Freeing memory for subjectAltName in parse_cert(), will give cert_cb
pointers to freed memory zone that may already been overwritten. Memory
for subjectAltName is released in parse_cert_free().
Signed-off-by: Cedric Izoard <cedric.izoard@ceva-dsp.com>
The "Accepting Additional Mesh Peerings bit == 0" means the peer cannot
accept any more peers, so suppress attempt to open a connection to such
a peer.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This moves pairwise, group, and management group ciphers to various mesh
data structures to avoid having to hardcode cipher in number of places
through the code. While CCMP and BIP are still the hardcoded ciphers,
these are now set only in one location.
Signed-off-by: Jouni Malinen <j@w1.fi>
The AMPE element includes number of optional and variable length fields
and those cannot really be represented by a fixed struct
ieee80211_ampe_ie. Remove the optional fields from the struct and
build/parse these fields separately.
This is also adding support for IGTKdata that was completely missing
from the previous implementation. In addition, Key RSC for MGTK is now
filled in and used when configuring the RX MGTK for a peer.
Signed-off-by: Jouni Malinen <j@w1.fi>
The previous implementation was incorrect in forcing the MGTK to be used
as the IGTK as well. Define new variable for storing IGTK and use that,
if set, to configure IGTK to the driver. This commit does not yet fix
AMPE element parsing to fill in this information.
Signed-off-by: Jouni Malinen <j@w1.fi>
This extends the data structures to allow variable length MGTK to be
stored for RX. This is needed as an initial step towards supporting
different cipher suites.
Signed-off-by: Jouni Malinen <j@w1.fi>
This is initial step in fixing issues in how PMF configuration for RSN
mesh was handled. PMF is an optional capability for mesh and it needs to
be configured consistently in both hostapd structures (to get proper
RSNE) and key configuration (not included in this commit).
Signed-off-by: Jouni Malinen <j@w1.fi>
Commit 0603bcb7fe ('hostapd: Process MAC
ACLs on a station association event (SME in driver)') processes MAC ACL
on a station association event for drivers which use AP SME offload but
does not consider the scenario where the drivers offload ACL. This can
result in station disconnection, though the driver accepts the
connection. Address this by avoiding the hostapd ACL check for the
drivers offloading MAC ACL.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This commit also introduces a new attribute MANDATORY_FREQUENCY_LIST
which aims for AP operation in a channel that ensures best concurrency
sessions.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
eloop deinit calls could trigger segmentation fault if the early error
path is hit before eloop_init() gets called.
Signed-off-by: Jouni Malinen <j@w1.fi>
Verify that fst_global_init() has been called before deinitializing the
global FST context. This makes it a bit easier to handle failure paths
from initialization.
Signed-off-by: Jouni Malinen <j@w1.fi>
Set max peer links to kernel even when wpa_supplicant MPM is used. This
sets the correct value for the "Accepting Additional Mesh Peerings bit"
in "Mesh Capability field" in "Mesh Configuration element" in the Beacon
frame.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
Recent versions of engine_pkcs11 are set up to be autoloaded on demand
with ENGINE_by_id() because they don't need explicit configuration.
But if we *do* want to explicitly configure them with a PKCS#11 module
path, we should still do so.
We can't tell whether it was already initialised, but it's harmless to
repeat the MODULE_PATH command if it was.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Tested-by: Michael Schaller <misch@google.com>
Any data accessible from nla_data() is freed before the
send_and_recv_msgs() function returns, therefore we need to allocate
space for info.flags ourselves.
Signed-off-by: Paul Stewart <pstew@google.com>
gas_address3=1 can now be used to force hostapd to use the IEEE 802.11
standards compliant Address 3 field value (Wildcard BSSID when not
associated) even if the GAS request uses non-compliant address (AP
BSSID).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Previously all TX status events with wildcard BSSID were ignored. This
did not allow Public Action frame TX status to be processed with the
corrected wildcard BSSID use. Fix this to be allowed. In practice, this
affects only test cases since Action frame TX status was not used for
anything else.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
IEEE Std 802.11-2012, 10.19 (Public Action frame addressing) specifies
that the wildcard BSSID value is used in Public Action frames that are
transmitted to a STA that is not a member of the same BSS. hostapd used
to use the actual BSSID value for all such frames regardless of whether
the destination STA is a member of the BSS.
Fix this by using the wildcard BSSID in cases the destination STA is not
a member of the BSS. Leave group addressed case as-is (i.e., the actual
BSSID), since both values are accepted. No such frames are currently
used, though.
This version is still using the AP BSSID value in the Address 3 field
for GAS response frames when replying to a GAS request with AP BSSID
instead of Wildcard BSSID. This is left as a workaround to avoid
interoperability issues with deployed STA implementations that are still
using the non-compliant address and that might be unable to process the
standard compliant case.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This adds the necessary changes to support extraction and use of the
extended capabilities specified per interface type (a recent
cfg80211/nl80211 extension). If that information is available,
per-interface values will be used to override the global per-radio
value.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Add a new "timeout" argument to the event message if the nl80211 message
indicates that the connection failure is not due to an explicit AP
rejection message. This makes it easier for external programs to figure
out why the connection failed.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This allows the simplest case of SAE group negotiation to occur by
selecting the next available group if the peer STA indicates the
previous one was not supported. This is not yet sufficient to cover all
cases, e.g., when both STAs need to change their groups, but at least
some cases are no covered.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
hostapd deinit functions were not ready to handle a case where the data
structures were not fully initialized. Make these more robust to allow
wpa_supplicant mesh implementation to use the current deinit design in
OOM error cases without causing NULL pointer dereferences.
