From: Oleg Nesterov <oleg@xxxxxxxxxx> If some system uses multi-threaded init the kernel panics when the main thread exits. Perhaps this is just a "quality of implementation" question. sys_exit() from the main thread should be OK, why /sbin/init should be special? That said, I personally do not think that multi-threaded init is terribly useful. So I think the patch below makes sense for now. Note that it removes the games with pid_ns->child_reaper: this doesn't work currently, and this has to be modified when we actually support pid namespaces anyway. Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx> Acked-by: Pavel Emelyanov <xemul@xxxxxxxxxx> --- exit.c | 17 +++++++++-------- 1 files changed, 9 insertions(+), 8 deletions(-) --- ./kernel/exit.c.ve3 2007-08-06 12:14:53.000000000 +0400 +++ ./kernel/exit.c 2007-08-06 12:15:19.000000000 +0400 @@ -895,6 +895,14 @@ static void check_stack_usage(void) static inline void check_stack_usage(void) {} #endif +static inline void exit_child_reaper(struct task_struct *tsk) +{ + if (likely(tsk->group_leader != task_child_reaper(tsk))) + return; + + panic("Attempted to kill init!"); +} + fastcall NORET_TYPE void do_exit(long code) { struct task_struct *tsk = current; @@ -908,14 +916,6 @@ fastcall NORET_TYPE void do_exit(long co panic("Aiee, killing interrupt handler!"); if (unlikely(!tsk->pid)) panic("Attempted to kill the idle task!"); - if (unlikely(tsk == task_child_reaper(tsk))) { - if (task_active_pid_ns(tsk) != &init_pid_ns) - task_active_pid_ns(tsk)->child_reaper = - init_pid_ns.child_reaper; - else - panic("Attempted to kill init!"); - } - if (unlikely(current->ptrace & PT_TRACE_EXIT)) { current->ptrace_message = code; @@ -965,6 +965,7 @@ fastcall NORET_TYPE void do_exit(long co } group_dead = atomic_dec_and_test(&tsk->signal->live); if (group_dead) { + exit_child_reaper(tsk); hrtimer_cancel(&tsk->signal->real_timer); exit_itimers(tsk->signal); } _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers