Stop AP command can be used by the driver as an event to indicate that
AP mode has stopped operation. WLAN driver may have encountered errors
that has forced the driver to report this event or concurrent operations
on virtual interfaces may have forced AP operation to be stopped. When
in P2P GO mode, wpa_supplicant uses this event to remove P2P group to
keep in sync with the driver state.
It turned out that this breaks lots of use cases where p2p_find is
issued while already in p2p_listen state. As such, we cannot reject
p2p_find this easily without checking for more specific cases.
Deepthi Gowri [Thu, 1 Aug 2013 12:39:47 +0000 (15:39 +0300)]
P2P: Reject p2p_find while P2P connection is in progress
Though p2p_find is not expected during ongoing P2P connection, it is
possible that any third party application issues a p2p_find resulting in
connection failure. Address this by rejecting any p2p_find command while
connection is in progress.
Sunil Dutt [Thu, 1 Aug 2013 07:47:23 +0000 (10:47 +0300)]
TDLS: Handle transmission failures of TPK Handshake messages
A transmission failure of the TDLS Setup Request frame (TPK Handshake
M1) results in no further retries and the peer entry being left in state
where all TDLS Setup Request frames from the peer with higher address
would be rejected, thus always resulting in a failure to establish a
TDLS link. Handle the failures in transmission by disabling the link
immediately to ensure the traffic to the peer goes through the AP. This
commit also handles similar transmision failures for TPK Handshake M2
and M3 frames (TDLS Setup Response and Confirm).
Wei-Jen Lin [Wed, 31 Jul 2013 20:22:04 +0000 (23:22 +0300)]
P2P: Fix invalid remain-on-channel duration for frame TX
cfg80211 does not allow the zero duration of remain-on-channel. Instead,
use 20 ms as default waiting time when remain-on-channel is used to
schedule offchannel transmission that does not expect a response.
Sean Lin [Wed, 31 Jul 2013 20:09:34 +0000 (23:09 +0300)]
P2P: Do not delay retries on failure during group formation
4-way handshake may fail under extremely noisy environment and if this
happens during P2P group formation, the 10 second extra delay added in
wpas_auth_failed() can result in running over the 15 second timeout.
Avoid this by skipping the delay mechanism in wpas_auth_failed() for the
P2P group formation case. The P2P formation timeout will take care of
stopping the attempts if the failure condition does not get resolved.
nl80211: Fix TDLS key configuration to not set TX key index
The nl80211 command for setting the TX index does not distinguish TDLS
vs. AP key and as such, the driver would not know what this set TX key
index operation is doing in the TDLS case. This could result in the TX
key index for AP being changed instead if static WEP is used in the AP
connection. Fix the issue by not setting TX key index when configuring a
TDLS key.
This is not used anymore after the previous commits that changed the
driver interface to use number of supported concurrent channels instead
of this flag.
Signed-hostap: Ilan Peer <ilan.peer@intel.com>
Signed-hostap: David Spinadel <david.spinadel@intel.com>
Use wpa_drv_shared_freq() if get_radio_name() is not supported
Some driver interfaces may not support the get_radio_name() design and
get_shared_radio_freqs() needs to be aware of such possibility when
determining shared radio frequencies.
Ilan Peer [Sun, 21 Jul 2013 17:03:53 +0000 (20:03 +0300)]
Change share_vif_oper_freq() to handle multiple freqs
There are devices that can operate several channels concurrently.
Change shared_vif_oper_freq() to get_shared_radio_freqs() that can
return an array of frequencies currently used by all the virtual
interfaces that share the same radio.
In addition, move it to wpa_supplicant.c, so it can be used by other
modules.
Signed-hostap: Ilan Peer <ilan.peer@intel.com>
Signed-hostap: David Spinadel <david.spinadel@intel.com>
Ilan Peer [Sun, 21 Jul 2013 16:49:47 +0000 (19:49 +0300)]
nl80211: Report the number of concurrent support channels
Previously, drivers only reported if they support multiple concurrent
channels, but did not report the maximum number of supported channels.
