exit: wait: don't use zombie->real_parent
authorOleg Nesterov <oleg@redhat.com>
Wed, 10 Dec 2014 23:54:48 +0000 (15:54 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 11 Dec 2014 01:41:17 +0000 (17:41 -0800)
commitf953ccd00615140b5e722ffe2b920da22dfb4db9
tree6c087e1e9f3c8bdd941bbdcf8d7bec472272b5a5
parentf6507f83bccd4a5f7dc7091079bf58128dc56d66
exit: wait: don't use zombie->real_parent

1. wait_task_zombie() uses p->real_parent to get psig/siglock. This is
   correct but needs tasklist_lock, ->real_parent can exit.

   We can use "current" instead. This is our natural child, its parent
   must be our sub-thread.

2. Read psig/sig outside of ->siglock, ->signal is no longer protected
   by this lock.

3. Fix the outdated comments about tasklist_lock. We can not race with
   __exit_signal(), the whole thread group is dead, nobody but us can
   call it.

   Also clarify the usage of ->stats_lock and ->siglock.

Note: thread_group_cputime_adjusted() is sub-optimal in this case, we
probably want to export cputime_adjust() to avoid thread_group_cputime().
The comment says "all threads" but there are no other threads.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Aaron Tomlin <atomlin@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Sterling Alexander <stalexan@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/exit.c