Jouni Malinen [Thu, 2 Jan 2020 09:51:04 +0000 (11:51 +0200)]
Maintain BSS entries for 5 seconds after interface is disabled
This is targeting the case of MAC address change for an association
which may require the interface to be set down for a short moment.
Previously, this ended up flushing the BSS table that wpa_supplicant
maintained and that resulted in having to scan again if the MAC address
was changed between the previous scan and the connection attempt. This
is unnecessary extra latency, so maintain the BSS entries for 5 seconds
(i.e., the same time that the old scan results are consider valid for a
new connection attempt) after an interface goes down.
Matthew Wang [Tue, 17 Jul 2018 17:56:21 +0000 (10:56 -0700)]
Improve roaming logic
Currently, wpa_supplicant may roam too aggressively; the need_to_roam()
function will return early with a roaming decision if the difference in
signal level or throughput between the current and selected APs is
"sufficiently large." In particular, if the selected AP's estimated
throughput is more than 5k greater than the current AP's estimated
throughput, wpa_supplicant will decide to roam. Otherwise, if the
selected AP's signal level is less than the current AP's signal level,
or the selected AP's estimated throughput is at least 5k less than the
current AP's estimated throughput, wpa_supplicant will skip the roam.
These decisions are based only on one factor and can lead to poor
roaming choices (e.g., a roam should not happen if the selected AP's
estimated throughput meets the threshold but the current signal and
throughput are already good, whereas a roam should happen if the signal
is slightly worse but the estimated throughput is significantly better).
This change standardizes the roaming heuristic for signal strength
difference requirements and will hopefully improve user experience. The
change can be summarized as follows: based on the current signal level,
a certain roaming difficulty is assigned. Based on the selected AP's
estimated throughput relative to the current AP's estimated throughput,
the difficulty is adjusted up or down. If the difference in signal level
meets the threshold, a roam happens.
The hard-coded values were selected purely based on the previous version
of this function. They may eventually need to be fine-tuned for optimal
performance.
Signed-off-by: Matthew Wang <matthewmwang@chromium.org>
Jouni Malinen [Wed, 1 Jan 2020 15:09:18 +0000 (17:09 +0200)]
Allow roam to lower signal level if throughput benefit is significant
Do not prevent roam to a different BSS based only on the signal level
with the current BSS being higher than with the selected BSS. If the
estimated throughput is significantly higher (> 20%), allow roaming if
the following conditions are met.
Jouni Malinen [Wed, 1 Jan 2020 14:56:40 +0000 (16:56 +0200)]
Skip roaming based on signal level difference if current SNR is good
If the current SNR with the associated BSS is sufficiently good (better
than GREAT_SNR = 25), there is limited benefit from moving to another
BSS even if that BSS were to have a higher signal level. As such, skip
roaming based on the signal level difference between the selected BSS
from scan results and the current BSS for such cases.
Jouni Malinen [Wed, 1 Jan 2020 11:38:11 +0000 (13:38 +0200)]
Use signal_poll noise information for roaming, if available
Using average signal strength from the driver and hardcoded noise floor
does not look like an ideal design since there can be significant
differences in the driver-reported noise floor values. Furthermore, even
though the current noise floor is a snapshot from the driver, it is
common for drivers to use a noise floor value from a longer calibration
step and that should not prevent the driver provided value from being
used. This makes the comparisons of the signal strengths between the
current AP (signal_poll) and other APs (scan) more accurate.
As an example, test runs in home environment showed 5 dB difference
between the driver reported noise floor and the hardcoded value and this
could result in significant differences in estimated throughput
calculation.
Jouni Malinen [Wed, 1 Jan 2020 09:26:52 +0000 (11:26 +0200)]
RSN: Do not add PMKSA candidates unnecessarily
Add PMKSA candidates from scan results only if they advertise an AKMP
that is used with RSN pre-authentication. Previously, candidates were
added but then ignored later if the AKMP was not suitable.
Jouni Malinen [Wed, 1 Jan 2020 10:03:42 +0000 (12:03 +0200)]
Clear last Michael MIC error timer on FLUSH
TKIP countermeasures were already terminated on FLUSH, but the timer for
detecting two Michael MIC errors within 60 seconds was left behind. This
resulted in test case failures with following test sequence:
ap_cipher_tkip_countermeasures_sta ap_cipher_tkip_countermeasures_sta2
Matthew Wang [Fri, 6 Dec 2019 22:27:48 +0000 (14:27 -0800)]
wpa_supplicant: Fall back to avg_signal in roaming decision
Some drivers (e.g. Marvell WiFi) don't report avg_beacon_signal, but
it's still useful to poll for the signal again when a roaming decision
needs to be made. Use si.avg_signal when si.avg_beacon_signal is not
available.
