If the peer provides a username with large part of it being non-ASCII
characters, the previously used buffers may not have been long enough to
include the full string in debug logs and database search due to forced
truncation of the string by printf_encode(). Avoid this by increasing
the buffer sizes to fit in the maximum result.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
The off-by-one error in printf_encode() bounds checking could have
allowed buffer overflow with 0x00 being written to the memory position
following the last octet of the target buffer. Since this output is used
as \0-terminated string, the following operation would likely read past
the buffer as well. Either of these operations can result in the process
dying either due to buffer overflow protection or by a read from
unallowed address.
This has been seen to cause wpa_supplicant crash on OpenBSD when control
interface client attaches (debug print shows the client socket address).
Similarly, it may be possible to trigger the issue in RADIUS/EAP server
implementation within hostapd with a suitable constructed user name.
Signed-off-by: Stuart Henderson <sthen@openbsd.org>
If PMF was enabled, the validation step for EAPOL-Key descriptor version
ended up rejecting the message if GCMP had been negotiated as the
pairwise cipher. Fix this by making the GCMP check skipped similarly to
the CCMP case if a SHA256-based AKM is used.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Move to PTKINITDONE state and mark PTK valid after successful completion
of FT protocol. This allows the AP/Authenticator to start GTK rekeying
when FT protocol is used. Previously, the station using FT protocol did
not get the new GTK which would break delivery of group addressed
frames.
Signed-off-by: Jouni Malinen <j@w1.fi>
I got the below prints on a particular Android platform:
I/wpa_supplicant( 2637): nl80211: send_and_recv->nl_recvmsgs failed: 20
I/wpa_supplicant( 2637): nl80211: send_and_recv->nl_recvmsgs failed: 20
In JellyBean libnl_2 code, I see that the nl_recvmsgs returns postive values
too. In some cases, nl_recvmgs return the output of nl_recv function. nl_recv
function can return Number of bytes read, 0 or a negative error code.
Looks like this positive return value for nl_recvmsgs may be specific to
Android. While this is not how the API is supposed to work, this does no
harm with upstream libnl which returns only 0 or -1 from the function.
Signed-off-by: Jithu Jance <jithu@broadcom.com>
The dev_found() callback from NFC connection handover message processing
ended up using the p2p_dev_addr pointer that points to the parsed
message. However, that parsed data was freed just before the call. Fix
this by reordering the calls.
Signed-off-by: Jouni Malinen <j@w1.fi>
If the randomly generated bignum does not meet the validation steps, the
iteration loop in sae_get_rand() did not free the data properly. Fix the
memory leak by freeing the temporary bignum before starting the next
attempt at generating the value.
Signed-off-by: Jouni Malinen <j@w1.fi>
Couple of code paths in hostapd.c could have left hapd->drv_priv
pointing to memory that was freed in driver_nl80211.c when a secondary
BSS interface is removed. This could result in use of freed memory and
segfault when the next driver operation (likely during interface
deinit/removal). Fix this by clearing hapd->drv_priv when there is
reason to believe that the old value is not valid within the driver
wrapper anymore.
Signed-off-by: Jouni Malinen <j@w1.fi>
These three calls were used already in three different paths. Use a
helper function to avoid adding even more copies of this.
Signed-off-by: Jouni Malinen <j@w1.fi>
If for some reason interface setup fails mid-way when setting up
multi-BSS AP it was possible to get segmentation fault because driver
was not properly cleaned up.
One possible trigger, when using nl80211 driver, was udev renaming an
interface created by hostapd causing, e.g., linux_set_iface_flags() to
fail.
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
This variable is updated when calling hostapd_if_add(), so it makes
sense to do the same thing when calling hostapd_if_remove().
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Performing, e.g. `wpa_cli -p /var/run/hostapd raw DISABLE` twice led to
hostapd segmentation fault if multiple BSSes were configured. Fix this
by checking if there is anything to disable at all before trying.
