When using MAC randomization wpa_supplicant can change the local MAC
address during roaming scenario:
1. We attach to AP1 (with MAC1/SSID1).
2. Roaming to AP2 (with MAC2/SSID2) is started:
a) we send DEAUTH(for AP1, with MAC1)
b) we change MAC to MAC2 due to randomization
c) we start authentication for AP2
d) we get notification about DEAUTH for AP1 (which we ignore)
e) we complete association with AP2
In point 2d we completely ignore the notification which later causes
problems. This happens if the deauthentication event is generated by the
local driver (e.g., due to beacon loss) instead of AP2 sending an
explicit Deauthentication frame.
The intended behavior is as follows: during roaming we generate DEAUTH
(2a) and signal this event right away. To protect from handling of our
own DEAUTH for the 2nd time supplicant marks 'ignore_next_local_deauth'
variable. In point 2d we should receive this notification and clear the
flag but this does not happen because MAC1 in the notification is not
the current MAC address (it has been changed in 2b) so this notification
is ignored as a one with a "foreign" address.
So we end up successfully at AP2 but with 'ignore_next_local_deauth'
still set which causes problems. For example if AP2 shuts down it has
been observed on some drivers that the DEAUTH notification is generated
as a local one and since we have flag to ignore it nothing is reported
over D-Bus.
To address the problem let's store the previously used MAC address and
use it for checking for foreign address (in combination with the current
one).
Signed-off-by: Andrzej Ostruszka <amo@semihalf.com>
SSID may include unprintable characters. This change converts
unprintable characters into printable form before printing SSID in the
function wpas_send_ctrl_req(). The conversion is based on the function
wpa_ssid_txt().
Signed-off-by: Kaidong Wang <kaidong@chromium.org>
In case of drivers supporting 4-way handshake offload, mark port
authorized and state completion only if the driver advertizes authorized
state in the connect event. Otherwise there are fair chances of the
driver port authorization API getting called while 4-way handshake is in
progress at the lower layer.
In order to avoid this possible race condition always update port
authorization and supplicant state WPA_COMPLETED setting from
EVENT_PORT_AUTHORIZED context when the driver is done with the 4-way
handshake.
Signed-off-by: Vinayak Yadawad <vinayak.yadawad@broadcom.com>
Add the support of multiple domains for interworking credentials in
D-Bus API AddCred() using an array of strings.
Signed-off-by: Damien Dejean <damiendejean@chromium.org>
Fix compilation issue when using the following build parameter:
CONFIG_CTRL_IFACE=udp-remote
Fixes: 0aae045af0 ("ctrl: Print the source address of the received commands")
Signed-off-by: Janusz Dziedzic <janusz.dziedzic@gmail.com>
This is a step in separating RSN and WPA(v1) processing of EAPOL-Key
frames into separate functions. This allows the implementation to be
simplified and potentially allows the validation rules to be made
stricter more easily. This is also a step towards allowing WPA(v1)
functionality to be removed from the build in the future.
Signed-off-by: Jouni Malinen <j@w1.fi>
Move testing to use the new Home OI parameters while maintaining a
couple of tests for the deprecated parameters.
Signed-off-by: Damien Dejean <damiendejean@chromium.org>
Move from the now deprecated roaming_consortium and
required_roaming_consortium credential parameters to home_ois and
required_home_ois.
Signed-off-by: Damien Dejean <damiendejean@chromium.org>
Until now Hotspot 2.0 credentials were only supporting one home OI (with
roaming_consortium option) and one required home OI (with
required_roaming_consortium option). To improve the compliance with
Passpoint specification, add the support for multiple home and required
OIs.
The lists of OIs are provided using two new configuration options
home_ois and required_home_ois that expect a list of OIs formatted as
the roaming_consortiums list. It allows to keep the old options to avoid
breaking currently running configurations and better fits the vocabulary
used in the spec.
The OI match algorithm is updated to implement the behavior described in
Passpoint specification v3.2 section 9.1.2 (Home OIs nodes description
PerProviderSubscription/<X+>/HomeSP/HomeOIList/<X+>).
Signed-off-by: Damien Dejean <damiendejean@chromium.org>
The default provider is being loaded here explicitly only because
OSSL_PROVIDER_load() disables the fallback provider loading (on either
success or failure). If the legacy provider fails to load, which it may
in some configurations, it will never load the default provider.
Just use the formulation which attempts to load without changing the
fallback behavior.
"default" will still be/only be loaded if no other provider (notably
FIPS) is loaded to provide algorithms.
Signed-off-by: Norman Hamer <nhamer@absolute.com>
DES and RC4 are not allowed in such builds, so comment out des_encrypt()
and rc4_skip() from the build to force compile time failures for cases
that cannot be supported instead of failing the operations at runtime.
This makes it easier to detect and fix accidental cases where DES/RC4
could still be used in some older protocols.
Signed-off-by: Norman Hamer <nhamer@absolute.com>
CONFIG_NO_RC4=y could have been used to remove this functionality, but
it might as well be done automatically based on CONFIG_FIPS=y as well.
