mirror/freebsd
1
0
Fork 0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-24 11:29:10 +00:00
Code Issues Releases Activity
3f7112a1f4
freebsd/contrib/wpa_supplicant/eap_testing.txt
Sam Leffler 7700b89cce Import of WPA supplicant 0.5.8
2007-07-11 15:48:36 +00:00

360 lines
12 KiB
Plaintext
Raw Blame History

Automatic regression and interoperability testing of wpa_supplicant's
IEEE 802.1X/EAPOL authentication
Test program:
- Linked some parts of IEEE 802.1X Authenticator implementation from
hostapd (RADIUS client and RADIUS processing, EAP<->RADIUS
encapsulation/decapsulation) into wpa_supplicant.
- Replaced wpa_supplicant.c and wpa.c with test code that trigger
IEEE 802.1X authentication automatically without need for wireless
client card or AP.
- For EAP methods that generate keying material, the key derived by the
Supplicant is verified to match with the one received by the (now
integrated) Authenticator.
The full automated test suite can now be run in couple of seconds, but
I'm more than willing to add new RADIUS authentication servers to make
this take a bit more time.. ;-) As an extra bonus, this can also be
seen as automatic regression/interoperability testing for the RADIUS
server, too.
In order for me to be able to use a new authentication server, the
server need to be available from Internet (at least from one static IP
address) and I will need to get suitable user name/password pairs,
certificates, and private keys for testing use. Other alternative
would be to get an evaluation version of the server so that I can
install it on my own test setup. If you are interested in providing
either server access or evaluation version, please contact me
(j@w1.fi).
Test matrix
+) tested successfully
F) failed
-) server did not support
?) not tested
Cisco ACS ----------------------------------------------------------.
hostapd --------------------------------------------------------. |
Cisco Aironet 1200 AP (local RADIUS server) ----------------. | |
Periodik Labs Elektron ---------------------------------. | | |
Lucent NavisRadius ---------------------------------. | | | |
Interlink RAD-Series ---------------------------. | | | | |
Radiator -----------------------------------. | | | | | |
Meetinghouse Aegis ---------------------. | | | | | | |
Funk Steel-Belted ------------------. | | | | | | | |
Funk Odyssey -------------------. | | | | | | | | |
Microsoft IAS --------------. | | | | | | | | | |
FreeRADIUS -------------. | | | | | | | | | | |
| | | | | | | | | | | |
EAP-MD5 + - - + + + + + - - + +
EAP-GTC + - - ? + + + + - - + -
EAP-OTP - - - - - + - - - - - -
EAP-MSCHAPv2 + - - + + + + + - - + -
EAP-TLS + + + + + + + + - - + +
EAP-PEAPv0/MSCHAPv2 + + + + + + + + + - + +
EAP-PEAPv0/GTC + - + - + + + + - - + +
EAP-PEAPv0/OTP - - - - - + - - - - - -
EAP-PEAPv0/MD5 + - - + + + + + - - + -
EAP-PEAPv0/TLS - + - + + + F + - - - -
EAP-PEAPv1/MSCHAPv2 - - + + + +1 + +5 +8 - + +
EAP-PEAPv1/GTC - - + + + +1 + +5 +8 - + +
EAP-PEAPv1/OTP - - - - - +1 - - - - - -
EAP-PEAPv1/MD5 - - - + + +1 + +5 - - + -
EAP-PEAPv1/TLS - - - + + +1 F +5 - - - -
EAP-TTLS/CHAP + - +2 + + + + + + - + -
EAP-TTLS/MSCHAP + - + + + + + + + - + -
EAP-TTLS/MSCHAPv2 + - + + + + + + + - + -
EAP-TTLS/PAP + - + + + + + + + - + -
EAP-TTLS/EAP-MD5 + - +2 + + + + + + - + -
EAP-TTLS/EAP-GTC + - +2 ? + + + + - - + -
EAP-TTLS/EAP-OTP - - - - - + - - - - - -
EAP-TTLS/EAP-MSCHAPv2 + - +2 + + + + + + - + -
EAP-TTLS/EAP-TLS - - +2 + F + + + - - - -
EAP-SIM +3 - - ? - + - ? - - + -
EAP-AKA - - - - - + - - - - + -
EAP-PSK +7 - - - - - - - - - + -
EAP-PAX - - - - - - - - - - + -
EAP-SAKE - - - - - - - - - - + -
EAP-GPSK - - - - - - - - - - + -
EAP-FAST - - - + - - - - - + - +
LEAP + - + + + + F +6 - + - +
1) PEAPv1 required new label, "client PEAP encryption" instead of "client EAP
encryption", during key derivation (requires phase1="peaplabel=1" in the
network configuration in wpa_supplicant.conf)
2) used FreeRADIUS as inner auth server
3) required a patch to FreeRADIUS to fix EAP-SIM
