WO2010043176A1

WO2010043176A1 - Memory leak detecting method and device

Info

Publication number: WO2010043176A1
Application number: PCT/CN2009/074435
Authority: WO
Inventors: 杨胜强
Original assignee: 华为技术有限公司
Priority date: 2008-10-17
Filing date: 2009-10-14
Publication date: 2010-04-22
Also published as: CN101414272A; CN101414272B

Abstract

A memory leak detecting method and device, the method comprises: first judgement process, judging whether the current occupied time of a detected memory exceeds a life period preset for the detected memory; and triggering second judgement process when the result of the first judgement process is yes; the second judgement process, judging whether the detected memory is being used by a service process requesting the detected memory; when the result of the second judgement process is yes, confirming that there is no memory leak of the detected memory, when the result of the second judgement process is no, confirming that there is memory leak of the detected memory. A memory leak can be detected accurately and error determination phenomenon is avoided according to the detecting method and device.

Description

Memory leak detection method and device

The present application claims priority to Chinese Patent Application No. 200810201396.0, entitled "Detecting Method and Apparatus for Memory Leakage", filed on October 17, 2008, the entire contents of which is incorporated herein by reference. .

Technical field

The present invention relates to the field of computer applications, and in particular, to a method and apparatus for detecting a memory leak.

Background technique

The memory used for data storage is a necessary resource for the business process to run. The business process needs to apply before using the memory, and starts to occupy the memory. When the memory is used, the memory is released in time. The memory occupied by the business process is usually divided into two cases. One is the exclusive use of memory. When the memory is occupied by a business process and is occupied, it cannot be applied and occupied by other business processes. When the memory is released into the memory resource pool by the currently occupied service process, the memory can be applied and occupied by other service processes. Another situation is the reuse of memory, that is, memory can be occupied by at least two business processes at the same time. In object-oriented programming techniques, memory can record its own references, indicating that it is referenced by multiple business processes, that is, each memory can be occupied by multiple business processes. A memory leak means that the business process does not release the memory after using the memory for various reasons. Memory leaks can cause fewer and fewer memory applications to be applied and occupied by business processes in the system, and eventually the system will exit the service due to insufficient memory. Therefore, the detection of memory leaks is very important.

In the prior art, the method for detecting a memory leak is: determining whether the occupied time of the memory occupied by the business process exceeds a life cycle preset for the memory, and when the occupied time of the memory exceeds the life cycle, determining the memory is Whether the number of references by the business process is zero. If it is zero, it is determined that the memory has a memory leak. If it is not zero, it is determined that there is no memory leak in the memory.

However, the inventors found problems in the prior art in the research: The number of references to memory is updated when the business process releases the memory. When the business process does not correctly execute the process of releasing the memory, it will cause the reference recorded by the memory itself. The number is wrong. Therefore, it is possible to judge whether the memory leaks through the number of references recorded by the memory itself. When the memory is erroneously released due to misjudgment or the leaked memory is not released in time, the normal operation of the existing system is further affected. Summary of the invention

Embodiments of the present invention provide a method and apparatus for detecting a memory leak, which can accurately detect a memory leak and avoid a false positive phenomenon.

The embodiment of the invention discloses a method for detecting a memory leak, the method comprising: first determining a life cycle of the memory setting; and when the result of the first determining process is YES, triggering a second determining process, the The determining process is specifically: determining whether the detected memory is being used by a service process of the detected memory; when the result of the second determining process is YES, determining that the detected memory has no memory leak When the result of the second determining process is no, it is determined that a memory leak occurs in the detected memory.

The embodiment of the invention discloses a device for detecting a memory leak, the device comprising: a life cycle of the first detection setting, if yes, triggering a second detection unit; and a second detection unit, configured to determine whether the detected memory is It is used by the business process that is being applied for the detected memory, and if so, it is determined that no memory leak has occurred in the detected memory, and if not, it is determined that a memory leak occurs in the detected memory.

It can be seen from the foregoing embodiment of the present invention that when it is determined that the current occupied time of the detected memory exceeds the expected life cycle, the independent reverse traceability judgment is performed, that is, the service process applying for the memory is located, and then the judgment is performed. The memory is actually used by the business process. If it is determined that the business process has ended or the memory has been released by the business process, it is finally determined that a memory leak has occurred in the detected memory. This can prevent the occurrence of memory leak misjudgment and avoid the problem that the system leaks due to memory leak misjudgment.

