commit 0c60c490830a1a756c80f8de8d33d9c6359d4a36 upstream.

In order to make propagation time estimation faster,
use current sample as ewma output value during 'late ack'
tracking
Tested-by: NKoen Vandeputte <koen.vandeputte@ncentric.com>
Signed-off-by: NLorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The flag needs to be cleared for the last packet in the list, not the
first one. Fixes some issues with multicast packet loss for powersave
clients connected to an ath9k AP.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Ensure that a buffer gets tracked as part of the block-ack window as
soon as it's dequeued from the tid for the first time. Ensure that
double calls to ath_tx_addto_baw (e.g. on retransmission) don't cause
any issues.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Fixes missed indications of end of U-APSD service period to mac80211
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Prevents spurious ieee80211_sta_eosp calls.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Some functions definitions have either the initial open brace and/or
the closing brace outside of column 1.

Move those braces to column 1.

This allows various function analyzers like gnu complexity to work
properly for these modified functions.
Signed-off-by: NJoe Perches <joe@perches.com>
Acked-by: NAndy Shevchenko <andy.shevchenko@gmail.com>
Acked-by: NPaul Moore <paul@paul-moore.com>
Acked-by: NAlex Deucher <alexander.deucher@amd.com>
Acked-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Acked-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NMartin K. Petersen <martin.petersen@oracle.com>
Acked-by: NTakashi Iwai <tiwai@suse.de>
Acked-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>
Acked-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NNicolin Chen <nicoleotsuka@gmail.com>
Acked-by: NMartin K. Petersen <martin.petersen@oracle.com>
Acked-by: NSteven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

When ath9k was switched over to use the mac80211 intermediate queues,
node cleanup now drains the mac80211 queues. However, this call path is
not protected by rcu_read_lock() as it was previously entirely internal
to the driver which uses its own locking.

This leads to a possible rcu_dereference() without holding
rcu_read_lock(); but only if a station is cleaned up while having
packets queued on the TXQ. Fix this by adding the rcu_read_lock() to the
caller in ath9k.

Fixes: 50f08edf ("ath9k: Switch to using mac80211 intermediate software queues.")
Cc: stable@vger.kernel.org
Reported-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NToke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This reverts commit e937b8da.

Turns out that a new driver (mt76) is coming in through
Kalle's tree, and will conflict with this. It also has some
conflicting requirements, so we'll revisit this later.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This adds an API to mac80211 to handle scheduling of TXQs and changes the
interface between driver and mac80211 for TXQ handling as follows:

- The wake_tx_queue callback interface no longer includes the TXQ. Instead,
  the driver is expected to retrieve that from ieee80211_next_txq()

- Two new mac80211 functions are added: ieee80211_next_txq() and
  ieee80211_schedule_txq(). The former returns the next TXQ that should be
  scheduled, and is how the driver gets a queue to pull packets from. The
  latter is called internally by mac80211 to start scheduling a queue, and
  the driver is supposed to call it to re-schedule the TXQ after it is
  finished pulling packets from it (unless the queue emptied).

The ath9k and ath10k drivers are changed to use the new API.
Signed-off-by: NToke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

ath_tx_count_airtime is doing a lot of unnecessary work:

- Redundant station lookup
- Redundant rcu_read_lock/unlock
- Useless memcpy of bf->rates
- Useless NULL check of bf->bf_mpdu
- Redundant lookup of the skb tid

Additionally, it tries to look up the mac80211 queue index from the txq,
which fails if the frame was delivered via the power save queue.

