It looks like GnuTLS (at least newer versions) is using random padding
on the application data and the previously used 100 byte extra buffer
for tls_connection_encrypt() calls was not enough to handle all cases.
This resulted in semi-random authentication failures with EAP-PEAP and
EAP-TTLS during Phase 2.
Increase the extra space for encryption from 100 to 300 bytes and add an
error message into tls_gnutls.c to make it easier to notice this issue
should it ever show up again even with the larger buffer.
This adds mostly feature complete external Registrar support with the
main missing part being proper support for multiple external Registrars
working at the same time and processing of concurrent registrations when
using an external Registrar.
This code is based on Sony/Saice implementation
(https://www.saice-wpsnfc.bz/) and the changes made by Ted Merrill
(Atheros) to make it more suitable for hostapd design and embedded
systems. Some of the UPnP code is based on Intel's libupnp. Copyrights
and licensing are explained in src/wps/wps_upnp.c in more detail.
Previous version assumed that the Flags field is always present and
ended up reading one octet past the end of the buffer should the Flags
field be missing. The message length would also be set incorrectly
(size_t)-1 or (size_t)-5, but it looks like reassembly code ended up
failing in malloc before actually using this huge length to read data.
RFC 2716 uses a somewhat unclear description on what exactly is included
in the TLS Ack message ("no data" can refer to either Data field in 4.1
or TLS Data field in 4.2), so in theory, it would be possible for some
implementations to not include Flags field. However,
EAP-{PEAP,TTLS,FAST} need the Flags field in Ack messages, too, for
indicating the used version.
The EAP peer code will now accept the no-Flags case as an Ack message if
EAP workarounds are enabled (which is the default behavior). If
workarounds are disabled, the message without Flags field will be
rejected.
[Bug 292]
We need to be a bit more careful when removing the WPS configuration
block since wpa_s->current_ssid may still be pointing at it. In
addition, registrar pointer in wps_context will need to be cleared
since the context data is now maintained over multiple EAP-WSC runs.
Without this, certain WPS operations could have used freed memory.
Windows Server 2008 NPS gets very confused if the TLS Message Length is
not included in the Phase 1 messages even if fragmentation is not used.
If the TLS Message Length field is not included in ClientHello message,
NPS seems to decide to use the ClientHello data (excluding first six
octets, i.e., EAP header, type, Flags) as the OuterTLVs data in
Cryptobinding Compound_MAC calculation (per PEAPv2; not MS-PEAP)..
Lets add the TLS Message Length to PEAPv0 Phase 1 messages to get rid of
this issue. This seems to fix Cryptobinding issues with NPS and PEAPv0
is now using optional Cryptobinding by default (again) since there are
no known interop issues with it anymore.
Changed peer to derive the full key (both MS-MPPE-Recv-Key and
MS-MPPE-Send-Key for total of 32 octets) to match with server
implementation.
Swapped the order of MPPE keys in MSK derivation since server
MS-MPPE-Recv-Key | MS-MPPE-Send-Key matches with the order specified for
EAP-TLS MSK derivation. This means that PEAPv0 cryptobinding is now
using EAP-MSCHAPv2 MSK as-is for ISK while EAP-FAST will need to swap
the order of the MPPE keys to get ISK in a way that interoperates with
Cisco EAP-FAST implementation.
This allows the same source code file to be shared for both methods. For
now, this is only in eap_aka_prime.c, but eventually, changes in
eap_aka_prime.c are likely to be merged into eap_aka.c at which point
the separate eap_aka_prime.c can be removed.
This is just making an as-is copy of EAP-AKA server and peer
implementation into a new file and by using the different EAP method
type that is allocated for EAP-AKA' (50). None of the other differences
between EAP-AKA and EAP-AKA' are not yet included.
It is likely that once EAP-AKA' implementation is done and is found to
work correctly, large part of the EAP-AKA and EAP-AKA' code will be
shared. However, it is not reasonable to destabilize EAP-AKA
implementation at this point before it is clearer what the final
differences will be.
The wps_context data is now managed at wpa_supplicant, not EAP-WSC. This
makes wpa_supplicant design for WPS match with hostapd one and also
makes it easier configure whatever parameters and callbacks are needed
for WPS.
Previously, wpa_supplicant as Enrollee case was handled using a
different callback function pointer. However, now that the wps_context
structure is allocated for all cases, the same variable can be used in
all cases.
Previously, hardcoded values were used in wps_enrollee.c. These are now
moved into shared data in struct wps_context. In case of
AP/Authenticator, these are initialized in wps_hostapd.c. In case of
client/supplicant, these are now initialized in EAP-WSC peer method,
but will probably end up being moved into higher layer for better
configuration.
EAP-WSC peer method for
The old (i.e., currently used) AP Settings are processed. For now, they
are copied as-is into M8 as new AP Settings to avoid changing
configuration. This should be changed to allow external programs (e.g.,
GUI) to fetch the old AP settings over ctrl_iface and then allow
settings to be changed before sending M8 with the new settings.
This adds WPS support for both hostapd and wpa_supplicant. Both programs
can be configured to act as WPS Enrollee and Registrar. Both PBC and PIN
methods are supported.