Add this reporting to the driver capabilities and add the implementation
to driver_nl80211.
Signed-hostap: Ilan Peer <ilan.peer@intel.com>
Signed-hostap: David Spinadel <david.spinadel@intel.com>
To better support the IBSS/RSN mechanism, wpa_supplicant has to be able
to detect a possible peer reboot and in this case it should start a new
EAPOL handshake.
To perform such reboot detection wpa_supplicant has to perform an Open
Authentication by sending an Authentication frame and then replying to
it. IF an Authentication frame is received when the key have already
been exchanged, wpa_supplicant understands that the peer has rebooted
and can reset its state machine.
Whenever a new peer is added to the IBSS wpa_supplicant will start the
Open Authentication and only after having accomplished it will start the
key exchange. If the driver does not support Authentication frame
exchange initiated from user space, this step is skipped to maintain
previous behavior (just go through EAPOL-Key frame processing).
The Open Authentication was partly supported by the Linux kernel but now
wpa_supplicant can register for Authentication frames, handle it in
userspace and so avoid any possible race condition.
Signed-hostap: Nicolas Cavallari <cavallar@lri.fr>
Signed-hostap: Antonio Quartulli <antonio@open-mesh.com>
Commit 1aef400bf03159d60109ce886943fbd26d88a6c3 implemented IBSS RSN
disconnect() call using sta_deauth() in a way that resulted in NULL
pointer dereference in driver_nl80211.c if SME was in user space. Fix
this by passing the own MAC address in the sta_deauth call.
TDLS: Tear down TDLS using wpas_drv_tlds_oper() if not external
When the device indicates to take care of TDLS operations the TDLS
setup is done calling wpas_drv_tdls_oper(). This patch does a similar
thing for the teardown. This fixes failure of teardown:
"TDLS: Could not find peer <mac> for link Teardown"
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Cancel delayed scheduled scan when wpa_supplicant cleans up
Because a delayed scheduled scan will access the members of struct
wpa_supplicant which is freed and this can result in a crash,
wpa_supplicant needs to cancel delayed scheduled scan during cleanups.
The static WEP keys have to be configured for the new VLAN
interface that is created for a 4addr WDS peer, not doing so
breaks WEP functionality in nl80211/4addr based WDS links.
Michal Kazior [Sat, 20 Jul 2013 14:28:42 +0000 (17:28 +0300)]
wpa_supplicant: Add support for VHT BSS membership selector
This allows wpa_supplicant to associate to an AP that has VHT BSS
membership selector set to indicate VHT support is required for the BSS.
Without the patch it was impossible to connect to, e.g., hostapd-based
AP that has require_vht=1. wpa_supplicant was complaining with:
hardware does not support required rate 63.0 Mbps
Signed-hostap: Michal Kazior <michal.kazior@tieto.com>
In a AP/STA concurrent setup, if the STA interface is continually
scanning, trying to connect to a network, the AP interface
is basically broken since beaconing would be erratic.
This option can be used in a WDS setup where one AP acts as a
Client/AP-Repeater. The Repeater AP interface has to start beaconing
only after the Client interface has established a WDS link with the
"Root AP".
nl80211: Ignore disconnect event in case of locally generated request
Previously, there could be two disconnection events in core
wpa_supplicant when going through a case of wpa_supplicant-requested
disconnection with a driver that implements SME internally. This could
result in undesired behavior when a disconnection is followed by a new
connection attempt before the extra event has been received (e.g.,
during fast reassoc or WPS provisioning). Avoid such issues by ignoring
locally generated disconnect events after requesting cfg80211 to
disconnect.
This makes the previously used ignore_next_local_disconnect more
consistent by setting the variable within
wpa_driver_nl80211_disconnect() so that both callers get the same
behavior.
