Hello Dave Many thanks for being so reactive 😊
Please see below > -----Original Message----- > From: Dave Thaler via Datatracker <[email protected]> > Sent: jeudi 7 novembre 2019 00:10 > To: [email protected] > Cc: [email protected]; [email protected]; > [email protected] > Subject: Intdir last call review of draft-ietf-6lo-minimal-fragment-04 > > Reviewer: Dave Thaler > Review result: Ready with Issues > > The title implies the document specifies a forwarding mechanism, but it does > not, it merely provides discussion of two mechanisms in other docs (RFC 4944 > and draft-ietf-lwig-6lowpan-virtual-reassembly). I would recommend at least > changing the title to be more clear as to the purpose of the doc. A suggestion would help : ) Does "On Forwarding 6LoWPAN Fragments over a Multihop IPv6 Network" go the right way? > > Technical confusion > ------------------- > 1) Page 3 says the reassembly buffer contains "the link-layer address that > node B uses to forward the > fragments". I cannot tell whether this is referring to B's link-layer > address that it received the fragment on, or B's link-layer address that it > uses as a source link-layer address for forwarding it on, or the link-layer > address of the next hop to which B forwards. The latter. B needs to send all the fragments with the same source link-layer address because that's part of the index for the datagram in C. Proposed change: " the Link-Layer address that node B uses as source to forward the fragments " > > 2) Page 3 also says the reassembly buffer contains "the link-layer address of > the next hop that is resolved > on the first fragment". I found this similarly confusing. What does it > mean to resolve something "on" the first fragment? Does it mean "during > processing of the first fragment"? Maybe I missed it, but I couldn't find > in RFC 4944 anywhere that says that it would do next-hop resolution before > the datagram can be reassembled. That would seem like a waste, if the > fragments are then discarded (e.g., due to timer expiry) without actually > doing any forwarding. > RFC 4944 reassembles and then routes. We make the routing decision on the first fragment before we receive the second fragment, forward the first fragment and store that state. Unsure how to reword, your suggestion would be appreciated. > 3) Section 3 talks about "MAC address" specifically whereas section 1 always > talked about the more > generic "link-layer address". Why the inconsistency? > Fixed to "link-layer address" consistently. > 4) Section 3 talks about "a 1280-byte reassembly buffer for each packet", but > section 2.2 talks assumes > a "1 kB reassembly buffer". 1k != 1280 bytes. Why the inconsistency? > Fixed to " Assuming a reassembly buffer for a 6LoWPAN MTU of 1280 bytes as defined in section 4 of [6LoWPAN], typical nodes only have enough memory for 1-3 reassembly buffers. " > 5) Section 3 explains that "the first fragment must always be forwarded > first", > but does not explain > what the behavior is if a fragment other than the first fragment is > received > before the first fragment. Figure 1 shows that the fragments can be > received > out of order, since there fragment 6 is received before 5, which is > received > before 4. Presumably it is either queued or dropped. If it's queued, > then > section 4 is insufficient, which talks about an attacker generating a large > number of bogus "fragment 1" fragments, since if you queue the first > fragment received even if it's not "fragment 1", then the same attack > presumably exists, it's not specific to "fragment 1" packets. > 6LoWPAN does not mandate that all the fragments are sent in order, thus Fig1. But fragment 1 is sent first Quoting section 5.5 of Rfc 4944 " The first link fragment SHALL contain the first fragment header as defined below. 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 1 0 0 0| datagram_size | datagram_tag | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 4: First Fragment The second and subsequent link fragments (up to and including the last) SHALL contain a fragmentation header that conforms to the format shown below. 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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 1 1 0 0| datagram_size | datagram_tag | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |datagram_offset| +-+-+-+-+-+-+-+-+ Figure 5: Subsequent Fragments " The first fragment is the one with the IPv6 address of the destination that enables to find the next hop that the other fragments will use. So it is always available before a fragment can be forwarded. So it is easy to mandate that we forward it first. If we do that then the only way for a next fragment to arrive first is that the first fragment was lost in the transmission by the previous node. The first fragment may be queued for retries in the previous hop but that's a really bad idea. Proposal: * we add text on the above to clarify * we mandate that on a link with ARQ, the node only forwards a next fragment if the first was acknowledged. * we clarify that a next fragment that is received with no state from a first fragment for that datagram should be dropped. > Grammatical nits: > ----------------- > > Abstract has "... to forwarding ...", which should be "to forward" or "for > forwarding" done > > Abstract has "to the virtual Reassembly Buffer", which seems incorrect both in > terms of capitalization (since sectoin 3 has VRB) and grammar. Suggest "to > using virtual reassembly buffers". I think we meant to the VRB draft. Applied your recommendation in the meantime: " This method reduces the latency and increases end-to-end reliability in route-over forwarding. It is the companion to using virtual reassembly buffers which is a pure implementation technique. " Does that read well? > > Section 1, first paragraph: "though possibly" is likely a typo for "through > possibly" > Fixed > Section 1, 6th paragraph: "a same datagram" is oddly worded. Suggest either > "a datagram" or "the same datagram" Yes French uses "a" to mean like any old datagram but still the same one; fixed : ) > > Section 2.2, grammar issue in "Assuming 1 kB reassembly buffer". Either > "buffers" plural or "Assuming a ..." > Fixed as discussed above Many thanks Dave. I'll be waiting for you answer on the open issues above before publishing. Take care Pascal _______________________________________________ 6lo mailing list [email protected] https://www.ietf.org/mailman/listinfo/6lo
