Support EHT 1024 aggregation size in TX

The 1024 agg size for RX is supported but not for TX.
This patch adds this support and refactors common parsing logics for
addbaext in both process_addba_resp and process_addba_req into a
function.

Reviewed-by: Shayne Chen <shayne.chen@mediatek.com>
Reviewed-by: Money Wang <money.wang@mediatek.com>
Co-developed-by: Peter Chiu <chui-hao.chiu@mediatek.com>
Signed-off-by: Peter Chiu <chui-hao.chiu@mediatek.com>
Signed-off-by: MeiChia Chiu <MeiChia.Chiu@mediatek.com>
Link: https://patch.msgid.link/20241112083846.32063-1-MeiChia.Chiu@mediatek.com
[pass elems/len instead of mgmt/len/is_req]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Introduce 'ieee80211_mgmt_ba()' to avoid code duplication between
'ieee80211_send_addba_resp()', 'ieee80211_send_addba_request()',
and 'ieee80211_send_delba()', ensure that all related addresses
are '__aligned(2)', and prefer convenient 'ether_addr_copy()'
over generic 'memcpy()'. No functional changes expected.

Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Link: https://patch.msgid.link/20240725090925.6022-1-dmantipov@yandex.ru
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

In MLO, the deflink BSSID is clearly invalid. Since we fill
the addresses as MLD addresses and translate later, use the
AP address here instead.

This fixes an issue that happens with HW restart, where the
DelBA frame is transmitted, but not processed correctly due
to the wrong BSSID (or even just discarded entirely).  As a
result, the BA sessions are kept alive; however, as other
state is reset during HW restart, this then fails (reorder,
etc.) and data doesn't go through until new BA sessions are
established.

Reviewed-by: Miriam Rachel Korenblit <miriam.rachel.korenblit@intel.com>
Link: https://msgid.link/20240510112601.f4e1effdea29.I98e81f22166b68d4b6211191bcaaf8531b324a77@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

We have 13 bits for the status_data, so restricting
type to 4 and subdata to 8 bits is confusing, even if
we don't need more bits now. Change subdata mask to
be 9 bits instead, just to make things match up.

If we actually need more types or more subdata bits
we can later also reshuffle the bits between these,
but we should probably keep them at 13 bits together.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Ayala Beker <ayala.beker@intel.com>
Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com>
Link: https://msgid.link/20240320091155.28ac7b665039.I1abbb13e90f016cab552492e05f5cb5b52de6463@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

For channel contexts, mac80211 currently uses the cfg80211
chandef struct (control channel, center freq(s), width) to
define towards drivers and internally how these behave. In
fact, there are _two_ such structs used, where the min_def
can reduce bandwidth according to the stations connected.

Unfortunately,  with EHT this is longer be sufficient,  at
least not for all hardware.  EHT requires that non-AP STAs
that are connected to an AP with a lower bandwidth than it
(the AP) advertises (e.g. 160 MHz STA connected to 320 MHz
AP) still be able to receive downlink OFDMA and respond to
trigger frames for uplink OFDMA  that specify the position
and bandwidth  for the non-AP STA  relative to the channel
the AP is using.  Therefore, they need to be aware of this,
and at least for some hardware (e.g. Intel) this awareness
is in the hardware. As a result, use of the "same" channel
may need to be split over  two channel contexts where they
differ by the AP being used.

As a first step,  introduce a concept of a channel request
('chanreq') for each interface,  to control the context it
requests.   This step does nothing but reorganise the code,
so that later the AP's chandef can be added to the request
in order to handle the EHT case described above.

Link: https://msgid.link/20240129194108.2e88e48bd2e9.I4256183debe975c5ed71621611206fdbb69ba330@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Even if there are a lot of possible ways drivers might
call this, at least knowing when they do and with what
settings can be useful. Add tracing for it.

Link: https://msgid.link/20240129195435.b20d2ead2013.I8213e65c274451d523a3397519ac578c3ed2df4d@changeid
[removed link-id contortions as suggested by Jeff]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

The new 320 MHz channel width wasn't handled, so connecting
a station to a 320 MHz AP would limit the station to 20 MHz
(on HT) after a warning, handle 320 MHz to fix that.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Link: https://lore.kernel.org/r/20231109182201.495381-1-greearb@candelatech.com
[write a proper commit message]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

We now hold the wiphy mutex everywhere that we use or
needed the A-MPDU locking, so we don't need this mutex
any more. Remove it.

Most of this change was done automatically with spatch.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Convert the A-MPDU work to wiphy work so it holds the
wiphy mutex and we can later guarantee that to drivers.
It might seem that we could run these concurrently for
different stations, but they're all on the ordered
mac80211 workqueue, so this shouldn't matter for that.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

The current SMPS status handling isn't per link, so we only
ever change the deflink, which is obviously wrong, it's not
even used for multi-link connections, but the request API
actually includes the link ID.

Use the new status_data changes to move the handling to the
right link, this also saves parsing the frame again on the
status report, instead we can now check only if it was an
SMPS frame.

Of course, move the worker to be a wiphy work so that we're
able to cancel it safely for the link.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>