JPH0265339A - Loop bus abnormality detecting method for distributed data processor - Google Patents

Abstract

PURPOSE:To improve detection accuracy by sending a reproducing token to a slave module when intermittent abnormality is generated in a loop bus, specifying a module which detects the intermittent abnormality at first and deciding that the intermittent abnormality is generated in the loop buys in the front step of the module. CONSTITUTION:When the disappearance of the token due to time regulation is detected in a token disappearance detecting part 14 of a master module 10, the disappearance is informed of a token reproducing part 16. The token is reproduced through a packet transmission part 13. This token is passed through a slave module 20 and inputted to a packet reception part 31 of a slave module 30 and here, the token is detected as the token. This token is passed through a slave module 50 and inputted to the master module 10. In a packet data processing part 19, it is detected from an address in an address memory part 18 that the token disappears between the slave modules 30 and 20.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a distributed data processing device using a loop bus, and more particularly to a loop bus abnormality detection method for a distributed data processing device that detects intermittent abnormalities in the loop bus. Regarding.

[Conventional technology]

In a distributed data processing device using a loop bus, in order to detect an abnormality in the loop bus, it is necessary to detect an error in the loop bus transmission path coding rule, or to check whether the frequency component of the frequency used in the transmission path exists above a specified value. A method of detection was used.

[Problem to be solved by the invention]

However, such an abnormality detection method has the following problems. That is, in the former case, although it is possible to detect errors on a per-via basis, it is almost impossible in terms of hardware to count values such as less than 10-9 as an error rate. Therefore, in the case of an error rate of about 10-5 to 10-7 or more, a detection method that is impressive is usually used. However, with such quality, there is a drawback that the location of the failure cannot be detected unless a retransmission process is performed in the transmission procedure of the system and the failure expands to a state where the system is barely operating.

In the latter case, a sequence is added to the functional unit that extracts the tarokk from the signal from the transmission line, and determines an abnormality based on whether the peak value of the voltage amplitude of the clock component is consistently larger than a predetermined value. Therefore, the peak value will naturally be integrated with a certain time constant. Therefore, it is possible to detect burst abnormalities or continuous abnormalities, but
It was impossible to detect bit errors of about 10-g or less.

[Means to solve the problem]

The present invention has, as a token format, an n-bit cycle display field that indicates the number of cycles, and an m-bit cycle abnormal module address field that specifies whether the module is a master module or a plurality of slave modules. Prepare the token you have,
When the ask module detects the disappearance of a token, the master module adds 1 to the lap display field,
It reproduces a token in which its own address, the master module address, is written in the address field, and sends the reproduced token to the slave module via the loop bus.On the other hand, in the slave module, the cycle of the cycle display field of the reproduced token is Compare the number of cycles with the number of cycles of the token passed last time, and if the difference is not 1 and the address written in the address field of the reproduced token is the master module address, reproduce the own slave module address. write to the address field of the token and send it to the loop bus,
When the regenerated token arrives at the master module, the master module examines the address written in the address field of the token and determines that an intermittent failure has occurred in the loop bus preceding the slave module corresponding to that address. It is characterized by the following.

[Effect]

As a result, when an intermittent error occurs on the loop bus, the regeneration token is sent to the slave module and the module that first detected the intermittent error is identified, so that the intermittent error occurs on the loop bus in front of that module. It can be determined that

〔Example〕

Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is an explanatory diagram of the token format of the present invention. The token has a flag F at the right of the beginning and at the end, and the part between these flags F contains a control field C indicating the token and an n-bit circular display field CTR.
, m bits m times g constant detection module address field A and frame check sequence FC3 are formed.

The cycle display field CTR indicates the number of cycles of the token, and is incremented by 1 each time the token cycles. Further, the circulating abnormal module address specifies which module is the master module or each slave module, and the frame check sequence FC8 is used to confirm the normality of the token itself.

FIG. 2 is a system block diagram of a distributed data processing device to which a loop bus abnormality detection method according to an embodiment of the present invention is applied. The master module 10 has a management function for the entire system and manages each slave module 20, slave module 30, and slave module 50. These slave modules 20.30.50 are
All have the same functionality, including the master module lO,
They are connected by a loop bus 60. Since the respective slave modules 20, 30, and 50 all have the same function, the slave module 30 will be represented and explained together with the master module 10.

First, in normal times, tokens are distributed between the master module 10, slave module 20, and slave module 3.
0 and slave module 50, and then rotates on the loop bus.

Here, consider a case where a token is destroyed between the slave module 20 and the slave module 30 due to, for example, an intermittent bit error.

