The reason station id and tid fields are both in baid data and
in the reorder buffer per queue is that we couldn't access the
baid_data in the reorder timer functions.
Now that we do some pointer math and access it anyway, those
fields can be removed.
This save some space and some code.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Now that we may have up to 256 entries per reorder buffer, and possibly up
to 16 queues, we can use a LOT of memory for this (64k for each station).
Allocate it according to what we need, which is of course much less for HT
stations (only 16k at a max of 16 queues).

However, this comes at the expense of complicating the code a bit to
calculate the right entry structure to use for each frame.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The trigger that collects data when a frame is released
because of the timer of the reordering buffer was not
implemented for 9000 devices.
Fix this.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

We always need to initialize the status argument to the success case
before calling iwl_mvm_send_cmd_status() or
iwl_mvm_send_cmd_pdu_status() (which calls the former) otherwise we
may get an uninitialized value back.  In this case, we use
ADD_STA_SUCCESS as success.

Fixes: 732d06e9 ("iwlwifi: mvm: add station before allocating a queue")
Reported by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Add a new state to enum iwl_mvm_agg_state, which is used between
queueing the work that starts tx aggregations and actually starting that
work (changing to state IWL_AGG_STARTING).
This solves a race where ieee80211_start_tx_ba_session is called a
second time, before the work queued by the first run has a chance to
change the agg_state. In this case the second call to
ieee80211_start_tx_ba_session returns an error, and the fallback is to
abort the ba session start.

Fixes: 482e4844 ("iwlwifi: mvm: change open and close criteria of a BA session")
Signed-off-by: NNaftali Goldstein <naftali.goldstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

One of the queue config params is the associated station id.
Hence the FW must know about the station prior to the queue allocation.
In a000 devices, allocating a queue without a valid station
results with assert 0x2B00.
In FW restart flow the queues are allocated before
adding the station so first add the station.
Signed-off-by: NShaul Triebitz <shaul.triebitz@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When an RX block-ack session times out, the firmware, which offloads
RX reordering but not the BA session negotiation, stops the session
but doesn't send a DELBA.  This causes the the session to remain
active in the remote device, so no more BA sessions will be
established, causing a severe throughput degradation due to the lack
of aggregation.

Use the new ieee80211_rx_ba_timer_expired API when the ba session timer
expires, since this will tear down the ba session and also send a delba.

The previous API used is intended for drivers that offload the
addba/delba negotiation, but not the rx reordering, while our driver
does the opposite.

This patch depends on "mac80211: add api to start ba session timer
expired flow".
Signed-off-by: NNaftali Goldstein <naftali.goldstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

All the firmware versions the driver supports enable DQA, and thus
the only way to get non-DQA mode is to modify the source. Remove
this mode to simplify the code.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The iwl_mvm_add_bcast_sta() and the iwl_mvm_rm_bcast_sta() functions
are only called in P2P flows.  Add _p2p_ to the function names to make
this explicit.
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

While waiting for queues to empty,
If txq_id == IWL_MVM_INVALID_QUEUE for all txq_ids,
ret is used uninitialized.
Found by Klocwork.

Fixes: d6d517b7 ("iwlwifi: add wait for tx queue empty")
Signed-off-by: NSharon Dvir <sharon.dvir@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The FIFO numbering is different in A000 devices. This
means that we routed BE packets to BK FIFO. Fix this.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Set the STA_FLG_RTS_MIMO_PROT bit in station_flags_msk of the add sta
command, so that when smps mode changes, the FW will know about it.

In particular, in AP mode, clients are added upon receival of an auth
request, at which point there's no knowledge of the client's smps mode.
When the assoc request arrives, the add_sta command is resent to modify
the station parameters. At this point the driver knows the smps mode,
but since the corresponding bit in the mask is not set, the fw doesn't
update this field so there's no rts protection for mimo.

Fixes: 5bc5aaad ("iwlwifi: mvm: set up initial SMPS/NSS station info")
Signed-off-by: NNaftali Goldstein <naftali.goldstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When IBSS was implemented for DQA, we missid a few places where it
should be handled in the same way as AP.

