netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Patrick McHardy <kaber@trash.net>
To: David Miller <davem@davemloft.net>
Cc: hadi@cyberus.ca, herbert@gondor.apana.org.au,
	netdev@vger.kernel.org, kuznet@ms2.inr.ac.ru
Subject: Re: [PKT_SCHED]: Add stateless NAT
Date: Fri, 28 Sep 2007 18:55:32 +0200	[thread overview]
Message-ID: <46FD3204.6090305@trash.net> (raw)
In-Reply-To: <20070927.120803.39482469.davem@davemloft.net>

David Miller wrote:
> From: jamal <hadi@cyberus.ca>
> Date: Thu, 27 Sep 2007 08:39:45 -0400
> 
>>>+config NET_ACT_NAT
>>>+        tristate "Stateless NAT"
>>>+        depends on NET_CLS_ACT
>>>+        select NETFILTER
>>
>>I am gonna have to agree with Evgeniy on this Herbert;->
>>The rewards are it will improve performance for people who dont need
>>netfilter.
> 
> 
> I agree that we should move the functions out.   However...
> 
> You have to realize that this logic is a complete crock
> of shit for %99.99999999999999999999 of Linux users out there
> who get and only use distribution compiled kernels which are
> going to enable everything anyways.
> 
> So we better make sure there are zero performance implications at some
> point just for compiling netfilter into the tree.  I know that isn't
> the case currently, but that means that we aren't helping out the
> majority of Linux users and are thus only adding these optimizations
> for such a small sliver of users and that is totally pointless and
> sucks.


The hooks themselves actually shouldn't be that much of a problem
since without any registered users they come down to a simple
list_empty check. The bigger problem is probably the fact that
the "okfn" is usually declared inline, but we also take its address
for nf_hook_slow, so at least with forced inlining, we end up
with one inlined and one out-of-line version.

Looking at ip_input.o as example (everything without forced inlining):

   text    data     bss     dec     hex filename
   2076       8       0    2084     824 net/ipv4/ip_input.o
   3483       8       0    3491     da3 net/ipv4/ip_input.o


So we have roughly 1.75 more code for the two hooks. For reference,
without using the function pointer it increases a lot less:

   text    data     bss     dec     hex filename
   2319       8       0    2327     917 net/ipv4/ip_input.o

similar with not inlining ip_rcv_finish() and ip_local_deliver_finish():

   text    data     bss     dec     hex filename
   2282       8       0    2290     8f2 net/ipv4/ip_input.o

Any ideas how to do this better? Unforunately we need the function
pointers for reinjecting queued packets ..

  reply	other threads:[~2007-09-28 16:55 UTC|newest]