Signed-off-by: Jouni Malinen <j@w1.fi>
This avoids internal access of structs and also removes the dependency
on the reimplemented TLS PRF functions when EAP-FAST support is not
enabled. Notably, BoringSSL doesn't support EAP-FAST, so there is no
need to access its internals with openssl_get_keyblock_size().
Signed-Off-By: David Benjamin <davidben@google.com>
Most protocols extracting keys from TLS use RFC 5705 exporters which is
commonly implemented in TLS libraries. This is the mechanism used by
EAP-TLS. (EAP-TLS actually predates RFC 5705, but RFC 5705 was defined
to be compatible with it.)
EAP-FAST, however, uses a legacy mechanism. It reuses the TLS internal
key block derivation and derives key material after the key block. This
is uncommon and a misuse of TLS internals, so not all TLS libraries
support this. Instead, we reimplement the PRF for the OpenSSL backend
and don't support it at all in the GnuTLS one.
Since these two are very different operations, split
tls_connection_prf() in two. tls_connection_export_key() implements the
standard RFC 5705 mechanism that we expect most TLS libraries to
support. tls_connection_get_eap_fast_key() implements the
EAP-FAST-specific legacy mechanism which may not be implemented on all
backends but is only used by EAP-FAST.
Signed-Off-By: David Benjamin <davidben@google.com>
Commit 68ae4773a4 ('OpenSSL: Use library
wrapper functions to access cert store') fixed most of these, but missed
a few.
Signed-Off-By: David Benjamin <davidben@google.com>
This makes changes such that hostapd (and wpa_supplicant AP mode)
registers to kernel for specific Action frames instead of generically
registering for all Action frames. This makes it easier for other
programs to register for some Action frames that hostapd does not handle
today without having to somehow coordinate directly with hostapd.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Add P2PS config flag only when config_methods are set. This restores the
pre-P2PS behavioer for the cases where Display or Keypad config method
is specified for a peer (i.e., do not add the new P2PS method in that
case).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
eap_proto_sake_errors test case where the result was used after the
failed eap_sake_compute_mic() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
eap_proto_pax_errors test case where the result was used after the
failed eap_pax_mac() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
eap_proto_fast_errors test case where the result was used after the
failed sha1_t_prf() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
wpas_ctrl_error test case where the result was used after the failed
sha256_vector() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
hostapd_oom_wpa2_eap_connect test case where the result is used after
failed md5_vector() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
hostapd_oom_wpa2_eap_connect test case where memcmp is used after failed
hmac_md5() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This gets rid of a valgrind warning on uninitialized memory read in the
hostapd_oom_wpa2_eap test case where memcmp is used after failed
md5_vector() call.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Check for hmac_sha256() failures and exit from wps_derive_psk() without
printing out the derived keys if anything fails. This removes a valgrind
warning on uninitialized value when running the ap_wps_m3_oom test case.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
In some terminal, verified with gtkterm and teraterm, backspace key is
not properly processed. For instance, type 'abc', 3 times of backspace
key press then '123' shows the result of 'abc123' instead of '123'. To
fix this, add a routine to process '\b' character input when using
edit_simple.c instead of edit.c (i.e., without CONFIG_WPA_CLI_EDIT=y).
Signed-off-by: Siwon Kang <kkangshawn@gmail.com>
If there's ever a driver that, like nl80211, requires radiotap,
we need to have a NEED_RADIOTAP variable to avoid trying to link
the radiotap helpers twice. Introduce that.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Commit 4104267e81 ('Fix memory leak on NFC
DH generation error path') modified the generic (non-OpenSSL)
implementation of dh5_init() to free the previously assigned public key,
if any. However, that commit did not modify the OpenSSL specific version
of this function. Add the same change there to maintain consistent
behavior between these two implementations of the same function.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Commit 4104267e81 ('Fix memory leak on NFC
DH generation error path') modified dh5_init() behavior in the
non-OpenSSL implementation to free the public key (if any was previously
set). However, this did not update one of the callers to make sure the
publ argument in the call is initialized. This could result in trying to
free invalid pointer and segmentation fault when hostapd or
wpa_supplicant was built against some other crypto library than OpenSSL.
Signed-off-by: Rujun Wang <chinawrj@gmail.com>
BoringSSL added OpenSSL 1.1.0's SSL_get_client_random() and friends in
working towards opaquifying the SSL struct. But it, for the moment,
still looks more like 1.0.2 than 1.1.0 and advertises
OPENSSL_VERSION_NUMBER as such. This means that there is no need to
define those in BoringSSL and defining them causes conflicts. (C does
not like having static and non-static functions with the same name.)
As requested, this is conditioned on defined(BORINGSSL_API_VERSION) so
wpa_supplicant may continue to support older BoringSSLs for a time.
(BoringSSL revisions without the accessors predate BoringSSL maintaining
a BORINGSSL_API_VERSION.)
Also add a missing opensslv.h include. tls_openssl.c is sensitive to
OPENSSL_VERSION_NUMBER, so it should include the header directly rather
than rely on another header to do so.
Signed-off-by: David Benjamin <davidben@google.com>
Spurious newlines output while writing the config file can corrupt the
wpa_supplicant configuration. Avoid writing these for the network block
parameters. This is a generic filter that cover cases that may not have
been explicitly addressed with a more specific commit to avoid control
characters in the psk parameter.
Signed-off-by: Paul Stewart <pstew@google.com>
