This changes 4-way handshake authenticator processing to decrypt the
EAPOL-Key frames using an AEAD cipher (AES-SIV with FILS AKMs) before
processing the Key Data field. This replaces Key MIC validation for the
cases where AEAD cipher is used. This needs to move the EAPOL-Key msg
2/4 RSN element processing to happen only after the PTK has been derived
and validated. That is done for all AKMs to avoid extra complexity with
having to maintain two code paths for this.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Suite B 192-bit addition from IEEE Std 802.11ac-2013 replaced the
previous fixed length Key MIC field with a variable length field. That
change was addressed with an addition of a new struct defined for the
second MIC length. This is not really scalable and with FILS coming up
with a zero-length MIC case for AEAD, a more thorough change to support
variable length MIC is needed.
Remove the Key MIC and Key Data Length fields from the struct
wpa_eapol_key and find their location based on the MIC length
information (which is determined by the AKMP). This change allows the
separate struct wpa_eapol_key_192 to be removed since struct
wpa_eapol_key will now include only the fixed length fields that are
shared with all EAPOL-Key cases in IEEE Std 802.11.
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>
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>
The SAE PMKID is calculated with IEEE Std 802.11-2012 11.3.5.4, but the
PMKID was re-calculated with 11.6.1.3 and saved into PMKSA cache. Fix
this to save the PMKID calculated with 11.3.5.4 into the PMKSA cache.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This FTIE needs to be an exact copy of the one in (Re)Association
Response frame. Copy the stored element rather than building a new copy
that would not have the correct MIC value. This is needed to fix PTK
rekeying after FT protocol run.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
wpa_insert_pmkid() did not support cases where the original RSN IE
included any PMKIDs. That case can happen when PTK rekeying through
4-way handshake is used after FT protocol run. Such a 4-way handshake
used to fail with wpa_supplicant being unable to build the EAPOL-Key msg
2/4.
Fix this by extending wpa_insert_pmkid() to support removal of the old
PMKIDs, if needed.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
In addition to the PTK length increasing, the length of the PMK was
increased (from 256 to 384 bits) for the 00-0f-ac:12 AKM. This part was
missing from the initial implementation and a fixed length (256-bit) PMK
was used for all AKMs.
Fix this by adding more complete support for variable length PMK and use
384 bits from MSK instead of 256 bits when using this AKM. This is not
backwards compatible with the earlier implementations.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This ensures that group key is set as long as the interface exists.
Additionally, ifconfig_up is needed as wpa_group will enter
FATAL_FAILURE if the interface is still down. Also vlan_remove_dynamic()
is moved after wpa_auth_sta_deinit() so vlan_remove_dynamic() can check
it was the last user of the wpa_group.
Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
It was possible for wpa_auth_sm_event(WPA_DEAUTH) to be called from
wpa_sm_step() iteration in the case the EAPOL authenticator state
machine ended up requesting the station to be disconnected. This
resulted in unnecessary recursive call to wpa_sm_step(). Avoid this by
using the already running call to process the state change.
It was possible to hit this sequence in the hwsim test case
ap_wpa2_eap_eke_server_oom.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The new CONFIG_NO_RC4=y build option can be used to remove all internal
hostapd and wpa_supplicant uses of RC4. It should be noted that external
uses (e.g., within a TLS library) do not get disabled when doing this.
This removes capability of supporting WPA/TKIP, dynamic WEP keys with
IEEE 802.1X, WEP shared key authentication, and MSCHAPv2 password
changes.
Signed-off-by: Jouni Malinen <j@w1.fi>
Previously, struct wpa_group was created when the first station enters
the group and the struct wpa_group was not freed when all station left
the group. This causes a problem because wpa_group will enter
FATAL_FAILURE when a wpa_group is running while the AP_VLAN interface
has already been removed.
Fix this by adding a reference counter to struct wpa_group and free a
group if it is unused.
Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
If the AP/Authenticator receives an EAPOL-Key msg 2/4 for an association
that negotiated use of PSK and the EAPOL-Key MIC does not match, it is
likely that the station is trying to use incorrect PSK/passphrase.
Report this with "AP-STA-POSSIBLE-PSK-MISMATCH <STA addr>" control
interface event.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This allows WPA2 mode AP to be re-enabled automatically after external
ifconfig down + up on a netdev used by hostapd.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Previously, it was possible for some corner cases to leave the WPA
authenticator state machine running if PMK could not be derived. Change
this to forcefully disconnect the STA to get more consistent behavior
and faster notification of the error.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This modifies struct wpa_ptk to allow the length of KCK and KEK to be
stored. This is needed to allow longer keys to be used, e.g., with
Suite B 192-bit level.
Signed-off-by: Jouni Malinen <j@w1.fi>
This extends Disconnect-Request processing to check against PMKSA cache
entries if no active session (STA association) match the request.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Previously, only WPA + WPA2 was covered. If FT is enabled in addition to
WPA, MDIE is included in the buffer between RSN and WPA elements. The
previous version ended up leaving only the MDIE after having skipped RSN
element. Fix this to skip MDIE as well to leave only WPA IE regardless
of whether FT is enabled in AP configuration.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This extends the earlier PeerKey station side design to be used on the
AP side as well by passing pointer and already validated length from the
caller rather than parsing the length again from the frame buffer. This
avoids false warnings from static analyzer (CID 62870, CID 62871,
CID 62872).
Signed-off-by: Jouni Malinen <j@w1.fi>
If the 4-way handshake ends up having to retransmit the EAPOL-Key
message 1/4 due to a timeout on waiting for the response, it is possible
for the Supplicant to change SNonce between the first and second
EAPOL-Key message 2/4. This is not really desirable due to extra
complexities it causes on the Authenticator side, but some deployed
stations are doing this.
This message sequence looks like this:
AP->STA: EAPOL-Key 1/4 (replay counter 1, ANonce)
AP->STA: EAPOL-Key 1/4 (replay counter 2, ANonce)
STA->AP: EAPOL-Key 2/4 (replay counter 1, SNonce 1)
AP->STA: EAPOL-Key 3/4 (replay counter 3, ANonce)
STA->AP: EAPOL-Key 2/4 (replay counter 2, SNonce 2)
followed by either:
STA->AP: EAPOL-Key 4/4 (replay counter 3 using PTK from SNonce 1)
or:
AP->STA: EAPOL-Key 3/4 (replay counter 4, ANonce)
STA->AP: EAPOL-Key 4/4 (replay counter 4, using PTK from SNonce 2)
Previously, Authenticator implementation was able to handle the cases
where SNonce 1 and SNonce 2 were identifical (i.e., Supplicant did not
update SNonce which is the wpa_supplicant behavior) and where PTK
derived using SNonce 2 was used in EAPOL-Key 4/4. However, the case of
using PTK from SNonce 1 was rejected ("WPA: received EAPOL-Key 4/4
Pairwise with unexpected replay counter" since EAPOL-Key 3/4 TX and
following second EAPOL-Key 2/4 invalidated the Replay Counter that was
used previously with the first SNonce).
This commit extends the AP/Authenticator workaround to keep both SNonce
values in memory if two EAPOL-Key 2/4 messages are received with
different SNonce values. The following EAPOL-Key 4/4 message is then
accepted whether the MIC has been calculated with the latest SNonce (the
previously existing behavior) or with the earlier SNonce (the new
extension). This makes 4-way handshake more robust with stations that
update SNonce for each transmitted EAPOL-Key 2/4 message in cases where
EAPOL-Key message 1/4 needs to be retransmitted.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This adds support for AKM 00-0F-AC:11 to specify the integrity and
key-wrap algorithms for EAPOL-Key frames using the new design where
descriptor version is set to 0 and algorithms are determined based on
AKM.
Signed-off-by: Jouni Malinen <j@w1.fi>
The new AKM uses a different mechanism of deriving the PMKID based on
KCK instead of PMK. hostapd was already doing this after the KCK had
been derived, but wpa_supplicant functionality needs to be moved from
processing of EAPOL-Key frame 1/4 to 3/4 to have the KCK available.
Signed-off-by: Jouni Malinen <j@w1.fi>
This makes hostapd create PMKSA cache entries from SAE authentication
and allow PMKSA caching to be used with the SAE AKM.
Signed-off-by: Jouni Malinen <j@w1.fi>
It looks like the use of sm->wpa == WPA_VERSION_WPA2 in two locations
within the function was a bit too much for clang static analyzer to
understand. Use a separate variable for storing the allocated memory so
that it can be freed unconditionally. The kde variable can point to
either stack memory or temporary allocation, but that is now const
pointer to make the design clearer.