DRAWINGS

1 is a flow chart of a first embodiment of a method for detecting a memory leak according to the present invention;

Figure 2 is a schematic diagram of the same memory being reused by three business processes;

3 is a structural diagram of a first embodiment of a memory leak detecting device of the present invention;

4 is a flow chart of a second embodiment of a method for detecting a memory leak according to the present invention

FIG. 5 is a structural diagram of a second embodiment of a memory leak detecting apparatus according to the present invention; FIG.

6 is a structural diagram of a third embodiment of a method for detecting a memory leak according to the present invention; detailed description

The above described objects, features, and advantages of the present invention will become more apparent from the aspects of the invention. Please refer to FIG. 1 , which is a flowchart of a first embodiment of a method for detecting a memory leak according to the present invention. The detecting method may include the following steps:

Step 101: Determine whether the time that the detected memory is currently occupied by the service process exceeds the life cycle set in advance for the detected memory, and if yes, proceed to step 102;

The detected memory can be occupied by only one service process at the same time, and can also be reused by multiple service processes. When the detected memory is reused by multiple service processes at the same time, a unified life cycle may be preset for each service process, or different life cycles may be set for different business processes according to the business process logic of the business process. value. When different business processes repeatedly refer to the same block of memory, here, three service processes sequentially refer to the same block of memory as an example to illustrate the process of determining the lifetime value of the detected memory. Service process 1 applies for a block of memory at time tl. In business process 1, the lifetime of the memory is T1. Then, when service process 1 has not released the memory, business process 2 applies for the same block of memory at time t2. In business process 2, the lifetime of the memory is T2. As shown in Figure 2, at this time, the service process 1 and the service process 2 reuse the same memory. When the service process 2 refers to the memory, the life cycle of the memory is updated. At this time, the updated memory life cycle is the business process 2 When the memory is referenced, the remaining time of the memory in the business process 1 and the maximum lifetime of the memory under the business process 2, or the remaining time of the memory in the business process 1 and the business process when the business process 2 refers to the memory 2 The sum of the lifetime values of the memory. After that, another business process 3 applies for the memory at time t3, and under business process 3, the lifetime of the memory is T3. When the service process 3 references the memory, and the life cycle of the memory is re-updated, the updated life cycle is the remaining time of the memory in the business process 2 when the memory is referenced by the business process 3 and the memory of the business process 3 The maximum value of the life cycle, or the sum of the remaining time of the memory in business process 2 and the life cycle of the memory under business process 3 when business process 3 refers to the memory, and so on.

In the above steps, when the determination result is no, it is determined that no memory leak has occurred in the detected memory. Step 102: Determine whether the detected memory is being used by the service process of the detected memory, if yes, proceed to step 103, if no, proceed to step 104;

The above steps only determine whether the detected memory is used by the service process when the detected memory is applied by one service process at the same time. When the detected memory is applied for and reused by multiple service processes at the same time, the above steps are performed. It is necessary to judge whether the detected memory is used by each business process one by one. When there is a business process that is actually using the detected memory, it is determined that there is no memory leak in the detected memory, when all the applications for detecting the memory are detected. When the process does not use the detected memory, it is determined that a memory leak occurs in the detected memory.

The present invention does not limit the order of judgment of various business processes, and the order of judgment can be arbitrarily set. When it is judged one by one whether the detected memory is used by all the business processes, a preferred method is that when it is first determined that a certain service process is using the detected memory, the determination process may be terminated, and the detected memory is determined. No memory leaks have occurred.

The determining process in the foregoing steps specifically includes: first identifying, according to the service process identifier, a service process that applies for the detected memory; and then determining whether the service process exists and whether a memory address saved by the service process is related to the The address of the memory is the same. When the service process exists and the memory address saved by the service process is the same as the address of the detected memory, the determination result is yes, when the service process does not exist or/and If the memory address saved by the business process is different from the address of the detected memory, the determination result is no.

