This allows the user to complete WPS provisioning using PBC by
selected a specific Enrollee even if there are other Enrollees
in active PBC mode at the same time. The other Enrollees will be
rejected should they try to connect at the same time.
Ignore possible mismatches in the source address of the frame and only
use UUID-E to check whether a Probe Request or M1 is from the same
Enrollee when figuring out whether there is PBC session overlap. This
is needed to avoid potential issues with Enrollee devices that may have
multiple interfaces indicating active PBC state.
If there is already a link the requested peer, request start of
renegotiation instead of completely new link. This seems to be needed
to allow some driver to accept the trigger for a new negotiation.
tdls_testing 0x80 can now be used to request wpa_supplicant to send
a TDLS Setup Request frame (i.e., try to initiate TDLS setup)
whenever a TDLS Setup Request frame has been received from a peer.
Blindly clearing all struct wpa_tdls_peer members is a risky
operation since it could easily clear pointers to allocated
memory, etc. information that really should not be removed.
Instead of hoping that new code gets added here to restore
the important variables, reverse the approach and only clear
structure members one by one when needed.
This allows the same code path to be used for both protected and
unprotected configurations to limit need for duplicated code in
both the driver and wpa_supplicant.
This adds code for validating that the same Key Lifetime is used
throughout TPK handshake and enables TPK lifetime expiration and
renewal/teardown. These part seem to be working, but the actual
handling of TDLS Setup Confirm and renewal are not fully
functional yet.
tdls_testing bits:
bit 3 = use short TPK lifetime (301 seconds)
bit 4 = use wrong TPK lifetime in TDLS Setup Response
bit 5 = use wrong TPK lifetime in TDLS Setup Confirm
These special test cases can be configured at run time with "wpa_cli
tdls_testing <value>" where <value> is an integer (either as a decimal
or as a hex value with 0x prefix) bitmap of special features with
following bits available at this point:
bit 0 = long frame (add dummy subelement to make FTIE very long)
bit 1 = use alternative RSN IE (different RSN capab value and no extra
replay counters)
bit 2 = send incorrect BSSID in Link Identifier of TDLS Setup Request
(e.g., 1 = long FTIE, 2 = different RSN IE, 3 = both of those)
This is disabled by default and can be enabled for the build by
adding the following line to .config:
CFLAGS += -DCONFIG_TDLS_TESTING
This allows driver wrappers to indicate whether the association was
done using Association Request/Response or with Reassociation
Request/Response frames.
CFLAGS += -DCONFIG_VALGRIND can now be used to enable support for
Valgrind client requests to help in removing some non-issues from
Valgrind reports. This is mainly aimed at allowing unknown ioctl
results to be marked as defined data.
If the os_time_t variable used for the expiration time (seconds)
overflows when the registered timeout value is being added,
assume that the event would happen after an infinite time, i.e.,
would not really happen in practice. This fixes issues with
long key timeouts getting converted to immediate expiration due
to the overflow.
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.
Build options can now be used to replace the location of client
sockets for UNIX domain socket control interface:
CFLAGS += -DCONFIG_CTRL_IFACE_CLIENT_DIR=\"/tmp\"
CFLAGS += -DCONFIG_CTRL_IFACE_CLIENT_PREFIX=\"wpa_ctrl_\"
The buffer size for routing socket is fixed to 2048.
This patch fix it to obtain the size from OS.
This patch worked on x86 platform with NetBSD 5.0.2.
This allows keystore:// prefix to be used with client_cert and
private_key configuration parameters.
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>
This will only retrieve information about peers that have been fully
discovered, not peers that are only half-discovered based on their Probe
Request frames.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The DBus code will want to have perfect matching of dev_found and the
dev_lost it adds so it doesn't need to keep track internally. Enable
that with a new flag in the core that tracks whether we have already
notified about this -- the existing users can ignore it.
The part where this is always set to 1 if the new device is discovered
by a driver that has P2P in the driver is buggy -- the driver should
feed the P2P peer database and then that should feed the notification
here instead.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
This embeds some information about each P2P peer that will be publically
visible in a struct that is shared.
The dev_found notification function is also passed the new struct, which
requires some work for the driver-based P2P management.
