[6lo] clarifications in minimal fragments [ was :Question on draft-ietf-6lo-fragment-recovery and VRB ]

["Pascal Thubert (pthubert)" <[e-mail]>]

Following Martine’s questions, please find proposed additions or updates to the minimal fragment draft:


Before:
“

   Conceptually, a reassembly buffer for 6LoWPAN contains:

   o  a datagram_size,
   o  a datagram_tag, associated to the link-layer sender and receiver
      addresses to which the datagram_tag is local,
   o  the actual packet data from the fragments received so far, in a
      form that makes it possible to detect when the whole packet has
      been received and can be processed or forwarded,
   o  a timer that allows discarding a partially reassembled packet
      after some timeout.

“
After
“

The reassembly buffer for 6LoWPAN is indexed in node B by
- a unique Identifier of Node A (e.g., Node A's link-layer address)
- the datagram_tag chosen by node A for this fragmented datagram

Because it may be hard for node B to correlate all possible link-layer addresses that node A may use (e.g., short vs. long addresses), node A MUST use the same link-layer address to send all the fragments of a same datagram to node B.

Conceptually, the reassembly buffer in node B contains, assuming that node B is neither the source or the final destination:
- a datagram_tag as received in the incoming fragments, associated to link-layer address of node A to which the received datagram_tag is local,
- the link-layer address that node B uses to forward the fragments
- the link-layer address of the next hop that is resolved on the first fragment
- a datagram_tag that node B allocated for this datagram and that is used when forwarding the fragments of the datagram
- the actual packet data from the fragments received so far, in a form that makes it possible to detect when the whole packet has been received and can be processed or forwarded,
- a datagram_size,
- a buffer for the remainder of a previous fragment left to be sent,
- a timer that allows discarding a partially reassembled packet after some timeout.
“


And later:

Before

“      Each datagram can be uniquely identified by the source and final
   destination link-layer addresses of the frame that carries it, the
   fragment size and the datagram_tag.

“
After
“
Each datagram can be uniquely identified by the sender link-layer addresses of the frame that carries it and the datagram_tag that the sender allocated for this datagram.
“

Please let me know if you see an issue with that. I wish to publish soon as we enter the IESG review for this spec.

All the best,

Pascal

From: 6lo <[e-mail]> On Behalf Of Pascal Thubert (pthubert)
Sent: vendredi 30 août 2019 13:44
To: Martine Lenders <[e-mail]>; [e-mail]
Subject: Re: [6lo] Question on draft-ietf-6lo-fragment-recovery and VRB

Hello Martine

Great to hear about your implementation. The group is interested to learn about your progress (keep us tuned) and happy to help you through. P
lease do not hesitate to ask for more clarification as you feel needed. Even if you’re not too sure you should ask, if you wonder there’s probably a reason and you’re helping the implementors who will be following you.

Please see below:

I have an implementation question for the virtual reassembly buffer with regards to [draft-ietf-6lo-fragment-recovery]. Section 6.1 of this draft states that a tuple (source address, tag) is used to identify a VRB entry and cites [draft-ietf-6lo-minimal-fragment] for that.

Ø   Correct, and the meaning is source MAC address not source IP address. The exact sentence is

Ø                                       Upon a first fragment (i.e. with a

Ø     sequence of zero), a VRB and the associated LSP state are created for

Ø     the tuple (source MAC address, datagram_tag) and the fragment is

Ø     forwarded along the IPv6 route that matches the destination IPv6

Ø     address in the IPv6 header as prescribed by

Ø     [I-D.ietf-6lo-minimal-fragment<https://tools.ietf.org/html/draft-ietf-6lo-fragment-recovery-05#ref-I-D.ietf-6lo-minimal-fragment>].

Ø
However, as far as I understand [draft-ietf-6lo-minimal-fragment] and [draft-ietf-lwig-6lowpan-virtual-reassembly] this is not the case. [draft-ietf-6lo-minimal-fragment] doesn't mention VRB entry identification and only refers to [draft-ietf-lwig-6lowpan-virtual-reassembly].

  *   It would not hurt to mention it though. Interesting. Minimal has:” All fragments are tagged with a 16-bit

   "    Each datagram can be uniquely identified by the source and final

   destination link-layer addresses of the frame that carries it, the

   fragment size and the datagram_tag.



Ø   That’s  misleading… I’ll correct it.
That in turn only recounts the classic (source address, destination address, size, tag) tuple defined in RFC4944

  *   When receiving a fragment, the destination is self and the size may vary from a datagram to the next. So the only thing that really identify the datagram is the source MAC and the tag.
and states that the only difference between the classic and VRB is the lack of the reassembled packet and addition of the next hop's address and the newly assigned tag:

Ø   Actually the forwarder sends the datagram with self as source to the source mac address is swapped naturally. It has to set the next hop’s mac address and the swapped datagram tag. The next hop’s MAC address identifies the node that should receive the packet and the datagram tag together with the source mac identify the reassembly buffer within the receiver. There is no contradiction.
To reduce the memory requirement for reassembly buffers, the implementation may opt to not keep the actual packet data in the reassembly buffer. Instead, it may attempt to send out the data for a fragment in the form of a forwarded fragment, as soon as all necessary information for that is available. Obviously, all fragments need to be sent with the same outgoing address (otherwise a full reassembly implementation would discard the fragments) and the same datagram_tag.

So my question is: Is the tuple definition in [draft-ietf-6lo-fragment-recovery] really correct?


  *   It is.

For exclusion datagram size I agree that it could be somewhat redundant. However, a node could have multiple destination addresses (either via multiple interfaces or IEEE-802.15.4-style short and long address). So, as the tag is link-specific (defined by a (source, destination) tuple), there could be distinct datagrams that have the same tuple (source, tag), or am I missing something?


Ø   Yes, the role of the destination MAC address is to reach the next hop, but does not change the processing within that node. I could change the L2 address I use to refer to the next hop in the middle of a fragmented packet, it is still the same fragmented packet. In other words, please do not use the dmac as to index the VRB. From your question I think the LWIG draft should clarify, and we can change minimal draft to clarify as well.

I already implemented the VRB with draft-ietf-6lo-minimal-fragment in mind and thus used the classic 4-tuple for indexing. But now I'm wondering: Can I reduce its index or would that be not advised?


Ø   You should reduce the index : ) There can not be 2 different datagrams coming in parallel from a same previous hop and with a same datagram tag. There must be a sentence somewhere in the LWIG draft that says that a new and currently unused datagram tag is chosen for a new incoming datagram, correct?  Minimal says

Ø                                                                Upon

Ø     receiving a fragment from node A with a datagram_tag previously

Ø     unseen from node A, node B allocates a buffer large enough to hold

Ø     the entire packet.



Ø   This text only indexes the VRB with any identifier of node A and the tag… And this is correct. What’s a bit more ugly is that Node A should not change the source MAC in between fragments because that will confise node B. I need to add text on that too.


Many thanks Martine for raising all this, and all the best for your implementation.

Pascal