JPH07319825A - Lock control system for shared table - Google Patents

Abstract

PURPOSE:To unlock a shared table by automatically restoring the shared table to an original state when an program interruption is generated during the locking of the shared table. CONSTITUTION:When a lock request to a table 111 is outputted from a process 4, a locking mechanism 6 locks the table 111 and then accesses a table retreating/restoring mechanism 10 to retreat the table 111 to a table retreating area 2. The process 4 executes processing for referring and updating the table 111, but if a program interruption is generated before the end of the processing, control is transferred to an exception handler 5. The handler 5 outputs an unlock request specifying a restoration request to an unlocking mechanism 7, which restores the table 111 retreated to the retreating area 2 to a table group 1 and then unlocks the table 111.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a lock control method for a shared table in a multiprocessor system, and more particularly, it automatically restores the shared table to its original state when a program interrupt occurs while the shared table is locked. The present invention relates to a lock control method for a shared table that is configured to release a lock.

[0002]

2. Description of the Related Art In a multiprocessor system,
Since different processes (also called tasks) are executed simultaneously for each processor, the table on the main memory that is shared by the processes is exclusively used, and only one process is temporarily applicable. It is necessary to lock the table so that it can be referenced and updated. Here, as a table on the main memory that is commonly used among processes, there is a management table (a table that manages the usage status of the resource, etc.) corresponding to each resource that is referred to and updated in resource allocation processing and the like. is there.

By the way, in a multiprocessor system for controlling the lock of a shared table, if a program interrupt occurs during the execution of a process that locks a certain shared table and the processing is interrupted, how to lock the shared table is decided. It becomes a problem.

For example, in Japanese Patent Laid-Open No. 1-261771, when a process that locks a certain shared table interrupts the process due to an interrupt, the shared table is kept locked. There is a problem that if you try to lock because it is necessary to refer to and update, a deadlock state occurs because it has already been locked.

This kind of problem seems to be solved by forcibly unlocking the shared table when the above process is interrupted, but the above process completely updates the shared table before interrupting the process. If not finished
The content of the shared table will be unlocked while it is incomplete, which will cause a malfunction in the operation of referring to and updating the shared table.

Therefore, conventionally, when the shared table is locked, the lock display is set at a predetermined position on the main memory, and the process requiring the lock is normally completed and unlocked. The exception handler refers to the lock display when a program interrupt occurs and the system is forcibly stopped if the display is set. ing.

[0007]

As described above, in the prior art, when a program interrupt occurred while the shared table was locked, the system was immediately stopped. Therefore, the occurrence of the program interrupt leads to an immediate system stop.
There was a problem that the throughput of the system was significantly reduced.

Among the program interrupts, in addition to the ones that essentially make it impossible for the system to continue, for example, the processing in which an abnormality has occurred is macro processing, and the abnormality is caused by the user, such as a designation error. There are some things that can be notified to the user of a specification error etc. without stopping, and even in the case of this latter program interruption,
Stopping the system due to problems with shared table locking is not really desirable.

In order to continue the system without stopping it when a program interrupt of a type that can originally continue the system occurs, the state of the shared table locked by the process interrupted by the program interrupt is consistent. It is necessary to unlock after restoring the state, and a technique that enables such processing is desired.

The present invention has been proposed in view of the above circumstances, and an object thereof is to automatically restore the shared table to the original state when a program interrupt occurs while the shared table is locked. The purpose of the present invention is to provide a lock control method for a shared table that can be unlocked by a lock.

[0011]

According to the lock control method of a shared table of the present invention, a process of saving a shared table in the main memory to a table save area and a shared table saved in the table save area to the original memory of the main memory In response to a lock request, the shared table on the main memory is locked, and the shared table on the main memory is called to return the locked shared table to the table save area. In response to an unlock request that specifies the lock mechanism for saving to the table and the need to restore the table, the table save / restore mechanism is called to restore the shared table saved in the table save area to the original location in main memory. And then unlock the shared table,
In addition, in response to an unlock request that specifies that the table need not be restored, an unlock mechanism that unlocks the shared table and deletes the unlocked shared table from the table save area, and a table interruption when a program interrupt occurs An exception handler for sending an unlock request designating restoration required to the unlock mechanism is provided.

