If a legacy client with no P2P PS support is trying to connect to
a P2P GO, the driver should know that, and change its PS behavior
accordingly.
Add a parameter to hostapd_sta_add_params() indicating if P2P PS is
supported by the station and pass this parameter to kernel with nl80211
driver when the station is added/set.
Signed-off-by: Ayala Beker <ayala.beker@intel.com>
Fix a possible memory leak in decode_tunnel_passwords() if an invalid
passphrase is received from the RADIUS server.
Signed-off-by: Ayala Beker <ayala.beker@intel.com>
This was previously handled for the case where the non-success
Invitation Response frame was sent out during the Listen phase. However,
in the case the Action frame TX ended up getting scheduled when the
Search phase scan had already started (e.g., due to the driver reporting
Invitation Request RX late enough for the Listen-to-Search transition
having already started), the postponed Action frame TX status processing
did not cover the specific case of non-success Invitation Response. This
could result in the p2p_find operation getting stopped (stuck in SEARCH
state) unexpectedly.
Fix this by calling p2p_check_after_scan_tx_continuation() from
Invitation Response TX callback handler if the invitation was rejected.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The Secure bit in the Key Information field of EAPOL-Key frames is
supposed to be set to 1 when there is a security association. This was
done for other frames, but not for the EAPOL-Key Request frame where
supplicant is requesting a new PTK to be derived (either due to Michael
MIC failure report Error=1 or for other reasons with Error=0). In
practice, EAPOL-Key Request frame is only sent when there is a PTK in
place, so all such frames should have Secure=1.
Signed-off-by: Jouni Malinen <j@w1.fi>
This struct in the union is empty, but the design of using a zero-length
u8 array here is not fully compatible with C++ and can result in
undesired compiler warnings. This struct is not used anymore, so it can
be removed from the struct ieee80211_mgmt definition to complete the
changes started in commit d447cd596f
('Updates for stricter automatic memcpy bounds checking').
Signed-off-by: Jouni Malinen <j@w1.fi>
This struct in the union is empty, but the design of using a zero-length
u8 array here is not fully compatible with C++ and can result in
undesired compiler warnings. Since there are no non-IE fields in the
Probe Request frames, get the location of the variable length IEs simply
by using the pointer to the frame header and the known header length.
Signed-off-by: Jouni Malinen <j@w1.fi>
This struct in the union is empty, but the design of using a zero-length
u8 array here is not fully compatible with C++ and can result in
undesired compiler warnings. Since there are no non-IE fields in the
Probe Request frames, get the location of the variable length IEs simply
by using the pointer to the frame header and the known header length.
Signed-off-by: Jouni Malinen <j@w1.fi>
This reverts commit 4ca16b5fd7.
Configuration for this will be done using a previously assigned more
generic command. This new command
QCA_NL80211_VENDOR_SUBCMD_SET_TXRX_AGGREGATION has not been used in any
driver version and won't be used, so the assigned command id can be
freed for future use.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
There is no strong need for pulling in linux/if_ether.h here since all
that is needed if ETH_P_IP and we already cover multiple other ETH_P_*
values in utils/common.h.
Signed-off-by: Jouni Malinen <j@w1.fi>
This was only used for providing an option to use linux/if_packet.h
instgead of netpacket/packet.h in src/ap/iapp.c. However,
netpacket/packet.h is nowadays commonly available and hostapd already
depends on it through src/l2_packet/l2_packet_linux.c, so there is no
need to continue to provide this option for the kernel header.
Signed-off-by: Jouni Malinen <j@w1.fi>
This gets rid of need to include linux/if_vlan.h and additional defines
in vlan_ioctl.c to avoid issues with missing definitions in libc
headers.
Signed-off-by: Jouni Malinen <j@w1.fi>
Due to both <netinet/in.h> (in "utils/includes.h") and <linux/in6.h> (in
<linux/if_bridge.h>) being included, the in6_addr is being redefined:
once from the C library headers and once from the Linux kernel headers.
This causes some build failures with for example the musl C library:
In file included from /usr/include/linux/if_bridge.h:18,
from ../src/ap/vlan_init.c:17:
/usr/include/linux/in6.h:32: error: redefinition of 'struct in6_addr'
/usr/include/linux/in6.h:49: error: redefinition of 'struct sockaddr_in6'
/usr/include/linux/in6.h:59: error: redefinition of 'struct ipv6_mreq'
Mixing C library and Linux kernel headers is a bit problematic [1] and
should be avoided if possible [2]. In order to fix this, define just the
macros needed from <linux/if_bridge.h> as done in Busybox for the brctl
applet [3].
