Re: [PATCH 1/2] signal: simplify deadlock-avoidance in lock_task_sighand()

[Peter Zijlstra <peterz@infradead.org>]

On Mon, Sep 22, 2014 at 06:44:37PM +0200, Oleg Nesterov wrote:
> __lock_task_sighand() does local_irq_save() to prevent the potential
> deadlock, we can use preempt_disable() with the same effect. And in
> this case we can do preempt_disable/enable + rcu_read_lock/unlock only
> once outside of the main loop and simplify the code. This also shaves
> 112 bytes from signal.o.
> 
> With this patch the main loop runs with preemption disabled, but this
> should be fine because restart is very unlikely: it can only happen if
> we race with de_thread() and ->sighand is shared. And the latter is only
> possible if CLONE_SIGHAND was used without CLONE_THREAD, most probably
> nobody does this nowadays.

If its unlikely to repeat, it shouldn't matter either way and we can
keep the preempt_disable() inside the loop to guarantee better worst
case behaviour.

And I suppose we can fix -rt, like Steve already mentioned, there's the
preempt_disable_nort() thing for just such cases.