Signed-off-by: Konguraj(Raj) Kulanthaivel <konguraj.kulanthaivel@intel.com>
Signed-off-by: Fabien Marotte <fabienx.marotte@linux.intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When using multiple vifs and dynamic vlan tagging is enabled on any
interface, the created AP VLAN interfaces get the BSSID of the first
AP mode interface instead of the BSSID of the corresponding AP mode
interface.
Example:
wlan0 - xx:xx:xx:xx:x0
wlan1 - xx:xx:xx:xx:x1
Assume a STA connects to the AP interface wlan1 and gets a dynamic
VLAN tag 100 assigned by the RADIUS server. Hostapd will create an AP
VLAN interface wlan1.100 but doesn't set an address for this interface
which results in wlan1.100 getting the same address as wlan0:
wlan1.100 - xx:xx:xx:xx:x0
As a result the STA that was moved to wlan1.100 isn't able to finish its
4-way handshake since mac80211 won't pass its frames to wlan1.100 due to
the different address.
To fix this issue make use of the address of the AP interface when
creating an AP VLAN interface.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Add MAC addresses for stations and use wpa_msg instead of printf
methods to make it easier to grep logs and find messages for the
station in question.
Signed-off-by: Ben Greear <greearb@candelatech.com>
If parsing of the certificate or private key succeeds using any of
the tried encoding types, clear the OpenSSL error queue without
showing the pending errors in debug log since they do not really
provide any useful output and can be confusing.
This patch fixes a problem I had when I tried to connect
an embedded system [wpa_supplicant, CONFIG_TLS=internal]
to my TLS secured network.
TLSv1: Send CertificateVerify
TLSv1: CertificateVerify hash - hexdump(len=36): ha .. ha
PKCS #1: pkcs1_generate_encryption_block - Invalid buffer lengths \
(modlen=512 outlen=454 inlen=36)
It turned out that a fixed 1000 byte message buffer was just
a little bit too small for the 4096 bit RSA certificates
I'm using.
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
In multi BSS setups it wasn't possible to set up an HT BSS in
conjunction with a WEP/TKIP BSS. HT needed to be turned off entirely
to allow WEP/TKIP BSSes to be used.
In order to allow HT BSSes to coexist with non-HT WEP/TKIP BSSes add a
new BSS conf attribute "disable_11n" which disables HT capabilities on a
single BSS by suppressing HT IEs in the beacon and probe response
frames. Furthermore, mark all STAs associated to a WEP/TKIP BSS as
non-HT STAs. The disable_11n parameter is used internally; no new entry
is parsed from hostapd.conf.
This allows a non-HT WEP/TKIP BSS to coexist with a HT BSS without
having to disable HT mode entirely. Nevertheless, all STAs associated to
the WEP/TKIP BSS will only be served as if they were non-HT STAs.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Use l2_packet with Ethernet header included so that the source
address for RRB packets can be forced to be the local BSSID. This
fixes problems where unexpected bridge interface address may end
up getting used and the recipient of the frame dropping it as
unknown R0KH/R1KH.
This is needed to keep kernel and hostapd in sync. In addition,
the obsolete entry in hostapd prevented use of Deauthentication
or Disassociation frame with reason code 6/7 mechanism to indicate
to the STA that there is no association if the STA missed the
broadcast Deauthentication frame for any reason.
The internal pointer to RADIUS client configuration needs to be
updated whenever a new hostapd configuration is loaded. Without
this, freed memory may be dereferenced and this can result in
segmentation faults.
This converts number of debugging messages to use wpa_msg() in order
to allow the interface name to be shown with the messages.
A new function, wpa_dbg(), is introduced to allow
CONFIG_NO_STDOUT_DEBUG=y builds to remove the debug strings. This is
otherwise identical with wpa_msg(), but it gets compiled out if stdout
debugging is disabled.
This is needed to allows WPS PBC session overlap detection to work
with drivers that process Probe Request frames internally. This
code is is run in hostapd, but the wpa_supplicant AP mode did not
have call to the hostapd_probe_req_rx() function even though it
registered handlers for hostapd Probe Request RX callbacks.
An optional parameter, p2p_dev_addr, can now be given to WPS_PBC
command on P2P GO to indicate that only the P2P device with the
specified P2P Device Address is allowed to connect using PBC. If
any other device tries to use PBC, a session overlap is indicated
and the negotiation is rejected with M2D. The command format for
specifying the address is "WPS_PBC p2p_dev_addr=<address>", e.g.,
WPS_PBC p2p_dev_addr=02:03:04:05:06:07
In addition, show the PBC session overlap indication as a WPS failure
event on an AP/GO interface. This particular new case shows up as
"WPS-FAIL msg=4 config_error=12".