Step 103: Determine that no memory leak occurs in the detected memory, and end the process;

After determining that no memory leak occurs in the detected memory, the step may further include: resetting a life cycle of the detected memory. The purpose of resetting the detected memory life cycle is to allow the detected memory to be re-detected in the next cycle.

Step 104: Determine that a memory leak occurs in the detected memory, and end the process.

After determining that a memory leak occurs in the detected memory, the step may further include: releasing the detected memory.

Corresponding to the above-mentioned method for detecting a memory leak, the embodiment of the present invention further provides a device for detecting a memory leak. Please refer to FIG. 3, which is a structural diagram of a first embodiment of a memory leak detecting apparatus according to the present invention. The apparatus includes a first detecting unit 301 and a second detecting unit 302. The internal structure and connection relationship will be further described below in conjunction with the working principle of the device. The life cycle of the detected memory, if yes, triggering the second detecting unit 302;

The second detecting unit 302 is configured to determine whether the detected memory is being used by a service process that is applied to the detected memory, and if yes, determine that no memory leak occurs in the detected memory, and if not, determine that the memory is Check for memory leaks in memory.

The embodiment may further include a releasing unit 305, configured to release the detected memory after the second detecting unit 302 determines that a memory leak occurs in the detected memory.

The embodiment may further include a lifecycle resetting unit 306, configured to reset the life cycle of the detected memory after the second detecting unit 302 determines that no memory leak occurs in the detected memory.

The second detecting unit 302 includes an identifying unit 303, configured to identify a service process for applying the detected memory according to the service process identifier, and a determining unit 304, configured to determine whether the service process exists and the service process is Whether the stored memory address is the same as the address of the detected memory, and if so, it is determined that no memory leak has occurred in the detected memory, and if not, it is determined that a memory leak occurs in the detected memory.

It can be seen from the foregoing embodiment of the present invention that when it is determined that the current occupied time of the detected memory exceeds the expected life cycle, the independent reverse traceability judgment is performed, that is, the service process applying for the memory is located, and according to the business process. In actual use, if it is determined that the business process has ended or the memory has been released by the business process, it is finally determined that a memory leak occurs in the detected memory. Therefore, the occurrence of memory leak misjudgment can be completely prevented, and the problem of normal execution of the service due to memory leak misjudgment is avoided.

In addition, when using a unified pre-set memory lifecycle, it is not possible to adapt to a variety of memory usage scenarios. For example, some time-consuming business processes still use memory normally, but repeated memory leak checks due to timeouts cause the system to waste resources. Some of the less time-consuming memories may not be detected in time because the current occupied time has not reached the life cycle. In the embodiment of the present invention, according to the service processing logic of the service process, the service process sets different memory lifecycle values for the memory applied for different scenarios, thereby avoiding the above problems and making the system operation more stable. Referring to FIG. 4, it is a flowchart of a second embodiment of a method for detecting a memory leak according to the present invention. The difference from the first embodiment is that, in the embodiment, when the Detection When the memory is being used by a business process, the lifecycle set in advance for the detected memory is modified. The detection method can include the following steps:

Step 401: Determine whether the time that the detected memory is currently occupied by the service process exceeds a life cycle set in advance for the detected memory, and if yes, proceed to step 402;

The detected memory can be occupied by only one service process at the same time, and can also be reused by multiple service processes. When the detected memory is reused by multiple service processes at the same time, a unified memory life cycle may be preset for each business process, or different memory may be set according to business process logic of the business process. Life cycle value.

In the above steps, when the determination result is no, it is determined that no memory leak has occurred in the detected memory. Step 402: Determine whether the detected memory is being used by the service process of the detected memory, if yes, proceed to step 403, if no, proceed to step 405;

Wherein, when the detected memory is applied and exclusive by only one service process at the same time, the foregoing step only judges whether the detected memory is used by the service process, and when the detected memory is applied and reused by multiple service processes at the same time, the above The steps need to determine whether the detected memory is used by each business process one by one. When there is a business process that is actually using the detected memory, it is determined that no memory leak occurs in the detected memory, when all applications and reuses are detected. When the memory of the service process does not use the detected memory, it is determined that a memory leak occurs in the detected memory.

