wpa_supplicant could crash due to dereferencing freed memory in a corner
case. When a reestablished P2P GO group gets removed because the
hardware mode is not supported when switching from 2.4 to 5 GHz. Group
removal in wpas_p2p_group_delete() forget to clean wpa_s->last_ssid due
to the missing wpas_notify_network_removed() call before removing the
network with wpa_config_remove_network().
That could cause a crash when flushing old BSSs which age out in
wpa_bss_flush_by_age() which uses wpa_s->last_ssid in some cases.
Fix this by invoking wpas_notify_network_removed() to clean
wpa_s->last_ssid when removing a temporary P2P group network.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Define an attribute for high RSSI roam trigger threshold. STA is
expected to trigger roam if the current connected AP's RSSI gets above
this high RSSI threshold. STA's roam attempt on high RSSI threshold aims
to find candidates from other better Wi-Fi bands.
Signed-off-by: Sunil Dutt <quic_usdutt@quicinc.com>
Add three vendor roam control attributes to configure the roaming
parameters dynamically.
QCA_ATTR_ROAM_CONTROL_CONNECTED_LOW_RSSI_THRESHOLD controls the
connected AP's low RSSI threshold to trigger the neighbor lookup.
QCA_ATTR_ROAM_CONTROL_CANDIDATE_ROAM_RSSI_DIFF and
QCA_ATTR_ROAM_CONTROL_6GHZ_CANDIDATE_ROAM_RSSI_DIFF control the RSSI
difference threshold between the connected AP and the new candidate AP
for the roam to trigger.
Co-authored-by: Sunil Dutt <quic_usdutt@quicinc.com>
Signed-off-by: Sunil Dutt <quic_usdutt@quicinc.com>
Extend enum qca_wlan_vendor_attr_mlo_peer_prim_netdev_event to add MLD
MAC address, the number of links, and link info. Link info contains
ifindex and MAC address of each link of a non-AP MLD that was negotiated
in ML association.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Add vendor attribute IDs QCA_WLAN_VENDOR_ATTR_ROAM_STATS_ORIGINAL_BSSID,
QCA_WLAN_VENDOR_ATTR_ROAM_STATS_CANDIDATE_BSSID, and
QCA_WLAN_VENDOR_ATTR_ROAM_STATS_ROAMED_BSSID for updating roaming AP
BSSID to user space to enable user space collecting the BSSID for
roaming issues.
Signed-off-by: Chunquan Luo <quic_chunquan@quicinc.com>
Add support to obtain the start channel when checking whether DFS
is required when using 320 MHz channels.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Determine the channel width by operating class for the 6 GHz band when
validating puncturing bitmap. This is needed to allow puncturing to be
used with 320 MHz channels.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The MIC length depends on the negotiated group when SAE-EXT-KEY or OWE
key_mgmt is used. wlantest can determine the group if the capture file
includes the group negotiation, i.e., the initial association when a PMK
was created. However, if the capture file includes only an association
using PMKSA caching, the group information is not available. This can
result in inability to be able to process the EAPOL-Key frames (e.g.,
with the "Truncated EAPOL-Key from" message).
If the negotiated group is not known and an EAPOL-Key frame length does
not seem to match the default expectations for group 19, check whether
the alternative lengths for group 20 or 21 would result in a frame that
seems to have valid length. If so, update the STA entry with the guessed
group and continue processing the EAPOL-Key frames based on this.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Use helper function is_zero_ether_addr() for checking empty bssid
value in hostapd_driver_init().
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Signed-off-by: Manaswini Paluri <quic_mpaluri@quicinc.com>
Add mld_addr configuration option to set the MLD MAC address.
The already existing bssid configuration option can be used to
control the AP MLD's link addresses.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Signed-off-by: Manaswini Paluri <quic_mpaluri@quicinc.com>
When STAs connect to transmitted BSS and nontransmitted BSS, the
AP should assign the aIDs from the same pool.
Use the transmitted BSS AID pool to assign AIDs when the AP enables
multiple BSSID.
Signed-off-by: Allen.Ye <allen.ye@mediatek.com>
In the original flow, after hostapd_config_tx_queue() successfully
parses a tx_queue variable, wpa_config_process_global() would not return
immediately. Then it would print out "unknown global field" later and set
return val to -1.
Return success (0) after hostapd_config_tx_queue() successfully parses a
tx_queue variable to fix this.
