When wpa_supplicant is running on a Linux interface that is configured in
promiscuous mode, and it is not a member of a bridge, incoming EAPOL
packets are processed regardless of the Destination Address in the frame.
As a consequence, there are situations where wpa_supplicant replies to
EAPOL packets that are not destined for it.
This behavior seems undesired (see IEEE Std 802.1X-2010, 11.4.a), and can
be avoided by attaching a BPF filter that lets the kernel discard packets
having pkt_type equal to PACKET_OTHERHOST.
Signed-off-by: Davide Caratti <davide.caratti@gmail.com>
Previously we set this flag to one in wpa_supplicant_init_iface() if
Wi-Fi controller does not have a dedicated P2P-interface.
This setting had effect only in scope of wpa_supplicant_init_iface() and
it contradicts with comment to struct wpa_interface::p2p_mgmt field.
This comment says that this flag is used only if Wi-Fi controller has
dedicated P2P-device interface.
Also it contradicts with usage of similiar p2p_mgmt field in struct
wpa_supplicant. Again struct wpa_supplicant::p2p_mgmt is set only for
dedicated P2P-device interface.
After this change wpa_interface become input argument to
wpa_supplicant_init_iface() that we are not modifying.
Signed-off-by: Vasyl Vavrychuk <vvavrychuk@gmail.com>
This fixes sending of FindStopped, GroupFormationFailure, and
InvitationReceived signals in the case of separate P2P-Device interface.
This extends the coverage of the earlier commit
745d62322b ("dbus: Redirect P2P request to
the managment device if present") to these three functions that were
missing the redirection.
Some wireless controllers might have separate P2P-Device interface, see
as example result of 'iw dev':
phy#0
Unnamed/non-netdev interface
...
type P2P-device
...
Interface wlp2s0
type managed
...
In this case there is separate 'struct wpa_supplicant' created for this
p2p-dev-* device as result of 'wpa_supplicant_add_iface >
wpas_p2p_add_p2pdev_interface > wpa_supplicant_add_iface'.
This instance of wpa_supplicant is not registered in D-Bus
(wpas_dbus_register_*) since for corresponding P2P device interface flag
'struct wpa_interface > p2p_mgmt' is set.
But this instance is saved in p2p_init_wpa_s and is used for handling
P2P related D-Bus commands. Therefore we should look for D-Bus path in
the parent of p2p_init_wpa_s instance.
Without this change test dbus_p2p_discovery starts failing if we set
support_p2p_device in vm-run.sh.
Signed-off-by: Vasyl Vavrychuk <vvavrychuk@gmail.com>
If any of the interfaces supports FILS (and similarly for FILS-SK-PFS),
include the "fils" (and "fils_sk_pfs") capability in D-Bus information.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
Add examples of relevant top level CONFIG clauses for wpa_supplicant
MACsec support to defconfig.
Extend the example of MACsec related network configuration. Also bring
them in line with the format of the other example network configurations.
Signed-off-by: Jaap Keuter <jaap.keuter@xs4all.nl>
In the current implementation, upon an EAP method failure, followed by
an EAP failure, the EAP Status is propagated up in wpa_supplicant with a
general failure parameter string "failure". This parameter is used for a
notification on the dbus.
This commit reports the EAP method failure error code in a separate
callback.
The solution in this commit is generic to all EAP methods, and can be
used by any method that need to pass its error code. However, this
commit only implements the reporting for EAP-SIM and EAP-AKA methods
where the Notification Code (in AT_NOTIFICATION) is used as the method
specific error code value.