The present invention does not limit the order of judgment of various business processes, and the order of judgment can be arbitrarily set. When it is judged one by one whether the detected memory is used by no business process, a preferred method is that when it is first determined that a certain service process is using the detected memory, the determination process may be terminated, and the detected memory is determined. No memory leaks have occurred.

Step 403: Determine that no memory leak occurs in the detected memory.

Step 404: Add a life cycle set in advance for the detected memory, and end the process; wherein, according to the business processing logic of each business process, different life cycle values are set for different business processes and the detected memory is simultaneously When a business process is reused, it is necessary to comprehensively determine the life cycle of a detected memory according to the lifetime value of each business process. The determining method has been described in detail in the first embodiment of the memory leak detecting method. Let me repeat. In this step, the added life cycle is the lifetime value of the detected memory determined by the life cycle value of each business process. Step 405: Determine that a memory leak occurs in the detected memory, and end the process.

Corresponding to the above-mentioned method for detecting a memory leak, the embodiment of the present invention further provides a memory leak detecting device. Referring to FIG. 5, which is a structural diagram of a second embodiment of a memory leak detecting apparatus according to the present invention, the apparatus includes a first detecting unit 501, a second detecting unit 502, and a life cycle increasing unit 503. The internal structure and connection relationship will be further described below in conjunction with the working principle of the device.

The first detecting unit 501 is configured to determine whether the time occupied by the detected memory by the service process exceeds a life cycle set in advance for the detected memory, and if so, triggering the second detecting unit 502;

The second detecting unit 502 is configured to determine whether the detected memory is being used by a service process that is applied to the detected memory, and if yes, determining that a memory leak does not occur in the detected memory, triggering a lifecycle increasing unit 503, If not, it is determined that a memory leak has occurred in the detected memory.

The life cycle increasing unit 503 is configured to increase the life cycle of the detected memory after the second detecting unit 502 determines that the detected memory does not have a memory leak.

The embodiment may further include a release unit, configured to release the detected memory after the second detecting unit 502 determines that a memory leak occurs in the detected memory.

The second detecting unit 502 includes an identifying unit, configured to identify a service process for applying the detected memory according to the service process identifier, and a determining unit, configured to determine whether the service process exists, and determine the detected memory. No memory leak occurred, and if not, it is determined that a memory leak has occurred in the detected memory.

It can be seen from the foregoing embodiment that when it is determined that the current occupied time of the detected memory exceeds a preset life cycle, the independent reverse traceability judgment is performed, that is, the service process of applying for the memory is located, and according to the business process. In actual use, if it is determined that the business process has ended or the memory has been released by the business process, it is finally determined that a memory leak occurs in the detected memory. Therefore, it is possible to completely prevent the occurrence of memory leak misjudgment, and avoid the problem that the memory leakage misjudgment affects the normal execution of the service. At the same time, if it is determined that the detected memory is not used by the business process, it is determined that the detected memory is not out. Now the memory leaks. At this time, by adding the life cycle to reset the lifetime value of the detected memory, it is prevented that the memory is detected again and the misjudgment occurs, and the memory detection is prevented from being too frequent, and the system CPU is wasted processing resources.

In addition, when using a unified pre-set memory lifecycle, it is not possible to adapt to a variety of memory usage scenarios. For example, some long-running business processes still use memory normally, but repeated memory leak checks due to timeouts cause the system to waste resources. Some of the less time-consuming memories may not be detected in time because the current occupied time has not reached the life cycle. According to the service processing logic of the service process, the service process sets different memory lifecycle values for the memory applied by different service scenarios in different service scenarios, thereby avoiding the above problems and making the system operation more stable. Referring to FIG. 6, which is a flowchart of a third embodiment of a method for detecting a memory leak according to the present invention, this embodiment is a preferred method for memory detection when a detected memory is exclusively occupied by one service process. After the service process successfully requests the memory, a memory information record is generated, and the related information of the memory is recorded, which may include: a memory address, a memory application time, a service process identifier, a lifetime of the detected memory, and a remaining time, and then The memory information record is placed into an in-memory repository. After the business process releases the memory successfully, the memory information record in the memory information library is released. After the system enters the detection process of the memory leak, the detection method includes the following steps:

Step 601: Access the memory information library, and find the memory with zero remaining time, and use this memory as the detected memory;

Among them, the remaining time in the memory record = the life cycle of the detected memory - the current occupied time of the detected memory, the current occupied time of the detected memory = the current time of the system - the memory application time. The memory application time is the time point recorded when the business process requests this memory successfully; the lifetime of the detected memory is the time occupied by the business process set in advance for the detected memory, which is an approximate estimated value.