Currently, hostapd has more complete configuration option for WPS
parameters and wpa_supplicant configuration style will likely change in
the future. External Registrars are not yet supported in hostapd or
wpa_supplicant. While wpa_supplicant has initial support for acting as
an Registrar to configure an AP, this is still using number of hardcoded
parameters which will need to be made configurable for proper operation.
It looks like [MS-PEAP] 3.2.5.6 points towards this being the expected
behavior (however, that chapter is very confusing).
In addition, remove Cryptobinding TLV from response if the received
Cryptobinding TLV is not valid. Add some more debug messages to the case
where the received Cryptobinding TLV is found invalid.
I fixed the engine issue in phase2 of EAP-TTLS. The problem was that you
only defined one engine variable, which was read already in phase1. I
defined some new variables:
engine2
engine2_id
pin2
and added support to read those in phase2 wheres all the engine
variables without number are only read in phase1. That solved it and I
am now able to use an engine also in EAP-TTLS phase2.
It looks like ACS did not like PAC Acknowledgment TLV before Result TLV, so
reorder the TLVs to match the order shown in a
draft-cam-winget-eap-fast-provisioning-09.txt example. This allows
authenticated provisioning to be terminated with Access-Accept (if ACS has
that option enabled). Previously, provisioning was otherwise successful,
but the server rejected connection due to not understanding the PAC Ack
("Invalid TEAP Data recieved").
Previously, hardcoded identity in the network configuration skipped both
IMSI reading and PIN verification. This broke cases where PIN is needed for
GSM/UMTS authentication. Now, only IMSI reading is skipped if identity is
hardcoded.
This adds all the attributes that are marked as mandatory for SoH in
IF-TNCCS-SOH v1.0. MS-Machine-Inventory does not contain correct data
(i.e., all version fields are just marked as inapplicable) and
MS-MachineName is hardcoded to wpa_supplicant@w1.fi for now.
The change to support fragmentation added extra function to generate the
EAP header, but forgot to remove the original code and ended up getting two
EAP headers and TNC flags field in the generated message. These header
fields need to be added only in the function that builds the final message
(and if necessary, fragments the data).
Even though we try to disable TLS compression, it is possible that this
cannot be done with all TLS libraries. For example, OpenSSL 0.9.8 does not
seem to have a configuration item for disabling all compression (0.9.9 has
such an option). If compression is used, Phase 2 decryption may end up
producing more data than the input buffer due to compressed data. This
shows up especially with EAP-TNC that uses very compressible data format.
As a workaround, increase the decryption buffer length to (orig_len+500)*3.
This is a hack, but at least it handles most cases. TLS compression should
really be disabled for EAP use of TLS, but since this can show up with
common setups, it is better to handle this case.
Process SoH Request in SoH EAP Extension method and reply with SoH TLV. The
contents of SSoH is not yet complete (i.e., some of the required attributes
are still missing). Use of SoH is disabled by default; it can be enabled
with tnc=soh in phase1 parameter for PEAP.
crypto_binding=# in phase1 network parameter can now be used to change the
PEAPv0 cryptobinding behavior (0 = do not use, 1 = optional (default),
2 = required).
The previous version was deriving unnecessarily large amount of output data
for the implicit challenge. With the TLS PRF, this was not causing any
problems because the output length is not explicitly bound into the
derivation. Anyway, it is better to use the correct output length should
the PRF be ever changed to a one that uses the output length to change the
beginning of the output data.
It looks like Microsoft implementation does not match with their
specification as far as PRF+ label usage is concerned.. IPMK|CMK is derived
without null termination on the label, but the label for CSK derivation
must be null terminated.
This allows cryptobinding to be used with PEAPv0 in a way that
interoperates with Windows XP SP3 (RC2) and as such, this functionality is
now enabled as an optional addition to PEAPv0.
Microsoft uses reverse order for MS-MPPE keys in EAP-PEAP when compared to
EAP-FAST derivation of ISK. Swap the keys here to get the correct ISK for
EAP-PEAPv0 cryptobinding. This resolves the cryptobinding interop issue
with WinXP SP3 RC2. However, it looks like MSK derivation does not
interoperate when using cryptobinding, so this code remains disabled for
the time being.
This implementation is complete for PEAPv0 (Microsoft extension), but the
use of cryptobinding is disabled in this version, i.e., this does not
change protocol behavior at all.
Interop tests between hostapd and wpa_supplicant work fine, but there are
some problems in getting hostapd to interoperate with Windows XP SP3 RC2
(peer replies with Result Failure to the attempted cryptobinding). The
implementation will remain disabled until this issue has been resolved.
EAP-PEAP was the only method that used the external eap_tlv.c peer
implementation. This worked fine just for the simple protected result
notification, but extending the TLV support for cryptobinding etc. is not
trivial with such separation. With the TLV processing integrated into
eap_peap.c, all the needed information is now available for using
additional TLVs.
Need to process EAP AVP after the non-EAP Phase 2 method. In addition,
EAP-TTLS/MSCHAPv2 needs special code for handling the starting of TNC after
extra roundtrip of MSCHAPv2 success message.
Number of TLVs were processed in groups and these cases were now separated
into more flexible processing of one TLV at the time. wpabuf_concat()
function was added to make it easier to concatenate TLVs. EAP Sequences are
now supported in both server and peer code, but the server side is not
enabled by default.
