JPH057208A - Multi-processor system - Google Patents

Abstract

PURPOSE:To quickly decide the presence or absence of the trouble of each processor in a multi-processor system with the small traffic of the communication between processors. CONSTITUTION:In a multi-processor system, during the communication between processors, when a transmitting source processor 21 detects the abnormality of a transmitting destination processor 22, it is informed that the abnormality exists at a master processor 30. A CPU 60 receives the information, adds '1' to an abnormality informing frequency counter 72 corresponding to the processor 22, and checks whether or not the abnormality informing frequency counter 72 exceeds the prescribed threshold. When it exceeds, it is decided that the trouble exists at the processor 22, and when it does not exceed, the trouble deciding processing (health checking) performed originally periodically is performed even when the period does not reach, and it is decided whether or not it is the trouble.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multiprocessor system in which a plurality of processors are connected via a bus and the processors communicate with each other. The present invention relates to a multiprocessor system having a function of determining whether a processor is a faulty processor.

[0002]

2. Description of the Related Art Conventionally, in a multiprocessor system of this kind, a master processor having a failure determination function periodically performs failure determination processing (health check) on each slave processor as a method of determining a failed processor. When the source processor detects an abnormality in the destination processor during communication between the processors and the method of determining the failure of the processor by performing the transmission, the source processor notifies the master processor of the abnormality (notice),
The master processor accumulates the number of times each slave processor is notified of an error, and when the error notification count exceeds a specified threshold, the processor is determined to be the faulty processor. ing.

[0003]

However, of the above-mentioned two methods, in the method of receiving an abnormality notification from the transmission partner and comparing the number of times of notification and a threshold value to determine whether or not there is a failure, If the number of error notifications of each processor does not exceed the threshold value, it is not judged as a failure. Therefore, there is a problem that it takes time to determine a failure in a processor with low traffic between processors.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a multiprocessor system capable of determining a failed processor early even when the traffic of processor mutual communication is small. It is in.

[0005]

According to the present invention, in a multiprocessor system in which a plurality of processors are connected via a bus and communicate with each other, a source processor is a destination processor during communication between the processors. When an abnormality is detected, the abnormality notification receiving means for receiving the abnormality notification, the cumulative processing means for accumulating the abnormality notifications received by the abnormality notification receiving means, and the threshold value of the abnormality notification frequency for each processor are set. Threshold value setting means, abnormality notification frequency check means for determining whether the cumulative value calculated by the cumulative value processing means has reached the threshold value set by the threshold value setting means, and the processor It is determined that the accumulated value does not reach the threshold value by the failure determination processing means for performing the failure determination processing and the abnormality notification frequency check means. It is obtained by a structure in which a failure determination starting means for starting the failure determination processing unit to focus.

In the multiprocessor system of the present invention,
The failure determination by the failure determination processing means is performed only for the processor that has received the abnormality notification, and when the abnormality notification frequency check means determines that the accumulated value has reached the threshold value. Determines that the destination processor has a failure.

With such a configuration, in the multiprocessor system of the present invention, when the source processor detects an abnormality of the destination processor during communication between the processors and reports the abnormality, the cumulative total of the abnormality notification and the threshold value are It is determined that there is a failure when compared and the cumulative value reaches the threshold value. On the other hand, if the accumulated value does not reach the threshold value, the failure determination processing that is originally performed periodically is executed even if the cycle has not come, and it is determined whether or not the processor is the failure processor. .

[0008]

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

FIG. 1 shows the configuration of a multiprocessor system according to an embodiment of the present invention.

In this multiprocessor system, n processors 21 to 2n and a master processor 30 are connected to a bus 1.
It is configured to be connected via 0, and the processing is performed while the processors 21 to 2n and the master processor 30 communicate with each other.

The multiprocessor 30 includes a CPU (Central Processing Unit) 60 for executing a failure determination process and a receiving unit 40 for receiving data on the bus 10 and transmitting it to the CPU 60.
And a transmission section 50 for transmitting data from the CPU 60 onto the bus 10, a counter 70 and a ROM (read only memory) 80 connected to the CPU 60. The counter 70 is composed of a plurality of abnormality notification number counters 71 to 7n corresponding to the processors 21 to 2n, respectively. The failure processing program according to the flow chart described later is stored in the ROM 80.

FIG. 2 shows the basic configuration of this multiprocessor system. In this multiprocessor system, when the transmission source processor detects an abnormality in the transmission destination processor during communication between the processors, the abnormality notification receiving means 61 for receiving the abnormality notification and the abnormality notification receiving means 61 for receiving the abnormality notification are received. Cumulative processing means 62 for accumulating the abnormal notifications, threshold setting means 63 for setting the threshold value of the number of abnormal notifications for each processor, and the cumulative value calculated by the cumulative processing means 62 is the threshold setting means. An abnormality notification frequency check means 64 for determining whether or not the threshold value set by 63 has been reached, and a failure determination processing means 6 for performing a failure determination process for the processor to which the abnormality notification is made.
6 and a failure determination activation means 65 that activates the failure determination processing means 66 when the abnormality notification frequency check means 64 determines that the cumulative total has not reached the threshold value.