[0012]

According to the lock control method for the shared table of the present invention, when a process executed on any of the processors requests a lock of a certain shared table, the lock mechanism responds to the lock request by the main unit. The corresponding shared table on the memory is locked, and the table save / restore mechanism is called to save (copy) the locked shared table to the table save area. A process that successfully locks a shared table refers to the corresponding shared table in main memory,
Update and execute predetermined processing. When this process terminates the processing that requires locking of the shared table normally and issues an unlock request specifying that table restoration is not required, the unlock mechanism responds to the unlock request and locks the corresponding shared table. At the same time as releasing it, the table saved in the table save area is deleted. On the other hand, if a program interrupt occurs while processing that requires the lock of the shared table is being executed in the above process, control passes to the exception handler, and the exception handler unlocks the unlock request specifying that the table needs to be restored. In response to the unlock request, the unlock mechanism calls the table save / restore mechanism to restore the shared table saved in the table save area to the original location on the main memory. The shared table is unlocked so that the shared table can be referred to and updated without conflict in subsequent processing.

[0013]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a main part of a multiprocessor system to which a lock control method for a shared table according to the present invention is applied. The multiprocessor system of this example includes two processors CPU1 and CPU2 and a main memory MEM shared between the two processors, and the main memory MEM includes various tables to be subjected to exclusive control. There is a table group 1 and a table save area 2 which is a save area for the table group 1. Further, processes 3 and 4 related to the user program and OS, an exception handler 5, a lock mechanism 6, an unlock mechanism 7, a save table element return mechanism 8, a restore table element return mechanism 9, and a table save. The restoration mechanism 10, the restoration permission / inhibition registration mechanism 11, and the pattern information setting mechanism 12 are present in the main memory MEM as functional units that can be executed by either of the processors CPU1 and CPU2. Incidentally, 13 is a pattern information table, and 14 is a restoration availability display.

The table group 1 includes a plurality of tables 111 each serving as an exclusive unit, and a lock table group 112 including a lock table corresponding to each table 111. In the case of this embodiment, each table 111 is
It has a configuration as shown in either (1) or (2) below.

(1) As shown in FIG. 2, a structure in which a plurality of table elements 212 to 214 holding data to be actually referred to and updated are chained from the queue terminal 211. Here, all of the table elements 212 to 214 have the same size in XXX, a pointer for pointing to the next table element is stored at a position of length a from the beginning of each table element, and the last table element 214 in the chain is stored. Has a NULL pointer. In this embodiment, the pattern identifier A is given to the table having such a configuration.

(2) As shown in FIG. 3, from the main table 311, a plurality of table elements 312 to 316 holding data to be actually referred to and updated are pointed. Here, the main table 311 stores pointers pointing to the table elements in the area of each length b from the head thereof, and all the table elements 312 to 316 have the same size of YYY. In this embodiment, the pattern identifier B is given to the table having such a configuration.

The lock tables included in the lock table group 112 have a one-to-one correspondence with the respective tables 111, and have a structure as shown in the lock tables 215 and 317 of FIGS. 2 and 3. Each lock table has an area E1 for setting a table name, a lock field E2,
An area E5 for setting a lock process name, an area E5 for setting a pattern identifier, and an area E5 for setting a pointer to a corresponding table. here,
The setting of the table name in the area E1, the setting of the pattern identifier in the area E4, and the setting of the pointer in the area E5 are performed in advance by the pattern information setting mechanism 12 started from the process 3. Also, at this time, area E
The lock field of 2 is reset to unlocked, and the lock process name of the area E3 is set to NULL.

The pattern information table 13 is a table for holding pattern information indicating, for each pattern identifier A and B, the structure of the table having such a pattern identifier, and an example thereof is shown. As shown in FIG. In FIG. 4, the size XXX and the pointer position a are pattern information related to the pattern identifier A, and
As the information that characterizes the table having the configuration as shown in FIG. 3, the size of the table elements 212 to 214 is XXX, and the pointer is set to the position of the length a from the head of each table element. Further, in FIG. 4, the size YYY and the pointer position Δb are pattern information regarding the pattern identifier B, and the size of the table elements 312 to 316 is YYY as the information that characterizes the table having the configuration shown in FIG. , Indicates that the pointer is set in the area of each Δb of the main table 311. Such a pattern information table 13 is created in advance by the pattern information setting mechanism 12 started from the process 3.

