RSN: Ignore IGTK configuration errors with swapped KeyID values
There are number of deployed APs with broken PMF implementation where
the IGTK KDE uses swapped bytes in the KeyID field (0x0400 and 0x0500
instead of 4 and 5). Such APs cannot be trusted to implement BIP
correctly or provide a valid IGTK, so do not try to configure this key
with swapped KeyID bytes. Instead, continue without configuring the IGTK
so that the driver can drop any received group-addressed robust
management frames due to missing keys.
Normally, this error behavior would result in us disconnecting, but
there are number of deployed APs with this broken behavior, so as an
interoperability workaround, allow the connection to proceed.
RSN: Report completion only after IGTK configuration
Previously wpa_supplicant_key_neg_complete() was called before the
attempt to configure the IGTK received from the authenticator. This
could resulted in somewhat surprising sequence of events if IGTK
configuration failed since completion event would be followed by
immediate disconnection event. Reorder these operations so that
completion is reported only if GTK and IGTK are configurated
successfully.
Furthermore, check for missing GTK KDE in case of RSN and handle that
with an explicit disconnection instead of waiting for the AP to deliver
the GTK later.
SAE: Use open authentication to reassociate for PMKSA caching
For reassociation with the same AP wpa_supplicant attempts to use cached
PMKSA. For this purpose PMKID is passed in RSNE in (Re)Association
Request frame. In the case of SAE AP, open authentication shall be used
during reassociation. Otherwise cached PMKID becomes invalid after full
SAE authentication.
The previous implementation correctly handles SME-in-wpa_supplicant
cases. However SME-in-driver cases, complete SAE authentication is
performed. As a result, first reassociation attempt fails.
Fix SME-in-driver behavior by reseting authentication algorithm to
WPA_AUTH_ALG_OPEN when reassociating with SAE AP with an existing PMKSA
cache entry.
Fix FILS ERP association event with 4-way HS offload drivers
When FILS authentication is used with ERP, no EAPOL frames are expected
after association. However, for drivers that set the
WPA_DRIVER_FLAGS_4WAY_HANDSHAKE_8021X capability flag, the EAP state
machine was not configured correctly and was waiting for EAPOL frames,
which leads to disconnection.
Fix this by reordering the if branches to set the EAPOL/EAP state
machines to success when FILS authentication was already completed.
When processing the NL80211_CMD_PROBE_CLIENT command response, the
nl80211 layer in the kernel sends a response containing the cookie
associated with the client probe request. This response was not handled
by driver_nl80211.c when sending the command, and it was mistakenly
handled as an asynchronous event. This incorrect event did not include
the MAC/ACK attributes, so it was ignored in practice, but nevertheless,
the command response should not be processed as an event.
Fix this by reading the response as part of the sending the command
flow.
SA Query wasn't stopped after disconnection, which could potentially
result in an unexpected SA timeout firing later when already connected
to another AP. Fix that by stopping SA Query when an association is
terminated.
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
pending_p2ps_group flag is not always cleaned, which may later result
in an unexpected GO bring up, after PD response is transmitted in
wpas_prov_disc_resp_cb().
This can be seen when running the following hwsim tests together:
- p2ps_channel_sta_connected_disallow_freq_mcc
- p2ps_channel_active_go_and_station_different_mcc
Fix this by clearing pending_p2ps_group flag also when processing new
PD requests. In addition, set this flag only when really needed.
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
mka: Avoid memory leak in unexpected case in RECEIVE
It looks like it is possible for the RECEIVE state to leak memory where
a previously allocated sm->lki is moved to sm->oki while sm->oki is
pointing to not yet freed entry. It is not clear how this can be
triggered, but it has come up in hwsim testing under heavy load.
Free sm->oki if it is still set in RECEIVE before replacing it with
sm->lki to avoid this memory leak.
If os_get_random() fails, llid must not be read/used since it could be
uninitialized. Handle that special case by clearing llid explicitly to 0
to continue iteration.
mesh: Fix operations after SAE state machine removing the STA
It is possible for the SAE state machine to remove the STA and free the
sta pointer in the mesh use cases. handle_auth_sae() could have
dereferenced that pointer and used freed memory in some cases. Fix that
by explicitly checking whether the STA was removed.
