From: Hangbin Liu <liuhangbin@gmail.com>
To: Eric Dumazet <eric.dumazet@gmail.com>
Cc: netdev@vger.kernel.org, "David S . Miller" <davem@davemloft.net>,
Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
Jakub Kicinski <kuba@kernel.org>,
Willem de Bruijn <willemb@google.com>
Subject: Re: [PATCH net 2/2] IPv6: reply ICMP error if the first fragment don't include all headers
Date: Fri, 9 Oct 2020 18:07:38 +0800 [thread overview]
Message-ID: <20201009100738.GK2531@dhcp-12-153.nay.redhat.com> (raw)
In-Reply-To: <f7272dda-0383-c7d0-1a8a-4a70a1aadb77@gmail.com>
On Thu, Oct 08, 2020 at 11:47:00AM +0200, Eric Dumazet wrote:
>
>
> On 10/8/20 10:30 AM, Hangbin Liu wrote:
> >>> @@ -282,6 +285,21 @@ static struct sk_buff *ip6_rcv_core(struct sk_buff *skb, struct net_device *dev,
> >>> }
> >>> }
> >>>
> >>> + /* RFC 8200, Section 4.5 Fragment Header:
> >>> + * If the first fragment does not include all headers through an
> >>> + * Upper-Layer header, then that fragment should be discarded and
> >>> + * an ICMP Parameter Problem, Code 3, message should be sent to
> >>> + * the source of the fragment, with the Pointer field set to zero.
> >>> + */
> >>> + nexthdr = hdr->nexthdr;
> >>> + offset = ipv6_skip_exthdr(skb, skb_transport_offset(skb), &nexthdr, &frag_off);
> >>> + if (frag_off == htons(IP6_MF) && !pskb_may_pull(skb, offset + 1)) {
> >>> + __IP6_INC_STATS(net, idev, IPSTATS_MIB_INHDRERRORS);
> >>> + icmpv6_param_prob(skb, ICMPV6_HDR_INCOMP, 0);
> >>> + rcu_read_unlock();
> >>> + return NULL;
> >>> + }
> >>> +
> >>> rcu_read_unlock();
> >>>
> >>> /* Must drop socket now because of tproxy. */
> >>>
> >>
> >> Ouch, this is quite a buggy patch.
> >>
> >> I doubt we want to add yet another ipv6_skip_exthdr() call in IPv6 fast path.
> >>
> >> Surely the presence of NEXTHDR_FRAGMENT is already tested elsewhere ?
> >
> > Would you like to help point where NEXTHDR_FRAGMENT was tested before IPv6
> > defragment?
> I think we have to ask the question : Should routers enforce the rule, or
> only end points ?
From IPv6 Core Conformance test[1], it applied to both router and host(It will
marked specifically if a test only for router).
>
> End points must handle NEXTHDR_FRAGMENT, in ipv6_frag_rcv()
Yes, I was also try put the check there, but it looks that would be too late
if module nf_defrag_ipv6 loaded
> >> Also ipv6_skip_exthdr() can return an error.
> >
> > it returns -1 as error, If we tested it by (offset + 1 > skb->len), does
> > that count as an error handler?
>
> If there is an error, then you want to send the ICMP, right ?
No, this is only for fragment header with no enough Upper-Layer header, which need
send ICMP Parameter Problem, Code 3 specifically. For other errors, I guess
the other code will take care of it.
So for -1 return, I just skipped it.
>
> The (offset + 1) expression will become 0, and surely the test will be false,
> so you wont send the ICMP...
