netfilter-devel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Phil Sutter <phil@nwl.cc>
To: Sriram Rajagopalan <bglsriram@gmail.com>
Cc: netfilter-devel@vger.kernel.org
Subject: Re: iptables-nft: Wrong payload merge of rule filter - "! --sport xx ! --dport xx"
Date: Fri, 8 Mar 2024 12:30:47 +0100	[thread overview]
Message-ID: <Zer252IUJr07J_eX@orbyte.nwl.cc> (raw)
In-Reply-To: <CAPtndGDEJVWXcggRkw66YLjhu3QyUjJ5j4YEbvJLj-qbPkQaPg@mail.gmail.com>

Hi Sriram,

On Fri, Mar 08, 2024 at 02:49:38PM +0530, Sriram Rajagopalan wrote:
> iptables-nft based on nftables has an issue with the way the rule
> filter - "! --sport xx ! --dport xx" is wrongly merged and rendered.

I agree with your analysis and the patches look fine. Could you please
submit them formally?

[...]
> % export IPTABLES=/usr/local/sbin/iptables-legacy; sudo $IPTABLES -A
> INPUT -p tcp ! --sport 22 ! --dport 22 -i vm2; echo -e "\n---- Before
> data ----\n"; sudo $IPTABLES -L INPUT -vvvn; sudo python -c "from
> scapy.all import *;
> sendp(Ether(dst='9e:00:fa:a3:c9:48')/IP(src='1.1.1.1',
> dst='2.2.2.2')/TCP(sport=23, dport=22), iface='vm1')"; echo -e "\n----
> After data with either one of tcp sport/dport being 22 ----\n"; sudo
> $IPTABLES -L INPUT -vn; sudo python -c "from scapy.all import *;
> sendp(Ether(dst='9e:00:fa:a3:c9:48')/IP(src='1.1.1.1',
> dst='2.2.2.2')/TCP(sport=23, dport=23), iface='vm1')"; echo -e "\n----
> After data with neither one of tcp sport/dport being 22 ----\n"; sudo
> $IPTABLES -L INPUT -vn; sudo $IPTABLES -D INPUT -p tcp ! --sport 22 !
> --dport 22 -i vm2
> 
> 
> ---- Before data ----
> 
> ip filter INPUT 41
>   [ meta load iifname => reg 1 ]
>   [ cmp eq reg 1 0x00326d76 ]
>   [ payload load 1b @ network header + 9 => reg 1 ]
>   [ cmp eq reg 1 0x00000006 ]
>   [ payload load 2b @ transport header + 0 => reg 1 ]
>   [ cmp neq reg 1 0x00001600 ]
>   [ payload load 2b @ transport header + 2 => reg 1 ]
>   [ cmp neq reg 1 0x00001600 ]
>   [ counter pkts 0 bytes 0 ]

You're fibbing here: That netlink debug output can't come from
iptables-legacy. I suspect it actually comes from your patched
iptables-nft or nft too. :)

[...]
> Author: Sriram Rajagopalan <bglsriram@gmail.com>
> Date:   Fri Mar 07 20:09:38 2024 -0800
> 
> iptables: Fixed the issue with combining the payload in case of invert
> filter for tcp src and dst ports
> 
> Signed-off-by: Sriram Rajagopalan <bglsriram@gmail.com>
> Signed-off-by: Sriram Rajagopalan <sriramr@arista.com>

Maybe avoid the double SoB? Apart from that:

Acked-by: Phil Sutter <phil@nwl.cc>

Thanks, Phil

  reply	other threads:[~2024-03-08 11:30 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-03-08  9:19 iptables-nft: Wrong payload merge of rule filter - "! --sport xx ! --dport xx" Sriram Rajagopalan
2024-03-08 11:30 ` Phil Sutter [this message]
2024-03-13  9:01   ` Sriram Rajagopalan
2024-03-08 13:37 ` Florian Westphal
2024-03-12 10:24   ` Sriram Rajagopalan
2024-03-12 10:34     ` Florian Westphal
2024-03-13  9:02       ` Sriram Rajagopalan

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=Zer252IUJr07J_eX@orbyte.nwl.cc \
    --to=phil@nwl.cc \
    --cc=bglsriram@gmail.com \
    --cc=netfilter-devel@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).