Fixes: 790026c3da ("Allow TX queue parameters to be configured for wpa_supplicant AP/P2P GO")
Signed-off-by: Michael Lee <michael-cy.lee@mediatek.com>
Pass bss->ctx instead of drv->ctx in order to avoid multiple reports for
the first bss. The first report would otherwise clear hapd->cca_color and
subsequent reports would cause the iface bss color to be set to 0.
In order to avoid any issues with cancellations, only overwrite the color
based on hapd->cca_color if it was actually set.
Fixes: 33c4dd26cd ("BSS coloring: Handle the collision and CCA events coming from the kernel")
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Commit b20991da69 introduced errors in
the order of arguments to the calls of ccmp_decrypt() and
ccmp_256_decrypt(). Correct the order of arguments.
Fixes: b20991da69 ("wlantest: MLD MAC Address in CCMP/GCMP AAD/nonce")
Signed-off-by: Henry Ptasinski <henry@e78com.com>
Fix the warning:
wpa_supplicant.c:2257:5: warning: conflicting types for
‘wpas_update_random_addr’ due to enum/integer mismatch; have ‘int(struct
wpa_supplicant *, enum wpas_mac_addr_style, struct wpa_ssid *)’
[-Wenum-int-mismatch]
2257 | int wpas_update_random_addr(struct wpa_supplicant *wpa_s,
| ^~~~~~~~~~~~~~~~~~~~~~~
In file included from wpa_supplicant.c:32:
wpa_supplicant_i.h:1653:5: note: previous declaration of
‘wpas_update_random_addr’ with type ‘int(struct wpa_supplicant *, int,
struct wpa_ssid *)’
1653 | int wpas_update_random_addr(struct wpa_supplicant *wpa_s, int
style,
| ^~~~~~~~~~~~~~~~~~~~~~~
Fixes: 1d4027fdbe ("Make random MAC address style parameters use common enum values")
Signed-off-by: Chien Wong <m@xv97.com>
This extends testing coverage to detect an issue that was fixed in
commit bf9cbb462f ("Fix writing of BIGTK in FT protocol").
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
A copy-paste issue in wpa_ft_bigtk_subelem() ended up encoding the IGTK
value instead of the BIGTK when providing the current BIGTK to the STA
during FT protocol. Fix this to use the correct key to avoid issues when
beacon protection is used with FT.
Fixes: 16889aff40 ("Add BIGTK KDE and subelement similarly to IGTK")
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
QCA vendor extension is used for NDP setup. This defines the new
attributes QCA_WLAN_VENDOR_ATTR_NDP_CSIA_CAPABILITIES and
QCA_WLAN_VENDOR_ATTR_NDP_GTK_REQUIRED to support GTKSA, IGTKSA, and
BIGTKSA for NDP setup.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The new PTK migth need to be copied to another MLO STA entry, but that
operation should not modify the MLD MAC address of unrelated AP MLDs.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The "Updated AP MLD MAC Address from EAPOL-Key 1/4" can be confusing
when there is actually no change.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
This makes the "Discovered new STA" entries in the debug log easier to
use when analyzing roaming cases with MLO.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
When FT over-the-DS is used with MLO, the Target AP Address field is
expected to identify the AP MLD using its MLD MAC address.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Allow either a link address or the MLD MAC address of a non-AP MLD to
match the MAC address that is being used to identify a source or
destination of a frame for the MLO cases.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
FT over-the-DS might have created the new STA entry on another
affiliated BSS during the FT Request/Response exchange, so use a wider
search to locate the correct STA entry when processing the Reassociation
Request/Response frames.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Use the Basic Multi-Link element in (Re)Association Request frames to
learn the non-AP MLD MAC address instead of having to wait until this
address is included in an EAPOL-Key frame. This is needed for FT
protocol (where 4-way handshake is not used) and it is also convenient
to have the MLD MAC address available as soon as possible to be able to
decrypt frames and even to recognize some special AP vs. STA cases when
either the BSSID or the AP MLD MAC address might be used.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Compare A1 against all the link addresses of a non-AP MLD when
determining whether a Data frame is from the non-AP MLD or the AP MLD
during a decryption attempt.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Make sta_find_mlo() more accurate by searching a non-AP MLD only from
the affialiated BSSs of the AP MLD instead of from any BSS. This might
help in some roaming cases where both the old and the new AP MLD have
their affiliated links in the BSS table.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Use the Basic Multi-Link element in Beacon frames (and Probe Response
frames for that matter) to learn the AP MLD MAC address instead of
having to wait until this address is included in an EAPOL-Key frame.
