whether the relative priority correction determination unit 202 holds the input predetermined task identifier, and it is determined that the predetermined task identifier is input or the predetermined task identifier is held. In this case, the priority change amount is derived so that the priority of the access request from one master device is lower than the access request from the predetermined task (that is, the access request from another master device). The derived priority change amount is output to the priority request determination unit 203.

  That is, the relative priority correction determination unit 202 has a lower priority of an access request from one master device than a basic priority of an access request from another master device input from the priority processing table register unit 201. Such a priority change amount is derived and output to the priority request determination unit 203.

  The priority request determination unit 203 corrects the basic priority input from the priority processing table register unit 201 with the priority correction amount input from the relative priority correction determination unit 202, thereby assigning the final priority to each access request. The priority is determined, and the determined final priority is output to the request selection unit 204.

  That is, for example, by subtracting the priority correction amount from the basic priority, a final priority lower than the basic priority is calculated, and the calculated final priority is output to the request selection unit 204.

  An access request is input from the master devices 101 to 103 to the request selection unit 204 via request signal lines corresponding to the master devices 101 to 103 on a one-to-one basis. Thereby, the request selection unit 204 recognizes what access request is currently issued.

  Further, the request selection unit 204 outputs an access permission signal for each access request to the master-slave data transfer unit 107 according to the final priority output from the priority request determination unit 203.

  The master-slave data transfer unit 107 receives parameters (for example, address information) related to access requests from the master devices 101 to 103. The master-slave data transfer unit 107 sets parameters regarding the access request permitted by the access permission signal input from the request selection unit 204 among the access requests input from the master devices 101 to 103 to the contention arbitration unit 10. Output to the slave device 104.

  The master-slave system 100 configured as described above is mounted on-chip.

  Next, the operation will be described.

  The outline of the operation of the entire system will be described with reference to FIG.

  The master devices 101 to 103 issue an access request to the slave device 104. When the issuance timings of access requests from any of the plurality of master devices 101 to 103 overlap, the contention arbitration unit 106 performs access request arbitration. The slave device 104 sequentially performs processing corresponding to each access request. That is, when the arbitration by the contention arbitration unit 106 is performed, processing corresponding to each access request is performed in the order determined by the arbitration. Then, the slave device 104 outputs response information, data, and the like for each access request to the bus control unit 105 as a result of processing according to each access request. The response information and data output from the slave device 104 are returned to the master devices 101 to 103 that issued the access request via the data transfer unit 107 of the bus control unit 105.

  Details of the operation of the contention arbitration unit 106 will be described with reference to FIG.

  The priority table register unit 201 derives a basic priority to be assigned to an access request issued from each master device 101 to 103, based on a task identifier representing the operation state of each master device 101 to 103. The priority table register unit 201 uses a preset priority processing table for deriving the basic priority.

  In the priority processing table, combinations of basic priorities given to access requests from the master devices 101 to 103 are specified for the combinations of task states that can be taken by the master devices 101 to 103.

  For example, if the number of master devices is k, each master device is designated as a master device M1, M2,. . . , Mk. Each master device M1, M2,. . . , Mk, let n1, n2,. . . , Nk, the number of rows in the priority processing table is (n1 × n2 ×... Nk).

  However, in practice, there may be many task combinations that do not require specification of the priority level, so that a priority processing table in a form that omits that amount can be used.

  The priority processing table may be set by the operating system when the master / slave system 100 is started, or may be rewritten as necessary by the master device.

  Assume that among the master devices 101 to 103, a task in one master device has a time-order relationship with execution of a predetermined task in another master device. When there is a task that enters a waiting state in the one master device in order to wait for the completion of the predetermined task in another master device, the other master device to be prioritized by the one master device or the operating system. The identifier of the task executed on the device is input to the relative priority correction determination unit 202 as task linkage information. The input task identifier (inter-task linkage information) is canceled by the task or operating system when the task of the one master device returns from the waiting state.

  In the relative priority correction determination unit 202, when inter-task linkage information is input from one master device, whether or not to lower the priority of an access request from one master device based on the inter-task linkage information Judgment and derivation of the reduction amount (priority correction amount) are performed. The amount of decrease is a value such that the priority of the access request from the one master device is lower than the priority of the access request from the predetermined task executed by the other master device. The basic priority of the access request from one master device input from the priority processing table register unit 201 is derived by comparing the basic priority of the access request from another master device. That is, for example, if the basic priority of an access request from one master device is 15 and the basic priority of an access request from another master device is 10, for example, the priority correction amount is set to 6, The basic priority of the access request from one master device after correction by the priority correction amount may be set to 9.

  A determination flow in the relative priority correction determination unit 202 will be described with reference to FIG.

  First, the relative priority correction determination unit 202 determines whether or not task linkage information indicating that there is a task of another master device that should be prioritized over the one master device is registered. That is, whether or not the task identifier of the predetermined task is input to the relative priority correction determination unit 202 as inter-task linkage information, or the input task identifier of the predetermined task is the relative priority correction determination unit. It is determined whether or not 202 holds (step S1).

