Add timeout to close incomplete DPP relay connections. This is needed to
avoid getting stuck with old entries that prevent new connections from
getting started.
Signed-off-by: Disha Das <dishad@codeaurora.org>
When CONFIG_ECC is not defined, openssl/ec.h is not included and EC_KEY
not known. Fix be not defining EVP_PKEY_get0_EC_KEY() when CONFIG_ECC is
not defined.
Signed-off-by: Wolfgang Steinwender <wsteinwender@pcs.com>
Simplify the implementation by using shared functions for parsing the
capabilities instead of using various similar but not exactly identical
checks throughout the implementation.
Signed-off-by: Jouni Malinen <j@w1.fi>
IEEE Std 802.11-2020 mandates H2E to be used whenever an SAE password
identifier is used. While this was already covered in the
implementation, the sae_prepare_commit() function still included an
argument for specifying the password identifier since that was used in
an old test vector. Now that that test vector has been updated, there is
no more need for this argument anymore. Simplify the older non-H2E case
to not pass through a pointer to the (not really used) password
identifier.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Do so for both wpa_supplicant and hostapd. While this was not explicitly
required in IEEE P802.11az/D3.0, likely direction for the draft is to
start requiring use of H2E for all cases where SAE is used with PASN.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
When a PTK derivation is done as part of PASN authentication flow, a KDK
derivation should be done if and only if the higher layer protocol is
supported by both parties.
Fix the code accordingly, so KDK would be derived if and only if both
sides support Secure LTF.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Define the new TWT attributes for configuring the broadcast TWT
parameters in enum qca_wlan_vendor_attr_twt_setup.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
The firmware rejects the TWT setup request when roaming and channel
switch is in progress. Extend enum qca_wlan_vendor_twt_status to
represent new reason codes for these cases.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Commit 3ab35a6603 ("Extend EAPOL frames processing workaround for
roaming cases") added a work around to address the issue of EAPOL frame
reception after reassociation replied to with an incorrect destination
address (the BSSID of the old AP). This is due to association events and
EAPOL RX events being reordered for the roaming cases with drivers that
perform BSS selection internally.
This mechanism relies on the fact that the driver always forwards the
EAPOL handshake to wpa_supplicant after the roaming (sets
last_eapol_matches_bssid during the EAPOL processing and resets on the
assoc/reassoc indication).
The above approach does not address the case where the driver does the
EAPOL handshake on the roam, indicating the authorized status to
wpa_supplicant but also forwards the EAPOL handshake to wpa_supplicant
for few other roam attempts. This is because the flag
last_eapol_matches_bssid is not set with the roam+authorized event from
the driver. Thus, the next reorder of roam and EAPOL RX events would
miss this workaround.
Address this by setting last_eapol_matches_bssid=1 on a roam+authorized
event from the driver.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
This changes the Send-Confirm value for the first SAE Confirm message to
be 1 instead of 0 for all cases to match the design shown in IEEE Std
802.11-2020, Figure 12-4 (SAE finite state machine).
Sc is defined to be "the number of SAE Confirm messages that have been
sent" which is a bit vague on whether the current frame is included in
the count or not. However, the state machine is showing inc(Sc)
operation in all cases before the "2" event to build the Confirm.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
If the maximum size of MAC ACL entries is large enough, the
configuration message may exceed the default buffer size of a netlink
message which is allocated with nlmsg_alloc(), and result in a failure
when putting the attributes into the message.
To fix this, calculate the required buffer size of the netlink message
according to MAC ACL size and allocate a sufficiently large buffer with
nlmsg_alloc_size().
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
NL80211_ATTR_MAC_ACL_MAX is a u32 attribute to advertise the maximum
number of MAC addresses that a device can support for MAC ACL. This was
incorrectly used as a u8 attribute which would not work with any values
larger than 255 or on big endian CPUs. Fix this by moving from
nla_get_u8() to nla_get_u32().
Fixes: 3c4ca36330 ("hostapd: Support MAC address based access control list")
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
If WMM is not set explicitly in the configuration, it can be set based
on HT/HE config. As HE can be used without HT/VHT (which was introduced
as a special behavior for the 6 GHz band), add a similar automatic
enabling of WMM for HE without HT.
Signed-off-by: Lavanya Suresh <lavaks@codeaurora.org>
wpa_supplicant_ctrl_iface_deinit() was executed only if the
per-interface control interface initialization had been completed. This
is not the case if driver initialization fails and that could result in
leaving behind references to the freed wpa_s instance in a corner case
where control interface messages ended up getting queued.
Fix this by calling wpa_supplicant_ctrl_iface_deinit() in all cases to
cancel the potential eloop timeout for wpas_ctrl_msg_queue_timeout with
the reference to the wpa_s pointer. In addition, flush any pending
message from the global queue for this interface since such a message
cannot be of use after this and there is no need to leave them in the
queue until the global control interface gets deinitialized.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Previous debug print used IGTK instead of BIGTK, so fix that to use the
correct key. Actual generation of the BIGTK subelement itself was using
the correct key, though, so this is only needed to fix the debug print.
Signed-off-by: Jouni Malinen <j@w1.fi>
Move from RS to PA country code to allow these test cases to work since
regdb was updated to require DFS for these operating classes in RS.
Signed-off-by: Jouni Malinen <j@w1.fi>
This is a 2 octet field, so need to use WPA_GET_LE16() here instead of
using only the first octet of the value.
Fixes: bbd3178af4 ("MSCS: Add support to process MSCS Response frames")
Signed-off-by: Jouni Malinen <j@w1.fi>
The new hostapd configuration parameters ext_capa_mask and ext_capa can
now be used to mask out or add extended capability bits. While this is
not without CONFIG_TESTING_OPTIONS, the main use case for this is for
testing purposes.
Signed-off-by: Jouni Malinen <j@w1.fi>
The conditional compilation block with only the opening brace included
in two variants was messing up auto indentation in emacs. Work around
this by defining the maximum value conditionally while leave the if
block outside any conditional building rules.
Signed-off-by: Jouni Malinen <j@w1.fi>
There is no need to determine the exact length of the element before
filling in the octets since this function is already capable of
truncated the fields based on what the actual values are.
Signed-off-by: Jouni Malinen <j@w1.fi>
Process the received comeback cookie and retry automatically if the AP
allows this. Otherwise, provide the cookie to upper layers to allow a
later attempt with the cookie.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Reuse the SAE anti-clogging token implementation to support similar
design with the PASN comeback cookie.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
