JPS63271646A - System for managing garbage collection - Google Patents

Abstract

PURPOSE:To perform garbage collection when the emergency level of garbage correction is low and to prevent other programs from being disturbed by the processing by adjusting the frequency of garbage collection according to the emergency level of the program group. CONSTITUTION:An execution time is assigned to respective programs including a garbage collection (gc) call program Tg unless a program which requires high-priority emergency is present among programs to be brought under the execution time assignment control of a scheduler 11. Then the securing of a memory region by a region securing part 14 and the recovery processing of a memory region by a (gc) processing part 15 are performed nearly by turns. In this state, when a program which requires emergency is started, the priority of the (gc) call program is lowered relatively, the execution time assignment of the (gc) program is reduced, thereby adjusting the (gc) processing frequency.

Description

[Detailed Description of the Invention] [Objective of the Invention] (Field of Industrial Application) This invention relates to a garbage collection generation method that manages the generation of garbage collection to recover memory areas that are no longer needed and make them available for reuse. Regarding management methods.

(Prior Art) In program processing of a sieving machine, a memory management method is sometimes used in which memory areas having a required size or data type are allocated one after another in response to requests during program execution. In this kind of memory management method,
In order to make effective use of memory, so-called garbage collection (hereinafter referred to as GC) collects unnecessary memory areas that are no longer referenced by any program and makes them available for reuse. It is generally accompanied by a process called.

The above memory management method will be specifically explained with reference to FIG. In FIG. 2, HBEG indicates the first address of the memory area available for the corresponding memory management, and HEND indicates the final address. HCtJR indicates the final address of the memory area cut out so far. In the initial state, HCUR matches the value of HBEG, and each time a new memory area is required, it advances in the direction of HEND by that amount and cuts out the advanced area as a reserved memory area. . In Figure 2, AI, A
A region indicated by 2...An is shown cut out. At the beginning of the cut out multi-headed 11IA1 to An,
The sizes 81 to 3n of the area are stored. Also, in area A1-80, area Ai that is not in a reusable state
(In Fig. 2, in the area A1゜A4, An >,
A count value (referenced count Ci) indicating from where the area Ai is referenced is stored. On the other hand, reusable areas form a chain starting from the position indicated by FREE, and when a memory area of the size is requested, if an area of the desired size is included in this chain, then , the area is removed from the chain and becomes a used area. Areas that can be reused and areas that cannot be reused are identified by the value (O or 1) of the least significant bit at the beginning of the area.

Now, the QC described above is a process of selectively connecting an area with a reference count of O out of the memory area that has been cut out to a chain of reusable areas (at this time, directly or indirectly from that area The referenced count of the referenced area is decremented by 1). Conventionally, the following two methods have been known regarding the timing at which gc processing is performed. One is a real-time processing method, in which a small-scale QC is performed every time a region with a reference count of 0 occurs. This method has a problem in that the processing efficiency constantly decreases. The other method is a batch processing method. For example, when an area is needed, if there is no one of the desired size in the chain of reusable areas, all areas with a subject count of O at that time will be processed. This is to perform gc. Although this method does not cause a steady decline in processing efficiency, the original processing is interrupted irregularly, so
During that time, there was a problem that the processing efficiency decreased significantly.

(Problems to be Solved by the Invention) As mentioned above, conventional methods have been to generate gC (garbage collection) every time a reusable area is available, or when the area cannot be secured, it can be reused at that point. In any of the methods that generate GC for all areas, QC (garbage collection) is started regardless of the program processing state, which leads to a decrease in processing efficiency and reduces real-time response performance. There was a problem of scarcity.

This invention was made in view of the above circumstances, and its purpose is to enable garbage collection (GC) generation by adjusting the degree of occurrence of garbage collection (GC) according to the status of a group of executed programs, thereby improving the real-time responsiveness of the system. The purpose is to provide a management method.

[Structure of the Invention] (Means for Solving the Problems) The present invention provides a first program having a garbage collection generation request function and a plurality of second programs not having a garbage collection generation request function. A scheduler that allocates time according to the execution priority of each program, and a memory area that is called by the allocated program when the scheduler allocates execution time to one of a plurality of second programs. A securing means and a garbage collection processing means called by the first program when execution time is allocated to the first program by the scheduler are provided, and when the memory area securing means is called,
In addition to performing processing to secure a memory area that can be used by the calling program, the execution priority of the first program is increased according to the memory area securing status, and when the garbage collection processing means is called, unnecessary Garbage collection processing is performed to recover the memory area and make it available for reuse, and if the execution priority of the first program is higher than the initial value, the priority of the same layer is lowered.

(Operation) According to the above configuration, garbage collection occurs only when the scheduler allocates execution time to the first program and the garbage collection processing means is called from the first program, so it can be reused as before. Compared to a method that generates garbage collection every time a new area is generated, or a method that generates garbage collection for all areas that can be reused at that time when the area cannot be secured, 2 The processing efficiency of the program is improved. Moreover, when the second program with a high execution priority is started, the execution priority of the first program becomes relatively low, so the occurrence of garbage collection is reduced and there is no risk of interfering with the execution of the second program. .

Additionally, if you are trying to allocate a memory area and the area cannot be allocated, you can temporarily increase the execution priority of the first program to increase the occurrence of garbage collection, so you can quickly allocate the memory area. can be secured.

(Embodiment) FIG. 1 is a block diagram showing an embodiment of the present invention. In the figure, TO is a program dedicated to QC (garbage collection) generation request (hereinafter referred to as QC calling program), and TI.

