Jouni Malinen [Sat, 3 Aug 2019 14:00:39 +0000 (17:00 +0300)]
SAE: Fix KCK, PMK, and PMKID derivation for groups 22, 23, 24
IEEE Std 802.11-2016 is not exactly clear on the encoding of the bit
string that is needed for KCK, PMK, and PMKID derivation, but it seems
to make most sense to encode the (commit-scalar + peer-commit-scalar)
mod r part as a bit string by zero padding it from left to the length of
the order (in full octets).
The previous implementation used the length of the prime (in full
octets). This would work for KCK/PMK, but this results in deriving all
zero PMKIDs for the groups where the size of the order is smaller than
the size of the prime. This is the case for groups 22, 23, and 24.
However, those groups have been marked as being unsuitable for use with
SAE, so this fix should not really have a practical impact anymore.
Anyway, better fix it and document this clearly in the implementation
taken into account the unclarity of the standard in this area.
Jouni Malinen [Sat, 3 Aug 2019 13:28:02 +0000 (16:28 +0300)]
OpenSSL: Fix crypto_bignum_to_bin() with padlen == 0
The earlier change to add support for BN_bn2binpad() and
BN_bn2bin_padded() broke this function for cases where no padding is
used (padlen == 0). Those would have always failed after the changes and
the function would return -1. There are no such cases in the current
hostap.git, so this did not have any real issues, but anyway, better fix
this function to match its documentation.
Fixes: 1e237903f5b5 ("OpenSSL: Use BN_bn2binpad() or BN_bn2bin_padded() if available") Signed-off-by: Jouni Malinen <j@w1.fi>
nl80211: Use separate flag for 4-way handshake offload
Commit d896874f8689 ("nl80211: Indicate 802.1X 4-way handshake offload
in connect") used the req_key_mgmt_offload flag to indicate to the
driver that it should offload the 802.1X handshake. However, this field
was existing and used for a different offload API. This causes
wpa_supplicant to send a connect request without the WANT_1X_HS flag and
the subsequent set-pmk is rejected causing the connection to fail. Fix
that by introducing a new flag req_handshake_offload so the offloads are
no longer entangled.
Fixes: d896874f8689 ("nl80211: Indicate 802.1X 4-way handshake offload in connect") Reported-by: Stefan Wahren <wahrenst@gmx.net> Tested-by: Stefan Wahren <wahrenst@gmx.net> Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Jouni Malinen [Thu, 1 Aug 2019 12:44:22 +0000 (15:44 +0300)]
tests: Make scan_bss_expiration_count more robust with UML
Test case sequence "persistent_group_channel scan_bss_expiration_count"
was failing with UML when using time travel. This seemed to be because
there was no explicit wait to confirm that the AP has been fully
disabled before running the next scan. Work around this by verifying
that hostapd has terminated the BSS and waiting a bit after that before
proceeding.
EAP-SIM/AKA server: Allow pseudonym/fast reauth to be disabled
The new hostapd configuration option eap_sim_id can now be used to
disable use of pseudonym and/or fast reauthentication with EAP-SIM,
EAP-AKA, and EAP-AKA'.
EAP-SIM/AKA: Do not allow anonymous@realm "pseudonym" to be cleared
If the EAP-SIM/AKA server does not provide a new pseudonym and the
locally configured "pseudonym" in anonymous_identity is actually an
anonymous identitity instead of a real EAP-SIM/AKA pseudonym, do not
clear the anonymous_identity network profile parameter. This is needed
to avoid forgetting the anonymous identity when going through
EAP-SIM/AKA authentication and then reverting back to using IMSI-based
(e.g., encrypted) identity.
Fixes: 4df4133917ab ("EAP-SIM/AKA: Add support for anonymous@realm") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Brian Norris [Wed, 31 Jul 2019 01:09:33 +0000 (18:09 -0700)]
D-Bus: Demote timeout/flush messages to MSG_MSGDUMP
We intentionally don't emit property-changed signals on every property
update -- for "less timing critical" messages we delay up to 5
milliseconds waiting to see if we can batch them together. When the
timer hits, we emit the signal anyway and (potentially) log this
message. This amounts to effectively tracing every property update,
which can be quite excessive.
