Linux Netfilter development
 help / color / mirror / Atom feed
From: Patrick McHardy <kaber@trash.net>
To: Fabian Hugelshofer <hugelshofer2006@gmx.ch>
Cc: Pablo Neira Ayuso <pablo@netfilter.org>, netfilter-devel@vger.kernel.org
Subject: Re: Conntrack Events Performance - Multipart Messages?
Date: Mon, 21 Jul 2008 17:59:44 +0200	[thread overview]
Message-ID: <4884B270.5010104@trash.net> (raw)
In-Reply-To: <4884B068.4050306@gmx.ch>

Fabian Hugelshofer wrote:
> It took me some time to set up profiling on the router. I am using 
> oprofile. nf_conntrack and nfnetlink are built into the kernel. Most of 
> the time is probably spent in nfnetlink.
> 
> If you need other data than the one provided here, just let me know. I'm 
> not very familiar with oprofile, so providing the needed arguments would 
> be helpful.

Thanks.

> opreport --symbols (first 20, full file attached):
> CPU: ARM/XScale PMU2, speed 0 MHz (estimated)
> Counted CPU_CYCLES events (clock cycles counter) with a unit mask of 
> 0x00 (No unit mask) count 100000
> samples  %        app name                 symbol name
> 20018     9.6520  vmlinux                  memcpy

Callgraph information would be useful since its unclear whether
this is the memcpy triggered by netlink message trimming in
af_netlink.c or something different. Unfortunately according
to the documentation this is only supported on x86. I think
selecting the netfilter options as modules should provide
slightly more detail though.

> 19493     9.3988  ath_pci.ko               ath_sysctl_register

This looks odd. I couldn't find this function in the current
kernel tree, which version are you using?

> 6676      3.2189  vmlinux                  __nf_conntrack_find
> 6098      2.9402  vmlinux                  ipt_do_table
> 5858      2.8245  wlan.ko                  ieee80211_input
> 5383      2.5955  nf_conntrack_netlink.ko  .text
> 4567      2.2020  vmlinux                  __memzero
> 4225      2.0371  vmlinux                  __kmalloc
> 4091      1.9725  vmlinux                  csum_partial

You can disable conntrack checksumming by executing:

echo 0 >/proc/sys/net/netfilter/nf_conntrack_checksum

> 3469      1.6726  vmlinux                  nf_nat_setup_info
> 3468      1.6721  vmlinux                  netlink_broadcast
> 3376      1.6278  vmlinux                  __hash_conntrack
> 3304      1.5931  vmlinux                  nla_put
> 2992      1.4426  vmlinux                  __nla_reserve
> 2933      1.4142  vmlinux                  __nf_conntrack_confirm
> 2926      1.4108  oprofiled                /jffs/usr/bin/oprofiled
> 2847      1.3727  ath_pci.ko               ath_intr
> 2698      1.3009  vmlinux                  kfree
> 2655      1.2801  vmlinux                  __nla_put
> 2563      1.2358  vmlinux                  nf_conntrack_in
> 


  reply	other threads:[~2008-07-21 15:59 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-07-16 16:42 Conntrack Events Performance - Multipart Messages? Fabian Hugelshofer
2008-07-17  9:16 ` Patrick McHardy
2008-07-17 10:03 ` Pablo Neira Ayuso
2008-07-17 14:34   ` Fabian Hugelshofer
2008-07-17 15:15     ` Fabian Hugelshofer
2008-07-18 15:56     ` Fabian Hugelshofer
2008-07-18  2:11   ` Patrick McHardy
2008-07-21 15:51     ` Fabian Hugelshofer
2008-07-21 15:59       ` Patrick McHardy [this message]
2008-07-21 17:49         ` Fabian Hugelshofer
2008-07-23 14:32           ` Fabian Hugelshofer
2008-07-23 14:38             ` Patrick McHardy
2008-07-23 16:12               ` Fabian Hugelshofer
2008-07-23 17:01                 ` Patrick McHardy
2008-07-23 17:07                   ` Patrick McHardy
2008-07-23 17:30                     ` Fabian Hugelshofer
2008-07-23 17:32                       ` Patrick McHardy
2008-07-23 17:38                         ` Fabian Hugelshofer
2008-07-23 17:40                           ` Patrick McHardy
2008-07-23 17:15                   ` Fabian Hugelshofer
2008-07-23 17:20                     ` Patrick McHardy
2008-07-24 13:21                       ` Fabian Hugelshofer
2008-07-25  8:51                         ` Fabian Hugelshofer
2008-07-25  9:32                         ` Pablo Neira Ayuso
2008-07-25 11:15                           ` Pablo Neira Ayuso
2008-07-27 17:23                             ` Fabian Hugelshofer
2008-07-28 18:31                             ` Pablo Neira Ayuso
2008-07-28 23:12                               ` Fabian Hugelshofer
2008-07-29 17:11                                 ` Pablo Neira Ayuso
2008-07-25  8:44                 ` Fabian Hugelshofer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4884B270.5010104@trash.net \
    --to=kaber@trash.net \
    --cc=hugelshofer2006@gmx.ch \
    --cc=netfilter-devel@vger.kernel.org \
    --cc=pablo@netfilter.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox