JPH0363769B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a control method for a communication control device for a data communication system.

[Background of the invention]

Figure 1 shows the central processing unit CPU and main memory MM
This shows a data communication system consisting of a communication control device CCP, and a data channel device DCH connecting the CPU and CCP. CCP consists of multiple processing units PUi (i=0, 1,
..., n-1) and the data bus that connects PUi
It has an interface control unit BC that controls the BUS and the connection between the BUS and DCH, and further includes a line control unit LCi and a storage device Mi corresponding to the processing device PUi. Figures 2 and 3 are from the CPU, respectively.
A method according to the prior art for loading a program onto a CCP is shown.

The problems of the prior art will be explained below using FIGS. 1, 2, and 3.

The program to be loaded from the CPU to the PUi must be divided into appropriate lengths and loaded, taking into account the probability of error occurrence on the BUS.
One unit of this divided program is written as a packet.
Let's call it PKT, the CPU that sends the PKT
PKT error detection control and retransmission control when an error occurs is required between the PKT and the receiving PUi. Furthermore, in order to reduce loading time, a simultaneous loading method is generally adopted in which one PKT sent from the CPU is received by all PUs at the same time.

First, FIG. 2 will be explained.

In FIG. 2, the program to be loaded is divided into m PKTs by the CPU, assigned an error detection code, and transmitted from the CPU. As shown in Figure 2, the first PKT
When 0 is sent from CPU, DCH, BC and
It is received simultaneously by all PUs via BUS. At this time, each PU checks the error detection code assigned to PKT0 and stores whether or not an error has occurred.
Next, the CPU first transmits an inquiry signal POL ₀ to PU ₀ in order to inquire all PUs whether PKT0 has been correctly received. This POL ₀ to PU ₀
If there is no error, an acknowledgment will be sent based on the previously memorized record of whether or not an error has occurred.
Send ACK ₀ to CPU. When the CPU receives this ACK ₀ , it next sends POL ₁ to PU ₁ , and when it receives ACK ₁ from PU ₁ in response to this POL ₁ , the CPU sends POL to the next PU. When this operation is performed in sequence and ACK _o-1 is received from PU _o-1 , it means that PKT0 has been correctly received by all PUs. Next, when the CPU transmits the next PKT1 and POL ₀ to PU ₀ as described above, PU ₀ may not be able to correctly receive PKT1 and transmit a negative response NAK ₀ . When the CPU receives this NAK ₀ , all PUs
PKT1 is retransmitted. In this way, the PKTs are sequentially transmitted, and the final response to the final PKT _(n-1), ACK _o-1, from PU _o- 1 is received, and the loading of all programs is completed.

The method shown in FIG. 2 has the problem that the program loading time becomes long because all PUs are inquired about the packet reception status by POL once and for all.

Next, FIG. 3 will be explained.

In this method, programs are loaded simultaneously to all PUs as shown in Figure 2, but all PKTs from PKT0 to PKT _{(n-1) are loaded simultaneously.}
, and then finally sends POL to all PUs to inquire about the reception status of PKT. This method completes program loading when ACK is received from all PUs. But for example
PU ₁ is any PK from PKT0 to PKT _(n-1)
is received by mistake, NAK ₁ is sent in response to POL ₁ from the CPU. When the CPU receives this NAK ₁ , it sends POL _o-1 to the last PU _o-1 , and after receiving the response from PU _o-1 , it restarts all programs only for PU ₁ from now on. Loading.

In this method, reloading for the error-generated PU is significantly delayed, and the retransmission program monopolizes the data channel and the BUS in the CCP, interfering with the communication operations of other PUs.

As described above, the conventional method has problems in that it takes a long time to load a program and when an error occurs during loading. This is a big problem, especially when the number of PUs is large in CCPs for large-scale systems.

[Purpose of the invention]

An object of the present invention is to provide communication control that eliminates the drawbacks of the prior art described above, shortens the initial loading time of a program from a central control unit to a communication control unit, and realizes prompt reloading when a program loading error occurs. The objective is to provide a control method for the device.

[Summary of the invention]

In the present invention, a positive response to program loading is confirmed by not receiving a negative response from any processing device for a certain period of time, and reloading is performed when a negative response is received within a certain period of time. It is.

[Embodiments of the invention]

The present invention will be explained below with reference to FIG.

After transmitting PKT0 to all PUs via DCH and BC, the CPU postpones transmission of the next PKT1 for a fixed time interval t. During this t,
If no signal is received from any PU, PKT
0 is interpreted as correctly received by all PUs. Then, the next PKT1 is sent. in the same way
Transmission is performed sequentially up to PKT _(o-1) ,
For example, if you send PKT2 and it is received by PU ₁ by mistake, PU ₁ will indicate that.
Send NAK to CPU. If the CPU receives this NAK within the time t after sending PKT2, it will receive the NAK again.
Send PKT2 to all PUs. After that, if NAK is not received from all PUs even after t has elapsed, the next PKT3 is transmitted.

As described above, the present invention confirms a positive response to a PKT by not receiving a NAK from any PU for a certain period of time.

As shown in FIG. 1, since the BUS is shared by all PUs, each PU cannot respond at the same time. This is because the conventional method stipulates a certain response format in order to identify the PU that sent the response, so as shown in Figures 2 and 3, when an inquiry from the CPU , a polling method is required in which responses are sent sequentially from specified PUs. On the other hand, in the method according to the present invention,
A positive response is indicated by not transmitting any signal from all PUs to the CPU for a certain period of time after transmitting the PKT, and a negative response is indicated by the PU detecting an error in the PKT transmitting some amorphous signal to the CPU. This can be realized by sending it to the CPU in time for display.

In addition, there is a certain time t for displaying the response after sending the PKT.
If the PKT length is long, there will be no effect of reducing the loading time, but it is sufficient to send a signal of several bytes to indicate a negative response, and on the other hand, the length of one PKT is at least several hundred bytes, so the length of the PKT The length of t can be made sufficiently short compared to .

〔Effect of the invention〕

Recently, data communication systems have become larger and larger, and efforts have been made to improve the processing power of communication control devices. However, when implementing this using a multiprocessor system, increasing the number of processors
There is a problem that the initial loading time of the program is large and the system start-up time is long. However, according to the present invention, the loading time depends only on the amount of the program, regardless of the number of processors. This has the effect of easily increasing the scale of the system.

[Brief explanation of drawings]

Figure 1 is a diagram showing a data communication system consisting of a central processing unit and a communication control unit, Figures 2 and 3 are
FIG. 4 is a system flow diagram showing the initial loading of a program from a central processing unit to a processor in a communication control device according to a conventional method, and FIG. 4 is a system flow diagram showing an embodiment of a control method according to the present invention. CPU...Central processing unit, CCP...Communication control unit, MM...Main memory, DCH...Data channel device, BC...Bus control unit, BUS...Data bus, PUi...Processing unit, Mi... Storage device, LCi
...Line control unit, PKT...Packet, POL...
Inquiry signal, ACK...affirmative response, NAK...negative response, t...response monitoring time.

Claims (1)

[Claims] 1 Data communication consisting of a communication control device configured using a multiprocessor system that connects multiple storage program control processing devices via the same bus, and a central processing unit that controls this communication control device via a data channel. In the system, when performing initial program loading from the central processing unit to each processing unit in the communication control unit, if the program is successfully loaded to all processing units,
Each processing unit does not send a response to the central processing unit, but if a program is loaded by mistake, the incorrectly loaded processing unit sends a negative response within a certain period of time to send a negative response to the central processing unit. A control method for a communication control device characterized by prompting reloading of a program in which a problem has occurred.