2 thoughts on - Question On Calculating PJSIP Md5 Authentication With NEC

• Just tracked down the code for the chan_sip MD5 REGISTER and have been able to verify that chan_sip is calculating the HA1 same as I am calculating the md5_cred for PJSIP
  3016:abc@xyz.com:3016

  Both chan_sip and PJSIP REGISTER traces show the uri as sip:10.100.102.82
  So the HA2 for both as MD5(method:uri) (which in this case is REGISTER:sip:10.100.102.82).

  chan_sip response formula with qop detected (sent by NEC) is ha1:nonce:noncecount:cnonce:auth:h2
  Using this formula I am able to match it with my Asterisk chan_sip trace. NEC accepts this REGISTER

  Can anyone point me to the area where Asterisk PJSIP would be doing the response for the REGISTER 401 reply?
  NEC does not like the way PJSIP calculates the response value for the REGISTER reply. I tried to calculate the response exactly like chan_sip does (using the values from the trace and the HA1/HA2) but it’s not matching what the sip trace shows Asterisk sending for the response.

  Does Asterisk PJSIP support handle this or is it all done by PJSIP?

  Dan

  From: asterisk-users On Behalf Of Dan Cropp Sent: Friday, July 12, 2019 2:09 PM
  To: Asterisk Users Mailing List – Non-Commercial Discussion
  Subject: Re: [asterisk-users] Question on calculating PJSIP md5 authentication with NEC

  I have done additional testing and I haven’t been able to figure out why it’s failing.

  Since my original testing we now set the realm on the authentication section to match what we receive from NEC. It’s of the format abc@xyz.com
  I have verified the md5_cred several times and it matches the user:realm:password formula 3016:insiph@something0a646666.com:3016 where username is 3016 and password is 3016

  We suspect it has something to do with the format of the realm that NEC is sending where it may not be working correctly supported by the Asterisk PJSIP code.

  Is there anyone who has used PJSIP outbound REGISTRATION using MD5 support that can provide some insight?
  Or even anyone who know chan_sip’s REGISTER and how it calculates it’s HA1, HA2 for the MD5 authentication?

  Dan

  From: asterisk-users > On Behalf Of Dan Cropp Sent: Wednesday, July 10, 2019 10:48 AM
  To: asterisk-users@lists.digium.com
  Subject: [asterisk-users] Question on calculating the md5_sum

  Using chan_sip, we are able to register with an NEC switch. When I try to REGISTER with PJSIP, the authentication is being rejected. Traces show it’s using md5 authentication. The packets looks almost identical. The one area that I suspect is causing the problem is the md5_cred for my pjsip.conf registration.

  I’m using a Poco MD5 utility to generate the MD5 passing username:realm:password Where username is 3016
  Realm is asterisk (default)
  Password is 3016 which is the same as chan_sip’s secret The value I’m setting the md5_cred in auth section to is 63e8aedc77335879c93123055d21211d

  Would this value match what chan_sip would pass as the md5 credentials?

  Our sip.conf looks like the following…
  [general]
  context = NECTEST
  bindaddr = 0.0.0.0
  bindport = 5060
  websocket_enabled = false srvlookup = no allowguest = yes debug = yes sipdebug = yes defaultexpiry = 480
  deny = 0.0.0.0/24
  permit = 10.100.102.0/24
  permit = 192.168.9.0/24
  canreinvite = yes callcounter = yes register = 3016:3016@10.100.102.82:5060/3016

  [3016]
  type = friend qualify = no nat = no host = 10.100.102.82:5060
  defaultuser = 3016
  secret = 3016
  incominglimit = 24
  accountcode = 33
  port = 5060
  context = NECTEST
  dtmfmode = auto disallow = all allow = ulaw defaultexpiry = 480
  insecure = invite fromdomain = 10.100.102.82
  acl = acl6

  Have a great day!

  Dan

• Thank you Joshua.

  We are confident the problem is with NEC.

  The value is given to PJSIP[1] and it does the rest. You could use the password option and trace how PJSIP is calculating it in that scenario, and then switch to MD5 cred after you determine that and make it match.

  [1] https://github.com/asterisk/asterisk/blob/master/res/res_pjsip_authenticator_digest.c#L180