There is a debugfs knob to configure the maximal length
of the A-MSDU. If this value is not 0 (which is the
default), allow Tx A-MSDU even if the rate control
disallows it.
While at it, add "unlikely" to the if that limits the
length of the A-MSDU based on the debugfs hook.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Support sending P2P device frames should be sent from
queue #2, as required in DQA mode.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In DQA mode, allocate a dedicated queue (#9) for P2P GO/soft
AP probe responses.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The TSO code creates A-MSDUs from a single large send. Each
A-MSDU is an skb and skb->len doesn't include the number of
bytes which need to be added for the headers being added
(subframe header, TCP header, IP header, SNAP, padding).

To be able to set the right value in the Tx command, we
put the number of bytes added by those headers in
driver_data in iwl_mvm_tx_tso and use this value in
iwl_mvm_set_tx_cmd.

The problem by setting this value in driver_data is that
it overrides the ieee80211_tx_info. The bug manifested
itself when we send P2P related frames in CCK since the
rate in ieee80211_tx_info is zero-ed. This of course is
a violation of the P2P specification.

To fix this, copy the original ieee80211_tx_info to the
stack and pass it to the functions which need it.
Assign the number of bytes added by the headers to the
driver_data inside the skb itself.

Fixes: a6d5e32f ("iwlwifi: mvm: send large SKBs to the transport")
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

This enum is already perfectly aliased to enum nl80211_band, and
the only reason for it is that we get IEEE80211_NUM_BANDS out of
it. There's no really good reason to not declare the number of
bands in nl80211 though, so do that and remove the cfg80211 one.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Declare and enable support of RX and TX checksum for 9000 family.
Configure offload_assist in the TX cmd accordingly to support
TX csum.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

"DQA" is shorthand for "dynamic queue allocation". This
enables on-demand allocation of queues per RA/TID rather than
statically allocating per vif, thus allowing a potential
benefit of various factors.

Please refer to the DOC section this patch adds to sta.h to
see a more in-depth explanation of this feature.

There are many things to take into consideration when working
in DQA mode, and this patch is only one in a series. Note that
default operation mode is non-DQA mode, unless the FW
indicates that it supports DQA mode.

This patch enables support of DQA for a station connected to
an AP, and works in a non-aggregated mode.

When a frame for an unused RA/TID arrives at the driver, it
isn't TXed immediately, but deferred first until a suitable
queue is first allocated for it, and then TXed by a worker
that both allocates the queues and TXes deferred traffic.

When a STA is removed, its queues goes back into the queue
pools for reuse as needed.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

"is_data_qos == true" is equivalent to "tid < IWL_MAX_TID_COUNT"
since tid is only assigned (and range-checked) in that case.

This removes a (harmless) smatch warning that occurs because it
can't seem to follow the above logic from the code.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

TX CMD API has changed to support offload assist.
Currently we do not enable checksum yet, but must set the
padding indication, to avoid FW errors.
Set other amsdu flag as well.
The rest of the flags will be configured only if HW csum
is enabled and will be set in future patches.
This change is backward compatible.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

We need to track the next packet that we will reclaim in
order to know when the Tx queues are empty. This is useful
when we open or tear down an A-MPDU session which requires
to switch queue.
The next packet being reclaimed is identified by its WiFi
sequence number and this is relevant only when we use QoS.
QoS NDPs do have a TID but have a meaningless sequence
number. The spec mandates the receiver to ignore the
sequence number in this case, allowing the transmitter to
put any sequence number. Our implementation leaves it 0.
When we reclaim a QoS NDP, we can't update the next_relcaim
counter since the sequence number of the QoS NDP itself is
invalid.
We used to update the next_reclaim based on the sequence
number of the QoS NDP which reset it to 1 (0 + 1) and
because of this, we never knew when the queue got empty.
This had to sad consequence to stuck the A-MPDU state
machine in a transient state.
To fix this, don't update next_reclaim when we reclaim
a QoS NDP.