The table save area 2 is a place for saving the table elements of the locked table 111. FIG. 5 shows how the table elements of the table 111 are saved in the table save area 2. In FIG. 5, each save entry 501 is a save area for one table element, and is composed of a header section 502 and a table content save area 503. In the header section 502, the saved table elements of the table group 1 are saved. Address 504, the size 505 of the table element thereof, and the table name 506 of the table 111 including the table element, and the table content save area 503 stores the content itself of the saved table element.

Process 4 is a process of referring to and updating the table 111 in the table group 1. Table group 1
When referring to or updating the table 111 in the table group 1, first, a lock request designating the table name is issued to the lock mechanism 6, and after receiving the notification of the lock success, the table 111 in the table group 1 is actually referred to or updated. Then, when the necessary processing is normally completed without causing an abnormality such as a program interruption, an unlock request specifying the table name and restoration unnecessary is issued to the unlock mechanism 7. Although only one such process 4 is shown in FIG. 1, a plurality of processes 4 normally exist and mutually refer to and update individual tables 111 in the table group 1 mutually.

Process 3 is the pattern information setting mechanism 12
Is a process of creating the pattern information table 13 and setting the above-mentioned pattern identifier and the like in the lock table group 112 in the table group 1 by activating.

The lock mechanism 6 responds to the lock request from the process 4 by applying the corresponding table 11 in the table group 1.
The lock table is returned to the table save and restore mechanism 10 by notifying the table save address and size of the table element of the lock target table obtained by activating the save table element return mechanism 8 A process of saving the contents of 111 immediately before the lock to the table save area 2 is performed, and an example of the process is shown in FIG.

The save table element returning mechanism 8 refers to the pattern information corresponding to the pattern identifier of the lock target table from the pattern information table 13 to recognize the configuration of the table, and saves the table group 1 according to the recognition result. This is a mechanism for sequentially notifying the lock mechanism 6 of the address and size of the table element to be processed, and an example of the processing is shown in FIG.

The unlock mechanism 7 is a mechanism for processing an unlock request from the process 4 or the exception handler 5. There are two types of unlock requests, an unlock request that specifies that the table needs to be restored and an unlock request that specifies that the table does not need to be restored. For the former request, the unlock mechanism 7 determines whether or not the table can be restored. On condition that the display 14 indicates that restoration is possible, the address of the save entry obtained by activating the restore table element return mechanism 9 is set to the table save / restore mechanism 10.
Is notified to restore the table element saved in the table save area 2 to the table group 1, and then the unlock processing and the clear processing of the table save area 2 are executed. In addition, in the latter case, in response to an unlock request that specifies that table restoration is unnecessary, unlock processing and table save area 2 clear processing are executed. FIG. 8 shows a processing example of the unlock mechanism 7.

The restoration table element return mechanism 9 retrieves the address of the save entry in which the table element of the table to be restored is saved by searching the table save area 2 and sequentially notifies the unlock mechanism 7. FIG. 9 shows an example of the processing.

The table save / restore mechanism 10 saves the table elements in the table group 1 to the table save area 2 and the table group 1 of the table elements saved in the table save area 2.
Is a mechanism for executing restoration and restoration. FIG. 10 shows a processing example of the table save / restore mechanism 10.

When a program interrupt is generated, the exception handler 5 analyzes the generated program interrupt and, if the system can be continued, sends an unlock request designating table restoration to the unlock mechanism 7, and ,
If the system cannot be continued, the restoration permission / prohibition registration mechanism 11 is activated to set the restoration permission / prohibition display 14 to be non-restorable, and then an unlock request designating restoration of the table is sent to the unlock mechanism 7. An example of such processing of the exception handler 5 is shown in FIG. 11 together with a processing example of the process 4 which caused the program interrupt.

The restoration propriety registration mechanism 11 uses the exception handler 5
This is a mechanism for changing the restoration propriety display 14 to non-restorable when there is a request from the user to set the restoration propriety display 14 to non-restorable. It should be noted that the restoration permission / prohibition display 14 is restoration possible in a normal state.

The operation of this embodiment having the above-mentioned structure will be described below.

Now, referring to FIG. 1, the processor CPU1,
When a lock request designating a certain table name is issued to the lock mechanism 6 from the process 4 executed by one of the CPUs 2, the lock mechanism 6 starts the processing shown in FIG.

First, the lock mechanism 6 uses the lock table group 112 of the table group 1 such as the lock table 215 of FIG. 2 and the lock table 317 of FIG.
To check whether or not the lock field in the lock table is in the unlocked state (S1). If it is not in the unlocked state but in the locked state, another process refers to and updates the table, and therefore the process 4 is notified of the lock failure (S2).

