0

I have gone through the RFC 4443 and 8201, perhaps I did not understand as I am new to some terminology described in these RFC but I want to understand the implication of using payload in ICMP packet to big message?

As per the RFC 4443 Linkt to 4443 RFC

The payload will contain as much of invoking packet not exceeding the minimum path MTU in packet to big message.

I don't understand the use case of such payload, even in the RFC 8201 there is no mention about the usages of payload. only one comment present was

Added clarification in Section 4, "Protocol Requirements", that nodes should validate the payload of ICMP PTB messages per RFC 4443, and that nodes should detect decreases in PMTU as fast as possible.

What can be the implication validating payload in the PTB message and how should we validate the payload and based on what conditions?

Packet Too Big Message as per RFC 4443.

   0                   1                   2                   3
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |     Type      |     Code      |          Checksum             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                             MTU                               |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                    As much of invoking packet                 |
  +               as possible without the ICMPv6 packet           +
  |               exceeding the minimum IPv6 MTU [IPv6]           |
Community
  • 1
  • 1
Aryaman Gupta
  • 616
  • 1
  • 8
  • 20
  • IPv6, unlike IPv4, cannot fragment packets in the path if the MTU shrinks somewhere along the path. A method needs to exist to tell the source host that its packets are too big to cross the path. This allows the source host to either pre-fragment the packets, or to adjust its MTU/MSS to fit the smallest MTU in the path. There are also other related RFCs. Search for `rfc pmtud`. – Ron Maupin Jan 07 '20 at 14:57
  • that is correct @RonMaupin, but I was not able to understand why we need payload to set with the invoking packet right after the MTU field in the packet too bing message. – Aryaman Gupta Jan 08 '20 at 04:58
  • 1
    If you look at IPv4 ICMP, at least the first eight bytes of the packet payload are included in the ICMP return messages. That allows the sending application to be identified and notified. For this message, IPv6, itself, and the transport protocol both need to be informed, and IPv6 has a header chain that may include option headers between the IPv6 and transport headers. Rather than waste time parsing everything before replying, let the originating source do that. ICMP is such a low priority, anyway. Routers are supposed to route as fast as possible, and taking time for ICMP impedes that. – Ron Maupin Jan 08 '20 at 06:07
  • Yes @RonMaupin that make sense, after going through some kernel code I came to know that the main purpose of the payload is to store the MTU value per path basis. i.e receiving PTB for some particular path, will make the source to set the MTU value of that particular route to this new MTU given the MTU is valid(i.e not less than minimum ipv6 MTU value) – Aryaman Gupta Jan 08 '20 at 09:16

0 Answers0