Use spinlock, this should not impact Tx as lock is always free
except for de-allocation.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Non-cacheable Tx descriptor for the last fragment of
multi-fragment frame used to be written back twice.

Refactor code to always write non-cacheable descriptor
only once
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In the Tx, work around used to force destination index 0
to be used. This is no more necessary, as firmware supports
multiple destinations
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Since the maximum number of configurable patterns
is chip-specific, use the HW capability instead
of a fixed value for checking if a free pattern
slot is available.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

* Remove unnecessary comments.
* Remove unused macros.
* Indent the macros.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

These macros are applicable to pre-AR9003 chips
and the addresses are different for the AR9003
family. Since they are unused anyway, remove them.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Since the number of patterns that can be configured in
the HW is higher for newer chips, store the chip-specific
value in ath9k_hw_wow.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This can be used to hold the WOW state in
ath9k_hw.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

There is no need to allocate a new structure and
free it for every user pattern, instead use local
variables.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This can be used to force WOW for cards that
are not present in the supported PCI ID list.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Printing the value of the wakeup status is sufficient.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

The various variables tracking bmiss interrupts
are not really used anywhere except in a debug
message. Remove them since they have no functional
purpose.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This patch makes sure that valid WOW triggers
are present before trying to suspend the device.
Also, introduce and use ATH_OP_WOW_ENABLED to
bypass PCI suspend and clear it in resume().
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Registering the card as a wakeup source needs to
be done once, during initialization. When the WOW
configuration changes, the card's status as wakeup
source needs to be changed too and this is done
via the set_wakeup() callback. Also, make sure
the device is removed properly using ath9k_deinit_wow().
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

If CONFIG_ATH9K_CHANNEL_CONTEXT is enabled, check whether
multiple contexts are active and if so, return 1 without
enabling WOW since we don't support it in this case.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

If multiple interfaces are active or there is no
associated station interface, bail out early and
return 1 so that mac80211 can proceed with the normal
suspend routine.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

WOW capability is registered with mac80211 only when
the device has the ability to wakeup, so there is no
need to check in the suspend() routine.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Do not try to cancel work instances and ANI when
the device is not present or WOW triggers are
not configured. Bail out early and use ath_err()
for such error conditions.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Enabling WOW based on the chip is incorrect since
it needs to be done for specific sub-devices which
have proper platform support.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

On some AR955x/QCA953x boards, noise floor calibration
gets stuck and the cause is a hardware/BB issue. To fix
this, peak detect calibration in the HW is disabled
and is done in the driver instead. There a few differences
with the calibration routine for older chips like
AR9331.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

The value is different for PCOEM cards and AR955x/AR953x.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

The LNA gain setting override needs to be done
only for AR9330 and PCOEM chips.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This can be used if we need to apply register settings
for all PCOEM solutions (in the AR9003 family).
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This patch disables HW peak detect calibration for
AR955x, since it is problematic on a few boards.
Instead, manual calibration will be done in the
driver.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This patch disables HW peak detect calibration for
QCA953x, since it is problematic on a few boards.
Instead, manual calibration will be done in the
driver.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This fixes issue reported by Dan Carpenter:
The patch 3277213f: "wil6210: ADDBA/DELBA flows" from Dec 23,
2014, leads to the following static checker warning:

        drivers/net/wireless/ath/wil6210/rx_reorder.c:205 wil_tid_ampdu_rx_alloc()
        error: scheduling with locks held: 'spin_lock:tid_rx_lock'

drivers/net/wireless/ath/wil6210/rx_reorder.c
   202  struct wil_tid_ampdu_rx *wil_tid_ampdu_rx_alloc(struct wil6210_priv *wil,
   203                                                  int size, u16 ssn)
   204  {
   205          struct wil_tid_ampdu_rx *r = kzalloc(sizeof(*r), GFP_KERNEL);
                                                                 ^^^^^^^^^^
   206
   207          if (!r)
   208                  return NULL;
   209
   210          r->reorder_buf =
   211                  kcalloc(size, sizeof(struct sk_buff *), GFP_KERNEL);
                                                                ^^^^^^^^^^^
   212          r->reorder_time =
   213                  kcalloc(size, sizeof(unsigned long), GFP_KERNEL);
                                                             ^^^^^^^^^^^
   214          if (!r->reorder_buf || !r->reorder_time) {
   215                  kfree(r->reorder_buf);
   216                  kfree(r->reorder_time);
   217                  kfree(r);
   218                  return NULL;
   219          }
   220

