There could be multiple pending TX operations and if the earlier ones
have used wait_time, but the last one did not, the driver call for
canceling pending wait was not done. This could result in the driver
getting stuck waiting for the previously scheduled wait time and not
being able to do new operations until that. Fix this by canceling the
wait if any of the past offchannel_send_action() calls since the last
offchannel_send_action_done() used non-zero wait_time.
This was showing up as issues in certain DPP Public Action frame
sequences when the same offchannel operation is used with multiple
frames and the last frame in the sequence does not need wait_time.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Introduction of the new base64 helper function changed the backtraces
for these OOM test cases and resulted in test failures. Update the test
scripts to work with the new implementation.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The additional SHA-384 and SHA-512 functionality is needed to support
DPP with various ECC curves.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This can be helpful in figuring out what happened if P2P_FIND operation
is unexpectedly rejected.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
wpa_s->p2p_in_provisioning needs to be cleared when group formation
fully completes. The change to postpone GO side handling to the first
data connection in commit 41f853235f
('P2P: Extend group formation timeout on GO to first data connection')
resulted in making this not happen in one P2P Client side case: EAP-WSC
timeout in PBC case. While that is quite special case since it requires
30 second timeout without receiving new EAPOL frames and not getting
disassociation, it can apparently happen in some cases in practice. This
would result in new P2P operations (e.g., P2P_FIND) getting rejected
until wpa_supplicant is restarted.
Fix this by clearing wpa_s->p2p_in_provisioning whenever processing a
group formation failure case. For group formation success,
wpa_s->p2p_in_provisioning is left set to non-zero value to avoid
breaking the earlier limits on the GO side.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This allows the eap_proxy mechanism to be used with multiple SIMs by
following the configured sim_num to index which SIM to use for when
fetching the IMSI through eap_proxy.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
For proxy based EAP methods, the EAP identity is constructed in
eap_proxy layer from IMSI when required. Realm information from identity
is used to do ERP eventually, hence construct the realm for proxy based
methods from IMSI in core wpa_supplicant to enable the ERP use case.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
wpa_supplicant tries to reinitialize an interface when a previously
removed netdev is restored (e.g., re-insert a USB dongle). If that
initialization fails (e.g., driver ejects ifconfig UP), the previous
implementation resulted in leaving the interface in incomplete state
while still claiming to upper layers that the interface status has
changed back to functional one.
Fix this by skipping the interface status update if reinitialization
fails. In other words, remain in INTERFACE_DISABLED state if the
interface cannot be re-enabled successfully.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit adds support for deriving ERP key information in EAP Proxy
based EAP method implementations.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
hostapd processes a received 20/40 BSS Coexistence management frame, but
if no separate callbacks are registered for handling Public Action
frames it eventually sends a reply with MSB of category code set to 1
thinking that the received frame is an invalid frame. This could happen
based on whether hostapd was built and enabled with functionality using
the callback functions.
Fix this by explicitly returning 1 from the function when the 20/40 BSS
Coexistence Management frame is processed.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The new acs_exclude_dfs=1 parameter can be used to request hostapd to
exclude all DFS channels from ACS consideration. This is mainly of use
for cases where the driver supports DFS channels, but for some reason a
non-DFS channel is desired when using automatic channel selection.
Previously, the chanlist parameter could have been used for this, but
that required listing all the acceptable channels. The new parameter
allows this to be done without such a list.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit is similar to the commit
783c2920cc ('P2P: Check if the pref_freq
reported by the driver supports P2P') but extends the check for
supported pref_freq to wpas_p2p_select_go_freq().
This avoids issues with P2P_GROUP_ADD ending up selecting an unexpected
channel when get_pref_freq_list() (i.e.,
QCA_NL80211_VENDOR_SUBCMD_GET_PREFERRED_FREQ_LIST) is used. Filter the
list by removing channels that do not allow P2P operation at all.
Previously, only the explicitly disallowed channels were removed and
that could have resulted in selecting an operating channel that is not
allowed for P2P and failing to complete the operation to start the
group.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
It was possible for the BTM Request with no matching BSSIDs to cause
wpa_supplicant-initated roam to a better BSS (5 GHz band preferred) when
finding the second AP in a scan started by that BTM Request. This could
make the following step in the test case fail. Fix this by asking
another channel to be scanned to postpone discovery of the other AP.
Signed-off-by: Jouni Malinen <j@w1.fi>
This makes it easier to synchronize log entries in the kernel log
(seconds from boot) and wpa_supplicant/hostapd (UNIX timestamp).
Signed-off-by: Jouni Malinen <j@w1.fi>
Add NULL checking for EAP name. If it is NULL, do not add the phase2
parameter autheap. This should not happen in practice due to earlier
checks for credential matching, but if there is a code path that would
allow this to be set, it is better to skip setting of the invalid value
and allow automatic selection of the Phase 2 parameters.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit renames the vendor attribute
QCA_WLAN_VENDOR_ATTR_CONFIG_DISABLE_OFFCHANNEL to
QCA_WLAN_VENDOR_ATTR_CONFIG_RESTRICT_OFFCHANNEL as intended by the
original commit d506c35efc ('Set Wi-Fi
Configuration attribute to restrict offchannel operations').
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit defines an attribute to
QCA_NL80211_VENDOR_SUBCMD_SET_WIFI_CONFIGURATION -
QCA_WLAN_VENDOR_ATTR_CONFIG_RESTRICT_OFFCHANNEL which can be used to
restrict offchannel operations on the AP/GO interface.