  If the inter-task linkage information is not registered (No in step S1), the priority correction amount is set to 0 and output to the priority request determination unit 203 (step S4). Therefore, the basic priority derived from the priority processing table unit 201 with respect to an access request from one master device becomes the final priority as it is.

  On the other hand, if it is registered that a predetermined task in another master device should be prioritized over one master device (Yes in step S1), whether or not the other master device is executing the predetermined task Is determined (step S2). Here, as a method of obtaining the identifier of the task being executed, a method of inputting the task identifier to the relative priority correction determination unit 202 from each of the master devices 101 to 103 executing the task, or task switching There is a method in which the task identifier after switching only at the timing is input from each of the master devices 101 to 103 to the relative priority correction determination unit 202, and thereafter the task identifier is stored and held in the relative priority correction determination unit 202.

  If a predetermined task is not being executed in another master device (No in step S2), the priority correction amount is set to 0 and is output to the priority request determination unit 203 (step S4), and an access request from one master device is received. On the other hand, the basic priority derived from the priority processing table unit 201 becomes the final priority as it is.

  If a predetermined task is being executed on another master device (Yes in step S2), a priority correction amount is derived according to the inter-task linkage information, and the derived priority correction amount is output to the priority request determination unit 203. (Step S3). Here, this priority correction amount is a correction amount such that the final priority assigned to an access request by one master device is lower than the priority of another master device.

  Therefore, the basic priority derived from the priority processing table unit 201 for the access request from one master device is corrected by the priority request determination unit 203 using the priority correction amount, so that the final priority is obtained. The degree is lower than the priority of other master devices.

  Here, the inter-task linkage information indicates a relative priority relationship between specific tasks. As a specific implementation method, for example, an instruction is issued so that the priority request determination unit 203 outputs a value obtained by subtracting 1 of the minimum unit value from the priority value for the access request by the task to be prioritized. A form given by the relative priority correction determination unit 202 is conceivable. At this time, a mode in which a function for calculating a difference between basic priorities given from the priority processing table register unit 201 is mounted in the priority request determination unit 203 is also conceivable.

  However, the priority correction amount output from the relative priority correction determination unit 202 is not necessarily a value that lowers the priority from the basic priority. That is, as a result of inputting inter-task linkage information such that a task of one master device increases its own priority to the relative priority correction determination unit 202, the priority request determination unit 203 determines the priority of the task of one master device. It is also possible to perform an operation that raises the priority of the task higher than the task priority of another master device. Note that since only relative values are valid values for priority, increasing a specific priority value and decreasing other priorities are equivalent to each other.

  In addition, when the final priority after the basic priority is corrected by the priority correction amount becomes the same value as the priority of the access request of another master device, that is, when a deadlock state occurs. Needs to further change the priority of a particular access request in order to resolve that condition. Specifically, for example, among the tasks included in the deadlock, the task having the highest basic priority held in the priority processing table register unit 201 is forcibly changed to each task included in the deadlock. Of these, the top processing task (the task that executes processing first) can be considered.

  Further, during the interrupt processing, since the task is not executed and the determination based on the task identifier cannot be performed, the determination using the interrupt number as an index indicating the operation state of the master device may be performed. That is, in the above description, the operation using the task identifier as an index indicating the operation state of the master device has been described. However, during the interrupt processing, the operation in which the task identifier in the above operation is replaced with the interrupt number can be performed. .

  The request selection unit 204 may accept only one access request according to the output of the priority request determination unit 203. However, the request selection unit 204 receives a plurality of access requests from each master device and holds them in the stack in order of priority. The form to do is also considered. In the latter case, the access request itself and its priority are held in the stack as a pair in the stack, but in the same way as described above, in order of the final priority of the access request given to each task, The access request may be permitted.

  According to the first embodiment as described above, the following effects can be obtained.

  The first effect is that the final priority given to an access request by one master device can be changed according to the operating state of the other master device, so that the relative relationship of the states among a plurality of master devices is considered. Competitive mediation is possible.

  The second effect is the specification required for the operation of the entire master / slave system 100 by assigning priority given to the access request by each task for each combination of tasks in the plurality of master devices 101 to 103 that issued the access request. This makes it possible to perform more detailed mediation.

  The reason for this is that an access request by a certain task is not determined uniquely from the operating state of the master device that issues the access request, but is a combination of tasks running concurrently on multiple master devices. This is because the relative relationship of the operation states between the master devices can be taken into the arbitration determination by considering the above.

  Furthermore, what should be arbitrated by transmitting information between physically separate devices such as a plurality of master devices is processed at a single location by a single hardware configuration (that is, the competitive arbitration unit 106). This is because high-speed processing becomes possible. In addition, necessary priority correction is performed in the same hardware configuration (competition arbitration unit 106), so that final arbitration information (that is, final priority) is equivalent to the bus transmission time or It will be obtained at a high speed of less than that.

