Hi Elwyn-

  Inline and trimmed a bit.

...

>> > Specifics:
>> > On-the-wire format:
...
>>
>> EO#
>> Your comments are around the on-wire format for PSC TLVs.  It seems
>> weird to specify the encoding *only* for TLVs and not for the rest of
>> PSC
.....
> The other parts of the PSC don't involve integers and so aren't
> affected.
>
> The usual solution is to add a single statement to the effect that all
> integer fields are encoded as unsigned integers in network bit order.
>
> Assuming things is always dangerous in standards.

All true.  OK, I'll add a such a statement.

...

>> > Malformed messages check:
>> > - Check values of fields prior to TLV Length are consistent with s4.2 of
>> > RFC 6378.
>> > - Check overall length of message matches value in TLV Length + 12.
>> > - Check Sum of lengths of TLVs matches value in TLV Length.
>> > - Check all TLV types received are recognized.
>> >
...
>>
>> Is this level of detail really necessary?
>
> Doubtless that is a matter of opinion, but in my experience of code
> inspection and functional specifications/standards, providing a
> checklist while we are thinking about it can help to avoid the sort of
> security doom that OpenSSL is coping with at the moment.
>

EO#  OK.  I'll add these cases and make clear that they are not the
only ones.  I don't want to make it too heavy as I'm wary of crossing
that line between protocol spec and implementation spec.

I will put something together today or tomorrow and get it out to this
list for review.

thanks!



eric

_______________________________________________
Gen-art mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/gen-art

Reply via email to