Signed-off-by: Ahmed ElArabawy <arabawy@google.com>
When connecting to a WPA-EAP network and the MAC address is changed
just before the association (for example by NetworkManager, which sets
a random MAC during scans), the authentication sometimes fails in the
following way ('####' logs added by me):
wpa_supplicant logs:
wlan0: WPA: RX message 1 of 4-Way Handshake from 02:00:00:00:01:00 (ver=1)
RSN: msg 1/4 key data - hexdump(len=22): dd 14 00 0f ac 04 d8 21 9d a5 73 98 88 26 ef 03 d2 ce f7 04 7d 23
WPA: PMKID in EAPOL-Key - hexdump(len=22): dd 14 00 0f ac 04 d8 21 9d a5 73 98 88 26 ef 03 d2 ce f7 04 7d 23
RSN: PMKID from Authenticator - hexdump(len=16): d8 21 9d a5 73 98 88 26 ef 03 d2 ce f7 04 7d 23
wlan0: RSN: no matching PMKID found
EAPOL: Successfully fetched key (len=32)
WPA: PMK from EAPOL state machines - hexdump(len=32): [REMOVED]
#### WPA: rsn_pmkid():
#### WPA: aa - hexdump(len=6): 02 00 00 00 01 00
#### WPA: spa - hexdump(len=6): 66 20 cf ab 8c dc
#### WPA: PMK - hexdump(len=32): b5 24 76 4f 6f 50 8c f6 a1 2e 24 b8 07 4e 9a 13 1b 94 c4 a8 1f 7e 22 d6 ed fc 7d 43 c7 77 b6 f7
#### WPA: computed PMKID - hexdump(len=16): ea 73 67 b1 8e 5f 18 43 58 24 e8 1c 47 23 87 71
RSN: Replace PMKSA entry for the current AP and any PMKSA cache entry that was based on the old PMK
nl80211: Delete PMKID for 02:00:00:00:01:00
wlan0: RSN: PMKSA cache entry free_cb: 02:00:00:00:01:00 reason=1
RSN: Added PMKSA cache entry for 02:00:00:00:01:00 network_ctx=0x5630bf85a270
nl80211: Add PMKID for 02:00:00:00:01:00
wlan0: RSN: PMKID mismatch - authentication server may have derived different MSK?!
hostapd logs:
WPA: PMK from EAPOL state machine (MSK len=64 PMK len=32)
WPA: 02:00:00:00:00:00 WPA_PTK entering state PTKSTART
wlan1: STA 02:00:00:00:00:00 WPA: sending 1/4 msg of 4-Way Handshake
#### WPA: rsn_pmkid():
#### WPA: aa - hexdump(len=6): 02 00 00 00 01 00
#### WPA: spa - hexdump(len=6): 02 00 00 00 00 00
#### WPA: PMK - hexdump(len=32): b5 24 76 4f 6f 50 8c f6 a1 2e 24 b8 07 4e 9a 13 1b 94 c4 a8 1f 7e 22 d6 ed fc 7d 43 c7 77 b6 f7
#### WPA: computed PMKID - hexdump(len=16): d8 21 9d a5 73 98 88 26 ef 03 d2 ce f7 04 7d 23
WPA: Send EAPOL(version=1 secure=0 mic=0 ack=1 install=0 pairwise=1 kde_len=22 keyidx=0 encr=0)
That's because wpa_supplicant computed the PMKID using the wrong (old)
MAC address used during the scan. wpa_supplicant updates own_addr when
the interface goes up, as the MAC can only change while the interface
is down. However, drivers don't report all interface state changes:
for example the nl80211 driver may ignore a down-up cycle if the down
message is processed later, when the interface is already up. In such
cases, wpa_supplicant (and in particular, the EAP state machine) would
continue to use the old MAC.
Add a new driver event that notifies of MAC address changes while the
interface is active.
Signed-off-by: Beniamino Galvani <bgalvani@redhat.com>
Add an event callback for EVENT_STATION_OPMODE_CHANGED to allow
user/application to get the notification whenever there is a change in a
station's HT/VHT op mode.
The new events:
STA-OPMODE-MAX-BW-CHANGED <addr> <20(no-HT)|20|40|80|80+80|160>
STA-OPMODE-SMPS-MODE-CHANGED <addr> <automatic|off|dynamic|static>
STA-OPMODE-N_SS-CHANGED <addr> <N_SS>
Signed-off-by: Tamizh chelvam <tamizhr@codeaurora.org>
To retain configurator information across hostapd/wpa_supplicant
restart, private key need to be maintained to generate a valid pair of
authentication keys (connector, netaccess_key, csign) for new enrollees
in the network.