Fixes: bb598c3bdd06 ("AP: Add support for full station state") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
ap_free_sta() frees the sta entry, so sta->addr cannot be used after
that call. Fix the sequence of these two calls to avoid use of freed
memory to determine which PMKSA cache entry to remove.
Fixes: 9f2cf23e2e0d ("mesh: Add support for PMKSA caching") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Fix hostapd BSS_TM_REQ handling of bss_term parameter
The TSF field in BSS termination information was not cleared correctly.
It was supposed to be cleared to all zeros, but the memset call did not
point at offset 2; instead, it cleared it with 0x02 octets and also
cleared the subelement header with 0x02 octets while leaving two last
octets uninitialized.
Do not use the buf argument to wpas_ctrl_iface_get_pref_freq_list() in
the debug print since it points to the response buffer ("OK\n" without
null termination). This was supposed to print the cmd argument instead.
The previous version ended up using uninitialized heap memory from the
response buffer.
Fixes: 983422088f00 ("nl80211: Add means to query preferred channels") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
D-Bus: Fix P2P GroupStarted signal not to use uninitialized IP buffer
wpas_p2p_completed() used wpa_sm_get_p2p_ip_addr() return value properly
for filling in the ip_addr[] string only if the IP addresses got
assigned, but the raw ip[] array was passed to
wpas_dbus_signal_p2p_group_started() regardless of whether the IP
addresses were assigned. This could result in using uninitialized stack
memory for constructing the GroupStarted signal values. Fix this by
filling those in only if the IP addressed are actually available.
tests: Avoid confusing "DETACH failed" exception prints in D-Bus tests
dbus_p2p_go_neg_init, dbus_p2p_group_idle_timeout, and
dbus_p2p_group_termination_by_go could end up print a "DETACH failed"
exception as a warning from WpaSupplicant.__del__ for the dev1 instance
used within the TestDbusP2p class. This did not cause the test cases to
fail, but the output is a bit confusing, so clean this up be explicitly
closing the control interface monitor sockets and furthermore by
ignoring the "DETACH failed" exception within __del__.
This external tool was used for testing data connectivity with
mac80211_hwsim. However, the last user for it was removed in 2014 by
commit 1131a1c8d29c ("tests: Replace last remaining hwsim_test uses with
DATA_TEST"), so there is not point in maintaining this forgotten test
tool in the repository anymore.
nl80211: Update assoc_freq and bss->freq based on real association info
Move event.assoc_info.freq selection to be after the
nl80211_get_assoc_ssid() call so that the current cfg80211 information
on the operating channel can be used should anything unexpected have
happened between the association request and completion of association.
Furthermore, update bss->freq based on assoc_freq to make that
information a bit more useful for station mode. It was already updated
after channel switches during association, but not at the beginning of
association.
nl80211: Clear bss->freq on station mode disconnection
This fixes some issues where bss->freq could have been used to replace
the current operating channel when sending out a management frame.
bss->freq has not been consistently used to track the current operating
channel in station mode, so it should not be trusted for this type of
uses. Clearing it makes this a bit more robust by at least avoiding the
cases of information from past association being used.
EAP-pwd: Remove unused checks for cofactor > 1 cases
None of the ECC groups supported in the implementation had a cofactor
greater than 1, so these checks are unreachable and for all cases, the
cofactor is known to be 1. Furthermore, RFC 5931 explicitly disallow use
of ECC groups with cofactor larger than 1, so this checks cannot be
needed for any curve that is compliant with the RFC.
Remove the unneeded group cofactor checks to simplify the
implementation.
tests: Remove testing of EAP-pwd groups 25, 26, and 27
This is in preparation of disallowing all use of these groups. Negative
test case for the groups will be added in a separate commit after the
implementation has been changed.
