[PATCH] coredump: don't take tasklist_lock
authorOleg Nesterov <oleg@tv-sign.ru>
Mon, 26 Jun 2006 07:26:08 +0000 (00:26 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Mon, 26 Jun 2006 16:58:27 +0000 (09:58 -0700)
commit7b1c6154fa8bb937e0b1b4f2adbb315d70270f10
tree1d58e5e7da61330639ebe62671fb666596d1b00a
parentd5f70c00ad24cd1158d3678b44ff969b4c971d49
[PATCH] coredump: don't take tasklist_lock

This patch removes tasklist_lock from zap_threads().
This is safe wrt:

do_exit:
The caller holds mm->mmap_sem. This means that task which
shares the same ->mm can't pass exit_mm(), so it can't be
unhashed from init_task.tasks or ->thread_group lists.

fork:
None of sub-threads can fork after zap_process(leader). All
processes which were created before this point should be
visible to zap_threads() because copy_process() adds the new
process to the tail of init_task.tasks list, and ->siglock
lock/unlock provides a memory barrier.

de_thread:
It does list_replace_rcu(&leader->tasks, &current->tasks).
So zap_threads() will see either old or new leader, it does
not matter. However, it can change p->sighand, so we should
use lock_task_sighand() in zap_process().

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