Jouni Malinen [Mon, 27 Jun 2011 16:02:24 +0000 (19:02 +0300)]
Clear WPA and EAPOL state machine config pointer on network removal
Make sure that the WPA and EAPOL state machines do not hold a pointer
to a network configuration that is about to be freed. This can fix
potential issues with references to freed memory.
Jayant Sane [Sat, 25 Jun 2011 09:08:43 +0000 (12:08 +0300)]
P2P: Update listen and operating channel from P2P D-Bus
Some P2PDevice properties were not updated in p2p->cfg structure:
reg_class, channel, op_reg_class, and op_channel. Hence, update p2p->cfg
parameters through p2p core calls in wpas_p2p_update_config().
Jayant Sane [Fri, 24 Jun 2011 08:20:19 +0000 (11:20 +0300)]
P2P: More complete persistent group management over D-Bus
Extend commit c2762e410fa319f75a174aeb12343beddf99fce4 to allow
applications to manage (add/remove) persistent groups and accepted
network object paths while invoking a persistent group.
Jayant Sane [Thu, 23 Jun 2011 18:29:10 +0000 (21:29 +0300)]
P2P: Show P2P peer signal level in D-Bus P2P device properties
Move level parameter from p2p_device to p2p_device_info in order to
expose this information and modify D-Bus P2P handler to return this new
parameter through the P2P device properties.
Jayant Sane [Thu, 23 Jun 2011 18:25:13 +0000 (21:25 +0300)]
P2P: Update D-Bus network object semantics during group formation
Do not emit network objects during P2P group formation since such
network objects can confuse certain apps. Instead, a persistent group
object is created to allow apps to keep track of persistent groups.
Persistent group objects only represent the info needed to recreate the
group.
Also fixes a minor bug in the handling of persistent group objects
during WPS operations.
Helmut Schaa [Thu, 23 Jun 2011 17:18:21 +0000 (20:18 +0300)]
hostapd: Don't mask out non-symmetric STA HT caps
Previously hostapd just masked the STAs HT caps with its own. However,
some HT caps are not symmetric and as such need to be handled
different.
hostapd shouldn't overwrite the STAs SMPS mode as otherwise the driver
cannot know it has to use RTS/CTS to wake the receiver from dynamic
SMPS for MCS rates > 7.
hostapd shouldn't mask the RX and TX STBC caps with it's own. They are
already handled in a special case below.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Jouni Malinen [Thu, 23 Jun 2011 16:53:45 +0000 (19:53 +0300)]
Fix CONFIG_NO_WPA_PASSPHRASE=y build
Need to define CONFIG_NO_PBKDF2 even though the crypto cleanup moved
the function to a separate file since there is conditional code in
wpa_supplicant/config.c. In addition, wpa_passphrase should not be
built at all if passphrase functionality is removed.
Jouni Malinen [Thu, 23 Jun 2011 13:39:26 +0000 (16:39 +0300)]
Add a copyright and license statement for a radiotap header file
This file is a part of the radiotap parser that Andy Green agreed to
relicense under the BSD license (per email, 11 Aug 2007 07:42:05
+0100). The copyright/license statement was updated in radiotap.c,
but this radiotap_iter.h file was forgotten at that point.
Zhu Yi [Thu, 23 Jun 2011 12:47:21 +0000 (15:47 +0300)]
bsd: Fix set_key() sequence number endian issue
In set_key handler, the seq[8] is in little endian order defined by
WPA. BSD kernel uses a u_int64_t value ik_keyrsc to represent it
internally. The kernel expects the native endian order for the value.
Thus, we need to detect the endian order and swap bytes when
necessary.
Jouni Malinen [Thu, 23 Jun 2011 12:39:00 +0000 (15:39 +0300)]
Fix AP selection to check privacy mismatch and IBSS with WPA/RSN IE
These checks were previously skipped if the scan result included WPA
or RSN IE. However, that can result in selecting a network that does
not match local configuration in some cases.
Jouni Malinen [Tue, 21 Jun 2011 17:55:46 +0000 (20:55 +0300)]
Add EVENT_RX_ACTION handler for hostapd
This fixes an issue with SA Query Response frames not being processed
anymore after wpa_supplicant started registering a handler for those.
This handler registration is in generic driver_nl80211.c code, so
hostapd uses it, too.
Jouni Malinen [Tue, 21 Jun 2011 08:47:03 +0000 (11:47 +0300)]
P2P: Start GO without extra scan step
There are some corner cases, where the wpa_supplicant_req_scan() call
may end up scheduling a scan even if we are about to start a GO. Avoid
this by explicitly marking the GO network to be selected for the next
connection.
