On 12/31/2015 04:42 PM, Vincent Lefevre wrote: > On 2015-12-30 16:11:39 +0100, Hans wrote: >> I changed the MAC cause of security purposes in this mail. > > FYI: > > http://security.stackexchange.com/questions/67893/is-it-dangerous-to-post-my-mac-address-publicly
I disagree here: while I don't think it's the end of the world if a MAC address is leaked, I do think it's better to not advertise it. For example, on standard Debian installations the IPv6 privacy extensions aren't activated, so the host part of the IPv6 address (the lower 64 bits) is built using the MAC address. It's bad enough that computers can easily be reidentified via this method (even for remote services), but it becomes worse if the MAC address can easily be associated with a real name via search engines (when e.g. mailing list postings become archived), you can then instantly track users at next to no additional cost. (Sure, without something like Tor there are other ways of doing the same even when using IPv6 privacy extensions, but they are much more expensive to do.) So I do think that it makes sense to keep MAC addresses private when possible, because while it's possible to obtain it in some other way, those are more difficult and one doesn't need to make it too easy for people who want to track oneself. (You shouldn't base authentication on MAC addresses, as was obviously explained in the posting you linked, but that's separate issue.) Off topic: I consider the change in RFC 4941 from RFC 3041 to turn the privacy extensions off by default (which is what Linux does as it implements the RFC correctly) to be idiotic: sure, having your outgoing IPv6 address change dynamically might require application programmers to be a bit more careful if they do in fact rely on the same address, but even IPv4 addresses can change (e.g. in Germany DSL lines are forced to reconnect every 24h for non-technical reasons); and most applications don't require a constant outgoing IP (on non-link-local) anyway, so when weighing the options of what the default should be, I just don't get why the IETF disabled it - the privacy advantages far outweigh the maybe 5 applications that now need to call setsockopt() to tell the kernel that they do need the permanent address. And this is not relevant to servers, which have a fixed IP address anyway. </rant> (I'm of a mind to open a bug that Debian should use privacy extensions by default anyway, regardless of what the RFC says, which Windows and Mac OS X already do, btw.) Regards, Christian
signature.asc
Description: OpenPGP digital signature
