signals: Fix more rcu assumptions
authorThomas Gleixner <tglx@linutronix.de>
Thu, 10 Dec 2009 00:53:21 +0000 (00:53 +0000)
committerThomas Gleixner <tglx@linutronix.de>
Thu, 10 Dec 2009 22:04:11 +0000 (23:04 +0100)
1) Remove the misleading comment in __sigqueue_alloc() which claims
   that holding a spinlock is equivalent to rcu_read_lock().

2) Add a rcu_read_lock/unlock around the __task_cred() access
   in __sigqueue_alloc()

This needs to be revisited to remove the remaining users of
read_lock(&tasklist_lock) but that's outside the scope of this patch.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <20091210004703.269843657@linutronix.de>


No differences found