#948 HIGH Trial-3: NM: key only allows hex entry

[Zarro Boogs per Child]

#948: NM: key only allows hex entry
---------------------+------------------------------------------------------
  Reporter:  bert    |       Owner:  tomeu  
      Type:  defect  |      Status:  new    
  Priority:  high    |   Milestone:  Trial-3
 Component:  sugar   |     Version:         
Resolution:          |    Keywords:         
  Verified:  0       |  
---------------------+------------------------------------------------------
Changes (by tomeu):

  * owner:  dcbw => tomeu


Comment:

 Sep 21 23:00:21 <dcbw>  tomeu: sure; WEP is an md5-based hash algorithm,
 WPA is sha1 based.  but I don't think you can just md5/sha1 the data
 straight
 Sep 21 23:00:53 <tomeu> dcbw: I cannot feed the s:<ascii> string directly
 to NM?
 Sep 21 23:00:59 <dcbw>  tomeu: no
 Sep 21 23:01:04 <dcbw>  tomeu: hashing is done client-side, not NM side
 Sep 21 23:01:09 <dcbw>  tomeu: I have no idea what bert's talking about
 Sep 21 23:01:34 <dcbw>  tomeu: I tihnk he's talking about ASCII
 passphrases which nobody actually uses anymore
 Sep 21 23:01:36 <tomeu> dcbw: gnome's nm applet allows entering an ascii
 passphrase
 Sep 21 23:01:41 <tomeu> dcbw: well...
 Sep 21 23:01:46 <tomeu> dcbw: I use...
 Sep 21 23:01:48 <dcbw>  tomeu: yes, and converts it internally to a hex
 key, which gets sent to NM
 Sep 21 23:01:55 <tomeu> that's why I want to do it
 Sep 21 23:02:04 <dcbw>  tomeu: NM does not do passphrase hashing, the
 client (in this case sugar) does the hashing
 Sep 21 23:03:55 <tomeu> dcbw: ok, will look at it
 Sep 21 23:08:18 <dcbw>  tomeu: it's not horrible for WEP at least:
 http://svn.gnome.org/viewcvs/NetworkManager/branches/NETWORKMANAGER_0_6_0_RELEASE
 /libnm-util/cipher-wep-passphrase.c?view=markup
 Sep 21 23:08:28 <dcbw>  tomeu: you care about
 cipher_wep_passphrase_hash_func()
 Sep 21 23:08:44 <tomeu> wow, great, dan!
 Sep 21 23:08:52 <dcbw>  tomeu: which repeats the data over 64-bytes, and
 then hashes it with md5
 Sep 21 23:08:57 <dcbw>  tomeu: so wep is easy
 Sep 21 23:09:57 <dcbw>  tomeu: WPA is harder; for wpa, look at
 http://svn.gnome.org/viewcvs/NetworkManager/branches/NETWORKMANAGER_0_6_0_RELEASE
 /libnm-util/sha1.c?view=markup and you care about pbkdf2_sha1()
 Sep 21 23:11:22 <tomeu> dcbw: regarding the UI, should we try to
 autodetect the format the key has been entered, or just add a checkbox to
 the dialog?
 Sep 21 23:11:49 <tomeu> dcbw: we have also the possibility of expecting
 passphrases to be prefixed by s:
 Sep 21 23:12:15 <dcbw>  tomeu: it is impossible to autodetect WEP
 passphrase types
 Sep 21 23:12:16 <tomeu> dcbw: I guess we don't need to bother eben about
 it, as it's not part of the main user experience
 Sep 21 23:12:26 <dcbw>  tomeu: so for that, my suggestion is to turn
 things around
 Sep 21 23:12:49 <dcbw>  tomeu: require $:<hex key> for hex, default WEP is
 md5 passphrase, and ":<ascii passphrase> for ascii passphrse
 Sep 21 23:13:48 <tomeu> dcbw: same thing for wpa?
 Sep 21 23:14:10 <dcbw>  tomeu: right, default to passphrase and must enter
 $:<key> for hex
 Sep 21 23:14:17 <dcbw>  tomeu: except that WPA makes it easy
 Sep 21 23:14:25 <dcbw>  tomeu: because the hex key _must_ be 64-bytes
 Sep 21 23:14:37 <dcbw>  tomeu: but WPA passphrase can only be from 8 - 63
 bytes, so length works there
 Sep 21 23:14:49 <dcbw>  tomeu: technically we wouldn't need any hints for
 WPA
 Sep 21 23:15:03 <dcbw>  tomeu: not sure if we want to keep the prefix if
 just for consistency
 Sep 21 23:15:22 <tomeu> yeah, perhaps would be better
 Sep 21 23:15:42 <dcbw>  tomeu: actually, lets just drop prefix for WPA and
 only use it for WEP
 Sep 21 23:15:57 <dcbw>  tomeu: it's easier to convert to prefix later than
 it is to change from prefix->key

-- 
Ticket URL: <https://dev.laptop.org/ticket/948#comment:4>
One Laptop Per Child <https://dev.laptop.org>
OLPC bug tracking system