Re: [patch 0/9] mutex subsystem, -V4

[Steven Rostedt]

On Thu, 22 Dec 2005, Linus Torvalds wrote:
>
>
> On Thu, 22 Dec 2005, Thomas Gleixner wrote:
> >
> > Semaphores need a counter, mutexes only a binary representation of the
> > locked/unlocked state
>
> Actually, that's not true.
>
> A _spinlock_ only needs a binary representation of the locked/unlocked
> state.
>
> A mutex needs a _ternary_ representation. It needs an additional
> "contention" state to tell the wakeup that extra action is needed.

True, and that's exactly what Ingo has.

1 unlocked, 0 locked, -1 locked with waiters.  But it still works well
with xchg.

-- Steve

>
> If you don't handle contention (and do extra action all the time), you're
> screwed from a performance standpoint.
>
> 			Linus
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/