JP2815588B2 - Forgetting return method of virtual memory - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Overview] Regarding a return-forgotten detection method for detecting a virtual memory that has been forgotten to be returned, without modifying a program that has already been created, the entire memory constituting the virtual memory can be used during operation of the computer system. On the other hand, for the purpose of detecting the address, size, and return source that has been forgotten to return, every time the virtual memory is acquired or returned, trace information consisting of the acquisition / return, area size, address information, etc. is collected. A tracing unit, a management unit that stores, in an acquisition table, trace information relating to a virtual memory that has not been returned at the present time when the trace information collected by the tracing unit exceeds a specified number, and a virtual unit that has not been returned at this current time. The number of unreturns for the memory trace information and the unreturned trace information A detection unit that generates a forgotten return candidate table that stores a value obtained by incrementing the degree by one, and outputs the trace information having the number of unreturns stored in the forgotten return candidate table.

[Industrial applications]

The present invention relates to a forgotten return detection method for detecting a virtual memory forgotten to be returned. The acquired virtual memory must be returned at the responsibility of the acquisition source, but may be forgotten due to a mistake in program design or coding. If the return is forgotten, the memory capacity that can be obtained is reduced by that amount. Therefore, it is desired to verify that the program has not forgotten to return the virtual memory.

[Problems to be solved by conventional technology and invention]

Conventional virtual memory management mainly manages free space for the purpose of speeding up acquisition processing and return processing and saving resources in the management table. For example, an operating system of the multiple virtual storage system automatically returns a virtual memory acquired as a job area by returning a segment table and a page table. However, such an automatic return service is effective at the end of a job or task, but has no effect on a program that operates for a long time. In addition, the program is returned unconditionally regardless of whether the user has forgotten to return it, and the creator of the program cannot notice that he / she has forgotten to return it. Furthermore, it was necessary for the program to instruct the automatic return request to the common area in advance. For this reason, the program cannot detect whether the virtual memory has been forgotten to be returned, and there is a problem that the virtual memory becomes insufficient during operation of the computer system.

The present invention, without modifying the created program,
The purpose of the present invention is to detect the address, size, and source of the forgotten return of the entire area constituting the virtual memory during operation of the computer system.

[Means to solve the problem]

 FIG. 1 shows a principle configuration diagram of the present invention.

In FIG. 1, a tracing unit 1 is called each time an acquisition request or a return request for all areas of the virtual memory is successful, and collects trace information including discrimination of acquisition or return, address information, area size, request source, and the like. Is what you do.

When the number of trace information items collected by the trace unit 1 exceeds the prescribed number N, the management unit 2 generates trace information that has not been returned at the present time and stores it in the acquisition table 2-1.

The detecting unit 3 generates trace information to which the number of unreturned times which is incremented by 1 is added each time the unreturned trace information stored in the acquisition table 2-1 is continuously detected as unreturned. This is stored in the return-forgotten-candidate table 3-1.

[Action]

As shown in FIG. 1, according to the present invention, the trace unit 1 called each time an acquisition request or a return request for the entire area of the virtual memory succeeds is used to distinguish between the acquisition or the return, the address information, the area size, and the request. When the number of collected trace information exceeds a specified number N, the management unit 2 generates trace information that has not been returned at the present time and obtains an acquisition table 2-2. 1 and the detecting unit 3 adds, to the unreturned candidate information table 3-1, the trace information to which the unreturned number is added by 1 each time the unreturned trace information is continuously detected as unreturned. It is stored. Then, trace information to which the number of unreturned times stored in the return forgotten candidate table 3-1 is added is output.

Therefore, without modifying the created program, generate trace information with the number of unreturns added every time a specified number of trace information is collected for all areas constituting the virtual memory during system operation and output it as a list By doing so, it becomes possible to easily verify the area, size, and request source (program) for which the return has been forgotten.

〔Example〕

Next, the configuration and operation of one embodiment of the present invention will be sequentially described in detail with reference to FIGS.

