This is a patch for OpenBSD wired IEEE 802.1X. This is only for wired,
not wireless, because OpenBSD uses wpa_supplicant only on wired now.
http://www.openbsd.org/cgi-bin/cvsweb/ports/security/wpa_supplicant/
I have tested with these.
OS : OpenBSD 4.5
EAP : EAP-TLS
Switch : CentreCOM 8724SL
Previously, both the command replies and unsolicited events were
received from the same socket. This could cause problems if an event
message is received between a command and the response to that command.
Using two sockets avoids this issue.
When Linux has Path MTU discovery enabled, it sets by default the DF bit
on all outgoing datagrams, also UDP ones. If a RADIUS message is bigger
than the smallest MTU size to the target, it will be discarded.
This effectively limits RADIUS messages to ~ 1500 Bytes, while they can
be up to 4k according to RFC2865. In practice, this can mean trouble
when doing EAP-TLS with many RADIUS attributes besides the EAP-Message.
[Bug 326]
When Linux has Path MTU discovery enabled, it sets by default the DF bit
on all outgoing datagrams, also UDP ones. If a RADIUS message is bigger
than the smallest MTU size to the target, it will be discarded.
This effectively limits RADIUS messages to ~ 1500 Bytes, while they can
be up to 4k according to RFC2865. In practice, this can mean trouble
when doing EAP-TLS with many RADIUS attributes besides the EAP-Message.
[Bug 326]
These could, at least in theory, be used to generate unexpected common
name or subject alternative name matches should a CA sign strings with
NUL (C string termination) in them. For now, just reject the certificate
if an embedded NUL is detected. In theory, all the comparison routines
could be made to compare these strings as binary blobs (with additional
X.509 rules to handle some exceptions) and display NUL characters
somehow. Anyway, just rejecting the certificate will get rid of
potential problems with the C string getting terminated and it should
not really be used in certificates, so this should not break valid use
cases.
The BLOCK_SIZE define can be made more specific by using AES_ prefix and
by moving it to aes.h. After this, most aes-*.c do not really need to
include anything from the internal aes_i.h header file. In other words,
aes_i.h can now be used only for the code that uses the internal AES
block operation implementation and none of the code that can use AES
implementation from an external library do not need to include this
header file.
Better not specify EVP_CIPHER again for the second init call since that
will override key length with the default value. The previous version
was likely to work since most use cases would be likely to use the
default key length. Anyway, better make this handle variable length
ciphers (mainly, RC4), too, just in case it is needed in the future.
This is not really of that much use since rc4_skip() can be used as
easily. In addition, rc4 has caused some symbol conflicts in the past,
so it is easier to live without that as an exported symbol.
wpa_supplicant can now be built with FIPS capable OpenSSL for FIPS mode
operation. Currently, this is only enabling the FIPS mode in OpenSSL
without providing any higher level enforcement in wpa_supplicant.
Consequently, invalid configuration will fail during the authentication
run. Proper configuration (e.g., WPA2-Enterprise with EAP-TLS) allows
the connection to be completed.
Instead of using low level, digest-specific functions, use the generic
EVP interface for digest functions. In addition, report OpenSSL errors
in more detail.
Some crypto libraries can return in these functions (e.g., if a specific
hash function is disabled), so we better provide the caller a chance to
check whether the call failed. The return values are not yet used
anywhere, but they will be needed for future changes.
It looks like GnuTLS does not know how to parse the previously used
DER-formatted PKCS#1 private key (server.key). To work around this, use
a PEM-formatted version of the same key. This format can now be used by
OpenSSL, GnuTLS, and the internal TLS implementation.
On NetBSD 5.0, when I use wired 802.1X, "Invalid argument" occurs
on SIOCADDMULTI ioctl and 802.1X fails.
I tried FreeBSD code, but "Address family not supported by protocol family"
occurs on SIOCADDMULTI ioctl and 802.1X fails, too.
This patch solves this issue.
I have tested with these:
OS : NetBSD 5.0
EAP : EAP-MD5
Switch : CentreCOM 8724SL
An SMC router was reported to use 0x22 (WPAPSK + WPA2PSK) in the
authentication type of the provisioned credential and wpa_supplicant
rejected this as invalid. Work around this by replacing WPAPSK + WPA2PSK
with WPA2PSK.
This is a (hopefully) temporary workaround to allow the same source code
tree to be used for building hostapd and wpa_supplicant without having
to manually force recompilation of some files. Currently, some of the
driver wrapper files need to be built separately for hostapd and
wpa_supplicant (#ifdef's in the files based on AP functionality).
This is somewhat racy as far as parallel make execution is concerned,
i.e., it may be necessary to run "make -j#" twice (plain "make" works
fine. Since this is supposed to be a temporary workaround, there is not
much point in trying to fix this with any more complex make processing.
Instead of having all driver stuff collected across wpa_supplicant
and hostapd, create a common snippet that they both include and
that handles the build configuration.
If base64_encode() were to be used with a huge data array, the
previous version could have resulted in overwriting the allocated
buffer due to an integer overflow as pointed out in
http://www.freebsd.org/cgi/query-pr.cgi?pr=137484. However, there
are no know use cases in hostapd or wpa_supplicant that would do that.
Anyway, the recommended change looks reasonable and provides additional
protection should the base64_encode() function be used for something
else in the future.