On the other hand, in the unlocked state, the lock mechanism 6 performs the lock process S3. In this lock processing S3, the lock field in the lock table is rewritten from the unlocked state to the locked state, and the process name of the requesting process 4 is set in the lock table as the lock process name.

Next, the lock mechanism 6 notifies the lock-requested table name and activates the save table element return mechanism 8. As a result, the evacuation table element returning mechanism 8 is configured as shown in FIG.
The process shown in is started.

First, the save table element return mechanism 8 retrieves the lock table having the notified table name from the lock table group 112 of the table group 1 and acquires the pattern identifier set therein (S11). That is, in the case of the lock table 215 of FIG.
, And in the case of the lock table 317 of FIG. 3, the pattern identifier B is acquired.

Next, the pattern information corresponding to the acquired pattern identifier is acquired from the pattern information table 13 (S12). That is, in the case of the pattern identifier A, the pattern information including the size XXX and the pointer position a in FIG. 4 is acquired, and in the case of the pattern identification B, the pattern information including the size YYY and the pointer position Δb in FIG. 4 is acquired. .

Next, the structure of the table is recognized from the acquired pattern information, and the address and size of the first table element are notified to the lock mechanism 6 (S13). That is, in the case of FIG. 2, the pointer set in the queue terminal 211 indicated by the pointer in the area E5 of the lock table 215 and the size XXX are notified to the lock mechanism 6, and FIG.
In this case, the lock mechanism 6 is notified of the head pointer of the main table 311 indicated by the pointer in the area E5 of the lock table 317 and the size YYY. Then, it waits for the next notification instruction from the lock mechanism 6 (S1).
4).

When the lock mechanism 6 receives the notification of the address and size of the first table element from the save table element return mechanism 8 (YES in S5 of FIG. 6), these and the table name for which the lock request has been made are designated. Then, the table save / restore mechanism 10 is activated in the table save mode (S
6). As a result, the table saving / restoring mechanism 10 is configured as shown in FIG.
The process shown in is started.

That is, since it is started by saving the table, the table saving / restoring mechanism 10 branches from the processing S41 of FIG. 10 to the processing S42, and the header portion including the address, size, and table name of the table element designated by the locking mechanism 6 is included. And then in step S43, the contents of the table element with the specified address and size are set in the table group 1
1 in step S44, including the header part created above and the contents of the acquired table element.
Write one save entry to table save area 2. Therefore, when no table element is saved in the table save area 2 and, for example, the address, size XXX, and table name of the table element 212 in FIG. 212
Size of the original address 504 in FIG.
A table content save area 50 storing the contents of the header part 502 and the table element 212 in which XX is the size 505 and the designated table name is set in the table name 506.
One save entry 501 consisting of 3 and 3 will be written to the table save area 2. If another save entry 501 already exists, as shown in FIG.
Immediately after that, a new save entry is written.

When the evacuation of the first table element is completed, the lock mechanism 6 instructs the evacuation table element return mechanism 8 to issue the next notification (S7 in FIG. 6), and the evacuation table element return mechanism 8 notifies this notification. When receiving (S14 in FIG. 7),
If the next table element is searched and found (YE in S15)
S), and notifies the address and size of the table element to the lock mechanism 8 (S16). If the next table element does not exist (NO in S15), the lock mechanism 6 is notified of the end of the table element (S17). Lock mechanism 6
Repeats the process of activating the table save / restore mechanism 10 each time the save table element return mechanism 8 notifies the table element address and size, so that all table elements of the currently locked table are stored in the table save area 2 When the end of the table element is notified from the save table element return mechanism 8 (NO in S5), the lock completion is notified to the requesting process 4 (S8),
Finish the process. Therefore, in the case of the table as shown in FIG. 2, the lock completion is notified to the process 4 after the table elements 212, 213 and 214 are saved in the table save area 2. In the case of the table shown in FIG. 3,
After the table elements 312 to 316 are saved in the table save area 2, the lock completion is notified to the process 4.

Upon receipt of the lock completion notice, the process 4 refers to and updates the locked table. Then, when the process that requires the lock ends normally, the unlock mechanism 7 issues an unlock request specifying the table name to be unlocked and the restoration unnecessary. Further, when the process 4 causes a program interrupt in the middle of the process requiring the lock, the control is transferred to the exception handler 5. Hereinafter, the operation will be described separately for the former case and the latter case.