P2P: Report group formation failure on error to start GO mode
There is no need to wait for the 15 second group formation timeout
before indicating P2P group formation failure if GO mode cannot be
started successfully for some reason.
It looks like some of the global control interface cases ended up
blocking in sendmsg() when trying to send an event. Since this can block
all wpa_supplicant processing for multiple seconds, this is very
undesirable. Avoid this by requesting sendmsg() to return an error
rather than waiting for the message to be sent.
Commit 3962b65858ae5d5226fa504d2d1b4f63e7686a95 removed the Xcode
project files, but missed the defines in build_config.h. Remove these
since there are no users for them in the current snapshot.
Interworking: Relax 3GPP info PLMN matching for MNC
3GPP TS 24.232 Annex A.3 allows network operator to advertise only two
digits of MNC even if MNC has three digits. Allow such matches in
network selection. In addition, allow three digit matches of MNC even if
MNC length was assumed to be two to avoid missing networks if MNC length
cannot be determined reliably. Remove the '-' separator from simulated
SIM/USIM cases to allow the new matching rules to work.
Fix the PLMN List information element parsing loop to use the length of
the PLMN List instead of the length of the full 3GPP Cellular Info to
avoid unexpected matches should a new element ever be added by 3GPP.
Finally, add more debug prints from PLMN matching to make the logs
easier to understand.
WPS: Disconnect when removing existing WPS network block
wpas_clear_wps() was just clearing the current wpa_s->current_ssid
pointer when removing a WPS network block which with the device was
associated. This could leave the association up even though the network
block had already been removed. Prevent this by explicitly disconnecting
from the network instead of such clearing current_ssid.
The new server_id parameter in hostapd.conf can now be used to specify
which identity is delivered to the EAP peer with EAP methods that
support authenticated server identity.
Jouni Malinen [Sun, 30 Jun 2013 21:13:11 +0000 (00:13 +0300)]
tests: Use wpa_supplicant global control interface for P2P
This allows more consistent interface to be used regardless of which
P2P driver design is used (especially for P2P management operations
over netdev vs. dedicated P2P_DEVICE).
Jouni Malinen [Sun, 30 Jun 2013 20:32:13 +0000 (23:32 +0300)]
P2P: Do not add ctrl interface for P2P_DEVICE (p2p-dev-*)
Adding a new wpa_supplicant control interface for the dedicated
P2P_DEVICE would be quite confusing for programs that manage P2P
operations. Remove this control interface and require the global control
interface to be used since it will provide consistent interface for both
the new dedicated P2P_DEVICE (non-netdev) and old style P2P management
through a netdev.
Jouni Malinen [Sun, 30 Jun 2013 17:02:25 +0000 (20:02 +0300)]
P2P: Fix TDLS and l2_packet init without P2P Device interface
Commit c68f6200a72f2b33304f583859b09e57757b9d2a made these calls
conditional on !p2p_mgmt, but forced p2p_mgmt=1 for cases where the
driver does not use the dedicated P2P Device. Fix this by making the
!p2p_mgmt condition apply only if the driver does indicate use of a
dedicated P2P Device.
Jouni Malinen [Sun, 30 Jun 2013 09:55:52 +0000 (12:55 +0300)]
Fix build with older OpenSSL versions
Check that SSL_clear_options and SSL_CTX_clear_options are defined
before using them to avoid compilation failures with older OpenSSL
versions that did not include these macros.
Arend van Spriel [Sun, 30 Jun 2013 07:19:31 +0000 (10:19 +0300)]
P2P: Derive group interface name bit more sensibly
The interface name for the P2P group interface is derived from the
P2P management interface. When the P2P management interface is a
P2P Device interface, i.e., p2p-dev-wlanX, the name for the group
interface is abbreviated to p2p-X (X being group index). When the
P2P management interface starts with p2p-dev- use its postfix
instead. So P2P management interface p2p-dev-wlan3 results in group
interface name p2p-wlan3-0.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Sun, 30 Jun 2013 07:17:47 +0000 (10:17 +0300)]
P2P: Create P2P Device interface if supported
If the capability flag of the driver indicates a dedicated P2P Device is
supported, a P2P Device interface is created.