Signed-off-by: Matthew Wang <matthewmwang@chromium.org>
Update throughput estimate for the current BSS based on signal poll
We saw that on certain platforms in certain places we keep switching
between two APs and eventually get the same RSSI. Debugging showed that
we have a very big difference between the two antennas.
Ant A can hear AP A very well (-60) but AP B very bad (-80)
Ant B can hear AP B very well (-60) but AP A very bad (-80)
When the device associates to AP A, it'll learn to use Ant A. If the
device uses one single antenna to receive the scan results, it may hear
the AP it is currently associated to on the second antenna and get bad
results. Because of that, the wpa_supplicant will roam to the other AP
and the same scenario will repeat itself:
Association to AP A (Ant A reports -60).
Scan on Ant A: AP A: -60, AP B: -80
Scan on Ant B: AP A: -80, AP A: -60 ==> ROAM.
Association to AP B (Ant B reports -60)
Scan on Ant A: AP A: -60, AP B: -80 ==> ROAM
Etc...
Improve this by querying the signal level of the current AP using
drv_signal_poll() instead of relying on the signal level that we get
from the scan results. Also update the throughput estimate based on the
likely more accurate values for the current association.
Jouni Malinen [Mon, 30 Dec 2019 17:32:28 +0000 (19:32 +0200)]
tests: Fix sigma_dut_sae_h2e_rsnxe_mismatch to clear sae_pwe
sae_pwe=1 could be left configured when exiting this test case since
sigma_dut does not guarantee the default value to be restored. This
could result in test case failures, e.g., in the following sequence:
sigma_dut_sae_h2e_rsnxe_mismatch sae_pwe_h2e_only_ap_sta_forcing_loop
Ethan Everett [Tue, 12 Feb 2019 22:20:04 +0000 (22:20 +0000)]
RADIUS client: fix extra retry before failover
This commit changes the failover behavior of RADIUS client. Commit 27ebadccfb2 ("RADIUS client: Cease endless retry for message for
multiple servers") changed the retry logic, causing RADIUS client to
wait RADIUS_CLIENT_NUM_FAILOVER + 1 timeouts before failing over the
first time. Prior to that commit, RADIUS client would wait
RADIUS_CLIENT_NUM_FAILOVER timeouts before each failover. This was
caused by moving the entry->attempts > RADIUS_CLIENT_NUM_FAILOVER
comparison to before the retry attempt, where entry->attempts is
incremented.
The commit in question set entry->attempts in radius_change_server to 1
instead of 0, so RADIUS client would still only wait
RADIUS_CLIENT_NUM_FAILOVER timeouts for subsequent failovers, the same
as the original behavior.
This commit changes the comparison so the initial failover now happens
after waiting RADIUS_CLIENT_NUM_FAILOVER timeouts, as it did originally.
It also changes the RADIUS_CLIENT_MAX_FAILOVER comparison to prevent an
additional attempt to the primary server after the final failover.
Jouni Malinen [Mon, 30 Dec 2019 17:12:33 +0000 (19:12 +0200)]
tests: radius_acct_unreachable2 to detect retransmissions earlier
It looks like this test case can start showing failures with a change in
the retransmission limit behavior for a server change. Check for
retransmissions every second instead of only at the end of the four
second wait to avoid this.
Ilan Peer [Wed, 22 Aug 2018 16:49:05 +0000 (19:49 +0300)]
tests: Extend the timeout for some SAE error case tests
Commit 407879b690ba ("mac80211: Adjust SAE authentication timeout") in
the kernel tree increased the SAE authentication timeout. This caused
some error case tests to fail. To fix this, extend the timeout for some
error case tests.
Jouni Malinen [Sun, 29 Dec 2019 21:24:10 +0000 (23:24 +0200)]
Call hostapd_allowed_address() directly from handle_probe_req()
ieee802_11_allowed_address() did not really do anything useful for the
call from handle_probe_req(), so replace that with a direct call to
hostapd_allowed_address() and make ieee802_11_allowed_address() a static
function.