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Make sure to not remove interfaces that were not created by
hostapd/wpa_supplicant. This was already done on number of the error
paths, but not all.
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
It is possible for the connect() call to fail (e.g., due to unreachable
network based on local routing table), so the current auth/acct_sock may
be left to -1. Use that as an addition trigger to allow server failover
operation to be performed more quickly if it is known that the
retransmission attempt will not succeed anyway.
Signed-off-by: Jouni Malinen <j@w1.fi>
The re-open socket to the current RADIUS server code path did not work
in the expected way here. The pending authentication messages do not
need to be flushed in that case and neither should the retransmission
parameters be cleared. Fix this by performing these operations only if
the server did actually change as a part of a failover operation.
Signed-off-by: Jouni Malinen <j@w1.fi>
The current RADIUS server pointer was updated after each SET command
which broke parsing of multiple RADIUS servers over the control
interface. Fix this by doing the final RADIUS server pointer updates
only once the full configuration is available.
Signed-off-by: Jouni Malinen <j@w1.fi>
Both IPv4 and IPv6 sockets were not closed consistently in the paths
that tried to change RADIUS servers. This could result in leaking
sockets and leaving behind registered eloop events to freed memory on
interface removal.
Signed-off-by: Jouni Malinen <j@w1.fi>
While iterating through RADIUS messages in radius_client_timer(), one
message entry may get flushed by "radius_client_retransmit -->
radius_client_handle_send_error --> radius_client_init_auth -->
radius_change_server --> radius_client_flush". This could result in
freed memory being accessed afterwards.
Signed-off-by: Jerry Yang <xyang@sonicwall.com>
This is one more possible send() error that should trigger RADIUS server
change if multiple servers are configured.
Signed-off-by: Jouni Malinen <j@w1.fi>
It is possible for the RADIUS authentication/accounting socket to not be
open even if partial RADIUS server configuration has been done through
the control interface SET commands. Previously, this resulted in send()
attempt using fd=-1 which fails with bad file descriptor. Clean this up
by logging this as a missing configuration instead of trying to send the
message when that is known to fail.
Signed-off-by: Jouni Malinen <j@w1.fi>
This makes invitation process more consistent with GO Negotiation as far
as the debug log entries are concerned and the resulting log is more
helpful for understanding channel selection.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Instead of using X509_print_fp() to print directly to stdout, print the
certificate dump to a memory BIO and use wpa_printf() to get this into
the debug log. This allows redirection of debug log to work better and
avoids undesired stdout prints when debugging is not enabled.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This configuration parameter was not used at all in the RADIUS server
implementation and instead, hard coded 0 was sent.
Signed-off-by: Jouni Malinen <j@w1.fi>
Commit 04eff7d5ba or something around that
timeframe may have caused a regression on how drv->if_indices gets used
with wpa_supplicant. Most (curretly likely all) wpa_supplicant virtual
interface use cases should not actually use this. This could result in
issues with P2P group interfaces delivering events to incorrect
interface (parent rather than the group interface). The previous commit
removed some of the issues, but more complete fix is to undo some of
those merged hostapd/wpa_supplicant operations.
Filter add_ifidx() uses based on hostapd vs. wpa_supplicant and iftype
to get closer to the earlier wpa_supplicant behavior for the driver
events from virtual interfaces.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
It is possible for a virtual interface to be added and removed by
different parent interfaces. This can happen, e.g., with P2P group
interfaces if the P2P parent interface does not happen to be the first
entry in the wpa_supplicant global interface list. That first entry is
used to remove the group interface while the addition would have
happened with the dedicated P2P management interface.
This can result in the interface that added a new virtual interface
getting stuck with an obsolete ifindex value in the drv->if_indeces list
and as such, deliver some extra events to incorrect destination wpa_s
instance. In particular, this can result in INTERFACE_DISABLED event
from deletion of a P2P group interface getting delivered incorrectly to
the parent wpa_s instance which would disable that interface even though
the interface remains in enabled state.
