Re: [PATCH] proc: Disable /proc/$pid/wchan

[Josh Poimboeuf]

On Wed, Sep 29, 2021 at 09:40:26PM +0200, Peter Zijlstra wrote:
> On Wed, Sep 29, 2021 at 11:54:55AM -0700, Kees Cook wrote:
> 
> > > > > > > It's supposed to show where a blocked task is blocked; the "wait
> > > > > > > channel".
> 
> > Since I think we're considering get_wchan() to be slow-path, can we just
> > lock the runqueue and use arch_stack_walk_reliable()?
> 
> Funny thing, when a task is blocked it isn't on the runqueue :-)
> 
> So if all we want to do is capture a blocked task and fail otherwise we
> don't need the rq->lock at all.
> 
> Something like:
> 
> 	unsigned long ip = 0;
> 
> 	raw_spin_lock_irq(&p->pi_lock);
> 	state = READ_ONCE(p->__state);
> 	smp_rmb(); /* see try_to_wake_up() */
> 	if (state == TASK_RUNNING || state == TASK_WAKING || p->on_rq)
> 		goto unlock;
> 
> 	ip = /* do actual stack walk on a blocked task */
> unlock:
> 	raw_spin_unlock_irq(&p->pi_lock);
> 
> 	return ip;

Ah, cool :-)

I'd also add that I don't see any reason to use the "reliable" unwinding
variant.  AFAIK, just basic stack_trace_save_tsk() should be sufficient.

-- 
Josh