c++ - 在线程应用程序(linux、pthreads)中读取文件大小时出错

Question

我正在尝试从 linux 中的一个文件夹中读取所有文件和目录，线程为获取当前目录和当前目录树下最大的文件大小和名称。

主线程扫描基目录查找文件，当找到一个目录时，它会生成一个新线程以继续扫描。

此时，线程加入，直到最后创建的线程结束。(我知道这不是最好的方法，但这只是一种练习。)

问题是程序返回了错误的结果，我不知道为什么。

我有以下文件树来测试应用程序:

    . ( Debug folder under codelite project / workspace )    ├── [       4096]  dir1    │   └── [          9]  arch-dir1.txt    ├── [       4096]  dir2    │   ├── [         27]  arch-dir2.txt    │   └── [      29083]  huge    ├── [      29053]  direxp    ├── [      27048]  direxp.o    └── [         68]  direxp.o.d

As you can see the highest file size under current directory it's direxp ( this program ) and the highest file size under tree it's huge

Running the binary, I got the following results:

    dir: .    dir: ..    arch: direxp.o.d    max dir & tree set to: direxp.o.d size: 68    arch: direxp.o    max file dir set to: direxp.o size: 27048    arch: .d    arch: direxp    max file dir set to: direxp size: 29053    dir: dir1     th dir: .     th dir: ..     th arch: arch-dir1.txt thsize: 4096    max tree file set to: arch-dir1.txt thsize: 4096    dir: dir2     th dir: .     th dir: ..     th arch: arch-dir2.txt thsize: 4096     th arch: huge thsize: 4096    Highest current directory file:    direxp tam:29053 bytes.    Highest tree file:    arch-dir1.txt tam:4096 bytes.

th-prefixed strings show data processed in another thread.

I use the functions readdir (main thread) and readdir_r (spawned thread) to read directory entries.

I thought this could be the trouble, but later compiled the program calling readdir_r under all threads and the wrong results persists.

Really I don't understand why file size it's returned wrong (4096 it's default cluster size in my filesystem. So why are files processed as directories ?

Could you give me a hand ? Thanks

main function code

#include <iostream>
#include <cstdio>
#include <cstring>
#include <cstdlib>

#include <sys/types.h>
#include <sys/stat.h>
#include <dirent.h>
#include <unistd.h>
#include <pthread.h>

using std::cout;
using std::cin;
using std::endl;

#define MAX_PATH 255

struct archivo 
{
char nombre[MAX_PATH+1];
off_t tam;
};

// thread args
struct thargs 
{
char nextdir[MAX_PATH+1]; // next dir
void* (*pth)(void*); // pointer to thread function
archivo* arch; // pointer to archivo
};


pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;


int main(int argc, char **argv)
{

char target[MAX_PATH+1] = {0}; // directorio inicial

archivo grande_dir ={{0}},grande_arbol = {{0}};


    // No params
    if ( argc < 2)
    {
        if ( ! getcwd(target,MAX_PATH) )
        {
            perror("Error en path:");
            exit(-1);
        }
    }

    if ( argc == 2)
        strncpy(target,argv[1],MAX_PATH);

    if ( argc > 2)
    {
        perror("Num params incorrecto");
        exit(-2);
    }

    DIR* midir = NULL;

    // try to open target dir
    if ( ! (midir = opendir(target) ) )
    {
        perror("Error abriendo dir:");
        exit(-3);
    }


    dirent* direntry;
    //dirent* rentry1 = NULL;

    struct stat estado = {0}; // struct needed for desambiguation

    bool primera = true; // control var to initialize the search

    // read current dir contents               
    //while( (readdir_r(midir,&direntry,&rentry1) == 0 ) && rentry1  )
    while( (direntry = readdir(midir) ) )
    {

        stat(direntry->d_name,&estado);

        // current entry it's a file
        if ( direntry->d_type == DT_REG )
        {

            cout << "arch: " << direntry->d_name << endl;

