Mesh interface would have used its channel parameters that were
configured during initialization even after channel switch due to DFS
radar detection during CAC which could result in a channel switch error.
Fix the error by updating the channel parameters when channel has been
changed from the initial one.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
ssid->frequency is one of the variables used to get the channel number
from a given frequency. Leaving it as unchanged when pri/sec channel
switch will cause picking up a wrong channel number after applying the
secondary channel offset for HT40 and leads to failing interface
bring-up.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
Mesh join function is the last function to be called during mesh join
process, but it's been called a bit earlier than it's supposed to be, so
that some mesh parameter values such as VHT capabilities were not
applied correct when mesh join is in process. Moreover, the current
design of mesh join that is called directly after mesh initialization
isn't suitable for DFS channels to use, since mesh join process should
be paused until DFS CAC is done and resumed after it's done.
The callback will be called by hostapd_setup_interface_complete_sync().
There is a possibility that completing mesh init fails, so add error
handling codes for that.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
wpa_supplicant_conf_ap_ht() has to happen before
hostapd_setup_interface() in order for its configuration settings to
have effect on interface configuration.
Disable primary and secondary channel switch because of missing tie
breaking rule/frames in mesh networks. A rather long comment about
this issue is placed in mesh.c in the corresponding place.
I was not able to reproduce the memory corruption during
mesh_secure_ocv_mix_legacy, which lead to a revert of a similar patch in
the past.
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Some places in the code base were not using the wrappers like
hostapd_set_oper_centr_freq_seg0_idx and friends. This could lead to
errors, for example when joining 80 MHz mesh networks. Fix this, by
enforcing usage of these wrappers.
wpa_supplicant_conf_ap_ht() now checks for HE capability before dealing
with VHT in order for these wrappers to work, as they first check HE
support in the config.
While doing these changes, I've noticed that the extra channel setup
code for mesh networks in wpa_supplicant/mesh.c should not be necessary
anymore and dropped it. wpa_supplicant_conf_ap_ht() should handle this
setup already.
Acked-by: John Crispin <john@phrozen.org>
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
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.
Signed-off-by: Jouni Malinen <j@w1.fi>
Hardcode this to be defined and remove the separate build options for
PMF since this functionality is needed with large number of newer
protocol extensions and is also something that should be enabled in all
WPA2/WPA3 networks.
Signed-off-by: Jouni Malinen <j@w1.fi>
Mesh points can partially support HE features (when requiring no
controlling STA/AP) as long as hardware supports it. The kernel just
requires support for HE mesh and wpa_supplicant can forward the peer
capabilities to the kernel for further processing.
Signed-off-by: Sven Eckelmann <seckelmann@datto.com>
The bandwidth values are shared between VHT and HE mode so remove the
VHT specific prefix.
Signed-off-by: Shashidhar Lakkavalli <slakkavalli@datto.com>
Signed-off-by: John Crispin <john@phrozen.org>
The driver was left in mesh mode (joined to the group) if RSN
authenticator initialization failed. This could result in next
operations failing due to unexpected driver state.
This was found with the following hwsim test case sequence:
mesh_wpa_auth_init_oom dpp_config_no_discovery
Signed-off-by: Jouni Malinen <j@w1.fi>
It does not look like wpa_s->ifmsg could be NULL here, but better be
more consistent anyway to keep static analyzers happier by avoiding
dereference of wpa_s->ifmsh in the function before the NULL check for
it.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
This reverts commit 2564184440.
Commit 2564184440 ("mesh: Apply channel attributes before setup
interface") triggers some channel configurations to result in leaking
memory. This seems to be caused by hapd->started not getting set when
going through a callback to start hostapd operation (e.g., when using
HT40 coex scan) due to hostapd_setup_bss() not getting called. This
results in hostapd_free_hapd_data() not clearing allocated
hapd->wpa_auth. This can be reproduced with the hwsim test case
mesh_secure_ocv_mix_legacy.
A more complete cleanup of the pending mesh patch for DFS support seems
to be needed to fix this properly, so the best approach for now is to
revert this patch and bring it back once rest of the mesh changes are
ready to be applied.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
Channel width in VHT mode refers HT capability when the width goes down
to below 80 MHz, hence add checking HT channel width to its max
operation channel width. So that mesh has capability to select bandwidth
below 80 MHz.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
wpa_supplicant configuration has country parameter that is supposed to
be used in AP mode to indicate supporting IEEE 802.11h and 802.11d.
