The currently unused conf_offset parameter used a mismatching type (enum
vs. unsigned int) compared to the prototype.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Need to handle the little endian 16-bit fields properly when building
and parsing Authentication frames.
Fixes: 5ff39c1380 ("SAE: Support external authentication offload for driver-SME cases")
Signed-off-by: Ashok Ponnaiah <aponnaia@codeaurora.org>
Instead of going through the configuration exchange, reject invalid
legacy configurator parameters explicitly. Previously, configuring
legacy (psk/sae) parameters without psk/pass resulted in a config object
that used a zero length passphrase. With this change, that config object
is not sent and instead, either the initialization attempts is rejected
or the incoming initialization attempt is ignored.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
The new conf variable was used only within the CONFIG_TESTING_OPTIONS
block and as such, added a warning about unused variable into
non-testing builds. Fix that by using that variable outside the
conditional block as well.
Fixes: a22e235fd0 ("OWE: Add testing RSNE for OWE assoc response with driver SME/MLME")
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
This commit enhances QCA_NL80211_VENDOR_SUBCMD_LINK_PROPERTIES to
also be an event, aimed to notify the link status (EX: connected
stations status on an AP link).
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
When external authentication is used, a specific netlink socket is used
to send the connect command. If the same socket is not used for
disconnect command, cfg80211 will discard the command. This constraint
was added into the kernel in commit bad292973363 ("nl80211: Reject
disconnect commands except from conn_owner"). That requires an update
for the hostap.git commit 40a68f3384 ("nl80211: Create a netlink
socket handle for the Connect interface").
Add a new flag into struct i802_bss to indicate if the special
nl_connect socket was used for the connect command. When sending
disconnect command this flag is tested to select the correct socket.
Signed-off-by: Cedric Izoard <cedric.izoard@ceva-dsp.com>
When selecting SSID to start external authentication procedure also
check the key_mgmt field as several network configuration may be defined
for the same SSID/BSSID pair. The external authentication mechanism is
only available for SAE.
Signed-off-by: Cedric Izoard <cedric.izoard@ceva-dsp.com>
The call to remove_group() may fail, in which case all following
cleanup is skipped. This may result in failing many tests since
cleanup did not complete successfully.
Fix this by calling remove_group() after other cleanup is done so
even it fails it will not affect the following tests.
Signed-off-by: Avraham Stern <avraham.stern@intel.com>
Add documentation to the wpa_signal_info structure.
Add a define for an invalid noise value.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
BSS additional/removal cases were not considered at all in the previous
implementation of hostapd configuration file reloading on SIGHUP. Such
changes resulted in num_bss values getting out of sync in runtime data
and configuration data and likely dereferencing of freed memory (e.g.,
when removing a BSS).
Fix this by forcing a full disable/enable sequence for the interface if
any BSS entry is added/removed or if an interface name changes between
the old and the new configuration.
Signed-off-by: Jouni Malinen <j@w1.fi>
Implement ECDSA signing functionality in the Python test script for
generating a valid signedConnector. This allows coverage of DPP config
object testing to be increased more easily.
Signed-off-by: Jouni Malinen <j@w1.fi>
A local memory allocation failuring during GAS Comeback Response frame
generation could result in freeing the response context without removing
it from the list. This would result in dereferencing freed memory when
processing the next comeback request.
Signed-off-by: Jouni Malinen <j@w1.fi>
If local memory allocation for the GAS response failed, couple of error
paths ended up leaking some memory maintaining the state for the
exchange. Fix that by freeing the context properly.
Signed-off-by: Jouni Malinen <j@w1.fi>
The GAS client processing of the response callback for DPP did not
properly check for GAS query success. This could result in trying to
check the Advertisement Protocol information in failure cases where that
information is not available and that would have resulted in
dereferencing a NULL pointer. Fix this by checking the GAS query result
before processing with processing of the response.
Signed-off-by: Jouni Malinen <j@w1.fi>
This lets us know whether we can attempt to use FT-PSK, FT-EAP,
FT-EAP-SHA384, FT-FILS-SHA256 or FT-FILS-SHA384.
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Based on Jouni Malinen's [76055b4c6 "tests: D-Bus Get/Set Pmf"], modified
to use the correct "s" signature for the "Pmf" property.
Removed the negative test cases, because the synthesized property doens't
seem to do error checking upon being set.
Signed-off-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
This reverts commit adf8f45f8a.
It is basically all wrong. The Pmf property did exist, with a signature of
"s" as documented in doc/dbus.doxygen. It was synthesized from
global_fields[].
The patch added a duplicate one, with a signature of "u", in violation
of D-Bus specification and to bemusement of tools that are careful
enough:
$ busctl introspect fi.w1.wpa_supplicant1 /fi/w1/wpa_supplicant1/Interfaces/666
Duplicate property
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Make the buf_len calculation match more closely with the following
wpa_buf*() operations. The extra room from the existing elements was
apparently sufficiently large to cover this, but better add the two
octet header explicitly.
Signed-off-by: Jouni Malinen <j@w1.fi>
The maximum size of a Mesh Peering Management element in the case
of an AMPE close frame is actually 24 bytes, not 23 bytes, plus the
two bytes of the IE header (IEEE Std 802.11-2016, 9.4.2.102). Found by
inspection.
The other buffer components seem to use large enough extra room in their
allocations to avoid hitting issues with the full buffer size even
without this fix.
Signed-off-by: Bob Copeland <bobcopeland@fb.com>
use 'printf' instead of 'echo -n', to suppress the following warning:
In POSIX sh, echo flags are undefined. [SC2039]
Signed-off-by: Davide Caratti <davide.caratti@gmail.com>
The hwsim's start.sh script spawns hostapd process using "sudo".
Since sudo forks a child process, $! holds the pid of sudo itself.
Fix that by storing the PID of the child process instead.
Since in VM "sudo" is replaced with a dummy script, pass an additional
argument to run-all.sh and start.sh scripts to indicate that they are
running inside a VM.
This is needed to fix ap_config_reload and ap_config_reload_file test
cases on some platforms where sudo is apparently not relaying the
signals properly.
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
In ap_acl_deny test, the AP doesn't send probe responses during scan due
to ACL reject. As the result, dev[0] might miss the AP's Beacon frame
because the dwell time is too short. Make the test more robust and
trigger passive scan, and by that increase the probability of hearing
the AP.
Signed-off-by: Ayala Beker <ayala.beker@intel.com>
Back in December 2017, Jouni fixed the output side since that was
causing a kernel message to be printed, but the input side should
also be fixed, otherwise it will not work correctly on big-endian
platforms.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
