c - 使用 fork() 计算接收到的命令行参数的总和时出现问题

Question

我正在尝试根据从命令行收到的数字集来计算总和，并且我使用一个名为 worker 的配套程序来为我计算。如果收到的数字数量是奇数，它会在数字数量上加一个零，使集合成为偶数。

这是一个易于理解的程序流程(归功于 Alok):

举个例子会更清楚:

假设您要添加 7 个数字:1 2 3 4 5 6 7

./协调员 1 2 3 4 5 6 7

1. 输入 = [1 2 3 4 5 6 7 0]
2. n = len(输入) = 8
3. m = n/2 = 4
4. 输出 = [0 0 0 0]
5. 我们 fork 4 个进程，第一个进程获得 [1 2]，第二个进程获得 [3 4]，...
6. 4个进程分别返回3、7、11、7，我们分配给output。
7. 输出有 4 个元素，所以我们为新输入分配 4+1 = 5 个元素的空间。
8. 设置输入 = [3 7 11 7 0]
9. n = len(输入) = 5
10. m = n/2 = 2
11. 输出= [0 0]
12. 我们 fork 2 个进程，第一个得到 [3 7]，第二个得到 [11 7]
13. 这 2 个进程返回 10、18，我们将其分配给输出。
14. 输出有 2 个元素，所以我们为新输入分配 2+1 = 3 个元素的空间。
15. 设置输入 = [10 18 0]
16. n = len(输入) = 3
17. m = n/2 = 1
18. 输出= [0]
19. 我们 fork 一个进程，它得到 [10 18]
20. 该过程返回 28，我们将其分配给输出。
21. 输出有 1 个元素，所以我们完成了。

虽然在这组特定的数字上我得到:

Process ID: 15195 
Sum of 1 and 2 is 3 

Process ID: 15196 
Sum of 3 and 4 is 7 

Process ID: 15197 
Sum of 5 and 6 is 11 

Process ID: 15198 
Sum of 7 and 0 is 7 

*** glibc detected *** ./coordinator: free(): invalid next size (fast): 0x080ec048 ***

后跟堆错误列表。

我相信我没有正确地重新分配指针的大小，我在第一次调用 next_step() 后尝试将旧输出重定向到新输入。所以它试图将数据放入没有空间的内存部分。

更新:

@诺曼

这是我收到的输出:

==3585== Memcheck, a memory error detector
==3585== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al.
==3585== Using Valgrind-3.5.0 and LibVEX; rerun with -h for copyright info
==3585== Command: ./coordinator 1 2 3 4
==3585== 
calc: 2:
input[0]: 1
input[1]: 2
input[2]: 3
input[3]: 4
==3585== Use of uninitialised value of size 4
==3585==    at 0x4076186: ??? (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x4079A81: vfprintf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x4080F7F: printf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x8048833: main (in /home/bryan/cpp/coordinator)
==3585== 
==3585== Conditional jump or move depends on uninitialised value(s)
==3585==    at 0x407618E: ??? (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x4079A81: vfprintf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x4080F7F: printf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x8048833: main (in /home/bryan/cpp/coordinator)
==3585== 
==3585== Conditional jump or move depends on uninitialised value(s)
==3585==    at 0x4077877: vfprintf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x4080F7F: printf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x8048833: main (in /home/bryan/cpp/coordinator)
==3585== 
==3585== Conditional jump or move depends on uninitialised value(s)
==3585==    at 0x407789B: vfprintf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x4080F7F: printf (in /lib/tls/i686/cmov/libc-2.10.1.so)
==3585==    by 0x8048833: main (in /home/bryan/cpp/coordinator)
==3585== 
input[4]: 0
==3586== Memcheck, a memory error detector
==3586== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al.
==3586== Using Valgrind-3.5.0 and LibVEX; rerun with -h for copyright info
==3586== Command: ./worker 1 2
==3586== 
Process ID: 3586 
Sum of 1 and 2 is 3 

==3586== 
==3586== HEAP SUMMARY:
==3586==     in use at exit: 0 bytes in 0 blocks
==3586==   total heap usage: 0 allocs, 0 frees, 0 bytes allocated
==3586== 
==3586== All heap blocks were freed -- no leaks are possible
==3586== 
==3586== For counts of detected and suppressed errors, rerun with: -v
==3586== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 11 from 6)
==3587== Memcheck, a memory error detector
==3587== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al.
==3587== Using Valgrind-3.5.0 and LibVEX; rerun with -h for copyright info
==3587== Command: ./worker 3 4
==3587== 
Process ID: 3587 
Sum of 3 and 4 is 7 