Jouni Malinen [Mon, 20 Jun 2011 07:17:33 +0000 (10:17 +0300)]
nl80211: Add support for driver-based PMKSA cache
Implement PMKSA cache operations add, remove, and flush using nl80211
commands NL80211_CMD_{SET,DEL,FLUSH}_PMKSA to support PMKSA caching
with drivers that select the AP and generate the RSN IE internally.
Johannes Berg [Sun, 12 Jun 2011 21:47:49 +0000 (14:47 -0700)]
DBus: Refactor array adding, add binary arrays
Some new code we're working on will require the dbus type "aay" (an
array of arrays of bytes). To add this, refactor the array code to
reduce code duplication by given a type string to the array starting
code, and also add code to create and parse such arrays from or into an
array of struct wpabuf respectively.
Since there's no unique DBus type for this, add a "fake"
WPAS_DBUS_TYPE_BINARRAY type that is separate from the regular DBus
types for parsing.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Jayant Sane [Sun, 12 Jun 2011 21:20:39 +0000 (14:20 -0700)]
P2P: Retry provision discovery requests in IDLE state
Since the peer may not be in Listen state when the provision discovery
request is sent, try to send the request again number of times when in
IDLE state. This was already done when p2p_find is in progress, but this
commit adds retries to the case where no other P2P operations are in
progress.
Signed-off-by: Jayant Sane <jayant.sane@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Jouni Malinen [Tue, 31 May 2011 17:07:11 +0000 (20:07 +0300)]
random: Add support for maintaining internal entropy store over restarts
This can be used to avoid rejection of first two 4-way handshakes every
time hostapd (or wpa_supplicant in AP/IBSS mode) is restarted. A new
command line parameter, -e, can now be used to specify an entropy file
that will be used to maintain the needed state.
Jouni Malinen [Fri, 27 May 2011 15:54:36 +0000 (18:54 +0300)]
nl80211: Enable more AP callbacks for non-hostapd AP mode
Some of these are required for proper functionality (like
get_seqnum); others may not be needed yet, but including them
allows some extra ifdef/endif blocks to be removed.
Jouni Malinen [Thu, 19 May 2011 14:52:46 +0000 (17:52 +0300)]
Fix regression in RSN pre-authentication candidate list generation
Processing of the scan results for RSN pre-authentication candidates
was moved to happen before the network was selected. This resulted in
all candidates being dropped due to no SSID having been configured.
Fix this by moving the processing to happen after the network has
been selected. Since the raw scan results are not available at that
point, use the BSS table instead of scan results to fetch the
information.
Vinay Adella [Thu, 19 May 2011 09:55:47 +0000 (12:55 +0300)]
WPS UPnP: Fix UPnP initialization for non-bridge case with some drivers
If the driver wrapper is setting up the interface up only at commit(),
UPnP initialization fails. Fix that by moving UPnP setup to happen after
the driver commit() call.
Jouni Malinen [Tue, 17 May 2011 16:53:02 +0000 (19:53 +0300)]
WPS: Add a workaround for Windows 7 capability discovery for PBC
Windows 7 uses incorrect way of figuring out AP's WPS capabilities by
acting as a Registrar and using M1 from the AP. The config methods
attribute in that message is supposed to indicate only the configuration
method supported by the AP in Enrollee role, i.e., to add an external
Registrar. For that case, PBC shall not be used and as such, the
PushButton config method is removed from M1 by default. If pbc_in_m1=1
is included in the configuration file, the PushButton config method is
left in M1 (if included in config_methods parameter) to allow Windows 7
to use PBC instead of PIN (e.g., from a label in the AP).
Jouni Malinen [Mon, 16 May 2011 18:07:05 +0000 (21:07 +0300)]
P2P: Include operating class 124 (channels 149,153,157,161)
When using the p2p_oper_reg_class configuration option instead of
p2p_group_add freq parameter, the operating class 124 was not
processed correctly. Include that class in the list of 5 GHz classes
for the p2p_oper_reg_class to resolve this.
Jouni Malinen [Mon, 16 May 2011 16:18:42 +0000 (19:18 +0300)]
nl80211: Filter out duplicated BSS table entries from scan results
cfg80211 maintains separate BSS table entries for APs if the same
BSSID,SSID pair is seen on multiple channels. wpa_supplicant does
not use frequency as a separate key in the BSS table, so filter out
duplicated entries. Prefer associated BSS entry in such a case in
order to get the correct frequency into the BSS table.