5) PEAPv1 required termination of negotiation on tunneled EAP-Success and new
label in key deriviation
(phase1="peap_outer_success=0 peaplabel=1") (in "IETF Draft 5" mode)
6) Authenticator simulator required patching for handling Access-Accept within
negotiation (for the first EAP-Success of LEAP)
7) EAP-PSK is not included in FreeRADIUS distribution; used external
rlm_eap_psk implementation from
http://perso.rd.francetelecom.fr/bersani/EAP_PSK/
EAP-PSKWindowsimplementations.html
8) PEAPv1 used non-standard version negotiation (client had to force v1 even
though server reported v0 as the highest supported version)
Automated tests:
FreeRADIUS (1.0pre and CVS snapshot)
- EAP-MD5-Challenge
- EAP-GTC
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / GTC
- EAP-PEAPv0 / MD5-Challenge
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-GTC
- EAP-TTLS / EAP-MSCHAPv2
- EAP-TTLS / CHAP
- EAP-TTLS / PAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-SIM
* not supported in FreeRADIUS
- EAP-PEAP / TLS (Unable to tunnel TLS inside of TLS)
- EAP-TTLS / EAP-TLS (Unable to tunnel TLS inside of TLS)
Microsoft Windows Server 2003 / IAS
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / TLS
- EAP-MD5
* IAS does not seem to support other EAP methods
Funk Odyssey 2.01.00.653
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / GTC
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / GTC
Note: PEAPv1 requires TLS key derivation to use label "client EAP encryption"
- EAP-TTLS / CHAP (using FreeRADIUS as inner auth srv)
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge (using FreeRADIUS as inner auth srv)
- EAP-TTLS / EAP-GTC (using FreeRADIUS as inner auth srv)
- EAP-TTLS / EAP-MSCHAPv2 (using FreeRADIUS as inner auth srv)
- EAP-TTLS / EAP-TLS (using FreeRADIUS as inner auth srv)
* not supported in Odyssey:
- EAP-MD5-Challenge
- EAP-GTC
- EAP-MSCHAPv2
- EAP-PEAP / MD5-Challenge
- EAP-PEAP / TLS
Funk Steel-Belted Radius Enterprise Edition v4.71.739
- EAP-MD5-Challenge
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / MD5
- EAP-PEAPv0 / TLS
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / MD5
- EAP-PEAPv1 / GTC
- EAP-PEAPv1 / TLS
Note: PEAPv1 requires TLS key derivation to use label "client EAP encryption"
- EAP-TTLS / CHAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-MSCHAPv2
- EAP-TTLS / EAP-TLS
Meetinghouse Aegis 1.1.4
- EAP-MD5-Challenge
- EAP-GTC
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / TLS
- EAP-PEAPv0 / GTC
- EAP-PEAPv0 / MD5-Challenge
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / TLS
- EAP-PEAPv1 / GTC
- EAP-PEAPv1 / MD5-Challenge
Note: PEAPv1 requires TLS key derivation to use label "client EAP encryption"
- EAP-TTLS / CHAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-GTC
- EAP-TTLS / EAP-MSCHAPv2
* did not work
- EAP-TTLS / EAP-TLS
(Server rejects authentication without any reason in debug log. It
looks like the inner TLS negotiation starts properly and the last
packet from Supplicant looks like the one sent in the Phase 1. The
server generates a valid looking reply in the same way as in Phase
1, but then ends up sending Access-Reject. Maybe an issue with TTLS
fragmentation in the Aegis server(?) The packet seems to include
1328 bytes of EAP-Message and this may go beyond the fragmentation
limit with AVP encapsulation and TLS tunneling. Note: EAP-PEAP/TLS
did work, so this issue seems to be with something TTLS specific.)
Radiator 3.9 (eval, with all patches up to and including 2004-08-30)
- EAP-MD5-Challenge
- EAP-GTC
- EAP-OTP
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / GTC
- EAP-PEAPv0 / OTP
- EAP-PEAPv0 / MD5-Challenge
- EAP-PEAPv0 / TLS
Note: Needed to use unknown identity in outer auth and some times the server
seems to get confused and fails to send proper Phase 2 data.
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / GTC
- EAP-PEAPv1 / OTP
- EAP-PEAPv1 / MD5-Challenge
- EAP-PEAPv1 / TLS
Note: This has some additional requirements for EAPTLS_MaxFragmentSize.
Using 1300 for outer auth and 500 for inner auth seemed to work.
