KR100315500B1 - Memory allocating method - Google Patents

Abstract

PURPOSE: A memory allocating method is provided to allocate a memory at high speed using an empty pointer which designates an empty area, thereby searching an empty area continuously in a memory structure. CONSTITUTION: In a memory structure having a flag for displaying an allocation/empty status of a memory area and an allocation available size, a TOP memory area(10) for displaying a header is set. In the case that a flag status is an empty status, empty memory areas(20,40) having an empty pointer for designating a memory area of the next empty status are set. A memory is allocated in an empty memory area, a flag is changed, and an allocated memory area(30) is set.

Description

Memory allocation method

The present invention relates to a memory allocation method, and more particularly, to setting an unallocated pointer indicating an empty memory area in a memory structure, thereby freeing the unallocated pointers from the unallocated memory areas when a memory allocation request is made. The present invention relates to an optimal memory allocation method for searching and allocating at high speed.

When running a program on a computer, memory for variables and workspaces is allocated from the operating system (O / S) or system manager (System Management). In order to use memory on the system efficiently, the memory must be allocated dynamically, and after the allocated memory is used, it must be returned to the system so that other programs can use the memory.

On the other hand, in a low-cost system with low memory, memory acquisition and return occur frequently, and the time spent allocating and returning the memory greatly affects system performance depending on the operation speed of the entire software. Products that currently use Reduced Instruction Set Computer (RISK) or Complex Instruction Set Computer (CISK) CPUs, such as NC, Internet Set Top Box, DBS Set Top Box, Digital TV (DTV) As such, the software used for them must dynamically use a finite resource of memory, and the performance related to the allocation and return of the memory has a great effect on the performance of the system.

1 is a memory chain for explaining a memory allocation according to the prior art, the memory chain is a pointer indicating the next memory area, starting with the TOP area (1) as the basis of the memory stack (Next) ) And a pointer to a previous memory area, and a flag indicating a state of each memory area. Herein, the pointer is one of data structures used in a high-level programming language such as C language or Pascal language, and refers to information indicating a location or address of a storage location in which one piece of data is stored. You can go directly to the location where the data is stored. The flag is divided into Alloc. And Empty. When the allocated memory Alloc returns the memory to the system, the flag becomes an empty memory. The memory size is a size of the memory allocated by the memory area and is represented by the number of bytes required by a utility or the like.

Table 1 shows the memory as a structure type on a program. In the programmatic structure, Struct, int, and char are function names, and Memory Block * prev, Memory Block * next, Size, Flag, TASK_ID, and Data [Size} mean variable names. Remarks, such as 'Prev Pointer', are used to describe the contents of the variable. Here, two memory area pointers (Prev. Pointer and Next Pointer) are used, and Data [Size] represents an actual memory address allocated to a user system.

Structure Memory Block in Program Struct Memory Block * prev Previous Memory Area Pointer (Prev. Pointer) Struct Memory Block * next Next memory area pointer int Size Memory size required by utility char Flag Status of memory area (Alloc, Empty) char TASK_ID Identification number of the job requiring memory char Data [Size} Physical memory address assigned to your system

The above-described structure of Table 1 may be used to configure the memory chain of FIG. 1, wherein the TOP memory region 1 is NULL without a previous region, and the A1 memory region 2 has a previous region prev. And the next area next is A2. In the same manner, the memory area is distributed and allocated, and memory allocation and destruction are shown in FIGS. 2A and 2B.

2A and 2B are the last memory areas before and after memory allocation according to the prior art, in which A2 is allocated when A2 (3) is the last memory area, and the next area next is NULL. If 20 Kbytes is required, A3 (4), which is the next area, is created, 20Kbytes are allocated to A2, and the last memory area is created. That is, FIG. 2A is the last memory chain before memory allocation with a size of 200Kbytes and the flag is in an Unallocated state, and FIG. 2B is an Alloc state with 20Kbytes allocated to the memory chain A2 after memory is allocated. The new A3 becomes the last memory chain, where A3 is flagged and the memory size is 180 Kbytes.

FIG. 3 is a diagram illustrating allocating memory to an unallocated area according to the prior art. When the required memory size is 20 Kbytes, the A3 memory area 4 in the unallocated state is 20 Kbytes. Allocates to a memory area. Here, the A3 memory area represents a free, ie, unallocated area existing between the A2 memory area 3 and the A4 memory area 5.

FIG. 4 is a structural diagram of a memory map according to the prior art, and the above-described memory area is allocated by dividing a heap area on the memory map into memory allocation areas. For example, the start address of the A1 memory area is the sum of the address 20000h and the header size (HS), where the previous area pointer points to the TOP, and the next area pointer points to the A2 area. Will be displayed. Here, the heap area refers to a storage area reserved for the operating system in order to allocate a storage space required during the execution of a program. For example, the heap area in the operating system requires a storage space to store data during execution. The storage location at will be allocated to the program. The TOP area is a header area that contains contents of information stored such as a file or a record of a storage device.