Add workaround for race condition with AssocResp TX status
It may take some time for the TX status to be delivered for a
(Re)Association Response frame and if any Data frames are received
during that time, they may end up getting dropped as Class 3 frames in
not-associated state. This results in a Disassociation frame being sent
to the station and it assuming that the association has been lost.
Work around the issue by remembering that the (Re)Association Request
has already been accepted and skip the Deauth/Disassoc sending because
of the possible Class 3 frames before the TX status callback is
received.
Guy Eilam [Fri, 15 Apr 2011 15:13:04 +0000 (18:13 +0300)]
hostapd: Fix interfaces.iface initialization
Set all the interfaces.iface pointers to NULL after the allocation of
that memory block for cases those pointers are accessed during each of
the interfaces initialization process (hostapd_interface_init()). One
example for such case is during WPS initialization when the code tries
to fetch the uuid from each of the interfaces.
Add sanity checks for fseek and ftell return values
In theory, these calls could fail, but it is not really likely to
happen in practice in the use case here. Anyway, check that they do
not return an error before accepting the length of the file.
Avoid theoretical NULL pointer dereference during TLS reassemble
This function does not get called with in_data == NULL in practice, but
it seems to be at least partly prepared for that case, so better make it
consistent by handling the NULL value throughout the function.
Add sanity checks to EVENT_RX_PROBE_REQ event data
Both the SA and IEs from the received Probe Request frames must be
included and the Probe Request RX callback functions may assume that
these are not NULL.
Avoid theoretical NULL pointer dereference from debug code
The change to use wpa_dbg() in wpa_sm_parse_own_wpa_ie() could result
in a NULL pointer dereference if the function were called when WPA
state machine has not been initialized. While this cannot really
happen in practice, it is better to be prepared for that since that
was the case before the wpa_dbg() change.
Distinguish empty concatenated vendor-specific IE from missing one
When comparing BSS table entries, handle empty concatenated
vendor-specific IE differently from completely missing IE. This
does not change anything for the only currently compared IE that
can be fragmented (WPS IE), but it is better to have the generic
code here ready for any possible new IE that could be used in
with zero-length data.
Use type cast to get rid of implicit sign extension
The size_t value here can be 64-bit and result in implicit sign
extension. In this particular case, that gets masked out by
host_to_be32(), so there is no practical difference, but it is better
to get rid of the 64-bit variable explicitly.
TNC: Fix TNC_{TNCC,TNCS}_ReportMessageTypes copy type
The supportedTypes parameter is a list of TNC_MessageType values
and the buffer to be copied should use size of TNC_MessageType, not
TNC_MessageTypeList. In practice, these are of same length on most
platforms, so this is not a critical issue, but anyway, the correct
type should be used.
This avoids reading past the end of the IE buffer should the FTIE
be too short. In addition, one debug hexdump was using uninitialized
pointer to the FTIE buffer, so fixed it to use the pointer from the
parse data.
P2P: Clear go_dev_addr for group started events if it is not known
Better make sure the address value included in the event message gets
cleared in case the GO Device Address is not known for some reason.
Previously, this could potentially have been some random data from
stack should we ever hit the case where wpa_s->current_ssid is not
set (which should not really happen in normal use cases).
Yi Zhu [Tue, 12 Apr 2011 20:09:14 +0000 (23:09 +0300)]
bsd: Fix buffer size for routing socket with IPv6 disabled
AF_INET6 is not always enabled by default, so use AF_INET instead. In
addition, use the old fixed length, 2048, as a failover value if the
sysctl fails for any reason.
It is now possible to configure the the time in seconds that
wpa_supplicant waits before requesting a new scan after failing to find
a suitable network in scan results.
Commit 03d3f28a698ec2c269fed35b88be30373595eee2 broke initialization of
EAPOL authenticator state machines since an error value from
wpa_auth_sta_key_mgmt() (-1) was not handled properly and the fixed
wpa_key_mgmt_wpa_psk() identified the value as a PSK-based AKM because
of all bits being set to 1. The special error value needs to handled
separately to avoid skipping EAPOL state machine initialization.
If WPA/WPA2 was enabled in the configuration, the non-WPA exception
could allow an incorrect AP to be selected from scan results. Do
not use the exception if WPA/WPA2 is enabled in configuration.
