Current ath9k code does not handle beacon timers on opmode
specific. One such example is that a STA beacon config overwrites
already configured AP vif's beacon timers during scan.

On multi station vif case, configure beacon timers beased
on primary vif selected. This also helps while moving back
to single STA vif from multi STA vifs, where the power save
is enabled and hw has to be reconfigured with proper
beacon and bssid/aid. Otherwise connection poll will be triggered
so frequently due to beacon loss.
Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

As of now bssid/aid is overridden with recently changed vif's
bss config. This may cause improper beacon updation due to
bssid/aid mismatch. On station mode, select an associated
sta vif as primary vif and configure that vif's bss into hw.
Update the primary vif on interface change and bss info change.
Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Remove unused bssid from ath_vif and set av_bslot on beacon
alloc/return.
Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The present check denies the IBSS interface addtion if we
already have any other vifs. But it fails to deny interface
addition if IBSS was already present.
Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Since the beacon timers are global, the individual vif type should not
be used to determine the beacon timer configuration mode, use the
global opmode instead.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The recent cleanups in the beacon code fixed SWBA backoff calculation,
however it did not remove a line of code that worked around the issues
from the earlier version of the code.
After the cleanup, the initial TSF based slot calculation now always
returns 0 instead of ATH_BCBUF-1, so the previous hack that reversed the
slot order needs to be removed, as ad-hoc mode does not use staggered
beacons.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NVasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This function is nowhere used.
Signed-off-by: NMohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This does not seems to be used anywhere so remove it.
Signed-off-by: NMohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NMohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

It has the same purpose (and value) as ah->config.max_txtrig_level
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

It is not used anywhere
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Similar to the number of tx queue, the number of keycache entries depends
on the chip and shouldn't be messed with based on EEPROM data.
Remove this field and stick to using AR_KEYTABLE_SIZE
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

It is not used anywhere and seems pointless
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

It is only used in one place, and the device id check that it's based on
can be moved there as well.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The EEPROM contains a field that can restrict the number of hardware queues,
however this is not only useless (all the known chips contain the same
number of hardware queues), but also potentially dangerous in case of a
misprogrammed EEPROM (could trigger driver crashes), so let's just ignore
it completely.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

RF_BANK_SETUP, REG_WRITE_RF_ARRAY and REG_WRITE_ARRAY are way too big,
so they shouldn't be inlined at every single callsite, especially since they
can easily be turned into real functions.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

It's easier to read and it slightly decreases code size
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Reduces the number of calls to register ops. On MIPS this reduces the
ath9k_hw binary size from 321k down to 310k
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

With this change, loading the address to a register read/write function
costs only one pointer dereference instead of two. On MIPS this reduces
ath9k_hw binary size from 326k down to 321k.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Add support for using RSN IBSS with ath9k. For now, this uses software
crypto for group addressed frames in RSN IBSS, but that may be
optimized in the future by extending the key cache design to support
per-STA RX GTK.
Signed-off-by: NJouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Stuck beacon detection is supposed to trigger when 9 consecutive beacons
could not be sent by the hardware. When the driver runs only one active
AP mode interface, it still configures the hardware beacon timer for
4 (ATH_BCBUF) beacon slots slots, which causes stuck beacon detection
to be reset if ath9k_hw_stoptxdma clears the stuck frames between
SWBA intervals.
Fix this by not resetting the missed beacon count for empty slots and
multiplying the threshold not by the maximum number of beacon slots
but by the configured number of beacon interfaces.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Single missed (i.e. not transmitted) beacons in AP mode are not very rare
and not necessarily an indicator of strong interference, so only trigger
noise floor recalibration when multiple consecutive beacons could not
be transmitted.
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

AP mode beacon timers in ath9k are configured in milliseconds, which breaks
when increasing ATH_BCBUF to 8 instead of 4 (due to rounding errors).
Since the hardware timers are actually configured in microseconds, it's
better to let the driver use that unit directly.