OpenSSL: Fix server side openssl_ecdh_curves configuration with 1.0.2
It looks like SSL_CTX_set1_curves_list() command alone is not sufficient
to enable ECDH curve selection with older OpenSSL versions for TLS
server, so enable automatic selection first and specify the exact list
of curves after that.
This fixes failures in openssl_ecdh_curves test case when hostapd uses
OpenSSL 1.0.2.
tests: Make regdom clearing in WNM tests more robust
It looks like the scan operation could end up reverting regdom back to
the previously configured one, so configure 00 country before starting
the disconnect-and-stop-scan operation to give some more time for the
regdom to be cleared.
tests: Update rrm_beacon_req_last_frame_indication to match implementation
The last beacon report indication was set in all the beacon report
elements in the last frame of the beacon report, while it should be set
only in the last beacon report element of the last frame. This is now
fixed in wpa_supplicant, so update the test case expectation to match
the fixed behavior.
RRM: Set last beacon report indication in the last element only
The last beacon report indication was set in all the beacon report
elements in the last frame of the beacon report, while it should be
set only in the last beacon report element of the last frame.
Fixes: ecef0687dc33 ("RRM: Support for Last Beacon Report Indication subelement") Signed-off-by: Avraham Stern <avraham.stern@intel.com>
Remove unnecessary copying of SSID and BSSID for external_auth
The external authentication command and event does not need to copy the
BSSID/SSID values into struct external_auth since those values are used
before returning from the call. Simplify this by using const u8 * to
external data instead of the array with a copy of the external data.
Srinivas Dasari [Mon, 18 Mar 2019 09:34:04 +0000 (15:04 +0530)]
hostapd: Support external authentication offload in AP mode
Extend commit 5ff39c1380d9 ("SAE: Support external authentication
offload for driver-SME cases") to support external authentication
with drivers that implement AP SME by notifying the status of
SAE authentication to the driver after SAE handshake as the
driver acts as a pass through for the SAE Authentication frames.
Srinivas Dasari [Mon, 25 Feb 2019 12:22:05 +0000 (17:52 +0530)]
nl80211: External authentication in driver-based AP SME mode
This extends driver interface to nl80211 by introducing the following
changes,
1. Register for Authenication frames in driver-based AP SME mode.
2. Advertise NL80211_ATTR_EXTERNAL_AUTH_SUPPORT in set_ap when
offloaded SAE authentication is supported.
3. Extend the NL80211_CMD_EXTERNAL_AUTH interface to also send PMKID
so that the drivers can respond to the PMKSA cached connection
attempts from the stations avoiding the need to contact user space
for all PMKID-based connections.
4. Send external auth status to driver only if it is a driver based
SME solution.
beacon_set_done did not get reset to zero on disabling interface using
DISABLE control interface command and the subsequent ENABLE command will
caused configuration of Beacon/Probe Response/Association Response frame
IEs twice. The unnecessary two step configuration can be avoided by
resetting beacon_set_done on DISABLE so that ENABLE can bring up the
interface in a single step with fully updated IEs.
EAP-pwd: Get rid of unnecessary allocation of temporary buffer
Binary presentations of element and scalar can be written directly to
the allocated commit message buffer instead of having to first write
them into temporary buffers just to copy them to the actual message
buffer.
EAP-pwd: Enforce 1 < rand,mask < r and rand+mask mod r > 1
RFC 5931 has these conditions as MUST requirements, so better follow
them explicitly even if the rand,mask == 0 or rand+mask == 0 or 1 cases
are very unlikely to occur in practice while generating random values
locally.
This adds an explicit check for 0 < x,y < prime based on RFC 5931,
2.8.5.2.2 requirement. The earlier checks might have covered this
implicitly, but it is safer to avoid any dependency on implicit checks
and specific crypto library behavior. (CVE-2019-9498 and CVE-2019-9499)
Furthermore, this moves the EAP-pwd element and scalar parsing and
validation steps into shared helper functions so that there is no need
to maintain two separate copies of this common functionality between the
server and peer implementations.