This is needed for FT protocol (where 4-way handshake is not used) and
it is also convenient to have the MLD MAC address available as soon as
possible to be able to decrypt frames and even to recognize some special
AP vs. STA cases when either the BSSID or the AP MLD MAC address might
be used.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The set of protected elements in the FTE in Reassociation Response frame
is different for MLO. Count RSNE and RSNXE separately for each link.
This implementation uses the number of links for which a GTK was
provided which does not fully match the standard ("requested link") and
a more accurate implementation is likely needed, but that will require
some more complexity and state information.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Defragment the FTE if it was fragmented. This is needed for MLO when the
FTE in Reassociation Response frame might be longer than 255 octets to
include all the group keys for all the links.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Allow the -r<file> command line argument to be used multiple times to
read more than a single capture file for processing. This reduces need
for external tools to be used first to merge capture files for wlantest.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The Key MIC field is of variable length when using OWE, so determine the
correct length based on which group was negotiated for OWE during
association.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Changing sae_pwe and leaving the modified value for the following test
cases can result in failures.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Parse link id and station MLD address received from the driver in the
NL80211_CMD_NEW_STA event.
Set MLO information of the station to the sta_info and wpa_sm.
Co-authored-by: Manaswini Paluri <quic_mpaluri@quicinc.com>
Signed-off-by: Veerendranath Jakkam <quic_vjakkam@quicinc.com>
Add check to skip the STA mode specific MLO link channel switch handling
in AP mode. Commit 1b6f3b5850 ("MLD STA: Indicate per link channel
switch") added this indication only for STA mode.
Signed-off-by: Manaswini Paluri <quic_mpaluri@quicinc.com>
Add support to parse association link id and MLD address from the
NL80211_CMD_UPDATE_OWE_INFO event.
Set MLO information of the station to the sta_info and wpa_sm.
Use station association link address for sending DH IE info to the
driver.
Signed-off-by: Manaswini Paluri <quic_mpaluri@quicinc.com>
Use station association link address for sending SAE authentication
status to the driver in AP mode external authentication status.
Signed-off-by: Manaswini Paluri <quic_mpaluri@quicinc.com>
Set the sme RRM used config if the RRM element is present in the
(Re)Association Request frame sent in association event to cover the
cases where the driver SME takes care of negotiating RRM capabilities.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Extend QCA_WLAN_VENDOR_ATTR_CONFIG_OPTIMIZED_POWER_MANAGEMENT
attribute to support enum qca_wlan_vendor_opm_mode.
Add QCA vendor attribute QCA_WLAN_VENDOR_ATTR_CONFIG_OPM_ITO and
QCA_WLAN_VENDOR_ATTR_CONFIG_OPM_SPEC_WAKE to configure inactivity
timeout and speculative wake interval in User defined optimized
power save mode.
Signed-off-by: Vishal Miskin <quic_vmiskin@quicinc.com>
Verify that the operation succeeds before a debug print indicating that
it did. This was already done in most callers, so be more consistent and
do it here as well.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
The second argument to memset() is only eight bits, so there is no point
in trying to set 0xffff values for an array of 16-bit fields. 0xff will
do the exact same thing without causing static analyzes warnings about
truncated value.
Fixes: 903e3a1e62 ("FILS: Fix maximum NSS calculation for FD frame")
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
ideal_chan is NULL here, so it is not really valid to try to debug print
something from it due to the implied NULL pointer dereferencing.
Fixes: af0f60e7dd ("EHT: Calculate puncturing bitmap for ACS")
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Check the ieee802_11_parse_elems() return code and do not proceed in
various cases if parsing failed. Previously, these cases would have been
allowed to continue by ignoring whatever might have followed in the IE
buffer after the first detected parsing failure. This is not really an
issue in practice, but it feels cleaner to explicitly stop when
receiving an invalid set of IEs.
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
Optimize the search for nonzero octets when checking for the need to
work around WPS M1 padding. The previous implementation was really
inefficient (O(n^2)) and while that was likely sufficiently fast for the
cases where the MMPDU size limit prevents long buffers (e.g., all P2P
Action frames), it might be able to take tens of seconds on low-end CPUs
with maximum length EAP-WSC messages during WPS provisioning. More
visibly, this was causing OSS-Fuzz to time out a test case with
unrealisticly long data (i.e., almost 10 times the maximum EAP-WSC
buffer length).
Credit to OSS-Fuzz: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=60039
Signed-off-by: Jouni Malinen <quic_jouni@quicinc.com>