This makes log files much more readable if multiple interfaces
are being controlled by the same process. The interface name is
added to stdout/file/syslog entries, but not to the messages
sent to control interface monitors to avoid issues with parsing
in external programs.
Signed-off-by: Ben Greear <greearb@candelatech.com>
Add a callback for station's authorized transitions
to allow wpa_supplicant to emit events in DBus.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
To enable making state change notifications on the WLAN_STA_AUTHORIZED
flag, introduce ap_sta_set_authorized(), and to reduce use of the flag
itself also add a wrapper for testing the flag: ap_sta_is_authorized().
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
If the driver advertises max_remain_on_chan data, use it instead of
the hardcoded value of 5000. Keep the default at 5000 since that is the
value used by earlier versions of cfg80211/mac80211 and not advertised
in nl80211 for those.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When authenticating, and the interface type is not already
NL80211_IFTYPE_STATION, we need to call wpa_driver_nl80211_set_mode()
only once. Remove the excessive call.
Signed-off-by: Eliad Peller <eliad@wizery.com>
We need to call this function after having completed the neighboring
channel scan to figure out whether HT40 can be used and if so, which
channel is the secondary channel.
ieee80211n_supported_ht_capab is called after ieee80211n_check_40mhz in
function hostapd_check_ht_capab. ieee80211n_check_40mhz can return 1 in
a valid scenario where the initialization is completed in a callback. In
this case ieee80211n_supported_ht_capab is skipped and hostapd does not
check the ht capabilities reported by the driver. Fix this issue making
sure ieee80211n_supported_ht_capab gets called.
Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
If WMM is not disabled explicitly (wmm_enabled=0 in hostapd.conf),
enable WMM automatically whenever HT (ieee80211n) is enabled. Use
the default WMM parameters for AP TX queues and the EDCA parameters
advertised for stations in WMM IE if no overriding values are
included in the configuration.
We can use the P2P interface types to check if the driver supports P2P
and to tell the kernel that a given interface is going to be used for
P2P (when it is created).
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
There is no real reason to maintain these in the current development
branch anymore. If someone really needs support for the obsolete
driver interfaces, these can be found in older wpa_supplicant
branches.
driver_atmel.c
- vendor-specific interface for ATMEL AT76C5XXx cards
- for some old out-of-tree driver; not for the upstream atmel*
drivers
driver_ndiswrapper.c
- vendor-specific interface for an out-of-tree driver
- ndiswrapper should work with driver_wext.c, too
driver_ipw.c
- vendor-specific interface for old ipw2100/2200 driver
- the upstream driver works with driver_wext.c (and does not work
with the old interface)
driver_hermes.c
- vendor driver that was not even included in the main wpa_supplicant
releases
In order to enable protection mechanisms for different HT opmodes the
driver needs to be aware of the current HT opmode that is calculated by
hostapd. Hence, pass the current opmode to the nl80211 driver via
the bss attribute NL80211_ATTR_BSS_HT_OPMODE.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
tdls_prohibit=1 and tdls_prohibit_chan_switch=1 and now be used to
disable use of TDLS or TDLS channel switching in the BSS using
extended cabilities IE as defined in IEEE 802.11z.
Make sure that received management frames are long enough before
processing them. This avoids a potential segmentation fault if a
driver delivers an invalid frame all the way to hostapd.
The changes are:
1. the word "and" in the hunting-and-pecking string passed to the KDF
should be capitalized.
2. the primebitlen used in the KDF should be a short not an int.
3. the computation of MK in hostap is based on an older version of the
draft and is not the way it's specified in the RFC.
4. the group being passed into computation of the Commit was not in
network order.
Previously, only the Configuration Error values were indicated in
WPS-FAIL events. Since those values are defined in the specification
it is not feasible to extend them for indicating other errors. Add
a new error indication value that is internal to wpa_supplicant and
hostapd to allow other errors to be indicated.
Use the new mechanism to indicate if negotiation fails because of
WEP or TKIP-only configurations being disallows by WPS 2.0.
This needs to be done both in the more normal location in
p2p_timeout_connect_listen() (internal timeout after driver event) and
in p2p_listen_end() as a workaround for the case where the driver event
is delayed to happen after the internal timeout.