This patch fixes all of these issues by passing down the right set of
pointers instead of doing extra work

Cc: stable@vger.kernel.org
Fixes: 63fefa05 ("ath9k: Introduce airtime fairness scheduling between stations")
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Acked-by: NToke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Also include common MAC alive check. This should make the hang checks
more reliable for modes where beacons are not sent and is used as a
starting point for further hang check improvements
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Starting with commit d94a461d ("ath9k: use ieee80211_tx_status_noskb
where possible") the driver uses rcu_read_lock() && rcu_read_unlock(), yet on
returning early in ath_tx_edma_tasklet() the unlock is missing leading to stalls
and suspicious RCU usage:

 ===============================
 [ INFO: suspicious RCU usage. ]
 4.9.0-rc8 #11 Not tainted
 -------------------------------
 kernel/rcu/tree.c:705 Illegal idle entry in RCU read-side critical section.!

 other info that might help us debug this:

 RCU used illegally from idle CPU!
 rcu_scheduler_active = 1, debug_locks = 0
 RCU used illegally from extended quiescent state!
 1 lock held by swapper/7/0:
 #0:
  (
 rcu_read_lock
 ){......}
 , at:
 [<ffffffffa06ed110>] ath_tx_edma_tasklet+0x0/0x450 [ath9k]

 stack backtrace:
 CPU: 7 PID: 0 Comm: swapper/7 Not tainted 4.9.0-rc8 #11
 Hardware name: Acer Aspire V3-571G/VA50_HC_CR, BIOS V2.21 12/16/2013
  ffff88025efc3f38 ffffffff8132b1e5 ffff88017ede4540 0000000000000001
  ffff88025efc3f68 ffffffff810a25f7 ffff88025efcee60 ffff88017edebdd8
  ffff88025eeb5400 0000000000000091 ffff88025efc3f88 ffffffff810c3cd4
 Call Trace:
  <IRQ>
  [<ffffffff8132b1e5>] dump_stack+0x68/0x93
  [<ffffffff810a25f7>] lockdep_rcu_suspicious+0xd7/0x110
  [<ffffffff810c3cd4>] rcu_eqs_enter_common.constprop.85+0x154/0x200
  [<ffffffff810c5a54>] rcu_irq_exit+0x44/0xa0
  [<ffffffff81058631>] irq_exit+0x61/0xd0
  [<ffffffff81018d25>] do_IRQ+0x65/0x110
  [<ffffffff81672189>] common_interrupt+0x89/0x89
  <EOI>
  [<ffffffff814ffe11>] ? cpuidle_enter_state+0x151/0x200
  [<ffffffff814ffee2>] cpuidle_enter+0x12/0x20
  [<ffffffff8109a6ae>] call_cpuidle+0x1e/0x40
  [<ffffffff8109a8f6>] cpu_startup_entry+0x146/0x220
  [<ffffffff810336f8>] start_secondary+0x148/0x170
Signed-off-by: NTobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Fixes: d94a461d ("ath9k: use ieee80211_tx_status_noskb where possible")
Cc: <stable@vger.kernel.org> # v4.9
Acked-by: NFelix Fietkau <nbd@nbd.name>
Acked-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: NGabriel Craciunescu <nix.or.die@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

These are one-line functions that just call spin_lock/unlock_bh(); turn
them into static inlines to avoid the function call overhead.
Signed-off-by: NToke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This reworks the ath9k driver to schedule transmissions to connected
stations in a way that enforces airtime fairness between them. It
accomplishes this by measuring the time spent transmitting to or
receiving from a station at TX and RX completion, and accounting this to
a per-station, per-QoS level airtime deficit. Then, an FQ-CoDel based
deficit scheduler is employed at packet dequeue time, to control which
station gets the next transmission opportunity.