Mathy Vanhoef [Sun, 31 Mar 2019 15:43:44 +0000 (17:43 +0200)]
EAP-pwd client: Verify received scalar and element
When processing an EAP-pwd Commit frame, the server's scalar and element
(elliptic curve point) were not validated. This allowed an adversary to
bypass authentication, and act as a rogue Access Point (AP) if the
crypto implementation did not verify the validity of the EC point.
Fix this vulnerability by assuring the received scalar lies within the
valid range, and by checking that the received element is not the point
at infinity and lies on the elliptic curve being used. (CVE-2019-9499)
The vulnerability is only exploitable if OpenSSL version 1.0.2 or lower
is used, or if LibreSSL or wolfssl is used. Newer versions of OpenSSL
(and also BoringSSL) implicitly validate the elliptic curve point in
EC_POINT_set_affine_coordinates_GFp(), preventing the attack.
Mathy Vanhoef [Sun, 31 Mar 2019 15:26:01 +0000 (17:26 +0200)]
EAP-pwd server: Detect reflection attacks
When processing an EAP-pwd Commit frame, verify that the peer's scalar
and elliptic curve element differ from the one sent by the server. This
prevents reflection attacks where the adversary reflects the scalar and
element sent by the server. (CVE-2019-9497)
The vulnerability allows an adversary to complete the EAP-pwd handshake
as any user. However, the adversary does not learn the negotiated
session key, meaning the subsequent 4-way handshake would fail. As a
result, this cannot be abused to bypass authentication unless EAP-pwd is
used in non-WLAN cases without any following key exchange that would
require the attacker to learn the MSK.
Mathy Vanhoef [Sun, 31 Mar 2019 15:13:06 +0000 (17:13 +0200)]
EAP-pwd server: Verify received scalar and element
When processing an EAP-pwd Commit frame, the peer's scalar and element
(elliptic curve point) were not validated. This allowed an adversary to
bypass authentication, and impersonate any user if the crypto
implementation did not verify the validity of the EC point.
Fix this vulnerability by assuring the received scalar lies within the
valid range, and by checking that the received element is not the point
at infinity and lies on the elliptic curve being used. (CVE-2019-9498)
The vulnerability is only exploitable if OpenSSL version 1.0.2 or lower
is used, or if LibreSSL or wolfssl is used. Newer versions of OpenSSL
(and also BoringSSL) implicitly validate the elliptic curve point in
EC_POINT_set_affine_coordinates_GFp(), preventing the attack.
Jouni Malinen [Tue, 5 Mar 2019 21:43:25 +0000 (23:43 +0200)]
SAE: Fix confirm message validation in error cases
Explicitly verify that own and peer commit scalar/element are available
when trying to check SAE confirm message. It could have been possible to
hit a NULL pointer dereference if the peer element could not have been
parsed. (CVE-2019-9496)
Jouni Malinen [Sat, 2 Mar 2019 14:05:56 +0000 (16:05 +0200)]
SAE: Use constant time operations in sae_test_pwd_seed_ffc()
Try to avoid showing externally visible timing or memory access
differences regardless of whether the derived pwd-value is smaller than
the group prime.
Jouni Malinen [Sat, 2 Mar 2019 10:45:33 +0000 (12:45 +0200)]
SAE: Use const_time selection for PWE in FFC
This is an initial step towards making the FFC case use strictly
constant time operations similarly to the ECC case.
sae_test_pwd_seed_ffc() does not yet have constant time behavior,
though.
Jouni Malinen [Sat, 2 Mar 2019 10:24:09 +0000 (12:24 +0200)]
SAE: Mask timing of MODP groups 22, 23, 24
These groups have significant probability of coming up with pwd-value
that is equal or greater than the prime and as such, need for going
through the PWE derivation loop multiple times. This can result in
sufficient timing different to allow an external observer to determine
how many rounds are needed and that can leak information about the used
password.
Force at least 40 loop rounds for these MODP groups similarly to the ECC
group design to mask timing. This behavior is not described in IEEE Std
802.11-2016 for SAE, but it does not result in different values (i.e.,
only different timing), so such implementation specific countermeasures
can be done without breaking interoperability with other implementation.
