Re: [PATCH 3/4] reparent_thread: fix a zombie leak if /sbin/init ignores SIGCHLD

[Oleg Nesterov <oleg@redhat.com>]

On 01/29, Oleg Nesterov wrote:
>
> If /sbin/init ignores SIGCHLD and we re-parent a zombie, it is leaked.
> reparent_thread() does do_notify_parent() which sets ->exit_signal = -1
> in this case. This means that nobody except us can reap it, the detached
> task is not visible to do_wait().

Just in case, for reviewers...

To verify that the problem does exist and it is really fixed, I used the
stupid patch below, it allows to change init's SIGCHLD handler to SIG_IGN
and then restore it via prctl(1000, 0/1).	

Oleg.

--- kernel/sys.c~	2009-01-19 10:44:33.000000000 +0100
+++ kernel/sys.c	2009-01-29 07:37:09.000000000 +0100
@@ -1703,6 +1703,9 @@ SYSCALL_DEFINE1(umask, int, mask)
 	return mask;
 }
 
+void __user *I_SC;
+#include <linux/pid_namespace.h>
+
 SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
 		unsigned long, arg4, unsigned long, arg5)
 {
@@ -1716,6 +1719,17 @@ SYSCALL_DEFINE5(prctl, int, option, unsi
 
 	error = 0;
 	switch (option) {
+		case 1000: {
+			struct task_struct *i = init_pid_ns.child_reaper;
+
+			if (!I_SC) I_SC = i->sighand->action[SIGCHLD-1].sa.sa_handler;
+
+			i->sighand->action[SIGCHLD-1].sa.sa_handler =
+				arg2 ? I_SC : SIG_IGN;
+
+			break;
+		}
+
 		case PR_SET_PDEATHSIG:
 			if (!valid_signal(arg2)) {
 				error = -EINVAL;