Tests have shown that we can meet low latency KPIs with A-MSDU
enabled so enable it to achieve max TPT.
Signed-off-by: NMordechay Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Link: https://lore.kernel.org/r/iwlwifi.20200425130140.e469ce6501e4.Ibdecebca830bdfbf5220693dd1f5367f7736242d@changeid

This is a straight-forward conversion case for the new function, and
while we're at it, we can remove a null write to skb->next by replacing
it with skb_mark_not_on_list.
Signed-off-by: NJason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When we transmit after TXQ dequeue, we aren't paying attention to
the return value of the transmit functions, leading to a potential
SKB leak.

Refactor the code a bit (and rename ..._tx to ..._tx_sta) to check
for this happening.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Fixes: cfbc6c4c ("iwlwifi: mvm: support mac80211 TXQs model")
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

It used to be the case that if we got here, we wouldn't warn
but instead allocate the queue (DQA). With using the mac80211
TXQs model this changed, and we really have nothing to do with
the frame here anymore, hence the warning now.

However, clearly we missed in coding & review that this is now
a pure error path and leaks the SKB if we return 0 instead of
an indication that the SKB needs to be freed. Fix this.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Fixes: cfbc6c4c ("iwlwifi: mvm: support mac80211 TXQs model")
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

As noted in the previous commit, due to the way we allocate the
dev_cmd headers with 324 byte size, and 4/8 byte alignment, the
part we use of them (bytes 20..40-68) could still cross a page
and thus 2^32 boundary.

Address this by using alignment to ensure that the allocation
cannot cross a page boundary, on hardware that's affected. To
make that not cause more memory consumption, reduce the size of
the allocations to the necessary size - we go from 324 bytes in
each allocation to 60/68 on gen2 depending on family, and ~120
or so on gen1 (so on gen1 it's a pure reduction in size, since
we don't need alignment there).

To avoid size and clearing issues, add a new structure that's
just the header, and use kmem_cache_zalloc().
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This is dead code, nothing uses the IWL_DEVICE_22560 macro and
thus nothing every uses IWL_DEVICE_FAMILY_22560. Remove it all.

While at it, remove some code and definitions used only in this
case, and clean up some comments/names that still refer to it.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

There are various of flows which require tids flushing
(disconnection, suspend, etc...).
Currently, when the driver instructs the FW to flush
he masks all the data tids(0-7).
However, the driver doesn't set the management tid (#15)
which cause the FW not to flush it.
When the FW tries to remove the mgmt queue he throws an assert
since it is not an empty queue.
instead of just set only the data tids set everything and let
the FW ignore the invalid tids.
Signed-off-by: NHaim Dreyfuss <haim.dreyfuss@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

OFDM rate used for all bands except to band 2.4 which use CCK rate.
Inverting the condition help that in future we won't need to expand the
condition for more bands.
Signed-off-by: NTova Mussai <tova.mussai@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

debugfs amsdu_len sets only the max_amsdu_len for ieee80211 station
so take it into consideration while getting max amsdu

Fixes: af2984e9 ("iwlwifi: mvm: add a debugfs entry to set a fixed size AMSDU for all TX packets")
Signed-off-by: NMordechay Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Add a pointer to the iwl_trans structure and point it to the trans
part of the cfg.  This is the first step in disassociating the trans
configuration from the rest of the configuration.
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In order to be able to select the cfg depending on the HW revision or
on the RF ID, we need to set up the trans before selecting the cfg.
To do so, move the elements from cfg that are needed by
iwl_trans_alloc() to a separate struct at the top of the cfg, so it
can be used by other cfg types as well, before selecting the rest of
the configuration.
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When printing a TX, add to the print the length of the frame.
That will help with BSEP (buffer status report poll) tests.
Signed-off-by: NShaul Triebitz <shaul.triebitz@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Accessing the hdr of an skb that was consumed already isn't
a good idea.
First ask if the skb is a QoS packet, then keep that data
on stack, and then consume the skb.
This was spotted by KASAN.

Cc: stable@vger.kernel.org
Fixes: 08f7d8b6 ("iwlwifi: mvm: bring back mvm GSO code")
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

In some buggy scenarios we could possible attempt to transmit frames larger
than maximum MSDU size. Since our devices don't know how to handle this,
it may result in asserts, hangs etc.
This can happen, for example, when we receive a large multicast frame
and try to transmit it back to the air in AP mode.
Since in a legal scenario this should never happen, drop such frames and
warn about it.
Signed-off-by: NAndrei Otcheretianski <andrei.otcheretianski@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Now that we have per station control over amsdu size no need for
multiple entries, especially that the old one is misleading due to not
setting it for all protocols as a limit.
Signed-off-by: NMordechay Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

As iwl_mvm_tx_mpdu() is not disabling BH while obtaining iwl_mvm_sta->lock
(which is being taken from BH context as well), it has to be always
invoked with BH disabled. Make that clear in a comment.
Signed-off-by: NJiri Kosina <jkosina@suse.cz>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This helps debugging when things go wrong.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When hostapd sends a probe response, it does not request an
acknowledgment on the probe response. This setting is propagated
to the driver and FW from mac80211, and as a result probe responses
are transmitted only once without retries. In congested
environments, this reduces the likelihood for finding the AP.

To overcome this, enable acknowledgment and retries for probe
responses, even in case no acknowledgment is requested. In such
a case the retry limit is set to 3.
Signed-off-by: NIlan Peer <ilan.peer@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Previously the protection flag wasn't set on multicast frames even if
the RA is unicast. Fix this by checking addr1 instead.
Signed-off-by: NAndrei Otcheretianski <andrei.otcheretianski@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Add an option to not send beacons and probe responses. This is
used for testing multiple-bssid.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The driver assumes certain sizes and lengths aren't crossed in some
places.  Make sure this indeed happens.

