【发布时间】：2018-10-15 14:45:32
【问题描述】：
我目前正在尝试对目录和子目录中的多个文件进行排序，并将排序后的文件输出到不同的指定目录。我应该为遇到的每个子目录和每个实际排序使用分叉进程。我还应该跟踪我创建了多少子进程并打印每个子进程的 PID（进程 ID）。
void iterate_dir(char* current_path, char* new_dir, char* column_name){
DIR *dd = opendir(current_path);
struct dirent *curr;
int status = -1;
while((curr = readdir(dd))!=NULL){
if((strcmp(curr->d_name,".")==0 || strcmp(curr->d_name,"..")==0 || strcmp(curr->d_name,new_dir)==0) && curr->d_type==DT_DIR){
continue;
}
//Just formating stuff, I'm only supposed to sort files ending with .csv
if(curr->d_type!=DT_DIR){
if(strlen(curr->d_name)<4){
continue;
}
char* extension = strrchr(curr->d_name,'.');
if(extension==NULL){
continue;
}
if(strcmp(extension,".csv")!=0){
continue;
}
}
int pid = fork();
if(pid==0){
(*num_of_child_procs)++;
//child process
if(curr->d_type==DT_DIR){
char new_path[strlen(curr->d_name)+strlen(current_path)+2];
sprintf(new_path,"%s/%s",current_path,curr->d_name);
iterate_dir(new_path, new_dir, column_name);
exit(getpid());
}else{
//more formatting stuff, I'm supposed to output a new sorted file to the specified output directory
char file_name[strlen(curr->d_name)-3];
memcpy(file_name,curr->d_name,strlen(curr->d_name)-4);
file_name[strlen(curr->d_name)-4] = '\0';
char new_path[strlen(file_name)+strlen(column_name)+strlen(new_dir)+12+1+1];
sprintf(new_path,"%s/%s%s%s%s",new_dir,file_name,"-sorted-",column_name,".csv");
char old_path[strlen(curr->d_name)+strlen(current_path)+2];
sprintf(old_path,"%s/%s",current_path,curr->d_name);
int output = open(new_path, O_WRONLY|O_CREAT, 0666);
int input = open(old_path, O_RDONLY);
sort(input,output)
if(ret==-1){
_exit(-1);
}
if(ret==-2){
remove(new_path);
}
exit(getpid());
}
}
}
//This is where I print the pids for each process, by using the return status from each child process exit()
while(wait(&status)>0){
printf("%d,",status);
}
closedir(dd);
}
在我的 main 我这样调用函数
printf("Initial PID:%d\n",getpid());
printf("PIDS of all child processes:");
//mmap is only used for obtaining the number of processes, not printing their pids
num_of_child_procs = mmap(NULL, sizeof *num_of_child_procs, PROT_READ |
PROT_WRITE,MAP_SHARED | MAP_ANONYMOUS, -1, 0);
*num_of_child_procs = 0;
iterate_dir(dir_name, new_dir_name, column_name);
printf("\nTotal number of processes:%d\n", *num_of_child_procs);
munmap(num_of_child_procs, sizeof *num_of_child_procs);
除了输出之外，一切都很好。在包含 2 个 .csv 文件的目录、包含 2 个 .csv 文件的子目录和包含 1 个 .csv 文件的子目录的子目录（总共 7 个进程）上运行代码，打印出
Initial PID:9773
PIDS of all child processes:PIDS of all child processes:PIDS of all child
processes:PIDS of all child processes:PIDS of all child processes:PIDS of
all child processes:13312,PIDS of all child processes:13056,12800,12544,PIDS
of all child processes:12032,12288,11776,
Total number of processes:7
如您所见，由于某种原因，它会多次打印“所有子进程的 PIDS：”，尽管只在我的主进程中调用它。谁能解释它为什么这样做？
【问题讨论】：
-
"我使用每个子进程 exit() 的返回状态打印每个进程的 pids" 您阅读了
wait()的哪些文档？