Add a DPP_CONFIGURATOR_GET_KEY control interface API through which the
private key of an existing configurator can be fetched.
Command format:
DPP_CONFIGURATOR_GET_KEY <configurator_id>
The output from this command can then be used with
"DPP_CONFIGURATOR_ADD key=<hexdump>" to create the same key again.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
This extends dpp_test functionality to allow DPP exchanges to be stopped
after authentication is completed on the Initiator, i.e., after sending
out the Authentication Confirm message. Previously, dpp_test=89 was used
only on the Responder side to stop after receiving the Authentication
Confirm message. The main use case for this extended functionality is to
be able to stop the protocol exchange on a device that acts as
authentication Initiator and Enrollee.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
In case MKA is initialized successfully, local copies of CAK and CKN
were allocated, but never freed. Ensure that such memory is released
also when ieee802_1x_kay_create_mka() returns a valid pointer.
Fixes: ad51731abf ("wpa_supplicant: Allow pre-shared (CAK,CKN) pair for MKA")
Signed-off-by: Davide Caratti <davide.caratti@gmail.com>
When the driver indicates that the connection is authorized (i.e., the
4-way handshake was completed by the driver), cancel the EAP
authentication timeout and set the EAP state machine to success state.
Signed-off-by: Avraham Stern <avraham.stern@intel.com>
wpa_passphrase requires libcrypto from OpenSSL (or another selected
library). User can set an alternative path to OpenSSL libraries by
defining LIBS at the top of .config but if $(LIBS) is not actually used
wrong libcrypto is used or compilation fails if there is no libcrypto in
the default locations cc is looking for it. It's especially bad for
cross-compilers that fail with 'cannot find -lcrypto' message.
Signed-off-by: Arkadiusz Drabczyk <arkadiusz@drabczyk.org>
This header file pulls in an OpenSSL header file and as such, should not
be included without CONFIG_DPP=y to avoid bringing in an unnecessary
build dependency on OpenSSL header files.
Signed-off-by: Jouni Malinen <j@w1.fi>
Force use of the wildcard BSSID address in GAS query frames with DPP
regardless of how the gas_address3 configuration parameter is set. DPP
specification mandates this and the use of GAS here is really outside
the context of a BSS, so using the wildcard BSSID makes sense even for
the corner case of Configurator running on a known AP (where IEEE 802.11
standard would allow the BSSID of the AP to be used).
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
If a DPP_STOP_LISTEN call happens to be received when there is a pending
gas-query radio work that has not yet been started, it was possible for
gas_query_stop() to go through gas_query_done() processing with
gas->work == NULL and that ended up with the pending GAS query getting
freed without removing the pending radio work that hold a reference to
the now freed memory. Fix this by removing the pending non-started radio
work for the GAS query in this specific corner case.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Add support for wpa_supplicant to try to initiate PKEX on 5 GHz and 60
GHz bands in addition to the previously available 2.4 GHz case. If no
response from a peer device is seen on the 2.4 GHz band (channel 6) for
the five attempts, try the other PKEX channels (5 GHz channels 44 and
149; and 60 GHz channel 2) if they are supported and allowed for
initiating radiation.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Previously, wpa_supplicant set PMF as optional for the DPP AKM since
there was no clear statement about this requirement in the tech spec.
Now that this requirement has been added, update the implementation to
match. In addition, set ssid->ieee80211w using the actual enum
mfp_options values instead of magic constants to make this a bit more
readable.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
If the -I<config> argument is used and the referenced configuration file
cannot be parsed, wpa_config_read() ended up freeing the main
configuration data structure and that resulted in use of freed memory in
such an error case. Fix this by not freeing the main config data and
handling the error case in the caller.
Signed-off-by: Jouni Malinen <j@w1.fi>
If option -I:config points to a non-existing file, the the previously
allocated config must not be freed. Avoid use of freed memory in such an
error case by skipping the incorrect freeing operation.
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>
The special case of returning from wpa_supplicant_set_suites() when OWE
transition mode profile is used for an open association did not clear
the wpa_ie buffer length properly. This resulted in trying to use
corrupted IEs in the association request and failed association
(cfg80211 rejects the request or if the request were to go out, the AP
would likely reject it).