Found by Klocwork.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When a frame is transmitted without a station, need to set the rate
and flags in the Tx command, as the FW does not have any information as
to what rate and flags should be used for this frame.
Signed-off-by: NIlan Peer <ilan.peer@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

We have a slightly better TCP performance with GSO.
Add it back, it can co-exist with the code that builds
AMSDUs in mac80211.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Support getting mac80211 building AMSDUs for us. Remove GSO
support from mvm - we don't need it anymore.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Move to use the new mac80211 TXQs implementation. This has
quite a few benefits for us. We can get rid of the awkward
mapping of DQA to mac80211 queues. We can stop buffering
traffic while waiting for the queue to be allocated. We can
also use mac80211 AMSDUs instead of building it ourselves.

The usage is pretty simple:
Each ieee80211_txq contains iwl_mvm_txq. There is such a
queue for each TID, and one for management frames. We keep
having static AP queues for probes and non-bufferable MMPDUs,
along with broadcast and multicast queues. Those are being
used from the "old" TX invocation path - iwl_mvm_mac_tx.

When there is a new frame in a TXQ, iwl_mvm_mac_wake_tx is
being called, and either invokes the TX path, or allocates
the queue if it does not exist.

Most of the TX path is left untouched, although we can consider
cleaning it up some more, for example get rid of the duplication
of txq_id in both iwl_mvm_txq and iwl_mvm_dqa_txq_info.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

There is no need to lock mvm station for transport reclaim.
Move the locking down, after the reclaim.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Make the coupling of station id and queue id clear. Group code
together. Remove outdated comment. Never use an undefined hw
queue as given from mac80211.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Assigning mvmsta to be NULL when we are about to exit the
function is pointless. Remove it. Move the variable declaration
to the scope it is used.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In iwl_mvm_tx_skb_non_sta(), in case of managed interface,
the AP station was supposed to be used for multicast frames
instead of the auxiliary station to avoid frames possibly
sent to an absent P2P GO. However, when moving to DQA mode,
this was broken as no valid queue was assigned. This is fixed
by a recent patch that directs all non-offchannel traffic to
ap station earlier in the TX path. However, the broken, and
now dead code, remained. Remove it.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

If tx fails during connection establishment, try another antenna for
the next tx. This will increase the chance to establish connection if
one of the antennas is blocked.  Note that the antenna is toggled even
when failing to tx data frames since connection establishment may use
EAPOLs for 802.1X authentication/ 4 way handshake.
Signed-off-by: NAvraham Stern <avraham.stern@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Switch the antenna used for management tx only if previous tx failed.
If previous tx succeeded, there is no reason to switch antennas.
Signed-off-by: NAvraham Stern <avraham.stern@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

All the queue management code runs under mvm->mutex, so there are
only very few cases of accessing the data structures without it:
 * TX path, which doesn't take any locks anyway
 * iwl_mvm_wake_sw_queue() and iwl_mvm_stop_sw_queue() where we
   just (atomically) read a bitmap, so the lock isn't needed.

Therefore, we can remove the spinlock. This enables some cleanup
in the ugly locking in iwl_mvm_inactivity_check().
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

We don't really need this state: instead of having an inactive
state where we can awaken zombie queues again if needed, just
keep them in their normal state unless a new queue is actually
needed and there's no other way of getting one.

We do this here by making the inactivity check not free queues
unless instructed that we now really need to allocate one to a
specific station, and in that case it'll just free the queue
immediately, without doing any inactivity step inbetween.

The only downside is a little bit more processing in this case,
but the code complexity is lower.

Additionally, this fixes a corner case: due to the way the code
worked, we could only ever reuse an inactive queue if it was
the reserved queue for a station, as iwl_mvm_find_free_queue()
would never consider returning an inactive queue.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The triplet of get trigger, is trigger enabled and is trigger stopped
repeats itself.  Group them in a function to avoid code duplication.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Current FIFO size calculation is wrong for two reasons:
- We access lmac 0 by default
- We don't take 11ax into consideration.
Fix both.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Probe responses were sent to the multicast station while
they should be routed to the broadcast station.
This has no negative effect since the frame was still
routed to the right queue, but it looked very fishy
to send a frame to a (queue, station) tuple where
'queue' is not mapped to 'station'.

Fixes: 7c305de2 ("iwlwifi: mvm: Direct multicast frames to the correct station")
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The Free Software Foundation address is superfluous and causes
checkpatch to issue a warning when present.  Remove all paragraphs
with FSF's address to prevent that.
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When we receive TX response, we may release a few packets
due to a hole that was closed in the transmission window.

However, if that frame failed, we will mark all the released
frames as failed and will send multiple BARs.

This affects statistics badly, and cause unnecessary frames
transmission.

Instead, mark all the following packets as success, with the
desired result of sending a bar for the failed frame only.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

There is a bug in FW where the sequence control may be
incorrect, and the driver overrides it with the value
of the ieee80211 header.

However, in BAR there is no sequence control in the header,
which result with arbitrary sequence.

This access to an unknown location is bad and it makes the
logs very confusing - so fix it.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

A probe response built by a P2P GO should contain:
1. CSA/eCSA IE when relevant
2. If the corresponding probe request had P2P IE, then
need to add P2P IE with NOA attributes.

However, the NOA attributes and the updated channel switch
counter are known only to the FW. The solution is that FW
will send a notification with the relevant probe response
data and the driver will save it and update the probe
response accordingly.
Signed-off-by: NGregory Greenman <gregory.greenman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>