JPH02158854A - Protocol monitoring system for arbitration - Google Patents

Abstract

PURPOSE:To prevent a whole system from being practically stopped due to a fault unit by providing a bus arbiter with a protocol monitoring circuit, and at the time of detecting a protocol error in a bus using right generating request, informing the error to a unit generating the request. CONSTITUTION:The protocol monitoring circuit 8 always monitors the protocol of arbitration in co-operation with a bus using right control circuit 3, and at the time of detecting an exception, informs the generation of the exception to a bus monitoring circuit 10 and the control circuit 3. The circuit 10 receiving the information informs the generation of the exception from the circuit 8 to an exception generating unit 100-1. When the abnormal request is continuously generated even when the exception generation information is received, the circuit 8 decides the generation of a fault in the unit 100-1, informs the decided result to the circuit 3 so that the bus using right is not given until the recovery of the defective unit 100-1 is confirmed. Consequently, the whole system can be prevented from being practically stopped due to the fault unit.

Description

[Detailed description of the invention] 〔overview〕 As multiple units are linked together on a line.

Equipped with a bus arbiter that issues bus usage rights.

Regarding the no-arbitration protocol monitoring method when performing bus control under the so-called sprint method.

The purpose of this is to prevent the entire system from substantially stopping due to the faulty unit when a fault occurs in the unit.

Create a response line within the above line.

The bus arbiter and the slave side unit monitor the error status of the data being transferred, and when an exception occurs, they can notify the unit of this fact, 91
In addition to configuring the bus arbiter to abort the data transfer that has occurred in 7), a protocol monitoring circuit is provided in the bus arbiter, and when a protocol error regarding a bus use right issuance request is detected, the response line is sent to the unit that issued the request. It has been configured to notify you of this via.

[Industrial application field]

The present invention provides an arbitration protocol monitoring method,
In particular, the present invention relates to a protocol monitoring method for arbitration when bus control is performed under the so-called sprint method, in which a plurality of units are connected on a line and are equipped with a bus arbiter that issues bus usage rights.

Regarding data transfer under the sprint method.

When the masked unit issues a command or command to the slave side, it temporarily releases the line, and when the slave unit completes the specified processing.

The slave side then acts as a mask and transfers the data to the other party. This improves bus usage efficiency.

[Conventional technology]

FIG. 4 shows a conventional configuration, and FIG. 5 shows a time chart in the conventional case.

In FIG. 4, 100-1.100-2. . . represents a husband unit, and 200 represents a bus arbiter. Furthermore, 1 is an information transfer control circuit, 3 is a bus usage right control circuit, 6 is a gate, 10 is a bus monitoring circuit, 2
00-1 represents a data bus, and 200-4 represents a bus start line (STT). Needless to say, each unit 100-1.100-2. . . . have substantially the same configuration and perform the same processing.

It is now assumed that unit 100-1 and unit 100-2 transfer data to other units not shown.

In this case, as will become clearer with reference to the time chart shown in FIG. 3 is the request (B
RQ-0) and returns permission (BGR, T-0) when granting the right to use the bus. The said permission (BCRT-
0) turns on the gate 6 and is taken into the information transfer control circuit 1. The information transfer control circuit l that received the permission (BCRT-0) issues the request (BRQ-0) according to the protocol.
Needless to say, the bus right control circuit 3 drops the bus right request (BRQ-0) from the unit too-2.
After handling the conflict with 1), the above permission (BGRT-Q
) is issued. Therefore, the other unit 100-2 issues a grant (BCRT-1) while sending a request (BRQ-1).
wait.

In unit 100-1, information transfer control circuit 1 sends information prepared on a register (not shown) onto line 200. Data bus 200-1 has a command (C)
, address (A), and data (D) in this order, and are synchronized with the sending of commands.

1τ on the bus start line 200-4.
Outputs bus start (STT) indicating the start of the sequence.

During command, unit 100, which is currently masked,
- Each unit 1 connected to the 0 line 200 where the ID of the other unit that should become a slave is displayed
00-i is the content of bus 200-1 and bus 200.
In this case, when another unit that becomes a slave detects its own ID, it operates as a slave.

