netdev - Re: [PATCH 0/11] use ether_addr_equal

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20140117224315.GC13930@amd.pavel.ucw.cz>
Date:	Fri, 17 Jan 2014 23:43:15 +0100
From:	Pavel Machek <pavel@....cz>
To:	Oleksij Rempel <linux@...pel-privat.de>
Cc:	Julia Lawall <Julia.Lawall@...6.fr>, ath9k-devel@...ts.ath9k.org,
	kernel-janitors@...r.kernel.org, ath5k-devel@...ts.ath5k.org,
	linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
	linux-wireless@...r.kernel.org,
	"John W. Linville" <linville@...driver.com>,
	users@...x00.serialmonkey.com
Subject: Re: [PATCH 0/11] use ether_addr_equal_64bits

On Fri 2014-01-17 23:02:06, Oleksij Rempel wrote:
> Am 17.01.2014 22:24, schrieb Pavel Machek:
> > On Mon 2013-12-30 19:14:56, Julia Lawall wrote:
> >> Ether_addr_equal_64bits is more efficient than ether_addr_equal, and can be
> >> used when each argument is an array within a structure that contains at
> >> least two bytes of data beyond the array.
> > 
> > I mean, yes, it is probably faster, and yes, most structures probably
> > contain two more bytes, but... is the uglyness worth the speedup? I'd
> > say this should not be done except in very time-critical places...
> 
> This code run on every received beacon, almost on every wifi driver (If
> i understand what you mean.)

That does not look like "sufficiently often" to me. Can you measure
the improvement at least in some microbenchmark? Is there even
theoretical chance to get one?

You are comparing few bytes, number of cacheline accesses stays same,
there is likely _0_ speedup. And even if you saved 1T, that will be
compeletely lost in the noise.

In some kind of routing code, cache-hot... maybe it would make
sense. But once per interrupt?

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html