Next, the failure determination processing operation in the multiprocessor system of this embodiment will be described with reference to the flow chart of FIG. In this embodiment, for example, the processor 2
2 is an abnormal processor, processors 21 and 23-2n
Each as a normal processor, and the number of processors n is 9
The threshold value for failure determination is set to 5. In the present embodiment as well, the above-described periodic failure determination processing (health check) is performed.

First, when the processor 21 detects an abnormality in the processor 22 while the processor 21 and the processor 22 are communicating with each other via the bus 10, this processor 2
1 notifies the master processor 30 via the bus 10 that the processor 22 is abnormal. Master processor 3
At 0, the CPU 60 receives the abnormality notification from the processor 21 via the receiving unit 40 (step S1), and thereby increments the abnormality notification number counter 72 corresponding to the processor 22 in the counter 70 by 1 (step S2).
To do. As a result, the count value of the abnormality notification number counter 72 becomes "1".

Next, the CPU 60 checks whether or not the abnormality notification number counter 72 exceeds a threshold value of 5 (step S3). Here, the abnormality notification frequency counter is "1".
Therefore, the processor 22 is not determined to be a failure, but the CPU 60 further executes the failure determination processing that is originally performed periodically as follows even if the cycle has not come.

That is, when the processor 21 gives an abnormality notification, the CPU 60 sends the notification from the transmitter 41 to the processor 22 via the bus 10 even if the number of notifications has not reached the threshold (step S3; N). A check signal is sent to (step S4). After the processor 22 receives the signal, when the response signal is returned via the bus 10, the master processor 30 receives the returned signal at the receiving unit 40 and transmits it to the CPU 60 (step S
5). The CPU 60 checks the returned signal to detect whether or not there is an abnormality (step S6), and if there is an abnormality (Y), determines that the processor 22 is the faulty processor (step S7). When it is determined that there is no abnormality (step S; N), the process returns to step S1.

Further, in the multiprocessor system of this embodiment, it is assumed that the faulty processor 22 is subjected to the fault judging process.
Processor 22 cannot detect
~ Processor 27, processor 22
Each of ~ to 27 gives an abnormality notification once. For this reason,
The count value of the abnormality notification number counter 72 becomes 6 (step S3; Y) and exceeds the threshold value of 5, so that the processor 22 is determined to be the faulty processor.

As described above, in the multiprocessor system of the present embodiment, when the source processor detects an abnormality of the destination processor and notifies the abnormality during the communication between the processors, the master processor 30 originally performs the periodical operation. The failure determination process is executed even if the cycle has not come. Therefore, even when the traffic of communication between processors is small, the failure of the processor can be determined early.

[0019]

As described above, according to the multiprocessor system of the first to third aspects, when the source processor detects the abnormality of the destination processor and notifies the abnormality during mutual communication between the processors, Since the failure determination processing that is originally performed periodically is executed for the processor notified of the abnormality, there is no need to perform useless operations such as the determination processing for a normal processor. Therefore, there is an effect that the failure of the processor can be judged at an early stage especially when the traffic of the communication between the processors is small.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a multiprocessor system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing the basic principle of the present invention.

FIG. 3 is a flowchart for explaining a failure determination operation of the multiprocessor system of FIG.

[Explanation of symbols]

10 bus 21-2n processor 30 master processor 40 Receiver 50 transmitter 60 CPU 61 Cumulative processing unit 62 Abnormality notification frequency check processing unit 63 Failure determination start-up unit 64 Failure determination processing unit 70 counter 71-7n Abnormality notification frequency counter 80 ROM (Read Only Memory)

Claims (3)

[Claims] 1. In a multiprocessor system in which a plurality of processors are connected via a bus and communicate with each other, when a source processor detects an abnormality of a destination processor during communication between the processors, the abnormality An abnormality notification receiving means for receiving the notification, and a cumulative processing means for accumulating the abnormality notifications received by the abnormality notification receiving means,
Threshold setting means for setting a threshold of the number of times of abnormality notification for each processor, and whether the cumulative value calculated by the cumulative processing means has reached the threshold set by the threshold setting means. An abnormality notification frequency check means for determining whether or not, a failure determination processing means for performing a failure determination process of the processor, and the abnormality notification frequency check means, if it is determined that the cumulative value has not reached the threshold value, A multiprocessor system comprising: a failure judgment starting means for starting a failure judgment processing means. 2. The multiprocessor system according to claim 1, wherein the failure determination by the failure determination processing means is performed only on the processor that has been notified of the abnormality. 3. When the abnormality notification frequency check unit determines that the cumulative value has reached a threshold value, it is determined that the destination processor has a failure. 1. The multiprocessor system according to 1.