JPH03282942A - Multiprocessor system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a multiprocessor system that operates as a general-purpose data processor by connecting a plurality of microprocessor boards each having a certain function to a general-purpose bus. .

[Conventional technology]

Generally, in multiprocessor systems, system faults caused by hardware (e.g. memory corruption,
When a bus error (such as an illegal interrupt or bus lock) or a bus error due to software accessing an illegal address occurs within the operating system (O8), the console terminal device for the system operator (SO) will display a message indicating that the system has failed and the details of the error. After outputting , the entire system is stalled (system down).

On the other hand, when the system goes down, the system operator restarts the system by turning off some of the main switches and then turning them on again to turn on the power.

[Problem to be solved by the invention]

However, when the system operator restarts the system, it is undeniable that it takes a certain amount of time for the system's functionality to be restored, and especially when stalls occur frequently due to system faults as mentioned above, the system operator restarts the system. There were inconveniences such as a large amount of effort required to start up the system, and a long period of system downtime, which hindered various operations.

Therefore, if this multi-processor system is a multi-user type, an operator using this system in a remote location will not be able to restart the system himself if the system suddenly goes down. In some cases, the outage period was long.

In addition, in this multiprocessor system, if the system stalls due to the occurrence of a system fault, the main memory
Even if the system is restarted because some or all of the data stored in the local memory on each microprocessor board and the internal registers of the microprocessor are lost, processing before a system fault occurs I was unable to continue.

The present invention has been made in view of the above points, and an object of the present invention is to shorten the period during which a system is stopped due to the occurrence of a system fault, and also to prevent data loss at that time.

[Means to solve the problem]

In order to achieve the above object, the present invention automatically restarts the system when a system fault occurs due to hardware or software in the multiprocessor system as described above, as shown in the functional block diagram in FIG. This is provided with a restarting means A that executes restarting.

In addition, when the system goes down due to a system fault, the data stored in the main memory B, local memory C on each microprocessor board, and internal register D of the microprocessor is transferred to the secondary memory E.
After the system is restarted by the data save means F and the restart means A, each of the above data saved in the secondary storage device E is restored to the main storage device B, the local memory C, and the internal register D, respectively. A data restore means G is provided.

[For production]

In a multiprocessor system according to this invention.

When a system fault occurs due to hardware or software, a part of the entire system is reset, but the restarting means A automatically restarts the system thereafter, thereby shortening the system stop period.

Note that if the system diagnosis (memory test, initialization check for each peripheral device, etc.) that is performed at startup in a normal multiprocessor system is also performed after the system is started up using restart means A, system downtime can be prevented. It becomes possible to easily investigate the root cause.

In addition, when the system goes down due to a system fault, the data save means F saves the data stored in the main memory B, the local memory C on each microprocessor board, and the internal register D of the microprocessor, respectively, to the secondary memory E. After the system is restarted by the restart means A, the data restore means G transfers the above data saved to the secondary storage ME to the main storage B.
Since the data is restored to the local memory C and the internal register D, data loss is reliably prevented, and after the restoration is completed, it is possible to continue the process that was performed before the system fault occurred.

Therefore, the reliability of the system is also improved.

〔Example〕

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

FIG. 2 is a block diagram showing an example of a multiprocessor system according to an embodiment of the present invention.

This multiprocessor system 1 includes a general-purpose
6 or 32 bit CPU and local memory (RAM)
) system console control processor with built-in 2. Main processor 3. System memory 4. File control processor5. It is composed of a system disk control processor 6, a terminal control processor 7, etc., and these are interconnected by a system bus 8.

A system console control processor (SCP) 2 is in charge of input/output control with the system console (SC) 9, and in terms of hardware, it constantly monitors the status of the system bus 8 and performs arbitration when a defect is detected. .

It also has the function of outputting a reset signal onto the system bus 8 to reset all other processors 3, 5-7.

Furthermore, this system console control processor 2
This system also has a low load (faults are unlikely to occur) and has the above-mentioned functions, so
It also has a bootstrap (system restart) function, which will be described later.

Note that a non-volatile memory is also included in order to store the state of a start flag, which will be described later, which is necessary to realize this function.

The main processor (MP) 3 is a processor having a main CPU that centrally controls the entire system, and the operating system and user programs are used here.

The system memory (SM) 4 is a memory corresponding to the main memory B shown in FIG. 1, which is commonly used in this system, and is mainly used for commands (clld) /
Response (data for rspL packet.

and store file data, etc.

A file control processor (FP) 5 is provided to distribute the addition of an operating system, and executes file l10 (reading and writing to and from the disk) according to instructions from the main processor 3.