When the unlock request is issued from the process 4, the unlock mechanism 7 starts the process shown in FIG. 8 when the unlock request is issued from the process 4. First,
It is checked whether or not this unlock request includes a restore designation (S21). In this case, it is not included, so
The save entry having the unlock-requested table name is deleted from the table save area 2 (S28), and then the unlock process is performed (S29). In this unlock process,
Rewrite the lock field in the lock table from the locked state to the unlocked state, and set the lock process name to NULL.
Set to.

○ When control is passed to the exception handler 5
1, the process 4 requests the lock (S6
1) When a program interrupt occurs during execution of processing S62 such as table updating, control transfers to the exception handler 5, and the exception handler 5 analyzes the cause and location of the generated program interrupt (S51). Then, based on the analysis result, it is determined whether or not the system needs to be stopped, and when it is determined that the system needs to be stopped (YES in S52), the correction allowance registration mechanism 11 is activated to display the restoration allowance display. After changing 14 from the restoration-enabled state to the restoration-disabled state (S53), the lock table in the lock table group 112 of the table group 1 is referred to and the table locked by the process 4 that caused the program interruption The name is acquired, and an unlock request specifying the table name and the restoration requirement is issued to the unlock mechanism 7 (S54). When it is determined that it is not necessary to stop the system (NO in S52), the process S53 is skipped, and the restoration permission / inhibition display 14 is left as restorable.
An unlock request specifying the table name locked by the process 4 and the restoration requirement is issued to the unlock mechanism 7 (S5).
4).

In response to this unlock request, the unlock mechanism 7 starts the processing shown in FIG. 8. Since restoration is required, the process proceeds from step S21 to step S22, and the restoration permission / inhibition display 14 indicates that restoration is possible. It is checked (S2
2). Then, when the restoration availability display 14 indicates that restoration is not possible (NO in S22), the system is immediately stopped. On the other hand, when it indicates that restoration is possible (YES in S22,
S), a process for restoring the table element saved in the table save area 2 to the table group 1 is executed (S2
3 to S27), and thereafter, the clearing process S28 and the unlocking process S29 of the table save area 2 are performed.

That is, the unlocking mechanism 7 first notifies the table name specified by the exception handler 5 and activates the restoration table element returning mechanism 9 (S23). As a result, the restoration table element return mechanism 9 starts the processing shown in FIG.
The save entry including the specified table name in the header is searched in order from the beginning of the table save area 2 (S31), and the address of the found save entry is notified to the unlock mechanism 7 (S32). The unlock mechanism 7 specifies the notified address and activates the table save / restore mechanism 9 in the table restore mode (S26). As a result, the table save / restore mechanism 9 starts the process shown in FIG.

That is, since the activation is performed by restoring the table, the table saving / restoring mechanism 10 branches from the processing S41 of FIG. 10 to the processing S45, reads the saving entry of the address designated by the lock mechanism 6 from the table saving area 2, and then In process S46, the contents of the table content save area 503 in the read save entry are read from the original address 504 in the header section 502 in the save entry to the size 5
One table element is restored by copying to the area in the table group 1 indicated by 05.

When the restoration of the first table element is completed, the unlocking mechanism 7 instructs the restoration table element returning mechanism 9 to issue the next notification (S27 in FIG. 8), and the restoration table element returning mechanism 9 Upon receiving the notification (S3 in FIG. 9)
3) When the next save entry is searched and found (S35)
If YES, the address of the save entry is notified to the unlock mechanism 9 (S36). If the next save entry does not exist (NO in S35), the unlock mechanism 7 is notified of the end of the save entry (S37). The unlocking mechanism 7 repeats the process of starting the table saving / restoring mechanism 10 every time the address of the save entry is notified from the restore table element returning mechanism 9, so that the table locked by the process 4 causing the program interrupt is locked. All table elements in the table group 1 are restored, and at the time when the restoration table element return mechanism 9 notifies the end of the save entry (YES in S24), the save entry used for this restore is saved from the table save area 2. The table save area to be deleted is cleared (S28), and then unlocked (S29). In this unlock processing, the lock field in the lock table is rewritten from the locked state to the unlocked state, and the lock process name is changed to NU.
Set to LL.