The unit 100-1 serving as a mask issues a transfer end (CPT) to the bus monitoring circuit 10 in synchronization with the last transfer data to be transferred. Based on this, the bus arbiter 300 drops the permission (BCRT-O), accepts the request (BRQ-1) from the unit 100-2, and grants the permission (BCRT-O).
BCRT-1). Unit 10 received this
0-2 drops the request (BRQ-1) and connects data bus 2.
Information is sent on 00-1 in the order of command and address. During this time, the bus monitoring circuit 10 monitors errors and the like on the line 200. When transferring data from the unit 100-1 and unit 100-2 to other units, the bus monitoring circuit 10 determines whether an error or the like has occurred in the transferred data. is sent onto data bus 200-1.

Unit 100-1 and unit 100 that became masks
-2 receives the status S and learns whether the previous data transfer was successful or not.

[Problem to be solved by the invention]

In the case of the conventional configuration described above, since a split method is adopted, once the units 100-1 and 100-2, which serve as masks, have finished transmitting their own information, they temporarily occupy the line 200. release. For this purpose 2
The line 200 is no longer occupied by one unit for an undesired long period of time.

However, problems still remain. That is, as shown by the dotted line in the time chart shown in FIG.
) must be negated (dropped), but cannot be negated due to the occurrence of one exception. In the figure, the dotted line indicates a signal that should be negated during normal operation. Unit) 100-1 is Since the information transfer has been completed, a transfer end (CPT) signal is asserted (issued) to declare the end of the information transfer and release the right to use the bus. However, since the request (BRQ-0) is still asserted, the bus arbiter again grants the right to use the bus to the unit 100-1 with a higher priority, and unit l-100-2 cannot obtain the right to use the bus. Unit 1001 requests (BR
Unit 100-2 cannot obtain the right to use the bus unless Q-0) is negated.

The operation of the entire system will stop.

In this way, if an exception occurs in arbitration, such as when the bus acquisition request signal of a unit with a higher priority continues to be asserted or when a unit does not release the right to use the bus, other units will not be able to obtain the right to use the bus. Even though most units were normal, the entire system had essentially stopped operating. That is.

Failure of a single unit could bring down the entire system.

An object of the present invention is to prevent the entire system from substantially stopping due to a faulty unit.

(Means for Solving the Problems) FIG. 1 shows a principle block diagram of the present invention, and the reference numerals 100-1 and 100- in FIG. 1 show a data processing system according to the present invention.
2. ... are each a unit, 200 is a line, 200-
1 is the data bus, 200-2 is the tag bus, 20
0-3 is response line (RL N), 300
1 is a bus arbiter, 1 is an information transfer control circuit, 3 is a bus right control circuit, 8 is a protocol monitoring circuit, and 10 is a bus monitoring circuit.

The information transfer control circuit 1 issues a bus usage right issuance request (BRQ).
) and end of transfer (CPT), and after receiving the bus use permission (bus ground mask BGRM>) that allows the mask unit to use the line 200, information such as commands is sent on the second line 200. Send out.

A response line 200-3 is created in the line 200, and after the bus supervisory circuit 10 receives the bus start (STT) sent on the tag bus 200-2, the slave unit .

Units 100-1 and 100-2 serving as masks
The response to the information transferred from the
It is possible to send out on line 200-3.

The protocol monitoring circuit 8 receives a bus right issuance request (BRQ) sent to the bus arbiter 300.

Upon receiving the contents of the tag bus 200-2 and information from the bus right control circuit 3 (i.e. permission BGRT), a protocol error (exception) is generated in the bus right issuance request (BRQ).
The bus monitoring circuit 10. monitors whether an exception has occurred, and if the occurrence of an exception is detected, the bus monitoring circuit 10. and when a 0 exception occurs, which is notified to the bus right control circuit 3.

Bus monitoring circuit 10 causes error information from protocol monitoring circuit 8 to be placed on response line 200-3.

[Operation] Upon receiving a request (BRQ-0) from the information transfer control circuit 1, the bus right control circuit 3 issues a permission (BGRM-0). In unit 100-1.

Upon receiving the permission (BGRM-0), the gate 6 is turned on, and the information transfer control circuit 1 sends information onto the data bus 200-1, and also sends the bus signal onto the tag bus 200-2.
Star) (STT) is sent.

