From mboxrd@z Thu Jan  1 00:00:00 1970
From: Matthew Wilcox <matthew@wil.cx>
Date: Thu, 09 Feb 2006 12:44:51 +0000
Subject: Re: [KJ] Correct imprecise >= comparisons
Message-Id: <20060209124451.GD1593@parisc-linux.org>
MIME-Version: 1
Content-Type: multipart/mixed; boundary="===============71461868490055691=="
List-Id: <kernel-janitors.vger.kernel.org>
References: <20060208185312.GA26914@rhlx01.fht-esslingen.de>
In-Reply-To: <20060208185312.GA26914@rhlx01.fht-esslingen.de>
To: kernel-janitors@vger.kernel.org

--===============71461868490055691==
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On Thu, Feb 09, 2006 at 01:20:07PM +0100, Andreas Mohr wrote:
> > > if (val >= max)
> > >   val = max;
> > > 
> > > which can instead be expressed as
> > > 
> > > if (val > max)
> > >   val = max;
> > >
> > > in order to cut down on both executed cycles and cache write invalidation.
> > 
> > I am wondering if it is actually worth the effort, since it seems to
> > me as if this would normally be an exception that val >= max. And
> > instead of returning with an error or aborting you normalize the value.
> 
> It probably is unlikely in several cases, but that makes it all the more a
> janitorial effort ;)
> It's just "unclean" to set a variable's limit if it didn't even actually
> step beyond the limit, which makes this more or less a janitorial task.

I wonder if we want a macro similar to min()/max().  Maybe we could
write it as:

limit(val, max);

which could be defined as:

#define limit(val, max)  { \
	typeof(val) _val = (val); typeof(max) _max = (max); \
	if (_val > _max) _val = _max; }

Thoughts?

--===============71461868490055691==
Content-Type: text/plain; charset="iso-8859-1"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
https://lists.osdl.org/mailman/listinfo/kernel-janitors

--===============71461868490055691==--