支持内存错误检测的改进gdb调试器的设计与实现

发布时间：2018-08-30 10:19

【摘要】：C语言在各种系统软件和应用软件的开发中有着广泛的应用,它的内存管理机制和指针操作机制,大大提高了其表达能力和灵活性,但也导致了大量内存错误的出现,这些错误不易重现且难以查找诱发错误的原因,所以内存错误通常很难检测。目前的内存错误检测方法根据是否需要执行程序大致可以分为两类：动态方法和静态方法。静态方法不需要执行程序,因此没有执行开销,检测效率高,但由于程序运行过程中的不确定性和不可预测性,静态方法不能准确地把握内存的实际运行状况,通常具有较高的误报率和漏报率。动态方法在程序执行时收集信息,并由此定位到程序发生内存错误的位置。但动态方法收集信息的方式一般是对程序的源代码进行插桩修改,这种改动可能破坏程序的结构并影响程序的功能。另外,现有的一些工具在静态数组的越界访问检测方面并不理想。针对以上问题,本文通过改进gdb调试器,提出了在gdb调试器中实现内存错误检测的解决方案。改进后的gdb调试器可以在不修改任何源代码的前提下,实现内存越界访问错误和内存泄露错误的检测,而且摆脱了静态方法产生的误报问题。其设计思想是：1)改进后的gdb调试器在调试程序的过程中,根据程序内存分布的不同,采用不同的方法抓取程序中的内存分配信息。2)根据此信息建立用于内存越界访问检测的内存调试符号表以及用于内存泄露检测的内存泄露链表。3)基于gdb调试器的watchpoint原理,利用内存调试符号表,对程序中的内存分配块进行监控,探测内存越界访问错误；并在程序结束前分析内存泄露链表,检测程序是否产生了内存泄露。4)通过进一步与gdb调试器相结合,使其能够定位程序中发生内存错误的位置。
[Abstract]:C language has been widely used in the development of various system software and application software. Its memory management mechanism and pointer operation mechanism greatly improve its expression ability and flexibility, but also lead to a large number of memory errors. These errors are difficult to reproduce and difficult to find to induce errors, so memory errors are often difficult to detect. Current memory error detection methods can be divided into two categories according to whether or not to execute programs: dynamic method and static method. The static method does not need to execute the program, so it has no execution overhead and high detection efficiency. However, because of the uncertainty and unpredictability in the process of running the program, the static method can not accurately grasp the actual running state of the memory. It usually has high false alarm rate and false alarm rate. The dynamic method gathers information during program execution and thus locates where a memory error occurred in the program. But the method of collecting information by dynamic method is to modify the source code of the program by inserting piles, which may destroy the structure of the program and affect the function of the program. In addition, some of the existing tools for static array cross-border access detection is not ideal. In view of the above problems, this paper proposes a solution to implement memory error detection in gdb debugger by improving the gdb debugger. The improved gdb debugger can detect memory access error and memory leak error without modifying any source code, and get rid of the false positive problem caused by static method. The design idea of the improved gdb debugger is: 1) in the process of debugging the program, according to the different memory distribution of the program, Different methods are used to capture the memory allocation information in the program. 2) based on this information, the memory debug symbol table for memory cross access detection and the memory leak chain table 3 for memory leak detection are established) based on the watchpoint principle of the gdb debugger. The memory debug symbol table is used to monitor the memory allocation block in the program, to detect the memory access error, and to analyze the memory leak list before the end of the program. By further combining with the gdb debugger, the program can locate the location where the memory error occurred in the program.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP311.11;TP333.1

【参考文献】