代码之家 › 专栏 › 技术社区 › M-J

是什么导致此程序出现分段错误?

stack-overflow segmentation-fault c

M-J · 技术社区 · 7 年前

请通过在我的程序中调用以下函数来帮助我解决此SIGSEGV错误:

int* calculateFitness(int** population, int** townDistancesMatrix, int chromoSize){
    int sum = 0;
    static int* Fitnesses;
    Fitnesses = malloc(sizeof(int)*chromoSize); 
    for(int i=0; i<chromoSize; i++){
        int indexOne = 0;
        int indexTwo = 0;
        for(int j=0; j<chromoSize-1; j++){
            indexOne = population[i][j];
            indexTwo = population[i][j+1];
            //printf("\n%d %d",indexOne-1,indexTwo-1);
            sum += townDistancesMatrix[indexOne-1][indexTwo-1];
        }
        indexOne = population[i][0];
        sum += townDistancesMatrix[indexTwo-1][indexOne-1];
        Fitnesses[i] = sum;
        sum = 0;
    }
    return Fitnesses;
}

对于较小的输入(如5个城镇),程序运行没有问题,所以我首先怀疑这是堆栈溢出,因为程序总是运行一段时间(直到所有运行的I值相似)( i=20 ))然后停止运行并给出此错误(在GDB中):

程序接收信号SIGSEGV,分段故障。计算能力为0x0000000008000b9b(人口=0x7ffffffedcd0,城镇距离矩阵=0x8403470,染色体大小=48),未计算1.c:97 97总和+=城镇距离矩阵[indexOne-1][indexTwo-1];

calculateFitness 所以我认为这可能是由于我的函数中有大量的局部变量造成的,但是局部变量很少,而且很小,我的数组也是动态创建的,不在堆栈上运行(可能是嵌套循环的问题?)。

我还运行了valgrind(虽然我还不太熟悉它的报告,我只是用它来获取一些提示),下面是报告:

==198== error calling PR_SET_PTRACER, vgdb might block
==198== Use of uninitialised value of size 8
==198==    at 0x108B41: calculateFitness (Untitled1.c:92)
==198==    by 0x108866: main (Untitled1.c:29)
==198==
==198== Use of uninitialised value of size 8
==198==    at 0x108B6E: calculateFitness (Untitled1.c:93)
==198==    by 0x108866: main (Untitled1.c:29)
==198==
==198== Invalid read of size 4
==198==    at 0x108B9B: calculateFitness (Untitled1.c:97)
==198==    by 0x108866: main (Untitled1.c:29)
==198==  Address 0x522d43c is 4 bytes before a block of size 192 alloc'd
==198==    at 0x4C2FB0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck- 
amd64-linux.so)
==198==    by 0x108A56: readDistances (Untitled1.c:74)
==198==    by 0x1087EB: main (Untitled1.c:19)
==198==
==198== Invalid read of size 8
==198==    at 0x108B87: calculateFitness (Untitled1.c:97)
==198==    by 0x108866: main (Untitled1.c:29)
==198==  Address 0x522d278 is 8 bytes before a block of size 384 alloc'd
==198==    at 0x4C2FB0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck- 
amd64-linux.so)
==198==    by 0x108A20: readDistances (Untitled1.c:71)
==198==    by 0x1087EB: main (Untitled1.c:19)
==198==
==198==
==198== Process terminating with default action of signal 11 (SIGSEGV)
==198==  Access not within mapped region at address 0xFFFFFFFFFC000018
==198==    at 0x108B9B: calculateFitness (Untitled1.c:97)
==198==    by 0x108866: main (Untitled1.c:29)
==198==  If you believe this happened as a result of a stack
==198==  overflow in your program's main thread (unlikely but
==198==  possible), you can try to increase the size of the
==198==  main thread stack using the --main-stacksize= flag.
==198==  The main thread stack size used in this run was 8388608.

//...

==198== LEAK SUMMARY:
==198==    definitely lost: 0 bytes in 0 blocks
==198==    indirectly lost: 0 bytes in 0 blocks
==198==      possibly lost: 0 bytes in 0 blocks
==198==    still reachable: 13,632 bytes in 70 blocks
==198==         suppressed: 0 bytes in 0 blocks

我搜索了这个错误的部分,比如“仍然可以到达”,这似乎不是我需要注意的事情,但即使在搜索了第一部分之后,我也不确定它们的含义。我做错了什么?如果它真的是stackoverflow,除了递归之外,还有什么其他原因导致stackoverflow?

1 回复 | 直到 7 年前

bruno 7 年前

int main(){
  int* population[POPSIZE];

和

for(int i=0; i<chromoSize; i++){
    int indexOne = 0;
    int indexTwo = 0;
    for(int j=0; j<chromoSize-1; j++){
        indexOne = population[i][j];
        indexTwo = population[i][j+1];

您可以从人口 (发出信号) 瓦尔格林 ),您给的值为48 chromo_size (初始化色浆 爆米花 等于20

之后 indexOne 和 索引2 具有随机值,因此 城市距离矩阵 产生valgrind发出的seg故障信号

主要的 scanf("%d",&chromo_size); 检查值是否为<= 爆米花 为了避免这个问题,增加 爆米花 能够使用输入文件运行

另请注意 free(population) 是一个局部变量,它不在堆中分配