diff options
author | Oleg Nesterov <oleg@tv-sign.ru> | 2007-10-17 02:26:49 -0400 |
---|---|---|
committer | Linus Torvalds <torvalds@woody.linux-foundation.org> | 2007-10-17 11:42:51 -0400 |
commit | 84eb646b6eabcd82ec563f30d2d9d40c2054a9c9 (patch) | |
tree | 96219c5844c80c40cbc9814a13a9a72de7430fc7 | |
parent | 8e9073ed027771bcdee4033eb900a3c09ac90a19 (diff) |
handle the multi-threaded init's exit() properly
With or without this patch, multi-threaded init's are not fully supported,
but do_exit() is completely wrong. This becomes a real problem when we
support pid namespaces.
1. do_exit() panics when the main thread of /sbin/init exits. It should not
until the whole thread group exits. Move the code below, under the
"if (group_dead)" check.
Note: this means that forget_original_parent() can use an already dead
child_reaper()'s task_struct. This is OK for /sbin/init because
- do_wait() from alive sub-thread still can reap a zombie, we iterate
over all sub-thread's ->children lists
- do_notify_parent() will wakeup some alive sub-thread because it sends
the group-wide signal
However, we should remove choose_new_parent()->BUG_ON(reaper->exit_state)
for this.
2. We are playing games with ->nsproxy->pid_ns. This code is bogus today, and
it has to be changed anyway when we really support pid namespaces, just
remove it.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Roland McGrath <roland@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-rw-r--r-- | kernel/exit.c | 31 |
1 files changed, 11 insertions, 20 deletions
diff --git a/kernel/exit.c b/kernel/exit.c index 3b8dfffd9329..8b64c0371ae9 100644 --- a/kernel/exit.c +++ b/kernel/exit.c | |||
@@ -592,17 +592,6 @@ static void exit_mm(struct task_struct * tsk) | |||
592 | mmput(mm); | 592 | mmput(mm); |
593 | } | 593 | } |
594 | 594 | ||
595 | static inline void | ||
596 | choose_new_parent(struct task_struct *p, struct task_struct *reaper) | ||
597 | { | ||
598 | /* | ||
599 | * Make sure we're not reparenting to ourselves and that | ||
600 | * the parent is not a zombie. | ||
601 | */ | ||
602 | BUG_ON(p == reaper || reaper->exit_state); | ||
603 | p->real_parent = reaper; | ||
604 | } | ||
605 | |||
606 | static void | 595 | static void |
607 | reparent_thread(struct task_struct *p, struct task_struct *father, int traced) | 596 | reparent_thread(struct task_struct *p, struct task_struct *father, int traced) |
608 | { | 597 | { |
@@ -710,7 +699,7 @@ forget_original_parent(struct task_struct *father, struct list_head *to_release) | |||
710 | 699 | ||
711 | if (father == p->real_parent) { | 700 | if (father == p->real_parent) { |
712 | /* reparent with a reaper, real father it's us */ | 701 | /* reparent with a reaper, real father it's us */ |
713 | choose_new_parent(p, reaper); | 702 | p->real_parent = reaper; |
714 | reparent_thread(p, father, 0); | 703 | reparent_thread(p, father, 0); |
715 | } else { | 704 | } else { |
716 | /* reparent ptraced task to its real parent */ | 705 | /* reparent ptraced task to its real parent */ |
@@ -731,7 +720,7 @@ forget_original_parent(struct task_struct *father, struct list_head *to_release) | |||
731 | } | 720 | } |
732 | list_for_each_safe(_p, _n, &father->ptrace_children) { | 721 | list_for_each_safe(_p, _n, &father->ptrace_children) { |
733 | p = list_entry(_p, struct task_struct, ptrace_list); | 722 | p = list_entry(_p, struct task_struct, ptrace_list); |
734 | choose_new_parent(p, reaper); | 723 | p->real_parent = reaper; |
735 | reparent_thread(p, father, 1); | 724 | reparent_thread(p, father, 1); |
736 | } | 725 | } |
737 | } | 726 | } |
@@ -882,6 +871,14 @@ static void check_stack_usage(void) | |||
882 | static inline void check_stack_usage(void) {} | 871 | static inline void check_stack_usage(void) {} |
883 | #endif | 872 | #endif |
884 | 873 | ||
874 | static inline void exit_child_reaper(struct task_struct *tsk) | ||
875 | { | ||
876 | if (likely(tsk->group_leader != child_reaper(tsk))) | ||
877 | return; | ||
878 | |||
879 | panic("Attempted to kill init!"); | ||
880 | } | ||
881 | |||
885 | fastcall NORET_TYPE void do_exit(long code) | 882 | fastcall NORET_TYPE void do_exit(long code) |
886 | { | 883 | { |
887 | struct task_struct *tsk = current; | 884 | struct task_struct *tsk = current; |
@@ -895,13 +892,6 @@ fastcall NORET_TYPE void do_exit(long code) | |||
895 | panic("Aiee, killing interrupt handler!"); | 892 | panic("Aiee, killing interrupt handler!"); |
896 | if (unlikely(!tsk->pid)) | 893 | if (unlikely(!tsk->pid)) |
897 | panic("Attempted to kill the idle task!"); | 894 | panic("Attempted to kill the idle task!"); |
898 | if (unlikely(tsk == child_reaper(tsk))) { | ||
899 | if (tsk->nsproxy->pid_ns != &init_pid_ns) | ||
900 | tsk->nsproxy->pid_ns->child_reaper = init_pid_ns.child_reaper; | ||
901 | else | ||
902 | panic("Attempted to kill init!"); | ||
903 | } | ||
904 | |||
905 | 895 | ||
906 | if (unlikely(current->ptrace & PT_TRACE_EXIT)) { | 896 | if (unlikely(current->ptrace & PT_TRACE_EXIT)) { |
907 | current->ptrace_message = code; | 897 | current->ptrace_message = code; |
@@ -951,6 +941,7 @@ fastcall NORET_TYPE void do_exit(long code) | |||
951 | } | 941 | } |
952 | group_dead = atomic_dec_and_test(&tsk->signal->live); | 942 | group_dead = atomic_dec_and_test(&tsk->signal->live); |
953 | if (group_dead) { | 943 | if (group_dead) { |
944 | exit_child_reaper(tsk); | ||
954 | hrtimer_cancel(&tsk->signal->real_timer); | 945 | hrtimer_cancel(&tsk->signal->real_timer); |
955 | exit_itimers(tsk->signal); | 946 | exit_itimers(tsk->signal); |
956 | } | 947 | } |