A special value p2p_group_idle=-1 can now be used to configure the P2P
group idle mechanism to terminate a P2P client group immediately on any
disconnection after the completion of the initial 4-way handshake.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
Instead of relying on the P2P group idle timeout before the group
connection has been fully established, re-start the group formation
timeout in the end of the WPS provisioning step and clear it at the
successful completion of the initial 4-way handshake. This allows the
P2P group idle timeout to be set to a small value without triggering it
during the initial scan and connection attempt.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
wpa_drv_scan() may fail for the initial p2p_connect join scan request,
e.g., if the driver happened to be scanning at the time the new
operation was initialized. Previously, a special scan result handler was
registered regardless of whether the new scan was started. This could
result in partial scan results (e.g., from p2p_find social scan) from
being used as full results for join (or now more importantly for
p2p_connect auto) purposes. Fix this by registering the new scan result
handler only if wpa_drv_scan() returns success.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
p2p_connect command can now be used with an optional "auto" parameter
to request wpa_supplicant to determine automatically whether to use
join-a-group operation (if the peer is operating as a GO) or group
formation. This makes it easier for external programs to handle
connection type selection by offloading this to wpa_supplicant. The
previously used p2p_connect join commands can be replaced with
p2p_connect auto to use this new mechanism.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
Unlike the unicast SD queries, the queries directed to all peers depend
on P2P_DEV_SD_INFO flag being cleared to allow the query to be sent to
a peer that has previously replied to any SD query.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
Previously, the timeout scheduled during the previous association was
used after reassociation if the STA entry had not yet been removed. The
next timeout does not need to happen that quickly, so reschedule it
during reassociation.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
This makes it easier to figure out what exactly was done with the
ap_handle_timer registration/cancellation based on a debug log.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
The next ap_handle_timer action was already cleared when association
is handled with user space -based SME. However, this step was missing
from the driver callback for indicating new association. This could
result in the first ap_handle_timer timeout on the new association
removing the station unexpectedly. Fix this by resetting the timeout_next
in hostapd_notif_assoc().
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
intended-for: hostap-1
A station may move from EAP/WPS key_mgmt to PSK using re-association. In
practice, this can happen most likely with WPS when the station runs
through the provisioning step (EAP/WPS) followed by PSK authentication.
If a deauthentication event is missed from the station for whatever
reason, the EAPOL authenticator state machine could have been left
running.
This can result in various issues, including unexpected disconnection of
the station while both the 4-way handshake and EAPOL authenticator state
machines (EAP) are running at the same time when the supplicant is
trying to use PSK. Fix this by explicitly clearing EAPOL authenticator
state machine on (re)association if the new association does not use it.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
intended-for: hostap-1
The previous commit did not use the correct pointer in all operations
and was specific to station mode interfaces. Fix and extend it to work
with AP/GO interfaces, too.
Signed-hostap: Jouni Malinen <j@w1.fi>
If the driver does not indicate support for multi-channel concurrency,
abort join-group operation if the end result would result in use of
multiple operating frequencies with the same radio.
Signed-hostap: Jouni Malinen <j@w1.fi>
When starting an AP with HT40 on 2.4 GHz, limit the set of channels
to scan for based on the affected frequency range to speed up the
AP setup.
Signed-hostap: Jouni Malinen <j@w1.fi>
Mark the debug print excessive and print it only in case the entropy
collection is used since this function can get called pretty frequently.
Signed-hostap: Jouni Malinen <j@w1.fi>
OpenSSL wrapper was using the same certificate store for both Phase 1
and Phase 2 TLS exchange in case of EAP-PEAP/TLS, EAP-TTLS/TLS, and
EAP-FAST/TLS. This would be fine if the same CA certificates were used
in both phases, but does not work properly if different CA certificates
are used. Enforce full separation of TLS state between the phases by
using a separate TLS library context in EAP peer implementation.
Signed-hostap: Jouni Malinen <j@w1.fi>
This brings up the changes from defconfig into the Android configuration
file, but does not change any of the actual build parameters.