Lower this to MSGDUMP, so MSG_DEBUG can remain slightly more sane.
Signed-off-by: Brian Norris <briannorris@chromium.org>
Brian Norris [Tue, 16 Jul 2019 23:43:21 +0000 (16:43 -0700)]
nl80211: Missing sysctl flags aren't fatal
The relevant flags were only added in Linux 4.6, so we shouldn't
complain because they're missing. Also, they're always missing if a
device is being removed (e.g., 'iw dev wlan0 del', or if the device is
in the process of resetting itself). So kill those 2 birds with 1 stone:
if we can't find the file, just silently skip it.
Also, we probably should *actually* propagate the error if we had a
write failure.
Signed-off-by: Brian Norris <briannorris@chromium.org>
Terry Burton [Sun, 21 Jul 2019 12:05:56 +0000 (13:05 +0100)]
Extra RADIUS request attributes from SQLite
Add an SQLite table for defining per station MAC address version of
radius_auth_req_attr/radius_acct_req_attr information. Create the
necessary table and index where this doesn't exist. Select attributes
from the table keyed by station MAC address and request type (auth or
acct), parse and apply to a RADIUS message.
Add radius_req_attr_sqlite hostapd config option for SQLite database
file. Open/close RADIUS attribute database for a lifetime of a BSS and
invoke functions to add extra attributes during RADIUS auth and
accounting request generation.
Signed-off-by: Terry Burton <tez@terryburton.co.uk>
mka: Accept last two used MNs in Peers List of a received MKPDU
Previously, check for local MI,MN in a peer's Peers List accepted only
the cases that include the last used MN from an MKPDU sent by the local
device. This was problematic since it was possible to synchronize MKPDU
transmission between two devices in a way that made them always miss the
last MKPDU from the other device before filling in the Peers List.
Relax this matching requirement of "acceptably recent MN" to mean both
the last used MN and the one used just before it (i.e., copied from
either of the last two MKPDUs sent by the local device) are accepted.
While this might help in some real world scenarios in making the
protocol converge more quickly, the main help from this is to fix
consistent hwsim test cases failures in macsec_psk_ns when using UML
with time travel option which happened to practically guarantee the
inconvenient timing of MKPDU transmission/reception that ended up with
the MKPDU processing to see MI,MN with MN being the last used MN minus
1.
dragonfly: Disable use of groups using Brainpool curves
Disable groups that use Brainpool curves for now since they leak more
timing information due to the prime not being close to a power of two.
This removes use of groups 28, 29, and 30 from SAE and EAP-pwd.
This removes the separate rerun step from the parallel-vm.py processing
and instead, simply requeues the failed test cases into the same queue
that is used for the initial run. This is simpler and more efficient
since reruns start as soon as any VM is ready for processing them
instead of having to wait for all VMs to complete the first round.
Furthermore, this allows VMs to be stopped sooner when no more test
cases remain and that is helpful especially with the time travel patches
that make the wait-for-next-test step in the VM use all available CPU.
nl80211: Add WMM parameters while updating TDLS peer entry
The AP mode fix for removing NL80211_ATTR_STA_WME from
NL80211_CMD_SET_STATION did not consider the TDLS case and that resulted
in incorrectly removing WMM parameters from TDLS STA entry updates. Fix
this by considering the WPA_STA_TDLS_PEER flag similarly to the other
update parameters.
Fixes: 6d14b98fc6fc ("nl80211: Do not add WMM parameters when updating an existing STA entry") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Avoid nested enum wpas_mode declaration to allow C++ compilation
Move enum wpas_mode declaration to the global scope to avoid issues with
the recently added inline function wpas_mode_to_ieee80211_mode() using
it as an argument. This fixes C++ compilation issues with cases that
include wpa_supplicant_i.h.
