Monitoring Child Processes

In many application designs, a parent process needs to know when one of its child processes changes state—when the child terminates or is stopped by a signal. This chapter describes two techniques used to monitor child processes: the wait() system call (and its variants) and the use of the SIGCHLD signal.

26.1 Waiting on a Child Process

26.2 Orphans and Zombies

zombie : children die firstly orphan : parent die firstly

Using the PR_SET_PDEATHSIG operation of the Linux-specific prctl() system call, it is possible to arrange that a process receives a specified signal when it becomes an orphan

Regarding zombies, UNIX systems imitate the movies—a zombie process can’t be killed by a signal, not even the (silver bullet) SIGKILL. This ensures that the parent can always eventually perform a wait()

The parent may perform such wait() calls either synchronously, or asynchronously, in response to delivery of the SIGCHLD signal.

不會出現大量的 zombie 無法被系統回收，但是如果 parent 是一個 long lived 並且不主動回收自己產生的大量 child 那麼就糟糕了

26.3 The SIGCHLD Signal

To get around these problems, we can employ a handler for the SIGCHLD signal.

26.3.1 Establishing a Handler for SIGCHLD

The SIGCHLD signal is sent to a parent process whenever one of its children terminates. By default, this signal is ignored, but we can catch it by installing a signal handler. Within the signal handler, we can use wait() (or similar) to reap the zombie child. However, there is a subtlety to consider in this approach.

The solution is to loop inside the SIGCHLD handler, repeatedly calling waitpid() with the WNOHANG flag until there are no more dead children to be reaped. Often, the body of a SIGCHLD handler simply consists of the following code, which reaps any dead children without checking their status:

while (waitpid(-1, NULL, WNOHANG) > 0)
  continue;

The above loop continues until waitpid() returns either 0, indicating no more zombie children, or –1, indicating an error (probably ECHILD, meaning that there are no more children).

26.3.2 Delivery of SIGCHLD for Stopped Children

if the flag(SA_NOCLDSTOP) is present, SIGCHLD is not delivered for stopped children

SUSv3 also allows for a parent to be sent a SIGCHLD signal if one of its stopped children is resumed by being sent a SIGCONT signal. (This corresponds to the WCONTINUED flag for waitpid().) This feature is implemented in Linux since kernel 2.6.9.

Ignoring Dead Child Processes

Explicitly setting the disposition of SIGCHLD to SIG_IGN causes any child process that subsequently terminates to be immediately removed from the system instead of being converted into a zombie. In this case, since the status of the child process is simply discarded, a subsequent call to wait() (or similar) can’t return any information for the terminated child

26.4 Summary

Using wait() and waitpid() (and other related functions), a parent process can obtain the status of its terminated and stopped children. This status indicates whether a child process terminated normally (with an exit status indicating either success or failure), terminated abnormally, was stopped by a signal, or was resumed by a SIGCONT signal.

A common way of reaping dead child processes is to establish a handler for the SIGCHLD signal. This signal is delivered to a parent process whenever one of its children terminates, and optionally when a child is stopped by a signal. Alternatively, but somewhat less portably, a process may elect to set the disposition of SIGCHLD to SIG_IGN, in which case the status of terminated children is immediately discarded (and thus can’t later be retrieved by the parent), and the children don’t become zombies.

@todo 哪里分析过的 ?

本站所有文章转发 CSDN 将按侵权追究法律责任，其它情况随意。