To be able to do that, the beacon interval parameter abuse for passing
certain flags needs to be removed. This is easy to do, because those flags
are completely unnecessary anyway. ATH9K_BEACON_ENA is ignored,
ATH9K_BEACON_RESET_TSF can be replaced with calling ath9k_hw_reset_tsf
from the driver directly.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Only AR9485 1.1 was sold. This debloats the driver by ~14 KiB.

   text    data     bss     dec     hex filename
 300413     624    1056  302093   49c0d drivers/net/wireless/ath/ath9k/ath9k_hw.ko

   text    data     bss     dec     hex filename
 310285     624    1056  311965   4c29d drivers/net/wireless/ath/ath9k/ath9k_hw-old.ko

$ du -b ath9k_hw*
6210541	ath9k_hw.ko
6225089	ath9k_hw-old.ko

Cc: Bill Wu <bill.wu@atheros.com>
Cc: Paul Shaw <paul.shaw@atheros.com>
Cc: Forbes Tsai <Forbes.Tsai@Atheros.com>
Cc: Jesmine Chen <jesmine.chen@atheros.com>
Cc: Marvian Chen <Hou-hua.Chen@Atheros.com>
Cc: Vivek Natarajan <vivek.natarajan@atheros.com>
Cc: Bernadette Yetso <bernadette.yetso@atheros.com>
Cc: Sarvesh Shrivastava <sarvesh.shrivastava@atheros.com>
Acked-by: NYi-Chen Su <yi-chen.su@atheros.com>
Acked-by: NJeffrey Chung <jeffrey.chung@atheros.com>
Signed-off-by: NLuis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This workaround called ath_txq_schedule whenever there were still pending
frames for a queue, but the queue depth was zero. Because of its its high
false positive probability (e.g. with paused TIDs) and because it is in
the way of other pending work (AP powersave fixes), it is better to remove
this code entirely.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Most AR9100 devices already have a chainmask of 7 (three antennas), however
on the ones that don't (rx and tx chainmask set to 5), problems with IQ
mismatch calibration have been observed.
This shows up as tx queue hangs (and subsequent hardware resets) if traffic
is sent during this type of calibration.
Forcing the rx chainmask to 7 fixes the calibration issues with no apparent
negative side effects on throughput and stability.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Fixes interop issues with aggregation in combination with multi-BSSID
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

On some devices the correct MAC address is not in the EEPROM data, but
stored somewhere else.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Some devices control antenna settings or other things through GPIO pins
of the wireless interface. Add a debugfs interface for changing those
and keeping them set across card resets.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Since the ath9k debugfs directory is cleaned up by debugfs_remove_recursive,
there's no point in checking the return code of every single debugfs create
line.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Fine-tuning register write operation and avoid unnecessay
delays for ath9k_htc driver, saves hw reset time which
improves scanning time and also solves one of the following
scenario.

Sometimes the ACK is sent by STA for assoc response is not
seen at AP side. So the AP continues to send retry assoc
responses. At the STA side, since the assoc response was
already forwarded to mac80211, it proceeded to channel change
which in turns does chip reset.

In most of the cases the chip reset was completed before
max retries are reached at AP side. Hence STA can able to ACK
the retried frames again. But in clear environment these retries
are completed within shortspan of time.

Since ath9k_htc consumes more time for hw reset, this latency
is causing dissociation by AP due to max reties are reached.
This issue was originally reported with Cisco Aironet 1250 AP
in HT40 mode in noise free environment.
Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Doing adc gain calibration for AR9287 chips is causing
throughput drops in HT40 mode. Remove ADC Gain from supported
calibration list.
Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NRajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

commit 86271e46 introduced a
regression that caused mac80211 queues in stopped state.

ath_drain_all_txq is called in driver flush which would reset
the stopped flag and the mac80211 queues were never started
after that. iperf traffic is completely stalled due to this issue.

Restart the mac80211 queues in driver flush only if the txqs were
drained.
Signed-off-by: NSenthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

With the recent tx status optimization in mac80211, we bail out as
and and when invalid rate index is found. So the behavior of resetting
rate idx to -1 and count to 0 has changed for the rate indexes that
were not part of the driver's retry series.

This has resulted in ath9k using incorrect rate table index which
caused the system to panic. Ideally ath9k need to loop only for the
indexes that were part of the retry series and so simply use hw->max_rates
as the loop counter.

Pasted the stack trace of the panic issue for reference.