A system disk control processor (SDP) 6 controls a system disk (SD) 10, which corresponds to the secondary storage device E in FIG. 1, according to instructions from the file control processor 5.

A terminal control processor (TP) 7 is a user terminal control processor, and controls the terminal devices 111 to lln according to instructions from the main processor 3.

It also controls the printer 12.

Figure 3 shows the master/processor of each processor 2, 3.
The relationship between slaves and the system memory 4 is shown.

As you can see from this diagram, main processor (MP) 3
Since this is a mask for all processors, if this main processor 3 causes a fault, the entire system will stop functioning.

Here, two types of faults that cause the main processor 3 to stop functioning will be briefly explained.

■Software fault A bus error, illegal interaction, etc. that can be recognized by the operating system occurs, making it impossible to operate any further and causing a system panic.

■Hardware fault The CPU in the main processor 3 runs out of control or stops due to hardware reasons.

For this reason, when a fault occurs internally, the main processor 3 immediately gives a predetermined instruction to the system console control processor 2 to cause it to perform the processes shown in FIG. 4 and subsequent figures, which will be described later.

Further, if any of the other processors 3, 5 to 7 faults, the main processor 3 detects it and resets that processor.

Performs initialization and diagnostics, and notifies system operators of errors.

For example, by giving instructions to the system console control processor 2 to send error information to the system console 9,
The information is displayed on the display screen to notify the system operator.

Incidentally, the system console control processor 2 also functions as the restarting means A, the data saving means F, and the data restoring means G in FIG.

Further, local memories on all processors 2, 3.5 to 7 and internal registers of the CPU correspond to local memory C and internal register D in FIG. 1, respectively.

Next, the operation of this embodiment configured as described above will be specifically explained with reference to FIG. 4 and subsequent figures.

FIG. 4 is a flowchart showing the processing contents performed by the system console control processor 2 when a system fault occurs.

This routine starts when called by a main routine (not shown), and first, in step 1, it is determined whether a handshake timeout signal, which will be described later, has been input from the main processor 3.

If it is input, the process proceeds to step 2; if not, the process proceeds to step 8, where it is determined whether a system fault notification has been received from the main processor 3.

Then, if a system fault notification has been received, proceed to step 2; if not, return to step 1.

In step 2, the file control processor (FP) 5 and system disk control processor (SDP) 6 determine whether or not they are operating normally. S
D) Instruct the file control processor (FP) 5 to save to 10, and instruct each processor 3.5 to 7 to write the data in its own internal register to the system memory (SM) 4 in step 4. give.

At this time, the system console control processor (SCP)
) 2 The internal body also transfers the data in the internal registers to the system memory 4
Write it down.

Then, in step 5, the file control processor (FP) 5 is instructed to write all the data in the system memory (SM) 4 to the system disk (SD) 10, and in step 6, the file control processor (FP) 5 is instructed to write all the data in the system memory (SM) 4 to the system disk (SD) 10, and in step 6, After turning on the start flag in the non-volatile memory built into the processor 2, the system is reset in step 7, that is, all processors (including the system console control processor 2 itself) are reset, and the process is terminated.

On the other hand, in step 2, the file control processor (
FP) 5 or system disk control processor (S
If the terminal control processor (TP) 6 is not operating normally, in step 9 the terminal control processor (TP) 7 determines whether it is operating normally.

If the terminal control processor 7 is operating normally, an instruction is given to the terminal control processor (TP) 7 so that the fault details can be printed out using the printer 12, and then the process proceeds to step 6.

If the terminal control processor 7 is not operating normally, the fault details are output to the system console (SC) 9 and displayed on its display screen, and then the process proceeds to step 6.

The main processor 3 and the system console control processor 2 regularly perform handshakes, that is, confirm whether or not each other is operating normally.
Output a timeout signal.

Here, each handshake process of the system console control processor 2 and the main processor 3 will be briefly explained with reference to FIGS. 5 and 6.

FIG. 5 is a flowchart showing handshake processing by the system console control processor 2, which starts when called by the main routine.

First, start the timer, and then set the timer value T.
It is determined whether or not the handshake flag has reached the set value T1. At first, the process directly returns to the main routine, and then when the handshake flag reaches the set value T1, it is determined whether or not the handshake flag is on.

If it is not on, the timer value T is set to T2.
It is determined whether the handshake flag has been reached, and if it has not been reached, the process returns to determining whether the handshake flag is on.

If the handshake flag turns on before the timer value T reaches the set value T2, it is turned off, the timer is reset, and the process returns to the main routine.