This allows the slave unit to connect to the bus 20 above.
Receive information on 0-1. The bus monitoring circuit 10
When the start (STT) is received, a permission (BC, R3-1) is issued to the slave unit, the gate corresponding to gate 12 is turned on, and a response is sent from the unit to unit 1004. Make it possible to send. That is, permission (BGR3-1) is a signal that allows the slave unit to use the response line 200-3.

The slave unit is unit 1 (to-
While information is being received from 1, error checks are performed on each piece of information received, and the response is
A response is sent back via line 200-3.

The unit 100-1 serving as a mask continues sending out information while confirming whether or not each piece of information being sent has been correctly received based on the above response.Then, it continues sending out all information. Once completed, it issues a transfer end (CPT) signal on tag bus 200-2.

When the unit 100-1 reports the occurrence of an exception in response while transmitting its own information, it issues a transfer termination (CPT) early so to speak.

It is configured to release line 200. Also.

By sending the end of transfer (CPT) signal onto the tag bus 200-2, the slave unit can know that the response has been correctly transmitted to the unit 100-1.

Assume that an exception occurs in which the unit 100-1 continues to issue requests (BRQ-0) even though it has received permission (BGRM-0).

The protocol monitoring circuit 8 constantly monitors the arbitration protocol in cooperation with the bus right control circuit 3, and when an exception is detected, it receives a zero notification that notifies the bus monitoring circuit IO and the bus right control circuit 3 of the occurrence of the exception. The bus monitoring circuit 10 notifies the protocol monitoring circuit 8 of the occurrence of the exception to the unit 100-1 that has generated the exception. At the same time, the bus right control circuit 3 is set to UNISO when an exception occurs) 10
0-1 bus usage 4m (BGRM-0) is canceled and the path operation is discarded and terminated. 0 The unit 100-1 that was notified of the exception occurrence requests to restore the arbitration protocol to normal operation (BRQ-0). O)
unit 100-1 is determined to be normal, and subsequent bus operations return to normal. If the abnormal request (BRQ-0) continues to be issued even after receiving the exception occurrence notification, the protocol monitoring circuit 8 determines that a failure has occurred in the unit IQO-1, and notifies the bus right control circuit 3 of this. The bus usage right is not granted to the failed unit 100-1 until recovery of the failed unit 100-1 is confirmed, and the failed unit 100-1 is logically separated from the system.

Note that even though an error or the like occurs while information is being transferred from the unit 100-1, the slave unit itself may send a response indicating that there is an error due to a failure, for example. 1 if there is no
Bus monitoring circuit 1 that monitors the status and contents of the line 200
0 sends a response indicating an error on behalf of the slave unit.

〔Example〕

2 and 3 show time charts in the data processing system of FIG. 1, in particular, FIG. 2 shows a time chart during normal times, and FIG. 3 shows a time chart when an exception occurs.

Below, with reference to FIGS. 2 and 3, the unit to
The following description assumes that o-1 and unit 100-2 transfer information to a unit not shown.

As shown in FIG. 2, the information transfer control circuit 1 requests (BR
Q-0), the bus right control circuit 3 issues permission (B
GRM-0) is issued. Upon receiving the permission (BGRM-0), the unit 100-1 turns on the gate 6 and drops the request (BRQ-0). is transmitted on line 200. Data bus 200-1
At the top are commands (C), addresses (A), and data (D).
The information is sent out in the order of 1 and 1 in synchronization with the sending of the command.
Bus start (
STT) is issued on tag bus 200-2.

Among the commands, unit 1, which is the bus master,
00-1 and the ID of the unit to become the slave are displayed, and the unit to become the slave takes in the information on the data bus 200-1, assuming that the information is addressed to itself.

bus star on tag bus 200-2) (STT)
When the bus monitoring circuit IO receives this, it issues permission (BGR3-1) to the slave unit.

The slave unit receives information on the data bus 200-1 every 1τ, and sends a response line (RLN) 20 to report the status at the time of receiving the information.
Sends a response on 0-3.

RC, RA, RD, . . . shown in FIGS. 2 and 3 represent responses as a result of receiving information C, A, D, . . . on the data bus.

The unit 100-1 serving as a mask performs the transfer while confirming that each piece of information has been normally received by the above response every 1τ, and in synchronization with the last transfer, as shown in FIG. Transfer ends on tag bus 200-2 (
CPT).

