netdev - Re: Stable regression with 'tcp: allow splice() to build full TSO packets'

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

Message-ID: <20120517172246.GM14498@1wt.eu>
Date:	Thu, 17 May 2012 19:22:46 +0200
From:	Willy Tarreau <w@....eu>
To:	Eric Dumazet <eric.dumazet@...il.com>
Cc:	netdev@...r.kernel.org
Subject: Re: Stable regression with 'tcp: allow splice() to build full TSO packets'

On Thu, May 17, 2012 at 06:49:47PM +0200, Eric Dumazet wrote:
> On Thu, 2012-05-17 at 18:40 +0200, Willy Tarreau wrote:
> >  try to port your patch (8d4057 tg3: provide frags as skb head) to
> > mv643xx_eth as an exercise. If I succeed and notice an improvement, I'll
> > send a patch :-)
> 
> 
> tg3 uses build_skb(), thats an API that allocates the skb after frame is
> filled by the device, to have less cache misses in RX path.
> 
> But for most devices, netdev_alloc_skb() would be more than enough.

OK, so that's out of reach for me right now.

> I'll send a patch once tested.

I'll happily give it a try.

BTW, net-next gave me around 13-14% improvement over 3.0/3.2+splice
patches ! Now I'm seeing splice() being 75% faster than recv/send, while
it used to be slower on this hardware not that long ago! That's really
impressive, great work!

Willy

--
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