Re: [Roll] draft-rahul-roll-rpl-observations-00 Section 2.1: Wear leveling

[Michael Richardson <[e-mail]>]

Rahul Arvind Jadhav <[e-mail]> wrote:
    > That's a good point you made about DTSN not been a lollipop counter
    > (this was clearly misinterpreted by us and others in
    > implementations). Section 7 explains "Sequence Counters" but DAO
    > Trigger Sequence Number (DTSN) is not considered as sequence counter in
    > that section!! 

    mcr> I looked again, and I don't see anything about DTSN being a lollipop
    mcr> counter that starts at 240, and wraps at 239.  That would, I think
    mcr> make it clear when a node has booted, and maybe that solves the
    mcr> problem. 

    > [RJ] Very interesting point. We assumed that DTSN is a lollipop counter
    > in our implementation. But looks like it is not. Section 7 explicitly
    > states that only DAOSequence, DODAGVersionNumber, and Path Sequence are
    > lollipop counters ... Looks like we read the spec wrong, also Contiki
    > (including ng) has got it wrong (i.e. uses lollipop counter for
    > DTSN). RIOT (rightly as per spec) does not assume lollipop counter for
    > DTSN.
    
    > I think DTSN needs to be a lollipop counter which alleviates the
    > problem but still does not completely avoid the persistent storage
    > (discussed in other mail in same thread). This point should be added to
    > the draft.

So I think that the errata against 6550 is that it should be a lollipop
counter, and I think that this solves the reboot problem, and eliminates
having to write it to flash.

-- 
]               Never tell me the odds!                 | ipv6 mesh networks [ 
]   Michael Richardson, Sandelman Software Works        | network architect  [ 
]     [e-mail]  http://www.sandelman.ca/        |   ruby on rails    [