JPS63146134A - Shared data control system - Google Patents

Abstract

PURPOSE:To easily control shared data without the occurrence of dead lock by detecting update of shared data and performing retry after restoration of shared data if necessary. CONSTITUTION:If shared data 4X and 4Y which are used as a basis of the processing in programs 1A-1C are updated by another program and results of processings are ineffective, related shared data 4X and 4Y are restored by a shared data control mechanism 6 if necessary and the control is transferred to a restart address preliminarily set by programs to perform retry. Thus, dead lock does not occur fundamentally. Programs easily escape the dead lock state even if dead lock is detected.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to control of shared data when data processing is performed by an electronic computer.

[Conventional technology]

Conventionally, as a method for controlling shared data, exclusive control by locking and unlocking the shared data has generally been performed. In other words, from the time the shared data is referenced until the value is updated based on that value, the value based on the shared data is prevented from being updated by other programs.

[Problem that the invention seeks to solve]

The conventional shared data control method described above basically has the disadvantage that deadlocks occur, and even if deadlocks can be detected, it is extremely difficult to escape from a deadlock state once it has occurred. there were.

[Means for solving problems]

The shared data control method of the present invention updates the shared data after referencing all the necessary shared data, and the shared data that is the basis of these processes is updated by other programs until the processes are completed. A program that can be retried from a restart address set in advance on the program side in the event of a failure, a program history that stores a history of references to or updates to the shared data of the program, and a history of references to or updates to the shared data. It is detected from the history that the shared data that is the basis of the process has been updated, and when the shared data is updated, the shared data that stores the shared data Bv: It has a shared data control mechanism that allows the program to be restarted and retried after recovery.

〔Example〕

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

FIG. 1 is a block diagram showing an embodiment of the present invention.

IA, IB, IC are programs that refer to or update shared data 4X, 4Y, 2A, 28.20 are each program I
A, IB, a register storing a restart intention when the IC retries; 3A, 3B, 3C each program IA;
IB and IC are shared data 4X.

This is a program history list that refers to or updates 4Y and arranges the 7'c history in chronological order, and the program history 3A-1 respectively.
...3A-n, 3B-4・---38-n
, 3C-1-...3cmn.

4X, 4Y are shared data referenced or updated by programs IA, IB, ic, 5X, 5Y are respective shared data 4X, 4
This is a shared data history list in which the history referenced or updated by Y is arranged in chronological order, and each shared data history 5X-1...
...5X-m, 5Y-1...5Y-m.

6. If the shared data on which programs IA, IB, and IC are based is updated by a program other than their own and the processing results become invalid, if necessary, after restoring the related shared data, This is a shared data control mechanism that causes the program to transfer control to a preset restart address and retry.

FIG. 2 shows the program IA used in this embodiment.

It is a flowchart of IB and IC. Program IA.

IB and IC handle shared data 4X and 4Y as common inventory amounts for their respective products. Program I
A consumes 2 pieces aI+" of each of 4X and 4Y only when there are 32 or more pieces of each a1+" at the same time.Frogram IB is 4
Only when there are b or more X, b are consumed. Program I
C consumes 2 cl and 2 c only when there are 01+02 or more of each of 4X and 4Y at the same time.

First, the operation when there is no interference between programs will be explained using program IB as an example.

When program IB starts operating, program I
B notifies the shared data control mechanism 6 of the beginning of the program as the restart address, and the shared data control mechanism 6 stores the value in the register 2B, thereby setting the restart address of the program IB. Next, when the program IB refers to the shared data 4X, the shared data control mechanism 6 changes the program history list 3B-1 at the beginning of the program history list 3B corresponding to the program IB to the shared data history list (5X) corresponding to the shared data 4X. The address of the shared data history 5X-1 is stored, and the address of the program history 3B-1 and a flag (hereinafter referred to as flag S) indicating that the access is by reference are stored in the shared data history 5X-1. Next, program IB refers to the shared data 4X
The content X and the value held within the program are compared, and if
Notify that the contents of 4X will be updated. The shared data control mechanism 6 first searches the program history list 3B-1 and the shared data history 5X-1, which were generated as a result of the reference previously made by the program IB, from the program history list) 3B and the shared data history list 5X, and searches the shared data history list 5X-1 and the shared data history list 5X. The flag S stored in the program file M5X-1 is changed to a flag indicating that the access is due to an update (hereinafter referred to as the flag), and the previously stored program file file is changed to this flag.
In addition to the address of B-1, the current value xf< of the shared data 4X is stored in the shared data 1iJfJ5X-1, and then the content of the shared data 4X is changed to x-b. Next, program IB declares that it has finished using the shared data. Along with this, the shared data control mechanism 6 has a shared data history 5X-
After confirming that the update indicated by 1 will not be invalidated by the operation of other programs using the shared data history list 5 Then, the program history 3A-1 is deleted from the program history list 3A, and the processing of the program IB and the contents x-b of the shared data 4X updated thereby are confirmed. The shared data controller 4i46 then invalidates the restart address previously set in the register 2B by the program IB. With the above steps, the processing of program IB is completed. Note that when declaring the end of use of shared data, a judgment is made as follows whether or not there is a possibility that the shared data will become invalid due to the operation of another program. First, find the address of the shared data history stored in the program history from the first program history to the last program history in the program history list corresponding to that program, and use that shared data history in the shared data history list that includes that shared data history. Shared data Y in which flags for access due to updates are stored before the data archive history
Check to see if there is an I history. If the corresponding shared data MW does not exist, it is determined that it will not be invalidated by the operation of another program. If the corresponding shared data history exists, it is determined that there is a possibility that it may become invalid. If it is determined that there is a possibility that the shared data will be invalidated, the corresponding shared data history is divided by 1 and the end of the declaration of end of use of the shared data is postponed. If the shared data history is deleted by confirming the update process,
The declaration of end of use of shared data ends. However, the shared data history becomes invalid due to the operation of other programs, which results in a sudden change! If this happens, this side will be invalidated in a chain reaction, so the declaration of end of use of the shared data will not be completed and will be retried.