Michael Braun [Sun, 28 Apr 2019 11:14:57 +0000 (13:14 +0200)]
Fix possible memory leak of RADIUS data in handle_auth()
When returning from handle_auth() after ieee802_11_allowed_address()
returned HOSTAPD_ACL_ACCEPT, but before ieee802_11_set_radius_info() has
been called, identity, radius_cui, and psk might not have been consumed.
Fix this by avoiding the need to free these variables at all.
Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
Jouni Malinen [Sun, 29 Dec 2019 18:38:40 +0000 (20:38 +0200)]
FT: Do not deliver RRB messages locally without matching FT/SSID
For FT protocol to work, the BSSs need to be operating an FT AKM with
the same SSID and mobility domain. The previous commit covered the
mobility domain, this covers the other prerequisites. This reduces
unnecessary load from having to allocate queued messages for interfaces
that cannot have valid data.
Jinglin Wang [Fri, 13 Dec 2019 08:30:28 +0000 (16:30 +0800)]
FT: Check mobility domain when sending RRB message to local managed BSS
Fast BSS Transition requires related APs operating in the same mobility
domain. Therefore, we can check whether the local managed BSS is
operating the same mobility domain before sending multicast/unicast
messages to it. This reduces unnecessary load from having to allocate
queued messages for interfaces that cannot have valid data.
Signed-off-by: Jinglin Wang <bryanwang@synology.com> Signed-off-by: MinHong Wang <minhongw@synology.com>
Jouni Malinen [Sun, 29 Dec 2019 17:32:29 +0000 (19:32 +0200)]
RRB: Do not reorder locally delivered messages
Add new messages to the end of the l2_oui_queue instead of inserting
them at the beginning so that the dl_list_for_each_safe() iteration in
hostapd_oui_deliver_later() goes through the messages in the same order
they were originally queued.
Jinglin Wang [Fri, 13 Dec 2019 08:30:27 +0000 (16:30 +0800)]
FT: Fix hostapd_wpa_auth_oui_iter() iteration for multicast packets
When using FT wildcard feature, the inter-AP protocol will send
broadcast messages to discover related APs.
For example,
12/6 16:24:43 FT: Send PMK-R1 pull request to remote R0KH address
ff:ff:ff:ff:ff:ff
12/6 16:24:43 FT: Send out sequence number request to
ff:ff:ff:ff:ff:ff
If you have multiple interfaces/BSSs in a single hostapd process,
hostapd_wpa_auth_oui_iter() returned 1 after the first interface was
processed. Iteration in for_each_interface() will be stopped since it
gets a non-zero return value from hostapd_wpa_auth_oui_iter().
Even worse, the packet will not be sent to ethernet because
for_each_interface() returns non-zero value. hostapd_wpa_auth_send_oui()
will then return data_len immediately.
To prevent this, hostapd_wpa_auth_oui_iter() should not return 1 after
any successful transmission to other interfaces, if the dst_addr of
packet is a multicast address.
Signed-off-by: Jinglin Wang <bryanwang@synology.com> Signed-off-by: MinHong Wang <minhongw@synology.com>
Daniel Kobras [Tue, 2 Jul 2019 14:19:38 +0000 (16:19 +0200)]
OpenSSL: Add support for TPM2-wrapped keys
If the header of a PEM-formatted certificate or key in private_key file
indicates that it is wrapped with a TPM2 key, try to autoload the
appropriate OpenSSL engine that can transparently unwrap the key. This
enables systems to use TPM2-wrapped keys as drop-in replacements to
ordinary SSL keys.
This functionality needs
https://git.kernel.org/pub/scm/linux/kernel/git/jejb/openssl_tpm2_engine.git
to be installed as an OpenSSL engine.
Signed-off-by: Daniel Kobras <kobras@puzzle-itc.de>
Jouni Malinen [Sat, 28 Dec 2019 21:19:44 +0000 (23:19 +0200)]
Fix ignore_broadcast_ssid behavior with SSID List and Short SSID List
ignore_broadcast_ssid=1 (or 2) were practically ignored if the Probe
Request frame included the SSID List or Short SSID List elements. Fix
this by requiring exact SSID match whenever ignore_broadcast_ssid is in
use regardless how SSID parameters are set in the Probe Request frame.
AP: Support Short SSID List element in Probe Request frames
According to IEEE P802.11ax/D6.0, 11.1.4.3.4 (Criteria for sending a
response), AP should answer Probe Request frames if either SSID or Short
SSID matches. Implement this part of the Short SSID use for the BSS (the
collocated 6 GHz BSS case is not covered in this commit).
Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Felix Fietkau [Sun, 17 Feb 2019 15:02:56 +0000 (16:02 +0100)]
mesh: Fix race condition in mesh mpm new peer handling
When wpa_supplicant receives another new peer event before the first one
has been processed, it tries to add a station to the driver a second
time (which fails) and then tears down the station entry until another
event comes in.
Fix this by only adding a station to the driver if it didn't exist
already.
Ensure authenticator session timer is applied with wired driver
We use the wired driver for wired port authentication with a slight
extension to add the port into a bridge upon successful authentication
and to remove it from the bridge when the session terminates.
Our expectation was that the Session-Timeout configuration at the RADIUS
server is respected, i.e. the session is terminated and would need
re-authentication - like it is working for WLAN sessions over the
nl80211 driver. Alas, it turned out the session is not terminated with
the wired driver.
It turned out that when ap_handle_session_timer() is executed, the
sta->flags of the wired port has only the WLAN_STA_AUTHORIZED bit set.
The WLAN_STA_AUTH bit, which is used to check whether the STA needs to
be de-authenticated, is missing.
Extend the check for any of the WLAN_STA_(AUTH | ASSOC | AUTHORIZED)
bits to solve this issue with the wired driver. That should not have any
side-effect for the WLAN cases since WLAN_STA_AUTH is expected to always
be set for those when there is an ongoing session and separate checks
for ASSOC and AUTHORIZED don't change this.
Jouni Malinen [Sat, 28 Dec 2019 16:17:44 +0000 (18:17 +0200)]
Allow debug log to be written to both syslog and file
If hostapd or wpa_supplicant is started with both -s and -f command line
arguments, debug log ended up being written only into syslog and the log
file was left empty. Change this so that the log entries will be written
to both places. Either -s or -f (or both) results in debug log to stdout
being disabled which was already the case.
Jouni Malinen [Sat, 28 Dec 2019 15:17:05 +0000 (17:17 +0200)]
tests: Ignore fe80::ff:fe00:300 NS from AP for now in proxyarp tests
This NS seems to go out from the AP interface every now and then and it
makes proxyarp_open_ebtables_ipv6 fails inconveniently often. That frame
should not really be there, but it's not clear what exactly is needed to
stop it going through. It does not come from the test operations
themselves, so ignore it for now to keep test results cleaner.
Jouni Malinen [Fri, 27 Dec 2019 21:17:43 +0000 (23:17 +0200)]
tests: Replace tcpdump with wlantest
This removes dependency on tcpdump by using an already included test
tool for capturing frames with Ethernet headers. There were some issues
in getting tcpdump working on Ubuntu 19.10, so this seems to be a clean
way of addressing that.
Jouni Malinen [Fri, 27 Dec 2019 21:15:14 +0000 (23:15 +0200)]
wlantest: Ethernet interface capture
Allow option (command line argument -e) to capture Ethernet headers
instead of IEEE 802.11 so that wlantest can be used as a replacement for
tcpdump/dumpcap for capturing.
Jouni Malinen [Fri, 27 Dec 2019 20:41:36 +0000 (22:41 +0200)]
tests: Make rrm_reassociation more robust
Flush cfg80211 scan cache explicitly when running this test case to
avoid issues with the ROAM command not working due to a scan result from
an earlier test case. This was causing failures in the following test
case sequence:
rrm_beacon_req_active_ap_channels rrm_reassociation
Field wpa_s->sme.ht_sec_chan keeps secondary channel for the 40 MHz
band. This field is used to prepare a list of channels for the STA OBSS
scan. Initially, the secondary channel is set to HT_SEC_CHAN_UNKNOWN.
Later on, in function wpa_obss_scan_freq_list() it is obtained from the
current BSS HT operation IE. However, the secondary channel information
is not updated after channel switch, which may lead to an incorrect list
of channels prepared for the STA OBSS scan.
Update ht_sec_chan according to the channel switch event data to fix
this.
Jouni Malinen [Fri, 27 Dec 2019 18:01:38 +0000 (20:01 +0200)]
tests: Move ocsp-resp-*-signed*.der generation into test case
There is no need to generate these OCSP responses for every single test
session. Generate these more dynamically if a test case that uses these
files is executed.
Jouni Malinen [Fri, 27 Dec 2019 17:44:35 +0000 (19:44 +0200)]
tests: Move ocsp-server-cache-{revoked,unknown}.der generation into test case
There is no need to generate these OCSP responses for every single test
session. Generate these more dynamically if a test case that uses these
files is executed.