[  754.093192] BUG: unable to handle kernel paging request at ffff88046a9025b0
[  754.093256] IP: [<ffffffffa02eac49>] ath_tx_status+0x209/0x2f0 [ath9k]
[  754.094888] Call Trace:
[  754.094903]  <IRQ>
[  754.094928]  [<ffffffffa051f883>] ieee80211_tx_status+0x203/0x9e0 [mac80211]
[  754.094975]  [<ffffffffa053e305>] ? __ieee80211_wake_queue+0x125/0x140 [mac80211]
[  754.095017]  [<ffffffffa02e66c9>] ath_tx_complete_buf+0x1b9/0x370 [ath9k]
[  754.095054]  [<ffffffffa02e6fcf>] ath_tx_complete_aggr+0x51f/0xb50 [ath9k]
[  754.095098]  [<ffffffffa05382a3>] ? ieee80211_prepare_and_rx_handle+0x173/0xab0 [mac80211]
[  754.095148]  [<ffffffff81350e62>] ? _raw_spin_unlock_irqrestore+0x32/0x40
[  754.095186]  [<ffffffffa02e9735>] ath_tx_tasklet+0x365/0x4b0 [ath9k]
[  754.095224]  [<ffffffff8107a2a2>] ? clockevents_program_event+0x62/0xa0
[  754.095261]  [<ffffffffa02e2628>] ath9k_tasklet+0x168/0x1c0 [ath9k]
[  754.095298]  [<ffffffff8105599b>] tasklet_action+0x6b/0xe0
[  754.095331]  [<ffffffff81056278>] __do_softirq+0x98/0x120
[  754.095361]  [<ffffffff8100cd5c>] call_softirq+0x1c/0x30
[  754.095393]  [<ffffffff8100efb5>] do_softirq+0x65/0xa0
[  754.095423]  [<ffffffff810563fd>] irq_exit+0x8d/0x90
[  754.095453]  [<ffffffff8100ebc1>] do_IRQ+0x61/0xe0
[  754.095482]  [<ffffffff81351413>] ret_from_intr+0x0/0x15
[  754.095513]  <EOI>
[  754.095531]  [<ffffffff81014375>] ? native_sched_clock+0x15/0x70
[  754.096475]  [<ffffffffa02bcfa6>] ? acpi_idle_enter_bm+0x24d/0x285 [processor]
[  754.096475]  [<ffffffffa02bcf9f>] ? acpi_idle_enter_bm+0x246/0x285 [processor]
[  754.096475]  [<ffffffff8127fab2>] cpuidle_idle_call+0x82/0x100
[  754.096475]  [<ffffffff8100a236>] cpu_idle+0xa6/0xf0
[  754.096475]  [<ffffffff81339bc1>] rest_init+0x91/0xa0
[  754.096475]  [<ffffffff814efccd>] start_kernel+0x3fd/0x408
[  754.096475]  [<ffffffff814ef347>] x86_64_start_reservations+0x132/0x136
[  754.096475]  [<ffffffff814ef451>] x86_64_start_kernel+0x106/0x115
[  754.096475] RIP  [<ffffffffa02eac49>] ath_tx_status+0x209/0x2f0 [ath9k]
Signed-off-by: NSenthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Kernel panic occurs just after AR2427 establishes connection with AP.
Unless aggregation is enabled we don't initialize the TID structure.
Thus accesing the elements of the TID structure when aggregation is
disabled, leads to NULL pointer dereferencing.