  The third effect is that arbitration can be performed so that not only access requests that are subject to contention but also access requests that will occur thereafter are processed at an appropriate stage.

  The reason is that in the case where the priority of a task executed by one master device is high, the high priority task needs to wait for the completion of processing of the task executed by another master device. Even when the master device task is in a waiting state and the access request does not appear in bus access, the contention arbitration device recognizes such a state from the inter-task linkage information given from each master device, and This is because contention arbitration is possible so that a task in a waiting state can be appropriately recovered from waiting and executed.

  In the above embodiment, an example in which the master slave system 100 includes one slave device 104 has been described. However, a form in which the master slave system 100 includes a plurality of slave devices is also conceivable.

  Further, a mode in which the contention arbitration unit 106 is arranged not between the bus control unit 105 but between the slave device 104 and the bus control unit 105 and performs contention processing for access requests to a specific slave device is also conceivable.

  In addition, the contention arbitration unit 106 includes, for example, a CPU or other processor, a program memory such as a ROM that stores a program for operating the processor, and a work memory such as a RAM that functions as a work area of the processor. The priority processing table register unit 201, the relative priority correction determination unit 202, the priority request determination unit 203, and the request selection unit 204 as described above may be executed by program processing. In this case, the priority processing table may be stored in the program memory or the working memory.

  In addition, the access request from each of the master devices 101 to 103 is input to the relative priority correction determination unit 202 via the request signal line, and the relative priority correction determination unit 202 performs step S1 in the operation of FIG. Before the process, it is determined whether or not an access request from any of the master devices has been issued. Only when the access request has been issued, the processes after step S1 in FIG. 3 may be performed.

It is a block diagram which shows the structure of the whole system which concerns on 1st Embodiment. 3 is a block diagram showing an internal configuration of a contention arbitration unit 106 in the contention arbitration processing apparatus according to the first embodiment. FIG. It is a flowchart which shows the processing flow of the data processor in the 1st Embodiment of this invention.

Explanation of symbols

100 Master / Slave System 101 Master Device 102 Master Device 103 Master Device 104 Slave Device 105 Bus Control Unit 106 Contention Arbitration Unit (Contention Arbitration Device)
107 Master-Slave Data Transfer Unit 201 Priority Processing Table Register Unit (Basic Priority Allocation Unit)
202 Relative priority correction determination unit (priority change amount deriving means)
203 Priority request determination unit (final priority determination means)
204 Request selection part

Claims (9)

In a contention arbitration device provided in a system including a plurality of master devices that issue access requests and a slave device that executes processing according to access requests issued from the plurality of master devices,
Basic priority assignment means for assigning a basic priority to access requests by the plurality of master devices;
Priority change amount deriving means for deriving a priority change amount of an access request by one master device of the plurality of master devices according to an operation state of another master device;
Final priority determination means for determining a final priority to be given to an access request by the one master device by correcting the basic priority by the priority change amount;
Competitive arbitration device comprising: The priority change amount derivation means includes:
Deriving a priority change amount such that the final priority of the access request by the one master device does not change from the basic priority,
Or
The priority change amount is derived such that a final priority of an access request by the one master device is lower than a priority given to an access request by the other master device. The contention arbitration device described. The priority change amount derivation means includes:
3. The contention arbitration apparatus according to claim 1, wherein a priority change amount of an access request by the one master device is derived according to a task identifier or an interrupt number of the other master device. The priority change amount derivation means derives a priority change amount of an access request by the one master device according to a task identifier of the other master device,
The priority change amount derivation means includes:
The one master device holds a task identifier of the task of the other master device preset as a task to be prioritized over the task of the one master device, and the held task identifier is the other master device. And a priority change amount such that the final priority of the access request by the one master device is lower than the priority given to the access request by the other master device. The contention arbitration device according to claim 2, wherein the contention arbitration device is derived. The basic priority assigning means includes:
The contention arbitration apparatus according to claim 1, wherein a basic priority is assigned to each access request by each master device based on a combination of operation states of the plurality of master devices.   6. The contention arbitration apparatus according to claim 5, wherein the basic priority assignment unit recognizes an operation state of each master device by a task identifier or an interrupt number of each master device. Multiple master devices issuing access requests;
A slave device that executes processing according to an access request issued from the plurality of master devices;
The contention arbitration device according to any one of claims 1 to 6,
A master-slave system comprising: In a method of arbitrating contention of access requests in a system comprising a plurality of master devices that issue access requests and a slave device that executes processing according to access requests issued from the plurality of master devices,
A first step of assigning a basic priority to access requests by the plurality of master devices;
A second step of deriving a priority change amount of an access request by one master device of the plurality of master devices according to an operating state of another master device;
A third step of determining a final priority given to an access request by the one master device by correcting the basic priority by the priority change amount;
Competitive arbitration method comprising:   A computer-executable program for causing a computer to execute the contention arbitration method according to claim 8.

Images