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.
The example GMK-to-GTK derivation described in the IEEE 802.11 standard
is marked informative and there is no protocol reason for following it
since this derivation is done only on the AP/Authenticator and does not
need to match with the Supplicant. Mix in more data into the derivation
process to get more separation from GMK.
When using WPS, we may end up here if the STA manages to re-associate
without the previous STA entry getting removed. Consequently, we need to
make sure that the WPA state machines gets initialized properly at this
point.
IEEE Std 802.11r-2008, 11A.4.2 describes FT initial mobility domain
association in an RSN to include PMKR1Name in the PMKID-List field
in RSN IE in messages 2/4 and 3/4. This makes the RSN IE not be
bitwise identical with the values used in Beacon, Probe Response,
(Re)association Request frames.
The previous versions of wpa_supplicant and hostapd did not add the
PMKR1Name value in EAPOL-Key frame and did not accept it if added
(due to bitwise comparison of RSN IEs). This commit fixes the
implementation to be compliant with the standard by adding the
PMKR1Name value into EAPOL-Key messages during FT 4-Way Handshake and
by verifying that the received value matches with the value derived
locally.
This breaks interoperability with previous wpa_supplicant/hostapd
versions.
It turns out that this is needed for both FT-over-DS and FT-over-air
when using mac80211, so it looks easiest to just unconditionally
re-configure the keys after reassociation when FT is used.
This seems to be needed at least with mac80211 when a STA is using
FT-over-DS to reassociate back to the AP when the AP still has the
previous association state.
IGTK get_seqnum needs to be skipped in the same way as GTK one when
rekeying group keys. Previously, the old PN value (the one from the
previous key) was indicated and that resulted in MMIE replay detection
at the station.
Must update sm->pairwise when fetching PMK-R1 SA.
Add a workaround for drivers that cannot set keys before association
(e.g., cfg80211/mac80211): retry PTK configuration after association.
wpa_sm_step() could theoretically free the statemachine, but it does
not do it in this particular case. Anyway, the code can be cleaned to
verify whether the state machine data is still available after the
wpa_sm_step() call.
The three existing enums were already depending on using the same
values in couple of places and it is just simpler to standardize on
one of these to avoid need for mapping between different enums for
the exact same thing.
Doxygen and some build tools may get a bit confused about same file
name being used in different directories. Clean this up a bit by
renaming some of the duplicated file names in src/ap.