In FIG. 1, a buffer 1-1 temporarily stores trace information collected by the trace unit 1. As shown on the right, this trace information includes the distinction between acquisition and return, acquisition and return address (location, start address), acquisition and return size (area size), acquisition and return request source (program, space ID) ), The time of acquisition and return (TOD of the computer). This buffer 1
The management unit 2 is called when the trace information has been stored for a predetermined number N or more for -1.

When the trace information extracted from the buffer 1-1 is an acquisition request, the management unit 2 creates a new acquisition table. On the other hand, when the trace information is a return request, a deletion process (deletion of the entire trace information, deletion of a part of the trace information and update of the rest, Alternatively, the middle part is deleted and divided into two parts on both sides). And
When the process is completed for the specified number N, the detection unit 3 is called.

The detection unit 3 extracts the unreturned trace information stored in the acquisition table 2-1 and stores it in the return-forgotten-candidate table 3-1 in a non-overlapping manner. Regarding the trace information, the number of unreturns of the corresponding trace information is incremented by one each time and updated.

Next, the configuration and operation of the embodiment arranged in the virtual space will be described with reference to FIG.

In FIG. 2, measured job spaces (1) and (2) are job areas used by the job (program) to be measured. The common area is provided with a trace section 1 according to the present embodiment and a buffer 1a for temporarily storing trace information under an area management supervisor.
In the job area, a virtual memory return forgetting detection space according to the present embodiment is newly provided, and when a predetermined number N or more of trace information is stored in the buffer 1a provided in the common area, the space (the management unit 2) is set. ) Is called. The management unit 2 that has been called stores the specified number N of trace information from the buffer 1a in the buffer 1b provided in its own space (see FIG. 3A), and then creates an unreturned trace information acquisition table. I do. Specifically, in the case of an acquisition request, a process for creating a new acquisition table is performed. On the other hand, in the case of a return request, any process of deleting the corresponding acquisition table, deleting a part and updating the rest, and deleting the center and dividing the acquisition table into two is performed. In the case of this update or division processing, the time in the trace information is updated to the latest return time, the latest return request destination, and the like. In addition, at the time of this return request, a similar deletion process is performed from the return-forgotten-candidate table 3-1.

The detecting unit 3 stores the unreturned trace information stored in the acquisition table 2-1 in the return-forgotten-candidate table 3-1 in a non-overlapping manner, and continuously counts the number of unreturned unreturned trace information. Each time it is incremented by one, it is updated. Therefore, the return-forgotten-candidate table 3-1 contains the latest unreturned trace information (acquired top address, acquired area size, acquisition request source (measured job space (1), etc.) at the present time. , The latest return request source, the acquisition time, the latest return time, etc.) are stored.
By printing the list as a list and viewing the list, it is possible to easily verify, for example, which request source has obtained how much virtual memory amount and for how long.

As shown in FIG. 2 “Example 1”, the operating system temporarily outputs the trace information to the external storage, takes in the trace information into the system when free, and generates the unreturned trace information. You may do so.

Next, the operation of the configuration of FIG. 2 will be specifically described using the specific example of FIG.

FIG. 3 (a) shows an example of trace information collected by the trace unit 1 in time series and stored in the buffer 1a. Here, “G” at the top of the trace information indicates an acquisition request, and “F” indicates a return request. The location indicates the start address, the size indicates the size of the virtual memory area, the request source indicates the request source (such as the job space to be measured) that has requested acquisition of the virtual memory, and the time indicates the time (TOD) at the time of the request. Represent.

FIG. 3 (b) shows an example of data stored in the acquisition table shown in FIGS. 3 (c) and 3 (d). From the top, “next address” indicates an address for pointing to the next acquisition table, “acquisition address” indicates the head address of the acquired virtual memory, “acquisition size” indicates the size of the acquired virtual memory, “Acquisition / return request source” is the acquisition request source /
The return request source is indicated, and “acquisition / return time” is the acquisition time /
The return time indicates the number of unreturned virtual memories.