Signed-off-by: Jouni Malinen <j@w1.fi>
This adds kek_len argument to aes_wrap() and aes_unwrap() functions and
allows AES to be initialized with 192 and 256 bit KEK in addition to
the previously supported 128 bit KEK.
The test vectors in test-aes.c are extended to cover all the test
vectors from RFC 3394.
Signed-off-by: Jouni Malinen <j@w1.fi>
This makes the implementation less likely to provide useful timing
information to potential attackers from comparisons of information
received from a remote device and private material known only by the
authorized devices.
Signed-off-by: Jouni Malinen <j@w1.fi>
Move to PTKINITDONE state and mark PTK valid after successful completion
of FT protocol. This allows the AP/Authenticator to start GTK rekeying
when FT protocol is used. Previously, the station using FT protocol did
not get the new GTK which would break delivery of group addressed
frames.
Signed-off-by: Jouni Malinen <j@w1.fi>
clang scan-build does not seem to like the 'd' suffix on floating
constants and ends up reporting analyzer failures. Since this suffix
does not seem to be needed, get rid of it to clear such warnings.
Signed-off-by: Jouni Malinen <j@w1.fi>
If the PMK-R1 needs to be pulled for the R0KH, the previous
implementation ended up rejecting the over-the-air authentication and
over-the-DS action frame unnecessarily while waiting for the RRB
response. Improve this by postponing the Authentication/Action frame
response until the pull response is received.
Signed-off-by: Jouni Malinen <j@w1.fi>
This allows hostapd to set a different management group cipher than the
previously hardcoded default BIP (AES-128-CMAC). The new configuration
file parameter group_mgmt_cipher can be set to BIP-GMAC-128,
BIP-GMAC-256, or BIP-CMAC-256 to select one of the ciphers defined in
IEEE Std 802.11ac-2013.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Incorrect PTK length was used in PMK-to-PTK derivation and the Michael
MIC TX/RX key swapping code was incorrectly executed for these ciphers
on supplicant side.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This new mechanism allows P2P Client to request an IPv4 address from the
GO as part of the 4-way handshake to avoid use of DHCP exchange after
4-way handshake. If the new mechanism is used, the assigned IP address
is shown in the P2P-GROUP-STARTED event on the client side with
following new parameters: ip_addr, ip_mask, go_ip_addr. The assigned IP
address is included in the AP-STA-CONNECTED event on the GO side as a
new ip_addr parameter. The IP address is valid for the duration of the
association.
The IP address pool for this new mechanism is configured as global
wpa_supplicant configuration file parameters ip_addr_go, ip_addr_mask,
ip_addr_star, ip_addr_end. For example:
ip_addr_go=192.168.42.1
ip_addr_mask=255.255.255.0
ip_addr_start=192.168.42.2
ip_addr_end=192.168.42.100
DHCP mechanism is expected to be enabled at the same time to support P2P
Devices that do not use the new mechanism. The easiest way of managing
the IP addresses is by splitting the IP address range into two parts and
assign a separate range for wpa_supplicant and DHCP server.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
It was already possible to configure hostapd and wpa_supplicant to use
FT-SAE for the key management, but number of places were missing proper
AKM checks to allow FT to be used with the new AKM.
Signed-hostap: Jouni Malinen <j@w1.fi>
The previous version could end up calling WPA authenticator routines
even though the authenticator had not been initialized and this could
result in NULL pointer dereference.
Signed-hostap: Jouni Malinen <j@w1.fi>
If configuration of the group key to the driver fails, move the WPA
group into failed state and indication group setup error to avoid cases
where AP could look like it is working even through the keys are not set
correctly.
Signed-hostap: Jouni Malinen <j@w1.fi>
Some driver wrappers may implement this by writing eight octets even
though IPN is only six octets. Use a separate WPA_KEY_RSC_LEN (8) octet
buffer in the call to make sure there is enough buffer room available
for the full returned value and then copy it to IPN field.
The previous implementation used the following igtk field as the extra
buffer and then initialized that field afterwards, so this change does
not fix any real issue in behavior, but it is cleaner to use an explicit
buffer of the maximum length for get_seqnum().