[  191.320358] Call Trace:
[  191.320364]  [<fd250ea7>] ? ath9k_tx+0xa7/0x200 [ath9k]
[  191.320376]  [<fd1ec7fc>] ? __ieee80211_tx+0x5c/0x1e0 [mac80211]
[  191.320386]  [<fd1edd2b>] ? ieee80211_tx+0x7b/0x90 [mac80211]
[  191.320395]  [<fd1edddd>] ? ieee80211_xmit+0x9d/0x1d0 [mac80211]
[  191.320401]  [<c014218f>] ? wake_up_state+0xf/0x20
[  191.320405]  [<c015dbc8>] ? signal_wake_up+0x28/0x40
[  191.320410]  [<c012a578>] ? default_spin_lock_flags+0x8/0x10
[  191.320420]  [<fd1ee308>] ? ieee80211_subif_start_xmit+0x2e8/0x7c0
[mac80211]
[  191.320425]  [<c058f905>] ? do_page_fault+0x295/0x3a0
[  191.320431]  [<c04c4a3d>] ? dev_hard_start_xmit+0x1ad/0x210
[  191.320436]  [<c04d96b5>] ? sch_direct_xmit+0x105/0x170
[  191.320445]  [<fd1f161a>] ? get_sta_flags+0x2a/0x40 [mac80211]
[  191.320449]  [<c04c780f>] ? dev_queue_xmit+0x37f/0x4b0
[  191.320452]  [<c04d75b0>] ? eth_header+0x0/0xb0
[  191.320456]  [<c04cc479>] ? neigh_resolve_output+0xe9/0x310
[  191.320461]  [<c053d295>] ? ip6_output_finish+0xa5/0x110
[  191.320464]  [<c053e354>] ? ip6_output2+0x134/0x250
[  191.320468]  [<c053f7dd>] ? ip6_output+0x6d/0x100
[  191.320471]  [<c0559665>] ? mld_sendpack+0x395/0x3e0
[  191.320475]  [<c0557f81>] ? add_grhead+0x31/0xa0
[  191.320478]  [<c055a83c>] ? mld_send_cr+0x1bc/0x2b0
[  191.320482]  [<c01535d9>] ? irq_exit+0x39/0x70
[  191.320485]  [<c055a940>] ? mld_ifc_timer_expire+0x10/0x40
[  191.320489]  [<c015b92e>] ? run_timer_softirq+0x13e/0x2c0
[  191.320493]  [<c0103a30>] ? common_interrupt+0x30/0x40
[  191.320498]  [<c055a930>] ? mld_ifc_timer_expire+0x0/0x40
[  191.320502]  [<c0153358>] ? __do_softirq+0x98/0x1b0
[  191.320506]  [<c01534b5>] ? do_softirq+0x45/0x50
[  191.320509]  [<c0153605>] ? irq_exit+0x65/0x70
[  191.320513]  [<c05917dc>] ? smp_apic_timer_interrupt+0x5c/0x8b
[  191.320516]  [<c0103df1>] ? apic_timer_interrupt+0x31/0x40
[  191.320521]  [<c016007b>] ? k_getrusage+0x12b/0x2f0
[  191.320525]  [<c039e384>] ? acpi_idle_enter_simple+0x117/0x148
[  191.320529]  [<c04a20da>] ? cpuidle_idle_call+0x7a/0x100
[  191.320532]  [<c01021d4>] ? cpu_idle+0x94/0xd0
[  191.320536]  [<c057ab88>] ? rest_init+0x58/0x60
[  191.320541]  [<c07a58ec>] ? start_kernel+0x351/0x357
[  191.320544]  [<c07a53c7>] ? unknown_bootoption+0x0/0x19e
[  191.320548]  [<c07a50aa>] ? i386_start_kernel+0xaa/0xb1
[  191.320550] Code: 03 66 3d 00 03 0f 84 7c 02 00 00 83 c3 18 0f b6 03
8b 4d e0 89 c3 83 e3 0f 6b c3 48 89 5d d8 8d 04 06 8d 50 0c 89 55 d0 8b
40 20 <8b> 00 3b 01 0f 85 8e 02 00 00 f6 47 20 40 0f 84 29 ff ff ff 8b
[  191.320634] EIP: [<fd2586d4>] ath_tx_start+0x474/0x770 [ath9k] SS:ESP
0068:c0761a90
[  191.320642] CR2: 0000000000000000
[  191.320647] ---[ end trace 9296ef23b9076ece ]---
[  191.320650] Kernel panic - not syncing: Fatal exception in interrupt

Cc: stable@kernel.org
Signed-off-by: NMohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Some clients seems to keep track of their reorder window even after an
aggregation session has been disabled. This causes issues if there are
still retried but not completed frames pending for the TID.
To ensure that rx does not stall in such situations, set sendbar to 1
for any frame purged from the TID queue on teardown.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

ath9k calls ath9k_hw_stoptxdma every time it sends a beacon, however there
is not much point in doing that if the previous beacon and mcast traffic
went out properly. On AR9380, calling that function too often can result
in an increase of stuck beacons due to differences in the handling of the
queue enable/disable functionality.

With this patch, the queue will only be explicitly stopped if the previous
data frames were not sent successfully. With the beacon code being the
only remaining user of ath9k_hw_stoptxdma, this function can be simplified
in order to remove the now pointless attempts at waiting for transmission
completion, which would never happen at this point due to the different
method of tx scheduling of the beacon queue.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>