Create the P2P Device in main interface creation loop when the added
interface flags support and P2P supplicant is not yet initialized
avoiding recursion of add_interface.
Do not register l2_packet for P2P Device interface (both for EAPOL and
for TDLS).
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Jouni Malinen [Sun, 30 Jun 2013 06:58:50 +0000 (09:58 +0300)]
nl80211: Do not indicate P2P_DEVICE support by default
Since the P2P_DEVICE support indication was added to kernel before
everything was working properly, there may be kernel versions in use
with the new mechanism breaking P2P functionality (especially with
mac80211_hwsim). For now, disable P2P_DEVICE support by default and
allow it to be enabled with driver_param=p2p_device=1 in the
configuration file. This default behavior may be changed in the future
once the kernel issues has been resolved in stable releases.
Arend van Spriel [Sun, 30 Jun 2013 07:48:44 +0000 (10:48 +0300)]
P2P: Ignore p2p_no_group_iface when driver advertizes P2P_DEVICE support
Setting p2p_no_group_iface means 'use P2P management interface as P2P
connection interface' because it attempts to change the interface type.
The P2P_DEVICE is a dedicated interface and can not be changed. As such
ignore the configuration option.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Sun, 30 Jun 2013 07:34:48 +0000 (10:34 +0300)]
nl80211: Fix determining phy name for P2P Device
The phy name was determined using /sys/class/net/<ifname> but the P2P
Device is not listed there since it does not have an associated net
device. This patch changes name determination to obtain the name from
the wiphy information provide by nl80211.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Sun, 30 Jun 2013 07:29:00 +0000 (10:29 +0300)]
nl80211: Fix nl80211_get_wiphy_index() for P2P Device
For P2P Device the netlink message should have wdev identifier
instead of the interface index. This fixes a failure which occurred
executing the P2P_GROUP_ADD command.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Jouni Malinen [Sat, 29 Jun 2013 21:54:24 +0000 (00:54 +0300)]
Add support for OCSP stapling to validate server certificate
When using OpenSSL with TLS-based EAP methods, wpa_supplicant can now be
configured to use OCSP stapling (TLS certificate status request) with
ocsp=1 network block parameter. ocsp=2 can be used to require valid OCSP
response before connection is allowed to continue.
hostapd as EAP server can be configured to return cached OCSP response
using the new ocsp_stapling_response parameter and an external mechanism
for updating the response data (e.g., "openssl ocsp ..." command).
This allows wpa_supplicant to verify that the server certificate has not
been revoked as part of the EAP-TLS/PEAP/TTLS/FAST handshake before
actual data connection has been established (i.e., when a CRL could not
be fetched even if a distribution point were specified).
Arend van Spriel [Tue, 25 Jun 2013 10:56:28 +0000 (13:56 +0300)]
P2P: Remove a call to wpas_p2p_deinit_global()
In wpa_supplicant_deinit(), the function wpas_p2p_deinit_global()
was called. Remove it as it will be called from wpas_deinit_iface()
upon removal of the P2P management interface.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Tue, 25 Jun 2013 10:52:58 +0000 (13:52 +0300)]
nl80211: Fix P2P group interface creating using P2P Device
When P2P Device is used as P2P management interface the creation of the
P2P group interface fails because MAC address retrieval fails for the
P2P Device interface.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Tue, 25 Jun 2013 10:51:59 +0000 (13:51 +0300)]
nl80211: Use wdev_id in nl80211_create_iface_once()
For P2P an interface may be created for the P2P client/group. The
create request is done on the P2P management interface, which may
be a P2P Device interface. In that case it needs to use the wdev_id.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Tue, 25 Jun 2013 10:51:14 +0000 (13:51 +0300)]
nl80211: Use wdev id to obtain P2P Device scan results
In order to get a P2P-DEVICE-FOUND event the supplicant needs to
see a peer device during SEARCH and LISTEN phase. The SEARCH
phase does a scan so obtaining the scan results for the P2P Device
interface needs to be supported, i.e., use the wdev_id.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Arend van Spriel [Tue, 25 Jun 2013 10:50:21 +0000 (13:50 +0300)]
nl80211: Add .get_mac_addr() callback for P2P Device
For P2P Device the MAC address is determined upon .init2(). The
wpa_s for this interface retrieves this address in
wpa_supplicant_update_mac_addr() using the .get_mac_addr() callback.