Signed-hostap: Jouni Malinen <j@w1.fi>
When using per-device PSKs, select the PSK based on the P2P Device
Address of the connecting client if that client is a P2P Device. This
allows the P2P Interface Address to be changed between P2P group
connections which may happen especially when using persistent groups.
Signed-hostap: Jouni Malinen <j@w1.fi>
This can be used to implement per-device PSK selection based on the
peer's P2P Device Address instead of P2P Interface Address.
Signed-hostap: Jouni Malinen <j@w1.fi>
This is needed to avoid allowing the STA to reconnect using a cached
PMKSA. ESS disassoc imminent notification is normally used to indicate
that the STA session will be terminated and as such, requiring full
authentication through the authentication server after this is needed.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
Since "pairwise" is defined as an integer, the current assignment leads
to it having the value 0 or 8, which is a bit strange in debug output:
WPA: Send EAPOL(version=2 secure=1 mic=1 ack=1 install=1 pairwise=8
kde_len=46 keyidx=2 encr=1)
Use !!(...) to normalize it to 0 or 1.
Signed-hostap: Johannes Berg <johannes.berg@intel.com>
For some testing it can be useful to force the Key MIC in group
EAPOL-Key frames to be corrupt. Add an option to allow setting a
probability for corrupting the Key MIC and use it in the WPA code,
increasing the first byte of the MIC by one to corrupt it if desired.
Signed-hostap: Johannes Berg <johannes.berg@intel.com>
When the OS is out of random bytes in SM_STATE(WPA_PTK, AUTHENTICATION2)
in ap/wpa_auth.c, hostapd sends the sm to state DISCONNECT without
clearing ReAuthenticationRequest, resulting in an infinite loop.
Clearing sm->ReAuthenticationRequest using gdb fixes the running hostapd
instance for me. Also sm->Disconnect = TRUE should be used instead of
wpa_sta_disconnect() to make sure that the incomplete ANonce does not
get used.
Fix this issue by resetting sm->ReAuthenticationRequest even if the STA
gets disconnected and use sm->Disconnect instead of
wpa_sta_disconnect().
Signed-hostap: Michael Braun <michael-dev@fami-braun.de>
The code initializing GMK Counter uses the group pointer value as extra
entropy and to distinguish different group instances. Some static
analyzers complain about the sizeof(pointer) with memcpy, so use a more
explicit type casting to make it more obvious what the code is doing.
Signed-hostap: Jouni Malinen <j@w1.fi>
Replace CONFIG_IEEE80211V with CONFIG_WNM to get more consistent build
options for WNM-Sleep Mode operations. Previously it was possible to
define CONFIG_IEEE80211V without CONFIG_WNM which would break the build.
In addition, IEEE 802.11v has been merged into IEEE Std 802.11-2012 and
WNM is a better term to use for this new functionality anyway.
Signed-hostap: Jouni Malinen <j@w1.fi>
Commit 296a34f0c1 changed hostapd to
remove the internal STA entry at the beginning of TKIP countermeasures.
However, this did not take into account the case where this is triggered
by an EAPOL-Key error report from a station. In such a case, WPA
authenticator state machine may continue processing after having
processed the error report. This could result in use of freed memory.
Fix this by stopping WPA processing if the STA entry got removed.
Signed-hostap: Jouni Malinen <j@w1.fi>
This introduces new AKM for SAE and FT-SAE and adds the initial parts
for going through the SAE Authentication frame exchange. The actual SAE
algorithm and new fields in Authentication frames are not yet included
in this commit and will be added separately. This version is able to
complete a dummy authentication with the correct authentication
algorithm and transaction values to allow cfg80211/mac80211 drivers to
be tested (all the missing parts can be handled with
hostapd/wpa_supplicant changes).
Signed-hostap: Jouni Malinen <j@w1.fi>
Try to share most of the cipher information like key and RSC lengths and
suite selector conversions, etc. in wpa_common.c to avoid having similar
code throughout the WPA implementation for handling cipher specific
behavior.
Signed-hostap: Jouni Malinen <j@w1.fi>
This allows both hostapd and wpa_supplicant to be used to derive and
configure keys for GCMP. This is quite similar to CCMP key
configuration, but a different cipher suite and somewhat different rules
are used in cipher selection. It should be noted that GCMP is not
included in default parameters at least for now, so explicit
pairwise/group configuration is needed to enable it. This may change in
the future to allow GCMP to be selected automatically in cases where
CCMP could have been used.