Reflect this configuration to Mesh also since Mesh is required to
support 802.11h and 802.11d to use DFS channels.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
RSN initialization should work together with mesh join when it's used.
Since mesh join could be called at a different stage if DFS channel is
used, relocate the RSN initialization call to mesh join. It is still the
same call flow of mesh join before this if non-DFS channels are used,
hence no significant side effect will occur.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
Mesh join function consists of two parts which are preparing
configurations and sending join event to the driver. Since physical mesh
join event could happen either right after mesh configuration is done or
after CAC is done in case of DFS channel is used, factor out the
function into two parts to reduce redundant calls.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
RSN initialization can be used in different phases if mesh
initialization and mesh join don't happen in sequence such as DFS CAC is
done in between, hence factor it out to help convering the case. This
can also get rid of unnecessary indentation by handling the
mconf->security != MESH_CONF_SEC_NONE functionality in a helper
function.
Signed-off-by: Peter Oh <peter.oh@bowerswilkins.com>
wpa_s->ifmsh needs to be allocated using hostapd_alloc_iface() instead
of a direct call to os_zalloc(), otherwise the linked list for station
taxonomy items remains uninitialized, leading to a crash on the first
attempt to traverse that list
Signed-off-by: Felix Fietkau <nbd@nbd.name>
The use of hostapd code for a mesh interface did not register
hapd->msg_ctx. This needs to be done similarly to the existing cases in
wpa_supplicant AP and IBSS mode uses so that wpa_msg() calls from the
hostapd/AP code get delivered properly.
Signed-off-by: Jouni Malinen <jouni@codeaurora.org>
The recently introduced sae_password parameter was only handled properly
in wpa_supplicant/sme.c while wpa_supplicant/mesh.c assumed that
ssid->passphrase exclusively holds the secret.
Import the logic from sme.c to mesh.c to allow having only sae_password
set which otherwise throws this error:
AP-ENABLED
mesh: Passphrase for SAE not configured
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
In some practical cases, it is useful to suppress joining to node in the
distance. The new field mesh_rssi_threshold could be used as RSSI
threshold for joining.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This leads to cleaner code overall, and also reduces the size
of the hostapd and wpa_supplicant binaries (in hwsim test build
on x86_64) by about 2.5 and 3.5KiB respectively.
The mechanical conversions all over the code were done with
the following spatch:
@@
expression SIZE, SRC;
expression a;
@@
-a = os_malloc(SIZE);
+a = os_memdup(SRC, SIZE);
<...
if (!a) {...}
...>
-os_memcpy(a, SRC, SIZE);
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The local custom version of allocating and initializing struct
hostapd_data within wpa_supplicant_mesh_init() is problematic. This has
already missed couple of initialization steps that are required. Instead
of trying to remember to keep this up to date, use
hostapd_alloc_bss_data() so that there is only one place for this
initialization.
This is fixing a recent issue where FILS HLP started using
hapd->dhcp_server and expected that to be initialized to -1. For the
mesh case, that did not happen and when removing the interface, the FILS
HLP implementation ended up unregistering eloop socket for
hapd->dhcp_server (= 0). This could result in missing socket callbacks
for an arbitrary socket.
Signed-off-by: Jouni Malinen <j@w1.fi>
If 20/40 MHz co-ex scan resulted in switching primary and secondary
channels, mesh setup failed to update the frequency parameters for
hostapd side configuration and that could result in invalid secondary
channel configuration preventing creating of the mesh network. This
could happen, e.g., when trying to set up mesh on 5 GHz channel 36 and
co-ex scan finding a BSS on channel 40. Switching the pri/sec channels
resulted in hostapd code trying to check whether channel 32 is
available. Fix this by swapping the channels for hostapd configuration
when needed.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Previously, this was initialized in hostapd_setup_bss() which made it
possible for a REMOVE_NEIGHBOR control interface command to be issued
prior to the list head pointers having been set. That resulted in a NULL
pointer dereference. Fix this by initializing the list head at the time
the data structure gets allocated.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Accoding to the comment of struct wpa_driver_mesh_bss_params, the
max_peer_links parameter should be under that struct.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
In mesh interface, the wpa_supplicant state was either
DISCONNECT/SCANNING in non-secured connection or AUTHENTICATING in
secured connection. The latter prevented the scan. Update the
wpa_supplicant state in mesh to be COMPLETED upon initialization. This
is similar to the P2P GO case.
