apache分析之二、post-commit后台运行

一、问题描述

希望在svn的post-commit中执行一个后台任务，但是发现该后台任务没有退出之前，svn提交始终不会返回。按照bash派生的后台任务的定义，就是在子进程派生之后，父进程不会把终端输入(终端的前台任务)派发给后台任务，也不会同步等该该子进程的返回。在父shell退出之后，内核会把退出进程的父进程设置为系统的始祖进程，也就是1号任务。此时svn已经不再是post-commit的父进程，所以它无权wait同步等待post-commit的退出，所以它一定使用了其它的同步机制。

二、内核中对于进程退出的处理

do_exit--->>exit_notify

tatic void

forget_original_parent(struct task_struct *father, struct list_head *to_release)

{

truct task_struct *p, *reaper = father

truct list_head *_p, *_

do {

reaper = next_thread(reaper)

if (reaper == father) {

reaper = child_reaper(father)

reak

}

} while (reaper->exit_state)

tatic inline struct task_struct *child_reaper(struct task_struct *tsk)

{

return init_pid_ns.child_reaper

}

这里的意思大致是这样的，如果退出的进程是一个进程中的一个线程，那么将自己子进程托付给线程组中的另一个线程，相当于自己的兄弟线程。可惜的是这种情况在通常的我们使用的工具都不会存在，一个进程里只有一个线程，不存在兄弟线程。此时就要执行第二步、或者叫做plan B，该方法就是把自己的子进程重新设置到系统的初始化进程，也就是1号进程。

三、做一个验证

在post-commit中添加一个

leep 1234 &am

这样通过ps可以方便的找到这个任务

root@Harry hooks]# ps aux | grep slee

daemon 19691 0.0 0.0 1892 400 pts/0 S 13:20 0:00 sleep 1234

root 22089 0.0 0.0 4220 696 pts/1 S+ 13:34 0:00 grep slee

[root@Harry hooks]# cat /proc/19691/statu

Name: slee

State: S (sleeping)

Tgid: 19691

Pid: 19691

PPid: 1

TracerPid: 0

Uid: 2 2 2 2

Gid: 2 2 2 2

Utrace: 0

FDSize: 256

Groups: 1 2 4 7

可以看到，此时sleep的父进程的确已经变成了1号进程，也就是系统的始祖进程。

[root@Harry hooks]# ps aux | grep post-commit

daemon 19690 0.0 0.0 0 0 pts/0 Z 13:20 0:00 [post-commit] <defunct>

root 24274 0.0 0.0 4220 696 pts/1 S+ 13:47 0:00 grep post-commit

此时它的父进程post-commit已经退出，成为僵尸进程。

四、此时apache的状态

同样是gdb附加到进程上，执行各个线程显示

(gdb) bt

#0 0x00cad424 in __kernel_vsyscall ()

#1 0x002e78f6 in epoll_wait () from /lib/libc.so.6

#2 0x009b8bd9 in impl_pollset_poll (pollset=0x99bce70, timeout=100,

um=0xa71e6284, descriptors=0xa71e6288) at poll/unix/epoll.c:256

#3 0x009ba574 in apr_pollset_poll (pollset=0x99bce70, timeout=100000,

um=0xa71e6284, descriptors=0xa71e6288) at poll/unix/pollset.c:343

#4 0x080aca7c in listener_thread (thd=0x99bd590, dummy=0xb6c014c0)

at event.c:1392

#5 0x009bfe0d in dummy_worker (opaque=0x99bd590) at threadproc/unix/thread.c:142

#6 0x00391925 in start_thread () from /lib/libpthread.so.0

#7 0x002e707e in clone () from /lib/libc.so.6

(gdb) info thread

18 Thread 0xb11f6b70 (LWP 6221) 0x00cad424 in __kernel_vsyscall ()

* 2 Thread 0xa71e6b70 (LWP 6237) 0x00cad424 in __kernel_vsyscall ()

1 Thread 0xb77bd9c0 (LWP 6210) 0x00cad424 in __kernel_vsyscall ()

(gdb) thread 1

[Switching to thread 1 (Thread 0xb77bd9c0 (LWP 6210))]#0 0x00cad424 in __kernel_vsyscall ()

(gdb) bt

#0 0x00cad424 in __kernel_vsyscall ()

#1 0x002df451 in select () from /lib/libc.so.6

#2 0x009c0baf in apr_sleep (t=500000) at time/unix/time.c:246

#3 0x080ae2fb in join_workers (listener=0x99bd590, threads=0x9a954c0)

at event.c:1959

#4 0x080ae895 in child_main (child_num_arg=0) at event.c:2109

#5 0x080ae9b8 in make_child (s=0x997cea8, slot=0) at event.c:2169

#6 0x080aeb02 in startup_children (number_to_start=3) at event.c:2233

#7 0x080af475 in event_run (_pconf=0x99580a8, plog=0x99b8aa0, s=0x997cea8)

at event.c:2561

#8 0x0806ef8f in ap_run_mpm (pconf=0x99580a8, plog=0x99b8aa0, s=0x997cea8)

at mpm_common.c:98

#9 0x08068eef in main (argc=2, argv=0xbfab1d24) at main.c:777

(gdb) thread 18

[Switching to thread 18 (Thread 0xb11f6b70 (LWP 6221))]#0 0x00cad424 in __kernel_vsyscall ()

(gdb) bt

#0 0x00cad424 in __kernel_vsyscall ()