The goal is to restrict any operations which would cause the AP/GO to
leave its operating channel.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit adds changes to not generate ERP information if the domain
name is not specified in the EAP identity. keyName-NAI needs the realm
part and as such, it is reasonable to require the main EAP configuration
to provide that realm.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
These test cases depend on ERP processing to reach the get_emsk handler
function. Since ERP really needs the realm to derive a proper
keyName-NAI, modify these test cases to pass the realm part in the
identity to allow error checking to be introduced for rejecting ERP
cases where the realm is not available.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Change in any network configuration at runtime will cause flush to
PMKSA cache. For most of the network parameters if there is no change
in value, PMKSA flush is not performed except 'bssid' and 'priority'.
Add 'bssid_hint' to exemption list of avoiding PMKSA flush on change.
This is needed to complete change in commit
43a356b268 ('Provide option to configure
BSSID hint for a network').
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Commit 3e0272ca00 ('nl80211: Re-read MAC
address on RTM_NEWLINK') added the detection of external changes to MAC
address when the interface is brought up.
If the interface state is changed quickly enough, wpa_supplicant may
receive the netlink message for the !IFF_UP event when the interface
has already been brought up and would ignore the next netlink IFF_UP
message, missing the MAC change.
Fix this by also reloading the MAC address when a !IFF_UP event is
received with the interface up, because this implies that the
interface went down and up again, possibly changing the address.
Signed-off-by: Beniamino Galvani <bgalvani@redhat.com>
This is to comply with uniform spreading requirement for ETSI domain
(section 4.7.2.7 in EN 301 893 - V1.8.1). ETSI uniform spreading
requires equal probability for the usable channels. The previous channel
selection logic after a radar detection did not fully comply with the
uniform spreading requirement for the domain by ignoring DFS channels.
Consider DFS channels also during channel selection when the current DFS
domain is ETSI.
Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qti.qualcomm.com>
Current DFS domain information of the driver can be used in ap/dfs
to comply with DFS domain specific requirements like uniform spreading
for ETSI domain.
Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qti.qualcomm.com>
This commit introduces the following two attributes to
QCA_NL80211_VENDOR_SUBCMD_KEY_MGMT_ROAM_AUTH events:
QCA_WLAN_VENDOR_ATTR_ROAM_AUTH_STATUS - Indicates the status of
re-association requested by user space
QCA_WLAN_VENDOR_ATTR_ROAM_AUTH_RETAIN_CONNECTION - Indicates whether
the old association was maintained when a re-association
is requested by user space and that re-association attempt
fails (i.e., cannot connect to the requested BSS, but can
remain associated with the BSS with which the association was
in place when being requested to roam).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This exposes user configurable option to set bssid_hint for a network.
bssid_hint indicates which BSS has been found a suitable candidate for
initial association for drivers that use driver/firmware-based BSS
selection. Unlike the bssid parameter, bssid_hint does not limit the
driver from selecting other BSSs in the ESS.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit defines an attribute
QCA_WLAN_VENDOR_ATTR_CONFIG_LISTEN_INTERVAL which allows the currently
used listen interval to be updated using
QCA_NL80211_VENDOR_SUBCMD_SET_WIFI_CONFIGURATION. This update applies
only during the association and is done without updating the AP about
the change.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The previous fix to the OCSP request construction ended up finally
moving from SHA-1 -based hash to SHA-256 for OCSP test cases. To
maintain coverage for SHA-1, add cloned versions of the two test cases
so that both SHA-256 and SHA-1 cases get covered.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Fix the openssl ocsp command line and check if it returns an error - so
that instead of having something unusable later we error out
immediately. Moving the -sha256 argument earlier fixes hash function use
for the OCSP request generation (the old version used SHA-1).
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Previously, only SHA1 hash -based server certificate matching was used,
but the OCSP response may use SHA256 instead of SHA1, so check the match
with both hash functions, if needed.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Previously, only SHA1 hash -based server certificate matching was used,
but the OCSP response may use SHA256 instead of SHA1, so check the match
with both hash functions, if needed.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Addition of remove_ies() handled the CONFIG_IEEE80211R dependency, but
missed the caller being within CONFIG_FILS as well.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
When reducing the configuration for MACsec/MKA to the bare minimum, so
no EAP authentication, just MACsec/MKA SA use with preshared key/name,
the EAPOL engine is still needed to run the protocol for MKA. Without
any EAP authentication options the IEEE8021X_EAPOL option is not set,
resulting in a non-working Key Agreement Entity.
Therefore the CONFIG_MACSEC block needs to move up and set the
IEEE8021X_EAPOL option.
Signed-off-by: Jaap Keuter <jaap.keuter@xs4all.nl>
In some practical cases, it is useful to suppress joining to node in the
distance. The new field mesh_rssi_threshold could be used as RSSI
threshold for joining.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This needs to allow all 20 MHz channel numbers to be converted even
though the Annex E table lists only channel _center_ frequencies 50 and
114. Neighbor Report (see IEEE Std 802.11-2016, 9.4.2.37 Neighbor Report
element) uses Channel Number field with "last known primary channel of
the AP" which refers to the 20 MHz channel and not the channel center
frequency.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This patch adds "TDLSCancelChannelSwitch" dbus method on
"fi.w1.wpa_supplicant1.Interface" interface to disable channel switching
with a TDLS peer.
Signed-off-by: Nishant Chaprana <n.chaprana@samsung.com>
This patch adds "TDLSChannelSwitch" dbus method on
"fi.w1.wpa_supplicant1.Interface" interface to enable channel switching
with a TDLS peer.
Signed-off-by: Nishant Chaprana <n.chaprana@samsung.com>
Derive PMK-R0 and the relevant key names when using FILS authentication
for initial FT mobility domain association.
Signed-off-by: Jouni Malinen <j@w1.fi>
Derive PMK-R0 and the relevant key names when using FILS authentication
for initial FT mobility domain association. Fill in the FT IEs in
(Re)Association Request frame for this.
Signed-off-by: Jouni Malinen <j@w1.fi>