As described above, in this embodiment, the process 4
Even if a program interrupt occurs while the table 111 in the table group 1 is locked by the above, if the program interrupt does not cause the system to stop, the state of the locked table 111 is changed to the state immediately before the lock. You can restore the system to continue operation of the system.

[0049]

As described above, according to the present invention, the shared table immediately before locking is saved (copied) in the table save area, and when a program interrupt occurs, it is saved in the table save area. The shared table in the original location on the main memory is restored to the state just before the lock by the shared table, so when the program interrupt of the type that can continue the system occurs, the system is uniformly stopped as before. There is no need to do so, and it becomes possible to restore the state of the shared table locked by the process interrupted by the program interrupt to a consistent state and continue the operation of the system.

[Brief description of drawings]

FIG. 1 is a configuration diagram of essential parts of a multiprocessor system to which an embodiment of the present invention is applied.

FIG. 2 is a diagram showing a configuration example of a table shared between processes and an example of contents of a lock table corresponding to the configuration example.

FIG. 3 is a diagram showing another configuration example of a table shared between processes and an example of contents of a lock table corresponding thereto.

FIG. 4 is a diagram showing an example of contents of a pattern information table.

FIG. 5 is a diagram showing an example of contents of a table save area.

FIG. 6 is a flowchart showing a processing example of a lock mechanism.

FIG. 7 is a flowchart showing a processing example of a save table element return mechanism.

FIG. 8 is a flowchart showing a processing example of an unlock mechanism.

FIG. 9 is a flowchart showing a processing example of a restoration table element return mechanism.

FIG. 10 is a flowchart showing a processing example of a table save / restore mechanism.

FIG. 11 is a flowchart illustrating an example of processing of an exception handler.

[Explanation of symbols]

 1 ... Table group 111 ... Table shared by a plurality of processes 112 ... Lock table group 2 ... Table save area 3,4 ... Process 5 ... Exception handler 6 ... Lock mechanism 7 ... Unlock mechanism 8 ... Save table element return mechanism 9 ... restoration table element return mechanism 10 ... table evacuation restoration mechanism 11 ... restorability propriety registration mechanism 12 ... pattern information setting mechanism 13 ... pattern information table 14 ... restorability propriety display MEM ... main memory CPU1, CPU2 ... processor

Claims (4)

[Claims] 1. A shared table lock control method in a multiprocessor system comprising a plurality of processors, wherein the shared table on the main memory is exclusively used by the processors, and the shared table on the main memory is transferred to a table save area. A table save / restore mechanism that performs the process of saving and the process of restoring the shared table saved in the table save area to the original location in main memory, and locks the shared table in main memory in response to a lock request. Calling the table save / restore mechanism to call the table save / restore mechanism in response to a lock mechanism that saves the locked shared table to the table save area and an unlock request that specifies the need to restore the table, After restoring the shared table saved in the table save area to the original location on the main memory, An unlock mechanism for releasing the lock of the shared table and releasing the lock of the shared table and deleting the released shared table from the table save area in response to an unlock request designating that the table need not be restored. And a lock control method for a shared table, comprising: an exception handler that sends an unlock request designating table restoration to the unlock mechanism when a program interrupt occurs. 2. A save table element return mechanism for recognizing the configuration of each shared table and notifying the address and size of the table element to be saved to the lock mechanism, wherein the lock mechanism is the save table element return mechanism. Notifies the table saving / restoring mechanism of the address, size, and table name to be saved, and the table saving / restoring mechanism notifies the header section including the notified address, size, and table name. 2. The lock control method for a shared table according to claim 1, wherein a save entry including a content of a table element specified by an address and a size is stored in the table save area. 3. A restore table element return mechanism for notifying an address of a save entry saved in a table save area, wherein the unlock mechanism restores the address notified from the restore table element return mechanism to the table save and restore. 3. A mechanism is notified, and the table saving / restoring mechanism restores the table element at the original location on the main memory based on the saving entry of the notified address.
Shared table lock control method described. 4. A restoration propriety registration mechanism for setting a restoration propriety display referred to by the unlock mechanism, wherein the exception handler analyzes the generated program interrupt and determines that the system operation cannot be continued. After the restoration permission / inhibition registration mechanism sets the restoration permission / inhibition to non-restoration, an unlock request is issued to the unlock mechanism, and the unlock mechanism stops the system immediately when the restoration permission / inhibition display indicates restoration is impossible. The shared table lock control method according to claim 3, wherein