Great summary Tero, thank you.
Can you say more about the reasoning behind "Disallow short addresses in
Nonce generation" ?

ksjp

On Thu, Apr 2, 2015 at 3:36 PM, Tero Kivinen <[email protected]> wrote:

> While I was doing the latest recirculation ballot for 802.15.4rev, I
> noticed something that can affect 6tisch, my comment and proposed
> resolution is:
>
> My Comment:
>
>         The TSCH pan formation should include not explaining that
>         beacons can be secured, but if they are encrypted new devices
>         cannot see the IEs and cannot join the network at all.
>
> Proposed Change:
>
>         Add note: “TSCH beacons cannot be encrypted, they can only be
>         authenticated. This is because if the beacons are encrypted,
>         then the TSCH Synchronization IE used to transmit ASN to
>         joining devices will be encrypted. The joining (or devices who
>         has lost synchronization with network) need to know the ASN
>         before they can decrypt the beacon frame, thus they cannot
>         decrypt the beacons and cannot join the network using
>         encrypted beacons..”
>
> I.e. this will affect the question whether the beacons can be
> encrypted with default key, which was one of the suggestions for the
> 6tisch drafts.
>
> This is something that is unchanged since 802.15.4-2011 and
> 802.15.4e-2012, i.e. the TSCH ASN has been carried in the Payload IEs,
> and Payload IEs are encrypted if security level >= 4. And to be able
> to decrypt the payload, you need to know ASN, which you cannot see as
> it is in encrypted part...
>
> The reason for this email is that people pointed me that there has
> been discussion about what changes there has been in the revision that
> would affect 6tisch and this is the latest one I found (just adding
> text stating the fact that has been true already before).
>
> There is also changes that might affect implementations, and those
> changes are fixing security issues:
>
>         * Disallow short addresses in Nonce generation
>
>         * Removal of encrypt only mode (this caused easy way to attack
>           TSCH mode, i.e. attacker could pick any encrypted and
>           authenticated frame from the air, remove the MIC, change the
>           security level in header to be encrypt only, then modify the
>           packet at will, and then send it to final destination, which
>           would then decrypt and receive the frame without noticing
>           anything, this was not problem for non-TSCH mode, as
>           security level was used to generate Nonce).
>
>         * Adding ability to filter whether MAC acts on information
>           elements. The 802.15.4e didn't add any method of saying
>           whether we should process information elements we receive in
>           the frame. I.e. the implicit default was that they were all
>           processed by MAC.
>
>         * Fixes in the inbound security processing to clearly allow
>           unsecured and secured frames to be processed at the same
>           time.
>
>         * Fixes in the inbound and outbound security processing for
>           TSCH saying that we do not keep normal 32-bit FrameCounter
>           when using TSCH mode, 802.15.4e didn't modify anything in
>           inbound or outbound security porcedures, meaning that
>           outbound processing rules still kept incrementing 32-bit
>           macFrameCounter for each outgoinf frame, and stopping at
>           counter error when it reached max value, even when it is not
>           used at all in TSCH mode. Also in the inbound rules it did
>           still got the frame counter from frame (not there in TSCH)
>           and compared it against previous value etc...
>
>         * Defined how new frames added in 802.15.4e are actually
>           encrypted and MACed, the 802.15.4e did not modify the table
>           53 listing private and open payloads, thus it was not clear
>           how those are encrypted etc.
>
>         * Changed the MAC command identifier to be encrypted for frame
>           versions 0b10. MAC command fames with Frame versions 0b10
>           can have payload IEs, and those are encrypted. The MAC
>           command identifier is after Payload IEs, and it would be in
>           clear, then the device would need to first n-bytes of the
>           frame (n is unknown, before the frame is decrypted, because
>           payload IEs have termination IE at the end), then get next
>           byte as is, i.e. without decrypting it, and then decrypt
>           rest of the frame. When polled from the 4e implementors they
>           said they do nto do that, but they will encrypt and decrypt
>           the MAC command identifier too for frame version 0b10. This
>           also caused changes to the inbound state machine as now we
>           cannot run filters based on the MAC command identifier for
>           MAC command frames before we decrypt the frame.
>
>         * Most likely still something more that I have forgotten...
>           And of course some changes which do not affect TSCH, or
>           which do not affect old devices (like adding new
>           secFrameCounterPerKey property for key).
>
> Ps. I have been too busy to really follow the 6tisch mailing list, but
> perhaps I need to find time to do it. I did read the "removing the 'e'
> thread" in the archives, and I think the correct thing would be to
> refer to the 802.15.4 without any year or specific list of amendments.
> --
> [email protected]
>
> _______________________________________________
> 6tisch mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/6tisch
>
_______________________________________________
6tisch mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6tisch

Reply via email to