Signed-hostap: Jouni Malinen <j@w1.fi>
This was never really completed and is of not much use at this point,
so clean up the repository by removing the Symbian-specific changes.
Signed-hostap: Jouni Malinen <j@w1.fi>
This optional parameter to the BSS command can be used to select which
fields are included in the output to avoid having to parse through
unneeded information and to reduce the buffer size.
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>
There is no need to subscribe to event messages during deinit process,
so skip this to avoid unnecessary operations and to keep the debug logs
a bit cleaner.
Signed-hostap: Jouni Malinen <j@w1.fi>
When a supplicant is deinited and shutting, disconnect from P2P groups.
This fixes a memory leak on variable dbus_groupobj_path on exiting
supplicant.
Signed-hostap: Nirav Shah <nirav.j2.shah@intel.com>
Signed-hostap: Angie Chinchilla <angie.v.chinchilla@intel.com>
When forming a P2P group using WSC PIN method, if the PIN is entered
incorrectly the P2P client supplicant instance will crash as a result
of cleanup happening on data that is still in use in a case where a
separate P2P group interface is used.
For example, here is the path for the first crash:
eap_wsc_process():
- creates struct wpabuf tmpbuf; on the stack
- sets data->in_buf = &tmpbuf;
- calls wps_process_msg()
- which calls wps_process_wsc_msg()
- which, in case WPS_M4: calls wps_fail_event()
- which calls wps->event_cb()
- wps->event_cb = wpa_supplicant_wps_event()
- wpa_supplicant_wps_event()
- wpa_supplicant_wps_event_fail()
- which calls wpas_clear_wps()
- which calls wpas_notify_network_removed()
- which calls wpas_p2p_network_removed()
- which calls wpas_p2p_group_formation_timeout()
- which calls wpas_group_formation_completed()
- which calls wpas_p2p_group_delete()
- which calls wpa_supplicant_remove_iface()
- which calls wpa_supplicant_deinit_iface()
- which calls wpa_supplicant_cleanup()
- which calls eapol_sm_deinit()
- ... which eventually uses the ptr data->in_buf to free tmpbuf, our
stack variable and then the supplicant crashes
If you fix this crash, you'll hit another. Fix it and then a segfault.
The way we're cleaning up and deleting data from under ourselves here
just isn't safe, so make the teardown portion of this async.
Signed-hostap: Angie Chinchilla <angie.v.chinchilla@intel.com>
Signed-hostap: Nirav Shah <nirav.j2.shah@intel.com>
intended-for: hostap-1
There is no need to force this configuration parameter to be written
into the configuration file for each network block that has not modified
the default value.
Signed-hostap: Jouni Malinen <j@w1.fi>
I think this comment, as originally stated, says the opposite of
what is meant. Reword to help the next guy.
Signed-hostap: Angie Chinchilla <angie.v.chinchilla@intel.com>
For concurrency scenarios we should not update the scan results without
P2P IE to P2P group interfaces. This patch is to take care if a scan has
been requested on STA interface and scan results are getting updated
from sibling to P2P group interface. A P2P GO Probe Response for a
legacy STA scan would be without P2P IE. So it will wrongly update the
BSS list of P2P group interface too when results are updated from
sibling.
Since P2P group interfaces are used only for connecting to a P2P group,
it is fine to filter out all non-P2P entries from their BSS table.
Signed-off-by: Neeraj Garg <neerajkg@broadcom.com>
This should keep us from getting into a state where we bounce large
numbers of stations all at once. Spreading out the bounce should cause
less stress on the network as the idle stations won't all be trying to
reconnect at once.
Signed-hostap: Ben Greear <greearb@candelatech.com>
For debugging I needed to look at the frames in a wpa_supplicant debug
log. Looking at the hexdump is a bit tedious, so I wrote a quick script
to convert the nl80211 debugging output to a PCAP file that can be
opened in wireshark.
I've polished the initial raw script a bit to add error messages and to
also record the timestamps to the pcap file.
