netdev - Re: pre-fetching skb for delayed send

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: <20120707141808.GE19462@kvack.org>
Date:	Sat, 7 Jul 2012 10:18:08 -0400
From:	Benjamin LaHaise <bcrl@...ck.org>
To:	Ben Greear <greearb@...delatech.com>
Cc:	netdev <netdev@...r.kernel.org>
Subject: Re: pre-fetching skb for delayed send

On Fri, Jul 06, 2012 at 09:15:29PM -0700, Ben Greear wrote:
> Well, to start with..I at least know the next skb to transmit,
> so I figured I'd prefetch it before starting tx of the current
> skb.

Prefetching data you're just about to immediately access doesn't actually 
help improve performance -- it's better to just access the data.  Prefetching 
subsequent skbs should be of more benefit.

> My question is more basic though:  Given an skb, how do you prefetch
> it...do you just prefetch the skb pointer, or do you need to dig into
> the guts of the skb?

See prefetch.h for details.  Just pass the pointer to the cacheline you want 
to trigger prefetch on to prefetch() or prefetchw(), or use prefetch_range() 
(probably useful for skbs given that they're larger than one cacheline).  
For an skb, you may have to prefetch the frag list as well.

		-ben
-- 
"Thought is the essence of where you are now."
--
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