The bus arbiter 300 responds with a grant (BGR).
M-0), and also sends permission (B
GRM-1) is issued. In response to this, Unit) 1
00-2 similarly performs bus operations.

On the other hand, as shown in FIG. 3, if an exception occurs for some reason in the unit 100-1, the unit 100-1
1, even if the bus usage right (BGRM-0) is acquired, the request (
Bus arbiter 300 that does not negate BRQ-0)
The arbitration protocol monitoring circuit 8 detects that the unit 100-1 does not negate the request (BRQ-0) at a normal timing. Therefore, the protocol monitoring circuit 8 receives permission from the bus right control circuit 3 (
Upon receiving a bus start (STT) from the bus monitoring circuit 10, the CPU determines the timing at which a request (BRQ-0) should be dropped.

The bus arbiter 300 that detected the exceptional operation transfers the response line 200-3, which is a bus for transferring response information.
ER (^rbitr) indicates that an arbitration protocol error has been detected in unit too-t.
cation Protocol Error) signal. At the same time, the bus operation is canceled due to the occurrence of an exception, and the bus operation ends abnormally.

That is, the bus arbiter 300 grants (BGRM-0)
(Gate 6 is turned off) and was waiting for permission) Permission for 100-2 (BGRM-
1) Issue. Accept this.

Unit) 100-2 performs bus operations.

If the unit 100-1 that has been notified of the occurrence of the exception negates the request (BRQ-Q) in accordance with the notification, it will not be determined to be a failed unit and subsequent bus operations will be performed normally. In other words, when UNISO) 100-1 issues a request (BRQ-0) to perform the previously abnormally terminated bus operation again, the permission (BRQ-0) is issued after the current pass over operation is completed.
BGRM-0) is given.

If the unit 100-1 continues to generate protocol errors even with the exception notification, it is determined that the unit 100-1 is a failed unit.

The unit 100-1 determined to be a faulty unit is not given the right to use the bus until it is confirmed that the fault has been repaired.
Logically separated from the system.

The bus monitoring circuit 10 issues a response on behalf of the slave unit.

〔Effect of the invention〕

As explained above, according to the present invention, while the unit serving as a mask is sending out individual information, the unit serving as a slave or the bus arbiter issues a response, and when a failure occurs, etc., the bus arbiter issues a response. In addition to being able to perform early release, even if a failure occurs in one unit, it is possible to prevent the entire system from substantially stopping due to this.

[Brief explanation of the drawing]

FIG. 1 is a diagram of the principle configuration of the present invention, FIG. 2 is a time chart during normal operation in the present invention, FIG. 3 is a time chart when an exception occurs in the present invention, and FIG. 4 is a conventional configuration diagram.
Figure 5 is a time chart during normal operation in the conventional example.
FIG. 6 shows a time chart when an exception occurs in the prior art. In the figure, 100 is a unit, 200 is a line, 200-1
is the data bus, 200-2 is the tag bus, 200
-3 is the response line, 300 is the bus arbiter,
Reference numeral 1 represents an information transfer control circuit, 3 a bus right control circuit, 8 a protocol monitoring circuit, and 10 a bus monitoring circuit.

Claims (1)

[Claims] A plurality of units (100) are connected on a line (200) and are provided with a bus arbiter (300) that issues a bus usage right for occupying the line (200), When one of the units (100-1) requests the bus arbiter (300) to issue a bus usage right and obtains permission, it issues information on the line (200) and ) in a data processing system configured to temporarily release the occupation of the data bus (200), at least the data bus (200)
A response line (200-3) is created together with the tag bus (200-2) and the bus arbiter (300) is connected to the unit (10
The bus arbiter (300) has a protocol monitoring circuit (8) that monitors the presence or absence of a protocol error with respect to the bus use right issuance request from the bus arbiter (200-1), and the bus arbiter (300) is connected to the data bus (200-1) and the tag. - Connected to the bus (200-2) and the response line (200-3), configured to monitor the communication status between the units and stop the communication when an exception occurs, and to monitor the protocol. When the circuit (8) detects the above protocol error, the unit (
100), the above response line (200-
3) An arbitration protocol monitoring method characterized in that the arbitration protocol monitoring method is configured to notify that fact through.