The life cycle may preset a unified memory life cycle for each business process, or set different memory life cycle values for different business processes according to the business process logic of the business process. A preferred solution is: Temporary memory applied within a function, the life cycle can be directly filled in 0, indicating that the requested memory is released immediately after the function is executed. If the memory is requested when sending a message, since the message memory is released from the application to the slave port, Generally it will not exceed 1000ms, so the life cycle can be set to 1000ms. If there is some complicated process processing, it is necessary to release the memory until the end of the process. The business module can basically determine the life cycle of the applied memory according to the process and system processing performance. When the result obtained by the calculation is a number smaller than 0, the remaining time takes a value of 0. When the remaining time of the detected memory is not 0, it indicates that the current occupied time of the detected memory does not exceed the life cycle of the detected memory, and no memory leak occurs in the detected memory at this time.

Step 602: Identify, according to the service process identifier in the memory record, a service process that applies for the detected memory.

A service process identifier uniquely identifies a service process, which may be the address of the service process control block or the number of the service process control block.

Step 603: Determine whether the detected memory is occupied by the service process indicated by the service process identifier, if yes, go to step 604, if no, go to step 605;

Wherein, when the remaining time of the detected memory is 0, it indicates that the current occupied time of the detected memory has exceeded the life cycle of the detected memory, and if the occupied time of the detected memory is related to the duration of the business process, in order to prevent misjudgment, It is necessary to further determine whether the detected memory is being occupied by a business process indicated by the service identifier.

The above step 603 can be specifically implemented by a business process detection function, which has two parameters: a memory address and a business process identifier. For the convenience of release, the service process generally records the memory address requested by the service process control block, determines the corresponding service process control block by the service process identifier, and determines whether the service process exists in the service process control block and the service process control block The saved memory address is the same as the memory address of the business process detection function input parameter. If yes, it is determined that the detected memory is occupied by the business process indicated by the service process identifier. If not, it is determined that the detected memory is not indicated by the service process identifier. Business process is occupied.

Step 604: determining that no memory leak occurs in the detected memory, increasing the life cycle of the detected memory, and ending the process;

When the detected memory is still occupied by the service process indicated by the service identifier, it indicates that the detected memory has no leakage, and the life cycle in the memory record of the detected memory needs to be increased. A preferred method is to double the original life cycle.

Step 605: determining that a memory leak occurs in the detected memory, according to the memory address in the memory record. Release the detected memory and end the process.

When the detected memory is not occupied by the service process indicated by the service identifier, it indicates that the detected memory has a leak, and the actual physical address of the detected memory is found according to the memory address in the memory record of the detected memory. And release the detected memory.

It can be seen from the foregoing embodiment of the present invention that when it is determined that the current occupied time of the detected memory exceeds the expected life cycle, the independent reverse traceability judgment is performed, that is, the service process applying for the memory is located, and according to the business process. In actual use, if it is determined that the business process has ended or the memory has been released by the business process, it is finally determined that a memory leak occurs in the detected memory. Therefore, the occurrence of memory leak misjudgment can be completely solved, and the problem that the service leakage is affected due to the memory leak misjudgment is avoided. At the same time, if it is determined that the detected memory is used by the business process, it is determined that there is no memory leak in the detected memory. At this time, by adding a life cycle, the life cycle value is reset for the detected memory to prevent the memory from being detected again. A case of misjudgment.

In addition, when using a unified pre-set memory lifecycle, it is not possible to adapt to a variety of memory usage scenarios. For example, some long-running business processes still use memory normally, but repeated memory leak checks due to timeouts cause the system to waste resources. Some of the less time-consuming memories may not be detected in time because the current occupied time has not reached the life cycle. According to the service processing logic of the service process, the service process sets different memory lifecycle values for the memory applied by different service scenarios in different service scenarios, thereby avoiding the above problems and making the system operation more stable.