Alesya saw this bug when testing u-APSD. Because the
A-MPDU state machine was stuck in EMPTYING_DELBA, we
updated mac80211 that we still have frames for that
station when it got back to sleep. mac80211 then wrongly
set the TIM bit in the beacon and requested to release
non-existent frames from the A-MPDU queue. This led to
a situation where the client was trying to poll frames
but we had no frames to send.
Reported-by: NAlesya Shapira <alesya.shapira@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The A-MSDU must be smaller than the Transmit FIFO in the
device. Since the size of the TXF can change depending
on the device / firmware compilation mode, take the size
of the FIFO dynamically from the what the firmware tells us.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Allow A-MSDU only when we are not downscaling and the
initial MCS is at least 5.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

in order to be able to tune the size of the desired A-MSDU
based on link condition, add a knob to modify the length
of the A-MSDU.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If the peer allows, we can have A-MSDU inside A-MDPU.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Now that PCIe knows how to create A-MSDUs, use this
capability and prepare SKBs that are large enough to
build an A-MSDU.
Advertise TSO support towards the network stack and
segment the packet with gso_size set to be the maximal
A-MSDU length (after having taken the headers to be added
into account) to make sure that the skb that is passed
down to the transport are not longer than the maximal
A-MSDU allowed.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If protocol offloading is configured, the fw might generate some
frames (e.g. arp response) on its own during d3/d0i3.

On d3/d0i3 exit the driver queries the updated PN (if relevant),
and updates its keys (for the d0i3 case, this is done by
iwl_mvm_d0i3_exit_work(), which is scheduled on d0i3 exit)

While in d0i3, iwlmvm defers tx frames until d0i3 exit, and
then continues their processing.

This is problematic with TKIP, since the frame's PN has already
been set at this stage (in contrast to CCMP, where the PN is
being set only later on), so both the frame's PN and the upcoming
PN update (from d0i3 exit work) might be wrong.

Fix it by moving the TX PN assignment (for TKIP) to the driver,
similarly to CCMP.
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Until this patch, when TXing non-sta the pending_frames counter
wasn't increased, but it WAS decreased in
iwl_mvm_rx_tx_cmd_single(), what makes it negative in certain
conditions. This in turn caused much trouble when we need to
remove the station since we won't be waiting forever until
pending_frames gets 0. In certain cases, we were exhausting
the station table even in BSS mode, because we had a lot of
stale stations.

Increase the counter also in iwl_mvm_tx_skb_non_sta() after a
successful TX to avoid this outcome.

CC: <stable@vger.kernel.org> [3.18+]
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

This will allow to collect the data when the firmware
sends a specific tx response status.
Signed-off-by: NGolan Ben-Ami <golan.ben.ami@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

FW behaviour changed and now updates driver about the used TPC
reduction in the following cases:
1. In tx response, which is used mostly for a single frame case
2. In BA notification

When tx aggregation fails with the initial rate, FW will send
to the driver BA notification and will try to transmit with the
next rate, but this time without tx power reduction. Thus, in case
of a failure with the initial rate, driver will get two BA notifications,
the first one with reduced tx power as in the LQ command and the second
one with 0 power reduction.

This patch adapts the TPC statistics according to the description above:
1. Use BA notifications instead of Tx response
2. For TPC only, drop the optimization which considers empty BA as one
MPDU. The reason is that with TPC we want to recover very quickly from
a bad power reduction and, therefore we'd like the success ratio to get
an immediate hit when failing to get a BA, so we'd switch back to a
lower or zero power reduction
Signed-off-by: NGregory Greenman <gregory.greenman@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The transport will need to use the info->driver_data
pointers. Since the op_mode has this memory hot in cache,
clear it there.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Differentiate between the cases where the skb is a large
send and the other cases.
Advertise TSO even if, at this stage, skb_gso_segment will
be called and it will do all the work.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

ilw@linux.intel.com is not available anymore.
linuxwifi@intel.com should be used instead.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

When we have holes in the BA window, there might be frames
that have been ACKed between the read and the right
pointers. This means that these frames won't be scheduled
again by the SCD and the firwmare won't see them.
This invalidates the number of frames we tell the firmware
to send. When we detect this case, tell mac80211 to close
the SP and to send an EOSP so that the firmware can be in
sync.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The fw debug functionality is big enough to warrant
a separate file. Move existing related functions to the new file.
Signed-off-by: NGolan Ben-Ami <golan.ben.ami@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Part of reorganising wireless drivers directory and Kconfig.
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Instead of only allowing the caller to decide whether the CMD_ASYNC
flag is set, let it pass the entire flags bitmask.  This allows more
flexibility and will be needed when we call this function in the
suspend flow (where other flags are needed).
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

"DQA" is shorthand for "dynamic queue allocation", with the
idea of allocating queues per-RA/TID on-demand rather than
using shared queues statically allocated per vif. The goal
of this is to enable future features (like GO PM) and to
improve performance measurements of TX traffic.

When RA/TID streams can't be neatly sorted into different AC
queues, DQA allows sharing queues for the same RA. This means
that DQA allows different ACs may reach the same HW queue.