Thread overview: 80+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-09-27  7:34 [PKT_SCHED]: Add stateless NAT Herbert Xu
2007-09-27  9:25 ` Evgeniy Polyakov
2007-09-27  9:33   ` Herbert Xu
2007-09-27 10:07     ` Evgeniy Polyakov
2007-09-27 10:29       ` Herbert Xu
2007-09-27 12:41         ` Evgeniy Polyakov
2007-09-27 12:45           ` Herbert Xu
2007-09-27 13:10             ` Evgeniy Polyakov
2007-09-27 13:16               ` Patrick McHardy
2007-09-27 13:25                 ` Evgeniy Polyakov
2007-09-27 13:30                   ` Patrick McHardy
2007-09-27 13:33                     ` Evgeniy Polyakov
2007-09-27 13:34                     ` jamal
2007-09-27 13:20               ` Herbert Xu
2007-09-27 13:29                 ` Evgeniy Polyakov
2007-09-27 13:39                   ` Patrick McHardy
2007-09-27 19:52                     ` David Miller
2007-09-28  8:19                       ` Evgeniy Polyakov
2007-09-28  9:16                         ` Evgeniy Polyakov
2007-09-28 16:31                       ` Patrick McHardy
2007-09-27 12:52           ` jamal
2007-09-27 13:06             ` Evgeniy Polyakov
2007-09-27 12:27     ` Herbert Xu
2007-09-27 12:46   ` jamal
2007-09-27 12:39 ` jamal
2007-09-27 12:58   ` Herbert Xu
2007-09-27 19:48     ` David Miller
2007-09-27 13:01   ` Herbert Xu
2007-09-27 13:16     ` jamal
2007-09-27 19:08   ` David Miller
2007-09-28 16:55     ` Patrick McHardy [this message]
2007-09-29  0:51       ` Herbert Xu
2007-09-29  3:37         ` Herbert Xu
2007-09-29 15:21           ` Patrick McHardy
2007-09-29 23:36             ` Herbert Xu
2007-09-30  0:13             ` David Miller
2007-09-30  0:26               ` Herbert Xu
2007-09-30  0:43                 ` Herbert Xu
2007-09-30 15:38                   ` Patrick McHardy
2007-10-14  4:26                     ` [0/10] Remove sk_buff ** from netfilter API Herbert Xu
2007-10-14  4:27                       ` [PATCH 1/10] [SKBUFF]: Merge common code between copy_skb_header and skb_clone Herbert Xu
2007-10-14 10:36                         ` jamal
2007-10-14  4:27                       ` [PATCH 2/10] [SKBUFF]: Add skb_morph Herbert Xu
2007-11-26  6:50                         ` Yasuyuki KOZAKAI
     [not found]                         ` <200711260650.lAQ6oOL9006774@toshiba.co.jp>
2007-11-26 15:10                           ` Herbert Xu
2007-11-27  6:50                             ` Yasuyuki KOZAKAI
2007-10-14  4:27                       ` [PATCH 3/10] [IPV4]: Make ip_defrag return the same packet Herbert Xu
2007-10-14  4:27                       ` [PATCH 4/10] [IPV4]: Change ip_defrag to return an integer Herbert Xu
2007-10-14  4:27                       ` [PATCH 5/10] [NET]: Avoid unnecessary cloning for ingress filtering Herbert Xu
2007-10-14 10:49                         ` jamal
2007-10-15 13:57                           ` jamal
2007-10-15 14:28                             ` Patrick McHardy
2007-10-15 14:55                             ` Alexey Kuznetsov
2007-10-16 10:28                               ` jamal
2007-10-14  4:27                       ` [PATCH 6/10] [BRIDGE]: Unshare skb upon entry Herbert Xu
2007-10-15  7:20                         ` Patrick McHardy
2007-10-15  8:04                           ` Herbert Xu
2007-10-15  8:50                           ` David Miller
2007-10-14  4:27                       ` [PATCH 7/10] [NETFILTER]: Do not copy skb in skb_make_writable Herbert Xu
2007-10-14  4:27                       ` [PATCH 8/10] [IPVS]: Replace local version of skb_make_writable Herbert Xu
2007-10-14  4:27                       ` [PATCH 9/10] [NETFILTER]: Avoid skb_copy/pskb_copy/skb_realloc_headroom Herbert Xu
2007-10-14  4:27                       ` [PATCH 10/10] [NETFILTER]: Replace sk_buff ** with sk_buff * Herbert Xu
2007-10-14  9:55                         ` Herbert Xu
2007-10-14 14:46                           ` [0/3] [IPV6]: Kill sk_buff ** usage on input path Herbert Xu
2007-10-14 14:49                             ` [PATCH 1/3] [IPV6]: Make ipv6_frag_rcv return the same packet Herbert Xu
2007-10-15  8:28                               ` David Miller
2007-10-14 14:49                             ` [PATCH 2/3] [IPV6]: Avoid skb_copy/pskb_copy/skb_realloc_headroom on input Herbert Xu
2007-10-15  8:29                               ` David Miller
2007-10-14 14:49                             ` [PATCH 3/3] [IPV6]: Replace sk_buff ** with sk_buff * in input handlers Herbert Xu
2007-10-14 15:43                               ` YOSHIFUJI Hideaki / 吉藤英明
2007-10-15  8:30                                 ` David Miller
2007-10-15  8:29                               ` David Miller
2007-10-15 14:21                                 ` Herbert Xu
2007-10-15  8:27                           ` [PATCH 10/10] [NETFILTER]: Replace sk_buff ** with sk_buff * David Miller
2007-10-15  6:25                       ` [0/10] Remove sk_buff ** from netfilter API Patrick McHardy
2007-10-15  7:57                       ` [IPV4]: Uninline netfilter okfns Patrick McHardy
2007-10-15  8:05                         ` Herbert Xu
2007-10-15  8:30                         ` [IPV6]: " Patrick McHardy
2007-10-15  8:51                           ` David Miller
2007-10-15  8:49                         ` [IPV4]: " David Miller

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=46FD3204.6090305@trash.net \
    --to=kaber@trash.net \
    --cc=davem@davemloft.net \
    --cc=hadi@cyberus.ca \
    --cc=herbert@gondor.apana.org.au \
    --cc=kuznet@ms2.inr.ac.ru \
    --cc=netdev@vger.kernel.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;
as well as URLs for NNTP newsgroup(s).