In the memory map structure of FIG. 4 described above, a memory allocation method uses a memory allocation method of an optimal size and a last memory allocation method. First, there is always a free area allocated and returned in the middle of the memory area, such as A3 of FIG. 3. To allocate the memory, an empty area, that is, an unallocated area (from the TOP area, which is the head of the memory area, through the Next pointer) Empty Block), and if the size of the empty area is suitable for the required size, memory is allocated, but if there is no size allocated, the required size is compared in the final area. Likewise, the optimal size memory allocation method is to allocate the required size by creating a new last region.

In order to increase the memory allocation speed, the memory is always allocated from the last memory, and if the last memory is insufficient, the method of using the above-described optimal size memory allocation method is the last memory allocation method.

However, in the conventional memory allocation method, the optimal size memory allocation method has a problem that the flag state is empty and it takes a long time to search and compare whether the memory size meets the required size.

In addition, the conventional last memory allocation method may be faster at first but later becomes slower, and is not an optimal memory allocation. Therefore, when a large amount of memory is required, unallocated areas are scattered in various places. Although the entire unallocated area is memory assignable, there is a problem that there is no memory of the assignable size.

Therefore, the present invention was created to solve the above problems, and in the memory allocation method, using an unallocated pointer that refers to an unallocated area while using an optimal size memory allocation method, The goal is to always allocate memory at high speed by searching for unallocated regions in the memory structure.

1 illustrates an embodiment of a memory chain according to the prior art,

2A and 2B show the last memory chain before and after memory allocation according to the prior art;

3 illustrates a memory chain for allocating memory to an unallocated area according to the related art.

4 illustrates a structure of a memory map according to the related art.

5 illustrates a memory chain using an Unassigned Pointer according to the present invention,

Figure 6 shows a memory chain omitting the next memory area pointer in accordance with the present invention,

7 shows a structure of a memory map according to the present invention,

8 is a flowchart illustrating a memory allocation method according to an embodiment of the present invention.

※ Explanation of code for main part of drawing

10: Top memory area 20: First memory area

30: second memory area 40: fourth memory area

According to an aspect of the present invention, there is provided a memory allocation method, comprising: a first step of setting an unassigned pointer indicating a memory area in which a next memory is empty in an area in which the memory is empty; And a second step of searching for and allocating a memory area based on the unassigned pointer when allocating a memory.

In the memory allocation method according to the present invention made as described above, in a memory structure composed of a flag, a request size, and the like, by adding an unassigned pointer indicating a place where the memory is empty, that is, the memory is unallocated to the area where the memory is empty. By setting the pointers, only memory areas having the unallocated pointers can be searched and allocated at high speed during memory allocation.

Hereinafter, a preferred embodiment of a memory allocation method according to the present invention will be described in detail with reference to the accompanying drawings.

FIG. 5 is a memory chain diagram illustrating a region in which memory is allocated using an unallocated pointer according to the present invention. FIG. 5 is a flag indicating an alloc / empty state of a memory region. 1. A memory structure having a size and an assignable size, comprising: a first step of setting a TOP memory area 10 representing a header; A second step of setting an unallocated memory area (20, 40) having an unallocated pointer indicating a memory area of a next unallocated state when the flag state is an unallocated state in which the memory is empty; And a third step of allocating memory to the empty memory area, changing the flag to set the allocated memory area 30, and having a flag, a required size, etc. By setting an unassigned pointer to, if the flag is in an unallocated state when allocating memory, a direct search is made from the A1 region 20 to the A3 region 40 by an unallocated pointer that refers to the memory region of the next unallocated state. That is, it is possible to search and allocate only the memory area having the unassigned pointer at high speed without passing through the A2 area 30 which is the memory area 30 to which the memory is allocated.

8 is a flowchart illustrating an operation of a memory allocation method according to the present invention, with reference to the memory chain structure of FIG. 5.

In Fig. 5, by assigning a non-allocating pointer to the next unallocated area in the memory area, when searching for the next unallocated area, A1 (20) → A3 ( It is possible to find the unallocated area immediately by the path of 40), and not to search the first unallocated area in the order of TOP (10) → A1 (20), but to set the unallocated pointer to find the first unallocated area. You can specify it immediately. However, as shown in FIG. 5, when the unallocated pointer is added to search the next unallocated region at high speed, the management memory becomes large.

