ndisc_snoop: Avoid misaligned read of IPv6 address
The IPv6 address in the frame buffer may not be 32-bit aligned, so use a
local copy to align this before reading the address with 32-bit reads
(s6_addr32[]).
trace: Use explicit alignment requirements to avoid misalignment
64-bit builds with CONFIG_WPA_TRACE=y resulted in the wpabuf pointers
getting misaligned (only 32-bit aligned) and that would result in reads
and writes of unaligned size_t values. Avoid this by indicating explicit
alignment requirement for wpabuf_trace to 8 octets (i.e., there will be
extra four octets of padding in case of 64-bit builds).
Similarly, struct os_alloc_trace resulted in some potential misalignment
cases, e.g., when CONFIG_ACS=y uses a 'long double' variable within
struct hostapd_channel_data. Avoid misalignment issues with explicit
alignment indication.
src/utils/list.h ended up defining a local version of offsetof() due to
stddef.h not getting included. This resulted in unnecessary warnings
from ubsan related to "dereferencing" of a NULL pointer.
Commit 38279bdb35c1b0c248289752197ebddd2fdd52ef ('D-Bus: Coding style
cleanup') handled the dbus_new_handlers.c change properly, but misplaced
the second 'const' in dbus_old_handlers.c in a way that resulted in
duplicated const rather than marking the actual value const.
The FT-specific check for valid group cipher in wpa_ft_gen_req_ies() was
not up-to-date with the current list of supported ciphers. Fix this by
using a generic function to determine validity of the cipher. In
practice, this adds support for using CCMP-256 and GCMP-256 as the group
cipher with FT.
Jouni Malinen [Tue, 30 Jun 2015 18:40:48 +0000 (21:40 +0300)]
TLS: Remove unused tls_capabilities()
This mechanism to figure out TLS library capabilities has not been used
since commit fd2f2d0489635d590930bc0945fbc438ba1387e2 ('Remove
EAP-TTLSv1 and TLS/IA') (Sep 2011).
Jouni Malinen [Sun, 28 Jun 2015 18:35:43 +0000 (21:35 +0300)]
Add backtrace-based error path testing mechanism
The new TEST_FAIL and GET_FAIL control interface commands can be used
similarly to the earlier TEST_ALLOC_FAIL/GET_ALLOC_FAIL design. The new
version is more generic framework allowing any function to be annotated
for failure testing with the TEST_FAIL() macro. This mechanism is only
available in builds with CONFIG_WPA_TRACE_BFD=y and
CONFIG_TESTING_OPTIONS=y. For other builds, the TEST_FAIL() macro is
defined to return 0 to allow the compiler to remove the test code from
normal production builds.
As the first test site, allow os_get_random() to be marked for failing
based on call backtrace.
Jouni Malinen [Sun, 28 Jun 2015 15:14:58 +0000 (18:14 +0300)]
P2P: Do not allow 40 MHz co-ex PRI/SEC switch to force MCC
Do not allow 40 MHz co-ex PRI/SEC switch to force us to change our PRI
channel if we have an existing connection on the selected PRI channel
since doing multi-channel concurrency is likely to cause more harm than
using different PRI/SEC selection in environment with multiple BSSes on
these two channels with mixed 20 MHz or PRI channel selection.
Stepanov, Max [Tue, 23 Jun 2015 13:47:59 +0000 (13:47 +0000)]
P2PS: Fix attribute addition in p2p_buf_add_service_instance()
Fix a condition when Advertised Service Info Attribute is added to
a probe response in p2p_buf_add_service_instance(). The issue is
that a 'found' value is increased even if 'test' and 'adv->hash' hashes
are different. As result 'found' may have a non-zero value when an
attribute data length is 0. In this cause an empty attribute is about to
be added. Fixing it by eliminating 'found' and checking 'total_len'
containing a real number of bytes added to Advertised Service Info
Attribute.
Stepanov, Max [Tue, 23 Jun 2015 13:52:22 +0000 (13:52 +0000)]
P2PS: Fix p2p_find last parameter handling
In p2p_find command line processing a loop searching for multiple
'seek=' parameters modifies cmd buffer adding '\0' terminators.
The 'freq=' parameter is handled after that and can be
lost if a 'freq=' follows 'seek=' in a command line.
Fix it by moving a handling of 'freq=' parameter to be processed
before 'seek=' handling loop.
Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Jouni Malinen [Sat, 27 Jun 2015 20:34:49 +0000 (23:34 +0300)]
AP: Increase maximum value accepted for cwmin/cwmax
The cwmin/cwmax parameters were limited more than is needed. Allow the
full range (0..15 for wmm_ac_??_{cwmin,cwmax} and 1..32767 for
tx_queue_data?_{cwmin,cwmax}) to be used.
Jouni Malinen [Sat, 27 Jun 2015 18:20:14 +0000 (21:20 +0300)]
SAE: Reject FFC commit-element with value p-1
The current P802.11 description of SAE uses "1 < element < p" as the
required range. However, this is not correct and does not match the
Dragonfly description of "1 < element < p-1". SAE definition will need
to change here. Update the implementation to reject p-1 based on the
correct rule here.
Jouni Malinen [Sat, 27 Jun 2015 14:10:19 +0000 (17:10 +0300)]
P2PS: Do not reply to ProbeReq on another channel when starting Listen
It was possible for the P2PS Probe Response frame to go out on the
channel on which a Probe Request frame was reported even when we are
just about to start Listen mode on another channel. This could result in
the peer device using incorrect channel for us. Fix this by skipping the
response in this special case while waiting for Listen mode to start.
This showed up as a hwsim test failure with test sequence "gas_fragment
p2ps_connect_display_method_nonautoaccept" in cases where the dev[0]
Listen channel was not same as the AP operating frequency in the GAS
test.
Jouni Malinen [Sat, 27 Jun 2015 09:41:40 +0000 (12:41 +0300)]
SAE: Reject commit-scalar value 1
IEEE Std 802.11-2012 description of SAE does not require this, i.e., it
describes the requirement as 0 < scalar < r for processing the Commit
message. However, this is not correct and will be changes to 1 < scalar
< r to match the Dragonfly description so that a trivial secret case
will be avoided explicitly.
This is not much of an issue for the locally generated commit-scalar
since it would be very unlikely to get the value of 1. For Commit
message processing, a peer with knowledge of the password could
potentially force the exchange to expose key material without this
check.
Dmitry Shmidt [Fri, 19 Jun 2015 17:33:50 +0000 (10:33 -0700)]
Send CTRL-EVENT-NETWORK-NOT-FOUND if no suitable network was found
This provides more information to upper layer programs on what happens
with connection attempts in cases where the enabled networks are not
found in scan results.
Kevin Cernekee [Sat, 20 Jun 2015 00:22:30 +0000 (17:22 -0700)]
Android: Fix nl80211 build if BOARD_*_PRIVATE_LIB is unspecified
wpa_supplicant has stub functions if the external p2p symbols are
unavailable, but the build still fails if the
wpa_driver_nl80211_driver_cmd symbol is missing. Fix this by leaving the
function pointer NULL. This is safe because wpa_drv_driver_cmd() performs
a NULL check.
Signed-off-by: Kevin Cernekee <cernekee@google.com>
Kevin Cernekee [Sat, 20 Jun 2015 00:22:29 +0000 (17:22 -0700)]
Android: Rename ANDROID_P2P_STUB to ANDROID_LIB_STUB
If BOARD_HOSTAPD_PRIVATE_LIB is not used on an Android build, we will
need to replace both the p2p functions *and* wpa_driver_nl80211_driver_cmd
in order to successfully link. Let's make the name more generic so it is
more obvious what it is used for.
Suggested-by: Dmitry Shmidt <dimitrysh@google.com> Signed-off-by: Kevin Cernekee <cernekee@google.com>
Jouni Malinen [Fri, 26 Jun 2015 12:29:29 +0000 (15:29 +0300)]
OpenSSL: Handle EC_POINT_is_on_curve() error case
Even though this OpenSSL function is documented as returning "1 if point
if on the curve and 0 otherwise", it can apparently return -1 on some
error cases. Be prepared for that and check explicitly against 1 instead
of non-zero.
Jouni Malinen [Fri, 26 Jun 2015 08:44:22 +0000 (11:44 +0300)]
SAE: Use random "password" in extra hunting-and-pecking loops
If PWE is discovered before the minimum number of loops (k) is reached,
the extra iterations use a random "password" to further obfuscate the
cost of discovering PWE.
Jouni Malinen [Thu, 25 Jun 2015 08:35:39 +0000 (11:35 +0300)]
SAE: Add side-channel protection to PWE derivation with ECC
This replaces the earlier IEEE Std 802.11-2012 algorithm with the design
from P802.11-REVmc/D4.0. Things brings in a blinding technique for
determining whether the pwd-seed results in a suitable PWE value.