Fix wpa_key_mgmt_*() helper functions to handle multiple bits
These can be used in some cases with a bitfield having multiple
AKMs enabled (e.g., WPA-PSK and WPA-PSK-SHA256). Address those
cases by checking whether any of the matching AKM are included.
nl80211: Add support for auth_alg auto-selection with connect command
When the SME is in the driver or cfg80211, the automatic selection
of auth_alg is done by leaving out the NL80211_ATTR_AUTH_TYPE
attribute from the NL80211_CMD_CONNECT command.
wpa_s AP mode: Fix regression in P2P AP mode setup
Commit c76e5d7f9b7a4ed8c201f280a04e579d52765572 moved the
ap_configured_cb() call into a completion callback from AP mode code.
However, this callback can be called before hostap_setup_interface()
returns. In that case, the ap_configured_cb() ended up getting
called before the wpa_supplicant connection information was updated.
Fix this by reordering code to set up everything before calling
hostapd_setup_interface().
Felix Fietkau [Sat, 2 Apr 2011 19:03:05 +0000 (22:03 +0300)]
nl80211: Send EAPOL frames as QoS data frames for QoS aware clients
This should fix EAPOL reauthentication and rekeying timeout issues
with Intel clients when using WMM (e.g., with IEEE 802.11n). These
stations do not seem to be able to handle EAPOL data frames as
non-QoS Data frames after the initial setup.
This adds STA flags to hapd_send_eapol() driver op to allow
driver_nl80211.c to mark the EAPOL frames as QoS Data frame
when injecting it through the monitor interface.
Paul Stewart [Thu, 31 Mar 2011 15:06:09 +0000 (18:06 +0300)]
bgscan_simple: Reschedule first fast scan
If the scan interval switches to the short interval soon after a
scan, bgscan_simple should not immediately scan again. However,
it should also make sure that the next scan occurs no later than
the new, short scan interval.
Paul Stewart [Thu, 31 Mar 2011 15:04:48 +0000 (18:04 +0300)]
bgscan_simple: Time out short scan interval
We gain diminishing returns by the short scan interval. The short
scan interval is used to hunt for a better AP if the RSSI of the
current AP drops. However, if we never roam, and the AP continues
to have low RSSI, at some point we should give up and return to
the slow background scan rate, otherwise we waste a lot of power.
Paul Stewart [Thu, 31 Mar 2011 15:02:37 +0000 (18:02 +0300)]
bgscan-simple: Poll for signal in init
It is not guaranteed that we will get a CQM signal shortly after setting
up monitoring. In order to establish the correct initial background
scanning rate, poll directly for the signal strength.
Paul Stewart [Thu, 31 Mar 2011 14:56:48 +0000 (17:56 +0300)]
driver: Function to directly poll signal quality
This provides a means for the supplicant to directly request signal
quality metrics from the driver. This is useful, for example for
background scan algorithms that might ask desire this information
out-of-band with CQM events.
Paul Stewart [Thu, 31 Mar 2011 14:38:08 +0000 (17:38 +0300)]
dbus: Add "excessive" to debug_strings
The "debug_strings" private array in dbus_new_handles.c has fallen out
of sync with wpa_debug.h. Add the new "excessive" level to the head of
the list, realigning everything.
Jouni Malinen [Tue, 29 Mar 2011 14:39:12 +0000 (17:39 +0300)]
Work around SNonce updates on EAPOL-Key 1/4 retransmission
Some deployed supplicants update their SNonce for every receive
EAPOL-Key message 1/4 even when these messages happen during the
same 4-way handshake. Furthermore, some of these supplicants fail
to use the first SNonce that they sent and derive an incorrect PTK
using another SNonce that does not match with what the authenticator
is using from the first received message 2/4. This results in
failed 4-way handshake whenever the EAPOL-Key 1/4 retransmission
timeout is reached. The timeout for the first retry is fixed to
100 ms in the IEEE 802.11 standard and that seems to be short
enough to make it difficult for some stations to get the response
out before retransmission.
Work around this issue by increasing the initial EAPOL-Key 1/4
timeout by 1000 ms (i.e., total timeout of 1100 ms) if the station
acknowledges reception of the EAPOL-Key frame. If the driver does
not indicate TX status for EAPOL frames, use longer initial
timeout (1000 ms) unconditionally.
Jouni Malinen [Mon, 28 Mar 2011 12:27:10 +0000 (15:27 +0300)]
P2P: Store more WPS attributes for peers
Store Manufacturer, Model Name, Model Number, and Serial Number
attributes for P2P peers and expose these through the existing
peer information mechanisms.