            // init search to find the highest file
            if (primera)
            {
                    strncpy(grande_dir.nombre,direntry->d_name,MAX_PATH);
                grande_dir.tam = estado.st_size;

                strncpy(grande_arbol.nombre,direntry->d_name,MAX_PATH);
                grande_arbol.tam = estado.st_size;

                primera = false;

                cout << "max dir & tree set to: " << direntry->d_name << " size: " << estado.st_size << endl;
            }

            // High file size
            if ( estado.st_size > grande_dir.tam)
            {
                pthread_mutex_lock(&lock);

                strncpy(grande_dir.nombre,direntry->d_name,MAX_PATH);
                grande_dir.tam = estado.st_size;

                pthread_mutex_unlock(&lock);

                cout << "max file dir set to: " << direntry->d_name << " size: " << estado.st_size << endl;
            }


        }

        // current entry it's a directory
        if ( direntry->d_type == DT_DIR )
        {

            cout << "dir: " << direntry->d_name << endl;

            // check not . or .. dir
            if ( (strcmp(direntry->d_name,".") != 0) && (strcmp(direntry->d_name,"..") != 0 ) )
            {   

                thargs args = {{0}};

                pthread_t th1;

                pthread_mutex_lock(&lock);

                sprintf(args.nextdir,"%s/%s",target,direntry->d_name);
                args.arch = &grande_arbol;
                args.pth = &procesadir;

                pthread_mutex_unlock(&lock);

                // new thread creation
                pthread_create(&th1,NULL,procesadir,&args);

                // main thread waits th1 completion
                pthread_join(th1, NULL);
            }
        }

    }

    closedir(midir);

    pthread_mutex_destroy(&lock);

    cout << endl << "Highest file in current directory file :" << endl
         << grande_dir.nombre << " tam:" << grande_dir.tam
         << " bytes." << endl;          

    cout << endl << "Highest file in tree:" << endl
         << grande_arbol.nombre << " tam:" << grande_arbol.tam
         << " bytes." << endl;

return 0;
}

线程函数代码

void* procesadir(void* args)
{

thargs* myargs = reinterpret_cast<thargs*>(args);

DIR* thdir = NULL;

if ( (thdir = opendir(myargs->nextdir) ) )
{

dirent thentry;
dirent* rentry = NULL;
struct stat thstat = {0};

//while( (thentry = readdir(thdir) ) )
while( (readdir_r(thdir,&thentry,&rentry) == 0 ) && rentry  )
{

stat(thentry.d_name,&thstat);

if ( thentry.d_type == DT_REG )
{

cout << " th arch: " << thentry.d_name << " thsize: " << thstat.st_size << endl;

if ( thstat.st_size > myargs->arch->tam)
{
    pthread_mutex_lock(&lock);

    memset(myargs->arch->nombre,0,MAX_PATH);
    strncpy(myargs->arch->nombre,thentry.d_name,MAX_PATH);
    myargs->arch->tam = thstat.st_size;

     pthread_mutex_unlock(&lock);

    cout << "max tree file set to: " << thentry.d_name << " thsize: " << thstat.st_size << endl;
}


}

if ( thentry.d_type == DT_DIR )
{

if ( (strcmp(thentry.d_name,".") != 0) && (strcmp(thentry.d_name,"..") != 0 ) )
{

    thargs largs = {{0}};

    pthread_t th2;

        sprintf(largs.nextdir,"%s/%s",myargs->nextdir,thentry.d_name);
        largs.arch = myargs->arch;
        largs.pth = myargs->pth;

        // thread creation                  
        pthread_create(&th2,NULL,procesadir,&args);

        // current thread waits th2 completion
        pthread_join(th2, NULL);
    }

    cout << " th dir: " << thentry.d_name << endl;

    }   
}   


closedir(thdir);

        else
        perror("Error abriendo dir en thread:");

return 0;
}

Answer 1

我建议您检查您正在进行的 stat() 调用的返回值。

在工作线程中，您正在打印看起来不错的 thentry.d_name，但是，如果没有相对于您的工作目录的路径信息，我相信调用 stat(thentry.d_name, &thstat); 会失败。