Signed-off-by: Jouni Malinen <j@w1.fi>
The description of how to use radius_accept_attr did not provide the
details on how the value is set. Extend it to cover the more complete
syntax.
Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
There is no way to create or remove a virtual interface with
wpa_supplicant dbus methods. The platform has to use out-of-band methods
to manage the virtual interfaces.
This change adds virtual interface create/remove logic to the dbus
methods CreateInterface and RemoveInterface to achieve similar
functionalities as wpa_cli commands interface_add and interface_remove.
Signed-off-by: Jintao Lin <jintaolin@chromium.org>
Need to copy own address and peer address locally and pass them to
pasn_deauthenticate(), because this pointer data will be flushed from
the PTKSA cache before sending the Deauthentication frame and these
pointers to then-freed memory would be dereferenced.
Fixes: 24929543 ("PASN: Deauthenticate on PTKSA cache entry expiration")
Signed-off-by: Vinay Gannevaram <quic_vganneva@quicinc.com>
Map to correct control interface commands for PASN start and stop.
Fixes: ad338cfe58 ("ctrl_iface: Add support for PASN authentication")
Signed-off-by: Veerendranath Jakkam <quic_vjakkam@quicinc.com>
PMKSA caching for the FT initial mobility domain association was fully
defined in IEEE Std 802.11-2020. The state before that was unclear and
there has been interoperability issues in this area, so use of PMKSA
caching with FT-EAP has been disabled in wpa_supplicant by default.
The wpa_supplicant and hostapd implementation of PMKSA caching for FT
ended up using an earlier default mechanism (SHA-1) for deriving the
PMKID when using the FT-EAP. This does not match what got defined in
IEEE Std 802.11-2020, 12.11.2.5.2 (SHA256). It is not really desirable
to use SHA-1 for anything with FT since the initial design of FT was
based on SHA256. Furthermore, it is obviously not good to differ in
behavior against the updated standard. As such, there is sufficient
justification to change the implementation to use SHA256 here even
though this ends up breaking backwards compatibility for PMKSA caching
with FT-EAP.
As noted above, this is still disabled in wpa_supplicant by default and
this change results in PMKSA caching not working only in cases where it
has been enabled explicitly with ft_eap_pmksa_caching=1. Those cases
recover by falling back to full EAP authentication.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Add QCA_WLAN_VENDOR_ATTR_CONFIG_WFC_STATE vendor attribute. Userspace
uses this attribute to configure wfc state to the driver/firmware. The
driver/firmware uses this information to optimize power savings, rate
adaption, roaming, etc.
Signed-off-by: Mukul Sharma <quic_mukul@quicinc.com>
wpa_supplicant used the WPS IE from a Probe Response frame, if one was
received, even if there might have been a more recent Beacon frame with
an updated WPS IE. This could result in using stale information about
active WPS registrar, e.g., when operating on the 6 GHz band.
Prefer WPS IE from a Beacon frame over the default selection of Probe
Response frame (if one has been received) in cases where the Beacon
frame is received more recently than the Probe Response frame and active
WPS Registrar information is being checked. Skip this for the case where
UUID-E is needed since that is not available in the Beacon frame.
Signed-off-by: Sai Pratyusha Magam <quic_smagam@quicinc.com>
OCV cannot be used without PMF and such a configuration were to be used
with wpa_supplicant, the AP would reject the association. hostapd is
already enabling PMF automatically whenever OCV is being enabled, so do
the same with wpa_supplicant.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
IEEE Std 802.11-2020, 12.4.7.6 says that status code CHALLENGE_FAILURE,
needs to be sent in case the verification action fails for SAE Confirm
message frame from a STA: "An SAE Confirm message, with a status code
not equal to SUCCESS, shall indicate that a peer rejects a previously
sent SAE Confirm message. An SAE Confirm message that was not
successfully verified is indicated with a status code of
CHALLENGE_FAILURE."
hostapd, however, did not use this status code for this case. In
ieee802_11.c the function sae_check_confirm() is called and in case of
verification failure (-1 is returned), the response is set to
WLAN_STATUS_UNSPECIFIED_FAILURE (status code = 1). Fix this to use
CHALLENGE_FAILURE.
Signed-off-by: Koen Van Oost <koen.vanoost@airties.com>
Signed-off-by: Mert Ekren <mert.ekren@airties.com>
If a higher priority BSS has invalid security parameters, e.g., an
invalid SAE password, and a lower priority BSS is discovered only after
the local network profile has been temporarily disabled, the BSSID
ignoring mechanism is not sufficient to allow the lower priority BSS to
be tried and all consecutive attempts will continue to use the higher
priority BSS. This might prevent connection in some unexpected cases
with invalid network configuration.
Extend BSSID ignoring mechanism to work in this type of a case by
ignoring the BSSID that resulted in disabling the SSID temporarily
during the first connection attempt after having re-enabled the SSID.
This allows a lower priority BSS, if any is available in scan results,
to be tried next to see if it might have working security parameters.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Explicitly check for invalid cases where the configured channel and
bandwidth might result in the full channel number range going beyond the
list of supported channels to avoid reading beyond the end of the
channel buffer.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The SA/DA checks needs to be reversed for the TX case.