Hopefully it's useful to somebody else as well.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
wps_vendor_ext_m1 configuration parameter can now be used to add a
vendor specific attribute into the WPS M1 message, e.g., for
Windows Vertical Pairing.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
This was previously indicated as disassociation, but the P2P group
session ending case expect a deauthentication notification. Since the
cfg80211 disconnection event does not really indicate which frame was
used, the easiest approach for now seems to be to just indicate these
as deauthentication events.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
The GO can indicate that the P2P Group session is ending by sending a
Deauthentication frame with reason code 3 (Deauthenticated because
sending STA is leaving) based on P2P specification section 3.2.9. Use
this reason code to remove the P2P client group without waiting for the
group idle timeout.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
Use reason code 3 (Deauthenticated because sending STA is leaving)
instead of reason code 2 (Previous authentication ot valid) in the
Deauthentication frame sent at the time a BSS is terminated. This is
more accurate reason for the message and allows the stations to get
better understand on why they were disconnected. This can be useful
especially for P2P clients that can use this as a signal that the
P2P group session was ended per P2P specification section 3.2.9.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
wpas_notify_state_changed() notified about disconnection when
new_state < WPA_ASSOCIATED, without checking the old state. This
results in extra processing which can be avoided by checking that
old_state was connected.
Signed-hostap: Eliad Peller <eliad@wizery.com>
intended-for: hostap-1
On P2P group removal, the GO is deinitialized correctly (and the vif
mode is set back to sta in case of nl80211), but the P2P client mode
wasn't deinitialized, and the nl80211 vif stays in P2P client mode.
Add a new deinit_p2p_cli op (similar to deinit_ap), which currently only
sets the interface back to station mode.
Signed-hostap: Eliad Peller <eliad@wizery.com>
intended-for: hostap-1
Commit 2d43d37ff2 broke EVENT_RX_MGMT
processing in wpa_supplicant AP mode. ap_mgmt_rx() needs to be called
for Probe Request frames even if they get notified through the new
D-Bus signal mechanism.
Signed-hostap: Jouni Malinen <j@w1.fi>
Some applications require knowing about probe requests to identify
devices. This can be the case in AP mode to see the devices before they
connect, or even in P2P mode when operating as a P2P device to identify
non-P2P peers (P2P peers are identified via PeerFound signals).
As there are typically a lot of probe requests, require that an
interested application subscribes to this signal so the bus isn't always
flooded with these notifications. The notifications in DBus are then
unicast only to that application.
A small test script is also included.
Signed-hostap: Johannes Berg <johannes.berg@intel.com>
The signal strength is currently never used as the only driver reporting
it is nl80211 which uses IEEE80211_RADIOTAP_DB_ANTSIGNAL which is never
populated by the kernel. The kernel will (soon) populate
IEEE80211_RADIOTAP_DBM_ANTSIGNAL instead though, so use that.
Also, since it was never really populated, we can redefine the signal
field to be in dBm units only.
My next patch will also require knowing the signal strength of probe
requests throughout the code (where available), so add it to the
necessary APIs.
Signed-hostap: Johannes Berg <johannes.berg@intel.com>
When hostapd is about to start an AP using HT40, it starts a scan to
check whether it is allowed or not. If OLBC is detected before the
scan has completed, it sets the beacons prematurely.
To fix this, instead of setting all beacons when OLBC is detected,
only update the ones that have already been started.
Signed-off-by: Luciano Coelho <coelho@ti.com>
Enable control of wpa_s->scan_interval via D-Bus. This parameter
controls the delay between successive scans for a suitable AP.
Also, update dbus.doxygen with ScanInterval, and some other
parameters that were undocumented.
Signed-hostap: Mukesh Agrawal <quiche@chromium.org>
ieee802_1x_encapsulate_radius() frees the RADIUS message if
radius_client_send() returns error. This could have resulted in use of
freed memory and double freeing of the RADIUS message if send() fails
since the message is also left in the retransmit list. Avoid this by not
returning error to the caller in such a case.
Signed-off-by: Alexander Couzens <lynxis@c-base.org>