Airtime fairness can significantly improve the efficiency of the network
when station rates vary. The following throughput values are from a
simple three-station test scenario, where two stations operate at the
highest HT20 rate, and one station at the lowest, and the scheduler is
employed at the access point:

                  Before   /   After
Fast station 1:    19.17   /   25.09 Mbps
Fast station 2:    19.83   /   25.21 Mbps
Slow station:       2.58   /    1.77 Mbps
Total:             41.58   /   52.07 Mbps

The benefit of airtime fairness goes up the more stations are present.
In a 30-station test with one station artificially limited to 1 Mbps,
we have seen aggregate throughput go from 2.14 to 17.76 Mbps.
Signed-off-by: NToke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The AR5416_VER_MASK macro does the same as get_eeprom_rev, except that
one has to know the actual EEPROM type (and providing a reference to
that in a variable named "eep"). Additionally the eeprom_*.c
implementations used the same shifting logic multiple times to get the
eeprom revision which was also unnecessary duplication of
get_eeprom_rev.

Also use the AR5416_EEP_VER_MINOR_MASK macro where needed and introduce
a similar macro (AR5416_EEP_VER_MAJOR_MASK) for the major version.
Finally drop AR9287_EEP_VER_MINOR_MASK since it simply duplicates the
already defined AR5416_EEP_VER_MINOR_MASK.
Signed-off-by: NMartin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This switches ath9k over to using the mac80211 intermediate software
queueing mechanism for data packets. It removes the queueing inside the
driver, except for the retry queue, and instead pulls from mac80211 when
a packet is needed. The retry queue is used to store a packet that was
pulled but can't be sent immediately.

The old code path in ath_tx_start that would queue packets has been
removed completely, as has the qlen limit tunables (since there's no
longer a queue in the driver to limit).

The mac80211 intermediate software queues offer significant latency
reductions, and this patch allows ath9k to realise them. The exact gains
from this varies with the test scenario, but in an access point scenario
we have seen latency reductions ranging from 1/3 to as much as an order
of magnitude. We also achieve slightly better aggregation.

Median latency (ping) figures with this patch applied at the access point,
with two high-rate stations and one low-rate station (HT20 5Ghz), running
a Flent rtt_fair_var_up test with one TCP flow and one ping flow going to
each station:

                                 Fast station        Slow station
Default pfifo_fast qdisc:            430.4 ms            638.7 ms
fq_codel qdisc on iface:              35.5 ms            211.8 ms
This patch set:                       22.4 ms             38.2 ms

Median aggregation sizes over the same test:

Default pfifo_fast qdisc:            9.5 pkts            1.9 pkts
fq_codel qdisc on iface:            11.2 pkts            1.9 pkts
This patch set:                     13.9 pkts            1.9 pkts

This patch is based on Tim's original patch set, but reworked quite
thoroughly.

Cc: Tim Shepard <shep@alum.mit.edu>
Cc: Felix Fietkau <nbd@nbd.name>
Signed-off-by: NToke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

For non-aggregated frames, ath9k was leaving handling of powersave
filtered packets to mac80211. This can be too slow if the intermediate
queue is already filled with packets and mac80211 does not immediately
send a new packet via drv_tx().

Improve response time with filtered frames by triggering clearing the
powersave filter internally.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It removes the need for undoing the padding changes to skb->data and it
improves performance by eliminating one tx status lookup per MPDU in the
status path. It is also useful for preparing a follow-up fix to better
handle powersave filtering.

A side effect is that these counters, available via debugfs, become now invalid:

* dot11TransmittedFragmentCount
* dot11FrameDuplicateCount,
* dot11ReceivedFragmentCount
* dot11MulticastReceivedFrameCount
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
[kvalo@qca.qualcomm.com: add a note about counters, thanks to Zefir Kurtisi]
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.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>

minor change, indenting is one tab out.
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

While mac80211 using null frames when connection polling,
we should queue this frames while NOA could be there, and
AP, P2P_GO could be not present.