When the token extinguishment detection unit 14 of the master module 10 detects the extinguishment of a token according to the time limit, the token regeneration unit 16 is notified of this. In addition, the value added by 1 to the counter value of the token circulation counting unit 15 is
The token is written in the circulation display field, and the circulation abnormality detection module address field is written. The token is then reproduced through the packet transmitter 13. This token passes through the slave module 20 and is input to the packet receiving section 3I of the slave module 30, where it is detected as a token. At this time, if there is data to be sent by the packet data processing section 40, it is held in the packet passing section 32. On the other hand, if there is no data to be sent, passing processing is performed.

In this case, the current number of cycles detected by the received token cycle number detection unit 35 and the previous number of token cycles that was previously held in the passing token cycle number memory unit 34 when passing the packet a pass 832. The values are compared by a comparator 37.

Then, if the output result of the comparator 37 is not +1 and the circulating abnormality detection module address of the token input this time is the address of the master module 10 in the circulating abnormality detection module address detection section 36, the timing decoder 38 activates the self-address writing control unit 39 and writes the self-address into the circulation abnormality detection module address field of the token held in the packet passing unit 32 or being passed. After that, the number of cycles of the current token is stored in the passing token cycle number memory unit 34, and if there is data to be sent from the packet data processing unit 4a, after controlling the transmission of that data packet, the number of cycles of the current token is stored. If there is no data, the token is output immediately through the packet transmitter 33.

This token passes through the slave module 50 and is input to the mask module 10. Master module 10
Then, the packet receiving section 11 detects that the token is a token, and the packet passing section 12 holds it if there is data to be sent from the packet data processing section 19, and if there is no data to be sent, it is passed through processing. At this time, the address detection control unit 17, which has already been notified of the occurrence of token regeneration from the token regeneration unit 16, examines the circulation abnormality detection module address field of this token, writes it to the address memory unit 18, and sends it to the packet data processing unit 19. At the same time, an instruction is given to the packet passing unit 12 to rewrite the circulation abnormality detection module address field of the token being passed to its own address. The token circulation counting unit 15 also instructs the packet passing B12 to add 1 to the token circulation display field.

At this time, the token extinction detection unit 14 clears the timer and starts counting again, assuming that the token has arrived within the specified time.

If there is data to be sent from the packet data processing unit 19, the token is used after controlling the transmission of the data packet.
If there is no data to be sent, the packet transmitter 13 immediately
output to the loop bus. In the packet data processing unit 19, from the address in the address memory unit 18,
It is detected that the token disappears between the slave module 30 and the slave module 20.

〔Effect of the invention〕

As explained above, according to the present invention, a lap display field and a lap abnormality detection module address field are provided in the token format, the master module adds the number of laps, and the slave module adds 1 to the regular number of laps. If the error detection module address field is the master module, it detects that the module that detected the error for the first time is its own module, adds its own address, and sets it as the master module. Since the system is configured to detect this error, it is possible to logically and accurately quickly discover the location of an intermittent error such as a bit error in a transmission line, which is a problem for the system.

[Brief explanation of the drawing]

Figure 1 is an explanatory diagram of the token format of the present invention, Figure 2 is an illustration of the token format of the present invention.
The figure is a system block diagram of a distributed data processing device to which a loop bus abnormality detection method according to an embodiment of the present invention is applied. lO... Master module, 11.31... Packet receiving section, 12.32.
...Packet passing unit, 13.33...Packet transmitting unit, 14...Token extinction detection unit,
15... Token circulation counting section, 16...
・Token regeneration unit, 17...Address detection control unit, 18...
・Address memory section, 19.40...Packet data processing section, 20.
30.50...Slave module, 34...
...Passed token lap number memory section, 35...
- Receiving token number of cycles detection unit, 36...Module detection module address detection unit, 37...Comparator, 38...Timing decoder, 39...
-Self address write control unit, 60...Loop bus. Applicant NEC Corporation Representative Patent Attorney Umeo Yamauchi

Claims (1)

[Claims] In a distributed data processing device in which one master module and multiple slave modules are connected by a loop bus and mutually transmit and receive packet data using a token ring access method, the token is an n-bit cycle indication indicating the number of cycles. field and an m-bit circulating abnormal module address field for specifying which module it is among the master module and the plurality of slave modules, and when the master module detects disappearance of the token. , the master module adds 1 to the lap display field;
The token is regenerated with its own address, the master module address, written in the address field, and the regenerated token is sent to the slave module via the loop bus, while the slave module reads the regenerated token. The number of laps in the lap display field and the number of laps of the passed token are compared, and the difference is not 1, and the address written in the address field of the reproduced token is the master module address. Sometimes, it writes its own slave module address into the address field of the regenerated token and sends it to the loop bus, and when the regenerated token arrives at the master module, the master module writes the address field of the regenerated token. Loop bus abnormality detection for a distributed data processing device, characterized in that an address written in an address field is checked and it is determined that an intermittent failure has occurred in the loop bus at the previous stage of the slave module corresponding to the address. Method.