linux-stable.git/net/mac80211, branch v6.7.7

wifi: mac80211: accept broadcast probe responses on 6 GHz

2024-03-01T12:41:41+00:00

[ Upstream commit 62a6183c13319e4d2227473a04abd104c4f56dcf ]

On the 6 GHz band, probe responses are sent as broadcast to
optimise medium usage. However, without OCE configuration
we weren't accepting them, which is wrong, even if wpa_s is
by default enabling OCE. Accept them without the OCE config
as well.

Link: https://msgid.link/20240129200907.5a89c2821897.I92e9dfa0f9b350bc7f37dd4bb38031d156d78d8a@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: adding missing drv_mgd_complete_tx() call

2024-03-01T12:41:41+00:00

[ Upstream commit c042600c17d8c490279f0ae2baee29475fe8047d ]

There's a call to drv_mgd_prepare_tx() and so there should
be one to drv_mgd_complete_tx(), but on this path it's not.
Add it.

Link: https://msgid.link/20240131164824.2f0922a514e1.I5aac89b93bcead88c374187d70cad0599d29d2c8@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: initialize SMPS mode correctly

2024-03-01T12:41:41+00:00

[ Upstream commit 86b2dac224f963be92634a878888222e1e938f48 ]

The SMPS mode is currently re-initialized too late, since
ieee80211_prep_channel() can be called again after we've
already done ieee80211_setup_assoc_link(), in case there's
some override of the channel configuration. Fix this.

Link: https://msgid.link/20240129195405.d6d74508be18.I0a7303b1ce4d8e5436011951ab624372a445c069@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: fix driver debugfs for vif type change

2024-03-01T12:41:41+00:00

[ Upstream commit 733c498a80853acbafe284a40468b91f4d41f0b4 ]

If a driver implements the change_interface() method, we switch
interface type without taking the interface down, but still will
recreate the debugfs for it since it's a new type. As such, we
should use the ieee80211_debugfs_recreate_netdev() function here
to also recreate the driver's files, if it is indeed from a type
change while up.

Link: https://msgid.link/20240129155402.7311a36ffeeb.I18df02bbeb685d4250911de5ffbaf090f60c3803@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: set station RX-NSS on reconfig

2024-03-01T12:41:41+00:00

[ Upstream commit dd6c064cfc3fc18d871107c6f5db8837e88572e4 ]

When a station is added/reconfigured by userspace, e.g. a TDLS
peer or a SoftAP client STA, rx_nss is currently not always set,
so that it might be left zero. Set it up properly.

Link: https://msgid.link/20240129155354.98f148a3d654.I193a02155f557ea54dc9d0232da66cf96734119a@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: fix race condition on enabling fast-xmit

2024-03-01T12:41:34+00:00

[ Upstream commit bcbc84af1183c8cf3d1ca9b78540c2185cd85e7f ]

fast-xmit must only be enabled after the sta has been uploaded to the driver,
otherwise it could end up passing the not-yet-uploaded sta via drv_tx calls
to the driver, leading to potential crashes because of uninitialized drv_priv
data.
Add a missing sta->uploaded check and re-check fast xmit after inserting a sta.

Signed-off-by: Felix Fietkau 
Link: https://msgid.link/20240104181059.84032-1-nbd@nbd.name
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: reload info pointer in ieee80211_tx_dequeue()

2024-02-23T08:51:51+00:00

commit c98d8836b817d11fdff4ca7749cbbe04ff7f0c64 upstream.

This pointer can change here since the SKB can change, so we
actually later open-coded IEEE80211_SKB_CB() again. Reload
the pointer where needed, so the monitor-mode case using it
gets fixed, and then use info-> later as well.

Cc: stable@vger.kernel.org
Fixes: 531682159092 ("mac80211: fix VLAN handling with TXQs")
Link: https://msgid.link/20240131164910.b54c28d583bc.I29450cec84ea6773cff5d9c16ff92b836c331471@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Greg Kroah-Hartman

wifi: mac80211: fix waiting for beacons logic

2024-02-16T18:14:20+00:00

[ Upstream commit a0b4f2291319c5d47ecb196b90400814fdcfd126 ]

This should be waiting if we don't have a beacon yet,
but somehow I managed to invert the logic. Fix that.

Fixes: 74e1309acedc ("wifi: mac80211: mlme: look up beacon elems only if needed")
Link: https://msgid.link/20240131164856.922701229546.I239b379e7cee04608e73c016b737a5245e5b23dd@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: fix unsolicited broadcast probe config

2024-02-16T18:14:20+00:00

[ Upstream commit 178e9d6adc4356c2f1659f575ecea626e7fbd05a ]

There's a bug in ieee80211_set_unsol_bcast_probe_resp(), it tries
to return BSS_CHANGED_UNSOL_BCAST_PROBE_RESP (which has the value
1<<31) in an int, which makes it negative and considered an error.
Fix this by passing the changed flags to set separately.

Fixes: 3b1c256eb4ae ("wifi: mac80211: fixes in FILS discovery updates")
Reviewed-by: Jeff Johnson 
Link: https://msgid.link/20240129195729.965b0740bf80.I6bc6f5236863f686c17d689be541b1dd2633c417@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin

wifi: mac80211: fix RCU use in TDLS fast-xmit

2024-02-16T18:14:19+00:00

[ Upstream commit 9480adfe4e0f0319b9da04b44e4eebd5ad07e0cd ]

This looks up the link under RCU protection, but isn't
guaranteed to actually have protection. Fix that.

Fixes: 8cc07265b691 ("wifi: mac80211: handle TDLS data frames with MLO")
Link: https://msgid.link/20240129155348.8a9c0b1e1d89.I553f96ce953bb41b0b877d592056164dec20d01c@changeid
Signed-off-by: Johannes Berg 
Signed-off-by: Sasha Levin