One of ordinary skill in the art can understand that all or part of the steps of implementing the above embodiments may be completed by a program instructing related hardware, and the program may be stored in a computer readable storage medium, and the program includes the following The first judging process, the second judging process has a life cycle; when the result of the first judging process is yes, the second judging process is triggered, specifically: determining whether the detected memory is being applied for The service process of the detected memory is used; when the result of the second determining process is YES, it is determined that no memory leak occurs in the detected memory, and when the result of the second determining process is no, the determining A memory leak occurs in the detected memory. The storage medium is, for example, a ROM/RAM, a magnetic disk, an optical disk, or the like.

The above description is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the present invention. These improvements and modifications should also be considered as protection scope of the present invention.

Claims

Rights request

A method for detecting a memory leak, the method comprising: setting a life cycle for the detected memory;

When the result of the first determining process is YES, the second determining process is triggered, where the second determining process is specifically: determining whether the detected memory is being used by a service process that is applied for the detected memory;

When the result of the second determining process is YES, it is determined that no memory leak occurs in the detected memory, and when the result of the second determining process is negative, it is determined that a memory leak occurs in the detected memory.

2. The method according to claim 1, wherein the method further comprises: increasing a life cycle of the detected memory after the second determining process determines that a memory leak has not occurred in the detected memory.

The method according to claim 1, wherein the method further comprises: releasing the detected memory after the second determining process determines that a memory leak occurs in the detected memory.

4. The method according to claim 1, wherein the life cycle set in advance for the detected memory comprises:

A life cycle that is set in advance for the detected memory of all business processes;

Or,

According to the business process logic of the business process itself, the life cycle set for the detected memory of each business process is pre-defined.

The method according to claim 1, wherein the second determining process comprises: identifying, according to the service process identifier, a service process for applying for the detected memory;

Determining whether the service process exists and whether the memory address saved by the service process is the same as the address of the detected memory, and if yes, the determination result of the second determining process is yes, if not, the second The judgment result of the judgment process is No.

The method according to claim 5, wherein the service process identifier comprises: an address of a service process control block or a number of a service process control block.

The method according to claim 5, wherein the according to the service process identifier, Identifying the business processes for applying for the detected memory includes:

After the service process is applied to the detected memory, the memory process includes the memory record containing the service process identifier, and the application for the detected memory is identified according to the service process identifier in the memory record. Business process.

8. The method according to claim 7, wherein the memory record comprises: a memory address, a memory application time, a business process identifier, and a memory life cycle.

The method according to any one of claims 1-8, wherein the detected memory comprises:

Memory that is exclusively occupied by one business process at the same time or memory that is reused by at least two business processes at the same time.

10. The method according to claim 1, wherein the life cycle of the detected memory is reset after the second determining process determines that no memory leak has occurred in the detected memory.

A device for detecting a memory leak, characterized in that: the device comprises: a life cycle of the detected memory setting, and if so, triggering the second detecting unit;

a second detecting unit, configured to determine whether the detected memory is being used by a service process that is applied to the detected memory, and if yes, determine that a memory leak does not occur in the detected memory, and if not, determine that the detected A memory leak has occurred in the memory.

The device according to claim 11, wherein the second detecting unit comprises: an identifying unit, configured to identify, according to the service process identifier, a service process for applying for the detected memory;

a determining unit, configured to determine whether the service process exists and whether a memory address saved by the service process is the same as an address of the detected memory, and if yes, determining that a memory leak does not occur in the detected memory, if not, A memory leak is determined to be detected in the detected memory.

The device according to claim 11, wherein the device further comprises: a life cycle increasing unit, configured to: add, after the second detecting unit determines that the detected memory does not have a memory leak The lifetime of the memory being tested.

The device according to claim 11, wherein the device further comprises: a releasing unit, configured to: when the second detecting unit determines that a memory leak occurs in the detected memory After that, the detected memory is released.

The device according to claim 11, wherein the device further comprises: a lifecycle resetting unit, configured to: after the second detecting unit determines that the detected memory has not experienced a memory leak, resetting The life cycle of the detected memory.