Re: [patch] PI-futex: -V2

[Ingo Molnar <mingo@elte.hu>]

* Andrew Morton <akpm@osdl.org> wrote:

> Ingo Molnar <mingo@elte.hu> wrote:
> >
> > > > +	for (;;) {
> >  > > +		if (top_waiter)
> >  > > +			plist_del(&top_waiter->pi_list_entry,
> >  > > +				  &owner->pi_waiters);
> >  > > +
> >  > > +		if (waiter && waiter == rt_mutex_top_waiter(lock)) {
> >  > 
> >  > rt_mutex_top_waiter() can never return NULL, so the test for NULL 
> >  > could be removed.
> > 
> >  it might be NULL if adjust_pi_chain() is called from remove_waiter(), 
> >  and next_waiter there is NULL (because !rt_mutex_has_waiters() after the 
> >  removal of the current waiter).
> 
> Yes, `waiter' might be NULL.  But rt_mutex_top_waiter() will never 
> return NULL.  So it might be possible to just do
> 
> 	if (waiter == rt_mutex_top_waiter(lock))

ah, indeed, you are right.

> Which might actually be less efficient, and more obscure.  Just 
> pointing it out.

ok, i left it as-is for now.

	Ingo