Without this patch, with no traffic we often saw disconnections
while we try to send nullfunc when AP/GO wasn't present.
Signed-off-by: NJanusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Add/extend debug messages when chanctx used.
Signed-off-by: NJanusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Use list_empty(&tid->list) instead
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

struct ath_atx_ac contains a list of active TIDs belonging to one WMM AC.
This patch changes the code to track active station TIDs in the txq directly.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

While compiling ath9k with some extra flags I've
found that:

 ath9k/xmit.c +2473 ## 16: warning: restricted __le16 degrades to integer
 ath9k/xmit.c +2474 ## 36: warning: invalid assignment: &=
 ath9k/xmit.c +2474 ## 36:    left side has type restricted __le16
 ath9k/xmit.c +2474 ## 36:    right side has type int

There's no way for frame ftype/stype to be
mistreated as the offending 'moredata' flag when
considering cab queue.

This could've however theoretically led sometimes
to increased power consumption on connected
stations as they would keep their Rx active
waiting for frames that would never come.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

A long time ago, ath9k had issues during reset where the DMA engine
would stay active and could potentially corrupt memory.
To debug those issues, the driver would print warnings whenever they
occur.

Nowadays, these issues are gone and the primary cause of these messages
is if the MAC is stuck during reset or busy processing a long
transmission. This is fairly harmless, yet these messages continue to
worry users.

To reduce the number of bogus bug reports, turn these messages into
debug messages and count their occurence in the "reset" debugfs file.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Do not use ieee80211_vif pointer in ath_get_rate_txpower() since it has been
overwritten by setup_frame_info() and it will result in a corrupted tx power
configuration. Set per-packet tx power in setup_frame_info() according to
current vif tx power.
Signed-off-by: NLorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In order to add per-vif TX power capability cap per-packet TX power to vif
configured power if the latter is lower than per-rate TX power and mac80211
per-frame power. Use vif TX power if TPC has been disabled for current the
interface
Signed-off-by: NLorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Check if the frame has been completed without any
error and use IEEE80211_TX_STAT_NOACK_TRANSMITTED to
indicate successful transmission of no-ack frames.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Add TPC capability to TX descriptor path for AR9002 based chips. Scale
per-packet TX power according to eeprom power bias, power adjustments for
HT40 mode and open loop CCK rates. Cap per-packet TX power according to
TX power per-rate tables
Signed-off-by: NLorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Fix typo
Signed-off-by: NLorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Move IEEE80211_TX_CTL_PS_RESPONSE to info->control.flags since
this is used only in the TX path (by ath9k). This frees up
a bit which can be used for other purposes.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Add TPC capability to TX descriptor path. Cap per-packet TX power according to
TX power per-rate tables. Currently TPC is supported just by AR9003 based chips
Signed-off-by: NLorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Enabling HW queue control for normal (non-mcc) mode
causes problems with queue management, resulting
in traffic stall. Since it is mainly required for
fairness in MCC mode, disable it for the general case.

Bug: https://dev.openwrt.org/ticket/18164

Cc: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Currently, ath9k uses a global counter for all
frames that need to be assigned a sequence number.
QoS-data frames are handled properly since they
have a per-tid counter. But, beacons and other
management frames use the same counter even if
multiple interfaces or contexts are present.

Fix this issue by making the counter per-interface
and using it when mac80211 sets IEEE80211_TX_CTL_ASSIGN_SEQ.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

On AR9003, tx control and tx status are in separate descriptor rings.
Tx duration is extracted from the tx control descriptor data, which
ar9003_hw_proc_txdesc cannot access.

Fix getting the duration by adding a separate callback for it.
Acked-by: NLorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

chan_lock is not required for checking if
we are in the middle of a HW reset, so do it
early. This also removes the small window
where the lock is dropped and reacquired.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Clearing IEEE80211_TX_CTL_PS_RESPONSE in a frame
that is not in the current context doesn't seem right.
Instead make sure that we don't add such frames
to the UAPSD queue by using a local variable.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

There is no reason why frames marked with
IEEE80211_TX_CTL_TX_OFFCHAN have to be sent using
the UAPSD queue. Since mac80211 makes sure that
RoC is done before pushing an offchannel frame
to the driver, we can use the normal TX queues
for transmission.
Signed-off-by: NSujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>