[1] https://sourceware.org/bugzilla/show_bug.cgi?id=15850
[2] http://www.openwall.com/lists/musl/2015/10/06/1
[3] https://git.busybox.net/busybox/commit/?id=5fa6d1a632505789409a2ba6cf8e112529f9db18
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
With this, vlan_init.c does not need any special header files anymore
and vlan_ifconfig.c does not need hostapd-specific header files that
might conflict with net/if.h on NetBSD.
Signed-off-by: Jouni Malinen <j@w1.fi>
Move the ioctl-based VLAN implementation to a separate file to avoid
need for conditional blocks within vlan_ioctl.c. This removes the
internal CONFIG_VLAN_NETLINK define, i.e., this is now used only in
build configuration (.config) to select whether to include the
vlan_util.c (netlink) or vlan_ioctl.c (ioctl) implementation of the
functions.
Signed-off-by: Jouni Malinen <j@w1.fi>
caddr_t is legacy BSD and should be avoided [1]. While glibc may still
use __caddr_t as the type, Linux kernel does not (it is "void __user *
ifru_data").
This fixes compile errors with the musl libc:
../src/ap/vlan_init.c: In function 'br_delif':
../src/ap/vlan_init.c:218:18: error: '__caddr_t' undeclared (first use in this function)
ifr.ifr_data = (__caddr_t) args;
[1] http://stackoverflow.com/questions/6381526/what-is-the-significance-of-caddr-t-and-when-is-it-used
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
cfg80211 reports a deauth/disassoc event when internally clearing
connection with the previous BSS. Ignore that event to allow the new
connect command to complete.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This makes it easier for drivers that use the Connect command instead of
separate Auth+Assoc commands to determine when to use reassociation
instead of association. Matching changes are still needed in cfg80211 to
allow this parameter to be used, but it is safe for wpa_supplicant to
start including this attribute now since it will be ignored by older
cfg80211 versions.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
pmksa->pmk or pmksa->pmkid cannot be NULL since they are arrays. Remove
the unnecessary NULL checks and use the provided pmksa pointer directly
to simplify the implementation. (CID 138519)
Signed-off-by: Jouni Malinen <j@w1.fi>
Interface additions/removals are not guaranteed to be for the driver
listening to the kernel events. As such, send the events to
wpa_supplicant_event_global() which can then pick the correct interface
registered with wpa_supplicant to send the event to.
Signed-off-by: Roy Marples <roy@marples.name>
Commit 1889af2e0f ('VLAN: Separate station
grouping and uplink configuration') added an ap_sta_set_vlan() function
that gets called from pmksa_cache_auth.c. This broke CONFIG_IBSS_RSN=y
build if src/ap/sta_info.c did not get included in the build, i.e., if
CONFIG_AP=y was not set.
Fix this by making the ap_sta_set_vlan() call conditional on
CONFIG_NO_VLAN being undefined and define this for CONFIG_IBSS_RSN=y
builds. This is fine for wpa_supplicant since CONFIG_AP=y case was
already defining this. For hostapd, this function call is not needed for
CONFIG_NO_VLAN case either.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This patch add functionality of mesh SAE PMKSA caching. If the local STA
already has peer's PMKSA entry in the cache, skip SAE authentication and
start AMPE with the cached value.
If the peer does not support PMKSA caching or does not have the local
STA's PMKSA entry in the cache, AMPE will fail and the PMKSA cache entry
of the peer will be removed. Then STA retries with ordinary SAE
authentication.
If the peer does not support PMKSA caching and the local STA uses
no_auto_peer=1, the local STA can not retry SAE authentication because
NEW_PEER_CANDIDATE event cannot start SAE authentication when
no_auto_peer=1. So this patch extends MESH_PEER_ADD command to use
duration(sec). Throughout the duration, the local STA can start SAE
authentication triggered by NEW_PEER_CANDIDATE even though
no_auto_peer=1.
This commit requires commit 70c93963ed
('SAE: Fix PMKID calculation for PMKSA cache'). Without that commit,
chosen PMK comparison will fail.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This extends the wpa_supplicant PMKSA_FLUSH control interface command to
allow the PMKSA list from the authenticator side to be flushed for AP
and mesh mode. In addition, this adds a hostapd PMKSA_FLUSH control
interface command to flush the PMKSA entries.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This extends the wpa_supplicant PMKSA control interface command to allow
the PMKSA list from the authenticator side to be listed for AP and mesh
mode. In addition, this adds a hostapd PMKSA control interface command
to show the same list for the AP case.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This group capability bit was previously added unconditionally which
could result in the P2P Client assuming the functionality is available
even though the GO would always reject the request (not reply to it with
an assigned IP address) during the 4-way handshake.