Signed-hostap: Arend van Spriel <arend@broadcom.com>
Johannes Berg [Tue, 25 Jun 2013 10:38:03 +0000 (13:38 +0300)]
nl80211: Allow Android P2P functionality
To support Android the kernel may have a "p2p0" netdev for a P2P Device
even though this isn't very useful, but Android requires a netdev. To
support this in the supplicant, if the interface mode is P2P_DEVICE,
re-set it to the same instead of STATION mode.
Note that this is only possible with a kernel that creates a
netdev for the P2P Device wdev.
Signed-hostap: Johannes Berg <johannes.berg@intel.com>
David Spinadel [Tue, 25 Jun 2013 10:35:05 +0000 (13:35 +0300)]
nl80211: Add a handler to create_interface
Add an option to pass a handler to nl80211_create_iface() and
nl80211_create_interface_once() that will be called after receiving the
message from the kernel. This handler will add the option to process the
message in different ways for different interfaces.
Signed-off-by: David Spinadel <david.spinadel@intel.com>
David Spinadel [Tue, 25 Jun 2013 10:33:45 +0000 (13:33 +0300)]
nl80211: Hold wdev identification for P2P Device
Add wdev_id to i802_bss. wdev_id_set indicates whether this id is
available. Use wdev_id if assigned, instead of ifindex. Use wdev_id for
events that come from the kernel to identify the relevant interface.
This commit does not assign wdev_id value for the BSS yet, i.e., this is
only preparation for the value to be used in a future commit.
Signed-off-by: David Spinadel <david.spinadel@intel.com>
Nirav Shah [Tue, 25 Jun 2013 10:29:48 +0000 (13:29 +0300)]
nl80211: Identify if nl80211 is capable of P2P Device abstraction
Check the supported interfaces attribute to identify support for
a dedicated P2P Device interface type. If set, update the driver
capabilities flag to notify the supplicant.
Signed-off-by: David Spinadel <david.spinadel@intel.com>
Michael Braun [Tue, 25 Jun 2013 09:03:02 +0000 (12:03 +0300)]
VLAN: Avoid access to non-existing interfaces
Currently, hostapd_get_vlan_id_ifname() is used to determine if a given
vlan is valid *and* to actually determine the interface. This leads to
wpa_set_keys() sometimes setting the key on the wildcard interface name,
which does not make sense.
This patch therefore adds hostapd_vlan_id_valid() and makes
hostapd_get_vlan_id_ifname() not return a wildcard interface.
Signed-hostap: Michael Braun <michael-dev@fami-braun.de>
Michael Braun [Tue, 25 Jun 2013 08:55:30 +0000 (11:55 +0300)]
bridge: Track inter-BSS usage
Currently, struct hostapd_vlan is a per-BSS data structure which
also contains informations about whether to remove the bridge
or clear wlan / tagged-vlan interface from the bridge.
In a multi-interface multi-BSS setup, this can lead to the following
race condition:
1. wlan0 creates VLAN A, sets DVLAN_CLEAN_BR and DVLAN_CLEAN_VLAN_PORT
2. wlan1 creates VLAN A, does not set DVLAN_CLEAN_BR and
DVLAN_CLEAN_VLAN_PORT as already there
3. wlan0 removes VLAN A, removes tagged-interface from the bridge
but not the bridge.
