Re: [PATCH] audit: Call tty_audit_push_task() outside preempt disabled region

[Thomas Gleixner <tglx@linutronix.de>]

On Tue, 1 Dec 2009, Oleg Nesterov wrote:
> On 12/01, Thomas Gleixner wrote:
> >
> > -void tty_audit_push_task(struct task_struct *tsk, uid_t loginuid, u32 sessionid)
> > +int tty_audit_push_task(struct task_struct *tsk, uid_t loginuid, u32 sessionid)
> >  {
> > -	struct tty_audit_buf *buf;
> > +	struct tty_audit_buf *buf = NULL;
> > +	unsigned long flags;
> >
> > -	spin_lock_irq(&tsk->sighand->siglock);
> > -	buf = tsk->signal->tty_audit_buf;
> > -	if (buf)
> > +	if (!lock_task_sighand(tsk, &flags))
> > +		return -ESRCH;
> > +
> > +	if (tsk->signal->audit_tty && tsk->signal->tty_audit_buf) {
> > +		buf = tsk->signal->tty_audit_buf;
> >  		atomic_inc(&buf->count);
> > -	spin_unlock_irq(&tsk->sighand->siglock);
> > +	}
> > +	unlock_task_sighand(tsk, &flags);
> > +
> >  	if (!buf)
> > -		return;
> > +		return -EPERM;
> 
> I think the patch is correct, but it changes the behaviour of
> audit_prepare_user_tty() a bit.
> 
> Suppose that signal->audit_tty != NULL but signal->tty_audit_buf
> is not allocated yet. In this audit_prepare_user_tty() returns 0
> before the patch and -EPERM after.
> 
> I do not know if this matters, just to be sure this is OK.

Hmm, true. Missed that. Thanks for catching it.

     tglx