Fixes: 8481c750 ("PASN: Fix Authentication frame checks")
Signed-off-by: Vinay Gannevaram <quic_vganneva@quicinc.com>
If an AP is started on a DFS channel (or any channels within its
bandwidth require DFS) and DFS is offloaded to the driver, hostapd needs
to wait for CAC to complete. But the driver may not do CAC and just
switches to a non-DFS channel instead. This would result in a failure to
start the AP because hostapd fails to receive a CAC complete event and
cannot finish interface setup.
Skip CAC and complete AP setup in the channel switch event handler for
this case.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Add CLOSE_LOG command to stop hostapd logging to file. This can be
followed with RELOG to restart logging to the same file path.
Signed-off-by: Sai Pratyusha Magam <quic_smagam@quicinc.com>
Add a new subcommand QCA_NL80211_VENDOR_SUBCMD_DOZED_AP to configure
doze mode state on an AP interface. This is also used as an event to
indicate the updated configuration. In doze mode, AP transmits
beacons at higher beacon intervals and RX is disabled.
Uses attributes defined in enum qca_wlan_vendor_attr_dozed_ap.
Signed-off-by: Purushottam Kushwaha <quic_pkushwah@quicinc.com>
During the roam scan, if there are no desired APs found in the partial
frequency list, an immediate full scan on all the supported frequencies
is initiated as a fallback. This would include the 6 GHz PSC
frequencies. Define an attribute to allow that behavior to be modified
to include PSCs only if 6 GHz use has been detected.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
When the RADIUS server requests a STA to be deauthenticated imminently
without providing a reason URL, there is no need to allow the STA spend
any additional time associated. Deauthenticate the STA immediately after
it has ACK'ed the WNM-Notification frame indicating imminent
deauthentication or at latest two seconds after having processes the
Access-Accept message.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Currently sta_mlo_info.req_links is not getting cleared before
populating the requested links information for a new connection/roam
event. This is causing wrong requested links bitmap in
sta_mlo_info.req_links if there is a change in requested link IDs
between the previous and the new connection. To avoid such issues fully
clear MLO connection information after disconnection and before
populating MLO connection information during (re)association event.
Fixes: cc2236299f ("nl80211: Get all requested MLO links information from (re)association events")
Signed-off-by: Veerendranath Jakkam <quic_vjakkam@quicinc.com>
IGTK and BIGTK MLO KDEs should be validated only when the AP sends them
in EAPOL-Key msg 3/4. Though IEEE P802.11be/D2.2 mandates MLO AP to
enable PMF and Beacon Protection features there is no text to mandate a
STA to discard connection when the MLO AP doesn't send IGTK and BIGTK
MLO KDEs in EAPOL-Key msg 3/4 for a link. Also, fix
wpa_sm->mgmt_group_cipher checks before processing MLO IGTK and BIGTK
MLO KDEs.
Fixes: f15cc834cb ("MLD STA: Processing of EAPOL-Key msg 3/4 frame when using MLO")
Fixes: 8f2e493bec ("MLD STA: Validation of MLO KDEs for 4-way handshake EAPOL-Key frames")
Signed-off-by: Veerendranath Jakkam <quic_vjakkam@quicinc.com>
During the roam scan, if there are no desired APs found in the partial
frequency list, an immediate full scan on all the supported frequencies
is initiated as a fallback. This flag controls the frequency list
creation for full scan on the following lines.
1 - Full scan to exclude the frequencies that were already scanned by
the previous partial scan.
0 - Full scan to include all the supported frequencies irrespective of
the ones already scanned by partial scan.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Taking sizeof(ptr) is incorrect to determine size of passed in hash and
results in hlen getting set to a very large value since MD5_MAC_LEN >
sizeof(ptr). Provide the actual size of the hash buffer from the caller
to fix this.
tls_key_x_server_params_hash() callers src/tls/tlsv1_client_read.c and
src/tls/tlsv1_server_write.c both pass in a large enough hash (hash[64]
or hash[100]) that this does not appear to have an impact, though it is
still wrong.
Signed-off-by: Glenn Strauss <gstrauss@gluelogic.com>
This new value was added to verify peer certificate if it is provided,
but not reject the TLS handshake if no peer certificate is provided.
Signed-off-by: Glenn Strauss <gstrauss@gluelogic.com>
Fix memory leak of intro.peer_key in wpas_dpp_rx_peer_disc_resp(). The
other code paths were updated to use dpp_peer_intro_deinit(), but this
non-privacy-protection version of the station implementation was missed.
Fixes: 148de3e0dc ("DPP3: Private Peer Introduction protocol")
Signed-off-by: Glenn Strauss <gstrauss@gluelogic.com>
The dev pointer could potentially be NULL here in some P2PS cases, so
check it explicitly before dereferencing it when checking for 6 GHz
capability.
Fixes: b9e2826b9d ("P2P: Filter 6 GHz channels if peer doesn't support them")
Signed-off-by: Jouni Malinen <j@w1.fi>
Try to make the bounds checking easier for static analyzers by checking
each length field separately in addition to checking them all in the end
against the total buffer length.
Signed-off-by: Jouni Malinen <j@w1.fi>