Fixes: ee48b722 ("iwlwifi: mvm: support ibss in dqa mode")
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Sometimes, we can have an firmware crash while trying to
recover from a previous firmware problem.
When that happens, lots of things can go wrong. For example
the stations don't get added properly to mvm->fw_id_to_mac_id.

Mac80211 tries to stop A-MPDU upon reconfig but in case of
a firmware crash we will bail out fairly early and in the
end, we won't delete the A-MPDU Rx timeout.
When that timer expired after a double firmware crash,
we end up dereferencing mvm->fw_id_to_mac_id[sta_id]
which is NULL.

Fixes: 10b2b201 ("iwlwifi: mvm: add infrastructure for tracking BA session in driver")
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

If the first frame on a given TID is received with seqno 0 and needed
to be retransmitted, we erroneously drop it because the deduplication
data is initialized to zero, and then comparing

        if (unlikely(ieee80211_has_retry(hdr->frame_control) &&
                     dup_data->last_seq[tid] == hdr->seq_ctrl &&
                     dup_data->last_sub_frame[tid] >= sub_frame_idx))
                return true;

will return in iwl_mvm_is_dup() since last_sub_frame is also set to
zero, and sub_frame_idx is usually zero since this only covers the
relatively rare case of A-MSDU.

Fix this by initializing the last_seq array to 0xffff, which is an
impossible value for hdr->seq_ctrl to have here because the lower
four bits are the fragment number, and fragments aren't handled in
this code but go to mac80211 instead.

Fixes: a571f5f6 ("iwlwifi: mvm: add duplicate packet detection per rx queue")
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

A SCD bug was fixed in the A000 family, allowing to
support aggregations of 64 frames (rather than 63).
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The sta_id variable is used as an index in an array, should be unsigned.
Found by Klocwork.

Fixes: 9f9af3d7 ("iwlwifi: mvm: re-aggregate shared queue after unsharing")
Signed-off-by: NSharon Dvir <sharon.dvir@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This new API allows flushing queues based on station ID and TID in A000
devices.  One reason for using this is that tfd_queue_mask is only good
for 32 queues, which is not enough for A000 devices.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NMordechai Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In case we need to move the scheduler write pointer by
steps of 0x40, 0x80 or 0xc0, the scheduler gets stuck.
This leads to hardware error interrupts with status:
0x5A5A5A5A or alike.

In order to work around this, detect in the transport
layer that we are going to hit this case and tell iwlmvm
to increment the sequence number of the packets. This
allows to keep the requirement that the WiFi sequence
number is in sync with the index in the scheduler Tx queue
and it also allows to avoid the problematic sequence.
This means that from time to time, we will start a queue
from ssn + 1, but that shouldn't be a problem since we
don't switch to new queues for AMPDU now that we have
DQA which allows to keep the same queue while toggling
the AMPDU state.

This bug has been fixed on 9000 devices and up.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

On A000 HW, the SCD rdptr has only 8 bits allocated
for it, thus when checking if a queue is full, or
when checking if the SSN is equal to the TID's
next_reclaimed, A000 HW should trim the SSN.

Fix this by "normalizing" the SSN to wrap around
0xFF when comparing to the next_reclaimed on A000
HW.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Install GTKs on AP side for new TX API.
Don't add IV space, it's added by the HW.

While at that fix GCMP abnd GCMP-256 GTK installation
which work similarly to the new TX API.
Signed-off-by: NDavid Spinadel <david.spinadel@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In DQA mode, there is no need to wait for the TXQ to
clear out after getting a DELBA, since traffic can
continue running on the queue.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The iwl_mvm_remove_sta_key() function handles removing a key when the
sta doesn't exist anymore.  Mistakenly, this was changed to return an
error while fixing another bug.

If the mvm_sta doesn't exist, we continue normally, but just don't try
to remove the igtk key.

Fixes: cd4d23c1 ("iwlwifi: mvm: Fix removal of IGTK")
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Allow working IBSS also when working in DQA mode.
This is done by setting it to treat the queues the
same as a BSS AP treats the queues.