T2...Tn are normal programs, such as user programs and system programs that perform processes other than gc. Each of these programs TQ.

TI, T2...Tn are programs that perform mutually independent processing, and each has an execution priority PΩ.

It has Pi, P2...pn. In this example,
The execution priority Pa of the gc calling program To is set to a relatively small value (here, the larger the value, the higher the priority) in the initial state. 11 is each of the above programs T
g, T1 to Tn, their execution priorities PJ, P1 to Tn
A scheduler that allocates execution time according to Pn, 12 is a physical memory area, 13 is a memory area Fi11
2 and provides logical memory areas to normal programs T1 to Tn. Memory management section 1
3 forms part of the operating system (O8).

The memory management unit 13 is called by a program to which execution time is allocated among the normal programs T1 to Tn,
It has an area securing unit 14 that performs processing to secure a memory area usable by the program, and a gc processing unit 15 that is called only by the gc calling program Tg and performs QC (garbage collection) processing. The area securing unit 14 and the QC processing unit 15 have a function of appropriately changing the execution priority Pa of the QC calling program To.

Next, the operation of the configuration shown in FIG. 1 will be explained. Scheduler 1
1, when there are a plurality of programs in the ready state among the programs Tg and 71 to Tn, execution time is allocated to these programs in the ready state according to the execution priority of the program. When the scheduler 11 allocates execution time to one of the regular programs T1 to Tn, the area securing unit 14 of the memory management unit 13 is called by the allocated program and requests to secure a memory area. As a result, the area securing unit 14
performs the well-known process of securing a memory area for use by the calling program. On the other hand, when execution time is allocated to the gc calling program Tg, the gc processing section 15 of the memory management section 13 is called from the program Tg and a gc process is requested.

This allows gc! The 2!1 processing unit 15 performs a well-known QC (garbage collection) process to collect unnecessary memory areas that are no longer referenced by any program and make them available for reuse.

It is now assumed that among the programs subject to execution time allocation control by the scheduler 11, there is no program that requires high priority urgency. In such a case, since execution time is appropriately allocated to each program including the gc calling program Tg, memory area securing processing by the area securing unit 14 and memory area recovery processing by the gc processing unit 15 are performed. (qc processing) is performed almost alternately. In this state, when a program Ti that requires g acute among the normal programs T1 to Tn is started, since the program T1 has a high priority, the gc calling program T! The priority level Pa of + is relatively low. As a result, the execution time allocation for the QC calling program T(+ is reduced, and the occurrence of QC processing by the gc processing unit 15 is reduced. In other words, according to this embodiment, depending on the urgency of the execution program group, In other words, the occurrence of QC can be adjusted according to the work load of the tightening machine system, and gC can be performed when there is less urgency or when the work load is small.

Now, if the occurrence of gc processing by the gc processing unit 15 is reduced, a situation may arise in which a memory area cannot be secured in the memory area securing process by the area securing unit 14. In such a case, the territory@securing section 14 uses -g
c The execution priority Pg of the called program Tg is temporarily changed to a high priority, for example, the highest priority. As a result, the scheduler 11 allocates execution time to the QC calling program Tg with priority over other programs T1 to Tn.

As a result, the gc processing unit 15 is called preferentially and QC
Processing is performed and the area necessary to secure the memory area is recovered. After performing the gc processing, the gc processing unit 15 returns the execution priority PJ of the gc calling program Tg to its original value (initial value). In this embodiment, the highest priority is given to the calling program T (+) only temporarily when there is no memory area that can be secured. There is no possibility that the execution of ~Tn will be hindered.

In the above embodiment, the execution priority Pa of the gc calling program Tg, which has been changed to a high priority, is returned to the original value (initial (direct)) by the gC processing unit 15 when the gc processing is performed. However, the present invention is not limited to this. For example, it may be lowered to a value between the current P(+ value and the original value. In this case, the degree of decrease in Pg is It may be determined according to the area recovery status of the gc processing unit 15, and if the recovery rate is sufficiently high, it can be lowered to the original value.

[Effects of the Invention] As detailed above, according to the present invention, the degree of occurrence of garbage collection <QC) can be adjusted according to the urgency of the execution program group, and garbage collection processing can be performed when the urgency is low. This reduces the possibility that other program processing will be interrupted due to garbage collection processing, and improves the real-time responsiveness of the system.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a diagram illustrating memory management accompanied by garbage collection. 11... Scheduler, 12... Memory area, 13
...Memory management unit, 14...Area securing unit, 15...
・gc processing unit, Tg...gc calling program (first
program), T1 to Tn...normal program (second
program). Applicant's representative Patent attorney Takehiko Suzue Figure 1 Figure 2

Claims (3)

[Claims] (1) a scheduler that allocates execution time to a first program that has a garbage collection generation request function and a plurality of second programs that do not have a garbage collection generation request function according to the execution priority of each program; Called by a program to which execution time has been allocated among the plurality of second programs, the program performs processing to secure a usable memory area, and gives priority to execution of the first program according to the memory area reservation status. and garbage collection processing that is called only from the first program to collect unnecessary memory areas and make it reusable, and the execution priority of the first program is set to the initial value. garbage collection processing means that lowers the same priority when the priority is higher, and the garbage collection processing means lowers the priority when the first program is higher than the first program; A garbage collection generation management method characterized by causing collection to occur. (2) The garbage collection generation management method according to claim 1, wherein the execution priority of the first program is configured so as not to be lowered from an initial value. (3) The degree of decrease in the execution priority of the first program is determined according to the area recovery status in garbage collection processing by the garbage collection processing means, as set forth in claim 2. Garbage collection generation management method.