[ snip ]

   331          spin_lock_bh(&sta->tid_rx_lock);
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
spin lock held.

   332
   333          wil_tid_ampdu_rx_free(wil, sta->tid_rx[tid]);
   334          sta->tid_rx[tid] = wil_tid_ampdu_rx_alloc(wil, agg_wsize, ssn);
                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
function called with the lock held.

   335
   336          spin_unlock_bh(&sta->tid_rx_lock);
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Access point require this API to check peer alive status.
Assume peer is alive when it is connected, because
firmware implements keep alive checks and will disconnect
peer if it is not alive.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Implement Tx status reporting using skb_complete_wifi_ack().
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Interrupt moderation parameters will never be passed as module
parameters. For product, they will be hard-coded after lab testing,
and for debugging, they can be altered via debugfs.
Signed-off-by: NVladimir Shulman <qca_shulmanv@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Due to HW limitation, inter-packet gap timeout max value is 13 usec.
Update of current thresholds from 15 to 13 usec.
Signed-off-by: NVladimir Shulman <qca_shulmanv@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Tuning rings size for performance optimization.
Increasing Tx ring size, allows buffering more packets for HW, thus
eliminating idle periods which were observed with smaller ring at
high throughput, because HW was fetching packets faster than driver was
filling them into the TX ring. Rx ring was similarly increased to
avoid same problems in Rx.
Signed-off-by: NVladimir Shulman <qca_shulmanv@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Tx queue was hard-coded to 1000 in ether_setup. Add wil_dev_setup
function which configures tx queue len to chosen default value
after calling ether_setup.
Signed-off-by: NVladimir Shulman <qca_shulmanv@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

upon cfg80211_stop_ap, a graceful AP shutdown is requested from firmware
followed by firmware reset.
In case graceful request failed, error was returned to cfg80211.

The change is to return success in this scenario, because firmware
reset will anyhow shutdown the AP.
Signed-off-by: NDedy Lansky <qca_dlansky@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

netif_carrier_on indication was too late. In case Rx packet received
before netif_carrier_on indication, upper layers could not send
Tx packet back.

The fix is to indicate netif_carrier_on earlier:
for STA, indicate netif_carrier_on when association starts.
for AP/PCP, indicate netif_carrier_on upon starting AP/PCP.
Signed-off-by: NDedy Lansky <qca_dlansky@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Incorporate changes from firmware.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

ath9k and ath9k_htc use the variable name "led_blink" to indicate
whether the module parameter "blink" is on. This name is easy to
conflict with other variables, and has caused a compiler error found
by kbuild test bot. The compiler error is as following:

drivers/net/wireless/ath/ath9k/ath9k_htc.o:(.data+0x47c): multiple definition of `led_blink'
drivers/net/wireless/ath/ath9k/ath9k.o:(.bss+0x20): first defined here

Fixes: 3a939a67 ("ath9k_htc: Add a module parameter to disable blink")
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NHong Xu <hong@topbug.net>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Coverity: CID 114932
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
Acked-by: NNick Kossifidis <mickflemm@gmail.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This clause is conditioned on htc_hdr != NULL, but it will only be NULL
when that check is reached.

Coverity: CID 114318
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Add an option "blink" to enable or disable the LED blink. The default
value is set to 1 so that existing users would not experience any
unexpected changes.
Signed-off-by: NHong Xu <hong@topbug.net>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>