Note: These MODP groups 22, 23, and 24 are not considered sufficiently
strong to be used with SAE (or more or less anything else). As such,
they should never be enabled in runtime configuration for any production
use cases. These changes to introduce additional protection to mask
timing is only for completeness of implementation and not an indication
that these groups should be used.
Jouni Malinen [Tue, 26 Feb 2019 17:34:38 +0000 (19:34 +0200)]
SAE: Avoid branches in is_quadratic_residue_blind()
Make the non-failure path in the function proceed without branches based
on r_odd and in constant time to minimize risk of observable differences
in timing or cache use. (CVE-2019-9494)
Jouni Malinen [Tue, 26 Feb 2019 11:05:09 +0000 (13:05 +0200)]
SAE: Minimize timing differences in PWE derivation
The QR test result can provide information about the password to an
attacker, so try to minimize differences in how the
sae_test_pwd_seed_ecc() result is used. (CVE-2019-9494)
Use heap memory for the dummy password to allow the same password length
to be used even with long passwords.
Use constant time selection functions to track the real vs. dummy
variables so that the exact same operations can be performed for both QR
test results.
Jouni Malinen [Tue, 26 Feb 2019 09:59:45 +0000 (11:59 +0200)]
EAP-pwd: Use constant time and memory access for finding the PWE
This algorithm could leak information to external observers in form of
timing differences or memory access patterns (cache use). While the
previous implementation had protection against the most visible timing
differences (looping 40 rounds and masking the legendre operation), it
did not protect against memory access patterns between the two possible
code paths in the masking operations. That might be sufficient to allow
an unprivileged process running on the same device to be able to
determine which path is being executed through a cache attack and based
on that, determine information about the used password.
Convert the PWE finding loop to use constant time functions and
identical memory access path without different branches for the QR/QNR
cases to minimize possible side-channel information similarly to the
changes done for SAE authentication. (CVE-2019-9495)
Jouni Malinen [Thu, 7 Mar 2019 22:24:12 +0000 (00:24 +0200)]
OpenSSL: Use constant time selection for crypto_bignum_legendre()
Get rid of the branches that depend on the result of the Legendre
operation. This is needed to avoid leaking information about different
temporary results in blinding mechanisms.
This is related to CVE-2019-9494 and CVE-2019-9495.
Jouni Malinen [Wed, 27 Feb 2019 16:38:30 +0000 (18:38 +0200)]
Add helper functions for constant time operations
These functions can be used to help implement constant time operations
for various cryptographic operations that must minimize externally
observable differences in processing (both in timing and also in
internal cache use, etc.).
This is related to CVE-2019-9494 and CVE-2019-9495.
Jouni Malinen [Tue, 26 Feb 2019 09:43:03 +0000 (11:43 +0200)]
OpenSSL: Use constant time operations for private bignums
This helps in reducing measurable timing differences in operations
involving private information. BoringSSL has removed BN_FLG_CONSTTIME
and expects specific constant time functions to be called instead, so a
bit different approach is needed depending on which library is used.
The main operation that needs protection against side channel attacks is
BN_mod_exp() that depends on private keys (the public key validation
step in crypto_dh_derive_secret() is an exception that can use the
faster version since it does not depend on private keys).
crypto_bignum_div() is currently used only in SAE FFC case with not
safe-prime groups and only with values that do not depend on private
keys, so it is not critical to protect it.
crypto_bignum_inverse() is currently used only in SAE FFC PWE
derivation. The additional protection here is targeting only OpenSSL.
BoringSSL may need conversion to using BN_mod_inverse_blinded().
This is related to CVE-2019-9494 and CVE-2019-9495.
This makes it easier to test various X.509 certificate validation steps
with the server certificate being generated and signed using pyOpenSSL
dynamically.