This commit does not included changes to WPS or P2P to allow GCMP to be
used.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
When using multiple VLANs, GKeyDoneStations counter is not updated
properly since wpa_auth_for_each_sta() call in wpa_group_setkeys() ends
up iterating through all STAs and not just the STAs of a specific
wpa_group (VLAN). Consequently, GTK rekeying gets initialized multiple
times if more than a single group state machine exists. Fix this by
iterating only through the STAs in the specific wpa_group.
Signed-hostap: Michael Braun <michael-dev@fami-braun.de>
intended-for: hostap-1
disable_dgaf=1 in hostapd.conf can now be used to disable downstream
group-addressed forwarding (DGAF). In this configuration, a unique
GTK (and IGTK) is provided to each STA in the BSS to make sure the
keys do not match and no STA can forge group-addressed frames.
An additional mechanism in the AP needs to be provided to handle some
group-addressed frames, e.g., by converting DHCP packets to unicast
IEEE 802.11 frames regardless of their destination IP address and by
providing Proxy ARP functionality.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
Definition of ANonce selection in IEEE Std 802.11i-2004 is somewhat
ambiguous. The Authenticator state machine uses a counter that is
incremented by one for each 4-way handshake. However, the security
analysis of 4-way handshake points out that unpredictable nonces help in
preventing precomputation attacks. Instead of the state machine
definition, use an unpredictable nonce value here to provide stronger
protection against potential precomputation attacks.
Signed-hostap: Jouni Malinen <j@w1.fi>
When there is not enough entropy and there are two station associating
at the same time, one of the stations will be rejected, but during
that rejection, the "reject_4way_hs_for_entropy" flag gets cleared. This
may allow the second station to avoid rejection and complete a 4-Way
Handshake with a GTK that will be cleared as soon as more entropy is
available and another station connects.
This reworks the logic to ban all 4-way handshakes until enough entropy
is available.
Signed-hostap: Nicolas Cavallari <cavallar@lri.fr>
Some deployed station implementations seem to send msg 4/4 with
incorrect type value in WPA2 mode. Add a workaround to ignore that issue
so that such stations can interoperate with hostapd authenticator. The
validation checks were added in commit
f8e96eb6fd.
Signed-hostap: Jouni Malinen <j@w1.fi>
Some supplicant implementations (e.g., Windows XP WZC) update SNonce for
each EAPOL-Key 2/4. This breaks the workaround on accepting any of the
pending requests, so allow the SNonce to be updated even if we have
already sent out EAPOL-Key 3/4.
While the issue was made less likely to occur when the retransmit
timeout for the initial EAPOL-Key msg 1/4 was increased to 1000 ms,
this fixes the problem even if that timeout is not long enough.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
The 100 ms timeout on retransmitting group key message can be too short
for stations that are in power save mode or if there is a large number
of association stations. While the retransmission of the EAPOL-Key frame
should allow this to be recovered from, it is useful to avoid
unnecessary frames to save soem CPU and power.
Signed-hostap: Jouni Malinen <j@w1.fi>
This previously helped when debugging some auth issues when hitting the
AP with 128 association attempts all at once.
Signed-off-by: Ben Greear <greearb@candelatech.com>
This is a workaround for Windows 7 supplicant rejecting WPA msg 3/4
in case it used Secure=1 in msg 2/4. This can happen, e.g., when
rekeying PTK after EAPOL-Key Error Request (Michael MIC failure)
from the supplicant.
Signed-hostap: Jouni Malinen <jouni@qca.qualcomm.com>
The GTK is renewed in the hostapd after a MIC attack dissassociation
without informing the driver, causing decryption failures. This patch
sends the new GTK/IGTK to the driver after it is updated by the hostapd.
Signed-off-by: Yoni Divinsky <yoni.divinsky@ti.com>
Some stations have been reported to send EAPOL-Key Error Reports
indicating Michael MIC failures even when the cipher is not TKIP
(e.g., when the network is using only CCMP). Ignore such reports
to avoid starting TKIP countermeasures unnecessarily. This can
prevent certaint types of denial of service attacks by insiders,
but mostly this is to work around invalid station implementations.
