[PATCH] exec: allow init to exec from any thread.

After looking at the problem of init calling exec some more I figured out
an easy way to make the code work.

The actual symptom without out this patch is that all threads will die
except pid == 1, and the thread calling exec.  The thread calling exec will
wait forever for pid == 1 to die.

Since pid == 1 does not install a handler for SIGKILL it will never die.

This modifies the tests for init from current->pid == 1 to the equivalent
current == child_reaper.  And then it causes exec in the ugly case to
modify child_reaper.

The only weird symptom is that you wind up with an init process that
doesn't have the oldest start time on the box.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 12 insertions, 1 deletions

diff --git a/fs/exec.c b/fs/exec.c
index c7397c4..d0ecea0 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -660,12 +660,23 @@ static int de_thread(struct task_struct *tsk)
 		struct dentry *proc_dentry1, *proc_dentry2;
 		unsigned long ptrace;
 
+		leader = current->group_leader;
+		/*
+		 * If our leader is the child_reaper become
+		 * the child_reaper and resend SIGKILL signal.
+		 */
+		if (unlikely(leader == child_reaper)) {
+			write_lock(&tasklist_lock);
+			child_reaper = current;
+			zap_other_threads(current);
+			write_unlock(&tasklist_lock);
+		}
+
 		/*
 		 * Wait for the thread group leader to be a zombie.
 		 * It should already be zombie at this point, most
 		 * of the time.
 		 */
-		leader = current->group_leader;
 		while (leader->exit_state != EXIT_ZOMBIE)
 			yield();