Jouni Malinen [Fri, 27 Dec 2019 17:37:06 +0000 (19:37 +0200)]
tests: Move ocsp-server-cache-key-id.der generation into test case
There is no need to generate this OCSP response for every single test
session. Generate this more dynamically if the test case that uses the
particular file is executed.
Jouni Malinen [Fri, 27 Dec 2019 16:14:09 +0000 (18:14 +0200)]
tests: Fix thread handling in P2P GO Negotiation test cases
Some of the error paths in go_neg_pbc() and go_neg_pin() did not wait
for the helper thread to complete processing. This could result in
unexpected behavior when the test case could have exited while the
thread was still performing tasks for the GO Negotiation. This could
result in getting stuck in one of the following test cases with
"go_neg_init_pbc thread caught an exception from p2p_go_neg_init: Group
formation timed out" showing up in the log.
This was hit, e.g., with the following test sequence:
no_go_freq p2p_channel_drv_pref_autogo
Jouni Malinen [Fri, 27 Dec 2019 15:12:34 +0000 (17:12 +0200)]
tests: Use python selector in the parallel-vm.py main loop
This gets rid of the loop that was polling for things to do every 0.25
seconds and instead, reacts to any data from VMs as soon as it becomes
available. This avoids unnecessary operations when no new data is
available and avoids unnecessary waits when new data becomes available
more quickly.
Jouni Malinen [Fri, 27 Dec 2019 08:09:43 +0000 (10:09 +0200)]
tests: Move VM starting delay control into parallel-vm.py
This is more efficient since we can now start only the necessary number
of VMs instead of always forcing all VMs to start with one second delay.
This can also control the starting delay by keeping at most two VMs
starting at a time instead of using the hardcoded one second wait for
each consecutive VM.
Jouni Malinen [Thu, 26 Dec 2019 22:03:24 +0000 (00:03 +0200)]
tests: Speed up wpas_mesh_gate_forwarding tshark operations
For some reason, running tshark in the test cases can take significant
time especially with UML time-travel. Optimize this by reducing the
number of times tshark needs to be executed in the loop.
Jouni Malinen [Thu, 26 Dec 2019 22:02:05 +0000 (00:02 +0200)]
tests: Optimize tshark operations for new versions
The wlan_mgt to wlan renaming is already included in most recent tshark
versions, so replace the backwards compatibility option to prefer the
new version so that current versions do not need to take the performance
hit.
Jouni Malinen [Thu, 26 Dec 2019 21:13:57 +0000 (23:13 +0200)]
tests: Check operating channel parameter in obss_scan*
It looks like these test cases can fail with the new "Undefined
secondary channel: drop OBSS scan results" case. Add more checks to
determine if something is wrong with the connection.
In addition, force clearing of the cfg80211 scan cache on the main AP
interface so that a scan result from a previous test case cannot prevent
40 MHz channel bandwidth from being used. This could apparently happen
in the following test case sequence:
ap_ht40_scan_conflict obss_scan
Jouni Malinen [Thu, 26 Dec 2019 15:01:02 +0000 (17:01 +0200)]
tests: Make ap_wpa2_disable_eapol_retry_group more robust
Wait for hostapd to report completion of the connection before going
through the disconnection and reconnection steps to avoid a race
condition especially with UML time-travel.
In the previous implementation connected STA performs OBSS scan
according to requests from its 20/40 MHz AP. However STA checks only 40
MHz intolerance subfield from HT Capabilities element in scan results.
Meanwhile, as per IEEE Std 802.11-2016, 11.16.12, STA should check
overlapping BSSs as well.
Note that all the required code to check overlapping BSSs did already
exist for AP mode since AP does those checks properly before operating
as 20/40 MHz BSS in the 2.4 GHz band. Use that existing code by replace
existing 40 MHz intolerance check in sme_proc_obss_scan() with the new
shared helper function check_bss_coex_40mhz().
Extract BSS coex 40 MHz check into a separate function
Move this implementation from check_40mhz_2g4() into a new helper
function check_bss_coex_40mhz function() so that it can be used in the
station mode case as well as the previously used AP mode case.
Ben Greear [Mon, 18 Mar 2019 20:46:34 +0000 (13:46 -0700)]
Allow removing neighbor DB entries by BSSID alone
Let users delete a neighbor by BSSID alone if they prefer. The
underlying code already properly handled a NULL SSID, so just relax the
control interface command calling restrictions.