[1] v6LC.1.3.6: First Fragment Doesn’t Contain All Headers part A, B,
C and D at https://ipv6ready.org/docs/Core_Conformance_5_0_0.pdf
Thanks
Hangbin
next prev parent reply other threads:[~2020-10-09 10:07 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-07 3:55 [PATCH net 0/2] IPv6: reply ICMP error with fragment doesn't contain all headers Hangbin Liu
2020-10-07 3:55 ` [PATCH net 1/2] ICMPv6: Add ICMPv6 Parameter Problem, code 3 definition Hangbin Liu
2020-10-07 3:55 ` [PATCH net 2/2] IPv6: reply ICMP error if the first fragment don't include all headers Hangbin Liu
2020-10-07 9:35 ` Eric Dumazet
2020-10-08 8:30 ` Hangbin Liu
2020-10-08 9:47 ` Eric Dumazet
2020-10-09 10:07 ` Hangbin Liu [this message]
2020-10-07 14:58 ` Jakub Kicinski
2020-10-08 8:36 ` Hangbin Liu
2020-10-21 4:20 ` [PATCHv2 net 0/2] IPv6: reply ICMP error with fragment doesn't contain " Hangbin Liu
2020-10-21 4:20 ` [PATCHv2 net 1/2] ICMPv6: Add ICMPv6 Parameter Problem, code 3 definition Hangbin Liu
2020-10-21 4:20 ` [PATCHv2 net 2/2] IPv6: reply ICMP error if the first fragment don't include all headers Hangbin Liu
2020-10-21 14:02 ` Willem de Bruijn
2020-10-22 9:12 ` Hangbin Liu
2020-10-22 15:46 ` Willem de Bruijn
2020-10-23 6:43 ` [PATCHv3 net 0/2] IPv6: reply ICMP error if fragment doesn't contain " Hangbin Liu
2020-10-23 6:43 ` [PATCHv3 net 1/2] ICMPv6: Add ICMPv6 Parameter Problem, code 3 definition Hangbin Liu
2020-10-23 6:43 ` [PATCHv3 net 2/2] IPv6: reply ICMP error if the first fragment doesn't include all headers Hangbin Liu
2020-10-23 18:18 ` Jakub Kicinski
2020-10-26 7:29 ` [PATCHv4 net 0/2] IPv6: reply ICMP error if fragment doesn't contain " Hangbin Liu
2020-10-26 7:29 ` [PATCHv4 net 1/2] ICMPv6: Add ICMPv6 Parameter Problem, code 3 definition Hangbin Liu
2020-10-26 7:29 ` [PATCHv4 net 2/2] IPv6: reply ICMP error if the first fragment don't include all headers Hangbin Liu
2020-10-26 8:09 ` Georg Kohmann (geokohma)
2020-10-26 12:55 ` Hangbin Liu
2020-10-26 14:49 ` Georg Kohmann (geokohma)
2020-10-27 2:28 ` [PATCHv5 net 0/2] IPv6: reply ICMP error if fragment doesn't contain " Hangbin Liu
2020-10-27 2:28 ` [PATCHv5 net 1/2] ICMPv6: Add ICMPv6 Parameter Problem, code 3 definition Hangbin Liu
2020-10-27 2:28 ` [PATCHv5 net 2/2] IPv6: reply ICMP error if the first fragment don't include all headers Hangbin Liu
2020-10-27 7:57 ` Georg Kohmann (geokohma)
2020-10-27 9:57 ` Hangbin Liu
2020-10-27 10:20 ` Georg Kohmann (geokohma)
2020-10-30 15:31 ` Willem de Bruijn
2020-10-30 18:39 ` Georg Kohmann (geokohma)
2020-10-27 12:33 ` [PATCHv6 net 0/2] IPv6: reply ICMP error if fragment doesn't contain " Hangbin Liu
2020-10-27 12:33 ` [PATCHv6 net 1/2] ICMPv6: Add ICMPv6 Parameter Problem, code 3 definition Hangbin Liu
2020-10-27 12:33 ` [PATCHv6 net 2/2] IPv6: reply ICMP error if the first fragment don't include all headers Hangbin Liu
2020-10-31 21:12 ` [PATCHv6 net 0/2] IPv6: reply ICMP error if fragment doesn't contain " Jakub Kicinski
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=20201009100738.GK2531@dhcp-12-153.nay.redhat.com \
--to=liuhangbin@gmail.com \
--cc=davem@davemloft.net \
--cc=eric.dumazet@gmail.com \
--cc=kuba@kernel.org \
--cc=kuznet@ms2.inr.ac.ru \
--cc=netdev@vger.kernel.org \
--cc=willemb@google.com \
--cc=yoshfuji@linux-ipv6.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).