Hi Renato,


> On Aug 11, 2023, at 9:41 PM, Renato Westphal <[email protected]> wrote:
> 
> Em dom., 16 de jul. de 2023 às 22:37, Christian Hopps
> <[email protected]> escreveu:
>>> On Jul 13, 2023, at 17:23, Acee Lindem <[email protected]> wrote:
>>> 
>>> Hi Renato,
>>>> 
>>>> Lastly, this might just be a small nitpick of mine, but I don't think
>>>> having a "length" leaf for all TLVs and Sub-TLVs adds much value. In
>>>> my opinion, it's only relevant for unknown TLVs that couldn't be
>>>> decoded; otherwise, it just adds unnecessary noise. If we take a look
>>>> at the IS-IS model, for instance, we can see that it doesn't have a
>>>> "length" leaf for the LSP TLVs and Sub-TLVs.
>>> 
>>> I’ve removed the three "length" leaves that were fixed length. I left the 
>>> ones that were variable due to contained sub-TLVS.
>>> Are you saying that these TLVs would be malformed if the length weren’t 
>>> correct?
>> 
>> I wonder what one would do with these length values? Only "unknown" things 
>> have a value leaf
>> 
>>  `+--ro value? yang:hex-string`
>> 
>> And in that case the length is given by the length of the hex-string.
> 
> Exactly. I'd probably keep the "length" leaf for unknown TLVs/Sub-TLVs
> to be consistent with other modules (including the OSPF base module).
> But I don't see much point for known TLVs/Sub-TLVs.
> 
>> I definitely think there could be value with having the length field when 
>> things are malformed i.e., where the `length` field in the [sub-]TLV 
>> disagrees with the actual length of the `yang:hex-string` value. In this 
>> case though we might want to call it out as `malformed-length` or something.
> 
> That's an interesting suggestion. In practice, however, I think most
> implementations will reject LSAs containing malformed TLVs/Sub-TLVs as
> a security measure.

Ok - I’ve relented and removed the TLV lengths for the TLVs/Sub-TLVs with the 
potential to have nested Sub-TLVs. I agree that including it isn’t the right 
way to debug LSA encoding issues. I also noted that we didn’t include lengths 
for TLVs with nested Sub-TLVs in RFC 9129. 



> 
>> On Jul 13, 2023, at 10:32, Renato Westphal <[email protected]> wrote:
>> [snip]
>> I think the "ospfv3-lsa-prefix" grouping from the base
>> module could be reused since it's identical, except it doesn't include
>> the "prefix-length" leaf.
> 
> So, it turns out it wasn't a good idea to reuse that grouping from the
> base module. The reason is that its "prefix-options" leaf-list uses an
> identityref to "ospfv3-prefix-option", whereas the
> "ietf-ospfv3-extended-lsa" module should use an identityref to
> "ospfv3-e-prefix-option" instead (notice the "e-" in the middle). My
> apologies for the bad suggestion ^^

I’ve restored the prior grouping only without prefix-length is already include 
in the ip-prefix type. 

These changes are in the -22 version that I just posted. 

Thanks,
Acee




> 
> Regards,
> -- 
> Renato Westphal

_______________________________________________
Lsr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lsr

Reply via email to