Re: [KJ] [third][PATCH 3/4][2.6 drivers/net/skfp/smt.c:]replace SIZE

[Matthew Wilcox <matthew@wil.cx>]

On Mon, Jul 16, 2007 at 07:06:36PM +0200, walter harms wrote:
> your patch seems ok but the stile here ...
> 
> pd=smt_pdef;
> for(i=ARRAY_SIZE(smt_pdef);i;i--) {
> 	if (pd->ptype = type)
> 		break ;
> 	else pd++;
> }
> 	
> normally i would say it should be more like for(i=0;i<ARRAY_SIZE(smt_pdef);i++) (rule of least surprise)
> but i have no clue why he did it not this way.

I can think of three possible reasons:

1. Speed.  Many processors have a 'branch if not zero' instruction.
You can often eliminate one instruction per loop by counting down to
zero instead of up to N.

2. Duplicate entries.  If you want to find the last item in the list
which has ptype = type, you need to start at the end, not the front.

3. Speed.  If you're likely to be removing entries from a list in the
opposite order to which you added them (a stack rather than a queue),
you'll find it more quickly on average if you start at the end than at
the beginning.

There may be other reasons for iterating in the opposite direction;
those are just three possible reasons that popped into my head when you
asked the question.

-- 
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
_______________________________________________
REMINDER: this mailing list moved to vger.kernel.org and current one will be discontinued soon.
To resubscribe, send email to majordomo@vger.kernel.org with
&quot;subscribe kernel-janitors&quot; in message body and follow instructions.

Kernel-janitors mailing list
Kernel-janitors@lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/kernel-janitors