Re: [PATCH] processes: reparent_thread: don't call kill_orphaned_pgrp() if task_detached()

[Oleg Nesterov <oleg@redhat.com>]

On 11/26, Roland McGrath wrote:
>
> > @@ -816,6 +816,8 @@ static void reparent_thread(struct task_
> >
> >  	list_move_tail(&p->sibling, &p->real_parent->children);
> >
> > +	if (task_detached(p))
> > +		return;
>
> Seems like it would be cleaner to reorganize the code a little.
> reparent_thread has only one caller.  How about we move:
>
> 	if (p->pdeath_signal)
> 		/* We already hold the tasklist_lock here.  */
> 		group_send_sig_info(p->pdeath_signal, SEND_SIG_NOINFO, p);
>
> 	list_move_tail(&p->sibling, &p->real_parent->children);
>
> into forget_original_parent and rename reparent_thread to something else,
> called only:
>
> 	if (!task_detached(p) && !same_thread_group(p->real_parent, father))
> 		orphaned_process(p);

Not that I think this really matters, but imho this code needs more
little trivial reorganizations.

reparent_thread() (or whatever) needs the "&ptrace_dead" parameter
too, if the new parent (init) ignores SIGCHLD we should release a
zombie. So we should rename "ptrace_dead" and ptrace_exit_finish().
And imho it makes sense to create the new helper which does
"list_for_each_entry_safe(father->children) {}", to make this code
more symmetrical() with ptrace_exit().

Oleg.