netdev - Re: cxgb3: Replace LRO with GRO

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: <20090116235837.GA22005@gondor.apana.org.au>
Date:	Sat, 17 Jan 2009 10:58:38 +1100
From:	Herbert Xu <herbert@...dor.apana.org.au>
To:	Divy Le Ray <divy@...lsio.com>
Cc:	davem@...emloft.net, netdev@...r.kernel.org,
	Steve Wise <swise@...ngridcomputing.com>
Subject: Re: cxgb3: Replace LRO with GRO

On Fri, Jan 16, 2009 at 03:12:07AM -0800, Divy Le Ray wrote:
>
> >> without the patch, lro off: 3.8Gbs
> >> without the patch, lro on: 6.1Gbs
> >> with the patch, GRO on: 4.8Gbs.
> > 
> > What about the case of GRO off with the patch? Just checking to
> > make sure that nothing else has changed.
> 
> I'm getting about 3.4Gbs with the patch and GRO off.

I wonder where the 3.8 => 3.4 slow-down is coming from.

> I did some profiling with GRO on and LRO on. Here are typical outputs:
> GRO on:
> CPU 1 :
>       26.445100  copy_user_generic_unrolled          vmlinux
>       10.050300  skb_copy_datagram_iovec             vmlinux
>        5.076900  memcpy                              vmlinux
>        4.636400  process_responses                  cxgb3.ko
>        4.625100  __pskb_pull_tail                    vmlinux
>        4.463300  eth_type_trans                      vmlinux
>        4.420600  inet_gro_receive                    vmlinux
>        4.310500  refill_fl                          cxgb3.ko
>        3.312700  skb_copy_bits                       vmlinux

Thanks, this is very helpful.  The first obvious deficiency
is that napi_fraginfo_skb is causing multiple pskb_may_pull
calls on every fragment which is bad.

I'll get a patch out for you test.

Cheers,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@...dor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
--
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