Signed-off-by: Ben Greear <greearb@candelatech.com>
Jouni Malinen [Wed, 25 Dec 2019 21:43:35 +0000 (23:43 +0200)]
mesh: Allow group management cipher to be configured
This allows BIP-GMAC-128, BIP-GMAC-256, or BIP-CMAC-256 to be used
instead of the previously hardcoded AES-128-CMAC as the group management
cipher when using mesh with PMF. For now, this can be configured by
setting a single group_mgmt value in the network block and doing that
consistently through all the STAs in the mesh.
Jouni Malinen [Wed, 25 Dec 2019 21:35:40 +0000 (23:35 +0200)]
mesh: Increase plink action frame AMPY buffer length for max GTK/IGTK
The previous calculation of the buffer length did not take into account
the possibility of 32-octet GTK and IGTK values and it was also missing
something to cover the 16 octet keys that are supported now. Other
buffer elements were likely sufficient to cover all these cases, but
anyway, it is better to allocate sufficient size specifically for AMPE
to avoid issues in the future.
Add "reconnect" cmdline argument to hostapd_cli/wpa_cli
When the newly added "-r" parameter is used, both clis will try to
reconnect forever on connection lost until signalled (ctrl+c) or
terminated. This is useful only when used with -a to take action to
retrieve events or get status and the cli process stays even if
hostapd/wpa_supplicant daemons restart for some reason (e.g.,
configuration change).
Signed-off-by: Veli Demirel <veli.demirel@airties.com> Signed-off-by: Bilal Hatipoglu <bilal.hatipoglu@airties.com>
Jouni Malinen [Wed, 25 Dec 2019 10:39:17 +0000 (12:39 +0200)]
DBus: Update dont_quote[] with new network profile parameters
This array had not been updated for years (since it was initially added)
and it was missing new network profile parameters that were not quoted
strings (but also not integers that could be set as integer types). Add
those missing non-quoted-strings to allow them to be set through the
DBus interface.
Jouni Malinen [Wed, 25 Dec 2019 10:26:33 +0000 (12:26 +0200)]
DBus: Remove not existing network profile params from dont_quote[]
These parameters are global parameters, not network profile parameters,
and as such, do not below in dont_quote[] which is used to determine
whether a network profile parameter needs to be quoted.
Matthew Wang [Fri, 11 Oct 2019 20:49:25 +0000 (13:49 -0700)]
dbus: Move roam metrics to the correct interface
These properties were in the wpas_dbus_bss_properties array when they
should have been in the wpas_dbus_interface_properties array. Move them
to the right place. This is the logical location for these properties
and it matches both the other parts of the implementation (e.g., being
in enum wpas_dbus_prop, not in enum wpas_dbus_bss_prop) and what
was originally documented for the interface in dbus.doxygen.
Fixes: 2bbad1c7c9cb ("dbus: Export roam time, roam complete, and session length") Fixes: 80d06d0ca9f3 ("dbus: Export BSS Transition Management status") Signed-off-by: Matthew Wang <matthewmwang@chromium.org>
Michal Kazior [Mon, 28 Oct 2019 12:49:29 +0000 (13:49 +0100)]
nl80211: Relax bridge setup
Normally nl80211 driver will attempt to strictly control what bridge
given interface is put in. It'll attempt to remove it from an existing
bridge if it doesn't match the configured one. If it's not in a bridge
it'll try to put it into one. If any of this fails then hostapd will
bail out and not set up the BSS at all.
Arguably that's reasonable since it allows to set the BSS up coherently
with regard to EAPOL handling as well as allows extra interactions with
things like FDB. However, not all hostapd drivers interact with bridge=
the same way. One example is atheros. Therefore it's not clear what the
desired behavior should be if consistency across drivers is considered.
There's a case where one might want to use a non-native Linux bridge,
e.g., openvswitch, in which case regular ioctls won't work to put an
interface into a bridge, or figure out what bridge an interface is in.
The underlying wireless driver can still be an ordinary nl80211 driver.
This change relaxes the bridge setup failure so that hostapd still
starts even if it fails to add an interface into a configured bridge
name. It still sets up all the necessary sockets (including the
configured bridge=) so EAPOL handling should work fine. This then leaves
it to the system integrator to manage wireless interface as bridge ports
and possibly fdb hints too.