Joy Latten wrote:

> > I think Tero's proposal about just noting this fact (i.e. not 
> > changing how this work) would be OK and sufficient.
> 
> I could be missing something, but RFC4301, section 4.1 allows
> implementations to use the SPI in conjunction with the IPsec protocol
> for SA identification. So, if someone is in that latter case, wouldn't
> they have a problem? 

Well... depends on whether the recipient of the notification actually
uses the SPI value for something (other than possibly debugging/logging).

The "INVALID_SPI" notification basically means "I've rebooted, or our
understanding of IPsec/IKEv2 state is otherwise screwed up".  If this
was an unprotected one-way notification, the recipient would it as a
hint that things might be wrong, and initiates a liveness test for the
IKE_SA.  If it was a protected notification, it probably means an
implementation bug somewhere, and a possible action would be to create
a new IKE_SA (and new CHILD_SAs) from scratch. In neither case, the
recipient really needs the SPI value for anything..

Best regards,
Pasi
_______________________________________________
IPsec mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ipsec

Reply via email to