I'm not sure about the exact scenario causing failure, but using the
libc pthread stubs isn't sufficient.
fixes one of the glmark2 crashes (at startup with llvmpipe)
ones found in the same sources online (mostly, Intel, AMD, Nvidia
and ASMedia web sites.) this includes:
- bunch of xeon e5 v3 devices
- bunch of radeon chipsets
- bunch of nvidia chipsets
- bunch of marvell chipsets
- bunch of asmedia chipsets
be more consistent with device naming within some groups.
SPL_SERIAL (SPL_HIGH) wedge. Gets sbmips to ~multiuser (usually with
a bunch of segfaults).
Possibly a gross hack, but not sure the current code ever worked...
that is equal or greater than the prime and as such, need for going
through the PWE derivation loop multiple times. This can result in
sufficient timing different to allow an external observer to determine
how many rounds are needed and that can leak information about the used
password.
Force at least 40 loop rounds for these MODP groups similarly to the ECC
group design to mask timing. This behavior is not described in IEEE Std
802.11-2016 for SAE, but it does not result in different values (i.e.,
only different timing), so such implementation specific countermeasures
can be done without breaking interoperability with other implementation.
Note: These MODP groups 22, 23, and 24 are not considered sufficiently
strong to be used with SAE (or more or less anything else). As such,
they should never be enabled in runtime configuration for any production
use cases. These changes to introduce additional protection to mask
timing is only for completeness of implementation and not an indication
that these groups should be used.
This is related to CVE-2019-9494.
attacker, so try to minimize differences in how the
sae_test_pwd_seed_ecc() result is used. (CVE-2019-9494)
Use heap memory for the dummy password to allow the same password length
to be used even with long passwords.
Use constant time selection functions to track the real vs. dummy
variables so that the exact same operations can be performed for both QR
test results.
timing differences or memory access patterns (cache use). While the
previous implementation had protection against the most visible timing
differences (looping 40 rounds and masking the legendre operation), it
did not protect against memory access patterns between the two possible
code paths in the masking operations. That might be sufficient to allow
an unprivileged process running on the same device to be able to
determine which path is being executed through a cache attack and based
on that, determine information about the used password.
Convert the PWE finding loop to use constant time functions and
identical memory access path without different branches for the QR/QNR
cases to minimize possible side-channel information similarly to the
changes done for SAE authentication. (CVE-2019-9495)
operation. This is needed to avoid leaking information about different
temporary results in blinding mechanisms.
This is related to CVE-2019-9494 and CVE-2019-9495.
for various cryptographic operations that must minimize externally
observable differences in processing (both in timing and also in
internal cache use, etc.).
This is related to CVE-2019-9494 and CVE-2019-9495.
involving private information. BoringSSL has removed BN_FLG_CONSTTIME
and expects specific constant time functions to be called instead, so a
bit different approach is needed depending on which library is used.
The main operation that needs protection against side channel attacks is
BN_mod_exp() that depends on private keys (the public key validation
step in crypto_dh_derive_secret() is an exception that can use the
faster version since it does not depend on private keys).
crypto_bignum_div() is currently used only in SAE FFC case with not
safe-prime groups and only with values that do not depend on private
keys, so it is not critical to protect it.
crypto_bignum_inverse() is currently used only in SAE FFC PWE
derivation. The additional protection here is targeting only OpenSSL.
BoringSSL may need conversion to using BN_mod_inverse_blinded().
This is related to CVE-2019-9494 and CVE-2019-9495.
when trying to check SAE confirm message. It could have been possible to
hit a NULL pointer dereference if the peer element could not have been
parsed. (CVE-2019-9496)
2.8.5.2.2 requirement. The earlier checks might have covered this
implicitly, but it is safer to avoid any dependency on implicit checks
and specific crypto library behavior. (CVE-2019-9498 and CVE-2019-9499)
Furthermore, this moves the EAP-pwd element and scalar parsing and
validation steps into shared helper functions so that there is no need
to maintain two separate copies of this common functionality between the
server and peer implementations.
(elliptic curve point) were not validated. This allowed an adversary to
bypass authentication, and act as a rogue Access Point (AP) if the
crypto implementation did not verify the validity of the EC point.
Fix this vulnerability by assuring the received scalar lies within the
valid range, and by checking that the received element is not the point
at infinity and lies on the elliptic curve being used. (CVE-2019-9499)
The vulnerability is only exploitable if OpenSSL version 1.0.2 or lower
is used, or if LibreSSL or wolfssl is used. Newer versions of OpenSSL
(and also BoringSSL) implicitly validate the elliptic curve point in
EC_POINT_set_affine_coordinates_GFp(), preventing the attack.
and elliptic curve element differ from the one sent by the server. This
prevents reflection attacks where the adversary reflects the scalar and
element sent by the server. (CVE-2019-9497)
The vulnerability allows an adversary to complete the EAP-pwd handshake
as any user. However, the adversary does not learn the negotiated
session key, meaning the subsequent 4-way handshake would fail. As a
result, this cannot be abused to bypass authentication unless EAP-pwd is
used in non-WLAN cases without any following key exchange that would
require the attacker to learn the MSK.
kamil