#1 0x00398eeb in read () from /lib/libpthread.so.0

#2 0x009ad41e in apr_file_read (thefile=0x9acff20, buf=0xb6d50020,

ytes=0xb11f5e18) at file_io/unix/readwrite.c:116

#3 0x0043da48 in svn_io_file_read (file=0x9acff20, buf=0xb6d50020,

ytes=0xb11f5e18, pool=0x9aca898) at subversion/libsvn_subr/io.c:3132

#4 0x0043b324 in stringbuf_from_aprfile (result=0xb11f5e9c, filename=0x0,

file=0x9acff20, check_size=1, pool=0x9aca898)

at subversion/libsvn_subr/io.c:2049

#5 0x0043b633 in svn_stringbuf_from_aprfile (result=0xb11f5e9c, file=0x9acff20,

ool=0x9aca898) at subversion/libsvn_subr/io.c:2106

#6 0x00bf72a9 in check_hook_result (name=0xc10126 "post-commit",

cmd=0x9acfd60 "/svnrepo/tsecer/hooks/post-commit", cmd_proc=0xb11f5f10,

read_errhandle=0x9acff20, pool=0x9aca898)

at subversion/libsvn_repos/hooks.c:71

#7 0x00bf77ef in run_hook_cmd (result=0x0, name=0xc10126 "post-commit",

cmd=0x9acfd60 "/svnrepo/tsecer/hooks/post-commit", args=0xb11f5f64,

tdin_handle=0x0, pool=0x9aca898) at subversion/libsvn_repos/hooks.c:211

#8 0x00bf81df in svn_repos__hooks_post_commit (repos=0xb6c032c8, rev=1,

ool=0x9aca898) at subversion/libsvn_repos/hooks.c:469

#9 0x00bf577f in svn_repos_fs_commit_txn (conflict_p=0xb11f6000,

repos=0xb6c032c8, new_rev=0xb11f5ffc, txn=0x9acc520, pool=0x9aca898)

at subversion/libsvn_repos/fs-wrap.c:64

---Type <return> to continue, or q <return> to quit---

#10 0x003c899e in merge (target=0x9aca338, source=0x9aa4948, no_auto_merge=1,

o_checkout=1, prop_elem=0x9aca0a0, output=0x9acb470)

at subversion/mod_dav_svn/version.c:1426

#11 0x00175f82 in dav_method_merge (r=0x9aca8d8) at mod_dav.c:4399

#12 0x00176a36 in dav_handler (r=0x9aca8d8) at mod_dav.c:4778

#13 0x0808aeda in ap_run_handler (r=0x9aca8d8) at config.c:169

#14 0x0808b5f2 in ap_invoke_handler (r=0x9aca8d8) at config.c:432

#15 0x080a2c68 in ap_process_async_request (r=0x9aca8d8) at http_request.c:317

#16 0x0809f3cf in ap_process_http_async_connection (c=0xb6c01c00)

at http_core.c:143

#17 0x0809f5a2 in ap_process_http_connection (c=0xb6c01c00) at http_core.c:228

#18 0x08095ee3 in ap_run_process_connection (c=0xb6c01c00) at connection.c:41

#19 0x080ab9ef in process_socket (thd=0x99bd390, p=0xb6c019c8, sock=0xb6c01a10,

cs=0xb6c01bb8, my_child_num=0, my_thread_num=9) at event.c:917

#20 0x080adb14 in worker_thread (thd=0x99bd390, dummy=0xb6c00a40) at event.c:1744

#21 0x009bfe0d in dummy_worker (opaque=0x99bd390) at threadproc/unix/thread.c:142

#22 0x00391925 in start_thread () from /lib/libpthread.so.0

#23 0x002e707e in clone () from /lib/libc.so.6

(gdb) frame 2

#2 0x009ad41e in apr_file_read (thefile=0x9acff20, buf=0xb6d50020,

ytes=0xb11f5e18) at file_io/unix/readwrite.c:116

116 rv = read(thefile->filedes, buf, *nbytes)

(gdb) p *thefile

$2 = {pool = 0x9aca898, filedes = 12, fname = 0x0, flags = 0, eof_hit = 0,

is_pipe = 1, timeout = -1, buffered = 0, blocking = BLK_ON, ungetchar = -1,

uffer = 0x0, bufpos = 0, bufsize = 0, dataRead = 0, direction = 0,

filePtr = 0, thlock = 0x0}

(gdb) shell ls /proc/6210/fd/12

/proc/6210/fd/12

(gdb) shell ls /proc/6210/fd/12 -l

lr-x------. 1 root root 64 2012-10-14 13:34 /proc/6210/fd/12 -> pipe:[4017708]

(gdb)

[root@Harry hooks]# ll /proc/19691/fd

total 0

lr-x------. 1 daemon daemon 64 2012-10-14 13:35 0 -> /dev/null

l-wx------. 1 daemon daemon 64 2012-10-14 13:35 1 -> /dev/null

l-wx------. 1 daemon daemon 64 2012-10-14 13:20 2 -> pipe:[4017708]

可以看到，apache的父进程在等待子进程的标准错误结束，如果子进程的标准错误一直没有关闭，那么父进程将会一直等待。

五、内核中关于文件进程和关闭

当派生一个新的线程时，此时子进程会增加文件的引用计数。

copy_process--->>>copy_files--->>>dup_fd

for (i = open_files; i != 0; i--) {

truct file *f = *old_fds++

if (f) {

get_file(f)

}

其中

#define get_file(x) atomic_inc(&(x)->f_count)

只是增加了文件的引用计数。

当进程退出关闭一个文件的时候，执行操作为

void fastcall fput(struct file *file)

{

if (atomic_dec_and_test(&file->f_count))

__fput(file)

}

此时由于sleep进程是打开了pipe文件的一个进程，所以它也会导致所有的文件关闭的时候也无法将pipe释放，即使它的父进程已经变成init进程。这也是为什么需要在svn的post-commit中将后台任务的标准错误也重定向到另外一个文件的原因。只有当管道关闭之后，read系统调用才会返回EOF。