Update the code to allow such queue sharing by having a mapping
between the HW queue and the mac80211 queues using it (as this
could be more than one queue).
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When building a Tx Command for management frames, we are lacking
a check for action frames, for which we should set a different
pm_timeout.  This cause the fw to stay awake for 100TU after each
such frame is transmitted, resulting an excessive power consumption.
Signed-off-by: NAvri Altman <avri.altman@intel.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Move the TX PN assignment (for CCMP only) to the driver. This prepares
the driver for future DSO (driver segmentation offload) where it will
split an SKB into multiple MPDUs by itself.

For TDLS, split out the CCMP TX command handling so that it won't get
a PN assigned, the firmware assigns the PN in that case.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

In iwl_mvm_tx_skb_non_sta(), in case of managed interface,
use the AP station for multicast frames instead of the auxiliary
station as otherwise the frames can be sent to an absent P2P GO as
the FW does not block transmissions for the auxiliary station
since it is not associated with the station MAC context.

Note that this is not possible for unicast frames, as a TDLS
discovery response is sent without a station entry, and in this
case the P2P GO NoA should not block transmission to the peer.
Signed-off-by: NIlan Peer <ilan.peer@intel.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

In the mvm driver, neither the old command nor the return value
are used, so remove them.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

When BT is active, we want to avoid the shared antenna for
management frame to make sure we don't disturb BT. There
was a bug in that code because it chose the antenna
BIT(ANT_A) where ANT_A is already a bitmap (0x1). This
means that the antenna chosen in the end was ANT_B.
While this is not optimal on devices with 2 antennas (it'd
disturb BT), it is critical on single antenna devices like
3160 which couldn't connect at all when BT was active.

This fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=97181

CC: <stable@vger.kernel.org> [3.17+]
Fixes: 34c8b24f ("iwlwifi: mvm: BT Coex - avoid the shared antenna for management frames")
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Currently, loading the firmware fails when it has higher API or CAPA
bits than the driver supports. That's an issue with integration.

At the same time, actually using api[0] and capa[0] will become
confusing when we also have api[1] and capa[1], and it's almost
certain that we'll mix up the bits and use the bits for api[1] with
api[0] by accident.

Avoid all this by translating the API/CAPA bits to the regular kernel
test_bit() format, and also providing wrapper functions. Also use the
__bitwise__ facility of sparse to check that we're testing the right
one.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

BlockAck sessions can have events that are interesting to
debug. When we send or receive a BAR, it is may indicate
that something bad is happening. Even more so when mac80211
tells us that a frame timed out in the reodering buffer.
Add a few triggers for BlockAck session debugging.
Allow per-TID debugging.
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The firmware frequently manages to trigger this, and there's
no known driver workaround, so stop warning.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

There are transport that must buffer frames in the driver.
This means that we have frames that are not in the op_mode
and not visible to the firwmare. This causes issues when we
flush the queues: the op_mode flushes a queue, and the
firmware flushes all the frames that are *currently* on the
rings, but if the transport buffers frames, it can submit
these while we are flushing. This leads to a situation
where we still have frames on the queues after we flushed
them.
Preventing those buffered frame from getting into the
firmware is possible, but then, we have to run the Tx
response path on frames that didn't reach the firmware
which is not desirable.
The way I solve this here is to let these frames go to the
firmware, but make sure the firmware will not transmit them
(by setting the station as draining). The op_mode then needs
to wait until the transport itself is empty to be sure that
the queue is really empty.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Currently rs uses the info in mac80211 tx status which is
a translation of the actual rate coming up from the FW in
the tx response. This is matched up against the LQ table first
rate to make sure this tx frame used the current LQ table.
Instead of using the translated mac80211 info it's easier and
cleaner to just pass the actual tx response rate in the driver
private data and use that for matching.
This becomes even more important once the FW begins to
decide on its own whether to use STBC/BFER/SISO.
Signed-off-by: NEyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

A new host command can be used to configure the scheduler
instead of accessing the scheduler's registers from the
driver. This is easier and less error prone since accessing
the hardware at certain moments can lead to races with the
firmware.
Prefer to use the host command whenever it is available.
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If Tx failed because the STA was in powersave there's no point
in sending a BAR so avoid indicating AMPDU_NO_BACK to mac80211.
Signed-off-by: NEyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

These are coming from the FW and are used to access arrays.
Bad values can cause an out of bounds access so discard
such ba_notifs and warn.

CC: <stable@vger.kernel.org> [3.10+]
Signed-off-by: NEyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>