phc-discussions - Re: [PHC] NFC v NFD UTF-8 Normalization Re: [Was output specifics]

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Wed, 11 Mar 2015 22:10:02 +0100
From: Thomas Pornin <pornin@...et.org>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] NFC v NFD UTF-8 Normalization Re: [Was output specifics]

On Wed, Mar 11, 2015 at 09:21:06PM +0100, CodesInChaos wrote:
> I think in this case we want the caps-lock switched equivalent, which
> is a question mark `?` for `ß` on the German keyboard layout.

It depends on the operating system...

Right now, I am trzing the German lazout on mz Linux szstem (which swaps
mz Z and mz Y). When I tzpe on 'ß' I get 'ß'. If I use the Shift kez I
get '?'. But if I use CapsLock I get 'ẞ'.

(Back to US layout)

So on that system (pretty standard Lubuntu), the CapsLock switched
equivalent of 'ß' is not '?' (U+003F) but U+01E9E (LATIN CAPITAL LETTER
SHARP S). Things may go differently on other operating systems.

There is a Wikipedia page dedicated to U+01E9E:

   http://en.wikipedia.org/wiki/Capital_%E1%BA%9E

Experimentally, .NET appears unable to map U+1E9E to anything else when
using String.ToLowerInvariant() (even with an explicit German
CultureInfo, which surprises me), so a C#/.NET server receiving a
CapsLock-switched password with that letter from a Linux client may fail
to recover in a Facebook-like behaviour. I would have to dig out the
Unicode casing tables to see who is wrong here, but I think this is
ample proof that such things are subtle and hard to get right.

In the Facebook case, this is a "recovery feature" designed to lower the
rate of calls to whatever serves as help desk, by detecting and fixing
MOST cases of users with the dreadful CapsLock key active. It does not
have to fix all the cases, so a non-unicodly perfect solution is still
useful. Similarly, if they have to work with NFC, I don't think it would
make their life so much harder than now; possibly, they already work
with NFC. In fact, it is possible that right now, Facebook uses "bytes"
and does its recovery attempt by a basic bytewise "xor 32" (it would
take some testing to reverse-engineer it, and since I don't have a
Facebook account I won't do it).

	--Thomas Pornin