Re: [PATCH] 2.6.16 - futex: small optimization (?)

From: Bill Davidsen
Date: Thu Mar 30 2006 - 15:22:25 EST


Ingo Molnar wrote:
* Bill Davidsen <davidsen@xxxxxxx> wrote:

Ulrich Drepper wrote:
On 3/27/06, Pierre PEIFFER <pierre.peiffer@xxxxxxxx> wrote:
I found a (optimization ?) problem in the futexes, during a futex_wake,
if the waiter has a higher priority than the waker.
There are no such situations anymore in an optimal userlevel
implementation. The last problem (in pthread_cond_signal) was fixed
by the addition of FUTEX_WAKE_OP. The userlevel code you're looking
at is simply not optimized for the modern kernels.
What are you suggesting here, that the kernel can be inefficient as long as the user has a way to program around it?

What are you suggesting here, that FUTEX_WAKE_UP is a "user way to program around" an inefficiency? If yes then please explain to me why and what you would do differently.

The point I'm making is that even if an application is "not optimized
for modern kernels" or whatever, there's no reason to ignore
inefficiencies. As Pierre Pfeiffer noted this happens independently of
user code. If a change can eliminate some CPU cycles and possible cache
activity, it would seem to be worth investigation.

The suggestion that the user code was inefficient was not mine...

Did I clarify it this time?

--
-bill davidsen (davidsen@xxxxxxx)
"The secret to procrastination is to put things off until the
last possible moment - but no longer" -me

-
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/