Fixes: 3459c54ac78b ("mesh: Add support for HE mode") Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
Report WPA/RSN protocol and AKM suite selector in STA MIB
The new "wpa" and "AKMSuiteSelector" entries in hostapd "STA <addr>"
control interface output can be used to determine the negotiated WPA/RSN
protocol and AKM suite of an associated station.
These SAE, OWE, DPP, and OSEN AKM suite selectors were covered in the
reverse conversion in rsn_key_mgmt_to_bitfield(), but were missing from
wpa_akm_to_suite(). Add them to make AKM suite selector reporting more
accurate in RADIUS Accounting messages (and future users of this
function).
dragonfly: SAE/EAP-pwd min PWE derivation iteration count to shared code
Use a shared function to determine the k parameter, i.e., the minimum
number of iterations of the PWE derivation loop, for SAE and EAP-pwd.
This makes it easier to fine-tune the parameter based on the negotiated
group, if desired.
EAP-pwd peer: Configurable set of groups with reduced default
Make the EAP-pwd peer use same default set of allowed groups as the SAE
implementation in wpa_supplicant uses, i.e., the groups 19-21 using NIST
curves P-256, P-384, and P-521. Previously, all groups that were
supported by the crypto library were allowed. In practice, this change
disables use of the Brainpool curves (groups 28-30) with recent OpenSSL
versions.
The default set of groups can be overridden with a new phase1 network
profile parameter, eap_pwd_groups=<list of allowed ranges>. For example,
phase1="eap_pwd_groups=0-65535" would restore previous behavior of
allowing all implemented groups to be used while eap_pwd_groups=19,20
would enable only the groups using NIST curves P-256 and P-384 to be
used.
tests: Prepare EAP-pwd test cases for allowed group configuration
Enable all supported groups in the existing ap_wpa2_eap_pwd_groups and
ap_wpa2_eap_pwd_invalid_group test cases to maintain current testing
functionality once wpa_supplicant is modified to use a different default
for the enabled groups.
Srinivas Dasari [Tue, 4 Jun 2019 07:20:17 +0000 (12:50 +0530)]
Add a vendor attribute to configure disconnect IEs
Add a new vendor attribute QCA_WLAN_VENDOR_ATTR_DISCONNECT_IES
to configure disconnect IEs to the driver. Driver shall fill
these IEs in disassoc/deauth frame.
These IEs are expected to be considered only for the next
immediate disconnection (disassoc/deauth frame) originated by
the DUT, irrespective of the entity (user space/driver/firmware)
triggering the disconnection.
The host drivers are not expected to use the IEs set through
this interface for further disconnections after the first immediate
disconnection initiated post the configuration.
If the IEs are also updated through cfg80211 interface (after the
enhancement to cfg80211_disconnect), host driver is expected to
take the union of IEs from both of these interfaces and send in
further disassoc/deauth frames.
Paul Zhang [Fri, 21 Jun 2019 06:11:11 +0000 (14:11 +0800)]
Add QCA vendor command to support OEM data
Add a QCA vendor sub command QCA_NL80211_VENDOR_SUBCMD_OEM_DATA
with attributes qca_wlan_vendor_attr_oem_data_params to support
OEM data. It is used to send OEM data binary blobs from
application/service to firmware. The attributes defined in enum
qca_wlan_vendor_attr_oem_data_params are used to deliver the
parameters.
OpenSSL: Fix TLS_CONN_TEAP_ANON_DH build with some library versions
The OPENSSL_VERSION_NUMBER ifdef block left out the local variable that
is needed with all versions. In addition, SSL_set_security_level() is
not available with LibreSSL or BoringSSL.
Fixes: 3ec65a8e38a0 ("OpenSSL: Allow anon-DH cipher suites to be added for TEAP") Signed-off-by: Jouni Malinen <j@w1.fi>
OpenSSL: Fix build with OpenSSL 1.0.2 and 1.1.0 and LibreSSL
The tls_connection_get_cipher_suite() implementation used
SSL_CIPHER_get_protocol_id which was added in OpenSSL 1.1.1. Need to use
compatibility code with older versions.