Fix this by clearing the removed interface from all if_indeces lists
instead of just the one that was used to delete the interface. This is
the simplest approach since the ifindex is unique and there is no need
to track which interface added the new virtual interface to always hit
the same one when removing the interface.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Add regulatory classes for the 60GHz band.
Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: Boris Sorochkin <qca_bsoroc@qca.qualcomm.com>
Instead of silently dropping the response that requires fragmentation,
send an error response to the station to make it aware that no full
response will be available.
Signed-off-by: Jouni Malinen <j@w1.fi>
If p2p_listen is called while previous listen command's
remain_on_channel event is pending, the p2p_listen would fail
and it used to clear pending_listen_freq. Now when the remain-
on-channel event comes from the driver, the pending_listen_freq
doesn't match and gets ignored. This was leading to a case
where listen state was getting stuck (in case of WAIT_PEER_CONNECT
state).
Signed-off-by: Jithu Jance <jithu@broadcom.com>
Previously, the NL80211_ATTR_AKM_SUITES was skipped if either of these
SHA256-based AKMs was negotiated.
Signed-off-by: Jithu Jance <jithu@broadcom.com>
This is similar to the existing functionality that parsed ASN.1-encoded
RSA public key by generating a similar public key instance from already
parsed n and e parameters.
Signed-off-by: Jouni Malinen <j@w1.fi>
Follow the PKCS #1 v1.5, 8.1 constraint of at least eight octets long PS
for the case where the internal TLS implementation decrypts PKCS #1
formatted data. Similar limit was already in place for signature
validation, but not for this decryption routine.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Based on PKCS #1, v1.5, 10.1.3, the block type shall be 01 for a
signature. This avoids a potential attack vector for internal TLS/X.509
implementation.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Verify that there is no extra data after the hash field. This is needed
to avoid potential attacks using additional data to construct a value
that passes the RSA operation and allows the hash value to be forged.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Previously, hostapd had to be started with at least one of the
configuration files enabling TNC for TNC to be usable. Change this to
allow TNC to be enabled when the first interface with TNC enabled gets
added during runtime.
Signed-off-by: Jouni Malinen <j@w1.fi>
This is for enabling easier testing of TNCS/TNCC functionality as part
of the test scripts without having to use the fixed /etc/tnc_config
location that could be used by the main system and would require changes
within /etc.
Signed-off-by: Jouni Malinen <j@w1.fi>
This patch adds epoll option for the eloop implementation. This can be
selected with the CONFIG_ELOOP_EPOLL=y build option.
[merit]
See Table1.
Table1. comparison table
+--------+--------+-----------+------------+-------------+
| | add fd | remove fd | prepare fd | dispatch fd |
+--------+--------+-----------+------------+-------------+
| select | O(1) | O(1) | O(N) | O(N) |
+--------+--------+-----------+------------+-------------+
| poll | O(1) | O(1) | O(N) | O(N) |
+--------+--------+-----------+------------+-------------+
| epoll | O(1) | O(1) | 0 | O(M) |
+--------+--------+-----------+------------+-------------+
"add fd" is addition of fd by eloop_sock_table_add_sock().
"remove fd" is removal of fd by eloop_sock_table_remove_sock().
"prepare fd" is preparation of fds before wait in eloop_run().
"dispatch fd" is dispatchment of fds by eloop_sock_table_dispatch().
"N" is all watching fds.
"M" is fds which could be dispatched after waiting.
As shown in Table1, epoll option has better performance on "prepare fd" column.
Because select/poll option requires setting fds before every select()/poll().
But epoll_wait() doesn't need it.
And epoll option has also better performance on "dispatch fd" column.
Because select/poll option needs to check all registered fds to find out
dispatchable fds. But epoll option doesn't require checking all registered fds.
Because epoll_wait() returns dispatchable fd set.
So epoll option is effective for GO/AP functionality.
[demerit]
The epoll option requires additional heap memory. In case of P2P GO, it is
about 8K bytes.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