Next, the operation when there is interference between programs will be explained.

First, when program IA starts operating, program I
A notifies the shared data control mechanism 6 of the beginning of the program as the restart address, and sets the restart address by storing this in two registers. Next, Program Sim I
When A refers to shared data 4X, shared data control +'U
46 stores the address of the first shared data record 1sx-1 of the shared data history list 5X corresponding to the shared data 4X in fifi3A-1 of the first program of the blockblum dependency list 3A corresponding to the program IA, and stores the address of the first shared data record 1sx-1 in the shared data record d5X -1 stores the address of program MIfi3A-1 and 7lag S. Next, the block IA uses the content X of the referenced shared data 4X and the internally held value a.
If X is larger or equal, shared data 4Y is referenced. Program history list) 3A's second program file fi3A-2 is the shared data record g5y at the head of the shared data history list corresponding to shared data 4Y.
-1 address, then shared data history 5Y-1
The address of program V history 3A-2 and flag S are stored in . Next, the program IA compares the content y of the referenced shared data 4Y with the internally held value a2,
If y is greater or equal, program history list) 3A
The order of program histories 3A-1 and 3A-2 is swapped. Next, the flag of the shared data JdQ5X-1 is changed from S to k, and in addition to this flag and the previously stored address of the program history 3A-1, the current value -1, and then changes the contents of the shared data 4X to x-al. Assume that control is transferred to program IB at the time when the program IA in FIG. 2 is operated in flowchart 1. Then, program IB performs the same operation as previously explained when there is no program interference. Then, when the operation reaches 70-Gouyat's A in the program IB in Figure 2, '+B
Suppose that IJ control is transferred to program IC.

When the program IC starts processing, it first sets the beginning of the program as the restart location, and then checks the shared data 4Y. As a result, the program Mum') ST3C corresponding to the program IC contains the first program nw3cm.
1 is shared data Jfjl history list 5 in which the address of shared data 'aM5Y-2 is stored and corresponds to shared data 4Y.
A flag S and the address of the program history 3C-1 are stored in Y's shared data vaginal cavity 5Y-2. Next, the program IC compares the content y of the referenced shared data 4Y with the internally held value c1, and if y is larger or equal, it then refers to the shared data 4X. As a result, the second program history of 3C (program history list) 3C-
The address of the shared data history 5X-3 is stored in the shared data history 5X-3.The third shared data history 5X-3 of IJ bust
address is stored. Next, the program IC compares the content X-al-b of the referenced shared data 4X with the internally held value c2, and if x-al-b is larger or equal, the shared data control mechanism 6 List 3
C program history 3C-1 and program history 3C-2
, and then change the flag of shared data fi15Y-2 from 8 to k, and in addition to this flag and the previously stored address of program history 3C-1, write the current value y of shared data 4Y. is stored in the shared data register m5Y-2, and then the content of the shared data 4Y is changed to kYc+.

At this time, access to the shared data history 5Y-1 located before the shared data history 75Y-2 that the program IC has just processed in the shared data employment history list) 5Y becomes invalid, so the shared data f1M history 5Y- 1 and the program history 3A-2 stored in the program are deleted, and the program IA is placed in a retry preparation state. Next, since the update of shared data 4X by program IA has become invalid, 3A-1, which is the remaining history in program history list 3A, and the corresponding shared data history 5X-1 are divided by 61'', and shared data 4X Return the contents to X, delete the shared data B #5 Shared data storage [5X-3
and the corresponding program register 13e-2 are deleted, and the program IC is brought into a retry preparation state. Then, the change of shared data 4Y of the program IC becomes invalid, and the program IC M3C-1 and its corresponding shared data register /ff become invalid.
15Y-2 is deleted and the contents of the shared data 4Y are returned to y. For programs in the retry preparation state whose corresponding program history list is empty, the retry preparation state jμ is released,
Resumes the program from the previously set restart address. Therefore, programs IA, IB, and IC are all restarted from the restart address.

〔Effect of the invention〕

As explained above, the present invention updates the shared data after referencing all the necessary shared data, and makes settings for when the shared data that is the basis of these processes is updated midway through. A program that can be retried from the restart address set, a reference or update history on the program side and the shared data side, and a recovery of the shared data if necessary by detecting that the shared data that was the basis of the process has been updated. The shared data control mechanism, which retries the system after performing the process, has the effect of relatively easily controlling shared data without causing a tedlock.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a flowchart of a program used in this embodiment. IA, IB, IC---Program, 2A, 2B,
20...Restart address storage register, 3A, 3B
, 3C...Program history list, 3A-1-3A-
n, 3B-1...3B-n, 3C-1...
Program history for 3C-n..., 4X, 4Y...
・・Shared data, 5X, 5Y・・・・・Shared data history list 1, 5x-t-・sx-m, 5y-1-・-s
y-m...shared data history, 6...river/shared data control mechanism. Agent Patent Attorney Uchihara 1-day \41. . Figure 2

Claims (1)

[Claims] A restart address that is set in advance on the program side when the shared data is updated after all the necessary shared data has been referenced, and the shared data that is the basis of these processes is updated from another program midway through. a program that can be retried from the start; a program history that stores a history of references to or updates to shared data of the program; a shared data history that stores a history of references to or updates to the shared data; Detects from the history that the shared data has been updated, restores the shared data using the pre-update value stored in the history beforehand when updating the shared data, and then retries the program from the restart address. A shared data control method having a shared data control mechanism.