Fix this by advertising the capability only if the GO configuration
allow IP address assignment to be completed.
Signed-off-by: Jouni Malinen <j@w1.fi>
Commit 732b1d20ec ('nl80211: Clean up
ifidx properly if interface in a bridge is removed') added
drv->if_indices_reason array similarly to the previously used
drv->if_indices. However, it had a copy-paste error here on the error
path where a reallocation failure after at least one successful
reallocation would result in the drv->if_indices being overridden
instead of restoring drv->if_indices_reason to the old value. Fix this
by setting the correct variable on the error path. (CID 138514)
Signed-off-by: Jouni Malinen <j@w1.fi>
SMPS mode is applicable only for HT and including an attribute to
configure it when HT is disabled could result in the AP start operation
failing. Fix this by adding the attribute only in cases where HT is
enabled.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Since this function needs to allocate memory, it might fail. Check that
the returned memory pointer is not NULL before trying to parse the
output.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
While this function could return NULL if the parameter issued to it were
NULL, that does not really happen here. Anyway, since this can result in
a warning from a static analyzer that does can see the return NULL
without fully understanding what it means here, check the return value
explicitly against NULL to avoid false warnings.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Since commit cb05808c46 ('nl80211: Generic
Linux master interface support for hostapd'), hostapd is listening for
EAPOL frames on any master which the interface is enslaved under.
This commit allows hostapd to support network hierarchy in which the
interface is enslaved under some master which in turn is enslaved under
a bridge.
Signed-off-by: Dedy Lansky <qca_dlansky@qca.qualcomm.com>
In scenarios where only DFS channels are available (e.g., outdoor,
special country codes), hostapd must be able to handle situations
where all are unavailable.
The two possibilities to get there are
1) while operating on the last available DFS channel a radar is
detected
2) hostapd is started while all channels are unavailable
In both cases, hostapd instead of terminating should better
wait for the NOPs to pass and re-try operation after the CAC.
This patch provides that feature by using the condition
(iface->state == HAPD_IFACE_DFS && !iface->cac_started)
as NOP mode signature to retry operation from within
hostapd_dfs_nop_finished().
Signed-off-by: Zefir Kurtisi <zefir.kurtisi@neratec.com>
Upon receiving FST setup request on old band, the peer is searched on
new band. The assumption is that MB IE from this peer on new band
previously received either in assoc event or in FST setup request.
There are cases in which above assumption doesn't work, peer is not
found and session setup fails. For example:
- STA connects over 11ac. Due to driver limitation, MB IE is not included
in assoc event
- STA connects over 11ad. MB IE included in assoc event.
- FST session established on 11ac band, with AP as initiator. i.e. FST
setup request sent in AP=>STA direction. STA searches for peer's (AP)
MB IE on 11ad band which exists.
- FST switch occur
- FST session established on 11ad band, with STA as initiator. i.e. FST
setup request sent in STA=>AP direction. AP searches for peer's (STA)
MB IE on 11ac band which are absent.
For fixing the above, consider also peers without MB IE as candidates in
the search algorithm.
Signed-off-by: Dedy Lansky <qca_dlansky@qca.qualcomm.com>
In the 60 GHz band, service discovery management frames are sent over
the control PHY and have a smaller maximum frame size (IEEE Std
802.11ad-2012, 21.4.3.2). Fix the code to use sufficiently small
fragment size when operating in the 60 GHz band.
The 60 GHz fragment size (928) is derived from the maximum frame size
for control PHY (1023) and subtracting 48 bytes of header size, and some
spare so we do not reach frames with the absolute maximum size.
Signed-off-by: Lior David <qca_liord@qca.qualcomm.com>
Add a new 'use_driver_iface_addr' configuration parameter to allow use
of the default interface address generated by the driver on interface
creation. This can be useful when specific MAC addresses were allocated
to the device and we want to use them for multi-BSS operation.
Signed-off-by: Eliad Peller <eliad@wizery.com>
There is a race condition in which AP might receive the EAPOL-Start
frame (from the just-associated station) before the TX completion of the
Association Response frame. This in turn will cause the EAPOL-Start
frame to get dropped, and potentially failing the connection.
Solve this by saving EAPOL frames from authenticated-but-not-associated
stations, and handling them during the Association Response frame TX
completion processing.
Signed-off-by: Eliad Peller <eliad@wizery.com>