FIG. 6 is a configuration diagram of a memory chain for streamlining memory management. In FIG. 4 and Table 1, it can be seen that the next memory area pointer Next Point becomes equal to an address after char Data [Size]. That is, Next Pointer = the sum of the Data Address and the Size, so the Next Pointer does not need to have it, and can be specified by operation.

Structure Memory Block in Program Struct Memory Block * prev Previous Memory Area Pointer (Prev Pointer) Struct Memory Block * Empty Next Empty Area Pointer int Size Memory size required by utility char Flag State of memory area; Alloc, Empty char TASK_ID Identification number of the job requiring memory char Data [Size} Physical memory address assigned to your system / * Struct Memory Block * next = address of Data + Size * /

In other words, as shown in Table 2, the Next Pointer does not actually exist by being allocated an address on the memory area, but is obtained by an operation.

FIG. 7 is a structural diagram of a memory map according to an embodiment of the present invention, which illustrates using an unassigned pointer separately in place of a next memory area pointer (Next Pointer) as compared to FIG. 4 described above. The pointer is obtained by operation.

Meanwhile, a method of allocating a memory using the aforementioned memory chain of FIG. 6 is as follows. In FIG. 8, when the utility requests a memory area to store data (S11), the memory area size is checked based on a preset unassigned pointer and the memory area is checked to meet the required memory size (S12). If the memory size does not match, the next unallocated area is checked. That is, if the size of the first unallocated area A1 is 20 Kbytes and the required memory size is 30 Kbytes, the first unallocated area is skipped, and the next unallocated area A3 area 40. Will be examined.

Then, the next unallocated area is checked, if it meets the size required by the utility, i.e. not smaller than the size requested by the utility (S13), and the size of the unallocated memory area is (required size + size of header). If not exceeded (S14), the corresponding memory area is allocated (S15), and the flag is changed to Alloc, (S16). Herein, it is determined whether or not the big memory area is based on the (required size + size of header).

If the size of the searched unallocated memory area exceeds the (required size) + header size, i.e., is too large than the required size (S14), once the area is a big area pointer (BBP) Will be assigned to). If there is an existing big memory area previously allocated to the big area pointer BBP (S17), the area is ignored and the first big area pointer BBP is allocated (S18). If there is no pre-allocated existing big memory area (S17), the unallocated memory area is allocated to the big area pointer BBP (S19). In this case, setting and allocating the big memory area is to use the small portion of the unallocated area, and selecting the minimum value of the memory area of the big area pointer BBP may increase the memory allocation efficiency.

If there is no area matching the required memory size by searching the unallocated area to the end (S20), that is, there is no memory area larger than the required size or smaller than (required size + header size). In this case, the memory area indicated by the above-mentioned big area pointer is allocated (S21), and the flag is changed to Alloc. (S22).

Accordingly, the present invention utilizes an optimal size allocation method when allocating memory, which is a main memory such as RAM, which is dynamically used in the system, while the memory area is empty, that is, the flag state is not assigned. Only the in-memory area is searched and allocated at high speed, and if the memory area to be allocated is a big memory area that is too large, the memory is allocated from a relatively small portion among the big memory areas.

The memory allocation method according to the present invention as described above is a useful invention that can allocate the optimal memory at the time of memory allocation using a non-allocating pointer at a high speed, and that a low-cost CPU can be used according to the speed of the software processing speed.

Claims (5)

A first step of setting, by an operation, an unassigned pointer indicative of the next empty memory area in an area where the memory is empty; And And searching for an empty memory area with reference to the unallocated pointer, and then allocating the searched empty memory area to a requested data storage space starting from a small portion of the searched empty memory area. According to claim 1, The non-allocating pointer of the first step is a memory allocation method characterized in that the summed value obtained by the arithmetic operation of adding the allocated data address and the required memory size is designated as a pointer value indicating a next memory area. . According to claim 1, The non-allocating pointer is a memory allocation method, characterized in that the first non-assigned region designation directly specified without passing through the header area. In the memory allocation method, Searching for an empty memory area with reference to an unassigned pointer indicating an empty memory area; A second step of comparing the size of the searched unallocated memory area with the size of data; And And a third step of allocating the searched unallocated memory according to the comparison result, or designating the memory as a big memory when the memory is larger than or equal to a predetermined size in comparison with the size of data. The method of claim 4, wherein The third step, A first substep of checking whether a predetermined big memory area exists when the size of the searched unallocated memory area is larger than a size of the data; A second substep of comparing a size of the searched unallocated memory area if a predetermined big memory area exists; According to the comparison result of the second sub-step, if the searched unallocated memory area is larger, the unallocated memory is set to the big memory, and if the unallocated memory area is smaller, to the last unallocated area, 1 A third substep of repeating the step; And And a fourth substep of allocating a memory to the big memory area if the big memory is not determined in the third substep.