Now wlan1 VLAN A is unusable due to the missing uplink.
4. wlan1 removes VLAN A, does not cleanup
Solution:
This requires an inter-BSS inter-interface data structure to track the
bridge / bridge port usage within hostapd. This data structure could
also be used to track any other device-has-been-created-by-hostapd
information or when regarding interface freeing.
Signed-hostap: Michael Braun <michael-dev@fami-braun.de>
Michael Braun [Tue, 25 Jun 2013 08:10:00 +0000 (11:10 +0300)]
bridge: Use safe default bridge interface
Currently by default, all BSS share the bridge brvlan%d.
While this is sane when no tagged-interface is given, this
is insane when different tagged interfaces are given, as
it would result in bridging those tagged interfaces.
This patch therefore uses br%s%d with %s=tagged_interface
and %d=VLAN ID as bridge name when a tagged-interface is given.
Signed-hostap: Michael Braun <michael-dev@fami-braun.de>
Johannes Berg [Sat, 22 Jun 2013 10:17:14 +0000 (13:17 +0300)]
wlantest: Always write radiotap pcap-ng
When writing pcap-ng files, always include a radiotap
header even if there wasn't one in the input. For now
just drop prism headers, I don't have any to test.
Signed-hostap: Johannes Berg <johannes.berg@intel.com>
Chengyi Zhao [Sat, 22 Jun 2013 09:09:09 +0000 (12:09 +0300)]
D-Bus: Emit signal when a station is authorized or deauthorized
Add "StaAuthorized" and "StaDeauthorized" D-Bus interface in AP mode.
After enabling the AP mode of wpa_supplicant, the other process need to
get the MAC address and authorization status of every station, so
wpa_supplicant emits signal when the station is authorized or
deauthorized.
Alexander Bondar [Sat, 22 Jun 2013 08:28:33 +0000 (11:28 +0300)]
Remove 802.11b rates only in case of P2P group operation
11b rates removal have had impact on SoftAP functionality in
wpa_supplicant. This patch verifies that only in case of P2P group
operation 11b rates will be eliminated. Refer also to commit 4c2c30289305b67c09f464ba29352b008b5ec433.
Signed-hostap: Alexander Bondar <alexander.bondar@intel.com>
Signed-hostap: Ilan Peer <ilan.peer@intel.com>
Sunil Dutt [Thu, 20 Jun 2013 12:10:15 +0000 (15:10 +0300)]
TDLS: Validate ext_supp_rates in copy_supp_rates
The ext_supp_rates passed to merge_byte_arrays would be invalid if not
advertized by the TDLS peer. Thus, validate the argument to avoid
crashes in such cases.
Jouni Malinen [Wed, 19 Jun 2013 16:16:23 +0000 (19:16 +0300)]
P2P: Do not reply to 802.11b-only Probe Request frames as GO
If AP mode SME/MLME within wpa_supplicant is used for processing Probe
Request frames in GO mode, drop Probe Request frames that include only
802.11b rates per P2P spec section 2.4.1.
Kyeyoon Park [Tue, 18 Jun 2013 14:40:51 +0000 (17:40 +0300)]
Fix session timeout after ANQP dummy STA entry with SME-in-driver
Upon association, disable the timer that removes the dummy STA. This
timer caused the STA that associates within 5 seconds of doing an ANQP
query to disassociate, thinking it's a dummy STA. Similar call was
already there for the SME/MLME-in-hostapd case in handle_auth(), but the
SME-in-driver case was not previously addressed.
Jouni Malinen [Fri, 7 Jun 2013 17:13:25 +0000 (20:13 +0300)]
Remove CONFIG_NO_WPA2 build parameter
There is not much use for enabling WPA without WPA2 nowadays since most
networks have been upgraded to WPA2. Furthermore, the code size savings
from disabling just WPA2 are pretty small, so there is not much
justification for maintaining this build option. Remove it to get rid of
undesired complexity.