Signed-hostap: Jouni Malinen <j@w1.fi>
Split WPA initialization into two parts so that the Beacon frames can be
configured fully before the initial keys (GTK/IGTK) are configured. This
makes it easier for drivers that depend on the AP security mode being
fully set before the keys are configured.
A new hostapd configuration parameter, disable_pmksa_caching=1, can now
be used to disable PMKSA caching on the Authenticator. This forces the
stations to complete EAP authentication on every association when WPA2
is being used.
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.
Previously, a bug in GKeyDoneStations count would remain in effect
until the authenticator is restarted. Make this more robust by
clearing the station count before each rekeying setup. While this
is not really supposed to be needed, there has been bugs in this
area in the past and it is safer to make the implementation recover
from such an issue automatically.
If the STA to be freed is still included in GKeyDoneStations count,
decrement the count when the STA is freed. This does not happen in
AP mode since there is enough time to go through the authenticator
state machine to clear the STA. However, in the current RSN IBSS
implementation, the authenticator state for the STA is freed in a
way that does not allow the state machine to go through the clearing.
To address this, make sure that wpa_free_sta_sm() decrements the
GKeyDoneStations count if the STA happened to be in the process of
GTK rekeying.
os_snprintf() can be a preprocessor macro and according to
C99 (6.10.3 clause 11) the results of having preprocessor directives
inside the macro arguments is undefined.
Previously, both NULL and ff:ff:ff:ff:ff:ff addr were used in various
places to indicate default/broadcast keys. Make this more consistent
and useful by defining NULL to mean default key (i.e., used both for
unicast and broadcast) and ff:ff:ff:ff:ff:ff to indicate broadcast
key (i.e., used only with broadcast).
On Linux, verify that the kernel entropy pool is capable of providing
strong random data before allowing WPA/WPA2 connection to be
established. If 20 bytes of data cannot be read from /dev/random,
force first two 4-way handshakes to fail while collecting entropy
into the internal pool in hostapd. After that, give up on /dev/random
and allow the AP to function based on the combination of /dev/urandom
and whatever data has been collected into the internal entropy pool.
By default, make hostapd and wpa_supplicant maintain an internal
entropy pool that is fed with following information:
hostapd:
- Probe Request frames (timing, RSSI)
- Association events (timing)
- SNonce from Supplicants
wpa_supplicant:
- Scan results (timing, signal/noise)
- Association events (timing)
The internal pool is used to augment the random numbers generated
with the OS mechanism (os_get_random()). While the internal
implementation is not expected to be very strong due to limited
amount of generic (non-platform specific) information to feed the
pool, this may strengthen key derivation on some devices that are
not configured to provide strong random numbers through
os_get_random() (e.g., /dev/urandom on Linux/BSD).
This new mechanism is not supposed to replace proper OS provided
random number generation mechanism. The OS mechanism needs to be
initialized properly (e.g., hw random number generator,
maintaining entropy pool over reboots, etc.) for any of the
security assumptions to hold.
If the os_get_random() is known to provide strong ramdom data (e.g., on
Linux/BSD, the board in question is known to have reliable source of
random data from /dev/urandom), the internal hostapd random pool can be
disabled. This will save some in binary size and CPU use. However, this
should only be considered for builds that are known to be used on
devices that meet the requirements described above. The internal pool
is disabled by adding CONFIG_NO_RANDOM_POOL=y to the .config file.
This commit adds a new wrapper, random_get_bytes(), that is currently
defined to use os_get_random() as is. The places using
random_get_bytes() depend on the returned value being strong random
number, i.e., something that is infeasible for external device to
figure out. These values are used either directly as a key or as
nonces/challenges that are used as input for key derivation or
authentication.
The remaining direct uses of os_get_random() do not need as strong
random numbers to function correctly.
This adds more time for the system entropy pool to be filled before
requesting random data for generating the WPA/WPA2 encryption keys.
This can be helpful especially on embedded devices that do not have
hardware random number generator and may lack good sources of
randomness especially early in the bootup sequence when hostapd is
likely to be started.
GMK and Key Counter are still initialized once in the beginning to
match the RSN Authenticator state machine behavior and to make sure
that the driver does not transmit broadcast frames unencrypted.
However, both GMK (and GTK derived from it) and Key Counter will be
re-initialized when the first station connects and is about to
enter 4-way handshake.