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).
When hostapd is removing a virtual BSS interface, the loop here was
incorrectly not updating the iterator during list traversal and
ended up in an infinite loop in some cases.
Use NULL instead of (u8 *) "" as the seq value and make sure the
driver wrapper implementations can handle NULL value. This was
previously already done in number of places, but not everywhere.
wpa_supplicant seems to crash from time to time on a NetBSD 4.0 MIPS
platform. The root cause turned out to be a MIPS alignment issue.
In my wpa_supplicant crash case, in function
wpa_driver_bsd_event_receive (from driver_bsd.c), the buf[2048] address
is started from i.e. 0x7fffd546, which is not 4 bytes aligned. Later
when it is casted to (struct if_msghdr *), and rtm->rtm_flags is used.
rtm->rtm_flags is "int" type, but its address is not 4 bytes aligned.
This is because the start address of rtm is not 4 bytes aligned.
Unfortunately in NetBSD MIPS kernel (unlike Linux MIPS kernel emulates
unaligned access in its exception handler), the default behavior is to
generate a memory fault to the application that accesses unaligned
memory address. Thus comes the early mentioned wpa_supplicant crash. An
interesting note is when I'm using the wpa_supplicant version 0.4.9, I
never saw this problem. Maybe the stack layout is different. But I
didn't look into details.
I used below patch to resolve this problem. Now it runs correctly for at
least several hours. But you might have a better fix (maybe we can use
malloc/free so that it is at least cache line aligned?). I'm also not
sure if other drivers should have the same problem.
This adds partial callbacks and events to allow P2P management to be
implemented in a driver/firmware. This is not yet complete and is
very much subject to change in the future.
CONFIG_WPS_REG_DISABLE_OPEN=y can be used to configure wpa_supplicant
to disable open networks by default when wps_reg command is used to
learn the current AP settings. When this is enabled, there will be a
WPS-OPEN-NETWORK ctrl_iface event and the user will need to explicitly
enable the network (e.g., with "select_network <id>") to connect to
the open network.
If the underlying driver supports off-channel TX, it will now be used by
the nl80211 driver wrapper, setting WPA_DRIVER_FLAGS_OFFCHANNEL_TX
accordingly.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
With the new kernel functionality coming to Linux to allow off-channel
TX, we can take advantage of that in the P2P code that currently uses
remain-on-channel. If a driver advertises support for it, it will be
asked to handle off-channel TX by itself.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The nl80211 driver can report low ACK condition (in fact it reports
complete loss right now only). Use that, along with a config option, to
disconnect stations when the data connection is not working properly,
e.g., due to the STA having went outside the range of the AP. This is
disabled by default and can be enabled with disassoc_low_ack=1 in
hostapd or wpa_supplicant configuration file.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
ClientTimeout changes from EAP peer methods were not supposed to
change behavior for other EAP peer methods or even other sessions
of the same method. Re-initialize ClientTimeout whenever an EAP
peer method is initialized to avoid this. This addresses problems
where WPS (EAP-WSC) reduces the timeout and consecutive EAP runs
may fail due to too small timeout.
If the peer you want to connect to is no longer available (does not
acknowledge frames) when wpa_supplicant sends GO Negotition Request
frames, retransmission of this frame is done until the associated
p2p_device structure is removed on timeout. In that case, no signal
is emitted to inform the GO Negotiation has failed.
When sending an Invitation Request frame, the same retransmission
mechanism is in place but limit the transmission to 100 and hitting
the limit generates an event.
This patch adds the same mechanism as the one in place for Invitation
Request, but with limit of 120 to match the existing wait_count for
for GO Negotiation.
If a station received unprotected Deauthentication or Disassociation
frame with reason code 6 or 7 from the current AP, there may be a
mismatch in association state between the AP and STA. Verify whether
this is the case by using SA Query procedure. If not response is
received from the AP, deauthenticate.
This implementation is only for user space SME with
driver_nl80211.c.
Some new code will require access to P2P group members, so add API to
retrieve the number of members and iterate the members themselves.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
If there is a pending GO Negotiation when p2p_cancel is used,
unauthorize the peer to avoid immediate reconnection from being
accepted without a new p2p_connect command.
This can be used by external programs (e.g., wlantest_cli) to inject
raw frames (hex dump of the frame header and body). The data can be
requested to be sent as-is or protected with the current key.