Jouni Malinen [Thu, 25 Jun 2015 19:17:28 +0000 (22:17 +0300)]
OpenSSL: Add support for Brainpool Elliptic Curves
This allows the IKE groups 27-30 (RFC 6932) to be used with OpenSSL
1.0.2 and newer. For now, these get enabled for SAE as configurable
groups (sae_groups parameter), but the new groups are not enabled by
default.
Jouni Malinen [Wed, 24 Jun 2015 17:47:08 +0000 (20:47 +0300)]
SAE: Increase security parameter k to 40 based on Dragonfly recommendation
draft-irtf-cfrg-dragonfly recommends implementation to set the security
parameter, k, to a value of at least 40. This will make PWE generation
take significantly more resources, but makes it more likely to hide
timing differences due to different number of loops needed to find a
suitable PWE.
Jouni Malinen [Thu, 25 Jun 2015 08:56:51 +0000 (11:56 +0300)]
tests: Try heavier SAE groups in sae_groups
This changes the sae_groups test case design to try with every group and
skip triggering test failure for the heavier ones that are likely to
fail in some VM setups under load. This provides more testing coverage
by not limiting the test based on lowest common setup.
Jouni Malinen [Tue, 23 Jun 2015 19:30:15 +0000 (22:30 +0300)]
SAE: Verify that own/peer commit-scalar and COMMIT-ELEMENT are different
This check explicitly for reflection attack and stops authentication
immediately if that is detected instead of continuing to the following
4-way handshake that would fail due to the attacker not knowing the key
from the SAE exchange.
OpenSSL 0.9.8 (and newer) includes SSL_CTX_get_app_data() and
SSL_CTX_set_app_data(), so there is no need to maintain this old
OPENSSL_SUPPORTS_CTX_APP_DATA backwards compatibility design.
Jouni Malinen [Tue, 23 Jun 2015 17:39:08 +0000 (20:39 +0300)]
libtommath: Fix mp_init_multi() stdarg use on error path
Previously, it would have been possible for va_end(args) to be called
twice in case mp_init() fails. While that may not cause issues on number
of platforms, that is not how va_start()/va_end() are supposed to be
used. Fix this by returning from the function without using va_end()
twice on the same va_list args.
Jouni Malinen [Tue, 23 Jun 2015 17:23:31 +0000 (20:23 +0300)]
wpa_gui: Initialize WpaGuiApp::w in the constructor
This gets rid of a static analyzer warning. The actual value for
WpaGuiApp::w will be set after the constructor has returned, so this
value was not really used uninitialized.
Jouni Malinen [Tue, 23 Jun 2015 16:11:35 +0000 (19:11 +0300)]
Use unsigned/signed printf format more consistently
These configuration parameters did not use matching printf format string
parameters (signed vs. unsigned). While these configuratin values are,
in practice, small unsigned integers, the implementation should use
matching types to write these.
Jouni Malinen [Tue, 23 Jun 2015 15:30:11 +0000 (18:30 +0300)]
P2P: Use offsetof() instead of local implementation
The construction used here to figure out the offset of variable length
IEs in Probe Request frames was a bit odd looking and resulted in a
warning from a static analyzer, so replace it with more standard use of
offsetof().
Jouni Malinen [Tue, 23 Jun 2015 15:25:35 +0000 (18:25 +0300)]
ERP server: Make erp_send_finish_reauth() easier for static analyzers
The flags argument is used to indicate a failure case (0x80) which
allows erp == NULL. This may be a bit too difficult combination for
static analyzers to understand, so add an explicit check for !erp as
another condition for returning from the function before the erp pointer
gets dereferenced without checking it.
Jouni Malinen [Tue, 23 Jun 2015 15:18:25 +0000 (18:18 +0300)]
Remove redundant NULL check in ieee802_1x_encapsulate_radius()
The eap argument to this function is never NULL and the earlier
ieee802_1x_learn_identity() call is dereferencing it anyway, so there is
no point in checking whether it is NULL later in the function.
Jouni Malinen [Tue, 23 Jun 2015 08:21:51 +0000 (11:21 +0300)]
AP: Add more 2.4 GHz channels for 20/40 MHz HT co-ex scan
This needs to find the PRI channel also in cases where the affected
channel is the SEC channel of a 40 MHz BSS, so need to include the
scanning coverage here to be 40 MHz from the center frequency. Without
this, it was possible to miss a neighboring 40 MHz BSS that was at the
other end of the 2.4 GHz band and had its PRI channel further away from
the local BSS.