==3587== 
==3587== HEAP SUMMARY:
==3587==     in use at exit: 0 bytes in 0 blocks
==3587==   total heap usage: 0 allocs, 0 frees, 0 bytes allocated
==3587== 
==3587== All heap blocks were freed -- no leaks are possible
==3587== 
==3587== For counts of detected and suppressed errors, rerun with: -v
==3587== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 11 from 6)
==3585== Invalid write of size 4
==3585==    at 0x8048A3A: main (in /home/bryan/cpp/coordinator)
==3585==  Address 0x417f0b4 is 8 bytes after a block of size 4 alloc'd
==3585==    at 0x4024C6C: malloc (vg_replace_malloc.c:195)
==3585==    by 0x4024CF6: realloc (vg_replace_malloc.c:476)
==3585==    by 0x8048A25: main (in /home/bryan/cpp/coordinator)
==3585== 
==3588== Memcheck, a memory error detector
==3588== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al.
==3588== Using Valgrind-3.5.0 and LibVEX; rerun with -h for copyright info
==3588== Command: ./worker 3 7
==3588== 
Process ID: 3588 
Sum of 3 and 7 is 10 

==3588== 
==3588== HEAP SUMMARY:
==3588==     in use at exit: 0 bytes in 0 blocks
==3588==   total heap usage: 0 allocs, 0 frees, 0 bytes allocated
==3588== 
==3588== All heap blocks were freed -- no leaks are possible
==3588== 
==3588== For counts of detected and suppressed errors, rerun with: -v
==3588== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 11 from 6)
==3585== Invalid read of size 4
==3585==    at 0x8048AB5: main (in /home/bryan/cpp/coordinator)
==3585==  Address 0x417f0e0 is 0 bytes after a block of size 0 alloc'd
==3585==    at 0x4024C6C: malloc (vg_replace_malloc.c:195)
==3585==    by 0x4024CF6: realloc (vg_replace_malloc.c:476)
==3585==    by 0x8048A77: main (in /home/bryan/cpp/coordinator)
==3585== 
The final sum is: 0==3585== 
==3585== HEAP SUMMARY:
==3585==     in use at exit: 28 bytes in 2 blocks
==3585==   total heap usage: 4 allocs, 2 frees, 32 bytes allocated
==3585== 
==3585== LEAK SUMMARY:
==3585==    definitely lost: 8 bytes in 1 blocks
==3585==    indirectly lost: 0 bytes in 0 blocks
==3585==      possibly lost: 20 bytes in 1 blocks
==3585==    still reachable: 0 bytes in 0 blocks
==3585==         suppressed: 0 bytes in 0 blocks
==3585== Rerun with --leak-check=full to see details of leaked memory
==3585== 
==3585== For counts of detected and suppressed errors, rerun with: -v
==3585== Use --track-origins=yes to see where uninitialised values come from
==3585== ERROR SUMMARY: 6 errors from 6 contexts (suppressed: 11 from 6)

Answer 1

你应该考虑编写一个函数，我们称它为step_once()，它将采用input 和n 个数字，并写入到相应的 output 与 m = n/2 元素。上面的n是输入数字的个数+1，input的最后一个元素等于0。

在您的驱动程序函数中，假设main():如果output 包含1 个数字，您就完成了。否则，您重新分配 input 以包含 n_new = m+1 元素，重新分配 output 以包含 m_new = n_new/2 元素，并再次调用函数 step_once()。你一直这样做，直到你得到一个数字:

function next_step(input, output, n, m):
    n := number of input numbers # this is 1 greater than
                                 # the number of numbers being summed
    m := n / 2 # C division
    n_children := m
    i := 0
    while i < m:
        fork worker with input[2*i] and input[2*i+1]
        get result in output[i]
        i := i + 1

function main:
    set n := length(input) + 1
    set m := n/2
    allocate memory for input # n+1 elements, last = 0
    allocate memory for output # m elements
    set values in input
    while True:
        next_step(input, output, n, m)
        if length or output == 1:
             done, return
        else:
            set n := length(output) + 1
            set m := n/2
            allocate space for new_input # n elements
            set new_input := output + [0]
            free input and output
            set input := new_input
            allocate memory for output # m elements

优点是您可以测试您的 next_step() 函数以确保其正常工作，从而使调试更容易。

举个例子会更清楚:

假设您要添加 7 个数字:1 2 3 4 5 6 7

1. 输入 = [1 2 3 4 5 6 7 0]
2. n = len(input) = 8
3. m = n/2 = 4
4. 输出 = [0 0 0 0]
5. 我们 fork 4 个进程，第一个进程获得 [1 2]，第二个进程获得 [3 4]，...
6. 4个进程分别返回3、7、11、7，我们分配给output。
7. output 有 4 个元素，因此我们为新的 input 分配 4+1 = 5 个元素的空间。
8. 设置输入 = [3 7 11 7 0]
9. n = len(input) = 5
10. m = n/2 = 2
11. 输出 = [0 0]
12. 我们 fork 2 个进程，第一个得到 [3 7]，第二个得到 [11 7]
13. 这 2 个进程返回 10、18，我们将其分配给 output。
14. output 有 2 个元素，因此我们为新的 input 分配 2+1 = 3 个元素的空间。
15. 设置输入 = [10 18 0]
16. n = len(input) = 3
17. m = n/2 = 1
18. 输出 = [0]
19. 我们 fork 一个进程，它得到 [10 18]
20. 该过程返回 28，我们将其分配给 output。
21. 输出有 1 个元素，所以我们完成了。