coredump: don't take tasklist_lock
Subsystems (2)
PATCH 4/4 coredump: don't take tasklist_lock
depends on
"[PATCH rc1-mm] de_thread: fix deadlockable process addition"
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, ¤t->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 <[email protected]>
fs/exec.c renamed from fs/exec.c~4_TLOCK +8 -4
@@ -1385,7 +1385,11 @@ static void zap_process(struct task_strustruct task_struct *t;unsigned long flags; -spin_lock_irqsave (&start->sighand->siglock, flags); +/* +* start->sighand can't disappear, but may be +* changed by de_thread() +*/ +lock_task_sighand (start, &flags); start->signal->flags =SIGNAL_GROUP_EXIT ; start->signal->group_stop_count =0 ;
@@ -1398,7 +1402,7 @@ static void zap_process(struct task_stru } }while ((t =next_thread (t)) != start); -spin_unlock_irqrestore (&start->sighand->siglock, flags); +unlock_task_sighand (start, &flags); }static void zap_threads (struct mm_struct *mm)
@@ -1416,7 +1420,7 @@ static void zap_threads(struct mm_structcomplete (vfork_done); } -read_lock (&tasklist_lock); +rcu_read_lock ();for_each_process (g) { p = g; do {
@@ -1427,7 +1431,7 @@ static void zap_threads(struct mm_struct } }while ((p =next_thread (p)) != g); } -read_unlock (&tasklist_lock); +rcu_read_unlock (); }static void coredump_wait (struct mm_struct *mm)
Quoted 1 line
This patch removes
tasklist_lockfromzap_threads().
That's a very worthwhile thing to do, and this change looks good to me. I have not 100% followed the recent related de_thread discussion, but certainly issues there should be resolved such that this change can be done.
Thanks,
Roland