Signed-off-by: Maital Hahn <maitalm@ti.com>
This moves pairwise, group, and management group ciphers to various mesh
data structures to avoid having to hardcode cipher in number of places
through the code. While CCMP and BIP are still the hardcoded ciphers,
these are now set only in one location.
Signed-off-by: Jouni Malinen <j@w1.fi>
This is initial step in fixing issues in how PMF configuration for RSN
mesh was handled. PMF is an optional capability for mesh and it needs to
be configured consistently in both hostapd structures (to get proper
RSNE) and key configuration (not included in this commit).
Signed-off-by: Jouni Malinen <j@w1.fi>
ssid->frequency cannot be 0 in wpa_supplicant_mesh_init() since
wpas_supplicant_join_mesh() rejects such a configuration.
Signed-off-by: Jouni Malinen <j@w1.fi>
This patch add functionality of mesh SAE PMKSA caching. If the local STA
already has peer's PMKSA entry in the cache, skip SAE authentication and
start AMPE with the cached value.
If the peer does not support PMKSA caching or does not have the local
STA's PMKSA entry in the cache, AMPE will fail and the PMKSA cache entry
of the peer will be removed. Then STA retries with ordinary SAE
authentication.
If the peer does not support PMKSA caching and the local STA uses
no_auto_peer=1, the local STA can not retry SAE authentication because
NEW_PEER_CANDIDATE event cannot start SAE authentication when
no_auto_peer=1. So this patch extends MESH_PEER_ADD command to use
duration(sec). Throughout the duration, the local STA can start SAE
authentication triggered by NEW_PEER_CANDIDATE even though
no_auto_peer=1.
This commit requires commit 70c93963ed
('SAE: Fix PMKID calculation for PMKSA cache'). Without that commit,
chosen PMK comparison will fail.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This allows a mesh peer connection to be initiated manually in
no_auto_peer mesh networks.
Signed-off-by: Natsuki Itaya <Natsuki.Itaya@jp.sony.com>
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
This command allows the specified mesh peer to be disconnected.
Signed-off-by: Natsuki Itaya <Natsuki.Itaya@jp.sony.com>
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
The full VHT channel information was not set in the hostapd data
structures which resulted in incorrect information (all zeros) being
used when building the VHT Operation element for peering messages while
the actual driver mode was set with the full details. We did not seem to
use the VHT information from peering messages, so this does not change
behavior with another wpa_supplicant-based mesh implementation. Anyway,
these elements should match the ones used in Beacon frames.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
This is needed to avoid hitting WEP/TKIP detection in
ibss_mesh_setup_freq() if the previous connection used WEP or TKIP.
Previously, that could have resulted in VHT and HT getting disabled for
the mesh connection.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Mesh Points themselves have capability to support VHT as long as
hardware supports it. However, supporting VHT in mesh mode was disabled
because no one had clearly tested and confirmed its functionality. Since
VHT80 has now been verified to work with ath10k QCA988X driver and
mac80211_hwsim, enable VHT support in mesh mode.
Signed-off-by: Peter Oh <poh@qca.qualcomm.com>
The mesh SAE auth often fails with master branch. By bisect I found
commit eb5fee0bf5 ('SAE: Add side-channel
protection to PWE derivation with ECC') causes this issue. This does not
mean the commit has a bug. This is just a CPU resource issue.
After the commit, sae_derive_pwe_ecc() spends 101(msec) on my PC (Intel
Atom N270 1.6GHz). But dot11RSNASAERetransPeriod is 40(msec). So
auth_sae_retransmit_timer() is always called and it can causes
continuous frame exchanges. Before the commit, it was 23(msec).
On the IEEE 802.11 spec, the default value of dot11RSNASAERetransPeriod
is defined as 40(msec). But it looks short because generally mesh
functionality will be used on low spec devices. Indeed Raspberry Pi B+
(ARM ARM1176JZF-S 700MHz) requires 287(msec) for new
sae_derive_pwe_ecc().
So this patch makes the default to 1000(msec) and makes it configurable.
This issue does not occur on infrastructure SAE because the
dot11RSNASAERetransPeriod is not used on it.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