Previously, pairwise and group cipher suites were configured only
when kernel SME (nl80211 connect API) was used. However, mac80211
needs this information even in the user space SME case for one
thing: to disable HT when TKIP/WEP is used. Add
NL80211_ATTR_CIPHER_SUITES_PAIRWISE to fix this special case with
user space SME. This allows mac80211 to disable HT properly when
the AP is configured with configuration that is not allowed.
This add preliminary code for setting the per-STA RX GTK for
RSN IBSS when nl80211 drivers. For some reason, this does not
seem to fully work, but at least driver_nl80211.c is now aware of
what kind of key is being set and the whatever is missing from
making this key configuration go through should be specific to
nl80211/cfg80211.
The frame needs to be sent from an individual (non-group) address,
so drop invalid frames before sending Deauth/Disassoc frames to
not associated STAs.
One of the pointers to the PAC buffer was not updated after realloc
and if the realloc ended up returning new pointer, the *pos pointer
was still pointing at the old location (i.e., freed memory at
this point).
When controlling multiple virtual interfaces on the same physical
radio, share the scan results events with sibling interfaces. This
decreases the time it takes to connect many virtual interfaces.
This is currently only supported on Linux with cfg80211-based
drivers when using nl80211 or wext driver interface.
Signed-off-by: Ben Greear <greearb@candelatech.com>
Some drivers are not providing exactly reliable error codes (e.g.,
with WEXT), but others may actually indicate reliable information.
Allow driver wrappers to indicate if that is the case and use
optimizations if so. For now, this improves nl80211 with
NL80211_CMD_CONNECT for a case where connection request fails.
driver_param=use_p2p_group_interface=1 can now be used to test
nl80211-drivers with separate P2P group interface. In other words,
the main interface (e.g., wlan0) is reserved for P2P management
operations and non-P2P connections and a new group interface (e.g.,
p2p-wlan0-0) is created for the P2P group.
This implementation is very minimal, i.e., it only support address
allocation for a single P2P group interface (if the driver does not
handle this internally). In addition, not all functionality has yet
been tested, so for now, this is disabled by default and needs that
special driver_param to enable.
WPA_DRIVER_FLAGS_P2P_MGMT_AND_NON_P2P flag can now be used to
indicate that the initial interface (e.g., wlan0) is used for
P2P management operations and potentially non-P2P connections.
This is otherwise identical to
WPA_DRIVER_FLAGS_P2P_DEDICATED_INTERFACE, but the possibility of
non-P2P connections makes some operations differ.
Getting rid of these inline functions seems to reduce the code size
quite a bit, so convert the most commonly used hostapd driver ops to
function calls.
This is not needed anymore and just makes things more difficult
to understand, so move the remaining function pointers to direct
function calls and get rid of the struct hostapd_driver_ops.
send_eapol, set_key, read_sta_data, sta_clear_stats,
set_radius_acl_auth, set_radius_acl_expire, and set_beacon
to use inline functions instead of extra abstraction.
Commit bf65bc638f started the path to
add this new abstraction for driver operations in AP mode to allow
wpa_supplicant to control AP mode operations. At that point, the
extra abstraction was needed, but it is not needed anymore since
hostapd and wpa_supplicant share the same struct wpa_driver_ops.
Start removing the unneeded abstraction by converting
send_mgmt_frame() to an inline function, hostapd_drv_send_mlme().
This is similar to the design that is used in wpa_supplicant and
that was used in hostapd in the past (hostapd_send_mgmt_frame()
inline function).
driver.h defines these functions to return 0 on success, not
number of bytes transmitted. Most callers are checking "< 0" for
error condition, but not all. Address this by following the driver
API specification on 0 meaning success.
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.
wlan0: RADIUS No authentication server configured
MEMLEAK[0x999feb8]: len 1040
WPA_TRACE: memleak - START
[3]: ./hostapd(radius_msg_new+0x33) [0x8074f43]
radius_msg_new() ../src/radius/radius.c:117
[4]: ./hostapd() [0x806095e]
ieee802_1x_encapsulate_radius() ../src/ap/ieee802_1x.c:439
ieee802_1x_aaa_send() ../src/ap/ieee802_1x.c:1496
For example, this error occured when I used WPS hostapd without
"eap_server=1" definition in configuration file.
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.
