JPS63157251A - Restoring system for information stored in main memory - Google Patents

Abstract

PURPOSE:To realize a continuous operation of an information processing system even in case a main memory has a fault during operation of said system, by separating the faulty area and sweeping the normal contents into an external memory to restore these contents into the main memory after the second rise. CONSTITUTION:When a main memory 2 has a fault, a normalcy check area 21 in a main memory constitution unit 20 is checked by a read-and-write operation via a main memory fault separating means 7. If an error is detected, it is decided that the unit 20 is defective and other normal units are written into an external memory 4. When a saving operation is through with the information on the main memory, the second rise is carried out by an OS rise means 6. Then the saved main memory information is read out to the memory 2 by means of an external memory reading means 5. Thus a continuous operation of an information processing system is ensured even after the second rise.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a fault handling method in an information processing system.

[Conventional technology]

Conventionally, when a failure occurs in a main memory, the system is restarted, ignoring the contents even if the contents of the main memory other than the failed part are readable.

[Problem that the invention seeks to solve]

As mentioned above, once a main storage failure occurs,
Even if the contents of the main memory other than the faulty part were readable, they were ignored and the system was restarted, so information that required continuity, such as file update information, became invalid. This was a serious obstacle to system operation.

[Means for solving problems]

The method for restoring information in main memory of the present invention is such that when a main memory failure occurs,
a main memory failure part isolating means for accessing the main memory and isolating the faulty part; an external storage writing means for writing the contents of the normal main memory other than the faulty part to the external storage; an O3 startup means for restarting the operating system after the main memory of the operating system is disconnected; and an external storage device reading means for transferring information from the external storage device to the main storage device after the operating system is started up. There is.

[Effect]

Therefore, even in the event of a failure of the main storage device, it is possible to avoid loss of information essential for continuity of system operation.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of an information processing system to which the main memory information restoration method of the present invention is applied.

This information processing system includes a central processing unit 1, a main storage device 2, a processing bag W3, and an external storage device 4.

The central processing unit 1 includes an external storage device reading means 5, the processing device 3 includes an oS startup means 6, a main memory failure part isolation means 7, and an external storage device writing means 8. There is.

When a fault occurs in the main memory device 2 during operation of the information processing system, the faulty part of the main memory in the processing device 3 is isolated using means 7 as follows. First, O
Before starting S, a normality check area 21 of about 1 page (1024 bytes) is secured in advance for checking the normal operation of the main memory 2 for each 20 main memory unit (bank or memory unit). Note that this check area 21 is a check-only area, so when the O3 startup means 6 notifies the OS of the usable main memory range, the normality check area 21 is not accessed from the OS. Addresses are excluded, and a special data pattern is written in the normality check area 21 by the means 7 for main memory failure site isolation. When a fault occurs in the main storage device 2 during operation, an error occurrence notification is sent from the main storage device 1 to the processing device 3, and a means 7 for isolating the faulty part of the main memory in the processing device 3 is activated. The main memory failure site isolation means 7 reads the contents of each normality check area 21 in the main memory unit 20, checks whether the previously written data pattern is read correctly, and then writes the same data pattern again. After writing, read and check. If a data comparison error or an error occurs in writing or reading, the main storage unit 20 containing the target to be checked is determined to be defective and is separated from the main storage unit 2.

For the main memory portion determined to be normal, the contents of the main memory whose operation has been determined to be normal are read out using the external storage device writing means 8 in the processing device 3, and then written to the external storage device 4. In addition, when the external storage device M4 is a magnetic disk, the external storage device writing means 8 of the processing device 3 is notified in advance of the evacuation position of the main memory contents after the OS has completed startup. Alternatively, the fixed area may contain only the pointer value of the evacuation area and the evacuation area may be secured at a variable position. However, if the external storage device writing means 8 writes to a wrong magnetic disk address, In order to avoid data destruction due to writing, the OS writes an identification code agreed upon in advance between the O3 and the processing unit 3 in the first few words of the main memory information saving area for checking the validity of the area. , the external storage device writing means 8 reads this identification code and checks the validity of the area before saving the main memory information to prevent accidental data destruction.After the main memory information is saved, the first word of the save area is read. After writing the information write flag to, the OS startup means 6 reconfigures the main memory and then restarts the OS. When O3 starts up, it checks whether information has been saved using the write flag in the first word of the main memory information saving area of the external storage device 4. If the information has been saved, the part of the main memory that can be read normally even in the event of a main memory failure by reading the saved main memory information to the main memory device 2 by the external storage device reading means 5 of the OS. can be taken over even after restarting O3.

〔Effect of the invention〕

As explained above, even if a failure occurs in the main memory during operation of an information processing system, the present invention isolates the faulty part and temporarily flushes out the contents of the main memory that can be read normally to an external storage. By restoring information from the external storage device to the main memory when the O3 is started up again, there is an effect that it is possible to easily take over information that particularly requires continuity in system operation.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of an information processing system in which the main memory information restoration method of the present invention is applied. l... central processing unit, 2... main storage device,
3... Processing device, 4... External storage device, 5
... External storage device reading means, 6 ... oS startup means, 7 ... Main memory failure part isolation means. 8... External storage device writing means, 20... Main storage device configuration unit, 21...
Area for health check. 1st rXJ

Claims (1)

[Scope of Claims] Main memory fault isolation means that accesses the main memory and isolates the faulty part when the main storage fails, and writes the normal contents of the main storage other than the faulty part to an external storage. an external storage device writing means for re-starting the operating system after the failed main memory has been disconnected; A method for restoring information in main memory, comprising an external storage device reading means.