Signed-off-by: Jouni Malinen <j@w1.fi>
This functionality can be used regardless of P2P and should not be under
the ifdef of CONFIG_P2P.
Signed-off-by: Simon Dinkin <simon.dinkin@tandemg.com>
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
This was supposed to be an array of const-pointers to const-char; not
something duplicating const for char and resulting in compiler warnings
with more recent gcc versions.
Signed-off-by: Jouni Malinen <j@w1.fi>
Extend the SME functionality to support the external authentication.
External authentication may be used by the drivers that do not define
separate commands for authentication and association
(~WPA_DRIVER_FLAGS_SME) but rely on wpa_supplicant's SME for the
authentication.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
This allows a separate configuration parameter (imsi_identity) to be
used in EAP-SIM/AKA/AKA' profiles to override the identity used in MK
derivation for the case where the identity is expected to be from the
last AT_IDENTITY attribute (or EAP-Response/Identity if AT_IDENTITY was
not used). This may be needed to avoid sending out an unprotected
permanent identity information over-the-air and if the EAP-SIM/AKA
server ends up using a value based on the real IMSI during the internal
key derivation operation (that does not expose the data to others).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
If the OWE network profile matches an open network which does not
advertise OWE BSS, allow open connection. The new owe_only=1 network
profile parameter can be used to disable this transition mode and
enforce connection only with OWE networks.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
GnuTLS-based builds can now be done using either libnettle or libgcrypt
for crypto functionality:
CONFIG_TLS=gnutls
CONFIG_CRYPTO=nettle
CONFIG_TLS=gnutls
CONFIG_CRYPTO=gnutls
Signed-off-by: Jouni Malinen <j@w1.fi>
Replace the internal HMAC MD5, SHA-1, and SHA256 implementations with
the ones from libgcrypt and also add the SHA384 and SHA512 versions.
Signed-off-by: Jouni Malinen <j@w1.fi>
Replace the internal SHA256 implementation with the one from libgcrypt
and also add the SHA384 and SHA512 versions.
Signed-off-by: Jouni Malinen <j@w1.fi>
If a specific DH group for OWE is not set with the owe_group parameter,
try all supported DH groups (currently 19, 20, 21) one by one if the AP
keeps rejecting groups with the status code 77.
Signed-off-by: Jouni Malinen <j@w1.fi>
Instead of requiring OpenSSL headers to be available just for the
SSL3_RANDOM_SIZE definition, replace that macro with a fixed length (32)
to simplify dependencies.
Signed-off-by: Jouni Malinen <j@w1.fi>
The new disable_fils parameter can be used to disable FILS functionality
in the driver. This is currently removing the FILS Capability bit in
Extended Capabilities and providing a callback to the driver wrappers.
driver_nl80211.c implements this using a QCA vendor specific command for
now.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Commit 265bda3444 ('OWE: Allow DH
Parameters element to be overridden for testing purposes') provided
means for using "VENDOR_ELEM_ADD 13 <IE>" in OWE protocol testing, but
that commit covered only the sme.c case (i.e., drivers that use
wpa_supplicant SME). Extend this to cover drivers that use internal SME
(e.g., use the nl80211 Connect command).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Wpa_supplicant's random pool is not necessary on Android. Randomness
is already provided by the entropymixer service which ensures
sufficient entropy is maintained across reboots. Commit b410eb1913
'Initialize /dev/urandom earlier in boot' seeds /dev/urandom with
that entropy before either wpa_supplicant or hostapd are run.
Signed-off-by: Jeff Vander Stoep <jeffv@google.com>
If a credential with encp type AES is received, add GCMP-256 and
CCMP-256 cipher options on station Enrollee based on local capabilities.
This is needed to allow connection with an AP using either of these
newer ciphers.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The credential MAC address is not necessarily that of the AP, i.e., it
is more likely to be that of the Enrollee. Check the scan results
against the current BSSID as well if match is not found otherwise when
going through the mixed mode workaround.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The previously used 10 second timer did not really make much sense since
the Initiator is not going to be waiting for the response that long.
Change this to 1 second based on the DPP tech spec change.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