EAP-TLS server: Add application data to indicate end of v1.3 handshake
This adds an encrypted version of a one octet application data payload
to the end of the handshake when TLS v1.3 is used to indicate explicit
termination of the handshake (either after Finished message or after the
optional NewSessionTicket message). The current
draft-ietf-emu-eap-tls13-05 defines this to be a zero length payload,
but since that is not allowed by OpenSSL, use a one octet payload
instead for now with hopes of getting the draft specification updated
instead of having to modify OpenSSL for this.
EAP-TLS peer: Handle possible application data at the end
EAP-TLS with TLS 1.3 uses an empty application data record from the
server to indicate end of the exchange, so EAP-TLS peer will need to
check for this special case and finish the exchange with an empty
EAP-TLS (ACK) so that the server can send out EAP-Success.
This will hopefully not be needed for EAP-TLS use cases since there
should not really be a middlebox that looks at the TLS layer details in
case of EAP authentication.
OpenSSL: Allow two server certificates/keys to be configured on server
hostapd EAP server can now be configured with two separate server
certificates/keys to enable parallel operations using both RSA and ECC
public keys. The server will pick which one to use based on the client
preferences for the cipher suite (in the TLS ClientHello message). It
should be noted that number of deployed EAP peer implementations do not
filter out the cipher suite list based on their local configuration and
as such, configuration of alternative types of certificates on the
server may result in interoperability issues.
Ervin Oro [Mon, 15 Apr 2019 17:05:49 +0000 (20:05 +0300)]
Add Type-Code context to EAP-TLS 1.3 exported Key_Material and Method-Id
Change to require the Type-Code in context for Key_Material and
Method-Id has now been published as draft-ietf-emu-eap-tls13-04.
https://tools.ietf.org/html/draft-ietf-emu-eap-tls13-04#section-2.3
EAP-TEAP server and peer implementation (RFC 7170)
This adds support for a new EAP method: EAP-TEAP (Tunnel Extensible
Authentication Protocol). This should be considered experimental since
RFC 7170 has number of conflicting statements and missing details to
allow unambiguous interpretation. As such, there may be interoperability
issues with other implementations and this version should not be
deployed for production purposes until those unclear areas are resolved.
This does not yet support use of NewSessionTicket message to deliver a
new PAC (either in the server or peer implementation). In other words,
only the in-tunnel distribution of PAC-Opaque is supported for now. Use
of the NewSessionTicket mechanism would require TLS library support to
allow arbitrary data to be specified as the contents of the message.
OpenSSL: Allow anon-DH cipher suites to be added for TEAP
Add a new TLS_CONN_* flag to provide a higher level mechanism for adding
(instead of fully replacing) allowed list of TLS ciphersuites for TEAP
provisioning purposes.
Silence static analyzer warning about null termination of a string
The buf[] array is initialized to zeros, so it was already null
terminated since the read() call did not allow the last character of the
buffer to be overwritten. Since that was apparently not enough to make
some static analyzers understand the design, use explicit null
termination after a successful read() call.
Now that the TLS peer_cert information is provided as a full struct to
handler functions, the altsubject pointer shows up as an array and
causes static analyzers to warn about unnecessary NULL comparison. Get
rid of that comparison now that it is clearly not needed anymore.
Jouni Malinen [Mon, 24 Jun 2019 20:02:51 +0000 (23:02 +0300)]
EAP-pwd: Run through prf result processing even if it >= prime
This reduces differences in timing and memory access within the
hunting-and-pecking loop for ECC groups that have a prime that is not
close to a power of two (e.g., Brainpool curves).
Jouni Malinen [Mon, 24 Jun 2019 20:01:06 +0000 (23:01 +0300)]
SAE: Run through prf result processing even if it >= prime
This reduces differences in timing and memory access within the
hunting-and-pecking loop for ECC groups that have a prime that is not
close to a power of two (e.g., Brainpool curves).
projects / thirdparty / hostap.git / log