On the other hand, if the timer value T reaches the set value T2 while the handshake flag remains off, the main processor 38 outputs a handshake timeout signal.

Note that the set values T1 and Tl have a relationship of Tl < Tl.

FIG. 6 is a flowchart showing handshake processing by the main processor 3, which starts when called by the main routine.

First, start the timer, and then set the timer value T.
It is determined whether or not the handshake flag has reached the set value T1. Initially, the process returns to the main routine as it is, and then, when the handshake flag reaches the set value T1, it is determined whether or not the handshake flag is in the off state.

If it is not off, the timer value T is set to T2.
It is determined whether or not the handshake flag has been reached, and if it has not been reached, the process is repeated by determining whether the handshake flag is off.

If the handshake flag turns off before the timer value T reaches the set value T2, it is turned on, the timer is reset, and the process returns to the main routine.

On the other hand, if the timer value T reaches the set value T2 while the handshake flag remains on, a handshake timeout signal is output to the system console control processor 2.

Next, the restart processing after system reset by the system console control processor 2 will be explained with reference to the flowchart of FIG.

This routine starts when called by the main routine, first tests read/write to the system memory (SM) 4, and then determines whether the write start flag in the nonvolatile memory is on.

If the restart flag is not in the on state, - Instruct each processor 3, 5 to 7 to initialize.

Immediately return to the main routine.

If the restart flag is on, it instructs the file control processor (FP) 5 to return the data saved in the save area of the system disk (SD) 10 to the system memory 4, and then returns the data to the system memory 4. An instruction is given to each processor 3, 5-7 to receive the corresponding data (previously saved data) from among the returned data.

At this time, the system console control processor 2 itself also retrieves the fault contents previously saved on the system disk 10 from the system memory 4.

Then, the fault contents are outputted to the system console 9 and displayed on the display screen, and then each processor 3, 5 to 7 is instructed to restart, and the process returns to the main routine.

As a result, all the processors 2, 3.5 to 7, including the system console control processor 2, ie, the entire multiprocessor system 1 is restarted.

In this manner, in the multiprocessor system 1 according to this embodiment, the entire system is temporarily reset when a system fault occurs.

After that, the system will be restarted automatically, so 1.
System downtime is reduced.

In addition, when the system goes down due to a system fault, the data stored in the system memory (main memory), local memory on each processor, and internal registers of the CPU is saved to the system disk (secondary storage), and the system After the system is restarted, each data saved on the system disk is restored to the system memory, local memory, and internal registers mentioned above, so data loss is reliably prevented, and after the restoration is completed, the processing before the system fault occurred can be resumed. You can proceed directly.

Therefore, failures due to system failure can be minimized.

Note that the present invention is applicable to various information processing apparatuses including multi-user, multi-task engineering workstations (EVS).

〔Effect of the invention〕

As described above, according to the multiprocessor system according to the present invention, the period during which the system is stopped due to the occurrence of a system fault can be shortened, and data loss at that time can be reliably prevented.

[Brief explanation of drawings]

FIG. 1 is a functional block diagram showing the basic configuration of this invention. FIG. 2 is a block diagram showing a multiprocessor system which is an embodiment of this invention. FIG. 3 is a master/slave relationship of each processor. FIG. 4 is a flowchart showing the processing by the system console control processor at the time of system failure; FIG. 5 is a flowchart showing the handshake processing by the system console control processor. , FIG. 6 is a flowchart showing handshake processing by the main processor of FIG. 2, and FIG. 7 is a flowchart showing restart processing after system reset by the system console control processor of FIG. 2. 1... Multiprocessor system 2... System console control processor 3... Main processor 4... System memory 5...
File control processor 6... System disk control processor 7... Terminal control processor 8... System bus 9... System console 10... System disk 11... Terminal device 1
2...Printer UL Figure 4 Figure 7

Claims (1)

[Claims] 1. In a multiprocessor system that operates as a general-purpose data processor by connecting a plurality of microprocessor boards with specific functions to a general-purpose bus, a system fault occurs due to hardware or software. 1. A multiprocessor system characterized by being provided with a restart means for automatically restarting the system when the system is restarted. 2. The multiprocessor system according to claim 1, wherein when the system goes down due to the system fault, the data stored in the main memory, the local memory on each microprocessor board, and the internal register of the microprocessor, respectively, is transferred to the secondary memory. data saving means for saving in a storage device; and after restarting the system by the restarting means, each data saved in the secondary storage device is saved in the main storage device, the local memory, and an internal register of the microprocessor. A multiprocessor system comprising: a data restore means for restoring data to each of the data processors;