Dear Pascal,

I am confused again on the order of headers in the packets.

For example, the packet:

*MAC header + (......) + ICMPv6*,

The words "IP in IP then  RPI then  6LoRH" equals to "*MAC header + (6LoRH
IPinIP, 6LoRH-RPI, 6LoRH RH3, IPHC) + ICMPv6*",

which we says RPI is RIGHT AFTER IPinIP is the first way I mentioned in the
beginning of this email, no?

Tengfei



On Wed, Jan 20, 2016 at 9:45 PM, Pascal Thubert (pthubert) <
pthub...@cisco.com> wrote:

> OK, let us see.
>
>
>
> With the new text I just proposed, the source (normally that’s the root)
> address of the packet with a RH3 is the reference for compression. So
> Ideally we’d parse that before we parse the RH3 so we can decompress the
> first address in the RH3. This is a change from the original text where I
> expected the first address in the RH3 to be mostly in the full. So in 6LoRH
> form,  IP in IP would come before the RH3.
>
>
>
> With the next text, the root address may be elided but that means placing
> an RPI to identify it if the network has more than one instance. To
> decompress the IP in IP where the root is elided, one really needs the RPI.
> So the RPI should be next to the IP in IP. I would probably have preferred
> to place the RPI before the IP in IP but to look more like the uncompressed
> format we might place the RPI right after the IP in IP.
>
>
>
> This would give (IP in IP then  RPI then  6LoRH *) *. With this we’d
> support multiple encapsulations though I cannot see where we’d need it for
> now.
>
>
>
> Works?
>
>
>
>
>
> Cheers,
>
>
>
> Pascal
>
>
>
> *From:* Tengfei Chang [mailto:tengfei.ch...@gmail.com]
> *Sent:* mercredi 20 janvier 2016 14:22
> *To:* Pascal Thubert (pthubert) <pthub...@cisco.com>
> *Cc:* 6...@ietf.org; 6tisch@ietf.org
> *Subject:* Re: [6tisch] The "BEFORE" and "AFTER"
>
>
>
> Thanks pascal for explaining!
>
>
>
> Yes, I vote to impose an order for the headers in the packet. It helps to
> understand the format of packet generally. Thanks a lot!
>
>
>
> Tengfei
>
>
>
> On Wed, Jan 20, 2016 at 10:52 AM, Pascal Thubert (pthubert) <
> pthub...@cisco.com> wrote:
>
> This is correct Tengfei, and quite classical.
>
>
>
> Headers are like a stack placed in front of the packet. One builds an
> LOWPAN-IPHC – compressed packet that does not have any RPL artifact in it.
> Then the RPL artifacts are added as 6LoRH headers. We have not imposed an
> order yet but it makes sense to place the RPI first if any, then the RH3 if
> any, then the 6LoRH.
>
>
>
> Would you wish that we impose an order to simplify the parsing?
>
>
>
> Pascal
>
>
>
> *From:* 6tisch [mailto:6tisch-boun...@ietf.org] *On Behalf Of *Tengfei
> Chang
> *Sent:* mercredi 20 janvier 2016 09:20
> *To:* 6tisch@ietf.org
> *Subject:* [6tisch] The "BEFORE" and "AFTER"
>
>
>
> Dear all,
>
>
>
> As recently more discussion in the ML about the format of packet,
> sometimes we say some header after/before the IPv6 header. I would like to
> clarify this.
>
>
>
> 1. For me, I say with the way that mac header is the first header in the
> packet and then, several Routing Headers are AFTER mac header (no mesh
> header/fragmenet header in between).  IPHC header is AFTER those Routing
> Headers.
>
>
>
> 2. However, with the view of constructing a packet, IPHC is first added
> into packet, then RHs are placed AFTER IPHC, MAC header is constructed at
> the end. (I feel pascal is using this way to describe the order of header,
> right?)
>
>
>
> What's the way when we describe something like this?
>
>
>
> Thanks
>
> Tengfei
>
>
>
>
>
>
>
_______________________________________________
6tisch mailing list
6tisch@ietf.org
https://www.ietf.org/mailman/listinfo/6tisch

Reply via email to