FIG. 3 (c) shows only the unreturned trace information at the time point n1 or n2.
-1 is stored. The acquisition table 2-1 is obtained by pointing each acquisition table from a corresponding pointer of the hash table B2 pointed from the acquisition table terminal B1 (pointer corresponding to the hash function obtained from the trace information) in order to perform a high-speed search. I have.

FIG. 3 (d) shows the unreturned trace information at the time points n1 and n2 in FIG. 3 (c).
Further, the number of unreturned times is added. The number of unreturns has a value of +1 each time continuous unreturn is performed. For example, the number of unreturns “2” in the trace information shown in FIG. Since it has not been returned as shown in the above, the value is incremented by 1 at the next time point n2 to "2". Hereinafter, similarly, as the number of unreturns increases continuously, the number of unreturns is sequentially added. In this way, the trace information to which the number of unreturns is added is stored in the return-forgotten-candidate table 3-1 and, if necessary, is sorted and printed out in a descending order of the number of unreturned times or in the order of the acquisition time.

Next, FIG. 3 (a) from time 1 to time 5 using FIG.
The process of storing only the unreturned information based on the trace information up to the time point n1 and generating the acquisition table B3 at the time point n1 in FIG. 3 (c) will be specifically described.

FIG. 4 (a) shows the trace information "G6050A1" (G: acquisition request) at time 1 in the position shown on the second line and below in FIG. 3 (b), respectively. This shows how the corresponding data is stored.

FIG. 4 (b) shows data corresponding to the trace information “G1050A2” at time 2 in FIG. 3 (b) corresponding to the positions shown in the second and subsequent rows of FIG. 3 (b). Show the situation.

FIG. 4C shows the trace information “G 2000 300 B 3” at time 3 in FIG.
This shows a state in which the corresponding data is stored at the positions shown in the rows below.

FIG. 4 (d) shows the trace information “F1050A4” (F: return request) at time 4 in FIG. 3 (b) as a return request. Since it was present, it is shown that this has been deleted and an unreturned acquisition table has been generated.

FIG. 4E shows data corresponding to the trace information “G1010A5” at time 5 in FIG. 3B, corresponding to the positions shown in the second and subsequent rows in FIG. 3B. Show the situation. Note that, at this time, the information is inserted at the head from the relation of points in the hash table in ascending order of location (address).

The acquisition table shown in FIG. 4 (e) created as described above is B3 at the time point n1 in FIG. 3 (c). Similarly, when the trace information relating to the unreturned virtual memory is generated from time 6 to time 10 in FIG.
B3 at the time of n2. In addition, while the acquisition table shown in B3 at the time of n2 is being generated, the return is performed in a plurality of times. In this case, the latest return time is stored.

〔The invention's effect〕

As described above, according to the present invention, the number of unreturned times is added every time a specified number of trace information items are collected for all the areas constituting the virtual memory during system operation without modifying the created program. Since the configuration is adopted in which the unreturned trace information is generated and output as a list, it is possible to easily verify the area, size, request source (program), and the like of the virtual memory that has been forgotten to return.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the principle of the present invention, FIG. 2 is a block diagram showing one embodiment of the present invention, FIG. 3 is a diagram illustrating a specific example of the present invention, and FIG. In the figure, 1 is a trace unit, 2 is a management unit, 2-1 is an acquisition table, 3 is a detection unit, and 3-1 is a return forgettable candidate table.

Claims (1)

(57) [Claims] In a forgotten return detection method for detecting a virtual memory that has been forgotten to be returned, each time a virtual memory is acquired or returned, the acquisition / return is performed.
A trace unit (1) for collecting trace information including return, area size, address information, and the like; and a virtual memory that has not been returned at the present time when the trace information collected by the trace unit (1) exceeds a prescribed number. Acquisition table of trace information related to (2-1)
And a management unit (2) that stores the unreturned virtual memory trace information at this point in time and a value obtained by incrementing the number of unreturned times for each of the continuously unreturned trace information by one each time. A detection unit (3) for generating a candidate table (3-1), and configured to output trace information having the number of unreturned times stored in the return-forgotten-candidate table (3-1). Forget return method of virtual memory to be returned.