Extend domain_match and domain_suffix_match to allow list of values
These wpa_supplicant network profile parameters could be used to specify
a single match string that would be used against the dNSName items in
subjectAltName or CN. There may be use cases where more than one
alternative match string would be useful, so extend these to allow a
semicolon delimited list of values to be used (e.g.,
"example.org;example.com"). If any of the specified values matches any
of the dNSName/CN values in the server certificate, consider the
certificate as meeting this requirement.
wolfSSL: Fix dNSName matching with domain_match and domain_suffix_match
Incorrect gen->type value was used to check whether subjectAltName
contained dNSName entries. This resulted in all domain_match and
domain_suffix_match entries failing to find a match and rejecting the
server certificate. Fix this by checking against the correct type
definition for dNSName.
RADIUS server: Accept ERP keyName-NAI as user identity
Previously the EAP user database had to include a wildcard entry for ERP
to work since the keyName-NAI as User-Name in Access-Request would not
be recognized without such wildcard entry (that could point to any EAP
method). This is not ideal, so add a separate check to allow any stored
ERP keyName-NAI to be used for ERP without any requirement for the EAP
user database to contain a matching entry.
tests: scan_multi_bssid_check_ie to allow for Multi BSSID Index IE
cfg80211 was modified to allow the Multiple BSSID Index element to be
included in the IEs for a nontransmitted BSS. Update the validation step
in this test case to allow that different with the IEs in the Beacon
frame (transmitted BSS).
DPP: Fix a regression in non-DPP, non-OpenSSL builds
Inclusion of common/dpp.h into hostapd/main.c brought in an undesired
unconditional dependency on OpenSSL header files even for builds where
DPP is not enabled. Fix this by making the dpp.h contents, and in
particular the inclusion of openssl/x509.h, conditional on CONFIG_DPP.
Fixes: 87d8435cf9fd ("DPP: Common configurator/bootstrapping data management") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
SAE: Reject unsuitable groups based on REVmd changes
The rules defining which DH groups are suitable for SAE use were
accepted into IEEE 802.11 REVmd based on this document:
https://mentor.ieee.org/802.11/dcn/19/11-19-0387-02-000m-addressing-some-sae-comments.docx
Enforce those rules in production builds of wpa_supplicant and hostapd.
CONFIG_TESTING_OPTIONS=y builds can still be used to select any o the
implemented groups to maintain testing coverage.
tests: Change most SAE test cases to use suitable groups
Reduce testing dependency on the unsuitable groups so that a test case
against a production build would not fail the test case unnecessarily.
This is in preparation of making production builds
(CONFIG_TESTING_OPTIONS not defined) of wpa_supplicant hostapd disable
all DH groups that have been indicated as being unsuitable.
identity_buf may be NULL here. Handle this case explicitly by printing
"N/A" instead relying on snprintf converting this to "(null)" or some
other value based on unexpected NULL pointer.
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Ilan Peer [Wed, 3 Apr 2019 15:17:13 +0000 (18:17 +0300)]
scan: Use normal scans after connection failure
In case of connection attempt failure, set 'normal_scans'
to zero, as otherwise it is possible that scheduled scan
would be used and not normal scan, which might delay the
next connection attempt.
hostapd: Reduce minimum beacon interval from 15 to 10 TUs
Very short beacon intervals can be useful for certain scenarios such
as minimising association time on PBSSs. Linux supports a minimum of
10[1] so let's reduce the minimum to match that.
Ilan Peer [Wed, 3 Apr 2019 12:16:14 +0000 (15:16 +0300)]
crypto: Fix unreachable code in tls_prf_sha1_md5()
While commit 1c156e783d35 ("Fixed tls_prf() to handle keys with
odd length") added support for keys with odd length, the function
never reached this code as the function would return earlier in
case the key length was odd. Fix this by removing the first check
for the key length.
Sunil Dutt [Fri, 29 Mar 2019 12:47:53 +0000 (18:17 +0530)]
Add a QCA vendor attribute to carry the reason for roaming
This commit introduces an attribute
QCA_WLAN_VENDOR_ATTR_ROAM_AUTH_REASON to carry the roam reason code
through QCA_NL80211_VENDOR_SUBCMD_KEY_MGMT_ROAM_AUTH event.