Note: Needed to use unknown identity in outer auth and some times the server
seems to get confused and fails to send proper Phase 2 data.
- EAP-TTLS / CHAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-GTC
- EAP-TTLS / EAP-OTP
- EAP-TTLS / EAP-MSCHAPv2
- EAP-TTLS / EAP-TLS
Note: This has some additional requirements for EAPTLS_MaxFragmentSize.
Using 1300 for outer auth and 500 for inner auth seemed to work.
- EAP-SIM
- EAP-AKA
Interlink Networks RAD-Series 6.1.2.7
- EAP-MD5-Challenge
- EAP-GTC
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / GTC
- EAP-PEAPv0 / MD5-Challenge
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / GTC
- EAP-PEAPv1 / MD5-Challenge
Note: PEAPv1 requires TLS key derivation to use label "client EAP encryption"
- EAP-TTLS / CHAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-GTC
- EAP-TTLS / EAP-MSCHAPv2
- EAP-TTLS / EAP-TLS
* did not work
- EAP-PEAPv0 / TLS
- EAP-PEAPv1 / TLS
(Failed to decrypt Phase 2 data)
Lucent NavisRadius 4.4.0
- EAP-MD5-Challenge
- EAP-GTC
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MD5-Challenge
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / GTC
- EAP-PEAPv0 / TLS
- EAP-PEAPv1 / MD5-Challenge
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / GTC
- EAP-PEAPv1 / TLS
"IETF Draft 5" mode requires phase1="peap_outer_success=0 peaplabel=1"
'Cisco ACU 5.05' mode works without phase1 configuration
- EAP-TTLS / CHAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-MSCHAPv2
- EAP-TTLS / EAP-GTC
- EAP-TTLS / EAP-TLS
Note: user certificate from NavisRadius had private key in a format
that wpa_supplicant could not use. Converting this to PKCS#12 and then
back to PEM allowed wpa_supplicant to use the key.
hostapd v0.3.3
- EAP-MD5-Challenge
- EAP-GTC
- EAP-MSCHAPv2
- EAP-TLS
- EAP-PEAPv0 / MSCHAPv2
- EAP-PEAPv0 / GTC
- EAP-PEAPv0 / MD5-Challenge
- EAP-PEAPv1 / MSCHAPv2
- EAP-PEAPv1 / GTC
- EAP-PEAPv1 / MD5-Challenge
- EAP-TTLS / CHAP
- EAP-TTLS / MSCHAP
- EAP-TTLS / MSCHAPv2
- EAP-TTLS / PAP
- EAP-TTLS / EAP-MD5-Challenge
- EAP-TTLS / EAP-GTC
- EAP-TTLS / EAP-MSCHAPv2
- EAP-SIM
- EAP-PAX
Cisco Secure ACS 3.3(1) for Windows Server
- PEAPv1/GTC worked, but PEAPv0/GTC failed in the end after password was
sent successfully; ACS is replying with empty PEAP packet (TLS ACK);
wpa_supplicant tries to decrypt this.. Replying with TLS ACK and and
marking the connection completed was enough to make this work.
PEAPv1:
Funk Odyssey 2.01.00.653:
- uses tunneled EAP-Success, expects reply in tunnel or TLS ACK, sends MPPE
keys with outer EAP-Success message after this
- uses label "client EAP encryption"
- (peap_outer_success 1 and 2 work)
Funk Steel-Belted Radius Enterprise Edition v4.71.739
- uses tunneled EAP-Success, expects reply in tunnel or TLS ACK, sends MPPE
keys with outer EAP-Success message after this
- uses label "client EAP encryption"
- (peap_outer_success 1 and 2 work)
Radiator 3.9:
- uses TLV Success and Reply, sends MPPE keys with outer EAP-Success message
after this
- uses label "client PEAP encryption"
Lucent NavisRadius 4.4.0 (in "IETF Draft 5" mode):
- sends tunneled EAP-Success with MPPE keys and expects the authentication to
terminate at this point (gets somewhat confused with reply to this)
- uses label "client PEAP encryption"
- phase1="peap_outer_success=0 peaplabel=1"
Lucent NavisRadius 4.4.0 (in "Cisco ACU 5.05" mode):
- sends tunneled EAP-Success with MPPE keys and expects to receive TLS ACK
as a reply
- uses label "client EAP encryption"
Meetinghouse Aegis 1.1.4
- uses tunneled EAP-Success, expects reply in tunnel or TLS ACK, sends MPPE
keys with outer EAP-Success message after this
- uses label "client EAP encryption"
- peap_outer_success 1 and 2 work
Reference in New Issue View Git Blame Copy Permalink