Fixes: 7948b873 ("iwlwifi: mvm: enable dynamic queue allocation mode")
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Now that we have 512 queues, add a wait for single TX
queue to gen2.
This replaces gen1 wait_tx_queues_empty, which was limited
to 32 queues.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The function flushed only agg queue and no more traffic
is queued on it.
However, it waits for all queues to empty, which is not
necessary and may take more time.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Avoid using the global flush and move to flush per
station whenever possible in DQA mode.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Rename current wait_tx_queue_empty to wait_tx_queues_empty since
it waits for multiple queues (up to 32).
Next patch will add a wait for single TX queue which is needed for
gen2 to be scalable for 512.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Support change to ADD_STA API to support station types.
Each station is assigned its type.
This simplifies FW handling of the broadcast and multicast
stations:
* broadcast station is identified by its type and not the mac
  address.
* multicast queue is no longer treated differently. The opening
  and closing of it is done by referring to its station.
  There is no need to specify it in the MAC command.
* When disabling TX to all station driver can disable the traffic
  on multicast station, so FW doesn't have to do it.
Change is backward compatible.
Change the order of adding and removing the stations according to
FW requirements.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Most of the fields aren't needed in new TX path.
Enlarging the struct to 512 queues will consume a lot of memory.
Remove all references to the struct in the new TX path.
Move mac80211 queue mapping outside, since it will be needed per
queue for TVQM mode.
Add warning in paths that shouldn't be hit.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

We can't use IEEE80211_INVAL_HW_QUEUE to mark a queue as
invalid since 255 will be a valid value for a TVQM queue
index.
Use IWL_MVM_INVALID_QUEUE instead for accessing txq_id.
reserved_queue can stay a u8 since reserved_queue is not
used when TVQM is enabled.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

cab_queue can now get bigger than u8, since in TVQM we will support
512 queues..
Support it by maintaining internal mapping between the actual number
and mac80211 queue (IWL_MVM_DQA_GCAST_QUEUE).
For pre-a000 the internal queue will be the same as the mac80211
queue.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The reserved queue is never used, save the trouble.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Remove an extra variable 'queue' that already exists.
Also, since there are no code paths that use 'queue'
without intializing it, remove the unnecessary zero
initialization.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

While freeing inactive queue, check mvmsta to be valid before
dereferencing it.  Found by Klocwork.
Signed-off-by: NSharon Dvir <sharon.dvir@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In the end, the firmware doesn't want the SP len as present
in the WMM IE, but rather the actual number of frames.

Fixes: bd3c6cf901a8 ("iwlwifi: mvm: tell the firmware about the U-APSD parameters")
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When we get SN that is smaller than SSN of the aggregation,
we shouldn't apply any reordering on them.
Further more, HW NSSN will be zeroed, which can cause us
to make some invalid decisions.
Detect the situation and invalidate the BAID.

Fixes: b915c101 ("iwlwifi: mvm: add reorder buffer per queue")
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In TVQM firmware returns the value of the queue ID and code
should accept it.
The TX queue config API was changed. Move to new API.
This has to be done in parallel in mvm and pcie.
Do not move yet to 512 queues since there are some opens
with enabling it.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Newer firmware versions will be able to handle all the
WMM-PS flows internally when we act as a GO. The firwmare
relies on the fact that the drivers puts frames for
different peers in different queues (DQA) to achieve this.
The driver will not be aware of the power state of the peers
anymore.

Tell the firmware about the WMM-PS parameters of earch peer
that connects to us so that it can know what are the
trigger-enabled ACs, the delivery-enableds ACs and the
Service Period length.

This API change is backward compatible since older firmware
versions will simply ignore the newly added values.

Since we don't support ieee80211 TSPECs for now, just copy
the trigger-enabled ACs to the delivery enabled ones.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

a000 devices queue management is going to change significantly.
We will have 512 queues. Those queues will be assigned number
by the firmware and not by the driver.

In addition, due to SN offload having TX queue shared between TIDs
is impossible

Also, the ADD_STA command no longer updates queues status.
The only point of changing queue in the SCD queue config API.

From driver perspective we have here a new design:

Queue sharing and inactivity checks are disabled.

Once this is done, the only paths that call scd_queue_cfg command
are paths that alloc and release TX queues - which will make future
accommodation to queue number assignment by FW easier.
Since allocating 512 queues statically is not advisable, transport
will allocate the queue on demand, fill the command with DRAM data
and send it. This is reflected in the new transport API.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>