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: 63fefa0504 ("ath9k: Introduce airtime fairness scheduling between stations")
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle 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: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
The most notable change here is the inclusion of airtime fairness
scheduling to ath9k. It prevents slow clients from hogging all the
airtime and unfairly slowing down faster clients.
Otherwise smaller changes and cleanup.
Major changes:
ath9k
* cleanup eeprom endian handling
* add airtime fairness scheduling
ath10k
* fix issues for new QCA9377 firmware version
* support dev_coredump() for firmware crash dump
* enable channel 169 on 5 GHz band
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
iQEcBAABAgAGBQJYalDTAAoJEG4XJFUm622ba2EH/ivoHG9CvmT6HgNw77odDr6b
Zc7plDi2dd9LIzp3rF/9j8Ietfzpu31Pp9Qz0o1Su/s8RJIlXKVF5UeG5IignnEG
02rv7Z+BGPW7SUNDb0cLa27PMYqUAh8vI39sxYSdQ9NQLIlpMLIxxx5iOZyAK5eV
HBRS0pVlSxbAntb1JgWiN/LjjXe/PdhX3CR/b9b468KAcWB6UWj9nXQxFmz7d3uM
dH3buDTgwaq7K24ZAFc1BsD7rgxprpauXxgIMWCKxRERf64WFfYkYdfdCER+PSPH
maV5Te1KzTJBrQHDOPs1ezVe8gDUhb5YE4BJY9SppZpOE1arucq2g3+3obQUeHE=
=DAVq
-----END PGP SIGNATURE-----
Merge tag 'wireless-drivers-next-for-davem-2017-01-02' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers-next
Kalle Valo says:
====================
wireless-drivers-next patches for 4.11
The most notable change here is the inclusion of airtime fairness
scheduling to ath9k. It prevents slow clients from hogging all the
airtime and unfairly slowing down faster clients.
Otherwise smaller changes and cleanup.
Major changes:
ath9k
* cleanup eeprom endian handling
* add airtime fairness scheduling
ath10k
* fix issues for new QCA9377 firmware version
* support dev_coredump() for firmware crash dump
* enable channel 169 on 5 GHz band
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
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: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle 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: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle 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: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Kalle 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: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle 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: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Kalle 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: Felix Fietkau <nbd@nbd.name>
[kvalo@qca.qualcomm.com: add a note about counters, thanks to Zefir Kurtisi]
Signed-off-by: Kalle 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: Johannes Berg <johannes.berg@intel.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: Janusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
Add/extend debug messages when chanctx used.
Signed-off-by: Janusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
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: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Kalle 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: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle 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: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Kalle 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: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: Kalle 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: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: Kalle 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Kalle 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: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: Kalle 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Johannes 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: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: John 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John 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: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John 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: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Since we use IEEE80211_HW_QUEUE_CONTROL now, the
CAB/Offchannel queues are registered as the last
two queues. There is no need to check and reassign
the queues in the TX start()/done() routines.
CAB frames will not reach the tx() callback since
we set IEEE80211_HW_HOST_BROADCAST_PS_BUFFERING and
pull the buffered frames during beacon transmission.
We also don't have a special HW queue for handling
off-channel frames.
Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Add sampling methods for ACK RX timestamp in ath_rx_tasklet() and for TX frame
timestamp in ath_tx_complete_aggr() and in ath_tx_process_buffer(). These
samples will be used in dynack processing for ACK timeout estimation
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Check if channel context usage is enabled before
calling ath_chanctx_event() from various parts of the
driver. Also, make sure that ath_chanctx_event() is
compiled only when CONFIG_ATH9K_CHANNEL_CONTEXT is
enabled.
Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
If an aggregation session fails, frames still end up in the driver queue
with IEEE80211_TX_CTL_AMPDU set.
This causes tx for the affected station/tid to stall, since
ath_tx_get_tid_subframe returning packets to send.
Fix this by clearing IEEE80211_TX_CTL_AMPDU as long as no aggregation
session is running.
Cc: stable@vger.kernel.org
Reported-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Packets originally buffered for the regular hardware tx queues can end
up being transmitted through the U-APSD queue (via PS-Poll or U-APSD).
When packets are dropped due to retransmit failures, the pending frames
counter is not always updated properly.
Fix this by keeping track of the queue that a frame was accounted for in
the ath_frame_info struct, and using that on completion to decide
whether the counter should be updated.
This fixes some spurious transmit queue hangs.
Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Use seperate tx queue for each AC in each channel context and expose
these information to mac80211 to avoid stopping one channel context
leads to stopping the entire traffic for that AC even on other contexts.
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Add a basic state machine for switch channel context
for beacon transmission.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Add remain on channel support in order to enable multi-channel
concurrency.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Force queueing of all frames that belong to a virtual interface on
a different channel context, to ensure that they are sent on the
correct channel.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The channel context worker is used to switch to next requested
channel context.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Add support to maintain per-channel ACs list.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
There are some corner cases where the driver could get stuck with a full
tid queue that is paused, leading to a software tx queue hang.
Since the tx queueing rework, pausing per-tid queues on aggregation
session setup is no longer necessary. The driver will assign sequence
numbers to buffered frames when a new session is established, in order
to get the correct starting sequence number.
mac80211 prevents new frames from entering the queue during setup.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Conflicts:
Documentation/devicetree/bindings/net/micrel-ks8851.txt
net/core/netpoll.c
The net/core/netpoll.c conflict is a bug fix in 'net' happening
to code which is completely removed in 'net-next'.
In micrel-ks8851.txt we simply have overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
we will need it for ath9k_htc, may be other drivers too
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
qi->tqi_readyTime is written directly to registers that expect
microseconds as unit instead of TU.
When setting the CABQ ready time, cur_conf->beacon_interval is in TU, so
convert it to microseconds before passing it to ath9k_hw.
This should hopefully fix some Tx DMA issues with buffered multicast
frames in AP mode.
Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Since commit 558ff225de (ath9k: fix
ps-poll responses under a-mpdu sessions) non-data frames would have
gotten a sequence number from a TIDs sequence counter instead of
using the global sequence counter.
This can lead to instable connections.
To fix this only select the correct TID if we are processing a
data frame. Furthermore, prevent non-data frames to get a sequence
number from a TID sequence counter by adding a check to
ath_tx_setup_buffer.
Cc: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau